Mathias Prokop Michael Galanski Co-Editors
Aartj. van der Molen Cornelia M. Schaefer-Prokop
'Thieme
Abbreviation Meaning AAA MO-HNS AAST ABPA ACC ACE ACKD ACS ACTH ACVB
abdominal aortic aneurysm American Academy ofOtolaryngology-Head Neck Surgery American Association for the Surgery ofTrauma allergic bronchopulmonary aspergillosis N-acetylcysteine acetylcholinesterase acquired cystic kidney disease anterior cervical space adenocorticotropic hormone aortocoronary venous bypass graft
AFLP AFP ACS AHA AlP AICC ALL AML APKD APUD ARDS ASD ASH ASNR ASSR ATAI ATS AUA AUSPV AV
acute fatty liver of pregnancy alpha-fetoprotein adrenogenital syndrome American Heart Association acute interstitial pneumonia American Joint Committee on Cancer acute lymphocytic leukemia acute myeloic leukemia acquired polycystic kidney disease amine precursor uptake and decarboxylation (cell line) adult respiratory distress syndrome atrial septal defect asymmetric septal hypertrophy American Society ofNeuroradiology American Society of Spine Radiology acute traumatic aortic injury American Thoracic Society American Urological Association anomalou~ unilateral single pulmonary vein atrioventricular
AVM BAL BALT BlP BMT BO BOOP BPH bpm BS CAD CBV CCC CCD CHF CLL CM CML CMN CMP CMV COLD COP COPD CPR CRP CS CSF CT CTA CTAP CTC CTm CTEPH CTHA CTP
arteriovenous malformation bronchoalveolar lavage bronchus-associated lymphoid tissue bronchiolitis obliterans interstitial pneumonia bone marrow transplant bronchiolitis obliterans bronchiolitis obliterans with organizingpneumonia benign prostatic hyperplasia beats per minute buccal space coronary artery disease cerebral blood volume cholangiocellular carcinoma charge couplet device chronic heart failure chronic lymphatic leukemia contrast medium chronic myeloid leukemia contrast medi urn nephropathy cardiomyopathy cytomegalovirus chronic obstructive lung disease [~COPDl ~togenic organizing pneumorna chronic obstructive pulmonary disease curved-planar reformation C-reactive protein carotid space cerebrospinal fluid computed tomography CT angiography CT during arterial portography CT cholangi ography CT dose index chronic thromboembolic pulmonary hypertension CT hepatic arteriography cr perfusion imaging
7 Principles ofCT, Spiral CT, and Multislice CT M. Prokop
2 Scanning
Principle
(p. 2)
Image
Reconstruction
Image
Display and
Scanning
(p .. 3) Documentation
Parameters
(p. 5)
(p. 7)
11
Spiral CT I Helical CT
Multislice CT
.. .. .. .... .. .. .. .. .. .... .. .. .. .. .. .... .. ~....
Scanning
(p. 18)
Principle
Detector Types (p. 20) System
Performance
Image
Reconstruction
Scanning
(p. 24) (p. 24)
Parameters
Workflow,
(p. 32)
Image Review,
Display
and Documentation (p. 36)
Cardiac Multislice CT Prospective
ECG Triggering
Retrospective
(p. 38)
ECG Gating (p. 39)
CT Fluoroscopy Principle Technique Radiation
and Image
. Reconstruction
(p. 42) Safety (p. 42)
(p. 42)
. ~.~~.~~~~~~.
1 Principles ojCT Spiral CT, and Multislice CT Computed tomography (CT) has been one of the biggest breakthroughs in diagnostic radiology. The first clinical CT scanner was developed by Godfrey N. Hounsfield for examinations of the head and was installed in 1971 at Atkinson-Morley's Hospital in Wimbledon, England. The first body CT scanner was installed in 1974, and before the end of the 1970s the basic technical evolution of CT was complete (Table 1.1). Technical details were refined during the 1980s, and CT technology remained on a plateau until the early 1990s, when the advent of spiral (helical) CT scanning sparked a further, rapid evolution leading to improved diagnostic capabilities, 3D imaging techniques, and CT angiography. The latest innovation is the introduction of multislice CT in 1998. This new technology is vastly expanding the performance of CT scanners: it truly transforms CT from a trans axial imaging modality to a 3D technique that yields high quality images in arbitrary planes and forms the basis for an expanding variety of 3D visualization techniques, including virtual endoscopy. In addition, these scanners have the potential to revolutionize cardiac imaging with CT.
Table 1.1 phy
Historical milestones in computed tomogra-
1929
Radon: basic mathematical principles
1963
Cormack: image reconstruction
1971
Hounsfield (EMI Laboratories): technology
1971
Head scanner (EMI Mark I)
1974
Whole-body scanner (ACTA)
1974
Third-generation scanner (Artronix)
1977
Fourth-generation scanner (AS&E scanner)
1979
Nobel Prize awarded to Hounsfield and Cormack
1980s
Technical refinements
1983
Dynamic spatial reconstructor
1983
Electron beam CT scanning
1987
Scanners with a continuously rotating tube
1989
Spiral CT (helical CT)
1991
Dual-slice spiral CT (Elscint)
1991
CTangiography
1995
Real-time reconstruction (CT fluoroscopy)
1998
Multislice CT (4 detector rows _~~~
1999
Multislice cardiac imaging
2001 /2
Multislice CT (8/16 detec~to"l'!u r--ow-~"tr
Future
Cone beam CT (a 128 detectorrows)
Computet Tomograpfy Scanning Principle
Computed tomography is an x-ray tomographic technique in which an x-ray beam passes through a thin axial section of the patient from various directions (Fig. 1.1).Parallel collimation is used to shape the x-ray beam to a thin fan, which defines the thickness of the scan plane. Detectors measure the intensity of the attenuated radiation as it emerges from the body. A mathematical image reconstruction (inverse Radon transformation) calculates the local attenuation at each point within the CT section. These local attenuation coefficients are translated into "CT numbers" and are finally converted into shades of gray that are displayed as an image. With conventional CT scanners the volume of interest is scanned in a sequential fashion, usually proceeding one section at a time. The first two generations of CT scanners (Table 1.2) were superseded in the late 1970s by third- and fourth-generation scanners, which are still in use today. In third -generation scan-
Fig. 1.1
Principle of CT scanning.
Image Reconstruction Table 1.2
Comparison of CT scanner generations
Type
First generation
and types
Second generation
Third generation
Translation-rotation
Principle
Single
Series
Arc (30°_60°)
Active detector rows
2
1-2
1-16
3-52
256-1000 0.4-10 s
Scan time
135-300 s
5-150 s
ners, tube and detector array rotate synchronously around the patient. The detector array covers the full width of the fan beam (Fig.1.2). In fourth-generation scanners, the detector elements cover a full circle around the scanner opening and remain stationary during the scan, while only the x-ray tube rotates around the patient (Fig. 1.2). Third-generation scanners, however, offer better scatter suppression and require less detector elements, which is the reason why all multislice CT scanners use third-generation technology. Attempts to speed up the imaging process led to the development of a multi-tube CT scanner called the dynamic spatial reconstructor (the "Mayo monster" equipped with 28 tubes, able to scan up to 240 sections, each of 1 mm thickness, in one 360° rotation), electron beam CTscanning, spiral CT, and, recently, multislice CT. Of these procedures, only spiral and multislice CT have achieved large-scale clinical impact.
Image Reconstruction The detector signals registered during a scan are preprocessed to compensate for inhomogeneities in the detector system and to correct for beam-hardening effects within the patient. After various correction steps and transformation from signal intensities into x-ray attenuation values these data are called the CT raw data (Fig.1.3).The raw data sets for third- and fourthgeneration scanners each consist of the attenuation profiles of some 500 to 1500 projections for each 360° rotation of the x-ray tube. Each projection in turn is composed of some 500 to 1500 single attenuation values. Image reconstruction from the raw data sets finally yields the image data set.
Electron beam scanning
Rotation
Detectors
Detector elements/row
Fourth generation
Electron beam deflection
Ring (360°)
Semicircular
(210°)
4 600-4000
432/864
1-5 s
~50 ms
Scanning principle
Image reconstruction
3rd Generation
Fig.1.2 Comparison of scanning principle and image reconstruction in third-generation (a, b) and fourthgeneration CT scanners (c, d). The fourth generation uses a stationary detector ring, and the data acquired by one detector are assembled into a projection for the various tube positions.
Image reconstruction starts with the selection ofthe desired field of view. Each ray from the tube to the detector that passes through this field of view is used for reconstruction. The attenuation coefficient for each image point is determined by averaging the attenuation values for all rays that cross this point (back projection). But this type of unfiltered back projection yields a very unsharp image with blurred edges, so multi-
3
I
1 Principles oICT, Spiral CT, and Multislice CT
Scan Data
-+
<:> Fig. 1.3
Id
Prepro-
Convolutionj -+ filtering
cessing
~
,...~~ , •. ~J
Imaae data Back projection
~~\~J.j
Processes involved in CT image reconstruction.
pie rays are assembled into a projection, and the resulting attenuation profile is subjected to an edge-enhancing mathematical filtering (convolution) process. The so-called "convolution kernel" determines the type offiltering. Back projection of the filtered attenuation profiles then yields a sharp image. The convolution kernel (reconstruction algorithm) used for the/itered back projection determines the properties of the reconstructed CT sections in terms of spatial reso-
a
Filtered raw data -+
b
Fig. 1.4 Effect of the convolution kernel on spatial resolution and image noise. In the lung parenchyma (high intrinsic contrast), the use of a sharp kernel (b) instead of a standard kernel (a) increases image sharpness. In
lution and image noise, and can vary from soft or smooth to sharp or edge-enhancing (Fig. 1.4). Third and fourth-generation scanners differ in the way that the attenuation values are assembled into a fan-shaped projection. The third generation uses the fan from a single tube position to the detector, while the fourth generation takes the data acquired by a single detector for various tube positions and assembles them into a projection (Fig. 1.2).
c
d
the liver (low intrinsic contrast), a soft kernel (c) is preferable to a sharp kernel (d) because the increased noise could otherwise obscure low-contrast structures.
Image Display and Documentation Fig. 1.5 The pixels in the CT matrix actually represent volume elements (voxels) in the scanned body region. Note that for a standard section thickness of 10 mm, each voxel has a matchstick shape.
I 0.5mm:10mm
I
FOV=25 em
Matrix= 5122
Image Display and Documentation •
Image Matrix and Field of View
ACT image is composed ofa square image matrix that ranges in size from 256 x 256 to 1024 x 1024 picture elements, or pixels. Since a CT section has a finite thickness, each pixel actually represents a small volume element. or voxel. The size of this voxel depends on the matrix size, the selected field of view (FOV) , and the section thickness (Fig. 1.5).In most CT examinations the voxel has a matchstick shape, i.e., the pixel size measured in the plane of the section, the x-y plane, is 10 to 20 times smaller than the section thickness, measured along the z-axis. This anisotropy (nonuniform shape) of the voxels can be decreased only by greatly reducing the section thickness. Only with multislice CT is it possible to obtain nearly isotropic (cube-shaped) voxels for larger body areas. (See also p. 46 for a more complete discussion of the issue.) There may be a difference between the actual image matrix that is reconstructed from the raw data and the display matrix that is shown on the viewing monitor or is printed out to film. Although the display matrix is usually identical to the image matrix, a larger display matrix may be selected (e.g., 1024 x 1024 instead of 512 x 512) to improve the image quality. It is not usually necessary to use data from the entire body cross section for image reconstruction, and afield of view (FOV) of more restricted size can be reconstructed from the raw data. This field of view is characterized either by its size in mm or by the zoomfactorrelative to the maximum field of view possible on that scanner. Depending on the scanner it can either be round or square. Some manufacturers call this the reconstruction field of view (RFOV) to distinguish it from the display field of view (DFOV) that can be
selected from that field and magnified for display on the monitor. Usually such a magnified image is less sharp than one directly reconstructed from the raw data because it uses only a portion of the image data rather than all the information contained in the raw data set. The reconstruction and display field of view have to be distinguished from the scan field of view (SFOV) available on some scanners. The SFOV is a reduced, centrally located area from which data are acquired, which can increase the sampling rate and thus improve spatial resolution. Usually this technique is applied in examinations of the extremities (calcaneus). spine or in the head and neck region. Since a reduced SFOV uses a fan beam with a narrower angle, the technique also reduces radiation exposure to those portions of the patient that are outside the SFOV. For this reason, it is used on some scanners to reduce radiation exposure for cardiac multislice CT imaging. H
CT Numbers
During image reconstruction, a numerical value (CT number) is assigned to each voxel according to the degree of x-ray attenuation in that voxel. To reduce dependence on the energy of the radiation and to obtain numerical values of convenient size, the CT number is defined as follows: CT=
1000
X
(u-U.water)/U-water
The unit for CT attenuation is called the Hounsfield unit (HU). The numbers are set on a scale in which -1000 represents the attenuation of air and 0 is the attenuation of water. Note that there is no upper limit to the scale. The available range of CT numbers varies between scanners and available bits per pixel (e.g., from -1024 to 3071 HU with 12 bits or up to 64,500 HU for 16 bits) .
5
pending on viewing conditions). Consequently, there is no point in assigning the complete diagnostic range of CT numbers (some 4000 HUl to the available range of gray levels (from white to black) because discrimination between structures with small differences in CT numbers would no longer be possible. It is therefore better to display just a portion of the CT scale. This so-called window is defined by its width, which affects image contrast. and by its level (center). which determines image brightness. Reducing the window width increases image contrast: lowering the window level brightens the image, and raising the window level darkens the image (Fig. 1.7). Examples of window settings shown in Table 1.3.
3000
2000 Cortical bone 1000 Cancellous bone Soft tissues Water
'ot ~Lung .1000~Air
•
Fig.lo6 The scale of CT attenuation numbers is defined by the values for air (-1000HU) and water (0 HU). Soft tissues occupy a narrow range around 50 HU.
The scale of IT attenuation numbers is shown graphically in Fig. 1.6. Except for fat, lowprotein fluids, and fresh blood, there are no "typical" values that would permit the specific characterization of soft [issues by their CT numbers (see also Fig. 7.5). •
Window
Settings
The human eye can distinguish only a limited number of gray levels (from about 40 to 100, de-
Image
Processing
and Analysis
The software used in CT scanners provides various options for processing and manipulating the cr sections. The features of greatest practical importance are the measurement of lengths and angles and the analysis of CT numbers within a selected region of interest (ROI). This ROJ may be selected interactively, allowing the operator to choose between predefined shapes (circle, ellipse, rectangle) or freely outlined regions of arbitrary shape. A computer program can then calculate mean CT numbers and standard deviations within the ROJ and can generate histograms. The scanned volume (an also be manipulated to produce reformatted images in any secondary plane of section ("multi planar reforms-
:::~:-I~j,
-1000 10
0
",~ •••••••••••• 1500/-650
-1000 . 1300 Fig. 1.7 Window settings. For optimum contrast, the gray levels available for vlewinq or filming are assigned to a designated portion of the CT scale. The window is
lung wIndow
.
Bone window
defined by its width, which determines contrast. and by itslevel,whlchdetermlnesbrightness.
Scanning Parameters Table13
Window
lung Emphysema
Scanning Parameters
settings Width
level center
1500
-650 -800
800
Soft tissues,
noncontrest uver.ncrcontrast Softtis5ues+CM Uver+CM Neck+CM
a Angiography
400 200 400 300 300
40 40 70 60-100 50
500
Bone
Gantry Angulation The gantry, or frame that houses the scanning equipment, can be tilted about the x-axis to perform oblique scans. The range of gantry angulation may be ::: 30°, depending on the scanner. Gantry angulation is used mainly for scanning the head, neck. or spine. It is unnecessary for most other applications and is increasingly being abandoned with multislice scanning and reconstruction of tilted sections from a volume dataset.
Section Thickness • tion." MPR) and various types of three-dimensional reconstruction (see Chapter 2, Image Processing). •
CT
Documentation
images from conventional scanners are usually printed on film for image review. It is importent to document all the window settings that are relevant for the examined body region. For example, chest scans should be documented in both the lung window and soft-tissue window settings. A bone window setting may be added to screen for skeletal metastases. Attempts to display two non-overlapping window settings simultaneously (e.g., mediastinum and lung) have been abandoned, because structures whose CT numbers lie between the two windows (e.g.. pleural lesions) are too easily missed. This dual window technique is no longer used. With modern spiral CT or multislice scanners, or with the introduction of picture archiving and communication systems (PACS), printing of CT images is increasingly abandoned in favor of direct viewing on CRT or flat screen monitors and storing image data in a digital archive.
Section
Profile
The beam collimation setting determines the thickness of a CT section. But as in conventional radiography. the x-ray tube emits a conically divergent beam rather than parallel rays. To obtain sections of reasonably uniform thickness, it is necessary to use collimators located directly behind the tube. Some scanners use additional post-patient collimators in front of the detectors for further optimization of the section profile. Despite these measures, a truly planeparallel section is never obtained in practice, and adjacent portions of the scanned object are always included. In addition, the finite size of the focus of the x-ray tube causes an area outside the primary beam to receive radiation of less intensity. This region is called the penumbra (Fig.l.8a). The effect on the acquired section can be described in terms of the section sensitiviry profile (SSP) or slice profile {Fig. 1.8), which shows how much a point in the object contributes to the image as a function of its distance from the center of the section. The ideal section profile is a rectangle whose width equals the desired section thickness, so that all points outside the section contribute nothing to the measured attenuation while all points within the section contribute uniformly to the CT number. Real section profiles have rounded "edges". meaning that adjacent regions also contribute slightly to the image. While the profile of thick IT sections (7-lOmm) closely approximates to an ideal rectangular shape, thin sections exhibit a more bell-shaped profile (Fig. 18b).
7
•
8
1 Principles
of CT, Spirol CT, and Multislice
IT
Sensitivity profiles Focal spot
lOmm 100%
50 %
~~
5mm
3mm
slice
2mm slice
t mm slice
'~'
I-+--+-+------f--H-H-\---I
---++--tt-
O%4.~~~-4~~~~~--~~~~ b
FWHM
FWTA
-4.0
4.0 8.0 Table position (mrn)
Hq. U:I Because of the x-ray beam geometry, regions located outside the selected section thickness are ineluded in the scan (a). This results in a rounded section sensitivity profile, which approximates to an ideal rectilinear shape only in thicker sections (b). The effective
section thickness (* section width SWj can be measured by the width of the section profile at half of its peak value, called the full width at half maximum (FWHM), or the width that encompasses 90% of the area under the curve, called the full width at tenth area (FWTA)(c).
In clinical practice, the increasingly large "tails" of thinner section profiles do not substantially affect image quality because the decreased width of the profile is more important and leads to improved resolution along the z-axis.
A more rigorous measure of the width of the section profile is the width that encompasses 90% of the area under the curve, called the full width at tenth area (FWTA). The FWTA indicates the width at which object elements located outside the section contribute just 10% to the cr number. FWHM and FWTA yield similar values in conventional cr when thick sections are chosen, but they differ markedly in spiral cr and when thin collimation is used in conventional cr. The slice profile quality index (SPQI) is another, more recent measure. It describes the percentage of the area under the section sensitivity profile that is bounded within an ideal rectangular profile of the same section width.
•
Effective
Section
Thickness
The width of the section sensitivity profile is usually quantified by measuring the width of the curve at 50% of its peak value, called thefull width at haJfmaximum (FWHM) (Fig.1Sc). This value is also known as the effective section thickness or section width (SW). It is by definition equal to the section collimation (SC) or nominal section thickness in conventional CT. It should be noted, however, that the effective section thickness is not equivalent to the section collimation in spiral cr. It is the commonest measure for describing the spatial resolution along the patient axis (a-axis),
Scanning Parameters 9 Partial Volume
Effect
The CT number of a pixel is determined by the x-ray attenuation that occurs in the corresponding voxel. If tissues with different attenuation properties occupy the same voxel (e.g., blood vessels and lung), the resulting CT number willwith good approximation-represent the sum of the different attenuation values (partial volume
effect or partial volume averaging):
tal.
C T =viXC Ti + V2 XC T 2 + ... , where the partial volume elements vi add
Given the much greater extent of a voxel along the z-axis than in the x-y plane, the section collimation contributes a great deal more to this partial volume effect than the field of view or pixel size (Fig. 1.9). • Section Collimation A number of anatomic structures (portions of the aorta, chest wall, or liver borders) are oriented parallel to the long axis of the body. The usual transaxial section orientation in CT means that the scan will cut the corresponding tissue boundaries at a perpendicular angle, which tends to minimize partial volume effects. For this reason, a section collimation of 7-10 mm is most commonly used in conventional body CT examinations. Partial volume effects are particularly troublesome in scans that are oblique or parallel in relation to tissue boundaries (diaphragm, apex
of lung, pole of kidney) and in the evaluation of small structures (small vessels, bronchi, adrenals). A collimation of3-5 mm can be used to evaluate structures that are parallel to the scan plane (e.g., the pancreas) or for evaluating small organs such as the adrenals. Thin collimation of 1 -2 mm is preferred for the lung, where a detailed structural analysis is required for the diagnosis of interstitial lung disease. With multislice CT scanning thin collimations have become standard (see below). Table Feed
In conventional CT, a tissue volume is scanned section-by-section. This is accomplished by moving the patient table by a designated amount, called the table feed (table increment), between consecutive scans. Contiguous imaging is generally performed, meaning that the section thickness and table feed are equal. Overlapping Scans
Reducing the table feed produces an overlapping scan pattern that increases the radiation dose to the patient. While overlapping scans have been recommended to improve the 3D imaging of skeletal structures with conventional CT, the introduction of spiral scanning has made them obsolete. Discontinuous
Respiratory
Fig. 1.9 Partial volume effect. Because of the matchstick-like configuration of the voxels, the CT number of a voxel is affected not only by the object of interest (e.g., a round lesion) but also by adjacent structures (e.g., lung parenchyma). This creates an averaging effect that distorts the CT number of the voxel.
Scanning
When the table feed is increased, intersection gaps are produced. This may be useful in selected cases where it is necessary only to search for gross pathologic changes that extend over larger ranges. Misregistration
Respiratory misregistration in CT occurs when structures that move with respiratory excursions are missed due to variations in the depth of respiration between sections. However large the efforts to reproduce a consistent depth of breath holding for 5 to 20 respiratory cycles, some gaps along the z-axis will generally occur. The thinner the collimation and the smaller the lesion, the greater the risk of respiratory misregistration will be (Fig. 1.10).This has only a moderate effect on the detection of hepatic lesions or the evaluation of the kidneys and adrenals, but it can seriously hamper the search for pul-
1 Principles ofCT, Spiral CT, and Multislice CT
.~ 15l· Fig. 1.10 Respiratory misregistration: different depths of breath holding can cause a small lesion (e.g., a pulmonary nodule) to be missed in sequential scans.
monary metastases. To obtain seamless coverage, a section collimation of less than 5mm should be avoided in areas that move with respiratory excursions. This creates conflicting demands in conventional Cl, where it is not possible both to minimize partial volume effects and avoid respiratory misregistration in the same scan. Reconstruction Algorithm (Convolution Kernel) The convolution kernel used in reconstructing an image from the raw data determines the relationship between spatial resolution and image noise. Noise limits contrast resolution, and thus the ability to differentiate objects that show very little attenuation difference from their surroundings. High contrast resolution is important for the detection of lesions in parenchymal organs like the liver and pancreas. High spatial resolution is necessary for the detection of very fine morphologic changes in lung or bone. Highresolution convolution kernels [HR kernels, sharp kernels) improve spatial resolution but disproportionately increase noise as well. Conversely, soft or smooth kernels lead to a concomi-
Image noise Fig. 1.11 The use of hiqher-resolution convolution kernels can improve spatial resolution but leads to a disproportionate increase in image noise.
tant reduction in noise and spatial resolution (Fig. 1.11). Standard kernels are designed as a compromise for good spatial resolution and reasonably low image noise for most body applications. Partial Scan Because attenuation of an x-ray beam is identical in both directions (from the tube to the detector and from the detector to the tube), enough data for reconstructing a cr image can be obtained from less than a full 360° rotation. Instead, data from a partial rotation of 180° plus the fan angle of the x-ray beam (usually some 60°) is sufficient for this purpose. This partial scan or half scan reconstruction can be used to reduce the scan time per CT section and thus reduce motion artifacts. In particular, it is employed for cardiac imaging using electron beam CT or multislice Cl, when temporal resolution is of utmost importance.
VIII
Abbreviations CTSI CUP CVS CVS CWP DAD DCF DEQCT DES DIP DLDCF DLP DORV DPB DRE DSA D-TGA DVT EAA EAP EBCT EBT ECA ECG ECST EHE ELCAP EMLA ERC ERCP ESR FBI FBSS FDG FIGO FLC FNAC FNH FOV FSH FWHM FWTA Gd-MRA GFR
CT severily index carcinoma of unknown primary calcium volume score continuous volume scanning (ECBn coal worker's pneumoconiosis diffuse alveolar damage deep cervical fascia dual-energy QCT diethylstilbestrol desquamative interstitial pneumania deep layer of the deep cervical fascia dose - length product double outlet right ventricle diffuse panbronchiolitis digital rectal examination digital subtraction angiography complete transposition of the great arteries deep venous thrombosis extrinsic allergic alveolitis early arterial phase electron beam CT electron beam tomography external carotid artery electrocardiogram European Carotid Surgery Trial epithelioid hemangioendothelioma Early Lung Cancer Action Project eutectic mixture oflocal anesthesia endoscopic retrograde cholangiography endoscopic retrograde cholangiopancreatography erythrocyte sedimentation rate fat - blood interface failed back surgery syndrome f1uorodeoxyglucose Federation Internationale de Gynecologie et d"Obstetrique fibrolamellar carcinoma fine needle aspiration cytology focal nodular hyperplasia field of view follicle-stimulating hormone full width at half maximum full width at tenth area Gadolinium-enhanced MRA glomerular filtration rate
GI GIP GIST GSD GTD HAE HAP HBV HCA HCC HD HDL HIV HL HMG-CoA HNP Ho:YAG HOCM HRCT HSV HVP ICA ICFT ICU IGCCCG IHE IHSS
gastrointestinal giant ceU interstitial pneumonia gastrointestinal stromal tumor glycogen storage diseases ~estational trophoblastic disease epatic alveolar echinococcosis hepatic arterial phase hepatitis B virus hepatocellular adenoma hepatocellular carcinoma Hodgkin's disease high density lipoproteins human immunodeficiency virus Hodgkins lymphoma hydroxymethylglutaryl coenzyme A
herniated nucleus pulposus ho lmiu m:yttriu m:alurninu rn:ga rnet hypertrophic obstructive cardiomyopathy high-resolution thin-slice CT herpes simplex virus hepatic venous phase internal carotid artery intracavitary fibrinolysis
intensive
care unit
International Germ Cell Cancer Collaborative Croup infantile hemangioendothelioma idiopathic hypertrophic subaortic stenosis
IIAC ILP IMA INR IPF IPH
idiopathic infantile arterial calcification interstitial laser photocoagulation inferior mesenteric artery international normal ratio idiopathic pulmonary fibrosis idiopathic pulmonary hernosiderosis
IPMT IV IVC IVF IVP KUB LA LAA LAD
intraductal papillary mucinous tumor
intravenous inferior vena cava
in vitro fertilization intravenous pyelogram kidney abdomen pelvis (plain abdominal radiograph) left atrium left atrial appendage left anterior descending coronary artery
Scanning Principle Spiral (Helical) CT
During the past decade, spiral (= helical) CT has become the standard technology for the maj ority of clinical indications for CT scanning.
(see Fig. 1.15). The higher the pitch, the lower the radiation dose to the patient and the greater the available range of scan coverage. Advantages
Scanning Principle
Spiral CT requires a scanner with a continuously rotating x-ray tube. A tube with a large heat capacity is needed that can operate continuously for the duration of the scan. More than 100 seconds scan duration are available with current scanners. In contrast to conventional CT, the patient is not scanned section-by-section but is translated through the scan plane at a uniform table speed during acquisition of the raw data (Fig. 1.12). The technique is named spiral or helical CT after the spiral or helical pattern that is traced by the scan. A CT image can be generated from any segment within the scanned volume, so the table feed is unrelated to the site of image reconstruction. Sectional images can be produced at arbitrary levels, and individual images can be overlapped as desired without increasing radiation exposure. The spacing between the reconstructed sections is called the reconstruction interval, increment or index. The section collimation (nominal section thickness) and table feed can be varied independently of each other in spiral CT. The pitch defines the ratio of the table feed (table increment) per gantry rotation to the section collimation
Position of
The advantages of spiral CT arise from continuous data acquisition and short total scanning time. With conventional CT, small lesions such as pulmonary or hepatic metastases may be missed due to respiratory misregistration (see Fig. 1.10). But spiral CT can acquire data in a seamless volume during one breath hold, eliminating respiratory motion and interscan gaps. The use of overlapping reconstruction intervals allows for optimum visualization of small lesions while eliminating partial volume effects. When thin collimation is used, 2 0 reformatted images can be generated in arbitrary planes of section, and 3D reconstructions of good quality can be produced. Owing to the short scan time, most spiral examinations can be performed during a single breath hold. Intravascular contrast medium can be used more effectively, providing either higher contrast or a reduction in the volume of contrast material. Arterial phase imaging has become available only with spiral scanning. These advantages can Significantly improve the detection of hepatic and pancreatic lesions compared with conventional scanning. High vascular contrast is the foundation for CT angiography (CTA), which is a technique that is not available with conventional CT. Volume acquisition and short scan times are utilized in this technique to capture the arterial enhancement and generate angiogram-like vascular images (see Chapter 24).
tube focus Disadvantages
Fig. 1,12
Principle of spiral CT.
Most disadvantages of spiral CT result from the use of older scanners. In these scanners the continuous tube operation necessitates a lower radiation dose per rotation and may lead to a marked increase in image noise. Modern scanners employ an improved tube technology, so noise is no longer a limiting factor. While the scan duration is shorter with spiral CT, there are more sections to be processed. This may increase the time required for image reconstruction, especially when overlapping
11
Abbreviations LAM LAO LAP LCA LCX LDH LDL LH LHA LHV LI LIMA LIP LITI LOCM LPV L-TGA LV MAC MALT MEN MFH MHV MIBC MinIP MIP mIP MLDCF MLI mLV Mn-DPDP MPNST MPR MRA MRCP MRI mRV MS MSAD MTF MTT NASCET NASH NASS Nd:YAG
lymphangioleiomyomatosis left anterior oblique late arterial phase left coronary artery left circumflex [coronaryartery[ lactate dehydrogenase low density lipoproteins luteinizinghormone left hepatic artery left hepatic vein linear interpolation left internal mammary artery lymphocytic interstitial pneumania
laser-induced thermotherapy low-osmolar contrast media left portal vein corrected transposition of the great arteries left ventricle Mycobacterium avium complex mucosa-associated lymphoid tissue
multiple endocrine neoplasia mal~nant fibrous histiocytoma mid e hepatic vein meta-iodobenzylguanidine minimum intensity projection maximum intensity projection minimum intensity projection middle layer of the deep cervical fascia multislice linear interpolation morphologic left ventricle mangafodipir-trisodium malignant peripheral nerve sheath tumor multiplanar reformation magnetic resonance angiography magnetic resonance cholangiopancreatography magnetic resonance imaging morphologic right ventricle masticator space multiple slice average dose modulation transfer function mean transit time
North American Symptomatic Endarterectomy Trial nonalcoholic steatohepatitis North American Spine Society neodymium: yttrium-aluminumgarnet
NHL NIPF n.p.o. nr-MIP NSCLC NSIP OLT OM 00 OP OR PA PACS PAI-l PAN PAP PAPVR PCP PCS PDA PDA PDV PE PEEP PEl PET pi PID PIE PLC
non-Hodgkin"s lymphoma nonspecific interstitial pneumania and fibrosis nothing by mouth (nil per as) noise-reduced MIP non-small cell lung cancer nonspecific interstitial pneumania
orthotopic liver transplantation obtuse marginal branch osteoid osteoma organizing pneumonia operating room posteroanterior picture archiving and com-
ill unication
system
plasminogen activator inhibitor-! polyarteritis nodosa pulmonary alveolar proteinosis partial anomalous pulmonary venous return Pneurnocysris carinii pneumonia
posterior cervical space patent ductus arteriosus posterior descending coronary artery pancreaticoduodenal vein pulmonary embolism positive end-expiratory pressure percutaneous ethanol injection positron emission tomography post injection pelvic inflammatory disease pulmonary interstitial emphyserna
pulmonary lymphangitis carcinornatosa
PLDD PMF PMMA PMS PNH PPH PPS PS PSA PT PTA PTC
percutaneous laser disk decom-
pression
progressive massive fibrosis polymethylmethacrylate pharyngeal mucosal space paroxysmal nocturnal hemoglobinuria primary pulmonary hypertension parapharyngeal space parotid space prostate specific antigen prothrombin time percutaenous transluminal angioplasty percutaneous transhepatic cholangiography
IX
1 Principles ofCT, Spiral CT, and Multislice CT sections are used. While this was a time-consuming process with older scanners. new scanners provide fast reconstruction and often require less than one second per image. An increased number of images needs to be viewed. documented and archived. which again may increase time and costs. There is a basic tradeoff in spiral CT between a large scan volume and a high spatial resolution in the z-direction. Short scan ranges. such as the inner ear, can be covered with thin collimation but long ranges, such as in thoracoabdominal CT examinations, require a thicker collimation. The short scan time also complicates the administration of contrast medium, and new types of artifact may occur. A faulty contrast technique can lead to suboptimum examinations or misleading findings (see Figs. 7.39-7.41).
Image •
Reconstruction
Interpolation
Algorithm
and
Section Profile
The table movement during the scan will produce motion artifacts if the raw data acquired during a 360° rotation are used directly for image reconstruction. This is because the first and last projections in the 360° rotation sample different data (due to the table motion during tube rotation). To avoid these artifacts, interpolation of the raw data before image reconstruction is required. The goal of the interpolation is to obtain a complete set (360°) of projections at the desired z-axis position in the scanned volume. The simplest linear interpolation of the projection data is called 360° LJ (Fig. 1.13a). At every angular position of a 360° rotation it interpolates between the two projections in the spiral data set that are closest to the chosen position along the z-axis. This interpolation from 720° of data results in a complete (360°) set of projections for the chosen z-position. The 360° Ll interpolation provides the least image noise but substantially broadens the section profile (Fig.1.l3b). More advanced interpolation algorithms exploit the fact that x-ray attenuation is independent of direction, i.e., the attenuation along a ray between the tube and detector is equal in both directions. This makes it possible to com-
pute a virtual second spiral (conjugated data) for the attenuation values along a ray from the detector to the tube, and to interpolate the projections at corresponding angles between the real and virtual spiral. This algorithm is called 180° LI (Fig. 1.13a) but actually uses data from 360° plus the fan angle of the x-ray beam. The resulting section profile is substantially narrower (Fig.U3b) because the distances between corresponding projections in the real and virtual spirals are less than between corresponding projections in the real spiral alone. Differences between the 360° and 180° interpolations are best appreciated in multiplanar reformations and are most pronounced when pitch factors> 1 are used (Fig. 1.14). The 180° Ll algorithm, however, results in a larger image noise since only half the data are used for interpolation as compared to 360° LI. In fact, the noise with 180° Ll is as high as it would be with a 360° Ll and half the exposure dose. Higher-order interpolation algorithms use not only two points from adjacent (real or virtual) spirals but instead, apply a more complex weighting function (longitudinal filtration, z-filtering) to the spiral projection raw data. This zfilter function defines how much each projection contributes to the final image depending on its distance to the reconstructed section. Such algorithms can be optimized to obtain more rectangular section profiles at the cost of more image noise (e.g., 180° HI). or they can reduce noise (and thus dose requirements) at the expense of a slightly broadened section profile (e.g., HRLF -10, SmartHelical, GE). Effective Section Thickness (Section Width)
While the width of the section profile in conventional CT is equal to the section collimation (nominal section thickness), the bell-shaped section profile in spiral scanning must be described in terms of the effective section thickness or section width (SW). The effective section thickness depends on the section collimation and a number of other factors that include the table feed and interpolation algorithm. The most frequently used measure for the effective section thickness is the full width at half maximum (FWHM), i.e., the width of the profile at one-half ofits peak value. For a pitch of 1 the section width, given as FWHM, is identical to the section collimation if
X Abbreviations PTCA PTLD FIT PV PVOD PVP PVP PVS QCT RA RAO RAS RBF RB-ILD RBV RCA RCC REAL RF RHA RHV RIMA RLD RN ROI RPS RPV RSV RV RVOT SAPHO SBO SC SCCa SCF SCLC SEQCT SFOV SI SLAP SLDCF SLE SLS SMA SMS SMV
percutaenous transluminal coronaryangioplasty post-transplantation lymphoproliferative disorder partial thromboplastin time portal vein puhnonary veno-occlusive disease portal venous phase percutaneous vertebroplasty prevertebral space quantitativeCT right atrium right anterior oblique renal artery stenosis regional blood flow respiratory bronchiolitis associated interstitial lung disease regional blood volume righ t coronary artery renal cell carcinoma Revised European-American Lymphoma (classification) radiofrequency right hepatic artery right hepatic vein right internal mammary artery right lateral decubitus regenerative nodule region of interest retropharyngeal space right portal vein respiratory syncytial virus right ventricle right ventricular outflow tract synovis tis-acn e-pustulosis- hyperostosis-osteitis (syndrome) small bowel obstruction slice collimation squamous cell carcinoma superficial cervical fascia small cell lung cancer single-energy QCT scan field of view sacroiliac
superior labral anterior - posterior (tear) superficial layer of the deep cervical fascia systemic lupus erythematosus sublingual space superior mesenteric
artery
superior
vein
submandibular space mesenteric
SPECT SPIO SSD SSP SIT SVC SVS TACE T-ALL TBC TEE TF TGA THAD THPE TIA TIPS TNB TNM TNMS
single photon emission computed tomography superparamagnatic iron oxide (particles) shaded surface displays section sensitivity profile scaphoid - trapezium - trapezoideum superior vena cava
step volume scanning (EBCT) trans catheter arterial chernoembolization T-cell acute lymphoblastic leukemia tuberculosis transesophageal echocardiography table feed transposition of the great arteries transient hepatic attenuation differences transient hepatic parenchymal enhancement transient ischemia attack transjugular intrahepatic portosystemic shunt transthoracic needle biopsy tumor, node, metastasis (staging classification tumor, node, metastasis,
serum
(staging classification) tubo-ovarian abscess TOA TRUS transrectal ultrasound thyroid stimulating hormone TSH time to peak TIP transurethral resection TUR ultrafastCT UFCT Union Internationale Centre Ie UICC Cancer DIP usual interstitial pneumonia UP) ureteropelvic junction US ultrasound ultrasmall superparamagnatic USPIO iron oxide (particles) V/Q, scanning ventilation-perfusion scintigraphy VATS video-assisted thoracic surgery VOl volume of interest VRT volume rendering technique VS visceral space VSD ventricular septal defect WHO World Health Organization ZEEP zero end-expiratory pressure 13-hCG P-human chorionic gonadotropin
Image Reconstruction
13
Image plane
.r ,
"
,, ,, ,
\'
' '
a
o
4
2 3 Z-position/collimation
a
360' L1:Linear interpolation between projectional data atzundz:» TF 180' L1:Linear interpolation between conjugated data at z" and projectional data atz' + TF
b Fig. 1.14 Quality comparison of coronal reformatted images using 360' LI (a) and 180' LI (b). The examination was performed with 3mm collimation, 6mm table feed, and a 2mm reconstruction increment.
measured spiral data set calculated conjugated data set Image plane
a 180 LI interpolation is used (see Fig.1.l3b). With a 360 LI interpolation, the section width is 28% larger. The same 28% larger section width is obtained if a 180 LI interpolation with a pitch of 2 is used (see Fig. 1.15b). 0
0
270
0
Image Noise
Image interpolation always includes a projection whose position corresponds precisely to the center of the section (Fig. 1.13a). This projection is not interpolated in the algorithms described, whereas its 180 counterpart undergoes maximum interpolation. This leads to a position-dependent discrepancy of spatial resolution and noise in the scan plane, as image areas near the tube show slightly better sharpness due to the lack of interpolation while opposite image areas show a reduction in noise (see Fig. 7.48). Newer interpolation algorithms (e.g., 180 adaptive interpolation) can correct for these discrepancies.
90
0
z"
3
4
Z-position/collimation
b % 100
"
- -,
, , ,
80 60
1800 LI 3600 LI
r'
u:
" "" ",
"
;, "
,.
_ 1__
"
" " " ,r , ,'' , , , ,
40 20
Pitch
p";l
0
, , , "
r
-5.0
, "" "
" ", , , ,
, , 0 -10
Collimation SC=Smm
,, ,'' , ,'
Standard CT ~ ,
.A
0
~5.010.0 Table position (mm)
Fig. 1.13 Principle of raw data interpolation, (a) Conventional diagram, (b) angluar diagram, (c) comparison of section profiles for 360' and 180' linear interpolation (L1).
XI Contents
Principles of cr, Spiral cr, and Multislice cr Image Processing and Display Techniques
p.45
Patient Preparation
p.83
Optimization
and Contrast Media
of Scanning Technique
p. 131
cr-Guided
p. 161
Interventional
Procedures
Image Analysis
p.187
Neck
p.223
Mediastinum,
system
Pleura, and Chest Wall
p.279 p.373
The Liver
p.405
The Biliary Tract
p.477
The Spleen
p.497
The Pancreas
p.513
The Gastrointestinal
Tract
p.541
Peritoneal Cavity and Retroperitoneum
p.595
Adrenal Glands
p.625
The Kidneys
p.639
Urinary Bladder
p.683
Female Pelvis
p.699
Male Pelvis
p. 725
Lymphatic System
p. 741
Heart
p. 759
Vascular System
p.825
Musculosceletal References Index
'.
p.109
Radiation Dose and Image Quality
Lungs and Tracheobronchial
4'"
p.1
System
p.929 p.998 p. 1052
1 Principles ofCT, Spiral CT, and Muitislice CT As mentioned above, a comparison of 360 LI and 180 LI interpolation shows that 360 LI yields a 28% larger section width (less spatial resolution along the z-axis) and a 29% reduced image noise, Variants such as SmartHelical (GE) lead to a 10% wider section profile than 180 LI but only suffer from 8-16% less noise (depending on the pitch), 0
Acquisition
Section Collimation
0
0
Scanning Parameters
The variable scan parameters in spiral CT examinations are reviewed in Table L4. Three basic parameters for spiral CT scanning are userselectable, In most scanners these are section collimation (SC), table feed per rotation (TF), and reconstruction interval (RI), In some scanners the table feed TF is substituted by the pitch factor P. All the other parameters are varied only in exceptional cases, For this reason, the basic scanning parameters provide an excellent idea of how a spiral CT scan was performed, In this book, we use a triplet of numbers {SCITFlRlj to describe these basic parameters, Section collimation, table feed and pitch are the most important acquisition parameters, while the reconstruction increment is the most important parameter for image reconstruction.
Table
1.4
Scan
parameters
Acquisition
parameters
SC
=
collimation
TF
=·Table
P P*
81.
=
Slice
Scan
RI
=
Reconstruction
width
Table Feed and Pitch
nominal
section
thickness
=N X PX SC=P' X SC = TF /[N x SC] = beam pitch = TF /S C = N X P slicepitch 'TI XTS= TI x N X PX SCIRT e
(mm) increment
(mm)
=
effective
=
reconstruction
section
thickness interval
parameters
=
Table
=
Scan time
"Equipment
•
The table feed per tube rotation (TF), also called the table increment, can be selected independently of the section collimation in spiral scanning, Pitch (P) is defined as the ratio ofthe table feed per tube rotation to the collimation (Fig, US), The following rules should be noted, Scanning at a pitch less than 1 produces an overlapping scan pattern that increases the radiation dose to the patient Overlapping scans may offer slight advantages for the 3D reconstruction of contours that are roughly parallel to the scan plane (e.g, the calvaria), but there are
parameters
Section
Derivative
The section collimation (SC) determines the spatial resolution that is achieved along the zaxis (the direction of table travel), SC can be varied in fixed increments, depending on the scanner. Usually the manufacturer predefines these increments, but with some scanner units they can be modified at installation, The following settings can be recommended as a good compromise between clinical requirements and number of collimation settings: SC=O,Smm, 1 mm, 2mm, 3mm, Smm, 7 mm, or 10mm,
CT
o
(mm)
(em)
=
=
rotation
pitch
length
Reconstruction
RT
and multislice
Pitch
-Volume =
i1 spiral
(mm)
feedltube
SW
N
Parameters
0
speed
(mm/s)
=
(s)
IF/RI
=
scan ciJration
=
duration
=
L1TS
=
(L X RD/(N
X P x S
parameters
Rotation
"Number
time
(s)
of detector
rows
Single
slice:
of one tube N
=
rotation
1; dual slice:
N
=
2; multislice:
Scanning Parameters 15
Pitch=2
Pitch=l
-SC
"
,I r
Pitch= 1 TF=Smm
j
\
/
Pitch = 2 TF= 10 mm
,
\
I I
80
I'
"
"
40
1\
1\ 1\ \
,, ,' I
\
, ,
20
,, ,,
:
o
-S.O
S.O
"
" ,_.• - -, .!I-
_..•.
.
,
I
1\
"
Table feed TF=Smm
'
" "
. - ....•
I,
"
TF
Pitch = 1 ----"--' : " SC=Smm ::
60 1180 LlI----
o~~~~~-,--~~~~-,--10
,
, ,
\'
-l-·-···_-\,
% 00
"\'
f
"
" I
se
Collimation SC=Smm
I
',
rf
60
20
I
••..1 I...•• 14-~
TF
SC
I
40
Pitch=2
--
TF
% 100 80
Pitch-'l
Pitch =2 SC= 2.Smm
J
:
"
"" ,, '' "
, ',
O~-r~/r'~,-,,-,~,"~,--,-10
-10
-S.O
o
S.O
10
Fig. 1.15 Increasing the pitch while keeping the section collimation SC constant has the effect of stretching outthe spiral (a). This widens the section profile (b). The resulting profile with 180 LI at pitch 2 is identical to the section profile with 360 LI at pitch 1 (compare Fig. 1.13 c). Thus, using a pitch of2 instead of 1 will yield
an only 30% wider section width but cover twice the scan length. Increasing the pitch by reducing the section collimation SC covers the same scan length (c). With 180 LI, this leads to a 35% narrower section width than with a pitch of 1 (d).
very few indications in which the minimal improvement of image quality justifies the increased exposure. Moreover, there is little if any effect on image noise, and the higher radiation dose essentially goes unutilized. The only exceptions are scanners with dual detectors (see below, p. 20), which have special z- filtering algorithms that can utilize the higher applied dose to improve image quality (useful in examinations of the intervertebral disks, for example). If the pitch is greater than 2, the volume of interest will be under-sampled, leading inevitably to artifacts. This may be acceptable in some instances, such as trauma cases where priority is placed on the rapid assessment of large volumes rather than on high image quality. But as a general rule, it is better to avoid scanning at pitch values greater than 2. One exception is scanning with very thin collimation, where even conventional CT sections have a bell-shaped profile. Due to broadening of the section profile,
a pitch as high as 3 can be used without causing serious artifacts if the collimation is 1 mm or less (Fig. 1.16).
0
0
0
•
Effective Section Thickness and Pitch
The most Widely used interpolation algorithms at present (180 U) do not increase the section width (effective section thickness) over the section collimation when a pitch of 1 is used (Fig. 1.15 b). Increasing the pitch to 2 reduces the radiation dose by half and doubles the scan length but does not increase the effective section thickness to a proportionate degree. The section profile obtained with 180 LI and a pitch of 2 equals the profile obtained with 360 LI and a pitch of 1, i.e., the scan length is increased and radiation exposure reduced, each by a factor of 2, with identical resolution along the z-axis (Fig.USb). 0
0
0
16
1 Principles oICT, Spiral CT, and Multislice CT reduces the radiation dose but also keeps the effective section thickness as small as possible. A pitch of 2 would be most favorable on theoretical grounds but is more vulnerable to motion and spiral artifacts. A pitch between 1.5 and 2 is the most reasonable tradeoff for most clinical investigations. Rotation Time Fig. 1.16 For narrow collimation, the pitch can be increased to 3 without markedly compromising image quality. Note the excellent quality of multiplanar reformations with a SCITF /RI = 1/3/1 protocol for visualization of subsegmental pulmonary emboli (arrow). Note that the vertebral end plates are well delineated but suffer from minor horizontal interpolation artifacts (arrow head).
Increasing the pitch may produce an increase or decrease in the section width. If the section collimation is kept constant while the table feed is increased, the section width increases (Figs. 1.15a, b). Conversely, the section width is reduced if the section collimation is decreased while the table feed remains unchanged (Figs. LISc, d). In clinical practice, the length L of the scan range and the available scan duration Tl (usually a maximum of 30 seconds for breath-hold scanning) determine the selection of the table feed (see formula in Table 1.4). In this situation it is generally best to use a pitch greater than 1 to maximize the spatial resolution along the z-axis for a given scan length (Table 1.5). This not only
Table 1.5 Effect of pitch on z-axis resolution (effective slice thickness) for a constant table feed
Scan parameters Pitch SC TF
1800 LI
3600 LI
3mm 4mm Smm 6mm
3.9mm 4.6mm 5.3mm 6.0mm
6.9mm
sc=
6mm 6mm 6mm 6mm
2.0 1.5 1.2 1.0
Section width (FWHlvlp
7.2mm 7.5mm 7.8mm
slice collimation, TF = table feed, FWTA full , width of section profile at half maximum Eo
The rotation time (RT) of the x-ray tube (duration of one revolution) is 1 second in most scanners but can vary from 0.42 to 2 seconds, depending on the scanner type. The table speed (TS), or scan length per unit time, can be calculated by dividing the table feed TF by the rotation time (see formula in Table 1.4). The IS is one-third higher in 0.75 second scanners and one-third lower in 1.5second scanners than in standard 1 second scanners. With a shorter rotation time, the z-axis coverage within a given scan time can be increased. All scanning protocols (basic parameters) for spiral CT presented in this book are based on scanners with a rotation period of 1 second. To achieve the same coverage with a slower scanner, the scan time must be increased by the factor RT compared with a 1 second scanner. This is feasible only in regions that do not show respiratory motion (usually skeletal examinations). In all other cases, slower scanners require wider section collimation. Conversely, faster scanners can apply narrower collimation. Because of the detector properties in some scanners, the number of projections available for image reconstruction must be reduced at the shortest rotation time available with these scanners. This leads to a degradation of image quality (loss of sharpness) in the scan plane. If the rotation time is user-selectable (not available in all scanners), it is advantageous to select a longer rotation time for studies that require higher in-plane spatial resolution (e.g., high resolution lung or skeletal examinations). Scan Time
The maximum scan time (scan duration) available in a spiral scanner depends on the x-ray
Image Review, Display and Documentation tube. The higher the dose requirement. the shorter the available scan time will be. Old scanners were limited to a scan time of 24 seconds. but curren t units can achieve scan times up to lOOseconds. In practice, scan times longer than 30 seconds are feasible only for body regions that can be examined without breath holding-the neck, the musculoskeletal system, and in some cases the pelvis. Thoracic and abdominal examinations generally require breath holding to ensure optimum image quality. In some regions (e.g., the lower abdomen), scanning can proceed while the patient slowly exhales after holding his or her breath and continues breathing shallowly. This will cause no substantial loss of quality and will enable scan times longer than 30 seconds.
Reconstruction •
Parameters
Raw Data Interpolation
Most modern scanners use the 180 LI raw data interpolation algorithm, but there are variants available among different units (e.g., SmartHelical on GE scanners, slim2 on Siemens scanners). Fourth-generation scanners (Marconi/Philips) use an interpolation (z-filtering) in which the effective section thickness is largely independent of the pitch (for a constant collimation), but image noise increases with the pitch. This differs from standard interpolations (360 LI. 180 LI), in which image noise is independent of pitch for a given collimation. 0
0
0
Reconstruction Algorithm (Convolution KerneO
The convolution kernels in spiral CT are -no different from those in conventional CT (see above).
The reconstruction increment has nothing to do with section collimation or section width. Rather, it defines the degree of overlap between the axial sections. For the detection of small structures (e.g., pulmonary nodules) it is important to choose Rl so that at least a 30% section overlap is generated. For optimum multiplanar reformations or 3D reconstructions, the reconstruction increment should not exceed half the section width (effective section thickness). This provides a 50% degree of section overlap and generates more than twice as many CT images as conventional scanning. Non-overlapping reconstruction wastes an important advantage of spiral CT, and small lesions located at the boundary between two sections may be missed due to partial volume effects. Overlapping reconstruction ensures optimum lesion display. The theoretic optimum for RI is even smaller than half the section width (see Chapter 4, p. 124), but the added value in clinical practice is generally low.
Image
Review,
cr
Axial
Increment
One of the main advantages of spiral scanning is the continuous acquisition of data, which makes itpossible to generate axial CTsections retrospectivelyat arbitrary z-positions within the scanned volume. The reconstruction increment (RI) defines the spacing of the reconstructed images.
Images
Spiral CT images differ very little in appearance from conventional CT sections obtained with a similar radiation exposure, an identical convolution kernel, and an identical number ofprojections. This is because the raw data interpolation eliminates image artifacts, and the spatial resolution in the axial scan plane corresponds to the spatial resolution of conventional CT images. Compared with conventional CT, however, the image noise for the same scan dose may be 18% lower (360 LJ) or 15% higher (180 LJ), depending on the type of interpolation algorithm used. Spiral CT with a small reconstruction increment generates a large number of images. For example, spiral CTofthe abdomen with 1 s rotation time, a table feed of8 mm, a scan time of 30 seconds, and image reconstruction at 2 mm intervals (as in CT angiography) will generate (30s x 8mm/s/2mm=) 120 images. This large number of images cannot be conveniently documented and reviewed with traditional methods. 0
Reconstruction
Display and
Documentation
0
Principles «ct. Spiral CT, and Muitislice CT •
Cine Mode and Image Documentation
As a rule, not all images can be documented on film One solution is to document every second to fourth image on film, depending on the type of examination, Abnormalities must be adequately documented, however, and a suspicious or abnormal finding must be reproducible, For this reason, selected overlapping sections or multiplanar reformations (MPRs) should also be documented if required, Image quality (noise) can be improved by retrospectively increasing the section width (approximately 7-10mm) for documentation instead of using the original sections, In some scanners, this can be done directly from the raw data by fusing projections from multiple tube rotations (similarly to multislice CT), These thicker sections reduce noise, which is particularly beneficial in scans taken with a low dose or a thin collimation, If this option is not available, thicker sections can be produced by averaging the data from several overlapping images (e.g, using thick axial reformations;' This is usually too time-consuming
for routine
examinations
but may be appropriate in selected cases where image noise is excessive, It is, however, the preferred technique for image documentation in almost all multislice CT applications,
If not all the reconstructed images are printed, all available images should be viewed in an interactive cine display so that an accurate evaluation can be made, This mode displays the individual images in the scanned volume on the monitor in a movie-like sequence, providing a three-dimensional impression of complex, nonlinear structures that have multiple intersections with the image plane, The radiologist can control the speed and direction of the image display sequence interactively by manipulating a computer mouse or trackball, Image Processing
A data set composed of overlapping axial spiral CT images is well suited for generating multiplanar reformatted images (MPRs) in arbitrary sectional planes through the scanned volume (see pp,48), In cases with equivocal findings, MPRs can assist the diagnosis by providing information in an extra viewing plane, When thin sections have been used, reformatted images of good quality can be produced, Three-dimensional images (see pp.Gtl) of the skeleton or vascular system improve anatomic orientation and can enhance the presentation of findings to referring colleagues,
Multislice CT
Multislice CT, or multidetector-row CT, multidetector CT, or volume CT, as it is also called, represents the next breakthrough in CT technology, Multislice CT scanners provide a huge gain in performance that can be used to reduce scan time, reduce section collimation, or to increase scan length substantially, Multislice CT transforms CT from a trans axial into a truly three-dimensional imaging technique, Multislice CT has gained a rapid acceptance by the radiological community, There is an almost exponential growth of the number of scanners: in 1998 there were 10 scanners installed, in 1999it was 100by the middle of the year, and by the end of 2000, over 1000 scanners were installed worldwide,
Scanning Principle
Unlike standard systems that use a single detector arc or detector ring, multislice CT systems are equipped with two or more parallel detector arrays and always utilize a third-generation technology with synchronously rotating tube and detector array as well as solid state detectors (Fig, 1.17),Although the very first CT scanners in the 1970s also employed a split detector, the first dual or split detector systems with continuously rotating tubes becarue available only in the early 1990s, Systems with four active detector arrays were introduced in 1998, and systems with 8, 10 or 16 active detector arrays are now available, The performance of many of these systems is further improved by a faster rotation time, As a result, a four-detector-row scanner with 0,5 s
Scanning Principle 19 Table 1.6 Advantages of higher scanner performance with multislice CT or subsecond spiral CT Shorter scan duration Reduced motion artifacts: • Children • Trauma patients • Acutely ill patients Improved scanning of parenchyma! organs • Well-defined phase of contrast enhancement Reduced volume of contrast medium Perfusion imaging Longer scan ranges Trauma
II»
Thoraco-abdominal
spine/blunt trauma
CT angiography: • Aorta and peripheral run-off • Thoraco-abdominal aorta • Carotids from arch to intracerebral circulation Thinner sections Fig. 1.17 Principle of multislice CT scanning. More than two detector rows are exposed simultaneously.
rotation has an about 8 times higher performance than a conventional 1 s single-detectorrow scanner. Multislice CT does not always involve spiral data acquisition, although spiral acquisition is implied in this book unless stated otherwise. A sequential mode (step-by-step scanning) as in conventional CT with possibilities of image fusion is also available, although it will be used only in a minority of applications, such as HRCT of the lungs or interventional procedures. Advantages
The performance of multislice CT is at least 4 times, with modern scanners 8-20 times higher than that of a conventional spiral CT scanner. This enormous increase in performance can be used for shorter scan duration, longer scan ranges, and thinner sections (Table 1.6). In general, the performance gain will be used to improve all of the above to a substantial amount. In consequence, multislice CT has overcome one of the most severe limitations of spiral CT, namely the inverse relation between scanning range and section collimation.
IMear-isotropic imaging: • • • • •
Temporal bone imaging Musculoskeletal imaging Arbitrary imaging planes Multiplanar reformats 3D rendering
Shorter scan duration will reduce the danger of motion artifacts, especially in children or critically ill patients. Marked improvements can be seen for trauma patients or in dyspneic individuals with suspected pulmonary embolism. Shorter scan duration also will allow for scanning of the liver or other parenchymal organs in a more well-defined phase of contrast enhancement, which again improves lesion detection and characterization. Shorter scans will also allow for substantial reduction of the volume of contrast material to be administered as long as only an arterial phase of contrast enhancement is required. Longer scan ranges are especially important for CT angiography. CTA-examinations become feasible that include the abdominal aorta and the peripheral run-off vessels down to the feet. Alternatively, the whole aorta can be scanned with high spatial resolution, or the carotids can be examined form the aortic arch to the intracranial circulation. Long scan ranges such as in thoraco-abdominal studies no longer pose a
1 Principles «ct. Spiral CT, and Multislice cr problem, even for indications that require a high spatial resolution. Finally, thinner sections and even isotropic imaging will become feasible yielding almost identical spatial resolution in all directions. This makes near-isotropic multiplanar imaging possible with a spatial resolution in any arbitrary plane that equals, or often exceeds, the resolutionofMRI. Disadvantages
The downside is a markedly increased data load, especially if near-isotropic imaging is performed. A scan of the chest and abdomen (60cm) can be performed with a 4 x 1 mm collimation in 50s and produces up to 500-800 images, depending on the degree of overlap. A chest scan with an identical collimation will produce as many images if reconstructed in a smoothing convolution kernel for the mediastinum and an edge-enhancing kernel for the lungs. ACTA of the aorta and peripheral arteries may produce 1000 images and more. The only way to avoid this enormous amount of data is to (acquire and) reconstruct thicker sections. This, however, will forgo many of the advantages of multi slice CT. Current scanning protocols frequently are modified versions of standard spiral CT protocols with a somewhat thinner section collimation. Most present scanners and workstations can easily handle the data created by such protocols. However, if near-isotropic imaging is performed, image reconstruction speed and data handling become limiting factors for some of the present scanners and workstations and make this technique cumbersome for clinical routine use. There is not yet an in-depth experience with near-isotropic imaging, nor do standardized protocols exist. This book suggests protocols optimized for clinical routine work as well as specialized protocols for more advanced applications and multiplanar imaging. Image processing adds to the increased time requirements for data analysis if a full-resolution scan is performed. No standardized protocols are available yet but manufacturers are working on techniques to include 3D processing in the standard workflow. Image noise grows as section collimation is reduced. For this reason, it is important to reconstruct thicker sections (MPR or axial sections) to keep image noise low. With very thin
collimation the geometric efficiency of the detector deteriorates (see Fig. 5.5). It varies between manufacturers and depends on scanner geometry, the implementation of beam collimation and image interpolation algorithms. With 16-slice scanners, the geometric efficiency approaches that of single slice CT. An increase in patient dose is only necessary if thin-section images of high quality are required. In all other cases, multislice CT requires less dose than conventional CT or a similar dose as spiral CTwith a pitch of2. Detector Types Dual Detector Systems
Dual or split detector systems are based on a detector array that is twice as wide as a conventional CT detector and is split in half (Fig. 1.18). Elscint was the first company to use this concept on a spiral CTsystem (ElscintTwin). Now, most vendors offer such dual-slice systems. Multidetector-row
Systems
True multidetector-row or multislice systems are able to acquire at least four simultaneous sections. To be able to choose between various section collimations, the detector arrays have to be subdivided into multiple detector rows. Data from each of the scanned sections is recorded by a data acquisition system (DAS), which consists of one detector row or a combination of detector rows, depending on the chosen section collimation (Figs. 1.19-1.21).The basic types of detectors available are described below. The width of each detector row is not given as its real width but instead as the width of the corresponding xray beam at the center of the scan field. Matrix Detector
Matrix detectors consist of multiple detector rows of identical width (Fig.1.l9a). The GE matrix detector is a typical example. It uses 16 parallel detector arrays with a width of 1.25mm each. One has to note that the detector width mentioned in this text (as well as most publications) is not the "real" width of the detector elements but the width of the detector-row beam
Detector Types
~~'.~.x-ray tube ..~.~~ Beam collimators
. X-ray beam ... Central trajectory
I I I I I Single-slice
Fig. 1.18 Comparison
of single and dual (split) detector systems.
collimation, measured in the center of the scan field. The real width of the elements of the x-ray detector is about twice as large, depending on the scanner geometry and the distance from the center of the scan field to the actual detectors. By exposing only the innermost four detector rows, a 4 x 1.25 mm collimation can be produced. For wider collimator settings the signals of groups of two or more adjacent detector rows have to be added to obtain 4 x 2.5 mm, 4x3.75 mm and 4x5 mm collimations (Fig. 1.19b). By partially covering the innermost two detector rows, also 2 x 0.63 mm sections can be generated. The same principle of grouping adjacent arrays holds true with 8-detector-row scanners, which will allow for 8 x 1.25 mm and 8x2.5 mm section collimation (Fig.1.l9c). The total collimation (e.g., 8 x 1.25 mm = 10 mm, or 8x2.5 rnrn Zfl mm) is limited by the total width of the detector array. e
•
Adaptive
Multislice
Dual-slice
Array Detector
Adaptive array detectors consist of detector rows that grow in width from the center of the section to the periphery (Fig.1.20a). Philips (Asymmetrix detector) and Siemens (AAD) have adopted this concept and use the same detector in their scanners. By collimating half of the two innermost 1 mm-wide detector arrays of 4-detector -row scanners, a section collimation of 2 x 0.5 mm can be obtained. By partially collimating the
next arrays, 4 x 1 mm sections can be gained. Adding the innermost two detector rows will yield 4 x 2.5 mm collimation (Fig. 1.20b). The rationale behind this approach is based on the fact that the x-ray beam hits the more peripheral detectors at an oblique angle. Any perpendicular septa in the more peripheral portions of the detector will block some of these oblique rays and will thus decrease the detection efficiency. Reducing the number of septa in the peripheral regions of the detector array should thus increase geometric dose efficiency. Indeed, the detector has a dose efficiency that is only slightly less than that of a single-slice scanner, if a section collimation s2.5mm is used (for more detailed discussion on geometric efficiency see Chapter 5 and Fig. 5.5). The detector geometry is optimized only for a particular number of active detector rows. A new detector configuration is needed as the number of detector rows progresses from 4 to 16 (Fig. 1.20 c). With more rows, also more septa have to be added. Hybrid
Detectors
Hybrid detectors are similar to matrix detectors with the exception that the innermost detector rows are thinner than the outer ones (Fig. 1.21a). This concept is embraced by Toshiba (selectable slice-thickness multirow detector, SSMD). These hybrid detectors are based on a detector array of a total width of 32 mm that con-
21
22
1 Principles of CT, Spiral CT, and Mu/tislice CT _16x1.25mm
__ 5mm
1.5mm 1 2.5mm :1 ':l.m
mm 2.5mm
.
Srnm
Center of scan plane ___
20mm_
20mm
2xO.63mm 4xl.25mm 4x2.5mm 4x3.75mm 4x5mm
2xlOmm
8xl.25mm 8x2.5mm 16xO.63mm 16xl.25mm Fig. 1.19 Matrix detectors consist of detector rows of identical widths. For the GE HiLight detector this width is chosen so that each detector row covers a 1.25 mm-thick section in the center of the scan field (a). Collimating the x-ray beam, or combining the signal from adjacent detector rows, or both, will yield various section collimations (b). An 8-row scanner provides 8 x 1.25 mm and 8 x 2.5 mm collimation. The 16-row scanner uses a hybrid detector with 16 x 0.63 mm and 16 x 1.25 mm collimation (c).
sists of 1 mm-wide detector rows, With 4-detector-row scanners the innermost 2 mm of this array are further subdivided into four 0.5 mm wide detector rows. With this detector configuration there is a multitude of possible collimations, ranging from 4 x 05 mm, 4 x 1 mm to
2xO.5mm
4x1 mm
4x2.5mm 4x5mm
2x8mm
•
Mll
II
i ••.
iiWIiI
1111t.tIIIIII!l%~
16xO.75mm 16x1.5mm
8x3mm
11111,1,llllllh
11111111111111111
Fig. 1.20 Adaptive array detectors have outer detector rows that are wider than the innermost detector rows (a). The detector used by Philips and Siemens covers a total section width of20 mm and allows for section collimations ranging from 2 x 0.5 mm to 4 x 5 mm (b). A new detector configuration is chosen for the new 16row scanner, which will be ableto provide 16 X 0.75 mm to 16 x 1.5 mm collimation (c). Philips will additionally offer a 8 x 3 mm collimation.
4 x 8 mm (Fig. 1.21b; Table 1.7). With 8- and 16detector-row scanners the innermost 8 or 16 rows of the 32 mm-wide detector array have a width of 05 mm (Fig. 1.21c). The Toshiba detector uses dividers between the various detector rows that are opaque to the
Detector Types
~-
4xO.5mm
15xl mm
15xl mm--
Fig. 1.21 Hybrid detectors are matrix detectors with thinner innermost detector rows (a). Toshiba uses a detector that covers 32 mm and allows for 4 x 0.5 mm to 4 x 8 mm collimation (b). The new detector configuration for a 16-row scanner will be able to provide 16 x 0.5 mm to 16 X 2 mm collimation (c).
32mm
scintillation light produced by each detector element (thus avoiding cross-talk) but which are transparent to the incident radiation. This effect
Cone Beam Systems (Volumetric
increases dose efficiency but also makes the de-
on cone beam technology in which a wide x-ray beam is used to cover a large range along the zaxis. Prototypes by various vendors include a
tector slightly more vulnerable radiation. The 16-detector-row
to scattered
CT)
Future generations of CT scanners will be based
scanners of all other
2S6-row detector and a 1024' flat panel detec-
vendors (Figs. 1.19c and 1.20c) also use thinner
tor. Such scanners may provide enormous spatial resolution and high-speed data acquisition.
innermost detector rows and therefore
be considered hybrid detectors.
should
Cone beam systems that are already on the market are reserved for rotational
angiography.
23
•
1 Principles ofCT, Spiral CT, and Multislice CT They use image intensifiers or (CCD) flat panel detectors as the radiation receptor and require a num ber of seconds for acquiring one (single-rotation) data set. The present systems are only able to depict high-contrast structures such as vessels during intra-arterial contrast injection. The reason is the large amount of scattered radiation that is not blocked from hitting the detector-as opposed to multislice CT where the relatively narrow beam and the various septa between the detector rows achieve adequate scatter suppression. This scattered radiation destroys low-contrast resolution if no adequate scatter-reduction techniques are developed. Unless this major obstacle is solved, true cone beam CTwill not become clinically feasible. Another major problem with cone beam detectors is the bandwidth of the signal transmission system that has to transfer the various detector signals out of the gantry to the image reconstruction computer. The data throughput in these systems will become enormous and will grow by a factor of 64 relative to a 4-detector row scanner if 256-detector rows are employed. At present there is no system that can transmit and process the acquired data with sufficient speed. The third and biggest problem is image noise, which will increase substantially. Image noise for isotropic resolution with voxels of a size of x3 is theoretically proportional to IIx'. This means that image noise will grow by a factor of 2*~ 16 if the size of an isotropic voxel is reduced from (Irnm}' to (0.5mm)3, and by 10,000 if the voxel size is decreased to (0.1 mm)3 As a consequence, sophisticated noise suppression and data reconstruction techniques will be necessary to create clinically useful images. For the reasons mentioned above, it will be several years until cone beam scanners become available for clinical use.
1a'~
System Performance The performance of multislice CT scanners increases with the number of active detector rows, the speed of x-ray tube rotation, and with the available pitch factor. This increased performance can be used to reduce scan time, reduce section collimation, or to increase scan length (Table 1.6). At the same time, increase in
performance has to be counterbalanced by the amount of artifact induced and by increased image noise with thinner sections. Rotation Speed
The rotation speed of the x-ray tube is commonly described by the time required for one tube revolution, the so-called rotation time RT. All multislice scanners have a rotation time of 0.8 seconds or less. Most high-end scanners even allow for a 0.5 s rotation. With the newest generation of scanners, rotation times of 0.42 s are available. Pitch
There are two definitions ofpirdi factor available with multislice CT scanners, depending on whether a single section collimationSC or the total cOiKmatwnofthedetectorarray(N x SC)ischosen as the reference. To distinguish between them, an asterisk is used to indicate the definition used by most manufactures for 4-slice scanners (volume pitch P*, also called sh'ce pitch or detector pitch), while P denotes the definition agreed upon internationally (lEC) and preferred by most physicists (pitch P), also called beam pitch:
P=TFI(NxSCj
P*~TF/SC
Pitch Volume pitch
While P is independent of the number of detector rows, P* increases as the number of detector rows grows. As with single-slice spiral CT, the pitch P can be increased up to 2, independent of the number of detector rows N. In multislice scanners with four active detector rows this corresponds to P*=8. Some manufacturers have limited the maximum pitch P to 1.5 (corresponds to P* = 6 for four-slice scanners). Relative System
Performance
System performance is proportional to the number of detector rows N and improves with a shorter rotation time RT of the x-ray tube. This concept can be applied to standard spiral CT as well as to dual detector scanners or multislice CT. Table 1.7 gives an overview of the relative performance of various scanner types and scanning protocols. It has to be noted that the quality of axial images may deteriorate when the highest possible pitch factors are chosen.
Image Reconstruction Table1.7 Examples of system performance relative to a 1 sec single-slice spiral CT scanner with pitch System
Rotation time
(5) 1s-scanner 1s-scanner 0.8 s-subsecond scanner 0.5 s-subsecond scanner 1s-dual detector scanner 0.5 s-dual detector scanner 0.8 s-multislice scanner 0.8 s-multislice scanner 0.5 s-multislice scanner 0.5 s-multislice scanner 0.5 s-multislice scanner 0.42 s-multislice scanner
1s 1s 0.8s 0.5 s 1s 0.5s 0.8s 0.8s 0.5 s 0.5 s 0.5 s 0.42s
Image Reconstruction
Image reconstruction in 2- and 4-detector-row (T systems does not have to use sophisticated interpolation if a section-by-section mode is used, but even then, the cone beam geometry of the system will lead to artifacts for regions outside the center of the gantry. With spiral acquisition, the situation is becoming more complex because the trajectories of the various detector rows may overlap and yield "redundant" data, depending on the pitch factor employed. For this reason, more sophisticated raw data interpolation schemes have to be used. •
The Cone Beam Problem
Due to the nature of the x-ray beam, which arises from a small focal spot, all (T scanning relies on cone beam geometry. With single-slice CTthese effects are hardly visible but they play a major role in multislice scanning because the same structure may be depicted on different detector rows during one revolution of the x-ray tube (Fig.1.22a). Only in the center of rotation are structures always captured by identical detector rows. The effect becomes more prominent the further a structure is from the isocenter (rotational axis), and the more detector rows are used (Fig. 122c). Simple raw data interpolation algorithms make the assumption that all beams are parallel (Fig.1.22 b), which yields reasonable results up to four active detector rows but fails with a higher number of rows. While adaptation for
Detector rows (N)
Pitch P
Pitch p.
=
2
Rei. performance
0.5 2 2 2 2 2 0.75 1.5 1.5 2 1.35
1 2 2 4 4 4 4 8 16
2 2 2 4 4 3 6 6 8 10.8 32
2
1.25 2 2 4 1.875 3.75 6 8 10.8 38
thin section and 8-slice scanners may still be possible, 16-slice scanners require more sophisticated cone beam algorithms for image reconstruction. •
Multislice Linear Interpolation
Algorithms for 4-slice systems that are analogous to 180 Ll and 360 LI from spiral (T are called 180 MLI (multislice linear interpolation) and 360 MLi (Fig. 1.23). For each projection angle, they use the projection data from the two detectors that are closest to the scanning pane (3600 MLI only real trajectories, 1800 MLI also virtual trajectories from the detector to the x-ray tube). Section profiles with these algorithms vary between those from conventional 180 Ll and 360 LI spiral (T interpolation, but the dependence on the pitch factor is more complex because of the sampling patterns described below. It is difficult to represent these effects using the standard diagrams for single-slice spiral (T. A diagram that displays the rotational angle of the x-ray tube for varying z-positions of each DASchannel (active detector row) is more suitable for this purpose (Fig. 123 b, compare also Fig. 113a). 0
0
0
0
0
•
0
The Sampling Problem
Depending on the pitch factor, the spiral trajectory of the first detector row of a 4-slice scanner may overlap with that of the second, third or fourth row for P* = 1, 2, and 3, respectively (Fig. 1.24). For a pitch of 0.5 (P* = 2), even the
25
•
7 Principles aICT, Spiral CT, and Multislice CT
Assumption for reconstruction: parallel rays
a
FOV z-axis
c
4 detector rows
8 detector rows
Fig. 1.22 Cone beam geometry. Peripheral structures will be seen by different detector rows during one revolution of the x-ray tube (a). Simple raw data interpolation algorithms make the assumption that all beams are parallel like in convention spiral CT scanning (b). Such algorithms will fail with more than four detector rows be-
cause the cone beam effect will become more prominent when more detector rows are used (c). The distance ASC traveled by the outermost rays within the field-of-view (FOV) becomes larger than the section collimation SC and will cause artifacts in standard reconstructions.
conjugate data from the virtual spiral overlap the real spiral at the isocenter (Fig. 1.24b). This makes data sampling particularly inefficient and only allows for an interpolation between data samples that are one section collimation apart. As a consequence, the section width is widened Similar to a 360 LI interpolation. Because of
varying degrees of overlap with changing pitch factors, the sampling density and thus the section width varies (Fig. 1.25). In principle this sampling problem applies also to 8- and 16-slice scanners but, because of the cone beam effect, the projection rays for the overlapping detector rows are not completely
0
Image Reconstruction Fig. 1.23 Multislice linear interpolation for a 4-slice scanner and a pitch P = 1.5 (volume pitch P* =6). The conventional representation (a) and the angular diagram (b) are equivalent but the angular diagram provides a better overview. Note that the trajectories of the original spiral (detector row 4) and the conjugated data (from detector row 1 to x-ray source) overlap.
lrnaqeplane
3
2
a
5
4
Z-position/collimation Image plane 360
dJ
en c:
" " ., ., " " ""
270
rtI
c:
0
BdJ
180
.,
""
'0 c, L-
"
90
., "
.,
., " " .,
"
o1
.,
., " ., /
3
5
4
""
360
270
"
0"
~" "
6
8
9
- - - - - - conjugated spiral trajectories
lmaqeplane 360 ,---_"..-----,r--_--,...,-----,
I
/
__ ----,
/
01 C
/
/
ro
1/1
180
Pitch 0.75 (P*
, = 3)
,
/
'0 c,
/
/
L-
/
§ 180
/
'B
I
/
OJ
'0'
/
/
'-
/
Q..
1
2
3
:4
5
6
Z-position/collimation
Fig. 1.24 Overlapping data sampling occurs when trajectory of one detector row superimposes another, e.g. for a pitch of 0.75 (P* = 3) (a). In the center of rotation, even the conjugated data from
b the on isothe
90
o
O~~-~~~-~-L--~i----£--L~~~
o
270
OJ
/
/
rtI
a
.,
/
I
en c:
g
., " "
"
Image plane
BdJ
., .;
""
/
Z-position/collimation original spiral trajectories
dJ
8/
/
"
2
b
., " "
.,
o
1 2 3 Z-position/collimation
=4
5
6
virtual spiral overlap the real spiral trajectory for a pitch of 0.5 (P* =2) (b). Note, however, that the cone beam geometry ameliorates this effect, especially for image points farther away from the isocenter.
27
28
1 Principles ofCT, Spiral CT, and Multislice CT
SW/SC 300%
200%
128% 100 % I-----~::::::-.------
---Pitch
f-----_.__---t-------,-------\
1.5
a
2
2.5
3
P
SW/SC
redundant outside the isocenter. Still, there are constellations in which the sampling is particularily advantageous or disadvantageous. While Philips and Siemens have made their interpolation algorithms independent of pitch, Toshiba and GE provide preferred pitch factors to ensure particularly advantageous data sampling for their 4-, 8- and 16-slice systems. Siemens has chosen another approach, in which a z- filtering for 4-slice systems and cone-beam reconstruction for 16-slice systems is chosen in a way that makes section width, image noise and radiation dose independent of the sampling pattern with varying pitch (see Figs. 1.25c and 1.28c). •
150% 360 MLI
..
..
:
:HS
• ]JH~························'···.·•··············· L--
0.25 b
__
~
Pitch
_.__-;----------'
0.5
0.75
2
3
4
1.25
1.5
1.75
5
6
7
2 8
P P*
SW/SC 200 %.... ,-----.---,--""""7'"""--;--.,.------,---, SW=2xSC I I
r
SW=,1.5xS
I
150%
I
I I I I
I
SW= 1.28xSC(Standard)
100%
c
II
HS,
I
.--+--...,
1------1--.[ HQ
SW= 1.0xSC(Thin slice)1 , " 0.25 0.5 0.75 1 1.25 1 2 3 4 5
I
1
1
1.5 6
,
Pitch
1
1.75 7
2 P 8 P*
Fig. 1.25 Effective section thickness (section width) as a function of pitch for 180 LI and 360 LI with single-detector-row CT (a) compared to 180 MLI and 360 MLI with 4-slice scanners (b). Adaptive interpolation schemes (c) keep section width independent of the pitch. HQ = high quality mode, MS = high speed mode on GE 0
0
0
scanners.
0
Z-Filter Interpolation
Z-filter interpolation is reserved for 4- and 8slice scanners uses a similar concept as higherorder interpolation algorithms for conventional spiral CT. Not only the two projections from the detectors that are closest to the scan plane are used but also adjacent projections [multipoint interpolation). These projections are weighted according to their distance from the scan plane (Fig. 1.26).This filter function may even contain negative portions that result in "edge enhancement" along the z-axis and thus can ensure that the section width is equal to the collimation even at unfavorable pitch factors. This, however, will come at an increased image noise. By using a wide filter function, noise is reduced and the section width becomes larger. Z- filtering controls the width of the slice profile (section width SW) of the reconstructed images. While most manufacturers (Philips, Siemens, Toshiba) always display the correct section width SW on their user interface, GE has chosen to display only multiples of SC as potential section widths SW on their scanners. This is correct for the vast majority of parameter combinations. For the thinnest section widths at pitch P> 1, however, the true width of the section profile at half maximum (FWHM) is approximately 30% larger than SC (for 4- and 8slice scanners but also for 16-slice units) and than the displayed value for SW. This widening of the section profile is also present with other manufactures but is correctly represented in the displayed section width SW. Depending on the manufacturer, various combinations between chosen section collimation and reconstructed section width are available (Tables 1.8-1.10). In principle, the only re-
Image Reconstruction
360
Fig. 1.26 Z-filtering is a technique that weights projection data according to distance from the reconstructed imaging plane.
/ / /
270
/
~
/
01
C
ro c 0
" ,,~
180
B Q.J
'0 L..
0...
90
/
Pitch 0.75
/ ,.
/ /
/ /
/ /
I
/
/ /
/
/
/ /
/ /
0
0
Z·Filterfunction
6
narrow
h(z)
Z-Filterfunction wide
straint is that the section width must be larger than or equal to the collimation. The main difference between manufacturers is how they treat the minimum available section width. If standard 180 MLI or 360 MLI algorithms (or variants) are used, the section width varies between 100% and" 128% ofthe section collimation, depending on the pitch factor. The minimum section width with the GE high speed mode (P *= 6) on 4-slice scanners is some 30% wider than the collimation SC (see Table 1.8). It is only in this mode and in the UF (P*=10.8, P=1.35) and US (P*=13.4, P= 1.675) modes in 8-slice scanners that the "reconstruction slice thickness' indicated on the user interface is not approximately equal to the section width. Philips provides the real section width on theiruserinterface. The available minimumnumbers vary with pitch (between SC and SC+30%). 0
0
Siemens varies their z- filter (adaptive array interpolation) with the pitch in such a way that section width, image noise, and dose requirements remain constant independent of pitch (Fig.1.25c). As a standard, they offer a section width that is some 30% wider than the section collimation. An edge-enhancing z- filter is also available for the 4 xl mm detector configuration that provides 1 mm section width independent of the pitch. The resulting images, however, suffer from a substantially higher noise Oust like the difference between 180 LI and 360 LI in spiral CT). Toshiba uses a z-filter of varying width (MUSCOT algorithm) that provides a given section width independent of the chosen pitch factor. The minimum section width is identical to the collimation and can be increased in steps of 0.5 mm. If SW is chosen equal to SC, the noise is substantially increased (like with all other vendors). 0
0
29
1 Principles ofCT, Spiral CT, and Multislice CT
•
Cone Beam Interpolation
Z-filter interpolation still relies on the assumption that rays are parallel (Fig. 1.22c). Real cone beam corrections require more complex calculations. Such algorithms are mandatory with 16 or more detector rows because cone beam artifacts (see Fig. 7.47) increase substantially with wider detector arrays. Most of these algorithms are still being refined, and use various types of compensation techniques for the cone beam geometry. Variants of 3D back projection (e.g., COBRA, Philips or ConeView, Toshiba) theoretically should yield the least artifacts. They rely on reb inning of projection data and back projection along the correct cone angles. For each voxel, the projectional rays that are next to it are actually used for backprojection. The Adaptive Multiple Plane Reconstruction (AMPR, Siemens) shifts the plane of interpolation from an axial orientation to an oblique position with a maximum angulation determined by the cone angle (Fig. 1.27). This yields a set of oblique planes (as an intermediate step) that rotate with different z-positions. Interpolation between these oblique planes then creates axial, coronal, or arbitrarily oriented sections of any desired section width without necessarily having to go through a real reconstruction of an orthogonal 3D data set. GE uses a combination of approaches (Crossbeam correction, Hyperplane and Conjugate Ray reconstruction) to address the various problems of cone-beam scanning, i.e. cone-beam artifact reduction and thin-slice, high-pitch scanning. At present, artifact behavior of these various algorithms has not yet been studied.
The cone beam interpolation used in the 16-slice scanners of Philips, Siemens and Toshiba allow for arbitrary pitch factors up to 1.5 to 2. However, Toshiba suggests preferred pitch values (at present P*=ll, 15 and 23; P= 0.6875, 0.9375, and 1.4375, respectively). Similar to their 4-slice and 8-slice systems, GE supports fixed pitch factors (P*=9, 15, 22 and 28; P= 0.5625, 0.9375, 1.375 and 1.75, respectively), for which their cone beam algorithms have been specifically optimized. •
Image Noise
Image noise in multislice CT depends on the zfilter (or cone beam algorithm) and the chosen section width. To understand how, we have to look back at conventional spiral CT. There, the noise was independent of the pitch but varied with the chosen interpolation algorithm. In fact, it was much smaller with 360 LI as compared to 180 LI; for identical noise, the dose with 180 LI would have to be doubled. The 180 LI algorithm had an identical performance to the 360 LI only at a pitch of 2 and twice the mAs setting (same section profile, identical noise at identical dose). This should have made the 360 LI algorithm the interpolation of choice for low pitch factors, and is another argument for using 180 LI only in combination with high pitch factors. The effects becomes clear when the noise data are normalized to identical patient exposure (Fig. 1.28 a), which can be done by increasing the mAs setting proportional to the pitch so that the effective mAs (=mAs/Pitch) remain constant. Multislice CT overcomes the limitation of single-slice spiral CT because of the higher number of detector rows and, thus, faster coverage. With 4-slice scanners, the 360 MLI algorithm has similar advantages in terms of image noise and dose requirements over 180 MLI as the corresponding algorithms from single-slice CT. When considering the pitch-dependence of the section width (Fig. 1.25) and normalizing the noise behavior to identical patient exposure (Fig. 1.28b), it becomes clear that a 360 MLI algorithm should be preferred for pitches up to 1 (slice pitch up to 4), and the 180 MLI algorithms should be used for pitches up to 2 (P" up to 8). Since a variant of 180 MLI is used for both the HO. (P*=3) and HS (P*=6) modes ofGE scanners, the noise per dose is higher for the HQ. mode than with the HS mode. 0
0
0
0
0
0
0
0
o
0
0
0
0
Fig. 1.27 Cone beam interpolation by interpolating raw data to obtain multiple angulated imaging planes that are then interpolated to form a 3D volume.
Image Reconstruction ReI. dose requirements for constant noise
0
lS0%
100 % i-------''<:----~---'''o,_=__----
i ~~~ ..···..··'··········" ..·····i~
180 LI
360 LI
:
0
Pitch
i------r-----r---r---,----,---,--
a
0.25
0.5
1.25
0.75
1.5
1.75
2 P
ReI. dose requirements for constant noise 180MLI
lS0%
"''''--·'f'''·--'' . --.--..'--'.----..+.~.--." --"."..
133 % 100%
Pitch
b
0.25 1
0.5 2
0.75 3
1 4
1.25 5
1.5 6
2 P 8 P*
1.75 7
ReI. dose requirements for constant noise ISO% SW=S --
133 %
--.--1': --
100%
: SW=1,2~=~ SW=l.Sa==------
67 %
--
SO%
,
--.. ,,-- ,---- .. 1':
-- --.-- ..
Conv. CTl::;=;;;;;::;;;::~::::::=---
--.----
t-v~
--.--.-- .. --, .. ----
~
SW=2xSC:,
Pitch
f----~------'-----
c
More complex z- filtering may improve the relation between noise and section width (e.g., SmartHelical) and can reduce the noise compared to a 180 LI algorithm (GE). Siemens use adaptive z- filtering to keep the noise and the section profile independent of the pitch at identical patient exposure (Figs.1.25c and 1.28c). The noise with this algorithm is 12-16% lower than the noise with conventional CT but the section profile is always some 30% wider than the collimation (as with 360 LI). Since 4-slice CT allows for using very thin collimation (4 xl mm), this broadening (to 1.3mm) is of little practical importance. With the Toshiba approach, the user selects a certain section width (in 0.5 mm steps) and the system chooses an appropriate zfilter width. Noise will increase substantially if the smallest section width (identical to the collimation) is used but decreases with the next larger section width. When choosing an even wider section width, more data are included in the reconstruction and image noise decreases further. The noise behavior with 16-slice scanners has not yet been published but in general, a similar behavior as for -slice units can be expecdet. If the section width SW is chosen identical to the section collimation SC, image noise should be substantially increased, while it should rapidly diminish if at least a 30% wider section width is employed.
0.25 O.S 123
0.75
1
4
1.25 567
1.5
1.75
2 P 8 P*
Fig. 1.28 Relative dose requirements for constant image nois as a function of pitch. Comparison of 180 LI to 360 LI for sinqle-slice CT (a), and 180 MLI for -l-stice CT (b). Adaptive interpolation keeps image noise fairly independent of the pitch (c). 0
0
0
31
1 Principles act. Spiral CT, and Multislice CT Scanning Parameters As with spiral CT scanning, section collimation (SC), table feed per rotation (TF) , and pitch (P) are the most important acquisition parameters in multislice CT In addition to the reconstruction increment (Rl), however, there is the effective section thickness or section width (SW) of the reconstructed images that contributes to the most important reconstruction parameters, All the other parameters are varied only in exceptional cases, Together with the number of active detector rows N, the acquisition parameters can be given as (N x SC ITF), and the reconstruction parameters can be given as (SW / Rl). Because more than one set of reconstructions is possible with multislice CT, it is reasonable to separate acquisition and reconstruction parameters,
Acquisition Parameters As has been painted out above, the section collimation is determined by the available detector configuration, In general, thinner sections (4x 1 to 4 x "1.25mm, and 8 x 1 to 8 x 1.25 mm) are used for near-isotropic volumetric imaging with the option to reconstruct arbitrary cut planes, Thicker sections (4 x 2 to 4 x 2,5 mm or more) are employed with fast spiral scanning for routine applications, and when data acquisition has to be particularly fast With 16-slice scanners, thin sections become standard even for routine fast spiral scanning, GE offers 16 x 1.25 nun: Philips and Siemens provide 16 x 1.5 mm, and Toshiba gives the choice between 16 x 1 mm and 16 x 2 mm for this purpose, For isotropic volumetric imaging, the scanners offer 16 x 0,62 mm (GE), 16 x 0,75 mm (Philips, Siemens), and 16 x 0.5 mm (Toshiba), There is a continuing discussion as to whether to use a high or low pitch factor, Lower pitch factors (e.g. , the HQ, mode in a 4-slice GE scanner) yield axial images with less cone beam artifact, but good results of a similar (and often better) quality can be obtained when images are acquired with a thin collimation and reconstructedwith a larger section width, In addition, low pitch factors require a higher patient dose for identical image noise, Using a thin collimation and a large pitch factor has the additional advantage that it is possible to go back to the
raw data and reconstruct it again with a thin section width, From such a data set, it will be possible to obtain superior quality multiplanar reformations (see Fig, 2,8), Such an approach is especially useful when findings on axial sections are not clear and require further evaluation by another imaging plane, With 8- and 16slice scanners this difference becomes less apparent because only thin sections are available for data acquisition, One has to bear in mind, however, that the geometric efficiency in 4-slice scanners (see Fig, 5,5) decreases with thinner sections, which will lead to an increase in image noise that may require using higher exposure settings, This limitation is overcome with 8- and 16-slice detector systems because of substantially higher geometric efficiency, There is a limit to such an approach of using thin collimation, however, on those scanners that do not allow for increasing the reconstructed section width arbitrarily, This is for example the case for current four-slice GE systems (see Table 1.8), A 4 x 1.25 mm collimation will not allow for reconstructing 5mm axial sections, In practice, many users will therefore not use such a thin collimation, but instead resort to 4x2,5 mm or 4x3,75 mm collimation, With the GE 8- and 16-slice scanners the maximum section width is 5mm even with the thinnest collimation, For the four-slice system there is a workaround described below that uses thick MPR from an overlapping thin-section secondary raw data set
Reconstruction Parameters The secrion width SW (effective section thickness) has to be larger than or equal to the section collimation Sc. Apart from this constraint, SW can be chosen independently from SC for most multislice CT units, The available choices for section width depend on the manufacturer and the type ofz-filtering (see Tables 1.8-1.10), Note that a section width that is identical to the collimation SC will result in a substantially increased image noise, and therefore should be reserved for only those applications with the maximum possible z-axis resolution (e.g, pulmonary or skeletal imaging), For most routine applications, and for reasons of image noise, a section width of 5-8 mm
Scanning Parameters 33 will be chosen. Only for special indications such as HRCT, skeletal imaging. imaging of the pancreas and adrenals, or preoperative staging of liver tumors will smaller section widths be necessary. The reconstruction increment RI can be chosen in a similar fashion as in spiral CT. For most routine applications, a moderate overlap of some 20% of the section width will suffice. For optimum quality of 3D reconstructions, at least a 50% overlap should be chosen unless the reconstruction increment is already as small as the pixel size. This will depend on the chosen field of view (see Table 4.4). For most body applications with an FOV of 30-40 cm, the pixel size is between 0.6 and O.S mm. Thus, a reconstruction increment of exactly the same size will yield an isotropic grid of data points. Note, however that this is not necessarily identical with isotropic data sampling (see also pp. 46). Chapter 4 gives an overview of recommended scan and reconstruction parameters for various clinical requests and various vendors (Tables 4.4-4.6). •
Secondary Raw Data Set
Whenever possible, a thin collimation should be preferred, because this will allow for creating a 3D data set of near isotropic resolution. We suggest calling such a set of overlapping axial images a secondary raw data set, because it can serve as the basis for reconstructing images in any desired plane and any desired section thickness using simple multiplanar reformatting (MPR)functions. Such a secondary raw data set typically consists of 0.5-1.5 mm-thick images reconstructed every 0.4 to 1 mm. The MPR function of the scanner or a workstation can be employed to create axial sections of arbitrary thickness and with arbitrary reconstruction increments. Such an approach can also be used with scanners that have no direct reconstruction of thicker sections from the raw data available (e.g., GE).ln addition, these sections can be anatomically adapted to compensate for improper positioning of the patient. Such a technique is especially helpful for symmetrical structures such as the inner ear or the cervical spine.
Vendor-specific Approaches In multislice CT, reconstruction algorithms are rather vendor-specific. All vendors use 2D fanbeam z- filtering algorithms for 4-slice scanners that neglect the cone-beam effect. These z-filtering variants largely determine which parameters the user may select and how. In the following paragraphs, the approach of the various vendors to 4-slice scanning is discussed. The approach to S- and 16-slice scanning is covered in chapter 4, optimization of scanning technique (p. 122, Table 4.6). General Electric Scanners
General Electric noticed that only at certain pitches the measurements from different detectors are relatively unique and complementary. From this the preferred helical pitch concept was developed, which made only two distinct pitch factors and corresponding section widths available. Scanning with P*=3 is called HQ (high quality) mode. while P*=6 is called HS (high speed) mode. The HO. mode provides good contrast resolution, low (cone beam) artifact levels and matches the quality of pitch 1 in single-slice CT. The HS mode is designed for high volume coverage and thin slices, comparable with pitch 2 in single-slice CT. For both pitch levels, image reconstruction is based on a variant of ISO° MLl. On GE scanners, the user chooses the section width SW (reconstruction slice thickness) first, depending on the clinical requirements. This choice is much as in single-slice helical CT. However, only multiples of the width of a single detector row (1.25 mm) are available (Table I.S). The user then has to decide whether to scan with 4 x 1.25 mm, or thicker collimation. Images can be reconstructed at arbitrary intervals, just as in conventional spiral CT. lfmultiplanar reformations are required, the data set can be reconstructed a second time with thinner section width. Consequently, quality of MPR is best if thin sections are chosen for acquisition. There is presently a constraint as to which combinations of section collimation and section width are actually available: for example, the largest section width that allows for using a 4 x 1.25 mm detector configuration is only 2.5 mm (Table I.S). This makes it necessary to use axial multiplanar reformations if section widths shall be reconstructed that are thicker than the maximum section width allowed by the scanner interface.
1 PrinciplesofCT, Table
1.8
Spiral CT, and Multislice CT
Selectable
Detector Configuration
options
in the GE Lightspeed
Available
table feed TF
system
(HU,
20(0)
Available
section
SW
interface)
(user
width
Measured
section
(FWHM') HQ_Mode2
width
HQ_Mode
HS_Mode2
4X 1.25mm
3.75mm
7.5mm
1.25 mm 2.5 mm
1.3mm 2.5mm
7.6mm 2.5mm
4x 2.S0mm
7.5mm
15.0mm
2.5mm 3.75mm
2.6mm 3.8mm S.Omm
3.2mm 3.8mm 5.0mm
3.9mm S.Omm 7.5 mm
S.Omm
7.5mm
5.0mm 7.Smm*
S.2mm 7.5 mm
2
S.Omm
4 x 3.75mm
11.25
mm
22.5mm
3.75mm S.Omm
4 x S.OOmm
15mm
30.0mm
IO.Omm* FWHM = full width at half maximum; section width SW is called face HQ = high quality, HS= high speed .• cnly available as 2 x 7.5 or 2 x 10 mm; with a special technique
1
10.0mm 'reconstruction
slice thickness'
HS_Mode2
NA 7.5mm 6.4mm 7.5mm 10.0mm on
user
inter-
2
For axial scans (step-and-shoot). raw data from multiple detectors can be averaged (1 i, 2i and 4i modes) before image reconstruction to reduce partial volume effects (e.g.. in the posterior fossa). Because of the lower pitch the patient dose (CTDlvoi. see Chapter 5) is twice as large for the HQ. mode compared to the HS mode if rnA settings are kept unchanged. Even if the scanner is allowed to change the rnA to provide equal image quality, HQ. requires about 50% more dose than HS.With the SmartHelical protocol. image noise can be reduced for both modes as compared to conventional helical CT. When switching from a GE single-slice scanner to amultislice unit, the patient dose increases dramatically if mAs settings are kept unchanged. The reason for this is the shorter scanner geometry, which yields a higher dose per mAs and problems with focal spot tracking. which were solved shortly after the first release of the LightSpeed scanner. For these reasons, a new user should not apply identical mAs settings but adjust the mAs downwardly to obtain a similar CTDlvoi with the two systems. Siemens Scanners
Siemens makes use of the adaptive array detector system and has developed an optimized fan beam reconstruction algorithm. called Adaptive Array Interpolation (AAI or SureView). This al-
2 x 0.625
mm is also available
gorithm ensures that the user-selected section width SW remains constant. independently of the chosen pitch factor (Fig. 1.25 c). Slice pitches P* between 2 and 8 are available for routine use. and even smaller values for cardiac CT. Available combinations of section collimation and section width are given in Table 1.9. In practice. pitch factors P* between 5 and 8 will be used in most situations. Siemens allows the user to prescribe multiple reconstruction protocols from one data set. thus speeding up workfiow if both a thick section for image review and thin sections for MPR and 3D imaging are required. To keep both image noise and patient exposure constant independent of the pitch (Fig.1.28c). the system raises the mAs settings automatically in proportion to the pitch factor. To make life easier for the user, the user interface on the scanner console does not display the real mAs but values for effective mAs, which are defined as mAseff= mAs I P = mAs x N I P*. These effective mAs correspond to the settings that would occur with a conventional step-and-shoot CT technique. The definition of mASeff can also be applied to other scanners and is valid even for conventional spiral CT. However. it also carries the risk that. should the user switch to a multislice scanner and apply the same mAs settings as on the single-slice unit. the change in definition of "mAs" on the
Scanning Parameters Table 1.9
Selectable options in the Siemens Volume loom; the Marconi MX 8000 is comparable
Detector
configuration 2 x O.5mm 4x 1.0mm 4x 2.5mm 4x Smm 2x8mm
FWHM
=
Available table feedTF
Available section width SW (FWHM)
0.5-2 mm 1.25-8 mm
0.5; 1.0; 3.0; 6.0; 8.0;
2.5-20mm 5-40mm 8-64mm
0.75; 1.0; 1,25; 1.5; 2.0 mm 1.25; 1.5; 2.0; 3.0; 4.0; 5.0; 6.0; 7.0; 8.0; 1O.0mm 4.0; 5.0; 6.0; 7.0; 8.0; 10.0 mm 7.0; 8.0; 10.0 mm 10.0 mm
full width at half maximum
multislice unit may not be accounted for and a substantially higher radiation dose would be applied. For example. a setting of 200 mAs at a pitch of 2 on a single-slice unit would correspond to a setting of 100 mAs(eff)on the multislice scanner. Keeping the 200 mAs setting would double the patient dose. Philips Scanners
Philips employs the same detectors as Siemens. The image reconstruction algorithm is termed Multislice Interpolation (MSlIMSSl). Exact details for this algorithm. however, are not yet public. User-definable options on the four-slice system are very similar to the Siemens system with collimations ofO.5-5.0mm, slice widths of 0.5-lO.0mm and pitch factors up to p~ 2 (see Table 1.9). Philips also employs the concept of effective mAs on most oftheir scanners. Toshiba
Scanners
Toshiba has developed a z-filtering fan-beam algorithm called multislice cone beam tomography reconstruction (MUSCOT) . Toshiba suggests preferred pitch values of P*~2.5, 3.0, 3.5, 4.5, 5.0, 5.5, 6.0, and 6.5. Toshiba prefers non-integer
Table
1.10
Selectable
options
in the Toshiba
Aquillion
Detector configuration
Available table feed TF
4xO.5mm 4 x 1.Omm 4X2.0mn 4x 3.0mm 4x4.0mm 4 x S.Omm 4x 8.0mm
1.2S-3mm 2.5-6 mm 5-12mm 7.5-18mm 10-24 mm 12.5-30 mm 20-48 mm
FWHM
=
full width at half maximum
values of P* to optimize sampling density and shift the position of the conjugate data. According to their specifications, however, best image quality is obtained with P*~3.0 and 5.5. The zfiltering is a multipoint interpolation using different filter widths (FW) that are automatically chosen by the system depending on which section width (SW) is selected by the user. The section width can be increased in 0.5 mm steps from a minimum identical to the collimation SC to a maximum of five times SC (Table 1.10). As with all other manufacturer, image noise is substantially increased for SW~SC. As with Siemens scanners, section width is independent of the pitch. Inpractice, apitchfactor P* ~ 5.5 (6.5 forCT A) will be used for most applications since it combines few multislice artifacts with a large coverage per time. For a constant rnA setting, patient dose decreases but image noise increases with larger pitch factors. For this reason, rnA settings have to be proportionately increased with larger pitch (constant effective mAs or CTDlvoi). Under such conditions of constant patient dose, noise is relatively independent of pitch.
Multi
"gAvailable
Section
width
SW (FWHM)
0.5-2.5 mm in 0.5 mm increments 1.0-5 mm in 0.5 mm increments 2-10 mm in 0.5 mm increments 3-15 mm in 0.5 mm increments 4-20 mm in 0.5 mm increments 5-20 mm in 0.5 mm increments 8-20 mm in 0.5 mm increments
7 Principles DICT, Spiral CT, and Multislice CT
Workflow, Image Review, Display and Documentation Multislice CT can be used as a technique for fast spiral scanning or as a true volumetric imaging modality (see also Chapter 4). With fast spiral scanning, thicker sections are reconstructed and viewed, very much like spiral CT. With volumetric imaging, a secondary raw data set of thin overlapping axial sections is created first. This data set is then used to reconstruct (anatomically adapted) thick axial sections as well as thick multiplanar reformations or various types of 3D images. The workflow of image processing, display, and documentation may change substantially. Fast Spiral Scanning
Fast spiral scanning is a technique that is excellently suited for routine imaging tasks. It provides at least equivalent, and often superior results to single-slice spiral CTwhile retaining the same basic imaging workflow. Display and documentation can remain similar to spiral CT (see above, p. 17). The thicker sections are used for making a diagnosis (preferably using cine displays for viewing) and can be printed on film. They are used to compare findings to older spiral-CT examinations, and are excellent for communicating findings to referring physicians. If there are remaining diagnostic problems, thin overlapping images can still be reconstructed (if thin-section scanning was performed) and further processing is done according to the volume imaging procedure described below. Volumetric
Imaging
Volume imaging takes full advantage of the three-dimensional capabilities of multislice CT. Reconstruction of a secondary raw data set of thin overlapping axial images gives the user full control over the available information contained in the data volume. However, such a secondary raw data set consists of several hundred images, depending on the covered scan range and the chosen reconstruction increment. This puts a lot of s train on image recons truction, image processing, data transfer, image reviewing techniques, archiving, and image demonstration. The imaging workflow has to be completely changed as compared to current scanners. With 4- and some 8-slice scanners, most of these
processes are not yet optimized because of substantial hardware and software constraints. As a consequence, the volumetric imaging mode will probably not be used as often as it should. Table 1.11 gives an overview of hardware and software requirements for optimum data handling when using a volumetric imaging technique withmultisliceCT. Image reconstmction from the raw data set can take a substantial time, even with sub-second reconstruction, if some 300-400 images are reconstructed for the chest or 400-500 images are reconstructed for one abdominal series. For this reason, very fast reconstruction (multiple images per second) is important to allow for a high patient throughput. Standard image reviewing is best performed on thick axial or multiplanar reformations that were reconstructed by the CT technician according to a predefined protocol that depends on the clinical imaging task (see Chapter 4, and the various organ chapters). The technicians can also perform various 3D reconstruction tasks according to specific protocols, such as curved planar reformations along the pancreatic duct, volume-rendered or MIP images of the vascular system, or exarticulated 3D views of joint surfaces (see Chapter 2). This allows for substantial time savings in a setting where imaging workstations are not real-time interactive. Interactive image reviewing is ideal if the imaging workstation is (near) real-time interactive. Mostworkstations on the market, however, suffer from substantial constraints because they are usually optimized for only a few specific processing tasks or they are too complex or too slow to be useful in a routine setting. Most important for interactive reviewing is a thick-section MPR mode that is truly real-time and allows for fast scrolling through the data set as well as for interactive change of the section thickness. Most software is real-time interactive for one-pixel-thin MPRs and only updates the image to the desired thick MPR as soon as the interaction is stopped. This may be sufficient for data sets obtained with high radiation exposure and thus little image noise. However, it is suboptimum for the vast majority of standard examinations, especially in the abdomen, where noise is very disturbing and can be reduced only by thick MPR. Interactive volume rendering is another option that gains increasing importance as a primary tool for image interpretation. Short data loading times, easy han-
Image Review, Disp/ayand Documentation Table 1.11
Hardware and software requirements for optimum workflow with multislice CT Volume imaging Minimum
Optimum
100-300
200-400
300-1200
> 1 image/sec > 10 patients 1 >6000
> 2 images/sec > 5 patients > 10.000
> 6 images/sec all pts. from one day > 40.000
CTworkstation RAM Hard disk" Volume accelerator board
512MB >40GB
1 GB >40GB
>2GB >80GB
Data transfer rate
100 Mbitlsec
Fast spiral scanning
Typical number of images Hardware
CTscanner Reconstruction speed Storage capacity of raw data Storage capacity for images2
100 Mbit/sec
+
III
;":
,1 1 111
/\ 1 Gbit/s
Software
CTscanner
Predefined scan protocols Predefined ~rocessing protocols (Thick) MPR MIP VRT
CTworkstation Real-timethick MPR Semi-automated segmentation MIP Volume rendering 1
2
3
depending on workflow. Raw data should remain available until it is decided whether additional thin sections are necessary depending on workflow. If further processing is done on the scanner, substantially larger storage capacities are necessary depending on workflow, Processing (MPR MIP, VRT) by the technicians can be done on the scanner or on a separate workstation
dling, proper opacity presets (see Table 2.4), and true real-time interaction are important for timeefficient image review and reporting. Processing workstations need to be equipped with sufficient computing power and memory. For multislice CT with several hundred images, the RAM should be more than 1 GB in size. For interactive volume rendering, accelerator boards are available that make real-time interaction possible (several frames per second). They differ with respect to the available matrix size for real-time interaction, and whether they allow for perspective rendering (important for virtual endoscopy) . Data networking may become another bottleneck. Fast networks are mandatory (at least IOOMbit/s, more for 16-slice units). Shared
databases and picture archiving and communication systems (PACS) for data archiving can help substantially to reduce data traffic due to multislice CT data sets. Multislice CT studies at present represent the largest data sets to be transferred over PACS networks. Transferring all necessary data (which may be well over 1000 images, if images are transferred to various workstations) from a multislice CT examination must not take longer than the time slot assigned to each patient on such a scanner. Appropriate planning prior to setting up a multislice CT system is essential to avoid clogged networks and suboptimum workflow. Image documentation will still require printing out of thick axial sections, especially when a patient is transferred to another institution. This
1 Principles oiCT, Spiral CT, and Multislice CT
is important as long as there are no generally
available standards for image reviewing outside of radiology departments in the offices of referring physicians. Most vendors now offer DICOM viewers that can be put on a CD-ROM together with the patient data. There is the additional
need to document critically important axial, multiplanar, or three-dimensional images on film or even on paper prints (color, black and white). Paper prints are generally well received by referring physicians because they can be more easily added to the patient record.
CardiacCT
Electron Beam CT (EBCT) is an established imaging modality for the noninvasive diagnosis of coronary artery disease (see also Chapter 23, Heart). Although cardiac CT is possible to some degree on single-slice or dual detector CT units (calcium scoring), only the introduction of multislice CT opened up cardiac CT for clinical practice. Multislice CT allows for morphologic imaging of the heart, calcium scoring, and coronary CTA. Phase-selective cardiac imaging adds the option offunctional heart studies. A high temporal resolution is needed to freeze cardiac motion and avoid artifacts. To eliminate breathing artifacts the complete heart should be scanned within one breath-hold. ECG synchronization of data is necessary to capture the heart in a (relative) motionless phase. EBCT uses non-mechanic beam rotation with acquisition times between 30 and 100 ms. Multislice CT is based on mechanically rotating CT units that are limited currently to 420-500 ms rotation time, but in the future may provide faster rotation of the order of 300ms. With current fourdetector-row systems with 0.5 s rotation time, a temporal resolution of 100-250 ms can be accomplished, depending on the heart rate, the
20
250-320
pitch, the rotation time and the segmental reconstruction capabilities. Even these systems are at their limit in terms of spatial and temporal resolution. These problems, however, probably will be solved with 16-slice cardiac scanners with faster rotation times and narrower section collimation which may reach a temporal resolution of 50-65 ms, comparable to current EBCT. Prospective ECG Triggering
Prospective ECG-triggering is used for sequential acquisition of four simultaneous sections. Temporal resolution is achieved by a partial scan technique. A prospective trigger is derived from the ECG to initiate scanning at a user-selectable time after the preceding R-wave or before the next R-peak. This delay is defined to coincide with the end-diastolic phase of the heart cycle. It is usually chosen between 40% and 80% of the RR interval. Due to limitations in scan cycle times (table movement), every other heart beat is used for data acquisition (Fig. 1.29). In order to
ms
15 c
o
:p
'is CL
10
N 5
Fig. 1.29 Prospective ECG gating. A partial scan is triggered by the R-wave of the ECG, Alternate heart beats are used for scanning and table incrementation, respectively.
Cardiac CT
capture the heart within one breath hold, in 4slice scanners a 4 x 2-3 mm collimation is employed, depending on the available detector configuration and the scan length, The table feed should be identical to the total collimation width (10-12 mm) or slightly less (6-10 mm) for overlapping sections, With 16-slice units collimation comes down to 1-2 mm with table feeds in the 12-32 mm range, Standard partial scan techniques require acquisition of data from half a rotation plus the fan angle of the x-ray beam, This will result in a temporal resolution of some 320-360 ms for a 0,5s rotation time, Some manufacturers offer optimized reconstruction that is achieved by rebinning the fan beam data to parallel beam geometry and using a reduced field of view of 240 mm or less, This results in an effective scan time of 50-55% of the tube rotation time, yielding a temporal resolution only slightly above 250ms, The exposure time should be less than 40% ofthe RR interval, i.e. the heart rate should be below 96 beats per minute (bpm) for 250 ms exposure time, and below 75 bpm for 320 ms exposure time, Administration of a beta blocker is advisable only ifthe heart rate is too high, and the diastole needs to be prolonged, Almost motion- free images of the heart are obtained up to aheart rate of70 bpm. Since all data are used for image reconstruction, no increase in radiation dose to the patient is necessary as compared to a conventional CT At identical noise levels, dose is similarto or smaller than thatofEBCT In 16-slice scanners, the decreased rotation times of 0.40-0.42 s results in routine temporal resolutions in the order of 80-200 rns, Also, the critical value of the heart rate for optimal temporal resolution has shifted to higher values (8085 bpm) , which is important in clinical scanning,
Retrospective
ECG Gating
In retrospective cardiac gating, a continuous spiral scan is acquired with simultaneous ECG-recording, In order to obtain enough projectional raw data during each part of the cardiac cycle, overs amp ling with a low pitch factor (P= 0,2-0.4) is required, The pitch is adapted to the heart rate, with higher pitch for higher heart rates, The oversampling will necessarily increase radiation exposure to the patient
The user can prospectively or retrospectively select a gate during the RR interval of the ECG for which data for image reconstruction are to be accessed, Gating can be performed with a relative approach (time delay as a fraction of the heart cycle) or with absolute approaches (either a fixed delay after the preceding R-peak or before the next R-peak). More sophisticated approaches take the relative duration of systole and diastole into account Overlapping reconstruction further improves image quality on MPR and 3D images, Retrospective gating reduces the sensitivity to cardiac arrhythmia, Multiphase reconstructions during various time points within the RR-interval allow for phase-selective or functional cardiac imaging, Single Sector Reconstruction
With single sector reconstruction, only data from the prescribed time range during one cardiac cycle are used for partial scan reconstruction of images, This yields a temporal resolution of about 250ms for 0,5s rotation time and 200 ms for 0.4 s rotation time, Images can be reconstructed using no interpolation at all by including only those data from the various detector rows that are closest to the desired plane of reconstruction, This works best for very low slice pitch factors P < 0,25 but, depending on the heart rate, pitch can be increased up to P = 0.4 without major artifacts (see below), Multislice Cardiac Volume Reconstruction (MSCV, Siemens) is an algorithm that combines partial scan reconstruction with multislice spiral weighting, For each projection angle in the multislice data set a two-point linear interpolation is performed between data acquired during the desired heart phase (Fig, 1.30), The z-range can be extended if nearest-neighbour instead of linear interpolation can be used, This algorithm produces a constant relation of the average slice width to the chosen slice collimation: SW -1.25 SC (lJ-L4 SC), Also, retrospective generation of thicker slices from the same dataset with SW= 1.5, 2 or 3 times the chosen collimation SC is possible, resulting in slices with lower noise and better low-contrast resolution at the expense of reduced z-rcsolution. Multisector
Reconstruction
Multisector reconstruction (e.g, ACV, Siemens or Snapshot Burst, GE) can increase the temporal
39
7 Principles orCT, Spiral CT, and Multislice
CT Fig. 1.30 Retrospective ECG gating
by multislice cardiac spiral interpolation. A two-point interpolation be-
15
c
o
:-e
tween corresponding projections acquired during the desired heart phase is performed. Oversampling with low pitch factors is a prerequisite.
10-
Vl
o
CL
N
Time
resolution of the above algorithm by using scan data from more than one heart cycle for image reconstruction (segmented reconstruction, Fig. 1.31a). The partial scan data set then consists of a number of projection sectors and a maximum temporal resolution of RT/2M (M = number of heart cycles used). For a 0.5 s 4slice scanner, a temporal resolution of up to 125ms can be reached with two sectors, and less than 80ms with up to four sectors. The more sectors that are used, the higher the overlap during data acquisition has to be. This requires lower pitches and results in an increased patient dose. The maximum number of sectors is limited by the number of detector rows. Due to phase-coherence between the tube rotation and the heart cycle, multiple sector reconstruction can only lead to improved temporal resolution if the patient's heart rate and the tube rotation are appropriately desynchronized so that the projection angles of the start and endpoints of each subsegment fit together and form a complete partial scan data segment. For data reconstruction using m sectors, the temporal resolution will depend on the heart rate and lie between RT/2 and RT/2 m. Temporal resolution will vary during the scan together with the patienf's heart rate. On a 4-slice scanner, the highest temporal resolution with 2-sector reconstruction is achieved for 65-70 bpm and 90-100 bpm, while 80 bpm leads to synchronization and therefore reduced temporal resolution. Spatial resolution will be decreased if data from two consecutive heart beats do not precisely match. At the same time, the diastole is long enough to allow for 250 ms reconstructions
for heart rates below 65. Thus, heart -rate-sensitive algorithms are being developed by most vendors, which use two or more sectors for higher heart rates and the conventional single sector reconstruction for slow rates (Fig. 1.31b). This gives an optimum compromise between temporal and spatial reconstruction. The relation between SW and SC remains fairly independent of heart rate: SW - 1.3 (± 0.2) SC. The first results with 16-slice scanners indicate that retrospective reconstruction techniques that neglect cone-beam geometry can be applied to the heart since it is sufficiently centered and does not contain very high contrast structures. Due to smaller section collimation (16xO.5-0.75mm) and improved axial sampling there is higher spatial resolution and less geometric artifact in coronary stents. Cone beam corrections, however, are able to further improve image quality. •
Pitch
Continuous volume coverage with single sector reconstruction requires limiting the pitch dependent on the heart rate. For optimum spiral weighting with two interpolation partners for every projection the following restriction applies: Pitch* < (N -1) x RT I (TRR + RToO (R'L-full rotation time: 500ms and RTA= partial scan rotation time" 360 ms). If faster volume coverage is needed and 50% of proj ections are generated with nearestneighbor interpolation, this restriction is reduced to P* < N x RT/TRR. Heart rates of 45-1 00 bpm correspond to pitch factors of P =0.3750.825 (PA 1.5-3.3 for 4-slice scanners) for 0.5s
Cardiac CT 41 Fig. 1.31 Retrospective EGG gating with multisector interpolation. The same phase of more than one heart cycle is used to collect data for interpolation, thus effectively improving the temporal resolution (a). This is only possible if data from each heart cycle provide different projection angles. Thus the temporal resolution varies with the heart rate (b).
125 ms 15
c
o
:e
~
10
Vl
o 0...
N 5
Time
a
600 -
500
Diastole Exposure time
400 Vl
Problematic Heart rates
E
300
/
I
-,
200 125 ms
100 Half scan reconstruction
Heart rate
o~--~--~--~--~--~--~--~--~--~--. 40 100 120 60 80
rotation time. Temporal resolution is in the order of 250 ms. If this type of interpolation is also used for muitisectorretrospective ECG-gating techniques, the pitch should be adapted as follows to the heart rate to achieve full coverage: P*£[l+(N-l)/mJxRT/TRR TRR is the maximum RR interval that is expected during the scan, N is the number of active detector rows, and m is the number of sectors for multisector reconstruction (m /\N) . However, fairly good results with this approach have only been seen with 4-slice scanners, which require a pitch of P= 0.2-0.375 (P *= 0.8-1. 5), depending on the heart rate. Lower pitches result in better image quality but are limited by the scan range that can be covered with the scanner. In practice, covering a 12cm scan range with 4x 1 mm collimation
within 40s on a 0.5s scanner requires a table feed of 1.5mmls (p*= 1.5; P= 0.375). Pitch limitations remain with 16-slice scanners but scanning speed increases substantially. In practice, the table speed (TF/RT, mm/s) is the limiting factor, which means that lower pitch factors have to be used with faster rotation times RT. Best results for 0.5-1 mm collimation are obtained with frequency-dependent multisector reconstruction and a table speed between 4 and 10 mmls, depending on collimation and manufacturer. •
ECG-gated
Reconstruction
with
Extended Coverage
Some clinical applications do not require complete suppression of motion artifacts but profit from reduced pulsation effects. Such applications include imaging of coronary artery
42
7 Principles
act. Spiral ct. and Multislice Acquired
projection
CT
data Half scan reconstruction
Fig. 1.32 Principle of CT fluoroscopy. Half scan reconstruction improves temporal resolution. Temporal overlap of images provides a fluoroscopic effect.
Overlapping half scan reconstruction
1.5
2
Tube revolutions
bypass grafts, cardiac morphology, or imaging of the central vessels. For such applications, ECGgating can be modified to allow faster scan speeds and reduced radiation exposure. This is done by excluding only those phases of the heart cycle from reconstruction that contain most motion artifacts (systole or phase of systolic pulse wave). The excluded time window ATs can be varied but usually some 250ms are used for this propose. The resulting temporal resolution is RT/2. The approach is phase-inconsistent but yields images of substantially better quality than a non-gated approach. The reconstructed image stacks should overlap in the zdirection giving a pitch limitation of: P* ss (N-l) -RTf (RTf2 + ATs). In the chest for ATs = 250ms, the resulting pitch P=0.75 (P'/\3.0) allows for covering 220mm in 36s with 4 x 1 mm collimation. Different values of ATs used for other applications will lead to different preferred pitch values. With 16-slice scanners, the whole chest and abdomen can be covered with such a technique in 16 x 1.5mm resolution. The technique is still a prototype application. •
Radiation
Exposure
Retrospective gating uses only a small portion of the acquired raw data for image reconstruction. As a result, the patient exposure required to obtain a given image quality increases substantially. The dose increases as the time window used for reconstruction (e.g., 125ms) gets shorter relative to the duration of the RR inter-
val (e.g., 750 ms for a heart rate of 80 bpm). This increase is larger for smaller pitch factors. The dose is usually increased by a factor of 3-4 as compared to ECG triggered techniques or nontriggered multislice scanning. Various dose reduction techniques are presently being developed by the manufacturers. By positioning the heart close to the center of the scan field (patient slightly shifted to the right of the examination table), the scan angle may be reduced, thus excluding the more peripheral portions of the chest. Such a technique will cause artifacts close to the periphery of the exposed scan field (see Fig. 7.45c) but these artifacts only involve the very periphery of this region and can be excluded from the display field of view. By choosing a smaller field of view « 24ern) centered on the heart such artifacts are no longer visible but the radiation dose to the skin and breasts is substantially reduced. ECG-controlled modulation ofthe tube output (ECG-pulsing) holds substantial potential for further dose reduction. With this technique, the rnA is substantially (80%) decreased during the (systolic) phases of the heart cycle that are usually not used for image reconstruction (see Fig. 23.15). Functional imaging, however, still remains possible because image quality generally suffices for volumetric assessment of the ventricles. Other new dose-reduction developments include edge-preserving image filtering and low-kvp scanning. In edge-preserving image filtering special postprocessing filters can reduce image noise while anatomical edges are pre-
CTF/uoroscopy
served. The technique may be' used to lower radiation exposure or to improve low-contrast resolution, alone or in combination with ECGpulsing. Scanning at 80 to 100kVp can increase contrast-enhancement in coronary CTA and coronary calcium density at the expense of a little increase in noise. The resulting dose reduction can be as high as 30%, while contrast-to-
noise ratios will be preserved. Reducing the scan field of view can be used to decrease the dose to outside the region of interest (heart). Similar effects can be generated by special bowtie filters placed behind the x-ray tube that reduce the radiation dose to the periphery of the body outside the region of the heart.
CT Fluoroscopy CT fluoroscopy is a technique that produces constant updates of the scanned data and therefore provides fluoroscopic cross-sectional images.
Principle and Image Reconstruction To gain a high temporal resolution despite amuch longer rotation time of the x-ray tube, images are generated in a temporarily overlapping fashion. For this purpose, partial scanning is employed. Data from onlyslightlymore thanhalfatube rotationare reconstructed (180 + fan angle). In addition, the 360 rotation is subdivided into sectors, and after the tube has scanned a new sector, the old one is dropped from the image and the new one is incorporated into the image. To speed up reconstruction, each sector may be reconstructed separately by filtered backprojection, and the final image is assembled simply by superimposing each sector. High refresh rates, up to 8 ormore images per second, are possible with this technique (Fig. 1.32). 0
0
Technique Fluoroscopy is employed for real-time monitoring of interventional procedures. A foot switch or table switch together with a monitor close to the gantry are necessary in order to follow the procedure. A moving table helps to find the region of interest or the needle tip interactively. However, it can sometimes be quite cumbersome to reposition the table if the needle leaves the scan plane because it is not always evident in which direction it does so. CT fl uoro scopyusing am ultislice scanner can be used to acquire multiple sections simultaneously and to see more easily in which direc-
tion a needle is deviating from the prescribed path. Some vendors offer the option to fuse the inner two sections (target plane), and simultaneously display and update the sections above and below this target plane (see Fig. 6.3). This way it is much easier to determine in which direction the needle or table have to be repositioned.
Radiation Safety When identical mAs settings are used as for diagnostic scanning, radiation dose to the scanned section increases substantially. As compared to a 1 s conventional section, a lOs fluoroscopy will cause 10 times the radiation exposure. For this reason, fluoroscopy always employs low mA settings, and-if possible-also low kVp settings. Too low an exposure, however, will substantially decrease image quality and may hamper the detection of the target lesion or a safe path towards the lesion. For this reason, the dose has to be adapted to the patient size (higher kVp settings in obese patients) and the organ region (higher mAs for the abdomen than for the lungs). Still, radiation dose in fluoroscopy is usually substantially higher than in a conventional CT-guided biopsy. Usually conventional step-and-shoot techniques may be almost as fast as fluoroscopyguided procedures, provided that there is a foot switch and display monitor in the examination room. Fluoroscopy should be reserved for difficult cases, in which conventional techniques are less safe or substantially slower (e.g., pulmonary nodules close to the diaphragm, difficult approaches in the abdomen). Protective gloves and special instrumentation (e.g., needle holders) should be used that protect the hands of the examiner from the radiation beam.
2 Image Processing and Display Techniques M. Prokop
"Data Volume
46
Cine Viewing
48
Multiplanar Reformations
__________________________ __________________________
Principle (p. 48) Artifacts (p. 51) Applications (p. 52) Image Analysis and Documentation
48
(p. 53)
Maximum Intensity Projection (MIP) and Minimum Intensity Projection (MinIP)
53
Principle (p.53) Thin-Slab MIP (p. 57) Minimum Intensity Projections (MinIP) (p. 57) Artifacts and Pitfalls (p. 59) Applications of MIP and MnlP (p. 60) 3D Surface
Rendering (Shaded Surface Display)
60
Principle (p. 60) Applications (p. 63) Volume Rendering Techniques
63
Principle (p. 63) Special Techniques (p. 67) Artifacts and Pitfalls (p. 67) Applications (p. 69) Segmentation Principle (p. 72) Automated Techniques, Applications (p. 74)
•... Computer-assisted
Vinual Endoscopy Principle (p. 78) Alternative Visualization Techniques Artifacts and Pitfalls (p. 81) Applications (p. 81)
.••...
72
Diagnosis (p. 74)
78 (p. 79)
46
2 Image Processing
and Display Techniques
Computed tomography is primarily a crosssectional imaging technique. Even with conventional scanning. a three-dimensional body region is mapped by CT but information along the patient axis (z-axis) is only available in discrete steps. With the advent of spiral and multislice scanning. CT has moved from a transaxial to a truly volumetric imaging tool. All techniques of two-dimensional (2D) and three-dimensional (3D) image processing rely on a set of individual axial sections acquired by CT scanning. These sections are stacked together in the computer to form a data volume that can be further processed and manipulated. Spiral and m ultislice CT have revolutionized data acquisition. With proper selection ofthe scan parameters. 2D and 3D images of exceptional quality can be generated. This contrasts with the relatively poor quality of many 2D and 3D reformatted images in conventional CT. With multislice CT data sets. near isotropic resolution becomes standard but images suffer from increased image noise. For this reason. new noise-suppression techniques have to be developed to take full advantage of the improved spatial resolution. There is a multitude of 2D and 3D processing techniques which are summarized in Table 2.1. Multiplanar reformations (MPR) are the most important 2D tools to create images in arbitrary. even curved planes through the volume. Maximum intensity projections (MIP) are popular for display of CT angiographic data sets. Shaded surface displays (SSD) have been used to visualize complex vascular anatomy and
Data •
Table 2.1
Overview
of selected
2D and 3D processing
tools 2D Display
Tools
Cine displays Multiplanar reformation Thick MPR Curved planar reformation Vessel
tracking
3D Display
Tools
Maximum intensfy projection Minimum intensity projection Shaded surface display Volume rendering technique S Tissue transition prqjection Raysum
MIP minlP/mlP
SSD VRT TTP
projection
Segmentation
Tools
Cutting functions Region growing Dilation/erosion Closing functions Removal of flying Watershed Automated Automated
MPR CPR
(removal of holes) pixels (floaters)
algorithms bone removal volume analysis
skeletal structures. Volume rendering techniques (VRT) are the most versatile rendering methods that allow for transparency as well as 3D effects. Interactive manipulation of data sets becomes more and more important the larger the data volumes become and the more complex and subtle the diagnostic questions to be answered.
Volume
Voxel-based vs. Grid-based Models
In conventional CT the data volume is subdivided into volume elements (voxels) in which the voxel cross-section (in the axial x-y-plane) equals the pixel cross-section. while the voxel height (along the z-axis) is defined by the section collimation (Fig. 2.1 a). With spiral CT. and with some conventional CT protocols as well. overlap of sections becomes possible. In addition. the section sensitivity profile is bellshaped. and a voxel may actually contain information from adjacent voxels. This makes a voxel-based model of 3D image data sets hard to handle: voxels no longer have sharply defined
upper and lower boundaries (due to the section profile). and voxels overlap substantially along the z-axis. Thus. a grid-based model is more helpful: here the center of each former voxel is used to define a 3D grid. The spacing along the z-axis then corresponds to the reconstruction increment. and the spacing in the x-y-plane corresponds to the pixel size (Fig. 2.1 b). Each grid intersection is assigned the CT number of the corresponding pixel in an axial CT image. 3D Sampling Function
The CT number samples information from the surrounding volume according to a sampling
Oata Volume
/
~
/ /
V ,
1
,c:. f---:,
V
r:'7 c---
V
'-- 1/
c--; V
/
--;
/
0
• <> •
-
--
/
/
V
/ /"
I
"c n plane
I
I
I
/
~
:;;<>-
--ol C---ol
•
e
•
I i I'
qe nter of·. i : ..
/
._...,
•
~Tt~~ '~.
-./
<:t
1
/
/
---' V -
•
0-
0
1--
I
I
1
• •• 0
q
.~
e
Y
V
Non-overlapping sections SW = 3mm, RI = 3mm
/ /
.'
1/
/
/
/~
Overlapping sections SW = 3mm, RI = 1.5mm
Grid of center points superimposed over voxels
Grid-based model with sampling function (PSF)
Fig. 2.1 Representations of a 3D data volume. The voxel-based model relies on box-shaped voxels (a) but has a problem dealing with overlapping image reconstruction and non-rectangular section profiles. A grid-
based model (b) assigns a node to the center of each voxel. Each node samples information along the z-axis (according to the section sensitivity profile) and within the scan plane (according to the point spread function).
function defined by the section profile in the zdirection and by the point spread function in the x-y-plane. This sampling function can be directly visualized by scanning a tiny lead bead and looking at the resulting image in the x-yplane anel the x-z- or y-z-plane (Fig. 2.2). As can be seen, the sampling function is not a voxel-
shaped box but instead an ellipsoid with a short axis that is determined by the spatial resolution in the x-y-plane (i.e., by the convolution kernel) and a long axis that is determined by the section profile. Fig. 2.2 b demonstrates that the sampling function is spherical for 1 mrn collimation, i.e., there is no distortion along the z-axis with
a
b
Fig.2.2 Sampling function (point spread function). The information sampled by each grid point in the data volume can be visualized by scanning a tiny lead bead (represented as a red dot in these images). Note that there is already blurring within the scan plane (x-y-
c plane) (a). Longitudinal distortion occurs along the zaxis for SC/TF = 3/4 (b). Resolution is isotropic with SCfTF = 1/1. The distortion effects along the z-axis are identical to those in the x-y-plane (c).
47
•
2 Image Processing and Display Techniques
the results that there is isotropic resolution with such a scanning protocol. Note, however, that results change if a different convolution kernel (e.g.) a high resolution algorithm is used, Secondary
Raw Data Set
The more isotropic the data set, the better the resulting image quality of 2D and 3D reconstructions, For multislice reconstruction of a
cr
secondary raw data set consisting of thin overlapping axial sections forms the basis of nearisotropic imaging, Image noise, however, may pose a problem with very thin sections, Increased noise may introduce artifacts in various rendering protocols and therefore should be kept low by using a sufficient radiation dose, a smoothing convolution kernel, or by applying noise-reducing prefiltering techniques,
Cine Viewing Cine viewing is an excellent tool for evaluating large numbers of sections, be they axial or multiplanar. It is important that the user has full control of the direction and speed of such cine displays, Such mouse- or trackball-controlled scrolling can be further improved if the system allows for fast scrolling through large ranges as well as finely controlled scrolling though just a few sections of interest It has been demonstrated that interactive cine viewing speeds up the evaluation of large data sets, allows for more accurate assessment of complex structures that cross the sectional plane multiple times (e.g. vessels or bowel loops), improves lesion detection (e.g. in the lungs or the liver), and increases productivity in picture archiving and communication systems
lanar
Pitfalls
In the initial phase, when moving from film reading to interactive cine viewing, there may be an increased risk of missing findings, The reason is that film reading is more patternbased than interactive viewing: any finding that lies outside the normal will be detected by an experienced viewer, even if it is located in usual sites (e.g, , cutaneous metastases), With cine viewing, the radiologist has to evaluate each organ system at a time, which means that many radiologists will have to change their normal search pattern.
Reformations
Principle Multiplanar reformations (MPRs) are two-dimensional reformatted images that are reconstructed secondarily in arbitrary planes from the stack of axial image data (Fig, 2,3), Coronal or sagittal reformations are generated by extracting and displaying only those voxels from the data volume that are positioned one above the other within the coronal or sagittal plane, Oblique or curved reformations are constructed in an analogous fashion, but the image data must be interpolated between adjacent voxels. Defining
(PACS) environments, Cine viewing becomes the primary mode of evaluation with multislice CT because of the large number of images to be viewed,
the Plane of Image Reformation
The image plane for a MPR is defined interactively on a workstation using a suitable refer-
ence image, which may consist of an axial image, another MPR, or a 3D image such as a maximum intensity projection (MIP) or a shaded surface display (SSD), By drawing a "cut line" on the reference image, the radiologist defines a section that is perpendicular to the reference image, The cut line may be drawn in any direction (orthogonal, oblique, or curved), Curved-planar reformations (CPR) are generally needed to depict structures that pass through multiple axial planes of section (e.g, a bronchus or blood vessel), In these cases it is necessary to have a workstation that allows the user to define the reformation plane on multiple reference images because the structure of interest usually leaves and enters the reference images repeatedly, For depicting tubular struc-
MPR-Principle
coronal sagitta
Fig.2.3 Principle of multiplanar reformation The images are reconstructed from a "stack" slices.
(MPR). of axial
tures, software is currently available that provides semiautomatic positioning of the cut line and keeps the line centered on the structure of interest (Fig. 2.4). Ideally only a few reference points have to be dropped on the structure of interest (usually a vessel), and the software should automatically generate a centered path through these regions {vessel tracking}. •
Thick MPR
As a rule, MPRs have a width equal to one voxel. Averaging a series of adjacent data points perpendicular to the reformatted plane can produce reformations with a greater section thickness (thick MPR). This is advantageous because it leads to reduced image noise and improved image quality (Fig. 2.5). This is an efficient tech-
nique to keep image noise low without losing in-plane spatial resolution, especially with thinsection multislice CT data sets. As with MRI, high-quality sections of arbitrary angulation can be obtained with a section width of 1-7 mm, depending on the clinical indication. The effect of noise-reduction by creating thicker sections is direction-dependent. Along the z-axis, image noise is highly correlated: even with highly overlapping reconstruction, the CT numbers on adjacent sections will not change much if a large section collimation was used. This will result in a suboptimal noise suppression along the z-axis when thick axial MPR are reconstructed (Fig. 2.6). In the axial (x-y) plane, image noise is hardly correlated between adjacent pixels unless very small fields-of-view have been used for image reconstruction. Thus, CT numbers in adjacent pixels change due to variations in noise. On MPRs perpendicular to the axial plane, noise therefore cancels out rapidly if multiple pixels are averaged and thicker MPRs are created (Fig.2.6c). With thin collimation, however, the difference between axial and coronal or sagittal reformations becomes smaller because correlation of noise along the z-axis decreases substantially. In summary, the image quality in coronal or sagittal reformations (perpendicular to the axial scan plane) can be greatly improved by increasing the section thickness to several millimeters. Thick axial reformations (parallel to the scan plane) can only be recommended when the CT data has been acquired with thin sections. Thick axial reformations are therefore mainly used in multislice CT. Note that excessive section thickness in the MPR leads to troublesome partial volume effects.
Fig. 2.4 Vessel tracking uses seed points that are dropped on a vessel (e.g., using MPR or MIP) to find the center of the vascular structure that connects these points (a). The resulting image can then be rotated around this centerline, allowing for detailed analysis of the vessel wall (b).
a
b
50
2 Image Processing and Display Techniques
a Fig. 2.5 If image noise is high (especially for thin-sectioruimaqinq) .the quality of MPRcan be improved by increasing the width of the section: (a) 1 pixel wide (0.6mm). (b) 5 pixels wide (3 mm) coronal sections from a 4 x 1 mm multislice CT data set. Note that low-
a
b
contrast resolution improves and the gastric folds can be better delineated (arrow) while the partial volume effects through the imaging plane also increase and small detail may vanish (arrow head).
c
Fig. 2.6 Thick axial MPR. Because section width (zaxis) is usually larger then the pixel size (x-y-plane). axial sections have to be thicker than coronal sections to efficiently reduce image noise unless very thin collima-
tion has been used. Primary 3 mm thick axial section from a 4x 2.5 mm multislice CT data set (a). Image noise on a 7.5 mm thick axial MPR (b), is identical to noise on a 3 mm thick coronal MPR (c).
•
Summed projections from an imaging volume in which the bony structures were removed result in DSA-like images. This effect can be enhanced if a threshold value above the softtissue range (e.g., > 80-100 HU) is used as well.
Ray-Sum Projection
For a ray-sum projection, the CT numbers encountered in the viewing direction are averaged or added together. Thus, a simple ray-sum projection is identical to a (very) thick MPR. The result simulates a conventional radiograph of the selected imaging volume (Fig. 2.7). Ifa threshold range is preassigned, only voxels with CT numbers within that range are averaged.
MPR-Arti facts 51
Fig. 2.7 Ray-sum projection (SO mm thick MPR) of the C02-filled colon (CT colonography) in a patient with Crohn's disease.
Artifacts
The quality of MPR images is best in imaging planes that deviate little from the scan plane. At the same time, the resolution along the z-axis affects all sections that are perpendicular to the scan plane. The smaller the section width and
the smaller the reconstruction interval, the better the image quality (Fig. 2.8). Section-by-section scanning and large section thickness in conventional CT generally lead to stair-step artifacts in reconstructions that are perpendicular to the axial plane. Spiral or multislice CT are much less vulnerable to such step artifacts because of overlapping image reconstruction. Good-quality MPRs require an overlap of some 30-50% of the section width. Thin collimation (for scanning) and thin section width (for data reconstruction) yield reformatted images of outstanding quality in any plane of section desired. Thicker collimation leads to loss of sharpness along the z-axis (Fig. 2.8). For objects outside the center of the gantry or interfaces that are oriented obliquely but almost parallel to the scan plane (such as the calvaria, tibia plateau or wrist bones) step artifacts and serrations of contours may also occur with spiral or multislice CT data sets if larger pitch factors are used (Fig. 2.9). These artifacts can be caused by the data interpolation process and by undersampling of data in z-direction.
Fig. 2.S Using thinner collimation (e.g.,4x1 mm instead of 4 x 2.5 mm) improves the quality of reformatted images, (a) SW/RI=3/2.5, (b)SW/RI =1.25/ 0.7. Note the improved delineation ofthe choledochal wall.
a
b
a
b
Fig. 2.9 Step artifacts and serration artifacts occur in multislice CT data sets at high-contrast interfaces oblique to the scan plane if higher pitch factors are used. MPR of the wrist bones scanned with 2 x 0.5 mm collimation at a pitch P= 1.5 (a) and ata pitch ofO.S (b).
2 Image Processing and Display Techniques
Applications
Multiplanar reformations are seldom worthwhile in conventional CT, because generally the image quality is too poor to yield substantial additional information. Spiral and multislice CT, however, open up many new potential applications for multiplanar imaging. The fact that MPRs have been underutilized in the past was due primarily to a lack of user-friendly software. •
Anatomically
Corrected
Data Sets
MPRs can compensate for faulty positioning and can provide symmetrical images for improved side-to-side comparisons (of the inner ear, neck, pelvis or shoulders, for example). Especially for coronal reformations, the MPR should be positioned parallel to the most important anatomic structures (e.g., the posterior ribs or the pelvis) to improve anatomical orientation. Tilted MPR may be helpful for the chest (parallel to the trachea or parallel to the sternum) or for the abdomen (parallel to the abdominal aorta). Appropriate tilting will ensure that important anatomic detail are sectioned longitudinally. In general, similar cut planes can be used as in MRI. •
Problem-solving
MPRs can be used as a problem-solving tool in cases where it is necessary to trace pathologic structures through multiple planes and in cases
a Fig. 2.10 Examples of applications of MPR: (a) The relation of this tumor recurrence to the muscles of pelvic floor is better appreciated on the coronal MPR.
that require imaging in a second plane. In spiral CT this applies consistently to skeletal investigations, frequently to CT angiography, and occasionally to investigations of tumors involving the liver, kidneys, or lower abdominal organs (Fig.2.10a). With thin-section multislice CT the role ofMPR is vastly increased. MPRs are important for defining the relation of focal lung lesions to the pulmonary fissures, chest wall and mediastinal structures, for assessing lymph nodes in all three planes, for defining tumor infiltration into adjacent structures, and for evaluating the bowel and structures of the small pelvis. Curved planar reformations are essential for delineation of soft plaques in CTA (see Fig. 2.4b) , evaluation of the pancreatic duct in one image (Fig.2.1Cb), or for the display and differentiation of ureteral stones from paraureteral calcifications. •
Noise Reduction
in Standard
Displays
Thick MPRs can be employed as the most important technique for noise reduction when thin collimation has been used to acquire a data set (see Fig. 2.5). The technique can be applied to spiral CT as well as to multislice CT data. Thick axial MPR can substitute for reconstructing thick axial sections directly from the raw data set. Thick MPR are a way to increase image quality substantially in low-dose applications of multislice CT (see Figs. 5.16-5.18).
b (b) Excellent overview ofthe pancreatic duct in a patient with chronic pancreatitis and stents in the pancreatic and common bile ducts.
MIPIMinlP-Principle
Image Analysis and Documentation
When MPR is used selectively to answer a question that cannot be resolved in the axial plane, it is usually best to perform the reformation interactively on a workstation. Software should be used that provides a "soft" transition between the reformatted sections when the imaging plane is changed. Abrupt transitions may cause lesions to be missed. In situations where it is important to document one or more additional planes, such as sagittal and coronal reformatted images of the calcaneus, a standard protocol can be established for computing and documenting the sections. To ensure the best image quality, the width of the MPR should be appropriate for the region under study (Table 2.2).
Table 2.2 Recommended section width (SW) of axial, coronal, and sagittal reformations Suggested width (SW) of MPR (mm)
I:Region
Neck Routine Larynx Chest Routine Bronchial system Peripheral tumors Chest wall Mediastinum
1.5/10 1.5 3
1.5 5
3
5
3-4 3 3-4 3-4 3-4
3-5 3-5 3-5 3-5 3-5
1
1 3 3 1 0.75
Abdomen 1 1
I 1
I. 1
Routine Liver (tumors):' Pancreas (tumors)" Kidne¥s (turners)" Bowel Small pelvis"
c
iMusculoskeletal system Cervical spine Thoracolumbar spine" I' Pelvis3 I., Foot Ifc Hand!
I
2 2
0.75 0.75
I ® depending on patient size
Maximum Intensity Projection (MIP)/Minimum Principle
Maximum intensity projection (MIP) and minimum intensity projection (MinIP or mIP) are volume-rendering techniques in which suitable editing methods are used to define the volume of interest (VOl). All of the CT image data set may be used, or the volume may be confined to a region of interest. In the most difficult case, only selected organ systems are included or excluded from the VOL The actual images are generated by projecting the volume of interest into a viewing plane and displaying the maximum CT numbers (for MIP) or the minimum CT numbers (for MinIP) that are encountered along the direction of the proj ection, called the viewing angle (Fig. 2.11). Both techniques ensure that optimum contrast
Intensity Projection (MinIP)
is produced between small, high-contrast structures and surrounding tissues. MIP views are used for CT angiography and for specialized pulmonary studies, while MinIP views are used mainly for visualizing the central tracheobronchial system. The following discussion of the use ofMIP in CT angiography also applies to other MIP applications and to MinIP views. •
Attenuation
Information
MIP preserves the attenuation information of structures with maximum CT numbers, so an anteroposterior viewing angle in most body regions would display skeletal structures rather than contrast-enhanced vessels. Thus, unlike MR angiography, some view angles in CTA make
53
54
2 Image Processing and Display Techniques Fig. 2.11 Principle of the maximum intensity projection (MIP). This technique displays the maximum CT numbers that are encountered along the view angle of the projection. Vascular imaging requires editing of the data volume to eliminate bony structures.
Projected VOl
Viewing direction
it necessary to edi t bony structures out of the CT image. This editing process may be done manually, semiautomatically, or automatically, depending on available workstation software (see Segmentation, p. 73).
a Fig. 2.12 MIP of the abdominal aorta, (a) When the projected volume encompasses the whole body, all vessels are superimposed over each other. Calcifications are well displayed, but thrombi are not directly visualized. There are linear artifacts in the periphery of the image (due to variations in image noise related to spiral
A major advantage of the MIP view is that contrast -enhanced vessels can be distinguished from wall calcifications by their different attenuations (Fig.2.12a). Even small vessels (1 mm or less in diameter) will be visualized as
b interpolation), (b) Reducing the width of the VOl (curved thin slab MIP) improves the contrast between small vessels and the background. Superimposing mesenteric vessels are removed and the thrombus becomes visible. Note that there are small bilateral renal artery aneurysms.
M/P-Pr'incip/e
long as they have a higher CT number than their surroundings within the VOL •
Image Background
The image background depends on the voxels with the highest CT numbers in the structures surrounding the vessels. If the surroundings are homogeneous, the MIP will display voxels that have the highest CT numbers based on statistical fluctuations (i.e., image noise). Because of this, an increase in image noise in the original data set is associated with an increase in background attenuation in MIP images. Arterial vessels generally have heterogeneous surroundings consisting of fat, unenhanced soft tissues, contrast -enhanced organs, and venous vessels. As the CT number of the surrounding structures increases, so does the background attenuation. It is therefore important to exclude as much contrast -enhancing surrounding structures from the VOl as possible in order to optimize vessel-to-background contrast (Fig.2.12). •
a
b
c
Image Contrast
Image contrast in MIP images is determined both by the CT numbers of the vessels of interest and by the difference between these CT numbers and the background attenuation. Accordingly, MIP image contrast increases with the level of intravascular contrast, and it decreases as a result of partial volume effects (which reduce the attenuation of small in-plane vessels) and as a result of higher background attenuation. For the evaluation of small blood vessels such as the renal arteries, the scans should be performed with a thin collimation, optimum intravascular contrast should be achieved, and background attenuation should be kept as low as possible. Background attenuation can be minimized by selecting a thin VOl in the direction of the projection and by eliminating all higher-attenuating superimposed structures such as contrast -enhanced organs or venous vessels by proper segmentation techniques.
Fig. 2.13 Image noise on MIP. (a) Axial section after bone removal in a very obese patient, (b) Noise in small vessels remains unchanged but noise in the aorta and the background is reduced on MIP images (here, 1 cm slab width), (c) Background attenuation on MIP images increases and the background noise decreases as the projected VOl becomes wider (here, 5 cm slab width)
IFor optimum image quality in the MIP (orl /\MinlP) view, the volume of interest (VOl) should be as thin as possible (Figs. 2.12 and g2.13). •
Viewing Angle
Given the different degrees of spatial resolution within the scan plane and along the z-axis, better image quality is obtained with axial viewing angles (i.e., nearly parallel to the z-axis) than with perpendicular view angles (e.g., anteroposterior or lateral).
2 Image Processing and Display Techniques
Where feasible, it is advantageous to render the MIP in a craniocaudal or slightly oblique direction. For evaluation of the renal arteries, the projection can be limited to the few sections that actually contain the vessels. An oblique projection can separate the aorta from the vertebral column, thereby eliminating superimposed structures and avoiding the need for a more complex segmentation process (see Fig.24.75a). •
Cine Loop
A single MIP conveys only two-dimensional information; it does not encode depth relationships. Consequently the foreground and background are not differentiated (Fig.2.12a), and a superimposed structure can obscure a less attenuating (usually smaller) vessel. Generating multiple images at constant angular intervals and viewing the images from various perspectives in a cine loop can improve 3D orientation. •
Image Noise
Image noise in the axial data set will propagate to MIP images in a more complex fashion than noise in thick MPR. In a homogenous region, noise leads to random deviation of CT numbers from the "real" CT attenuation is this region. MIPs display the pixels with maximum CT num-
a
b
Fig. 2.14 Noise-reduced MIP. High image noise (due to an high resolution filter kernel) on the MPR (a) leads to MIP of suboptimum quality with losss of bony detail (c).
bers, i.e., those pixels with maximum deviation from the real CT numbers. Because of noise, the CT numbers on MIP images of a homogenous region grow, the wider this region is (and thus, the more pixels are traversed). The bell-shaped histogram of CT numbers in a homogenous region indicates that the probability of encountering a higher CT number decreases the more this number deviates from the real attenuation. Thus, the wider the homogenous region that is projected, the smaller the resulting image noise on a MIP. If only a small homogenous region (such as a small vessel) is projected, image noise is largely unaffected by MIP. • Noise in small vessels remains unchanged I but background noise is reduced on MIP images (Fig. 2.13b). • Background attenuation on MIP images increases and the background noise decreases as the projected VOl becomes wider (Fig. 2.13c). • Contrast between small vessels and the background is improved if a thin slab is used (Fig. 2.12). Only if noise on the original image is excessive, such as for low dose applications of multislice CT, image noise on MIP images remains high
c
d
Reconstructing 3mm thick coronal sections (b) and using those for MIP substantially improves image quality (d).
Minimum Intensity Projections (MinIP)
and may obscure diagnostically relevant information. In such a situation, we suggest improving image quality by creating an overlapping set of thick MPR first (in the same orientation as the final MIP; e.g., coronal MPR for anterior-posterior MIP), and then reconstructing an MIP from this noise-reduced data set. Such a technique (noise-reduced MIP, nr- MIP) is excellently suited for large vessels but may reduce the contrast of small structures (Fig. 2.14). Thin-Slab MIP
As noted above, a thin VOl can improve the quality of the MIP image. In a technique called thin-slab MIP, this is accomplished by dividing the imaging volume into multiple thinner subvolumes, called "slabs," that have a specified thickness ranging from a few millimeters to several centimeters. A maximum intensity projection is then performed on each slab. An axial view angle is preferred with spiral CT because of better spatial resolution. In multislice CT, thinslab MIPs in arbitrary planes become feasible. By generating overlapping thin-slab MIPs and surveying the images interactively [sliding thin-slab MIP) one can improve spatial orientation and can cover the whole scan range. Comparing a thin-slab MIP view of designated thickness with an axial section or MPR of the same thickness, we note that the spatial resolution is identical. Small structures of high contrast (e.g., pulmonary vessels) decrease in contrast with increasing width of the axial section or MPR as a result of partial volume effects, but the contrast on MIP images remains fairly constant regardless of the width of the slab used for the MIP (Fig. 2.15). Minimum Intensity Projections (MinIP)
The imaging characteristics and artifact behavior of minimum intensity projections (MinIPs) are analogous to those of MIPs. They will be discussed for imaging of the tracheobronchial system. As a rule, only the central portions of the tracheobronchial system can be evaluated with this technique.
b
a
Fig.2.15 Comparison of a thin slab MIP (a) and a thick CPR (b) each of 10 mm width. Note the better contrast with MIP and the lowernoisewith MPR.
•
Image Contrast,
Noise and VOl
The greater the attenuation difference between air- filled bronchi and surrounding tissues, the more clearly the bronchi will be visualized. While MinIPs will clearly define bronchi that are surrounded by structures of soft -tissue density (mediastinum or consolidated lung), they will rarely delineate bronchi that are surrounded by aerated lung parenchyma or emphysematous lung (Figs. 2.16 and 2.17). MinIP created from high resolution data sets will suffer from the increased noise, which will cause a reduction in background density (similar to the increase in background density with MIP) and will obscure bronchi. This effect increases with wider VOl (Fig. 2.16) As a rule, intrapulmonary bronchi or bronchi obscured by superimposed lung parenchyma are depicted only if they have an attenuation value of approximately -1000 HU. The attenuation values of smaller bronchi are increased due to partial volume effects. Especially when oriented roughly parallel to the scan plane, small and even medium-size bronchi will not be visualized in sections that are more than 3 mm thick. The thinner the sections, the better the image quality.
57
2 Image Processing and Display Techniques
a
c
b
Fig. 2.16 Minimum intensity projection (MinIP). (a) A narrow VOl (1 cm) in this patient with pulmonary fibrosis and centrilobular emphysema displays the central bronchi and the distribution of the small bullae to good advantage. Note that the fibrotic changes cannot be securely detected, (b) Increasing the width to 3cm sub-
a
b
Fig. 2.17 Comparison of MPR and MinlP in a patient with a tracheal diverticulum (arrow), multiple tracheobronchial polyps (arrowheads), mucus in the trachea (wide arrow) and marked emphysema, (a) 1 pixel thick MPR (b) 10mm thick MPR (c) 10mm thick MinlP, (d)
•
Imaging Volume and Viewing Angle
A basic concern in the segmentation ofMinlPs is to eliminate superimposing air from the imaging volume. This air may be extracorporeal or be any air present in soft tissues or in a pneumothorax. Otherwise the extremely low attenuation value of the air will obscure the bronchi.
stantially decreases the distinction of bullae and the display of the upper lobe bronchi, (c) Using a noisy highresolution data set for MinlPdecreases background density and makes evaluation of the lung parenchyma virtually impossible.
c
d 3 cm thick MiniP. Note that the polyps and the mucus are easily missed on MinlP, and that the emphysema makes evaluation of more peripheral bronchi impqssible.
When an anteroposterior viewing angle is used, the imaging volume should be selected so that the central portions of the tracheobronchial system are displayed while the VOl is kept as thin as possible (Figs. 2.16 and 2.17). An attempt should be made to include a minimal amount of paramediastinal lung tissue in the VOl, as this would cause an unacceptable reduction in bronchial contrast.
MIPIMinlP-Artifacts and Pitfalls
When a craniocaudal (axial) viewing angle is used, a somewhat broader imaging volume (2-5 ern) can be selected. Any greater thickness of the VOl would generally lead to excessive superimposition of bronchial structures. A lateral viewing angle is rarely feasible. Artifacts and Pitfalls •
CT Angiography
If the selected VOl is too wide or if surrounding structures of higher attenuation are present, smaller vessels with CT numbers less than or equal to the background attenuation may be missed. This is often a result of partial volume effects or insufficient arterial contrast. Intravascular lesions such as mural thrombi or soft plaques often cannot be directly visualized with MIP. However, partial volume effects between a soft plaque and the opacified vessel lumen can cause an apparent decrease in luminal attenuation in the affected region. Intimal flaps due to dissection can be detected only if they are precisely parallel to the MIP viewing angle (Fig. 2.18a, b). If the true lumen and false lumen show different degrees of contrast enhancement, the width of the higher attenuating lumen will generally be overestimated unless the projection is parallel to the intimal flap. Calcifications have a higher attenuation and therefore superimpose on vascular structures. They are optimally detected on MIP images. At the same time it may become impossible to estimate the degree of a stenosis caused by a hard plaque (Figs. 2.18c.d). For optimum evaluation
of stenoses caused by hard plaques, curved planar reformations are necessary. •
Pulmonary MinlP and MIP
Contour irregularities can be appreciated only when they are defined in a profile view. Thus, minimum intensity projections tend to underestimate the extent of endobronchial lesions, or miss them entirely if they are small, due to the obscuring effect of superimposed endobronchial air (Fig. 2.17). Consequently, MinIPs are not useful for the detection of intraluminal tumors. Vascular pulsations transmitted to the lung tissue and respiratory artifacts can cause areas of low attenuation to appear at the point of vascular branching. This creates multiple focal hypoattenuating lesions on MinIP images, especially in the retrocardiac space, that may simulate emphysema. Pulsation of the retrocardiac vessels can lead to artifactual beading of the vessels on pulmonary MIP, even when individual sections are reconstructed in an overlapping fashion. When relatively thick slabs are used for pulmonary MIP, there may be rare cases in which small structures (e.g., pulmonary nodules) are obscured by larger ones (e.g., pulmonary vessels). If the selected imaging volume is too broad, MinIP can only visualize the central potions of the bronchial system (Fig.2.17 d). Failure to eliminate extracorporeal air from the imaging volume results in "black images".
Fig.2.18 Artifacts in CT angiography. (a, b) Intimal flaps in aortic dissection are only displayed when the flap is parallel to the projecting direction, (c. d) Calcifications may superimpose on vascular structures and can make the evaluation of stenoses impossible. Compare MIP (c) to CPR (d).
a
b
c
d
59
2 Image Processing and Display Techniques
a
~ b
Fig. 2.19 Sliding thin-slab MIP for nodule detection. Compare (a) 1.25 mm thick axial section, (b) 5 mm thick axial section, and (c) 5 mm thick MIP. Note that the high
Applications of MIP and MinlP
MIPs are mainly used to present findings of CT angiography. They are best suited for visualizing abdominal arteries (see Fig. 2.12) and the pelvic and lower limb vessels. MIPs are useful for thoracic studies only if the anatomic relationships in the imaging volume are relatively simple. MIPs are not recommended for the evaluation of complex vascular malformations, aortic dissections, central pulmonary emboli, or free- floating clots. Thin-slab MIPs are useful for imaging small pulmonary lesions in cases where thin collimation has been used for data acquisition. They improve the detection of small pulmonary nodules (Fig. 2.19). Diffuse lung diseases can be detected more accurately and at an earlier stage
130 Surface
c contrast of the 1.25 mm thick section is retained in the MIP while differentiation between vessels and nodules is improved.
than with other methods. The images combine the high contrast of a thin section with the superior anatomic orientation of a thicker section. Minimum intensity projections are used in evaluations of the central tracheobronchial system. They may be useful for the localization ofextrabronchial air collections or bronchial abnormalities (Fig. 2.17) and for demonstrating strictures, concentric stenoses, and dilatations, but MinIPs have very limited applications in the detection of tumor-associated changes. Minimum intensity proj ections are a sensitive tool to disp lay parenchymal density, in particular obstructive pulmonary disease and emphysema (Fig. 2.16). Thin-slab MinIP can be applied for evaluation of the intra.hepa.tic bile ducts and the pan-
creatic duct.
(Shaded Surface Display)
Principle
A shaded surface display (SSD) is a surface-rendered image that provides a realistically looking three-dimensional view of the surface of a structure of interest within the acquired volume data set. When creating a SSD, it is first necessary to define the "3D object" of interest, such as the
bony pelvis or abdominal aorta. This process of separating the object from the background, called segmentation, may be quite simple or extremely difficult depending on the object contrast. The simplest segmentation process is to define a 3D object by selecting a suitable range of CT numbers (e.g., all voxels whose attenuation exceeds a threshold value of 150 HU).
Threshold Selection and Size Representation
A three-dimensional display of the object surface is created by illuminating the object with one or more virtual light sources and then computing and displaying the intensity of the light that is scattered back into the plane of observation (Fig. 2.20). At each point in this plane, the data points of the object that are encountered first in the viewing direction are used to generate a 3D surface image. The distance of the surface from the light source and the surface gradient of the CT densities are used to create shadowing effects and heighten realism of the 3D display. Most programs allow only one virtual light source to be used. By varying the position of the light source details of the object surface can be high-lighted differently. With some programs, multiple objects can be displayed simultaneously and encoded in various colors. The projection of the object into the viewing plane can be done either by casting parallel rays (orthographic rendering) or by casting rays from a virtual eye on to the object [perspective rendering). While orthographic rendering is excellent for gaining an overview of the spatial relationship of structures, perspective rendering with various viewing angles (usually 15°-90°)
allows for maneuvering through the data volume and is the prerequisite for techniques such as virtual endoscopy. The quality of the SSD depends on the scan parameters, as in a 2D reformatted image, but generally the effect is less obvious because of the considerable amount of information that is lost in the thresholding process. A more important determinant of SSD image quality is object segmentation, especially the selection of the segmentation threshold. Threshold Selection and Size Representation
The optimum threshold value for segmentation is the value that depicts a structure in its true, complete size. In theory, this value is midway between the CT number of the object (e.g., the aorta) and that of its surroundings (Le., surrounding fat or other soft tissues). We know from practical experience that the theoretical value should be reduced by about 10%, depending on various imaging parameters. However, only structures that are larger than the section width SW will be portrayed at their true size.
Virtual light source Scattered
light --------..
Reflected light -----,. Vector perpendicula to object surface
Fig. 2.20 Shaded surface display (SSD).A virtual light source illuminates the 3D object, and software calculates the light intensities that are scattered and reflected toward the observer.
Eye of the virtual observer
ILightthat is scattered in the direction of the virtual observer. The length of the vector determines the gray level of the object surface.
61
2 Image Processing and Display Techniques
The larger the section width, the more object elements will be falsely represented in the display. Relatively small or thin details that are oriented parallel to the scan plane will lose contrast due to partial volume effects, with smaller objects appearing progressively diminished, and finally vanishing completely, as their diameter decreases. The choice of threshold, therefore, is always a compromise (Table 2.3). When SSD is used in skeletal examinations, this diminishing effect can lead to bony pseudodefects or the non-visualization of thin bony plates (e.g., the orbital floor). In CTA, small vessels that course horizontally may appear narrowed, may show pseudo-stenoses or pseudoocclusions, or may not be visualized (Fig. 2.21). •
Lowering
the Threshold
When the threshold is lowered, more voxels contribute to the displayed object, and the apparent diameter and volume of the object increase. This can partially compensate for partial volume effects, enabling smaller structures to be portrayed more realistically. Lowering the threshold also leads to longitudinal distortion of larger structures of higher contrast. If the threshold value is lowered even further, the CT number of single pixels will exceed the threshold due to image noise. This will cause flying pixels to appear in the 3D display (see Fig. 7.32), and useful image details will be obscured. Noise can also cause "stalactite" artifacts to appear on the upper and lower surfaces of low-contrast structures. Soft tissues whose CT numbers lie above the threshold value (e.g., after in-
Table 2.3
Fig. 2.21 The selected threshold value profoundly affects the appearance of the SSD. In this patient with a high-grade renal artery stenosis, no apparent pathology is seen with a threshold of 120 HU (a), while a threshold of 200 HU simulates a vascular occlusion (b).
travenous contrast administration) may also obscure structures of interest. The remedy is either to raise the threshold value or use more sophisticated segmentation techniques to separate the object of interest from other structures that also lie above the selected threshold value (see Fig. 2.40). •
Raising the Threshold
When the threshold is raised, fewer voxels contribute to the object display, and the apparent diameter and volume of the object decrease. This can eliminate flying pixels and superimposed structures of lower attenuation, but it
Threshold values for segmentation in 3 D surface rendering Threshold
Application
Comments
Bone
>150
Skeletal studies
Lower values are used in osteoporosis
Vessels
>150
eTA
Lower values can be used after smoothing
I Tissue
Soft tissues
Feasible only in exceptional cases
Fluid
Feasible only in exceptional cases
Fat
Not feasible
Lung Air
>-600 <-200 >-500 <-500 <-500 AA __§1CY';
Bronchogenic carcinoma Lung surface
Relation of tumor to vessels and pleura (Sub)pleural abnormalities, lung volume
Skin Colon, larynx Central bronchi Peripheral bronchi
Skin surface "Luminal cast" "Luminal cast" "Luminal cast"
VRT-Principle
also increases artifacts such as the pseudo-stenoses noted above. Frequently the optimum threshold value in skeletal applications is quite low. especially in patients with osteoporosis. but higher values must be selected to eliminate superimposed structures despite the fact that more "holes"appearin the bone.
Applications Shaded surface displays yield impressive images that can define surface structures and clarify complex three-dimensional relationships, for example among bone fragments or vascular structures (e.g., see Fig. 24.42 or Fig. 25.15). Objects rendered by SSD can be rotated and viewed from any desired angle, and SSDs can be used preoperatively to give the surgeon a three-dimensional impressionofthe operative situs.
Volume
Rendering
a tool for the presentaused as an aid to diagnosis in complex anatomic or pathologic situations (e.g., acetabular fractures, complex vascular relationships). As a result. SSDs are used predominantly in skeletal studies (see Chapter 25) and eTA (see Chapter 24) and they have occasional applications in bronchial examinations (e.g., see Fig. 9.14). The rendering of other types of soft-tissue structures is a laborious process that cannot be recommended for routine clinical investigations. SSD have a role in virtual endoscopy (see p. 70) because the rendering process is much faster than with volume rendering techniques. SSD is therefore well suited for interactive navigation though a virtual endoscopic data set (see Fig. 2.50). SSDs are primarily
tion offindings and are only occasionally
Techniques
Principle Maximum and minimum intensity projections are simple examples of volume rendering techniques. The image is generated by casting rays through the volume of interest (ray tracing), and projecting anumeric value that is derived from the data encountered along each ray according to prescribed rules (e.g., the maximum CT number in MIP). Generally, however, volume rendering technique (VRT) is the term applied to a complex procedure that is very versatile and can combine characteristics of surface rendering and MIP. VRT assigns a range of opacity values to CT numbers and thus yields better definition of object contours or semitransparent display of structures (Fig. 2.22). VRT is quickly becoming an established standard technique for rendering of spiral CT or multislice CT data sets. Image quality is high and the user has a whole range of rendering effects at his disposal. VRTcan best be understood when comparingitto shaded surface displays (SSD). Such surface rendering is a "binary' process in which all CT numbers belonging to the 3D object (within the chosen threshold range) have maximum
opacity, while all CT numbers outside the range have zero opacity and do not contribute to the image. Since all the voxels within the CT range have maximum opacity, only the surface of the object is depicted in the SSD. In volume rendering, opacity values are continuous and can vary between 0% and 100%. The behavior of VRT is determined by the opacity curve. This curve may be drawn man ually, a procedure that is tedious and not very reproducible. A very comfortable way of creating the curve is to use preset curve forms and to control its precise location on the CT number scale by window and level operations (Figs.2.22a and 2.23, Table 2.4). Alternatively, opacity curves may be produced by a number of (partially overlapping) trapezoids that can be altered in form, height and position on the CT scale (Fig.2.22b). These trapezoids can be chosen to represent the attenuation range of various tissues, e.g., fat, other soft tissues, contrast-enhanced blood vessels, and bone. A different color can then be attributed to each of these trapezoids to be able to differentiate between these tissues on the final image. VRT can either create a transmission display (similar to a ray-sum projection or MIP) or
63
64
2 Image Processing and Display Techniques
Opacity(%) 100r
Opacity(%)
100.---------------~--------~--~
50 _ _
__ i WI L = 40Q1100 •.
O+---~~--~----~,----._--~----~ -200
-100
o
100
200
300
-200
400
-100
CT number (HU)
Fig. 2.22 Principle of volume rendering (VRT). During the creation of a volume rendered image, an opacity curve determines the opacity of various tissues depending on the CT numbers of each voxel. For most instances, a simple technique based on a single opacity ramp from full transparency to full opacity yields good results (a). This ramp can be determined similar to the
o
100
200 300 400 CTnumber(HU)
window/level operation of normal CT images. Alternatively, multiple trapezoids with various colors that are assigned to tissues of different CT numbers can be used as an opacity function (b). Note that structures with low opacities (e.g. fat) will appear semitransparent with both techniques. No gradient shading (reflectivity =0) was used in these examples.
a Fig. 2.23 The opacity curve and the reflectivity determine the resulting volume rendered image. A narrow opacity ramp (W/L= 400/400) yields an excellent surface display of the abdominal vasculature (a). Note the
calcified gallstone (arrow). Shifting the opacity curve towards lower values yields semitransparent displays of soft tissues (b. W /L= 400/200). Additional gradient shading was used on these images.
a surface display (similar to the SSD) depending on the amount of gradient shading that is used to simulate reflectivity. This "reflectivity constant" may be determined either by using presets or by adjusting its numeric value.
•
Color Coding
Color coding may be performed on a per-trapezoid basis as described above. Ideally, various tissues are colored in a different way, thus allowing for visual differentiation. In practice, however, there is substantial overlap between
VRT-Principle Ta~e
2.4
Opacity
settings
Application
for volume Opacity
rendering s<sttings
W/L Skeletal
studies
CT angiography Diagnostic viewing Softtissue
Comments Range
Curve
300/200
100 ..400
Ramp
Lower values
400/300
100 ..500
Ramp
Values
100 ..300 -1050 .. 450 -900 .. 300
Ramp Ramp Ramp
MRP-like
400/100 1500/-300 1200/-300
depend
are used
500/-500
-750 .. -250
Spike
Air/soft
tissue
interfaces
projection Luminal casts
500/-500
-750 .. -250
Inv. ramp
Air/soft
tissue
interfaces
1200/-300 1200/-300 1200/-300 200/150
-900 -900 -900 100
Ramp Ramp Ramp Inv. ramp
Double
spike
depending
on contrast
Angioscopy Opooty
settings
In this table, Inv.
may vary a single
.. -300 ..-300 ..-300 .. -300
enhancement
and
opacification
rendering
Tissue-transition
Virtual endoscopy Bronchoscopy Laryngoscopy Coionoscopy
in osteoporosis
on vessel
implementation
for calcifications of volume
rendering
is used for most applications (which is sufficient for most imaging tasks) full opacity for lower CT numbers and zero opacity for hgh CT numbers
opacity function, maximum best achieved by fixed color
opacity schemes,
for center of range but may require use of multiple
trapezoids
on some
.workstations
various tissue types, e.g., between contrast-enhanced organs and blood vessels, and between vessels and bone, and no secure discrimination is possible (Fig. 2.22 b). Alternatively, a color scale can be assigned to the CT number scale. This color scale may be fixed relative to the CT numbers thus providing an absolute reference color depending on the attenuationortissue type (Fig. 2.24a). This technique is ideal for displaying non-contrast-enhanced structures. Because contrast enhancement may vary substantially between patients, a flexible assignment of colors and CT numbers is advisable, depending on the actual setting of the opacity curve. This can be done by assigning the color scale to the slope of the opacity curve (Fig.2.24b) . •
Spatial
Resolution and Matrix Size
Fast response is usually obtained by reducing the spatial resolution of the VRT image during user interaction. A 128-matrix will only give a rough orientation but a 256-matrix already provides enough detail to find the optimum viewing angle for demonstrating a structure ofinterest. A 512-matrix, however, further improves spatial resolution and has a positive effect on
the display of small vessels or bony surface detail (Fig. 2.25). A 1024- or 2048-matrix is also used in some systems, at least as an intermediate step during the rendering process. Such large matrices are especially helpful if structures have to be greatly enlarged from the original data set (such as the environs of a cerebral aneurysm), or ifvery long ranges have to be displayed (e.g., chest and abdomen or abdomen and legs). Interactive
Rendering-Movies
Presets make it easy to choose a proper display setting, and in most of the recent software. selection of parameters can be performed interactively close to real time. With specific volume rendering boards that can be installed in standard PCs, real-time interaction with> 8 frames per second becomes feasible. Movie sequences can be generated that allow for an off-line viewing of VRT data without having real-time interactive VRT software available. If standard digital movie formats (e.g. AVI, MPEG) are used, these sequences may be displayed on any Pc. Alternatively, images can be transferred to videotapes and be displayed on any videocassette recorder. Such
65
66
2 Image Processing and Display Techniques
Air I skin
Fat
Vessels / bone Soft tissues
o~~~~~~~~~~~~~~~~~ -1000 -500 o
500
relative color scale assigns colors to the opacity window but not to the CT number scale (b). Additional gradient shading was only used in (a).
Fig. 2.24 Color coding. An absolute color scale assigns colors to various soft tissues, depending on their CT numbers but independent of the opacity setting (a). A
a Fig. 2.25
b 2
Matrix size and image quality, (a) VRT with 256 resolution, (b) VRTwith 5122 resolution.
VRT-Artifacts and IPitfalls
a
c
b
Fig. 2.26 VRT of air-containing structures in the paranasal sinuses, pharynx and larynx, (a) Display of surface (cast). (b) Tissue transition projection of the airway wall. (c) Virtual endoscopy of the larynx.
movies are especially helpful for displaying complex anatomy to referring physicians. They already are being used for the display of renal cell carcinomas prior to nephron-sparing surgery, for the display of the anatomy in the evaluation of living renal donors, and for surgical planning of cere bral aneurysms. Special Techniques •
Air Casts
Inverted opacity curves that provide a high opacity for low-attenuation regions and low opacity for high-attenuation regions can be used to selectively display a "cast" of the tracheobronchial system, the larynx, or the colon (Fig.2.26a, see also F.igs.2.36a and 2.50a). •
Tissue Transition Projections
Tissue transition projections selectively render the walls of the structure of interest. With VRT this can be reached by using an opacity curve that has the shape of a "spike" that is centered on the CT number between the two structures of interest (e.g., -500 HU for imaging of soft tissue-air interfaces such as in the colon or the trachea). This display technique, for example, produces images that simulate a double-contrast examination (Fig. 2.26b, see also Figs. 2.36 and 2,50).
•
MPR-like
Rendering
MPR-like rendering is a technique that can substitute for traditional MPR. With VRT such images can be obtained by using a linear opacity curve with a slope that covers a range similar to conventional window settings. No reflectivity should be employed, to allow for optimum evaluation of cut surfaces. By using interactive cut planes, the radiologist can section through the data volume in a fashion similar to interactive MPR but has the additional advantage of depth information provided by these images (Fig. 2.27, see also see Fig. 2.22a). Table 2.4 provides suggestions for proper opacity settings, depending on the clinical imaging task. Artifacts and Pitfalls
--------
VRT is less susceptible to pseudo-stenoses or pseudo-occlusions compared to SSD, although suboptimum opacity settings also lead to image artifacts (Fig. 2.28). Venetian blind artifacts depend on the viewing angle and may be found incertain implementations of volume rendering (Figs.2.29a, b).
67
68
2 Image Processing and Display Techniques
b
a Fig. 2.27 MPR-like VRT. Cutting into a volume rendered image (with reflectivity set to 0) creates MPR-like views that add the advantages of volume rendering
(depth information) to that of MPR (evaluation of cut planes), (a) Soft tissue setting (W/L= 400/100), (b) lung setting (W/L= 15001-300).
a Fig. 2.28 Volume rendering uses the whole 3D data set for image creation. However, only parts of the 3D data are displayed, depending on the opacity function. With an opacity window ofW/L=400/250 (a) there is
Image noise may cause irregularities of the
object surfaces (Fig.2.29c) or may even obscure underlying pathology. Depending on the opacity function, noise may also cause a veil-like shadowing of structures that are deeper inside the displayed volume.
superimposition of the right renal vein (arrows). With an opacity window of 400/300 (b) portions of the accessory renal arteries on the left are lost (arrow heads) but the lower pole artery on the right is visualized (arrow).
It has been stated that VRT does not lose information because the whole data volume is projected. However, VRT does not display all the information in the data volume but, like all 3D rendering techniques, selectively displays portions of the scanned object, depending on its CT attenuation and position relative to the viewer (Fig. 2.28).
VRT-Applications
a
c
b
doscopic views I (upper IPortion of image), (c) Ilmage noise is especially disturbing in VRT with gradient shad2 ing (reflectivity) and a 512 resolution.
IFig..2.29 Artifacts.rfajVenetiari blindartifactsaretypical for some software implementations of volume rendering. (b) They are especially disturbing on virtual en-
a
b
Fig. 2.30 Differentiation between calcified plaques and vessel lumen is simplified on color displays (a) compared to black-and-white rendering (b). In this patient
Applications
Volume rendering can be used for CTA, skeletal imaging, tracheobronchial imaging, display of the lungs, the colon, abdominal organ and even as a primary tool for image analysis. •
CTAngiography
In CTA, volume rendering is becoming the new standard for the displayofthe pulmonaryvessels, the aorta, and the abdominal vessels. Colorcoded VRT provides display of vessel lumen and
c
d with a superior vena cava occlusion, color display (c) improves the visualization of the collaterals in the mediastinum relative to black-and-white displays (d).
calcifications in separate colors, thus making it easy to localize calcified plaques (Fig. 2.30). Color coding also may allow differentiation between arterial and venous vessels, as well as between organs with different contrast enhancement (Fig. 2.31). When the aorta is eliminated by editing procedures, excellent displays of the celiac and mesenteric vessels become available (Fig. 2.32). Segmentation can be used to eliminate superimposing soft tissues and make it possible not only to display the vessel lumen but
69
70
2 Image Processing and Display Techniques
also thrombosed portions of the vessel (e.g., of an abdominal aneurysm) as well (Fig. 2.33). •
Fig. 2.31 Differentiation between arteries, veins, and contrast-enhanced organs is possiblewith VRT ifthere is differential enhancement.
Skeletal Imaging
For skeletal imaging, VRT is an excellent tool for classifying complex fractures (e.g., of the acetabulum), for demonstrating the spatial position of fracture fragments, and for planning of complex or minimally invasive surgery. In orthopedic applications, VRT is superior to SSD because it is less susceptible to partial volume effects with pseudo-defects in thin bony laminae or osteoporotic patients. In addition, VRT allows for simultaneous display of bones as well as tendons or muscles, and even allows for assessment of the skin contours (Fig. 2.34), e.g., in patients prior to corrective surgery. In the spine, the vertebrae as well as the intervertebral disks can be displayed, making it an ideal tool for assessing and demonstrating pathology of the spinal column (Fig. 2.35).
Fig. 2.32 Removal of the aorta allows for excellent display of the celiac and mesenteric vessels. In this patient the relationship of the aneurysm to the mesenteric side branches can best be demonstrated on a PA view (b).
a
b
Fig. 2.33 With a soft tissue setting forVRT, and elimination ofsuperimposing soft tissue by image segmentation, direct display of thrombi is possible.
a
b
VRT-Applications
a
b
c
Fig. 2.34 VRT not only allows for a selective reconstruction of the skin (a) or bone (b), but also semitransparent displayof skin that make the underlying bone visible (c). Note the bilateral obstruction of the brachiocephalic veins.
•
Tracheobronchial
Imaging
For imaging ofthe tracheobronchial system, VRT provides either casts or semitransparent walls of the tracheobronchial tree (Fig. 2.36). VRTis especially helpful when it comes to virtual bronchoscopy (see p.81) because even segmental bronchi may be evaluated if a thin-section multislice CT data acquisition was used. •
Lungs
For the lungs, volume (or surface) rendering can be employed to create images of the lung surface (see Fig. 2.47) and to assess the spatial relationship of focal lung lesions to the chest wall or mediastinum (Fig. 2.37). It can be used to create color-coded density maps that help detect and
a
b
Fig. 2.36 The tracheobronchial system can be visualized as an "intraluminal cast" using SSD or VRT (a) or using a tissue transition projection (b). Similar effects
Fig. 2.35 Volume rendering of the spinal column allows for demonstration of the vertebrae as well as the intervertebral disks.
c are possible using a 20 mm thick MPR after removal of the mediastinal tissues by region growing (c).
71
2 Image Processing and Display Techniques
quantify regions with focal air trapping or perfusion abnormalities (Fig. 2.38). • Colon In the colon, VRT is an excellent technique for virtual colonoscopy but also for surveying the data with the help of sliding thin slabs (see below) that combine the advantages of axial, coronal, or sagittal MPR with the depth information provided by VRT. Again, casts or semitransparent displays of the colon become available, some very similar to conventional barium enema studies (see Fig. 2.50). Fig. 2.37 Volume rendering allows for excellent displays of the location of focal lesions relative to the chest wall and the pulmonary vessels. Note that this metastasis is due to a tumor thrombus spreading within a peripheral pulmonary artery.
Fig. 2.38 Color-coded density maps can be created when using a rainbow spectrum and a lung-centered opacity curve (W/L= 500/-750). Note the wedgeshaped perfusion defects (arrowheads) in this multislice CT A of chronic pulmonary embolism.
•
Routine Reporting
In the chest and abdomen, VRT with proper opacity settings provides semitransparent or color-coded displays of soft tissues (see Fig. 2.27). In the abdomen, in particular, VRT allows for excellent display of the liver, portal vessels, bowel structures, the pancreas, and even the suprarenal glands. VRT with opacity settings that mirror those of conventional CT imaging have been suggested as a primary viewing mode for making the diagnosis. By analogy with a soft -tissue window, the opacity settings should be chosen in a way that makes normal fatty tissue transparent but is able to demonstrate increased density in the fat on the rendered images (see Fig. 2.27). For the purpose of using VRT as the primary diagnostic tool, however, cut planes have to be available that expose cross-sections of organs that would otherwise be rendered opaque. In addition, interaction has to be real-time to allow for fast maneuvering through the data volume. There is still no hard data, however, that such a procedure is safe and does not miss important findings.
Segmentation Principle
Both 3D volume rendering (e.g., MIP, VRT) and 3D surface rendering (SSD) require a segmentation process that defines the volume of interest and separates it from structures that should not be represented in the 3D image. The process of manipulating the data set to control what structures are included or excluded in the rendering
is termed "editing". Positive editing is based on marking the structures of interest that will be retained in the 3D image. Negative editing involves marking the unwanted structures that will be removed. A distinction is drawn between 2D editing functions, which are applied section by section, and 3D editing functions, which are applied to the 3D data set as a whole.
Segmentation-Principle
•
Cutting Functions
Cutting functions may be used on the original data volume before 3D rendering or on the 3Drendered object itself. Cutting can be performed in the original data volume on a section-by-section basis, a technique that is very time consuming and can only be recommended in exceptional and complex cases in which a small set of sections are edited. It is more sensible to perform cutting functions on 3D-slabs, i.e., stacks of images that are a few mm to several cm wide. This can substantially reduce the effort ofediting butis still comparatively time consuming iflarger ranges, such the whole chest or abdomen, have to be processed. The most efficient technique is based on cut lines that are drawn on a number of unevenly spaced reference sections. The program then performs a three-dimensional interpolation between these cut lines and thus defines the volume to be included or excluded from the volume of interest that is chosen for further processing (Fig. 2.39). This technique allows the lungs or the abdominal vessels to be edited in a few minutes, and even allows for complex procedures in which portions of organs are removed for better visualization of neighboring structures (see Figs. 2.31-2.33). Cutting in a 3D-object first requires a view in which the cut plane can be drawn. This plane will be perpendicular to this view, so the user
a
b
Fig. 2.39 Cutting functions using a "rubber sheet" algorithm are highly efficient for complex editing tasks. A number of arbitrary regions of interest (ROI) are drawn at a few levels throughout the chest (a). Display of reo
has to make sure that the structures to be separated do not overlap. The cutting functions are best used to exclude unwanted objects but they can also be employed to focus on certain regions and include the structures of interest. •
Threshold Techniques
Threshold techniques are simple processes that use a threshold value or range of CT numbers to define the volume that will undergo segmentation (see Table 2.3). Threshold techniques are generally used in the rendering of SSDs. When threshold techniques are used alone, it is common for an image to include different anatomic structures that need to be displayed separately. This would apply, for example, to opacified vessels and the vertebral column, or to the femoral head and acetabulum. For this reason, threshold techniques are often combined with other methods that can separate objects within the same threshold range. •
Connectivity
Connectivity algorithms are used to identify contiguous image regions that have a certain property in common. Such a "region-growing" algorithm starts at a seed point and detects all voxels that satisfy the specified condition and are interconnected to the seed point. The usual condition is that the voxels lie within a prescribed range of CT numbers.
c maining volume with indicated ROllevels (b). A volumerendered vascular display (W/L= 600/200) demonstrates multiple AV -shunts in either lung.
2 Image Processing and Display Techniques
Connectivity algorithms can be applied to single axial sections, to slabs consisting of multiple sections, or to the 3D volume as a whole. The larger the volume, the faster the theoretical rate of segmentation, but there is a greater risk that artifactual "bridges" will form between image regions that are actually separate. Various methods can be used to eliminate these bridges (Fig. 2.40). Frequently, object separation can be achieved simply by changing the threshold range. Other methods involve the use of manually drawn separating lines or "morphologic operators" to erase the unwanted connections. Watershed algorithms are very helpful for separating two structures that are connected by a bridge. They determine the plane with the lowest CT number within that bridge and cut at that level. In practice this means that the operator only has to place a seed point on the structure to be included and others on the structures to be excluded and let the watershed algorithm do the separation. In many cases such a procedure will be sufficient; in some more complex cases, a repeat of this procedure at another level of the data volume will become necessary. •
Morphologic
Operators
Morphologic operators depend on the morphology of a structure rather than its attenuation values. The most simple but useful operators are those that remove rows ofvoxels from the surface of an object (erosion) and those that attach rows of voxels to the surface of an object {dilation}. Erosion operators can be used to remove bridges between adjacent objects. Dilation can then be per-
a
b
c
Fig. 2.40 Segmentation. Connectivity algorithms can be used to mark contiguous regions that have common characteristics. "Bridges" that persist between structures to be separated (a) can be eliminated by the use of
formed to restore the object to its approximate original size, although surface details will be lost (Fig. 2.40). Thesequenceoferosionfollowedbydilation is called "opening"because it opens up connections between structures, while the combination of dilation followed by erosion is called "closing" because it removes holes within an object. After successful detection of objects (e.g., the skeletal structures) to be excluded by region growing, it is advisable to perform a dilation procedure on these objects to include and eliminate also voxels at the object boundary that contain higher CT numbers than the surrounding structures. If these voxels are not excluded (Fig. 2.41), ghosting artifacts will result in 3D rendered images. Other morphologic operators can recognize small groups ofvoxels and can therefore be used to erase "flying pixels" from an image (Fig. 2.42). Automated Techniques, Computer-assisted Diagnosis
Thereisanincreasingamountoffullyorsemiautomated techniques being developed that require no or only minimal input (such as positioning of various seed points) for complex editing tasks. Such tools allow for automated extraction of the lungs, or bone removalin CTA (Fig. 2.43), or auromated vessel analysis (Fig. 2.44). Computer-aided diagnosis tools are able to automatically detect, extract, and measure the volume of pulmonary nodules for 1ung cancer screening. Similar tools are being developed for colon cancer screening.
d
e
cutting functions (b), erosion operators (c), or by increasing the threshold (d). Subsequent dilatation can be performed to remove the higher-attenuating areas that remain around bony structures (e).
Automated
Techniques, Computer-assisted
Fig. 2.41 Bone removal by region growing. Region growing with a threshold above 200 HU allows separation of bones from vessels but leaves high-attenuation areas around bony structures that lead to ghosting artifacts on MIP (a). This ghosting may completely obscure smaller vessels. Dilatation of the bone and removal of holes in the vertebral body using "closing operators", yields high quality MIP (b).
a
b
a
b
a
b
Fig. 2.42 Flying pixels ("floaters") due to image noise are especially disturbing on low dose data sets (a) but can be removed by size-dependent filter functions or regiongrowing of the structures of interest (b).
Fig. 2.43 Automated bone removal is now feasible on some workstations. By clicking on the structures to be removed or, in this case, retained (a), automatic removal is quite effective (b). Note the type II endoleak after aortic stent grafting (arrow).
Diagnosis
75
76
2 Image Processing and Display Techniques
Fig. 2.44 Automated vessel analysis is based on automatic vessel tracking but adds measurements perpendicular to the vascular course. Depending on the software, local cross-sectional area, and the minimum and maximum diameter of the vessel of interest are calculated and a rubber-band image ofthe vessel is displayed.
There are programs that include some form of artificial intelligence to account for anatomic variations. Many of these tools have just entered the market or are about to be introduced soon, pending FDA approval. Applications
Segmentation is a prerequisite for eTA with MIP images in an anteroposterior direction. Positive editing is generally sufficient if there is a focused question. For the presence of a renal artery stenosis, for example, best results are obtained if a thin curved slab is included that just encompasses the renal arteries from the origin
to the intrarenal branches and excludes the renal veins as well as portions of the anterior and posterior renal cortex (see Fig. 2.13b). Negative editing with exclusion of the skeletal structures can be recommended if a survey of the thoracic, abdominal, or peripheral vasculature is required (Table 2.5). Segmentation can improve visualization of complex vascular anatomy or pathology (posterior-anterior views) if SSD or VRT are used (see Fig. 2.31). In addition, it can provide selective views of certain vascular territories (e.g., mesenteric vessels) by eliminating the aorta or other superimposing structures (see Fig. 2.32). For visualization of the pulmonary vasculature with MIP, SSD, or VRT, prior removal ofthe chest wall is necessary (see Fig. 2.39). Segmentation of soft tissues is only useful for VRT. Using such a technique, visualization of complex anatomic relationships becomes possible (see Fig. 2.33). For the lungs, segmentation can be employed to selectively display the lung tissue in order to obtain density maps or to depict the pulmonary surface (see Fig. 2.47). Removal of all soft tissue structures in the chest that are connected to the mediastinum (connectivity with a threshold> -200 HU followed by a 2-3 pixel dilatation) will yield an overview of the aerated lung. Thick MPR from such a data set will create images that are similar to tissuetransition projections ("tissue transition MPR", Fig. 2.45). In the skeleton, segmentation can be used for bone exarticulation and provide selective views of the joint surface, e.g., of the acetabulum, but also of the glenoid or the calcaneus (Fig. 2.46, see also Fig. 25.41).
Fig. 2.45 Tissue-transition MPR can be obtained by removing all extracorporeal air as well as all soft tissues (> -200 HU), followed by a 2-pixel dilation, and then using a thick MPR to visualize interfaces of air-filled structures. Note that also the pulmonary vessels appear translucent (a). If no dilation is performed, the resulting image will selectively display the aerated lung
(b). a
b
Segmentation-Applications Table 2.5
Segmentation procedures for various clinical tasks
Application
Display goal
eTA
Bone removal Chestwall
Pulmonary vessels
Abdomen
Abdominal vessels
Neck
Carotid arteries
Bone exarticu- joint surface lation
technique
Automated bone removal or Cutting (rubber sheet): include all except chest wall Mark bone (RG/WS > 180 HU), dilate bone (2-3 pixels), remove bone Cutting (rubber sheet): include all except abdominal wall Mark carotids (RG/WS = 150-400 HU) using lower + upper threshold, create two VOl roughly including skull and neck/cervical spine, mark bone (RG/WS > 500 HU - skull, RG/WS > 180 HU spine), dilate bone (2-4 pixels) remove bone Draw center line through carotids (manualfsemiautomated) , produce 3-5 mm thick slab curved MPR/MIPIVRT Mark bone of interest (RG/WS > 180 HU), dilate 2-3 pixels, mark bone to be removed (RG/WS), dilate 2-3 pixels, remove bone
Lung extraction
Volume measurements
Mark trachea + large bronchi (RG <-900HU), dilate 2-4 pixels, extract each lung (RG <-200 HU)
Colon extraction
Selective display
Block ileocecal valve if necessary (0 HU line), mark colon (RG<-500HU) use multiple seeds if necessary, dilate 3 pixels for SSD orVRT
Tracheobronchial extraction
Selective display
Mark tracheobronchial tree (RG/WS <-900HU), increase threshold until overflow into lung parenchymajust does not yet occur, dilate 2-3 pixels forVRT/SSD
Tissue-transition MPR
Transparent wall of air-filled organ (larynx, trachea, bowel)
Mark air-filled organ (RG <-200HU), dilate 2-3 pixels, remove rest
Surface display
RG= region growing; WS = watershed algorithm; suggested thresholds in brackets have to be individually ,§4
Fig. 2.46 Bone exarticulation can be used to create an > unobscured view of articular surfaces (here: complex fracture ofthe acetabulum).
77
2 Image Processing and Display Techniques For volume quantification, of organs or lesions, the structure has to be isolated first by proper editing procedures. The techniques may be fully automated (pulmonary nodule detection and quantification, see Fig. 9.18) or will require substantial user input (e.g., quantification of hepatic or splenic volume). Often, a combination of connectivity (seeding) and manual cutting functions will provide the quickest results (Fig. 2.47). Fig. 2.47 Quantification of the lung volume may be helpful prior to lung transplantation (especially for living donors) or for follow-up after surgery. The total volume of the aerated lung including the tracheobronchial tree or an isolated lung are easily determined using region growing techniques.
Virtual Endoscopy Principle
Virtual endoscopy is a 3D-rendering technique that simulates endoscopic views. The first step is to select suitable threshold ranges (SSD) or opacity settings (VRT) that define the internal surfaces of interest, such as the bronchi, paranasal sinuses, blood vessels, or the gastrointestinal tract (Table 2.6). The endoscopic effect is produced by perspective rendering along a path within the data set. This provides an "endoscopic" view of internal objects (Fig. 2.48 and 2.49) . •
Maneuvering through the Data Volume
The viewer can interactively change his or her location, viewing direction, and the visual field of the virtual endoscope. Good viewing angles Table 2.6
that come close to those of real endoscopes vary between 60° and 90°. Orientation can be established from multiplanar reconstructions and from the virtual endoscopic image itself. In some systems the operator can preprogram a "flightpath" through the organ system or define the path interactively. A good first approximation of the flight path through tubular organs can be gained by calculating a centerline using so-called skeletonization algorithms. Techniques to view the walls are helpful, and back-mirror techniques that inverse the viewing direction are essential for some applications such as virtual colonoscopy. Still, at present, maneuvering through the data sets is a tedious and time-consuming effort. For this reason, virtual endoscopy is mainly used as a problem-solving tool, for example to
Threshold values for virtual endoscopy
Application
Interface
Opacity settings 1,
Threshold" W/L
Bronchoscopy Laryngoscopy Colonoscopy Angioscopy 3
Mediastinum/air Bronchial wall/air Soft tissue/air Soft tissue/air Soft tissue/contrast material
surface rendering (SSD), b volume rendering (VRT)
>-soo >-900 >-SOO >-SOO < 1S0
1200/300 300/700 1200/300 1200/300 200/1 SO
Range -900 -950 -900 -900
so
... 300 ... -550 ... 300 ... 300 ... 2S0
/\
Curve Ramp Ramp Ramp Ramp Inv. Ramp
Alternative
Visualization
Techniques
Fig. 2.48 Virtual endoscopy yields perspective views by using a central projection instead of the parallel projection used in other techniques. Internal surfaces can be displayed using SSD (a) or volume rendering (b).
a
b
a
b
Fig. 2.49 Virtual endoscopy relies on a high contrast between lumen and surrounding tissues. Most applications such as virtual cystoscopy (a) rely on air as a negative contrast material. For virtual angioscopy ( b) or cholangioscopy, the positive contrast in the structure of interest requires an inverted opacity curve. Note the small entry tear in this type B.
differentiate a colonic fold from a real polyp in CT colonography. Alternative Visualization Techniques
Casts ofa luminal organ are known from CTA and there represent the standard display technique with SSD or VRT. Similar techniques are possible with the tracheobronchial system, and yield a tree-like structure. Distal branches are displayed best if the data was acquired with thin-section multisliceCT. Forthecolon, these casts resemble a monocontrast barium enema, but they are superior because they provide a three-dimensional display of the colonic surface (Fig.2.50a). Casts, however, have an intrinsic drawback: they are less sensitive for displaying small endoluminal protrusions such as polyps orwall-adherent tumors. These lesions appear as small indentations on the surface of the cast and may easily be overlooked.
Tissue transition projections render the walls of the structure of interest in an opaque or translucent fashion. They can be created either by using SSD or VRT with a narrow range ofCT numbers that correspond to the transition zone between luminal contents (air or contrast medium) and the surrounding tissues. This display technique, for example, produces images that simulate a double-contrast barium enema (Fig.2.50b). Alternatively, a tissue transition MPR (see also Fig. 2.36) can be chosen that provides similar results (Fig. 2.50c). ThickMPRis useful for displaying small vessels (thickness 3-5 mm) or the central tracheobronchial system (thickness 7-20 mm). They provide an overview of anatomy and pathology (see Fig. 2.36) and make it easier to detect focal lesions because these will alter the attenuation of the vessel or bronchus. Sliding thin-slab VRT is a method similar to sliding thin-slab MIP but uses volume rendering
79
80
2 Image Processing and Display Techniques
a
b
d
e
9 Fig. 2.50 Alternative visualization techniques may provide a helpful adjunct to virtual endoscopy, as demonstrated in this patient with a colonic lipoma and small polyps in the proximal transverse colon, (a) Intraluminal cast. (b) Tissue transition projection using VRT. (c)
Tissue transition MPR. (d) Thin-slab VRT. (e) Coneddown VRT. (f) Virtual colonic dissection ("virtual gross pathology"), (g) Virtual colonoscopyusing a panoramic view ("unfolded cube"), (h) MPR demonstrating fat attenuation in the lesion, proving it to be a colonic lipoma.
Virtual Endoscopy-Applkations
instead of maximum intensity projections. This technique can be employed in CTA. tracheobronchial imaging as well as CT colonography (Fig.2.50d). It provides a good overview of the anatomy even in complex and spatially contorted cases. Coned-down VRT allows for in-detail analysis of a subvolume of interest and is helpful for vascular structures as well as CT colongraphy (Fig.2.50e). Virtual dissection (virtual gross pathology) is a new technique for CT colonography that finds the centerline of the colon. stretches it out and performs a virtual longitudinal cut in order to obtain a fiattened view of the inner colonic surface (Fig.2.50f). In order to avoid spatial distortion. only slim stripes of the colon wall are displayed. A panoramic view (Fig.2.50g) yields endoscopic projections in and against the direction of movement as well as views of the sides of the colon. This allows for almost complete endoscopic evaluation of the colonic surface and improves detection of polyps hidden behind larger folds. Other techniques have been tried as an alternative to virtual colonoscopy and include Mercator's projection, bull's eye views and others, but they have not yet galned clinical acceptance. Finally, multiplanar reformations (Fig. 2.50 h) must not be neglected because they are able to yield information about CT attenuation, contrast enhancement as well as transmural extent of a lesion.
Artifacts and Pitfalls Threshold-based techniques (SSD) are inherently more susceptible to artifacts than volume rendering techniques. Even with optimum thresholding, SSDs are degraded by pseudo-occlusions and pseudo-stenoses in small vessels (virtual angioscopy) and by pseudo-defects in the bronchial or colonic wall (virtual bronchoscopy and colonoscopy, respectively). These effects are less pronounced in volume rendering when suitable parameters are selected. With VRT even smaller vessels and airways can be evaluated. In virtual colonoscopy, it is often difficult to distinguish between intraluminal contaminants and polyps. In virtual colonoscopy and gastros-
copy, residual intraluminal fluid can obscure the surface of the stomach or bowel and prevent their evaluation. Optimum preparation is critically important. Thin bowel walls, especially when two bowel loops are adjacent to each other, or haustra may appear translucent and require readjustment of threshold or opacity values. In virtual bronchoscopy, breathing and pulsation may lead to ring-like distortion of the bronchial walls. Mucous can mimic polypoid lesions, although it often appears elongated and is therefore easily distinguished from real pathology. Measurements of luminal diameters and degrees of stenosis are strongly dependent on the viewer's perspective and, especially, on the selected threshold value.
Applications
Virtual colonoscopy is the most widely used virtual endoscopic technique (see also Figs. 15.1618). It is part of a CT colonography study and has been proposed for colon cancer screening as well as for detection of concomitant lesions in stenosing colon cancers. However, very few people use it to examine the whole colon. Instead it is used mainly as a tool to differentiate colonic folds that mimic polyps from real polypoid lesions. With further refinement of the technique, however, virtual colonoscopy may also become a method for screening the whole colon because it should have a higher sensitivity for detecting small polyps that are easily missed on cross-sectional images alone. Virtual bronchoscopy at present has very few clinical indications because most diagnostic decisions can also be made based on alternative display techniques such as thick MPR. Because of the strong enlargement of structures, however, it might gain a role with multislice scanning of the tracheobronchial system. Virtual bronchoscopy can more readily detect small polypoid protrusions into the bronchial lumen and is able to depict lymph nodes that bulge the bronchial wall at the level of a bronchial bifurcation, a sign considered indicative of malignant disease by bronchoscopists. What remains is a gutding tool for fiberoptic bronchoscopy that
81
2 Image Processing and Display Techniques helps in biopsies of more peripheral lesions, or for determining the safest point for transbronchial biopsies in a central location close to a main vessel. Virtual cystoscopy has the potential to detect very small polypoid bladder lesions, even in regions that may be hard to inspect by real cystoscopy, For best results, the technique requires filling of the bladder with contrast material or air. Virtual angioscopy has provided almost no additional information over other visualization
techniques, The only indication may be the display of the relative location of the struts of aortic stent grafts relative to the ostium of branching vessels, A major disadvantage of this techniques is the overestimation of stenoses due to calcified plaques (blooming effect),
Virtual cholangioscopy, virtual gastroscopy, virtual enteroscopy, or endoscopy of other organ
systems have been suggested as well, but most techniques fail to provide helpful information over other, less complex visualization tools,
3 Patient Preparation M. Prokop, A. J.
Preparation
and Contrast Media Application
van der Molen
for CT Examinations
.
Preparation of Patients with Impaired Renal Function (p. 84) Preparation of Patients with Allergy to Contrast Media (p. 86) Preparation of Patients with Hyperthyroidism (p. 87) Bow •• Preparation (p. 87) Positioning (p, 88) Vascular Access (p. 88) Planning the Scan (p. 89) Gastrointestinal Contrast Me
(p. 91)
Parenteral Contrast Media Types of Contrast Media (p. 93) Complications of Contrast Material Injection (p. 93) Treatment of Adverse Reactions (p. 95) Contrast Injection Parameters (p. 96) Intravenous Contrast Administration (p. 96) Basic Principles (p. 96) Injection Parameters and Organ Enhancement (P, Contrast Injection (p. 98) NaCi Bolus (Saline Flush) (p. 99) Contrast. Material Concentration (p. lC Scan D••ay (p. 101) Individualized Contrast Material Inject Perfusion Studies (p. 105) Direct Arterial, Portal, or Venous Contrast Administration Intrathecal Contrast Administration (p. 107) Intracavitary Contrast Administration (p. 107)
3
Patient Preparation and Contrast Media Application
The strategies applied in CT examinations are determined by the technical characteristics of the CT scanner and by the nature of the clinical question. Examinations without contrast medium are suitable only for intrinsically highcontrast structures (bone. stone detection. and lung parenchyma) and for the detection of he-
Preparation
for CT Examinations
When a patient is scheduled for CT examination. it must be decided whether special patient preparation is required (Table 3.1). In patients who have significant renal function impairment or other relative contraindications to contrastenhanced CT. other imaging modalities (ultra-
Table
3.1
Patient
preparation
Vv'hen the examination the
I
is schedulded,
determine
following:
• Renal function
status
(serum
creatinine>
130
u.moi/l)
Plasmacyto,m,a or amyloidosis? • Allergy to contrast materials? • Hyperthyroidism? Papillary or follicular thyroid carcinoma? • Bowel preparation (CT-Sellink prep, virtual colonoscopy)? Fasting (for CT of stomach or pancreas)? • Coordinate with anesthesiologist or pediatrician Intensive care patient on ventilatory support? Small child requiring sedation? • Determine
scan
When patient
protocol
arrives
for
(If not already
done).
examination:
Review the points listed above. Informed consent: Contrast administration Hypotensive agent (Buscopan lnterventional procedure • Bowel opacification • Hydration necessary? • Place needle for CM
orglucagon)
[positioninq • Radiation shielding) Comfortable
protection supine,
of the gonads prone,
or lateral
(apron,
other
decubitus
position • Padding
beneath
head, knees,
• If necessary, gastrointestinal tration is repeated • Buscopan
matomas. Almost all other types of soft-tissue examination will profit from the parenteral administration of contrast medium. Newer techniques rely on the analysis of multiple perfusion phases in contrast-enhanced scans. Bowel opacification is necessary for most abdominal CT examinations.
or glucagon
or lower legs contrast
administration?
adminis-
sound. MRl) should be considered as an alternative. The patient should come in advance if «al contrast will be administered. For CT examination of the upper abdomen. patients should be scheduled some 30minutes before the planned scan. and 60-90 minutes before the scan if the whole abdomen is to be examined with good bowel opacification. When the patient arrives on the day of the examination. the radiologist should review the contra indications to the procedure and inform the patient of potential side effects from the examination (contrast administration. use of antispasmodics or Hi/Hz blockers) or from any CTguided intervention. Gastrointestinal contrast material is administered orally if required. lfthe definitive protocol for the examination has not already been prescribed. it is done at this time. Preparation of Patients with Impaired Renal Function Iodinated contrast media (CM) can induce renal injury through a combination of changes in renal hemodynamics (initial vasodilatation followed by prolonged vasoconstriction) and direct tubular toxicity. Contrast nephropathy (CMN) can occur in patients with risk factors and is generally defined as an increase in serum creatinine >25% or >O.5mg/dl (45[Lmolll) within 48-72 hours compared to baseline. Established risk factors are pre-existing renal insufficiency, diabetic nephropathy, large contrast volumes, and dehydration. Possible risk factors include congestive heart failure, recurrent contrast procedures, and multiple myeloma patients with dehydration. For patients with normal renal function « l.Smg/dl or < 130Amolll) there are no renal
Preparation
contraindications for contrast -enhanced cr. In patients with risk factors. the serum creatinine levels provide an indicator ofrenal function. The creatinine clearance (CCr) as a better indicator of the glomerular filtration rate (G FR) can be estimated using the Cockroft-Cault formula: CCr(mllmin) ~ (140- age) x body weight (kg)/ (serum creatinine 0jimolll) x 0.81) For females a correction factor of 0.85 is usually used. Based on this formula. patients can be adequately stratified in risk categories (Table 3.2). As large volumes of CM are considered an independent risk factor. it is advisable to keep the contrast volume as low as reasonably possible. A useful rule was described by Cigarroa (1 mg/dl = 88 iimol/l): 5 ml CM per kg body weight (max. 300 ml) /
serum creatinine (mg/dl)
Hydration is an important part of prevention of CMN (Table 3.3). The inpatient can be prepared by infusing "100-150 ml/h of 0.45% saline in 5% glucose starting 4-12 hours before the examination. In outpatients. 1000ml can be infused starting approximately 30-60 minutes prior to contrast administration. Intravenous or oral hydration should be continued for 12 to 24 hours after the examination. Urine output should be monitored if possible. There is no associated benefit from using mannitol or diuretics. Nonionic low-osmolar contrast media. often already used routinely for all intravascular injections. are always indicated in patients at risk
of Patients with Impaired Renal Function
for CMN. The use of iso-osmolar contrast medium has not shown additional advantages. It is important to reduce other exogenous risks by withholding nephrotoxic medications such as non-steroidal anti-inflammatory drugs.
Table 3.3 phropathy
Suggestions for prevention of contrast (according to Waybill and Waybill, 2(01)
8» Identification
of patients
High risk patients - Patients < 25m - Patients
with stable
travenous administration. • Estimate creatinine clearance
clearance
creatinine
clearance
with stable
25-50 ml/min and risk factors" Moderate risk patients - Patients
with stable
creatinine
clearance
creatinine
clearance
25-50 ml/min - Patients
with stable
50-75 ml/min and risk factors" - Risk factors - Diabetic nephropathy - Congestive heart failure Recent administration of eM - High volume of eM required
risk
']
and identify
patients who are at moderate (c.f. Table 3.2). pPreparation
of patients
or high risk
at moderate
or high
risk
• Choose alternative imaging modality if possible. • Discontinue non-steroid anti-inflammatory drugs and Dipyridamole 48-72 hours before the procedure. • Withhold diuretics or before the procedure.
ACE inhibitors
for
24
hours
• Hydration in 0.45% saline
moderate risk patients: + 5 % glucose @ 1.0-1.5 ml/kg/hr (based on volume status) beginning 4 hours before procedure, and continuing 8-12 hours after the procedure.
• Hydration in 0.45% saline
high risk patients: + 5 % glucose @ 1.0-1.5 ml/kg/h (based on volume status) beginning 12 hours before procedure, and continuing 12-24 hours after the procedure. of patients
• Use low osmolar • Minimize volume Follow-up
creatinine
11m in
increased
• Measure serum creatinine in patients with (suspected) decrease in renal function, diabetes mellitus, or other risk factors scheduled for any in-
Examination Table 3.2 Stratification of contrast material-induced risk in patients with impaired renal function
with
ne-
at moderate
or high
risk.J
contrast material. of contrast material.
of patients
at moderate
or high
risk
• Discontinue Metformin 48 hours following the procedure, and restart only if creatinine levels are normalized. • Closely monitor urine output; and increase intravenous fluid rate to maintain input greater than output. The goal is to maintain positive fluid balance with hi~ urine flow rate. • If patient develops decreased urine output or progressive increase in creatinine, consult a nephrologist. • Check blood urea nitrogen and creatinine 24 hours after the procedure. Ifthere is any increase, admit the patient,
con-
tinue hydration and observation, and recheck daily until serum creatinine levels have returned to baseline (this may also be done on an outpatient-basis
in selected
cases).
85
86
3 Patient Preparation and Contrast Media Application scheduled shortly before the patient's ular dialysis session.
ACE inhibitors, diuretics, or the antiplatelet agent dipyramidole for 48 hours prior to the examination ifclinicallypossible. Metformin (e.g., Glucophage), an oral antihyperglycemic agent, has attracted interest recently because of the risk of causing lactic acidosis (pH <7.Z5, lactate >5mmol) in diabetic patients with impaired renal function. Current guidelines (e.g., European Society of Uroradiology, ESUR) advise the following: if the study is elective and renal function is normal, metformin should be discontinued for 48 hours following the examination. It can be reinitiated if renal function (serum creatinine) remains normal. If renal function is impaired (> l30'molll), alternative diagnostic modalities should be considered first. If CT is still required, metformin should be stopped and the exam should be postponed for 48 hours. Metformin can be reinitiated 48 hours after the study if renal function has not deteriorated. If the study is urgent and imperative, metformin should be stopped, patients should be hydrated as described above and renal function should be closely monitored. Patients with preterminal renal disease may need an individualized regime for preventive hydration because of an increased risk of developing renal failure. A recent publication suggests that a twoday regimen of hydration combined with acetylcysteine 600 mg twice daily-one day before and on the day of the examination-can protect the kidneys from CMN, even in patients with already decreased renal function. The role of other agents, like theophylline, prostaglandin El, dopamine, and the newer endothelin antagonists, in the prevention of CMN is still unclear. Patients on chronic dialysis can receive a normal contrast dose but the examination should be Table
3.4
Preparation
of high risk patients
140-50
Time
mg (methyl)prednisolone'
300 mg cimetidine2 50 mg
in 20-50 ml
diphenhydramine"
p.o. or i.v.
saline:'
I'
use nonimic
i.v.
contrast
• prophylactic short-term infusion of Hi (and HZ) antihistamine agents • prophylactic treatment with corticosteroids. When LOCM are used routinely for CT, only those patients with a history of a proven moderate to severe adverse reaction to contrast media should be premedicated if there are no alternative options (MRl, US). When corticosteroids are used in elective studies, it is important that the first dose is given at least "lZ hours prior to the CT examination. The chemotoxic effects are probably not improved by these agents. In most strategies the steroids are combined with Hi-antihistamines. There is conflicting evidence about the proved benefit of adding l-h-antihistamines, but at least a theoretical benefit is suggested. When LOCM are only used selectively, they should be used in all patients with known risk factors for adverse reactions: proven history of multiple allergies, bronchial asthma, severe congestive heart failure or recent heart attack, in children,
slim or obese
patients)
Category
of administration 2 h before the examination
Glucocorticoidsteroid
2 h before the examination
Hz-antihistamine
directly
Hi-antihistamine
during low-osmolar
The necessity to prepare patients depends whether or not non ionic low-osmolar contrast media (LOCM) are routinely used. There is controversy as to whether this risk can be reduced pharmacologically in the case of non ionic contrast media, but this has been proven for ionic media. Two basic precautionary measures are available:
12 hand
i.v.
18 G i.v. line Always
Preparation of Patients with Allergy to Contrast Media
(dose has to adapted
Ei~dicatipn
prior to the examination examination
media
A dose of 50 mg prednisolone is equivalent 12 Alternative: 50 mg ranitidine (e.g. Zantace') 13 Dilution at the discretion ofthe radiologist, 14 Alternative: 2 mg clemastine (e.g. Taveqile)
next reg-
to 250 mg hydrocortisone either
as a slow
injection
cr 10 mg dexamethasone or short infusion
Bowel Preparation
diminished renal function, and proven history of any adverse contrast reaction, For patients with a moderate or severe reaction in the past, proceed as above, A practical protocol is outlined in Tab Ie 3.4.
from these medications may be as high as the risk of llT in unselected patient groups, Premedication (Table 3.5) can be useful, however, in high-risk patients with known hyperthyroidism caused by Graves' disease, an autonomous adenoma in multinodular goiter, or a papillary or follicular thyroid carcinoma, Patients who are scheduled for diagnostic thyroid scintigraphy, or who are in the course of therapy with radioactive iodine should not undergo CT studies with iodinated contrast media without physician consultation, Injecting iodinated contrast material in non-premedicated patients with papillary or follicular thyroid cancer will make radioactive iodine treatment ineffective for a number of months and may seriously influence the overall prognosis of the disease,
Preparation of Patients with Hyperthyroidism Iodinated contrast media may contain free iodide in concentrations up to 20 [ig/ml apart from the organic iodine content During a routine CT examination there may be an administration of 2-3 mg free iodide, which corresponds to 10-40 times the minimum required dailyintake. In iodine-deficient areas (e.g. , parts of central Europe) there is ahigher incidence of multinodular goiter and autonomous thyroid tissue that may result in (non overt) hyperthyroidism especially in the elderly, When such patients are subjected to an excessive iodine load, this may result in iodine-induced hyperthyroidism or thyrotoxicosis (Il'T): which is often difficult to treat There are few controlled studies on the risk of developing 1lT after exposure to iodinated contrast media, One such study in unselected patients showed a very low incidence of llT of 0,03-0,2% with cases of llT occurring both in euthyreotic and hyperthyreotic patients, Prophylaxis is a matter of controversial discussion, even in patients from iodine-deficient regions, because the incidence of side effects
Table 3.5 cording
Possible
prophylactic
strategies
for
patients
Bowel Preparation The contents of the gastrointestinal tract can hamper or prevent evaluation of the stomach, duodenum, and colon on CT scans, Unless there is special emphasis on these organs, no special preparation is needed but it can be recommended to discontinue solid foods some 3-4 hours prior to the examination, In special investigations of the stomach and pancreas, solid foods should be discontinued the evening before the examination, The patient may continue to take oral medications, however. This makes it possible to achieve optimum dis-
at increased
risk for
iodine-induced
hyperthyroidism
to Hehrrnann. 1996)
Patients at risk Hyperthyroidism, Graves Latent hyperthyreodism Papillary thyroid cancer Follicular thyroid cancer Elective
CT
Sodium
Perchlorate
disease in patients
with
autonomous
adenoma
studies:
Thiamazol
3 dd 300 mg 1 dd 30 mg
Start 1 day before CT and continue for 8-14 days Start 1 day before CT and continue for 28 days
EmergencyCTstudies: Sodium
Thiamazol
perchlorate
1 dd 800 mg 1dd30mg
Therapy should be clinically Barium 2% should be used
controlled for bowel
Directly Directly
priorto CT and continue prior to CT and continue
with measurements opacification
of serum
with 3 x 300 mg for 8-14 days for 28 days
free T sand
TSH
(ac-
87
3
Patient Preparation
Table
3.6
'Stomach,
and Contrast Media Application
Bowel cleansing duodenum,
reduce motion artifacts. One should consider protecting highly radiosensitive areas such as the breasts and gonads with lead shielding in those examinations in which they are outside the scanned field. A wrap-around lead apron may be used in place of a gonadal shield. Keeping a lead shield in place within the scan range will cause major artifacts and mandates a repeat examination of that region. It has been reported, however, that partially radio-opaque shielding with bismuth garments (for the breasts) allows the shield to be kept within the scan field. To avoid artifacts, at least one arm (preferably both) should be removed from the scanning field (e.g., extended above the head). All metallic objects (zippers, wallets, ECG electrodes, metalreinforced ventilation tubing, etc.) should also be removed from the field.
for CT examinations
pancreas
No solid food for 12 h No food or liquids for 6 h Small
intestine
(CT enteroclysis)
No solid food for 24 h Low-residue diet High fluid intake (31/d) Mild laxative on day before
examination
Colon No solid food for 2 days, Laxative medication preparations) Mini-enema
bu.
copious
for 2 days
on morning
liquids
(preferably
'dry'
of examination
tension of the stomach and duodenum following the administration of butylscopolamine or glucagon and (negative) oral contrast medium. For special CT examinations of the small bowel (CT enteroclysis), solid foods are discontinued 24 hours before the examination. Although multiple methods will provide good results, a regimen of a low-residue diet, high fluid intake (3000 ml/day) and a combination of laxative medications (magnesium sulfate and bisacodyl) for one day is simple and works well. (Table 3.6). Special CT examinations of the colon (CT colonography) require the same kind of bowel preparation that would precede a barium enema. The regimen suggested above (low-residue diet, high fluid intake, and combination of laxative medications) can be used for two days, but prepackaged kits (e.g., Fleet) are also available (Table 3.6). "Dry" preparations, typically given for barium enemas, are superior to "wet" preparations (e.g. Golytely), typically given for colonoscopy, because there is less fluid left within the colon, and air can be used as a negative contrast agent.
Vascular Access Peripheral Venous Access
Intravenous access is established with a sufficiently large cannula (18-20 G), preferably placed in an antecubital vein. The vein should be substantial enough to accommodate the 3-5 mlls injection rate that is used in most protocols. Higher flow rates mandate larger cannulas(l6-17G). Central Venous Access
_
A central venous catheter (triple-lumen catheter or other catheter with a 6 G lumen) is suitable for contrast injection only at relatively low flow rates of 1.5-2.5 mlls (see manufacturer's guidelines). Depending on the type and length of the catheter it may become ruptured or dislodged due to excessive flow rates. If doubt exists, a trial NaCI injection should be performed manually to test catheter function, and the power injector should be coupled to a pressure-limiting device. Large central venous catheters such as an introducer sheath, a dialysis catheter, etc. allow for higher flow rates of 4 mlls or more, again depending on the catheter type used.
Positioning includes placing the patient in a comfortable, symmetrical position and providing adequate radiation protection. Padding should be placed beneath the head, knees, or lower legs to keep the patient comfortable and
Femoral venous access may be useful for CT angiography of the thoracic aorta but it is generally avoided where possible because of the higher risk of extravasation associated with
Positioning
Femoral Venous Access
Planning the Scan such a technique when performed on the CT table. It should be noted that many ordinary antecubital venous cannulas are not long enough to provide secure access in obese patients. In these cases a small insertion set can be used to thread a somewhat longer catheter safely through the femoral vein into the iliac vein (i.e.. using the Seldinger technique). If femoral access is required or if contrast medium must be injected through a pedal vein (especially in children) for abdominal scanning. it should be noted that high-contrast streak artifacts can persist in the inferior vena cava for up to 10 seconds after the end of the injection and can hamper the evaluation of adjacent structures. For this reason. lower contrast material concentrations can be recommended in these situations. •
Planning the Scan Planning the examination should be done as early as possible so that the patient can be adequately prepared and any missing information can be acquired prior to the examination. A scanning protocol that is appropriate for the scanner and for the actual requirements of the examination should be formulated and documented. In most cases. one will chose from a set of standard protocols. which may be modified according to individual requirements (e.g., in slim or obese patients, patients with impaired renal function or bad venous access). Such standard protocols should include information about patient preparation, data acquisition and image reconstruction technique as well as 3D postprocessing and filming. We recommend taking a written record of the individual scanning protocol on the request form, using an abbreviated code that simplifies the adaptation of standard protocols. We suggest using the notations listed in Table 3.7. This information should also be included in the patient comment on the CT scanner so that any subsequent problems can be investigated retrospectively to find explanations for unexpected effects in a particular examination.
Arterial Access
Certain examinations. such as CT during arterial portography (CTAP). require the placement of an intra-arterial catheter under fluoroscopic guidance in an angiography room. On the CT table a scanogram should be taken prior to contrast administration to recheck the catheter position.
Table 3.7 Scan
Suggested
notation
for examination
Spiral CT
parameters
Example
parameters
SCiTF/RI f /
5/8/4 f
Acquisition:
NXSCITFfi
Reconstruction:
SWIRl ax/cor/sag
4X1/5.5 T 5/4 ax, 1.5/3
Multislice CT:
sc=
TF = table feed direction, SW section width (mm), ax/cor/sag section
collimation
(mm),
A
Contrast
cor
(mmlrotation), =
RI
thin section
acquisition
5 mm axial,
1.5 mm coronal
=
reconstruction
interval
reformations
(mm),
f /\
«scan
section orientation
Example
parameters
Standard:
V/FID
100/2/100
Conrasr dilution:
V:NiFID
60:60/4/30
1: 1 dilution
= 150 mg/ml
NaCI bolus
V+NiFiD
100+50/4/20
SOml saline
flush
Bolus triggering:
V!FIDX
120/4/5A
bolus
Biphasic injection:
Vi/Fi+V2/Fz/D
50/5+50/2/20
fast injection
Biphasicscan:
V/F/Di+Dz
150/4/20+
arterial
70
triggering
S s after aortic
followed
and portal
threshold
by slower
phase
injection
scans
V= contrast volume (ml), F -flow rate (mils), De s can delay (s), Ne volume (ml) of dilution medium bolus, DX" delay after reaching trigger level at position X, contrast concentration 300 mg iodine/ml
or NaCI
Patient Preparation
and Contrast Media Application
To optimize image quality and to minimize radiation exposure of the patient either automated programs can be employed (most of which are just about to enter the market), or estimates of the required mAs settings based on the patient's size and weight may be used, Most of these approaches are still in their infancy, and no general recommendations can be made (see Chapters, p, 149), Documentation ofkVp, mAs, scanning mode, pitch and section collimation is
Gastrointestinal
Contrast
Media
Bowel opacification is necessary for the adequate evaluation of the gastrointestinal tract and for differentiating it from other structures of soft-tissue density, With good bowel opacification, pathologic changes in the bowel wall and lumen can be appreciated on the opacified scans.
Types of Gastrointestinal Contrast Media Contrast media for the bowel are classified as positive or negative, depending on whether the material is hyperattenuating or hypoattenuating relative to the walls of the gastrointestinal tract (Table 3,8), Water or methyl cellulose preparations are negative contrast media that facilitate evaluation of the mucosa following intravenous administration, They also do not superimpose on abdominal vessels and are therefore becoming a new standard for multislice CT of the abdomen,
Table
3.8
Types
of gastrointestinal
Medium contrast
(non-carbonated)
Methyl
cellulose
Paraffin
media
vegetable
Air
oils
pancreas
Tumor
diagnosis
Small
bowel,
colon
CT enteroclysis
Small
bowel,
colon
Recta!
Colon,
suspension
Applications
Stomach,
preparations
suspension,
Positive contrast
For use in CT examinations water or juice should not be carbonated, Methyl cellulose preparations can be administered orally (given adequate additives for taste) and have the advantage of a higher viscosity, The higher viscosity makes methyl cellulose ideally suited for CT enteroclysis or colonography. Water-based negative contrast media, however, provide insufficient detail if the bowel is not fully distended, and can hamper the detection of cystic lesions in the female pelvis, For this reason they are often used in conjunction with a spasmolytic compound (n-butylscopolamine, e,g" Buscopan) or are administered by enema or pump injection, Oily or fat-containing contrast media are not recommended for oral administration because of their unpleasant taste, which may lead to noncompliance, However, (full fat) milk has been described as a useful contrast medium for the upper abdomen, For administration by enema (colon) or pump injection (small bowel
media
Water
BaS04
contrast Region
Negative
required to be able to determine the patient's individual radiation exposure, If available, it is easier to use the volume CT dose index (CIDlYoi) instead, This number is an indicator of the locally applied dose and is displayed on the scanner console of most modern CT units for the specific protocols selected by the user. However, some manufacturers only display the CIDlYoiif explicitly asked by their customer.
stomach
application,
Only for virtual
poor oral compliance
endoscopy
media Stomach
+ bowel
Stomach
+ bowel
Universal
(contraindicated
if perforation
(contraindicated
in hyperthyroidism)
may be
present)
lodinated
solutions
Universal
Technique
enteroclysis) an oily contrast preparation improves the contrast between the bowel wall and the lumen, and makes it easier to distinguish between extra-enteric fluid and bowel, Air or COA as contrast media produce a very high negative contrast, which is particularly useful in virtual endoscopy of the colon (or stomach), C02 has the advantage over air that it is readily absorbed by the body and is eliminated by respiration, It induces less spastic response of the bowel wall and therefore is better tolerated by most patients, In addition, colon distension was reported to be superior with C02 as compared to air. Bowel distension can be further improved with injection of 20-40 mg of n-butylscopolamine (e.g. Buscopan) as soon the first symptoms (bowel colics) occur, Barium sulfate (BaS04) suspensions and l'odinated solutions are positive contrast media that have universal applications, Intestinal structures are clearly identifiable as such even when they are poorly distended, One disadvantage of positive contrast media is that they make mucosal surfaces more difficult to evaluate following IV administrationofcontrastmateriaL Barium suspensions have an agreeable taste and reportedly produce somewhat better opacification of the upper gastrotntesttnal tract They are contraindicated, however, in patients who may have a perforated bowel (clinical suspicion, immediate postoperative period, or after endoscopic intestinal biopsy), Iodinated contrast media should be avoided in hyperthyroidism and are contraindicated when patients are scheduled for scintigraphy or radiotherapy with iodine 131. Technique of Gastrointestinal Administration
Contrast
With few exceptions, oral contrast medium should be used routinely forCTexaminations of the abdomen. With spiral CT the bowel is opacified with a positive contrast medium, in which case an iodinated solution or BaS04 suspension may be used with equal success, Care should be taken that the medium is sufficiently diluted (3-4% for agents such as Telebrix Castro or Castrografin, 2% for barium suspensions), Spasmolytic agents are administered only if adequate distension of the duodenum is required,
of Gastrointestinal
Contrast Administration
Iodinated contrast media produce better opacification of the stomach and duodenum if the last cup of oral contrast is more highly concentrated (5%~IOml in 200ml water) and is ingested in the examination room, This will correct for the frequent increase in gastric secretions that occurs before the examination, For examinations of the upper abdominal organs, only the proximal small bowel requires opacification, This is accomplished by having the patient drink 500ml of contrast medium over a 30 minute period, Examinations covering the entire abdomen also require opacification of the distal ileum and perhaps the colon (for pelvic imaging), achieved by having the patient drink I to LSI of contrast medium, The doses should be fractionated to distribute them evenly over the ingestion period of 60 to 90 minutes prior to the examination, Starting the scan too early leads to insufficient opacification of the distal bowel segments, while an excessive delay between the last ingested portion of contrast medium and the CT scan leads to contrast problems in the proximal small bowel. Optimum bowel opacification requires accurate timing and adequate patient instruction, Excellent colon opacification is achieved when oral contrast administration is started the evening before the examination day, With their appointment, patients receive a bottle of 20 ml of contrast agent that they have to dilute to 600mL This can be taken with food in two portions and gives a good large bowel delineation, With multislice CT, especially when thinsection imaging is performed, delineation of the abdominal vessels becomes more important and requires use of negative contrast agents, Distinction between lymph nodes, masses and nonopacified bowel loops is less of a problem because of the increased spatial resolution of multisliceCT For the stomach and pancreas we recommend methyl cellulose preparations because of their increased viscosity, while water is sufficient for the upper abdominal examinations, The contrast material (500-1000 ml) should be given over a short period of time to ensure proper distension of the stomach and duodenum, For examinations of the whole abdomen, larger amounts of negative contrast material are required (1-1.51), and should be given 30-60 minutes prior to the examination, For the lower abdomen, some authors suggest a
91
92
3 Patient Preparation
and Contrast Media Application
combination of 500-1000 ml of positive contrast material given over a period of 30-90 minutes prior to the examination, followed by 500-1000 ml of negative contrast material given during the last 15 minutes prior to the scan, Sufficient time between ingestion of the two types of contrast material is required to avoid mixing, Even then, some portions of the small bowel or even the stomach may retain a
Table
3.9
Gastrointestinal
Region
contrast
luminal opacification that is identical to the attenuation of the bowel wall, which makes it impossible to evaluate mucosal or intraluminal abnormalities, Special techniques to achieve optimum opacification of the stomach, small and large bowel are summarized in Table 3,9 and are discussed in Chapter 15, Gastrointestinal Tract
administration
Anti
Technique
Comments
spasmodics Standard
protocols
Upper abdomen
500 mI positive
contrast
medium
Taken orally in fractionated over 30min period
amounts
Abdomen
1-1.51
contrast
medium
Taken orally in fractionated over a 60-90 min period
amounts
medium
Done occasionally
Special
positive
GI protocols 200 ml positive
Esophagus
contrast
to mark the lumen
on examination table or thick barium paste on exam table Stomach
500 ml water
before
250 ml water
on table
Tumor detection/staging In left lateral decubitus
exam
repeat for contrast Small
bowel
CT enteroclysis
1.5 I methylcellulose or 1.51 highly dilute BaS04
Rectum
500 ml positive
contrast
rectally
staging
in the small
1-1.51 positive contrast rectally or 1-1.51 negative contrast
Inflammatory disease Tumor staging/inflammatory
Stomach
3 packets
N.p.o. for 6 h
Coon
Rectal
Virtual
catheter
Done only to mark the rectum tumor
Colon
position,
series
via duodenal
for
pelvis disease
endoscopy
Parenteral
of effervescent
powder
air or C02 insufflation
Contrast
Bowel preparation
essential
Media
Intravascular contrast medium is necessary for most CT examinations of blood vessels and parenchymal organs, The contrast volume and mode of contrast administration vary with the nature of the study, CT scanning with intravascular contrast may have any of six objectives:
• interstitial enhancement (inflammatory lesions, tumors) • Urinary enhancement (excretion, pyelocalyceal system) • perfusion analysis (lesion characterization, tissue perfusion),
• CT angiography (vessel evaluation) • vascular opacification (to differentiate vessels from lymph nodes in nodal staging) parenchymal enhancement (lesion detection and characterization)
The same contrast media can also be used for the opacification of anatomic and pathologic spaces, like the spinal canal, lower urinary tract, pleura, peritoneal cavity, abscesses, or fistulas,
Complications of Contrast Material Injection
Number
Total
population
Known allergy Known
contrast
allergy
Severe adverse
reactions:
of cases
All adverse Ionic
169,284/168,363 12,913115,058 5785/9667
12.7%
3.1%
0.22%
0.04%
23.4%
6.9% 11.2%
0.53%
0.10%
severe
dyspnea,
Contrast
fall in blood
Media
Ionic contrast media are moderately well tolerated, depending on the injection rate used for IV administration (Table 3.10). Some patients experience warmth, pain, or burning at injection rates higher than 2mIls. For this reason, ionic contrast media are not recommended for use in spiral or multislice CT. They are still used in intracavitary applications such as fluoroscopic studies of the bladder and bowel. In many Western countries, most radiological departments have completely switched to nonionic media for all (intravascular) indications and therefore only few ionic media are still on the market. •
Nonionic
low-osmolar
Contrast
IMonionic
44.0%
Typesof Contrast Media Ionic High-osmolar
reactions
(ioniclnonionic)
Media
Nonionic contrast media are generally preferred over ionic media because of their lower osmolarity and significantly lower rate of adverse events (Table 3.10). There is also a lower incidence of reactions related to chemo- or osmotoxicity during contrast injection, making it safe to use rapid injection (flow) rates. The main disadvantage of nonionic media is their higher cost. Most contrast media used for intravenous application are monomeric molecules. They are in general better tolerated than dimeric molecules despite the fact that the latter are nearly iso-osmolar. Dimeric agents have a substantially higher viscosity and different hydrophilicity, and need to be heated up to body temperature prior to intravenous injection. Dimeric contrast agents are preferentially used for intrathecal and intra-articular applications (eT arthrography).
pressure,
cardiac
Complications Injection Contrast
0.72% arrest.
0.18% ansciousness
of Contrast
Material
Extravasation
The commonest complication of intravascular injections is contrast extravasation. Most often, this involves small volumes « l Oml), but when large volumes extravasate serious injury to skin and subcutaneous tissues may be produced. A number of risk factors have been identified: non-communicative patients (children, elderly), severely debilitated patients, multiple punctures in the same vein, and injections on the dorsum of the hand and foot. Treatment guidelines are outlined in Table 111.
Adverse Reactions to Contrast Media Parenteral contrast media can incite a variety of adverse reactions (Table 3.12), which may occur early (usually < 20 minutes) or late. The reactions
Table
3.11
Treatment
of contrast
extravasation
of the affected extremity heart • Ice packs - 15-60 min applications,
• Elevation
above the three
times
dally (day 1-3) • Close observation for 2-4 hours • Inform referring physician Plastic
surgery
consultation
when
• Extravasated volume >30 ml of ionic highosmolarCM • Extravasated volume >100 ml of nonionic lowosmolarCM • Skin blistering • Altered tissue perfusion (capillary refill) or change in sensation in hands
93
3 Patient Preparation and Contrast Media Application Table media
3.12
Symptoms
K Early reactions
of adverse
reactions
to contrast
(up to 60 min postinjection)
Nausea or vomiting Urticaria Diffuse erythema, angioedema Bronchospasm
pulmonary
edema
Respiratory arrest arrest
(
reactions
• Muk.pie
Most
(60 rrm to 3 days post-
irjection) Skin reactions
• Skin rash, itching, Systemic
risk factors
for adverse
reactions
risk factors
Chemotoxic
shock
Cardiac Delayed
General
• Previous • Bronchial
(requiring treatment)
Anaphylactic
media
Anaphylactoid
Vasovagal reaction Laryngeal edema, Hypotension
Selected
to contrast
• Ionic contrast media • Overt hyperthyroidism (contraindication) • Female gender (especially delayed reactions)
Mild to moderate
Severe
Table 3.13
• • • • •
reactions
serious contrast reaction asthma requiring treatment allergies (atopy) requiring treatment reactions
important
ones
include:
Cardiovascular instability Congestive heart failure grade III-IV Diabetic nephropathy Autonomous thyroid nodules in elderly Liver and kidney disease
swelling
reactions
• Headache, dizziness • Nausea, diarrhea • Chills, rigors
• Flu-like symptoms
Delayed Arm pain " Other reactions
may have an anaphylactoid etiology but can also result from osmolar effects (e.g .. direct irritation of the venous wall) or the number of carboxylic and hydroxylic side chains (chemotoxicity). The incidence of adverse reactions depends on whether the medium is ionic or nonionic and on the specific product that is used. It is also influenced by the concentration and volume of the contrast medium. the rate of contrast injection. and a number of patient-related factors (Table 3.13). Published incidence data vary dramatically depending on how adverse reactions are defined. A possible grading is as follows: • Minor reactions: nausea, vomiting, heat sensation. flushing. mild urticaria. sneezing • Intermediate reactions: mild hypotension. generalized urticaria. mild bronchospasm. diffuse erythema. angioedema. vasqvagal reaction • Severe reactions: hypotension (systolic pressure <70mmHg). pulmonary edema. epiglottic edema. severe bronchospasm. cardiac arrhythmia, cardiac arrest.
As a general rule, nonionic contrast media are well tolerated but are still associated with a 2-4% incidence of early adverse reactions and a 4-30% incidence oflate reactions (depending on definition). According to the study by Katayama in 1990, most early reactions are mild to moderate (3.1 %), and only 0.4% are severe (Table 3.10). The incidence of very severe reactions (causing critical illness or death) was 0.1 %. Late reactions are not uncommon but occur more frequently in patients who received interleukin2 therapies (10-30%). They often cause nonspecific symptoms like rash, mild fever, dizziness, chills, or rigors. In most cases no or only supportive treatment is indicated. Cutaneous
Reactions
Patients with cutaneous reactions are usually cooperative. Urticaria are anaphylactoid reactions resulting in raised focal areas of the skin and are most common in the face, neck and chest. They are associated with pruritus (itching) and can be mild or generalized. Patients may also develop diffuse erythema or subcutaneous (angio)edema that may develop into hypotension or alrway edema and therefore should be closely monitored. Respiratory
Reactions
Respiratory reactions usually result in agitation. Airway or laryngeal edema is associated with a feeling of tightness in the throat, a change of
Treatment of Adverse Reactions voice, or even inspiratory stridor. It can be preceded by swelling of lips or tongue, Bronchospasm can be seen in asthmatics and cause dyspnea with wheezing and difficulty exhaling, Pulmonary edema may be seen in patients with underlying congestive heart failure, The symptoms often develop slowly and dyspnea is the most important symptom Anxiety is a diagnosis of exclusion, but it can cause hyperventilation with chest tightness, •
Cardiovascular
Reactions
These are usually the more dramatic reactions in which patients are looking for help but may become unresponsive, Hypotensive patients often complain of light-headedness or weakness, Mild reactions often respond well to fluids and leg elevation, Vasovagal reactions are associated with bradycardia (pulse < 60bpm) and are more frequent in younger males, while tachycardia is suspect for an anaphylactoid reaction, It may be difficult to differentiate these two in patients that use (3-blocking agents, If patients do not improve after initial measures or if they become unresponsive, direct assistance should be sought
Treatment of Adverse Reactions
Oxygen should be administered
at high dosage (l0-1211min) via a face mask, preferably with an oxygen reservoir. This enables administration of 100% concentrations, Nasal cannulae are suboptimal for delivery of such high doses, All patients should receive a high dose regardless of the existence of chronic obstructive pulmonary disease, HI and HZ antihistamines are used primarily to reduce symptoms of skin reactions such as urticaria and diffuse erythema, They can also be used as a second-line drug for respiratory reactions after epinephrine has been given, Corticosteroids do not play an im portan t role in the treatment of the acute reaction, They are effective in prevention of delayed recurrences for as long as 48 hours, especially in respiratory reactions, When administered, a high loading-dose of 500-1000 mg hydrocortisone is advised which can be followed by constant infusion of 500 mg in 250ml saline at a rate of 60mllh, Overdosing in the acute phase is generally not problematic, 500-1000 mg hydro cor-
tisone is therapeutically equivalent to 125250 mg Prednisolone, 100-200 mg methylprednisolone or 20-40 mg dexamethasone, Epinephrine (adrenalin) is the most important medication, The 1:1000 dilution (l mg/ml) is used for subcutaneous or intramuscular injection for milder symptoms and is often available in 0,3 ml preloaded syringes (e.g, , Epipen ®), The 1: 10000 dilution (0.1 mg/ml) is used for intravenous injection in moderate to severe symptoms and may also be available in preloaded syringes, In a recent UK report from the Project team of Resuscitation Council, IV administration by non experienced personnel is strongly discouraged and requires monitoring, Instead, 1M injection of 0,5 ml 1: 1000 adrenalin is recommended, Care should be taken in patients using [3-blocking agents, as the selective a-adrenergic stimulation with vasoconstriction may lead to a hypertensive crisis (use isoproterenol instead), A detailed outline of the most important reactions and suggested treatment is given in Table 3,14,
Table 3.14 Treatment suggestions for adverse reactions to contrast media (protocols have to be checked with local regulations and available medication)
Ht.utancous symptoms Nausea or vomiting • No treatment needed Mild or asymptomatic urticaria: • No treatment needed Generalized
or symptomatic
urticaria:
• 50 mg diphenhydramine po/im/iv ·300 mg cimetidine or 50 mg ranitidine saline slowly iv Angioedema
or diffuse
n 20 ml
erythema:
• 50 mg diphenhydramine iv ·300 mg cimetidine or 50 mg ranitidine in 20 ml saline slowly iv • 0.5 ml epinephrine 1:1000 im-repeat after 5 min-if patient is hypovolemic • Saline 0.9% or Ringer's lactate rapid iv Respiratory
symptoms
Mild laryngeal
edema:
• Oxygen 10 I/min by face mask • 0.5 ml epinephrine 1:1000 im-repeat
after
5min Severe
laryngeal
edema-add:
• 300 mg cimetidine or 50 mg ranitidine in 20 saline slowly iv - Call code-consider tracheal intubation
ml
Continue
A
95
3 Patient Preparation and Contrast Media Application Table
3.14
Contrast Injection
(Continue)
lllliild
bronchospasm: • Oxygen 10 Ifmin by face mask • Albuterol, terbutaline or metaproterenol doseinhaler 2-3 deep inhalations (or nebulizer Smg i1 2 ml saline)
Severe bronchospasm-add: • O.Smlepinephrine 1:1000 • 500-1000 mg hydrocortisone (optional-see text) • Call
im-repeat
code t patient tracheal intubation
emergency
-consider I; Cardiovascular
after 5
rrm
iv or equivalent does
not respond
symptoms
Hypotension
with
bradycardia
(vagal
reaction):
• Elevate legs and release any abdominal pression • Oxygen 10 Ifmin by face mask • Saline 0.9% or Ringer's lactate rapid iv
com-
• 0.5-1.0 mg atropine slowly iv-repeat every 5minupto3mg Mild, isolated hypotension with tachycardia: • Elevate legs and release pression • Oxygen 10 I/min by face • Saline Severe
0.9%
or Ringer's
hypotension
with
any abdominal
com-
mask lactate
rapid
tachycardia
iv (shock)-
add: • 0.5 miepinephrine 1:1000 im-repeat after 5 rrm • 400 mg dopamine in 250 ml glucose 5%-0.2 mil mn infusion by pump (or 2-5 u,g/kg/min) • Call emergency code if patient does not respond Pulmonary edema: • Elevate head or have patient sit up • Oxygen 10 Ifmin by face mask • 40-80 mg furosemide slowly iv • 500-1000 mg hydrocortisone iv or equivalent (see text) • Call emergency
code
if patient
does
Angina: • 0.4 mg nitroglycerin sublingual • Oxygen 10 Ifmin by face mask • Call emergency code for cardiologist Cardiac arrest: • Start CPR • Call emergency Neurological
code
ECG
Anxiety reaction: • Monitor vital signs and reassurance • Paper bag for breathing (when hyperventilating) • If severe -2.5 mg midazolam or 5 mg diazepam slowly iv aspiration slowly
iv
All adverse reactions should be adequately documented in the radiology report-type of CM used, patient symptoms, 2 mg Clemastine dramine
iv
I 2 3 4 5
Contrast volume V (ml) Flow rate F(mlls) Scan delay D (s) Saline flush N (ml) Position X (ofthe reference region for bolus triggering) 6 Concentration C (mg iodine/ml) 7 Osmolarity (osmoll1) 8 Viscosity (kP) Parameters I through 3 are the main ones for contrast injections, while parameters 6 through 8 are specific for the product used and generally are not modified by the examiner. The injected contrast material can be more fully utilized for contrast enhancement ifthe contrast material is pushed forward and the injection vein is flushed by a bolus of isotonic NaCI (normal saline, parameter 4), Ifthe initiation of scanning is determined by the appearance of contrast medium in a reference region X (parameter 5), the scan delay D (parameter 3) denotes the interval from that point until a designated level of enhancement is achieved in that region, The contrast parameters should be documented as a comment on the images, as they may have important implications for image interpretation, We suggest using the notations from Table 3.1,
Intravenous and
immediately
• Oxygen 10 Ifmin by face mask ·2.5 mg midazolam or 5 mg diazepam
The most important parameters for intravascular contrast injection are:
not respond
symptoms
Seizures: ! Tum on side to avoid
Parameters
therapy, outcome and follow-up. iv alternative to 50 mg diphenhy-
Contrast Administration
In the great majority of cases, parenteral contrastmedium is administered by an intravenous route, Such an access route is quite long and leads to a defined sequence of vascular and organ enhancement with various mixing processes along the way to the target organ,
Basic Principles Contrast material has to flow through the injection veins into the vena cava, enter the right atrium, pass the pulmonary circulation and finally arrive in the aorta, Mixing with nonopacified blood occurs along the way to the right atri urn, In the rightventricl e, mixing 0 fop a cifi ed and non-opacified blood should be completed,
Intravenous Contrast Administration Once the contrast material has entered the aorta, it enhances various capillary beds during an arterial phase, enters draining veins that eitherjoin the vena cava or enter the portal venous system. Contrast material in the portal system enhances the liver parenchyma and drains into the liver veins before it reaches the right atrium again. As the contrast material flows back to the right heart from various organs, recirculation effects occur. Typical arrival times for various organ systems are shown in Table 3.15. Aortic enhancement ideally reaches a plateau phase shortly after the contrast material arrives in the aorta. The slope of this initial enhancement is an indicator of the cardiac output of the patient but may be slowed if there are any obstructions along the venous inflow. As more contrast medium is injected, the aortic "plateau" increases further from cumulative effects that culminate as the final amount of contrast material enters the aorta. Thus, the aortic enhancement "plateau" is not a real horizontal plateau but increases as the scan progresses (Fig. 3.1). Peak aortic enhancement is seen at the end of this phase. This peak enhancement should not be regarded mistakenly as the ideal time for imaging the arteries but it is well suited for imaging arterially perfused organs. Since the spontaneous flow in the injected veins is often slower than the injection rate, the inflow of contrast material slows as the injection is completed. This leads to a premature end
Table 3.15 Contrast arrival times after injection into the right cubital vein 6-12s 9-15s 13-20s 15-22S 16-24s 18-27s 22-33 s 22-30 s 22-30 s 24-32 s 120-250 s 30-45 s 35-50 s 50-80 s 120-250 s
Right atrium Main pulmonary artery Left atrium Aorta Carotids Renal arteries Femoral arteries Jugular vein Renal veins Suprarenal IVC Infrarenal IVC Splenic vein Mesenteric veins Liver veins Femoral veins
of the aortic plateau and does not take full advantage of the injected contrast material. A saline flush, injected at identical flow rates immediately after contrast administration, pushes the contrast material forwards and thus prolongs the aortic plateau phase. In theory, the time from contrast arrival to peak enhancement in the aorta should last about as long as the injection time (= VIF), but it often ends earlier in patients with a high cardiac output and lasts longer in patients with a low cardiac output. Most organs have an exclusively arterial blood supply. Only the liver and the lungs have a
300
200
Liver veins
100
Fig. 3.1 Contrast enhancement for the liver: example of a time-density curve for 150ml contrast material injected at 4mils.
o
20
40
60
80
100
120
140
Time (5)
3 Patient Preparation and Contrast Media Application
dual blood supply. In the lungs, there is a minor contribution from the bronchial arteries, which can increase in patients with pulmonary hypertension. In the liver, the main blood supply is via the portal vein (75-90%), and the remainder via the hepatic artery. If there is obstruction to portal flow or ifhypervascular lesions are present, arterial flow will increase. Organs with purely arterial perfusion such as the pancreas, the bowel, the bladder, or the adrenals are best imaged during peak organ enhancement, which occurs some 5-15 s after peak aortic enhancement. The kidneys are an exception because their perfusion characteristics are also determined by contrast excretion, and homogenous enhancement requires scanning in the nephrographic phase (80-100 s post injection). The liver requires scanning in at least the portal venous phase, which is always a phase that combines arterial enhancement with enhancement via the portal veins. Lesion detection and characterization is markedly improved if arterial phase scanning is performed as well. This phase should be timed so that there is optimum enhancement of tumor capillaries but there is still little portal enhancement.
Injection Parameters and Organ Enhancement The initial level of the aortic contrast plateau is important for imaging the arteries (CT angiography). It increases with the amount of iodine injected per second, i.e., with flow rate and contrast material concentration (see Fig. 3.4). If only very small contrast volumes are injected « 50ml in most patients with a normal circulatory state), the contrast enhancement curve (time-density curve) of the aorta shows only a single peak and does not level off (see Fig.3.3b). The maximum enhancement of this curve strongly depends on contrast material volume, a fact that is important to know if a very short scan duration is sufficient, such as with 16-slice CT angiography. For such a purpose, too Iowa volume of contrast material yields insufficient contrast enhancement. The peak aortic enhancement is important for imaging arterially perfused organs and tumors. It, too, increases with flow rate and contrast material concentration, but also, to a lesser degree, with the total volume of injected contrast material (see Fig. 3.3 b).
The duration of the aortic enhancement plateau increases with longer injection times, i.e., it increases with larger volumes and slower flow rates. In practice, there is a trade-off between the level and the duration of aortic enhancement. The faster the scanners (multislice CT), the higher are the flow rates and, thus, the higher the enhancement that can be maintained during the scan. At the same time, however, optimum timing of the scan becomes cruicia!. Portal venous enhancement depends mainly on the total amount of iodine injected. There are reports that suggest that faster flow rates increase early portal venous enhancement. For biphasic or multiphasic liver studies, the flow rate ("4 mlls) determines arterial enhancement, while the iodine load (=" 2 mllkg body weight at 300 mg iodine/ml) determines portal enhancement (see Fig. 3.4). Parenchymal enhancement of arterially perfused organs (e.g., pancreas) depends mainly on peak aortic enhancement and the total amount of iodine injected. The most important patient-related factors are weight, cardiac output and venous (or arterial) inflow obstruction. The higher the cardiac output, the greater the dilution effect and the lower the level of vascular enhancement. In patients with low cardiac output (see Fig. 3.6), arterial enhancement levels can be very high but one has to keep in mind that vascular and parenchyma! enhancement can be substantially delayed (the aortic plateau starting as late as 60s post injection). Thus, proper timing is an important factor (see below). Patient weight has moderate influence on peak aortic enhancement but strongly influences organ enhancement (pancreas and liver).
Contrast Injection For most contrast-enhanced CT studies, a mechanically controlled injection is necessary to achieve optimum results in terms of enhancement levels and contrast utilization. Peak vascular and organ enhancement and the duration of the plateau phase can be controlled by the proper selection of contrast volumes and flow rates. Based on the short scan time in spiral or multislice CT, imaging should be performed during a specific phase of enhancement (Ta-
Intravenous Contrast Administration Suggested start delay for specific contrast Table 3.16 phases (35 s duration of contrast injection)
Delay
Phase
Triggering' SA
Arterial phase (CTA)
205
SA
Portal venous phase
60s
155
Hepatic venous
80s
Hepatic
arterial
phase
phase
Parenchymal (pancreas)
phase
405
20A
Parenchymal
phase
405
20A
255
lOA
(bowel) Corticomedullary Nephrographic
phase
1005
phase
Excretion phase
5-15min
Systemic venous phase
150s
20V
(IVe) 1
Trigger ROI: A
cava
=
=
aorta,
S=
spleen, V
=
inferior vena
IVC
ble3.16) so that optimum results are achieved. In scanning protocols that include two enhancement phases. contrast administration is geared toward the arterial phase of enhancement. A power injector is essential for this type of study. using initial flow rates in the range of 3-5 mlls. For most clinical applications. monophasic contrast injections with a constant flow rate suffice. It has been shown, however, that an individually adjusted biphasic contrast material injection actually produces a more arterial constant plateau. This technique (see p. 103) is still too complex for routine use.
NaCl Bolus (Saline Flush) The injection of a saline bolus (saline flush) is useful only in spiral CT examinations that require arterial phase imaging. If only later phases are of interest, there is usually no need for a saline bolus. All injection techniques that employ small amounts of contrast material should include a saline bolus. Rationale
Saline bolus injection is based on the following considerations: when a contrast bolus is injected into a vein, the flow velocity in the vein usually returns to normal after the injection is completed. These normal velocities are highly
variable but generally are less than the injection velocity of the contrast medium. At the same time, venous pooling of contrast medium occurs in the veins through which the medium is delivered. Injecting a saline bolus immediately after the contrast bolus will flush the pooled contrast from the veins and should propel the contrast bolus ahead of the saline bolus at a constant velocity. This technique has been successfully used for some time in intravenous DSA and MR angiography, and can be suitably adapted for CT scanning. A saline bolus thus can be used to increase the duration of contrast enhancement, and may allow for a reduction in the amount of contrast material or for the use of higher injection rates at identical duration of contrast enhancement. In general, we recommend 40-60 ml of saline. Above that amount, almost no further improvements can be noted. A saline bolus yields the effect of contrast material for the cost of saline. Single-barrel CT Injectors
When using a single-barrel injector, at least a 200 ml syringe should be used for the injection. The contrast medium and NaCI are drawn successively into the syringe, using a technique that will cause the NaCI to form a supernatant layer on the heavier contrast medium. To perform the injection, the pump should be directed vertically downward (this is not possible with all devices and should be considered when the device is acquired). There are alternative techniques used to fill the syringe:
• Tip of injector pointing vertically downward: the saline is drawn first, taking approximately 10ml more than the planned volume of the NaCI bolus (a precaution against air injection). Air is expelled from the syringe, and the selected contrast volume is drawn carefully (slowest setting) into the barrel. Slight mixing will usually occur at the interface between the contrast medium and NaCI, but this will not significantly affect opacification.
Tip of injector pointing vertically upward:
the selected followed by ting) of the inadvertent
contrast volume is drawn first, careful drawing (slowest setNaCI bolus (+IOml to prevent air injection). This creates a
99
Patient Preparation
and Contrast Media Application
smooth meniscus at the interface of the contrast medium and saline. Air is expelled from the syringe, which is then carefully rotated 180 into the injection position, the NaCI bolus remaining above the contrast medium. This method provides a sharp demarcation between the injection solutions but is more laborious than the first technique. 0
Double-barrel
CT Injectors
and Roller
Pumps
Double-barrel syringe-based injection systems for CT are just appearing on the market. Like roller pumps with two or more reservoirs they markedly simplify the saline flush technique. The injector automatically switches to a separate saline reservoir or the second saline-filled barrel after the contrast bolus has been injected. Such injectors can also be used to vary the dilution of the injected contrast material by simultaneously injecting contrast and saline. Allowance must be made for the deadspace volume (5-30 ml) within the tubing system. This means that the volume of the saline bolus must be increased by an amount equal to the dead-space volume. Another potential difficulty is the fact that the tubing system is filled with saline rather than with contrast medium after each injection. This will not create problems if bolus triggering is consistently used. Otherwise the delay time to scan initiation must be increased correspondingly, or the tubing must be flushed with contrast medium.
Contrast Material Concentration Most institutions will use one concentration of contrast material as a standard for the vast majority of their CT examinations. All suggestions for contrast material injection given in this book rely on contrast material with 300 mg iodine/ml (unless stated otherwise). It has to be noted that the contrast enhancement in CT only depends on the local iodine concentration in the vasculature or tissues. In the vessels, this concentration is reached by a mixture of the injected contrast material with (non-opacified) blood. It depends on the injection rate of iodine in mg/sec. It should therefore be irrelevant whether contrast material of a concentration of 400 mg/ml is in-
jected at 3 mlls or whether a concentration of 300 mg/ml is injected at 4 mlls. Lower concentrations of contrast medium have the advantage of a more physiologic osmolarity, resulting in better tolerance and less intracapillary dilution from the back-diffusion of interstitial fluid. They are also less likely to cause high-contrast artifacts in the injected veins (important in chest CT), resulting in a theoretical preference for less concentrated media. However, they produce less vascular and organ enhancement than more concentrated media injected at an equal rate. To achieve the same level of enhancement, the fiow rate must be increased correspondingly. This means that the contrast volume also must be increased to achieve a similar duration of enhancement, but this taxes the limitations of CT injectors and may impose an excessive volume load on the patient. Only a few manufacturers supply reduced concentrations of contrast media in injection bottles of adequate size. Users can make their own dilutions by adding NaCI solution or injectable distilled water. It should be noted that 0.9% NaCI will cause less change in the solution osmolarity than distilled water. Excessive dilution of contrast media with distilled water may produce hypotonic solutions, with an associated
risk of edema
Higher concentrations of contrast medium allow the same amount of iodine to be injected at a lower flow rate than standard concentrations. Higher concentration of contrast material are particularily suited for short scan duration with multislice CT (4- and especially 16-slice scanners) because improved arterial enhancement over lower concentrations has been reported, even if identical amounts of iodine are injected per second. High-concentration contrast materials (>350mg iodine/ml) have a substantially increased viscosity and require pre-warming of the contrast material. Without the use of a saline flush, the duration of contrast enhancement drops early because the (more viscous) contrast material remains in the injection" veins.
Intravenous ContrastAdministration dium by IV bolus injection (preferablyusingmechanical injection at the same rate as the diagnostic scan) and performing several trial scans in a reference region using the lowest possible tube output. The scans may be taken at 1-2 second intervals, depending on the dose and type of scanner. Even for these trial scans, it is advantageous to have an 8-12 second delay from the start of the injection to the initial scan. The delay time depends on the patient's presumed circulatory status but rarely exceeds 20 seconds (except in patients with heart failure or valvular disease). The test is evaluated by identifying the image that shows maximal enhancement in the target region (e.g., the aorta). The time to peak At, measured from the start of the injection to the time index ofthe selected image, equals the optim um scan delay time. Experience shows, however, that the best results are achieved by adding 2 to 5 seconds to the calculated delay time (Table 3.17).
Scan Delay Ind ivid ual d iff erences in circulation tim e are difficult to assess since they are based on heart rate, patient age, and previous illnesses. Because ofthese differences, the scan delay time should be individualized so that imaging can be performed in the optimum phase of enhancement. This is the more important the faster the data acquisition becomes. Modern multislice scanners require just 6 s for a liver scan with 4x 2.5 mm collimation, and the next generation of 16-row scanners requires even less time (2-4 s. depending on collimation). The risk of mis-timing the injection increases with such equipment. The arterial phase of perfusion is the most difficult one to capture. Individual adaptation is mandatory for fast scanners. The parenchymal or portal phase of perfusion is less critical, thus standard delay times can be used for most routine examinations. Even in these cases, better results can be achieved by individualizing the scan delay (Table 3.17). •
Bolus Triggering
A test bolus is not the preferred method for determining the scan delay. A more efficient method is bolus triggering, in which the contrast bolus itself is used to initiate the scan. A growing number of manufacturers are offering this technique, in which "monitor scans" are
Test Bolus
The injection ofa test bolus is a robust method available for determining the individual circulation time and the optimum scan delay. It consists of administering 10-20 ml of contrast meTable
3.17
Tailoring
the scan delay to the patient's
circulation
time
Test bolus Reference
Contrast Scan
region:
parameters:
parameters:
Scan delay 0
Ascending
aorta Start of scanning
(Thoracic
(CTA
region
CTA)
of neck or abdomen)
10/5/8-12 5 mm section collimation, ning sequence: 20-30 s Calculated D = At + 2
minimal
from time to peak
dose,
scans
at 2-4
s
intervals;
duration
of scan-
At.
s (At < 20 s) D=At+5s(At>20s)
Bolus triggering Target
region:
Spleen
(5) or portal
Contrast parameters:
According
Scan
Section
parameters:
(Thorax, eTA)
Ascending aorta (A) Abdominal aorta (A) to protocol,
collimation
Scan delay 0
Depending
Notation:
V/FIDX, aorta
(Arterial phase in liver or pancreas) (Portal venous phase)
veins
no test injection
as in spiral scan,
on target
region:
e.g.: 120 + 60/4/5
A
minimal
dose,
delay starts when =
scan
initiated 5
scans at 1 to 4s intervals
enhancement
s
in
after appearance
ROI
exceeds
of contrast
50
HU
in the
101
3 Patient Preparation and Contrast Media Application
taken at a suitable table position using a greatly and the contrast material is injected rapidly C'mlls). On the other hand, a delay of <4s reduced dose, and a region of interest (trigger ROI) is defined in which the progression of en- does not allow adequate time to give breathing hancement is monitored (Fig. 3.2). Once an ade- instructions. We suggest that patients should quate level of enhancement is achieved (deter- hyperventilate immediately prior to contrast inmined visually or by a threshold CT number, the jection but breathe shallowly and not hold their trigger threshold), the table moves to the breath during the trigger scans. The position of the trigger ROI should be tailored to the clinical starting level, and helical scanning is initiated. For technical reasons, there is a delay of problem and to the time delay that is necessary 3-9 seconds between the last monitor scan and for giving breath-hold commands (Table 3.17). the start of helical scanning. Up to 50 ml of con- Because of a delay of at least 5s in most trast medium may be injected during that time, patients, it may be advisable to go upstream depending on the flow rate. Thus, bolus trigger- with the trigger ROI (e.g., descending aorta for ing for purely arterial imaging (CT angiography) abdominal studies, or ascending aorta for the should not be performed if the delay exceeds 5 s chest), and to use a relatively low trigger threshold of 50 H, which is above the level of most artifacts, and markedly below the enhancement level reached later during the examination. Alternatively, the scan can be started manually as soon as the contrast material is seen in the target region. Problems may arise if the software is cumbersome and makes the procedure time-consuming. Artifacts may occur in trigger ROIs in the aorta at the level of the diaphragm (respiratory artifacts). Thus, this level should be avoided. Instead, we suggest a trigger level either in the descending aorta, well above the diaphragm, or in the abdominal aorta, at the level of the SMA. Positioning the trigger ROI in the cardiac ventricles may be tricky because the scanograms are acquired during inspiration b a while the monitor scans are performed during 90 shallow respiration. A table position about (HU) 80 ,,",Enhancement ,., , ..",' " ,.. """" "'" ..,,'" 4-5 em below the tracheal bifurcation yields 70 _. - . ... \ i goods results in most patients. 60 "", "' .., ""..,'" "'"'' ,"" """",~"".'''".."'" ,. '"''''''..,.." ! Triggering can also be performed for exami50+-------~--------~--~--- nations of parenchymal organs. Choosing a trig40 ger ROI within the organ parenchyma, however, 30 is risky. If desired trigger level is not reached, ... h··············t--' 20 10 ,,,", .,,,,, ,,,,, ..',, the examination fails or a "default delay" has to , . --... ..... -. _. ... be used. In patients with excellent parenchymal o enhancement, the trigger level is reached too 10';'" --" , Time (5) 20+-------~1--------~----~~ early, starting the scan before the optimum en10 15 o 5 hancement could be reached. For this reason, triggering in the aorta is more advisable, but the Fig. 3.2 Bolus triggering. Placement of trigger ROI (a), delay has to be adjusted to the time expected last image of series of low dose trigger scans (80 kVp, for proper organ enhancement. 2mGy) (b). Scans were started 10s after begin of the I .••
,.'
•
- •..•••••••
p
•••••••
-
.~
••
!
.;
_
-
i
contrast material injection and were repeated every 1.5s. Time-density curve (c). As soon as the trigger threshold of 50 h over baseline is surpassed, the spiral scan is initiated. Note that there is an additional delay of 2-8 s with most scanners before the actual data acquisition starts.
Intravenous ContrastAdministration can calculate how the patient would respond (in terms of enhancement curves) to larger amounts of contrast material or higher flow rates. Such calculations show that the aortic enhancement after injection small volumes of contrast material has a short peak that is substantially less than with larger contrast volumes (see Fig. 3.3b). This behavior suggests that contrast volumes cannot be proportionally reduced as the scan duration decreases with new 8- and 16-slice scanners. For larger volumes there is a bent in the enhancement curve with a more rapid increase in the initial phase and a smaller increase as more contrast material is injected. Maximum enhancement is reached close to the end of this" enhancement plateau".
Individualized Contrast Material Injection
Circulatory parameters such as cardiac output, patient size, and inflow obstructions have a substantial influence on contrast enhancement in individual patients. Two main approaches towards a more individually adapted contrast material injection have been suggested: mathematic modeling and Fourier-deconvolution of a test bolus injection. The test bolus technique (Fourier-deconvolution technique) uses a test bolus injection to determinetheindividual patientresponsefunction (for the aorta) to the injection of contrast material. Assuming a linear behavior of the patient's circulatory system towards contrast material, one
8
16 ml@4 mils
Enhancement (HU)
300 4
2
o
Time (s)
o
8
16
24
100
o
32
8
I~-----~:: 8
16
24
32
40
48
56
64
72
00
40
48
56
64
72
80
-uu
Contrastbolus
Enhancement (HU)
8x16ml@4ml/s
6+-----------------·~------
300
4
zoo
2
100
o
o
8
16
24
32
Fig. 3.3 Individual adaptation of contrast material injection [modified from Fleischmann 1999]. The timedensity curve derived from a test bolus injection (a) can be usedto calculate an individual patient function that allows the effect of varying injection parameters on contrast enhancement to be predicted (b). Note that the aorticenhancement "plateau" grows over time and that the maximum enhancement also grows as larger
o
o
8
16
24
32
volumes of contrast material are injected. By setting a target level for contrast enhancement and the duration of the plateau phase, an ideal contrast injection curve can be calculated (c). Approximating this ideal injection curve with a biphasic injection with a rapid initial and a slower later injection phase agrees closely with an ideal enhancement plateau (d). Fig.3.3c.d
>
104
3 Patient Preparation and Contrast Media Application
10
i Theoretical ,ideal' enhancement
Theoretical ,ideal' bolus
8
I
I
200
+-' C QJ
E 100 QJ U C
ro
.s: c u.J
o+---~~~~-=~~~~-=~--~---
-4~-----------------
5 (
20
35
time (5)
50
65
~------------------------------------------------------------------~ Optimized enhancement
Optimized bolus 40 ml
@
6.7/5 + 73 ml
@
2.8/5
10 8 +-' C QJ
E 100QJ U C
ru .i: c
2
u.J
o d ~
o-·~~~~~~~~~~~~~~~~ 10
20
30
Injection duration (5)
5
20
35 Time (5)
50
65 ~
Fig. 3.3(, d
Conversely, it is also possible ttl pre-determine the level of contrast enhancement and the duration of the plateau phase, and then calculate the volume and flow rates required for this individual patient (Fig. 3.3). It could be shown that an individualized biphasic injection with a smaller amount of contrast material injected at higher flow rates followed by a larger amount injected with slower flow rates can produce a more constant enhancement than a monophasic injection. Up to now, however, the goal of reaching a predefined height of the plateau phase cannot be achieved in all cases. Mathematic modeling uses patient-specific data to determine the individual organ enhancement produced by a specific contrast injection protocol. It is also possible to calculate estimates for the volume and flow rates of contrast mate-
rial required to obtain a specific enhancement level in a target organ. This approach could show that an exponentially decreasing flow rate is ideally suited for a constant enhancement plateau in the aorta and various organs (Fig. 3.4). The model predicts that aortic enhancement increases as higher flow rates are used but hepatic enhancement remains almost independent of flow rate (Fig. 3.5). It also predicts that maximum liver enhancement can only be increased by injecting higher amounts of contrast material. With this model, the effect of body weight and cardiac output can also be determined. Liver enhancement increases almost proportional with the injected amount of iodine per kg body mass. Conversely, liver enhancement for a given contrast material volume depends on
Intravenous Contrast Administration Exponential decrease of flow rate
Optimized
enhancement
5 4 ---------------.-----
E
~
u,
E
2
0~~~~~~~-=~4-o 10
20
100- ---
O-'~~-=~~~~~~~~~~ 5 20 35
30
65
Time (s)
Injection duration (s)
Fig. 3.4 Mathematic modeling predicts that an exponentially decreasing flow rate for contrast constant plateau of contrast enhancement [modified from Bae and Heikenj.
Fig. 3.5 Mathematic modeling allows the effect of increased injection speed on aortic and hepatic enhancement to be predicted (a). Note that maximum aortic enhancement increases while maximum hepatic enhancement remains largely independent offlow rate [modified from Bae 1998].
50
injection will yield a
Contrast enhancement (HU)
350 300 250 200 150 100
liver parenchyma'
.. --
-- ..
50 O+---~--~--~---,--~---,----~--.---~--,
o
2
4
6
8
10
Flow rate F (mils)
patient weight and decreases for more obese patients. The effect of weight on aortic enhancement is much less. Conversely, cardiac output strongly effects maximum arterial enhancement (Fig. 3.6 a) but dose not effect maximum hepatic enhancement. However, hepatic enhancement is substantially delayed with low cardiac output (Fig.3.6b).
The system is not yet commercially available and suffers from the fact that cardiac output cannot be precisely determined beforehand.
Perfusion Studies
The administration of 30-80 ml of contrast medium by rapid bolus injection (4-10 mlls) can be used to study the perfusion characteristics in a specific organ region (e.g., for evaluating a focal hepatic lesion). CT scanning in these cases is performed with a stationary table, and contrast dynamics are analyzed through the inflow, peak, and clearance phases of enhancement. For optimal results, administration with a power injector and standardized flow rates is advisable. Two basic concepts for image analysis are available: one is based on the slope of contrast
105
3 Patient Preparation and Contrast Media Application
500
Fig. 3.6 Effect of decreased cardiac output on aortic and hepatic parenchymal enhancement in an experimental porcine model [modified from Bae 1998].
Cardiac output
==>
·-·-·---·-··-------···-······-····--····--··--·---··--·-·----··--·1
Reduced cardiac output • increased aortic enhancement 1 delayed aortic enhancement • decreased slope of enhancement .------- -- ..--- -- ---~- -----.--.- ..•.-.--.-.--.-~~- -
400
l
~
E 300 c
.
---.-,
Q)
u
c ro .s:
~ 200
u :.;:::;
'-o
« 100 Baseline
o
I~---'----'-----~---'----'-----'----'--
o
20
40
80
60
100
120
140
Time (s) ~
a ~
Cardiac output
-13%
60
-:::J I
~
c
Q)
E
40
Baseline
Q)
u c ro
..c
c
Q)
u
'.j:i
ro
0. Q)
20
_-_··_--_·_·_····· 1
I
__ ·_---_·_··_---_··_··_-_·_·_··
o ~,~~~-L~----~~--~--~~--~~--~_ 20 40 80 100 o 60 _
==> __ ·--
... _-
... ---.-
Reduced cardiac output • strong delay of liver enhancement • decreased slope of enhancement • little change in peak enhancement _- _ __ -- -.......•.. _ .......•. -..__ _-_ ..----_ _-_ _ ..-.-_.
b ~
enhancement (see Fig. 24.71) while the other depends on Fourier deconvolution techniques. Both techniques are able to calculate various circulatory parameters for each pixel in the image. The simpler, slope-based technique is more robust but requires higher injection rates and does not allow for absolute flow measurements. The other technique could also work with slower injection rates and provides calibrated numbers, but is more susceptible to noise and data inconsistencies. Both techniques, however, require data sets that are free from respiratory motion, a fact that makes them difficult to use for most body applications.
120
140 Time (s) ~
Direct Arterial, Portal, or Venous Contrast Administration
Arterial Contrast
Direct intra-arterial contrast administration through an angiographically placed catheter is necessary only in very selected cases. The commonest application is CT arteriography of the liver for the detection of hypervascular lesions. This procedure has been superseded almost completely by dual-phase scanning with intravenous contrast or MRI. Other applications include determination of vascular territories that are supplied by a particular vessel.
/\
DirectArterial, Portal, or Venous ContrastAdministration Because the injected artery is usually located within the target region, the contrast concentration C should be reduced to 150mg/ml to avoid artifacts, Flow rates should be similar to the ones used for angiographic injection of the same vessel.
Portal Contrast Generally the portal vein is, opacified indirectly by means of splenic or mesenteric portography, This procedure, called CT during arterial portography (or CTAP) , relies on angiographic insertion of a catheter into the superior mesenteric or splenic artery, It is unclear which of the two vessels is better suited for the procedure, Some authors recommend the simultaneous catheterization of both arteries for best results, A minimal dose of contrast medium should be used for angiography to avoid the premature enhancement of hepatic lesions, CT scanning is performed during intra-arterial injection of the contrast medium, A great variety of scanning protocols have been recommended in the literature (see also Chapter 11), The only means of direct portal opacification is through a catheter that has been inserted intraoperatively into the portal vein or superior mesenteric vein, This would be done only at the time of an arterial port insertion for regional infusional chemotherapy of the liver. Given the large tumor size in most of such cases, this procedure is very susceptible to artifacts and can be recommended only in patients with small lesions.
Venous Contrast There are few situations in which the high, uniform contrast enhancement of systemic venous structures can be achieved in CT examinations, Homogeneous venous opacification does not appear until the venous phase of enhancement, i.e, approximately 40-100 seconds after the start of the intravenous contrast injection, depending on the vascular region, Generally only a moderate level of enhancement is observed, Given the differences in the perfusion times of different organs, the mixing of opacified and nonopacified blood at venous confluences can
mimic an intravascular filling defect (pseudothrombus artifact, see Fig, 7.39), If scanning is performed during the direct enhancement, the contrast medium should be diluted by about 1:3 to 1:5 to avoid high-contrast artifacts, Even when this technique is used, artifacts can still occur due to mixing effects, Artifacts can also result from layering of the heavier contrast medium in dependent areas of laminar flow (e.g, within the inferior vena cava), Consequently, the injection should be performed bilaterally if possible (in both arms or legs), and the inferior vena cava should be imaged during an interval of adequate flow to promote mixing of the contrast medium with non opacified blood,
Intrathecal Contrast Administration Intrathecal contrast administration is useful for the optimum CT evaluation of spinal lesions or intervertebral disk disease, CT scanning is preceded by lumbar myelography using a standard technique (nonionic dimeric contrast material), The patient remains supine for about 1 to 2 hours and is then scanned prone, This interval is necessary to reduce the concentration of contrast material and avoid high-contrast artifacts,
Intracavitary Contrast Adm inistration Parenteral (preferably nonionic) contrast media can also be used for the opacification of anatomic and pathologic spaces, This is most commonly done in urinary tract studies (antegrade contrast administration through an indwelling nephrostomy tube or retrograde administration through a bladder catheter), Direct contrast injection can also be used to visualize abscess cavities and fistulous tracts, which is especially useful in the pelvis, Intraperitoneal or pleural contrast administration (dilution 1:10-30) is used only in highly selected cases for the imaging of encapsulated areas, demonstration of continuity between regions (Fig, 3.7), or for imaging small peritoneal metastases,
108
3 Patient Preparation and Contrast Media Application Fig. 3.7 Intraperitoneal contrast material (diluted to < 30 mg iodinel ml) demonstrates communication between abdominal ascites and an intractable left-sided pleural effusion.
a
b
jC;liiAagsgaa;'>SsA'aAiAsAA
High-contrast artitacts can be prevented by adequate dilution of the contrast medium (1:3 to 1:5, i.e., < 100 mg iodine/ml). If necessary they can be somewhat reduced by using the highest possible tube current (usually 140kV).
4 Optimization of Scanning Technique M. Prokop
Scanning
Strategy
.
Scanogram (p. 110) Settnq up the Scan Sequence (p. 110) Scan Parameters
.
General Acquisition Parameters (p. 112) Gantry Angulation (p. 112) Scan Field of View (SFOV) (p. 113) Scan Time and Rotation Period (p. 113) Scan Duration and Phase of Respiration (p. 113) Scan Length (p. 114) Scanning Direction (p. 114) Tube Voltage (kVp) (p. 115) Tube Load (mAs) (p. 115) Tube Output Characteristics (p. 115) Single-slice Spiral Acquisition Parameters (p. 116) Section Collimation (p. 116) Table Feed and Ptch (p. 117) Multislice Acquisition Parameters (p. 118) Section Collimation (p. 119) Table Feed and Pitch (p. 120) Vendor-specific Approaches (Four-slice Scanners) (p. 121) Eight- and 16-row Multislice Scanners (p. 122) Reconstruction and Display Parameters (p. 123) Section Width and Raw Data Interpolation (p. 123) Reconstruction Interval (p. 124) Convolution Kernel (Reconstruction Field of View (p. 127) Window Selection (p. 127) Image Processing (p. 128) 2D Processing (p. 128) 3D Processing (p. 129) Special Scanning Techniques (p. 130) Ouantitative CT (OCT) (p. 130) Dual-Energy CT (p. 130)
Algorithm)
(p. 124)
4 Optimization of Scanning Technique
Scanning Strategy Scanogram
The scanogram, known alternatively as a scan projection radiograph, topogram, scout view, or pilot scan, is used to locate and define the regions to be examined by CT. A scanogram is obtained by keeping the x-ray tube stationary while the patient is moved through the scan plane, producing a projection radiograph that resembles a conventional plain film. In the scanogram, however, the contiguous slit scans and collimated detectors greatly reduce scattered radiation compared with plain films. This has numerous advantages, such as a marked improvement in the delineation of the abdominal organs (important in determining the scan length). Spatial resolution is relatively poor, however, depending on the selected slice thickness (0.5-2 mm). Especially when low-dose scanning is performed, the scanogram may contribute substantially to overall patient dose. For this reason, the lowest possible mAs and kVp settings should be used in such situations (Fig. 4.1). Setting up the Scan Sequence
The following guidelines are helpful in devising a strategy for CT examinations:
additional, secondary target regions. For example, the pancreas is the primary target organ for thoracoabdominal staging in a patient with known pancreatic carcinoma while the thorax, liver, and the rest of the abdomen represent secondary target regions. (Intravenous) contrast administration should be tailored to the primary target organ. If timing is not critical for the primary target region, use the secondary target organ for which contrast timing is most critical. Determine the enhancement phases in which imaging should be performed (see chapters dealing with specific body regions). Next decide how many scans are required and determine the sequence in which the primary and secondary target regions will be scanned. The choice of scan parameters strongly depends on the available technology. With multislice CT scanning, either a volumetric imaging protocol with near-isotropic resolution or a fast spiral scanning mode with rapid data acquisition will be chosen. With single slice spiral CT the length of the individual scanning regions and the available time window determines the choice of scan parameters.
• The clinical problem will dictate the primary target organ for scanning as well as
Fig. 4.1 Scanogram using 120 kVp and 100 mAs (manufacturer's standard) (a) and 80 kVp and 50 mAs (approximately one fifth ofthe dose)(b). a
b
Setting up the Scan Sequence
Increasing attention is being given to radiation safety issues in CT examinations. The level of radiation exposure in CT is multiple times higher than that incurred in conventional radiography. This mandates ajudicious selection of imaging parameters, for even slight changes in the parameters can lead to marked absolute differences in radiation exposure. An effort should be made to reduce radiation exposure in accordance with the patient's size and the clinicalpro blem. The diagnostic accuracy of CT will generally be enhanced with a more specific clinical request and a more limited target region • Non-contrast scanning followed by a contrast -enhanced examination are usually indicated in patients with suspected hemorrhage, renal lesions, or suspected hypervascular hepatic tumors. • For most contrast-enhanced studies, the upper abdominal organs should be scanned first during the contrast injection, because organ perfusion and enhancement in that region will critically influence the quality of the examination. • Thoracic scanning should precede examination of the lower abdomen, because
111
vascular opacification in the pulmonary hilar region is more critical. The neck should be scanned last (interstitial enhancement), preceded if necessary by the injection of additional contrast medium. Repositioning of the arms should be considered when neck and chest have to be examined. Keeping the arms in the scan filed will cause major image degradation (Fig.4.2) The scan parameters should be tailored to the scanned region according to the anticipated lesion size, bearing in mind the limitations imposed by the available scanner technology. For CTA and virtual endoscopy, the smallest possible FOV (for better resolution) and a soft convolution kernel (for less noise and better 3D rendering) should be used. Sample protocols for CT examinations of varying complexity are shown in Table 4.1.
b Fig. 4.2 Raising the arms above the head should be tried whenever possible for scanning of the chest and abdomen: There are substantial streak artifacts with the
arms placed next to the body (a) while raising at least one arm above the head leads to a substantial reduction in artifacts and decreases dose requirements (b).
112
4 Optimization of Scanning Technique Table 4.1 Single-slice
Sample protocols of varying complexity D
for a single-slice scanner (left) and 4-slice scanner (right)
MultisliceD SC/TF/RI
NxSC/TF
V/F/DX
SW/RI'
MPRb
V+N/F/DX
7{12{6
1. Chest, inspiration
1{10 1{30
t t
5{1O{5 2. Tumor + hila:
t
2{4{2
1 i
noncontrast
4x2.5{15
1.Abdomen
noncontrast
8{8
HR'
1{1
HR'
1{1
noncontrast
1. Adrenals:
4xl16
j
3{2
90{3{30
2. Chest: tumor region
4x1{6
i
3{3 1.5{1.5
non contrast cor
90 + 50{3{30
cor+saq''
lymph node staging token over 1~hr p'eriod
- Preparation: 7 L oral eM 1. Liver + chest:
7{12{6
i
120{2{40
1. Liver to neck:
2. Rest of abdomen:
7{12 6
t
105 ISO
2. Rest of abdomen:
- Reposition arms next to body 3. Neck:
3{5{3
t
jl t
Suspected retroperitoneal hemorrhage 1. Abdomen
d
50{2{20
4x2.5{15
t
5{4
~=~~~:=====:;:i::;:;i===1:::::50{4{25 lOs ISO
non contrast
CPR
150+50{5{5A'
cor
lOs ISO
105 ISO
lOs ISO
SC = slice collimation (mm), TF = table feed (mm{rotation), RI = recon. increment (mm), i t = scanning direction, SW = effective section width (mm), MPR = multiplanar reformats. axial = axial sections. cor= coronal, sag = sagittal eM = contrast material, Contrast concentration = 300 mg/ml iodine V=volume of CM (rnl). N = volume of saline flush (ml) F=flow rate (ml{s), 0= start delay, X=trigger region a axial section for diagnostic review, an additional reconstruction to create a "secondary raw data set" may be necessary b MPR are reconstructed from a .secondary raw data set: with SWIRl = 1-1.5/0.7 Pancreas: CPR along pancreatic duct; CTA: MIP I VRT for routine display mode, CPR for stenoses C sequential CT d reconstruction from the same raw data set e read: 5 A = start delay 5 s after reaching 50 HU enhancement in the infrarenal aorta
Scan Parameters User-selectable scan parameters are tailored to the specific clinical request so that adequate spatial and contrast resolution will be achieved. Reasonable trade-offs should be made that will balance the requirements of an acceptable radiation dose with acceptable image quality. The selection of scan parameters is significantly influenced by the available scanner technology. For the notation of scan parameters, we recommend the abbreviations listed in Table3.7.
The following section describes the various trade-offs for the choice of data acquisition, reconstruction, and display.
General Acquisition •
Parameters
Gantry Angulation
Most IT examinations do not require gantry angulation, i.e., the scan plane remains perpen-
General Acquisition Parameters dicular to the table plane. In intervertebral disk studies, however, the gantry angle should be adjusted on the lateral-projection scanogram so that the scan plane is parallel to the center of the intervertebral disk space for each level that is examined. Gantry angulation can also be used in neck scanning to minimize dental artifacts and should always be used in head examinations to exclude the eyes from the scan field. With thin-section multislice scanning, gantry angulation is almost never necessary for body imaging because any desired imaging plane can be reconstructed after the data set has been acquired (see Fig. 25.45). In some scanners, gantry angulation will cause distortion in 3D reconstructions or will not be available for spiral scanning. •
Scan Field of View (SFOV)
On some scanners, the size of the scanned field can be reduced to increase the sam pling rate and thus improve spatial resolution. This is feasible mainly for relatively small object diameters (neck, extremities). If the object diameter exceeds the SFOV, "out-of-field artifacts" will appear at the periphery ofthe object (see Fig. 7.45). For this reason the object should be precisely centered within the gantry aperture, and, if in doubt, a trial scan may be advisable. Out-of- field artifacts influence only a relatively narrow band in the periphery of the (round) SFOV (see Fig. 7.45b). This can be used for indications with a focused interest on only a portion of the body cross-section (suchas in multislice cardiac imaging, see Chapter 23). When positioning the structure ofinterest exactly in the center ofthe gantry aperture, and reconstructing a slightly smaller fieldofview than the actually exposed field, the image quality will be high, and radiation exposure to the more peripheral body portions (e.g., skin,breast) will be reduced. Information outside
the SFOV, however, will be lost. This type of scan is often called a "target scan". For multislice scanning, cone beam artifacts are least severe for structures close to the rotational axis. For this reason, the examined body part should be placed as close as possible to the center of the gantry (com pare Fig. 1.22). •
Scan Time and Rotation Period
Shorter scan times per slice are associated with fewer motion artifacts. The scan time can be shortened by limiting the scan arc to an angle equal to 180 plus the fan angle (partial scan) rather than performing a full 360 rotation. Motion artifacts can be reduced by assembling multiple rotations into one image. Short rotation periods (rotation time, RT) in spiral and multislice CT increase the available scan length and shorten the time required to scan a particular volume. A short rotation time will also allow for using thinner sections while covering a given scan volume. In some scanners, very short rotation periods will cause some degradation of spatial resolution because the number of available projections decreases (e.g., for the "flying focal spot" technology used by Siemens). For high-resolution application, such ultra-short rotation periods should be avoided on these scanners (Fig. 4.3). For cardiac scanning, adapting the rotation time to the heart rate may improve temporal resolution. This will be the case if a two- or threesector reconstruction instead of a one-sector reconstruction can be used (see Chapter 23, Heart). 0
0
•
Scan Duration and Phase of Respiration
When selecting the scan duration, the radiologist should make certain that the patient has adequate breath-holding capacity and that imaging can be performed during the desired enhancement phase. When brief hyperventilation precedes the scan, most patients can tolerate a 30 second breath hold in inspiration.
Fig.4.3 Image quality of HRCT is improved with maximum number of projections. Some scanners using flying focal spot technology require a reduced rotation time (0.75s) (a) for optimum spatial resolution. Resolution with 0.5 s rotation time and the same exposure dose is slightly reduced (b).
a
b
4 Optimization of Scanning Technique
Most body spiral CT examinations (singleslice or multislice) are performed in inspiration. Scans during forced expiration are only required for the chest (high-resolution CT) to detect air trapping or emphysematous lung areas. Multislice scanning during expiration may be used to quantify the amount of air trapping in the whole lungs. There are a few cases in which scanning may be performed during shallow respiration, e.g., for patients in respiratory failure or sedated children. Spiral scanning will su bstantially reduce artifacts on the axial scans if only mild movement is present. If a breath-hold is attempted, the amount of artifacts will commonly increase because such a patient will usually have to continue breathing during the scan with deep inspirations to compensate for his dyspnea (Fig. 4.4). Breath-hold scanning should not be attempted in dyspneic patients, unless the scan duration is very short. Breath holding is unnecessary in examinations of the head, axial skeleton, and extremities. •
Scan Length
The scan range can be determined from the scanogram in most instances. For reasons of radiation protection, pre contrast scans to precisely define the scan range should be avoided.
Fig. 4.4 In dyspneic patients, trying to have the patient hold the breath may cause severe artifacts when the patient has to start breathing again. This can be avoided if the patient is instructed to breathe shallowly. Note the lack of artifact in the upper portion of this craniocaudal scan (arrows) and massive artifact in the lungs and bronchi further distally (arrowheads).
The radiation exposure to the patient (dose-length product, DLP) increases with the scan length. Therefore, the scan length should be kept as short as possible without omitting any portions of relevant organs. Proper patient instruction, training of the technicians and the use of standardized scan protocols will help to reduce the "safety margins" for examinations of the chest and abdomen that are commonly used to make sure all organs are covered, even in a patient with breathing problems. Generally the extension of the scan range will make a prolonged breath-hold and a wider section collimation necessary. This trade-off is less relevant with multislice CT but remains important when deciding on the total amount of contrast material or the scan duration that can be tolerated by the patient. •
Scanning Direction
The direction in which scanning proceeds can substantially influence image quality, depending on the contrast protocol and anticipated respiratory artifacts. A caudocranial direction is best for spiral scanning ofthe thorax, because it reduces motion artifacts toward the end of the scan in the lung apex. For optimum results of breath-holding enough time should be calculated for patients to breathe in and hold their breaths before the scan is started. If the scan is initiated a few seconds after the breath-holding instructions the results
a Fig. 4.5 Starting a caudocranial scan too early causes breathing artifacts (a) that are most prominent on MPR
(b).
General Acquisition
will be substantially improved (Fig. 4.5). This can be accomplished by a start delay of 4-5 s. Caudocranial scanning generally reduces contrast -induced streak artifacts in the cranial mediastinum particularly, if the IV contrast injection is followed immediately by a saline flush at the same inj ection speed: the brachiocephalic veins are then captured at a point in time when they already contain saline solution, the highly concentrated contrast material being flushed to therightatrium. When the liver is examined early during the portal contrast perfusion phase, caudocranial scanning will improve visualization of the hepatic veins because they will be covered during the late part of the scan when they are more readily opacifed. This effect is of course less pronounced with multislice CT because scan duration is much shorter. •
Tube Voltage (kVp)
If the tube voltage is reduced'from 140kV to 80 kV, a greater part of the radiation is absorbed in the prefilter and in the outer fatty tissue layers of the patient. As a result, the CTDIvoi for identical mAs settings is reduced by a factor of 3-4. This is accompanied, however, by a strong increase in noise but also an increase in CT attenuation (and therefore an enhanced radiation contrast) of substances containing high atomic numbers, such as bone, iodinated contrast agents, or metals. Hence, the relationship between image quality and kVp settings is complex (see more detailed discussion Chapter 5). Our own results suggest that a high kVP setting (140kVp) is advantageous for obese patients, while a low kVp setting (80-100kVp) is advantageous in children or in low-absorption areas such as the chest, especially when contrast
Fig. 4.6 Low-dose scan of a patient with chronic thromboembolic pulmonary hypertension performed with 80 kVp and a CTDlvoi=2.3 mGy. There are multiple peripheral pulmonary artery stenoses (arrows). Note the excellent contrast (window settings W/L= 1500/300) on the thin-slab MIP (a) and the good quality ofthe volume rendered display (b).
Parameters
material is injected (Fig. 4.6, see also Figs. 5.16, 5.19, 24.18, 24.19 a, 24.20 a, 24.64). •
Tube Load (mAs)
For constant tube voltage and beam filtration, the tube current is proportional to both, the patient dose and the detector dose. The mAs per tube rotation provides a useful measure of image noise (detector dose), while the mAs per tube rotation divided by the pitch (effective mAs = mAs per rotation/pitch) provides a measure of the patient dose. A high mAs setting results in low noise, permits the use of high-resolution kernels, and improves image quality (see Chapter 5) but also increases radiation exposure to the patient. On most newer scanners, the volume CT dose index (CTDIvoi) is displayed on the user interface (as CTDI) and provides a better information about radiation dose than the mAs settings ever could. Radiologists and technicians therefore should get adapted to using CTDIvoi instead of mAs as a measure of exposure settings to compare scanning protocols. •
Tube Output Characteristics
The x-ray tube output, heat capacity, and heat output are the tube performance characteristics that limit the mAs settings available for the duration of the spiral CT scan. This has a direct effect on image quality. The longer the scan duration or the greater the number of successive spiral scans that are acquired at short intervals, the higher the necessary tube output and the lower the available mAs per tube rotation. In older scanners, trade-offs have to be made in terms of image quality or available enhancement phases. With multislice CT, two, four, or more sections can be acquired simultaneously. Although
115
116
4
Optimization of Scanning Technique
this feature can be used to reduce the requirements on the x-ray tube by a similar factor, most users choose thinner sections instead, which increase the scan duration and keep mAs requirements high. In addition, subsecond rotation times (e.g., 0.5 s) and high pitch factors further increase the rnA settings that have to be used for obtaining a good signal-to-noise ratio. With a pitch of 2 (P * =8) and 0.5s rotation time, the rnA settings have to be four times higher than for a conventional incremental CT (1 s rotation time) at identical patient dose and signal-to-noise ratios.
ters are discussed below. Table 4.2 gives an overview of suggested scanning parameters depending on the clinical indication. Table 4.3 describes the trade-offs for the choice of various acquisition parameters. •
Section Collimation
A narrow section collimation (SC) has the advantage of reduced partial volume effects, better spatial resolution along the z-axis, and a con- f: sequent improvement in the image quality of • multiplanar reformations and 3D reconstrucReduced mAs, soft convolution kernels, tions. The resulting increased image noise can shorter scan lengths, and thicker slice collimation be reduced subsequently by increasing the seccan be used in an attempt to minimize image tion width in axial reformations, which will ennoise without overloading the tube. These tech- hance the signal/noise ratio without corn- ji y niques also reduce the radiation dose to the promising z-axis resolution, The main advantage of a wide section collipatient and, therefore, should be considered in mation is that less radiation dose (lower mAs the formulation of scan protocols. Maximum dose per section is required for setting) is required at a constant level of image very obese patients. This can be achieved by noise, whereas at a constant dose, there is less decreasing the rotation time and increasing the noise resulting in a better contrast resolution, section collimation and mAs. Image noise can be The most practical advantage of a wide collimareduced by reconstructing thicker sections (sec- tion is the ability to scan a given volume more , tion thickness SW= 10mm) and by using soft rapidly because thicker sections can also be re- ( convolution kernels. In single-slice CT, a pitch of constructed from thinner ones using MPR func1 and a 360 LI interpolation algorithm may be tions (or in the case of multislice scanning, the & used (twice as dose-efficient as 180 LI). With raw data of the scan). multislice scanning, low pitch factors are reThe choice of a collimation setting depends F on the clinical request and the nature of the required to allow for high mAs settings. gion that is to be examined. Since most anatomic structures are oriented along the longitudinal axis ofthe body, a section collimation of Single-slice Spiral Acquisition Parameters 5-8 mm can often be used without introducing Spiral CT has become the standard technique for detrimental partial volume effects. Large most applications in the body. The most impor- volumes are imaged with a large section thickness in order to reduce the scan duration. The tant trade-offs in the choice of the scan parame0
0
Table 4.2
Suggested scan parameters for single-slice spiral CT ofthe body
Ai; ! 1 "Indication I
Of
------------
Single-slice cT SC/TF/RI"
High-resolution CT of the lungs 9 Skeleton (facial bones, upper cervical spine, hand, foot, elbow, knee) Tracheobronchial system, CTA (renal arteries) Adrenals, skeleton (spine, pelvis, shoulder, elbow, knee), CTA (carotid artery: 2/411) 1 Head, neck, pancreas, kidneys, CTA (aorta: 3/6/2) 1 Thorax, liver, pancreas, kidneys, lower abdomen I Screening (chest + liver), abdomen 9
1
1-2/10-2013 1/2/1 2/3/1 2/4/2 3/5/3 5/8/4 7/12/6
SW 1 1.3 2.3 2.6 3.6 5.9 8.5
For optimum 3 D results (MPR MIP, SSD, VRT), choose RI = SW/2 incremental scanning technique SC = section collimation (rnrn). TF = table feed (mrn). RI = recon. increment (rnrn). SW =section width (mm)
b
Single-slice Spiral Acquisition Table 4.3
Trade-offs
Acquisition
that influence
Parameter
Section collimation
SC
Table feed TF
the choice
of scanning
Parameters
parameters
Trade-off large small
--
large
->
less nose, better low-contrast resolution, more partial volume effects less partial volume effects, better z-axis resolution, more noise faster coverage,
longer scan range per time, less motion
artifacts
on
MPR smalls
FItch P
smaller
large
->
small-> T uberotationtime
RT
resolution
large
-> more projections coverage _"' less motion
small Scan length L
better z-axis
or faster coverage,
increased
patient
+ better in-plane
artifacts,
better z-axis
less patient
resolution
(some scanners),
resolution
or faster
-> long scan range, thinner collimation possible, more contrast material required -> shorter scan range, less artifacts, less contrast better
anatomic
dose
dose
large
-» large coverage,
small
longer scan duration TI required -> less patient dose, thinner collimation
overview,
more motion material
thicker
possible,
less
coverage artifacts,
required
collimation
shorter
SC or
scan time
mAs settings
high low
-> less noise, better contrast resolution, _" less patient dose, more image noise, low-contrast details
kVp
high
-> better penetration,
low
-r
large small
-> less noise, better low-contrast resolution, more partial volume effects -> less partial volume effects, better z-axis resolution, more noise
large small
->
settings
Reconstruction
more patient dose decreased detection
less dose for given signal-to-noise
patients, especially in abdomen higher contrast (especially with iodinated contrast minirrum dose, less dose for given signal-to-noise patients, chest, and neck
Recon. increment
materia), lower for children, slim
RI
less images bu lesions may be missed more images, better MPR/3D representations,
more lesions
Z-filter
wide -> less image noise and dose requirements, lower z-axis narrow _"' increased noise, higher z-axis resolution
Recon. filter kernel
smoothing
less image noise better 3D images
high resolution Field of view (FOV)
large small
-> substantially
-> better overview, -c-» higher maximum
section collimation should be reduced, however, if it is necessary to depict fine anatomic details, With low-contrast lesions such as hepatic metastases, the signal-to-noise ratio, and thus detectability decreases for lesions that are smaller than the section width, Table
In spiral used for improves phase, or
of small
in obese
Parameter
Section width SW
•
SC possible
less spiral artifacts,
small large
Scan duration TI
collimation
Feed and Pitch
scanning, a pitch of 1.5 to 2 should be the vast majority of indications, This scan coverage during a breath-hold it can be used to improve spatial reso-
and dose requirements,
lower spatial
increased
spatial
noise,
higher
detected
resolution resolution
resolution
smoothing filter kernels are less disturbing spatial resolution (requires HR filter kemels)
lution by reducing section collimation (see Fig, US), At the same time, less dose is required than with conventional slice-by-slice scanning (see Fig, 5.12), A pitch in this range provides an optimum z-axis resolution for a given scan length and simultaneously reduces the radiation exposure, If the collimation is very thin (SC < 1 mm): even a pitch as high as 3 can be used without causing serious additional artifacts (Fig, 4.7),
118
,
4 Optimization of Scanning Technique Fig. 4.7 A pitch of 3 may be possible with thin collimation (SCI TF /RI = 1 /3/1) for CTA of the pulmonary vessels (a) and even the lung parenchyma (b).
b
a
MultisliceAcquisition Parameters
For the vast majority of applications there are two essentially different ways to use a multi-
Table 4.4
Comparison
of multislice CT scanning Volumetric
slice CT scanner: as a fast spiral CT scanner or as a volumetric imaging tool (Table 4.4). Fast spiral scanning takes advantage of the increased acquisition speed but still uses the
protocols for fast spiral scanning
and volumetric
imaging
Fast spiral scanning
imaging
16-slice
Protocol
4-slice NXSClTF
16-slice NXSClTF
4-slice NXSCITF
NXSC/TF
GE
4x1.25/7.5
16XO.625/13.75
4X2.5/15
16X1.25/27.5
Philips/Siemens
4X1/6
16XO. 7 5/16-18
4X2.5/15
16X1.5/32-36
Toshiba
4X1/5.5
16XO.5/11.5
4X3/16.5
16X1-2/23-46
Performances
TT TT TT TT
T
Coverage
T TTT TTT
Speed Z-resolution
lrnaqenolse" Pulsation
artifacts
Breathing
artifacts
TT TT
U U
A
1
Applications Standard,
Tumor
Chest
Tumor staging I nterstitial disease
Metastases Mediastinum
Abdorne?
Pancreas Liver, biliary system Bowel Preoperative evaluation
Standard abdomen Liver (noncontrast phase) Kidneys
Carotids Pulmonary vessels Abdomen
Aorta Oysponeic Veins
Small joints Cervical + thoracic spine Small bones
Pelvis Lumbar spine Long bones
Skeleton
3
c
staging
lymph node staging
Neck
relative to 1 s-single-slice 7/12/6 use 16x1-1.5 mm for volumetric
protocol imaging
b
in obese patients
patients
at identical patient dose (CTDlvoi)
Multislice Acquisition Parameters scanner primarily as an axial imaging tool. It has the advantage that a thinner collimation than with single-slice CT can be used, although the acquisition time is substantially reduced in comparison with single-slice spiral CT Images are reconstructed with a section width that is equal to or slightly less than section width in conventional spiral CT These images serve as the primary evaluation tool. Thinner sections are reconstructed retrospectively only if questions arise, and may be used to create good quality multiplanar reformations, This technique has the advantage of a known workflow, high patient throughput, limited numbers of images on a routine basis, and the option for problem solving with thin sections and multiplanar imaging, However, it does not take full advantage of the capabilities of multi slice CT, and fine detail may be missed if thick axial sections are to be used as the primary tool for making a diagnosis, Volumetric imaging takes full advantage of the near isotropic spatial resolution that becomesavailablewithmultisliceCT Ittransforms CT into a true multiplanar imaging tool that is, in this respect, superior to MRL It is the preferred, if not only, mode used with 16-slice scanners, Volumetric imaging requires the reconstruction of thin, overlapping axial images that serve as a secondary raw data set, These images are used as a 3D data volume from which further images are generated for diagnostic purposes, In most instances, these thin images will suffer from increased noise and require the reconstruction of thicker multiplanar reformations with a section width that is tailored to the clinical question, Such MPR are reconstructed as part of the standard data evaluation and can be "pretailored" to the anatomical site (such as coronal MPRofthe tracheobronchial tree, sagittal MPR of pulmonary masses, curved planar reformations of the pancreatic duct) or they may be created interactively at the CT workstation to probe the data volume interactively and obtain problem-adapted sections which is, in the interactive 3D data evaluation, similar to ultrasound, This approach requires very fast workstations and networks with proper software to be timeand cost-efficient Volume rendering may become the primary tool for evaluating such data because it combines the advantages of MPR with additional depth information,
Volumetric imaging takes full advantage of the capabilities of multislice CT scanning, However, there are marked differences in workflow, and the patient throughput is lower due to the longer processing times, Increased data volume has to be reconstructed, processed, analyzed, and stored, which requires a proper network and archiving system as well as fast, intuitive CT workstations and trained personnel. At present, such constraints are major and lead to markedly increased effort with such a technique, Therefore, volumetric imaging is used most often only in selective cases, such as planning of surgery, CT angiography, or complex cases that cannot be resolved with conventional techniques, Section
Collimation
Fast spiral scanning can rely on a 4x2,5 mm (4 x 2-3 mm) detector configuration for most indications, With a volume pitch (P*) around 6, a table feed of 15 mm per rotation can be obtained, which translates into a table speed of 18.75 mmls (0,8s rotation time) to 30 mmls (0,5s rotation time), This allows for covering the chest (24cm) in 8-12,8 s. which is a very comfortable breath hold period for most patients, A 40 em long abdomen can be covered in 13,3-21 s. which also can be met by most patients, Even combined thoracoabdominal scans become possible: a 60cm range can be covered in 20-32 s. Using a larger collimation but a lower pitch will reduce multislice artifacts but a similar if not better result is obtained if such thicker sections are reconstructed from the 4 x 2,5 mm data, Thicker collimation may be useful for very rapid scanning or in very obese patients as the image noise can substantially be reduced, Volumetric imaging relies on the highest possible spatial resolution at a reasonable patient dose, With current four-slice scanners, this can be achieved with a 4x 1 mrn or 4 x 1.25 mm detector configuration and the reconstruction of thicker sections for image analysis from the raw data or from the secondary raw data set that should consist of LCl-L5 mm thick sections reconstructed every 0.5-1 mm With current four-slice scanners, dose efficiency is reduced by some 20-30% with this detector configuration, In obese patients, a 4x2 mm or 4x2,5 mrn acquisition may be used instead, which requires reconstruction of 2.5-3 mm wide sections every 1-1.5 mm
4 Optimization of Scanning Technique
Similar to single-slice spiral CT a high pitch factor allows thinner collimation for a given scan coverage per time to be used. It has been shown that it is better to use thin collimation with a high pitch factor and then reconstruct thicker sections than to use thicker collimation with a low pitch factor primarily. For an identical reconstructed section width, artifacts are less with
thinner sections, and there is still the opportunity to reconstruct thinner sections if multiplanar imaging has to be performed. Thus, a volume pitch P* of 5.5 to 6 with thinner sections should be preferred over a volume pitch of3 for most applications (Fig. 4.8). Exceptions from this are situations in which the thinnest possible sections have to be reconstructed. In such cases (e.g., for the cervical spine or the extremities, Fig. 4.9), the amount of artifact can be reduced by using lower pitch factors (P *=3, P < 1). Another exception are obese patients, in whom image noise is a limiting factor. If too thin a collimation is employed electronic noise may dominate the images (see Fig. 7.33). Image quality then can no longer be recovered by reconstructing thicker sections, For such patients, thicker sections and maximum mA settings should be used with low pitch factors (p*= 3, P< 1). Adjusting the pitch P even higher, P < 1 to a maximum of 2 (P*=8 for 4-slice units-if allowed by the scanner) will result in an increase of artifacts. However, this may be advisable in cases in which faster scanning or longer coverage is required. Such artifacts can be decreased efficiently by either reconstructing thicker sections or reformatting thicker MPRs. In addition, such high pitch factors may be advisable for low-dose scanning because the mA settings at an identical patient dose is higher for a larger
a
c
Ultra-thin collimation (0.5-0.625 mm) is possible with most multislice CT scanners but there is a substantial increase in patient dose with current systems (see Fig.5.5b), and an increased susceptibility towards electronic noise which further decreases the signal/noise ratio. Such protocols are useful only for very selective applications, e.g., for the lung parenchyma or for small bones, where it can provide exquisite spatial detail. With 16-slice scanners, volumetric imaging is used in almost all cases. Thin (SC= 1-1.5mm) to very thin (SC = 0.5-0.75 mm) sections become available. Because of noise characteristics, thin sections will be the standard mode for the abdomen, while very thin sections can be recommended for the neck, chest and extremities. Only in slim patients and specific diagnostic questions (CT of the pancreas and biliary system, CTA of the mesenteric arteries) very thin sections may be used in the abdomen as well. •
Table Feed and Pitch
b
Fig. 4.8 For identical volume coverage per time (identical table feed), using a higher pitch factor and thinner sections may improve the quality both ofMPRaswell as thick axial sections, at identical dose to the patient. Comparison of a 4x2.5 mm collimation with volume pitch P* of 3 (a, c) to a 4 x 1.25 mm collimation with a
d
volume pitch of 6 (b. d). A CTDlyoi= 12 mGy was usedfor both examinations. Axial sections of 3 mm thickness (a, b) yield comparable image quality but coronal sections of 3 mm thickness (c. d) are substantially better with thinner collimation.
Multislice Acquisition Parameters
a
thickness) as multiples of a single section collimation (1.25 mm, 2.5 mm, 3.75 mm, etc., see Table 1.8) and then decide which detector configuration to use. This approach favors the fast spiral scanning technique and is very intuitive and time-efficient. GE provides only two volume pitches, which are named high-quality (HQ.) mode (P" = 3) and high-speed (HS) mode (P" = 6). As discussed above, the high-speed mode should be employed for optimum system performance (see also Table 1.7). For similar image quality the high-quality mode will require about 50% increase in dose. There are only limited combinations of section width and section collimation possible with the software on current four-detector row scanners (see Table 1.8): it is, for example, not possible to choose a 4 x 2.5 mm collimation for a 7.5 mm section width (reconstruction slice thickness), or a 4 x 1.25mm collimation for a 5mm section width, respectively. To take advantage of direct axial image reconstruction from the raw data, users will have to use the next thicker collimation. Note that the minimum section width is some 28% larger that the numbers displayed on the user interface if a high speed mode is chosen (see Table 1.8). To obtain thick axial sections even when thin-section scanning (volumetric imaging) was employed, the data have to be reconstructed with the minimum possible section width and an overlap of some 50% (e.g., with 1.25mm sections, every 0.7 mm for 4 x 1.25mm configura-
b
Fig. 4.9 For the skeletal system, a lower pitch factor yields fewer artifacts. Comparison of coronal MPRofthe cervicalspine obtained with a volume pitch P* = 3 (a) and a volume pitch P* = 6 (b) from a 4 X 1 mm detector configuration.
pitch, which leads to a higher detector signal and thus less contribution of electronic noise (Fig.4.10). •
Vendor-specific Approaches (Four-slice Scanners)
Table 4.5 compares the parameters settings for the volumetric imaging and fast spiral modes that are available with the various vendors. General Electric (GE) lets the user first choose the section width (reconstruction slice
a
b
Fig.4.10 For low-dose applications, noise decreases substantially with larger pitch factors despite an identical patient dose. Comparison of images obtained with a volume pitch P*=6 (a, c), and P*=8 (b. d) from a 4 x 1 mm detector configuration with 80 kVp and
c
d 40 mAs (CTDlyoi = 1.2 mGy) each. Note that there is a reduction of image noise not only in the 1.25mm thick secondary raw data set (a, b) but also on the 5 mm thick sections reconstructed for image review (c. d).
121
4 Optimization of Scanning Technique tion). After transfer to a CT workstation (e.g.. Advantage Windows) thick axial sections are then reformatted using the MPR mode, and additional MPR can be reformatted in any desired plane. Philips provides the real section width (effective slice thickness) and the pitch factor, not the volume pitch, on its user interface. Fast spiral scanning and volumetric imaging can be applied as described above. Good results are found for a pitch of 1.5 (corresponding to P*=6). Thick sections can be reconstructed from the raw data in a similar fashion as with all other vendors. Siemens has implemented an adaptive interpolation that assures that the section width is independent of the chosen pitch factor. The noise is also kept constant due to the increase of mA with increasing pitch (constant mASeff and thus, patient dose). Artifacts are somewhat lower at lower pitches but the advantage of increased coverage or higher spatial resolution should prompt the user to employ a volume pitch of 6-8, unless bone imaging is performed. Axial images of almost arbitrary width can be reconstructed from any detector configuration as long as the section width is larger or equal to the collimation (see Table 1.9). A special z-filter is available that allows for reconstructing sections of 1 mm width from a 4 x 1 mm collimation. This will increase image noise substantially and therefore cannot be recommended on a routine basis. Fast spiral scanning and volumetric imaging can be applied as described above. Toshiba uses a z- filtering algorithm (MUSCOT) that also assures that the chosen section width differs only minimally from the section width provided on the user interface. Axial images of almost arbitrary width can be reconstructed as long as the section width is larger or equal to the collimation. If the reconstructed section width is identical to the single section collimation, noise will be substantially increased (see Table 5.5). lt is therefore advisable to use a slightly (by 0.5mm) wider section width for most indications of volumetric imaging. The Toshiba detector gives the largest choice of detector configurations. For fast spiral scanning the user has the option of choosing between a 4x2 mm and a 4 x 3 mm collimation. In practice, the thicker collimation can be recommended for obese patients and in all cases where faster data
acquisition is required. The Toshiba hybrid detector is the only one that has four rows of 0.5 mm collimation, making ultrahigh resolution scanning more feasible and less dose-intense. Although the pitch can be varied freely, Toshibasuggests preferred pitch factors (see Table 1.10). Again, higher volume pitches of 5.5 or 6 yield the highest performance. Otherwise, fast spiral scanning and volumetric imaging can be applied as described above. Eight- and 16-slice Scanners
With the advent of more detector rows, there is a shift from fast spiral acquisitions to volumetric scanning with thin sections. These scanners have the advantage of better geometric dose efficiency compared to four-slice scanners, and most will also use more sophisticated detector technology that reduces electronic noise. Nevertheless, ultrathin collimation (0.5mm) will cause a substantial increase in image noise and will require reconstruction of thicker axial and multiplanar sections as described for volumetric imaging. This provides excellent image quality for the lungs as well as for the peripheral skeleton and the cervical spine. It is doubtful whether such thin sections can be used in the abdomen, especially in more obese patients. They prove most useful for the chest, neck and peripheral skeleton, where radiation absorption is not too high. The main difference between 8- and 16slice scanner types and four-slice scanners will be the substantially increased scanning speed, which will all but eliminate movement artifacts, Pulsation effects are spread out over longer distances, thus improving the display of the thoracic aorta and the origin of the supraaortic vessels. Optimization of contrast material administration requires special attention because of the short scanning time. Precise separation of scan phases is possible but timing is a crucial issue and frequently requires longer short delays. Only a limited number of applications allow for savings in contrast material (mainly CTA and chest imaging), most other applications (especially in the abdomen) require similar amounts of intravenous contrast material because organ enhancement is more dependent on the volume than the flow rate during contrast injection. I The application that profits most is cardiac imaging because eight-to 16-row scanners
Reconstruction an d Display Parameters
overcome many of the limitations of spatial and temporal resolution with present four-slice scanners, especially because some new scanners allow for 0.42 s or less tube rotation time. In addition. new techniques for dose reduction (ECGgated mAs modulation) and reduced scan field of view substantially reduce dose requirements with cardiac cr. Since the coverage is substantially increased. lower pitch factors can be chosen to keep artifacts to a minimum without compromising scan speed. However. all manufacturers use more elaborate reconstruction schemes that sufficiently compensate for cone beam gecmetry and allow for a maximum pitch Ps 1.5. Apart from these issues, the same general principles described above for volumetric scanning and fast spiral scanning hold true, and acquisition parameters only have to be adapted to the collimation settings avallable with a specific scanner (Table 4.6). GE supplies an eight-detector row scanner with 0.5 s rotation speed. Similar to their fourslice scanner, multiples of 1.25mm can be chosen as section collimation and reconstructed section width. The manufacturer allows for a volume pitch of 5 (UQ, P= 0.625), 7 (UM, P=0.875), 10.8 (UF, P=1.35), and 13.4 (US, P= 1.675). A volume pitch of 10.8 is suggested for most applications because it provides the best compromise between coverage, dose requirements, and image quality. A detector configuration of 8 x 1.25 mm at a table feed of 13.5 mm/rot, can be suggested for volumetric imaging, and a 8x2.5 mm configuration at 27 mm/rot, yields best results for fast spiral imaging. A special one-slice mode is available for HRCT of the chest. GE is now also supplying a 16-slice scanner that gives the choice between 16 x 0.625 mm and 16 x 1.25 mm collimation. The pitch can be varied in four steps. A volume pitch P*=9 (P =0.5625) and P*= 15 (P= 0.9375) yields a minimum SW that is similar to SC, while P*=22 (P= 1.375) and P*=28 (P=1.75) are more doseefficient but provide a 30% wider SW (0.8mm and 1.6mm, respectively). Apart from high-resolution applications, a pitch P*=22 (28 for CTA) provides the best compromise. Philips and Siemens have introduced a 16-detector row scanner that allows for 16 x 0.75 mm and 16 x 1.5 mm collimation. Like the present scanners, standard section width will be some
30% larger than the collimation (SW= 1 mm for SC=0.75mm, and SW=2mm for SC=1.5mm). This ensures improved signal-to-noise ratios. In addition, thin sections (identical to the collimation) can also be reconstructed but at the cost of a higher image noise. Such thin sections are only useful for the lung parenchyma and the peripheral skeleton. Pitch factors can be increased up to 1.5 (volume pitch P* up to 24) for Siemens and up to 2 (P* = 32) for Philips scanners. A standard pitch around 1.5 (P'" = 24) will suffice for most applications. Both vendors allow for continuous selection of pitch factors. For volumetric imaging ofthe chest, as well as for the abdomen in slim patients, a 16x0.75 mm detector configuration is best suited. In obese patients, a 16 x 1.5 mm configuration should be employed in the abdomen. Fast spiral scanning relies on a 16 x 1.5 mm collimation. Toshiba offers an 8-slice as well as a 16-slice scanner. On the latter, Toshiba provides a 16 x 0.5 mm, 16x 1 mm and 16x 2 mm collimation. Preferred pitch values are P*=ll (P=0.6875), P*=15 (P=0.9375), and 23 (p= 1.4375). When reconstructing sections SW that are identical to the collimation SC, image noise will be increased. Thus, this should only be done for high-resolution applications but not for standard techniques in the chest or abdomen. In case of doubt, it is better to use the next smaller section collimation instead and reconstruct (at least 30%) wider sections. Only for the peripheral skeleton, the neck, and the lungs can a 16xO.5 mm ultrahigh resolution protocol be recommended. In most standard situations in the neck, chest and abdomen, a 16 x 1 mm collimation yields less noise and thus a better image quality for volumetric imaging. A 16x2 mm collimation should be used for volumetric imaging in obese patients as well as for fast spiral scanning in standard patients. Fast image reconstruction is possible if the reconstruction increment RU is chosen 0.2, 0.5,1.0 or 2.0times SC.
Reconstruction and Display Parameters •
Section Width and Raw Data Interpolation
In single-slice spiral CT, the chosen collimation, the pitch factor, and the raw data interpolation algorithm determine the section width (see Figs. 1.13, 1.15). The section width SW grows continuously from 100% to 130% of the section collimation SC when the pitch is increased from 1 to 2 at a 180 LI algorithm. 0
123
Optimization of Scanning Technique A 360' LI algorithm is available only on a few scanners and requires the use of a pitch = 1 (SW=1.3xSC). lts main advantage is a better utilization of the radiation dose. In fact this algorithm reduces the noise as much as an increase in mAs by a factor of 2. Because scan coverage is low. this algorithm has been used mainly for dual slice systems in which this disadvantage is compensated by simultaneous acquisition of data from two detector rows. Reconstruction of thicker section width is possible with both single and multislice spiral CT but there are hardly any implementations in single-slice scanners available yet. Thicker sections for single slice CT have been employed for partial volume artifact reduction in the posterior fossa but have seen few if any use outside this application. They may. however. be used in a similar fashion to multislice CT, and then can serve as a tool to reduce image noise when thinsection spiral scanning had been employed. In multislice CT, the collimation, pitch factor, and reconstruction algorithm determine the minimum available section width SW (see Fig. 1.25). SW varies in a complex fashion between 100% and 130% of the section collimation SC if pitch is increased from 1 to 2 and a 180' MLI interpolation is used. A similar behavior is found for 360' MLI up to a pitch of 1. Adaptive interpolation schemes fix the section width at some 130% of the collimation, independent of the pitch (see Fig. 1.29). For fast spiral scanning, thick sections are reconstructed from the raw data set, even if thin collimation was used to obtain a scan. The section width depends on the size of the patient and the clinical indication (Table 4.5, see also Table 2.2). Thin sections and MPR are only reconstructed as a problem-solving technique if thick axial sections alone cannot resolve the clinical question. For volumetric scanning, overlapping thin sections of 0.5-1.5 mm width are reconstructed every 0.4-1 mm. This "secondary raw data set" serves as the basis for arbitrary cut planes using thick MPR. lt can be employed as a standard for indications such as staging of bronchogenic cancer or CT colonography (see organ chapters for more details), or as an adjunct if suspicious lesions are found (see Fig. 7.29). For indications, in which standard axial, coronal, and sagittal sections are desired for image interpretation, optimum results are gained if the axial plane-
as well as all other required planes-are tioned orthogonally relative to the patient and not relative to the scan table. Hence, positioning of the patient can be easily rected by imageprocessing. •
Reconstruction
posiaxes malcor1
Interval
The reconstruction interval Rl (synonyms: reconstruction increment, reconstruction index, or spacing) in spiral or multislice CT can be chosen independently from the collimation or section width. The advantage of a large reconstruction interval is that it reduces the number of images that must be reconstructed and interpreted. The main disadvantage of non-overlapping sections is that they do not provide optimum lesion detection, resulting in a less confident evaluation and a lower detection rate (up to 30% lower in the liver and lung). Overlapping reconstruction results in a greater number of images that have to be interpreted, processed, and stored. The quality of multiplanar and 3D reconstructions will generally improve with a higher degree of image overlap. Theoretical considerations suggest that a reconstruction increment that is some 30% less than half the section width is an optimum balance between practicality, accuracy, and image quality. In thin-section multislice CT, however, it makes little sense to choose a reconstruction interval that is smaller than the pixel size in the scan plane. For practical reasons, a reconstruction increment that is identical to half the section width suffices for most indications. Usually, in fact, conventional spiral CT as well as multislice CT require an overlap of only 30%, which is sufficient for most routine applications, e.g., lesion detection (Tables 4.2 and 4.5). Convolution Algorithm)
Kernel (Reconstruction
The selection of the convolution kernel always involves a trade-off between spatial and contrast resolution (see Fig. 5.11). High-resolution kernels always result in a disproportionate increase in image noise and may even introduce edge-enhancement artifacts. They are suitable only for regions in which image contrast is high, such as lung parenchyma, and bone. Soft kernels are suitable for cases in which high contrast resolution is required or image noise is excessive due to a low detector signal. This applies particularly to abdominal scanning in obese patients
Reconstruction Table 4.5
Suggested
scan parameters
for multislice
CT scanning
and Display Parameters
of the body (4-slice scanners)
Acquisition
Pitch factor"
N-SC/TF
P
P*
SW/Rlb
SWIRl'
2XO.625/1.S75
1.5
3
0.65/0.5
0.65/0.5
Volumetric resolution)
imaging
(ultrahigh
4X1.25/7.5
1.5
6
1.6/0.Sd
5/3"
Volumetric
imaging
(standard)
4X1.25/3.75
0.75
3
1.25/0.5
1.25-2.5/2
Volumetric
imaging
(skeleton)
4x2.5/7.5
0.75
3
2.5/1.3
7.5/5"
Volumetric patients)
imaging
(obese
4X2.5/15
1.5
6
3/1.5d
5/5
Fast sprial scanning
(standard)
4x3.75/22.5
1.5
6
5.0/2.5
7.5/5
Fast sprial scanning patients)
(obese
0.S-1.5
1.6-3
0.5/0.5
0.7/0.5
Volumetric resolution)
(ultrahigh
Image reconstruction
Application
General Electric
Philips. Siemens 2XO.5/0.S-1.5
imaging
4X1/6-S
1.5-2
6-S
1.25/0.7
3-5/3
Volumetric
imaging
(standard)
4X1/3
0.75
3
1.0-1.5/0.5
1-2/2
Volumetric
imaging
(skeleton)
4X1/S
2
S
2/1
6-S/5
Volumetric patients)
imaging
(obese
4x2.5/15
1.5
6
3/1.5
6/5
Fast spiral scanning
(standard)
4x2.5/20
2
S
4/2
S/5
Fast spiral scanning patients)
(obese
4XO.5/1.5-2.75
0.75-1.375
3-5.5
0.5/0.5
0.5/0.5
Volumetric resolution)
imaging
(ultrahigh
4x 1/5.5
1.375
5.5
1.5/0.7
3-5/3
Volumetric
imaging
(standard)
4X1/3
0.75
3
1.0-1.5/0.5
1-2/2
Volumetric
imaging
(skeleton)
4x2/11
1.375-1.75
5.5-7
2.5/1.3
S/5
Volumetric patients)
imaging
(obese
4X2/11
1.375
5.5
2.5/1.3
6/5
Fast spiral scanning
(standard)
4X3/16.5
1.375
5.5
4.0/2
S/5
Fast spiral scanning patients)
(obese
N = number of active detector rows; SC = section (mm), RI = recon. Increment (mm)
collimation
(mm); TF = table feed (mm), SW = section
width
a Pitch factor
P=TF(NxSC); Volume pitch P* =TF/SC parameters for reconstruction of an overlapping 3D volume (secondary raw data set) from which MPR and 3D displays can be generated; not necessary for fast spiral scanning unless MPR or 3D display is required c parameters for image review and reporting (may vary depending on clinical indication), may be reconstructed by thick MPR from 3D volume (secondary raw data set) d user interface displays SW = 1.25 and 2.5, respectively e needs to be reconstructed by thick MPR from 3D volume (secondary raw data set)
b
as well as to situations that require the use of a greatly restricted dose (e.g., pediatrics and pregnancy). When switching to a more smoothing kernel,images at first appear "blurred" but radiologistsquickly get used to the new appearance.
Apart from classic HRCTapplications, most soft tissue imaging will not require the extra spatial resolution provided by a sharper kernel (see Fig. 1.4). Hence, they can be used to decrease image noise at a lower exposure dose. In obese
125
126
4 Optimization of Scanning Technique Table 4.6
Suggested scan parameters for multislice
Acquisition
CT scanning of the body (16-slice scanners)
Image reconstruction
Pitch factor"
Application
P
p.
SW/Rlb
SW/Rlc
16xO.625/5.625
0.5625
9
0.625/0.4
0.625/0.5
Volumetric olution)
imaging (ultrahigh res-
16xO.625/13.75
1.375
22
1.25/0.7
5/4
Volumetric
imaging (standard)
16xO.625/9.375
0.9375
15
1.25/0.5
1-2/2
Volumetric
imaging (skeleton)
16X1.25/27.5
1.375
22
1.25/0.7
7.5/5
Volumetric imaging (abdomen or obese patients)
16X1.25/27.5
1.375
22
1.25/0.7
5/4
Fast spiral scanning (standard)
8x2.5/27
1.35
10.8
2.5/1.25
7.5/5
Fast spiral scanning (obese patients)
16XO.75/7.5
0.625
10
0.8/0.5
1/1
Volumetric olution)
imaging (ultrahigh res-
16xO.75/18
1.5
24
1/0.7
5/4
Volumetric
imaging (standard)
16XO.75/11.25
0.9375
15
1/0.7
1-2/2
Volumetric
imaging (skeleton)
16x 1.5/36
1.5
24
2/1
7/5
Volumetric imaging (abdomen or obese patients)
1.5
24
2/1
5/4
Fast spiral scanning (standard)
1.5
24
3/1.5
7/5
Fast spiral scanning (obese patients)
16XO.5/5.5
0.6875
11
0.5/0.3
1/1
Volumetric imaging (ultrahigh olution, peripheral skeleton)
16XO.5/11.5
1.4375
23
1/0.5
5/4
Volumetric
imaging (standard)
16XO.5/7.5
0.9375
15
0.5/0.5
1/2
Volumetric
imaging (axial skeleton)
16x 1/23
1.4375
23
2/1
7/5
Volumetric imaging (abdomen or obese patients)
16x1/23
1.4375
23
1/1
5/4
Fast spiral scanning (standard)
16X2/46
1.4375
23
3/1.5
7/5
Fast spiral scanning (obese patients)
NxSC/TF General Electric
Philips/Siemensd
16X1.5/36 16X1.5/36
e
Toshiba" res-
N = number of active detector rows; SC = section collimation (mm); TF = Table feed (mm); SW = section width (mm); RI = reconstruction increment (mm) "Pitch P = TF / (N x SC); Volume Pitch P* = TF/SC b parameters for reconstruction of an overlapping 3 D volume (secondary raw data set) from which MPR and 3 D displays can be generated; not necessary for fast spiral scanning unless a second plane of reconstruction of 3 D display is required C parameters for image review and reporting (vary depending on clinical indication, and may be reconstructed by thick MPR from secondary raw data set) d pitch factors may have to be varied according to newest developments by the manufacturers e optional: 8-slice mode (8x3 mm) may be used in very obese patients on a Philips unit
patients with a large body cross-section, the pixel size is large (0.8-1 mm) and limits spatial resolution. In such a situation, the smoothing kernels not only reduce noise, but also do not markedly influence spatial resolution.
Judicious selection of the convolution kernel and mAs will help reduce the radiation dose. A smoothing kernel should be used in obese patients.
Reconstruction
•
Field
of View
The field of view (FOy) selection limits image reconstruction to a particular body region of interest to obtain the best possible spatial resolution (Table 4.7). For example. portions of the subcutaneous fat can usually be omitted from the scanned field. A constant FOV should be used throughout the scanned body region to allow for multiplanar reformatting and 3D reconstructions. For this reason. the FOV should be based on the widest portion of the scanned region. For high-resolution thoracic CT, some authors recommend a separate reconstruction of each lung to obtain the smallest possible FOV. However, the gain is minimal compared with a tightly limited field confined to both lungs. In addition, some modern scanners allow for reconstruction of a 10242 matrix, which will provide a large field of view at optimum spatial resolution.
Table 4.7 a Effect of reconstruction filter kernel spatial resolution (example of single-slice scanner) Kemel
3.2 Lp/cm
5 Lplcm
Smooth
4.0 Lp/cm
6.2 Lp/cm
Standard
4.8 Lp/cm
High resolution
8.5 Lp/cm
7.8 Lplcm 12.6 Lplcm
Very smooth
Exact
numbers
resolution hgher
a small objects b
vary
between
for multi slice
manufacturers,
4J
scanners
that size are displayed
original contrast boundary resdutim
at 4
to two times with 20% of
%MTF
Table 4.7 b Effect of FOV selection attainable spatial resolution
FOV
on the
maximum
Boundary
Pixel size
resolution
S'l Zmm
1mm
5 Lplcm
256mm
0.5 mm
10 Lp/cm
200rrrn
O.39mm
12.0 Lp/cm
150mm ~xel size limits
O.29mm spatial
Choice of reconstruction small FOV
on
Boundary resomon"
Resolution at 20%MTFa
17.6 Lp/cm
resolution filter
for large
kernel
FOV
is limiting
for
and Display Parameters
A limited field (FOV of approximately 25cm) can be useful in studies of the adrenals and pancreas. Reconstructions including the whole cross-sectional body area should additionally be obtained if abnormalities outside this field of view are suspected (e.g., pancreatic cancer). For CT angiography, it is generally unnecessary to reconstruct a 'whole-body cross-section". A FOV of 20-25 em can substantially improve the delineation of small vessels and intraluminal pathology (pixel size: 0.4-0.5 rnrn). Excessive magnification (FOV too small) only amplifies the apparent noise without, improving spatial resolution (FOV < 15 cm 5 is rarely advised). 2. With a large FOV, high-resolution kernels give poor results and are therefore not recommended (Table 4.7). Window
Selection
The window level (or center) should be chosen close to the CT number of the primary structure of interest (e.g., 50-70 HU for most soft tissue imaging). The window should be wide enough to capture all relevant attenuation numbers (e.g., fat should not be completely black and enhanced vessels not completely white) but as narrow as possible if low attenuation differences shall be detected (e.g., liver imaging). In general, no structure of interest should be completely white or completely black, unless more than one window setting is available (e.g., for the chest). The window width should also be increased for imaging high-contrast structures. This particularly applies to the lung, where a window setting that is too narrow (800-1000 HU) can cause pleural lesions to be missed or can exaggerate the apparent size of small lesions and bronchial wall thickness. Narrow windows are useful only in the detection of emphysematous changes. A large window width decreases the visibility of image noise and may be used for low-dose examinations in which local contrast is not critical. In studies that are strictly confined to high-contrast structures (e.g., dedicated skeletal studies), a wide window setting can be used along with a substantially reduced radiation dose.
127
Optimization
of Scanning
Technique
sponse of the human observer. This may improve local contrast for wide window settings but makes it difficult to compare window settings to other scanners.
Some manufacturers (e.g., Philips) use a gradation curve that assigns gray levels nonlinearly to the CT numbers within the chosen CT window to compensate for the gray level re-
Image Processing
cr.
Image processing gives the best results with data sets that were acquired with thin collimation. For spiral and multislice CT, a thin section width and a reconstruction increment with at least 50% overlap should be chosen for a "secondary raw data set" that serves as the basis for optimum quality of reformations and 3D reconstruction with the acquired data set. In most cases, the section width should be chosen some 20-30% wider than the collimation in order to reduce noise and optimize image quality for a given patient exposure (see above). Image processing, apart from thick axial reformations, is required whenever further evaluation of a volumetric data set is intended. Table 4.8 gives an overview of the types of processing that we suggest for various clinical indications.
single-slice spiral The technique of thick axial reformations is the more effective the thinner the primary sections and the higher the overlap. If such axial sections are reformatted, the operator can also correct for positioning errors, e.g., at the skull base, the neck, the shoulders, or the pelvis. In general. axial. coronal or sagittal reformations should be aligned not to the CT table but to the patient anatomy. The workstation should have a simple tool to perform this realignment of data to patient anatomy. The commonest further processing techniques are coronal or sagittal multiplanner reformations. They are routinely indicated for all situations that require a second imaging plane for optimum evaluation. This is the case for many musculoskeletal applications, such as for the cervical, thoracic, and lumbar spine. For joint trauma in the shoulders, elbows, hands, hip, knee, and foot, biplanar reformations are mandatory. In the chest and abdomen, most diagnostic questions can be resolved by axial sections alone. Coronal, and sometimes sagittal reformations are mainly required for planning surgery, e.g., for staging of bronchial cancer, complex liver lesions, pancreatic tumors, kidney
2D Processing Axial images with a larger section width can be reformatted from the secondary raw data set, even if the scanner does not allow for direct reconstruction of such thick axial sections. The technique can be used for multislice CT as well as for
Table 4.8
Suggested
Processing
image
processing
Technique
Thick axial sections Correct
Standard
for positioning
Coronal
or sagittal
techniques
depending
on clinical
indication
Application
errors
reformations
Petrous Tumor
viewing
technique,
bone,
shoulders,
staging,
bronchi,
noise reduction
pelvis,
intervertebral
bile ducts,
stomach,
disks bowel,
urinary
bladder,
kidneys Interactive
reformation
Curved
planar
Shaded
surface
Maximum
displays
intensity
Minimrm Volume
reformations
intensity rendering
projection projections techniques
Complex
pathology,
Vessels,
pancreatic
Complex
anatomy
CT angiography Bronchi,
pathology aorta,
(standard),
rendering
Virtual endoscopy
CT angiography Preoperative Colon,
interfaces
bile duct
skeletal
lung nodules,
structures interstitial
lung disease
biliary system (standard),
skeleton,
planning Perspective
near tissue
duct, common ofthe
planning
rarely bronchi
or arteries
complex
pathology,
preoperative
3D
tumors, and gastrointestinal malignancies, The evaluation of bowel diseases is improved when coronal reformations are available, Interactive reformation is required in more complex cases that cannot be resolved adequately on axial sections alone, Interactive reformations would be an ideal primary diagnostic interpretation mode but hardly any present workstation is fully capable of the (relatively simple) prerequisites for time-efficient work with this viewing modality (Table 4,9), Basic prerequisites include real-time interaction even for a large data set, full-screen display of the reformatted plane, and easy and intuitive tools for interaction, Interactive choice of the width of the reformatted plane is especially desirable with thin-section multislice imaging because image noise becomes the single most important limiting factor for image quality, In addition, a simple way of aligning the three imaging planes to the actual morphology makes image interpretation more easy, For more complex pathology, interactive variation of the imaging plane is helpful but thick sections should be available for the same reasons as described above, With most present workstations interactive scrolling speed is still insufficient when thick (oblique) sections are reformatted, although appropriate PC-based hardware has been available since the late 1990s, Curved planar reformations (CPR) are used mainly for vessel tracking in CT angiography and allow for optimum display of wall changes and eccentric plaques, and are the prerequisite for accurate measurement of curved vessel segments (e,g" prior to interventions), CPR are mandatory to evaluate the vessel lumen in case of vessel calcification, e,g" of peripheral runoff vessels, but semiautomated programs are required to make this procedure time-efficient Coronal CPR can be Table
4.9 CT workstations: prerequisites use of multi planar capacities of multi slice Real-time
irteraction
even for large data sets
Interactive choice ofthe width ofthe matted plane (noise reduction) Full-screen
Easy
display ofthe
and intuitive
Interactive
reformatted
reforplane
tools for interaction
control of scrolling
speed
Tools for aligning the three orthogonal planes to the actual anatomy • Interactive
for optimum
CT
adjustment
(with thick sections)
of the imaging
imaging plane
Processing
used as a standard display tool for longitudinal evaluation of the sternum, the spinal column, and the larynx and trachea, Coronal CPR through the pancreatic or common bile duct are used for CT cholangiography and CT pancreatography, Coronal CPR through the ureters are excellent for demonstrating the position of ureter calculi and for differentiating between calculi and ph leboliths. Such CPR, however, are quite time-consuming and difficult, and are made easier if the curve can be interactively corrected,
3D Processing
Shaded surface displays (SSD) have been all but replaced by high-quality (>5122 matrix) volume rendered displays, They are helpful for complex skeletal fractures (especially for the acetabulum and the shoulders) and can be used to display complex vascular anatomy (especially of the thoracic aorta), Maximum intensity projection (M I P) is an excellent display modality for CT angiography, especially where small vessels are concerned, For larger vessels and the display of three-dimensional relationships, volume rendering is superior. With the advent of (semi)automated bone removal, MIP has become a time-efficient imaging tooL Still, time can be saved if the images are prepared by the CT technologists, Sliding thin-slab MIP, however, is an excellent tool for evaluating complex anatomy of small vessels, MIP holds promise for the evaluation of pulmonary parenchymal disease but is still not routinely used for this application, Minimum intensity projections (mIP or MinIP) have only fringe indications in the chest (display of air-filled tracheobronchial structures and emphysematous bullae in the lungs) but other display techniques such as volume rendering are often superior. Volume rendering techniques (VRT) have replaced many other 3D display modalities with the advent of fast workstations and dedicated accelerator boards, VRT have become the standard technique for rapid display and evaluation of vascular disease in CTA Often, only minimum editing is needed for good results, VRT is replacing SSD for skeletal trauma, VRT displays should become standard for initial evaluation of trauma cases, At some point in the future, VRT will probably replace conventional axial sections and even
129
130
4 Optimization of Scanning Technique multiplanar reformations, especially when thinslab techniques and opacity settings that simulate conventional CTwindowing are used, Perspective rendering may make interactive movies look more realistic but it rarely yields additional information over other rendering tools, especially when such tools are combined with sectional planes that may reveal otherwise hidden structures, Using it depends very much on individual user preferences,
Virtual endoscopyis an endoluminal visualization technique which has the advantage that the evaluated structures are substantially enlarged, thus makingit possible to detect even subtle abnormalities, It has a proven value for the differentiation of colonic polyps and folds while the practical value for other organ systems (such as the larynx or bronchi) is very limited,
Special Scanning Techniques Quantitative CT (QCT) Quantitative CT is used primarily for the assessment of bone mineral density, It is most commonly used to assess the bone mineral content of the first three lumbar vertebrae, but in principle QCT can be used in other spinal segments and in the neck of the femur. It requires special analytical software and a suitable calibration phantom, The technique is described fully in Chapter 25 (Musculoskeletal System),
Dual-Energy CT Dual-energy CT is used for the detection and quantification of substances with a high atomic number, such as calcium, iodine or iron, The technique is based on the principle that the xray attenuation and CT numbers of these substances are strongly dependent on the tube voltage that is used, The method has found clinical application in bone densitometry (DEQCT) and for assessing liver iron in patients with hemosiderosis or hemochromatosis, The technique can reduce measurement errors due to variations in the fat content of the bone marrow or liver. Dual-energy CT is used only sporadically at present The base material decomposition technique, although more accurate, is no longer available with current scanners, The postprocessing dual-energy technique is still available and can be performed on any scanner. It involves taking two successive CT scans at the level of interest using two different tube voltages (e.g. , 80kVp and -140 kVp) , Though less precise than the base-material decomposition technique, the postprocessing technique is useful for assessing liver iron and bone mineral
content, which are estimated from the mean CT numbers in the region of interest (ROl) measured at the two energies, The scans can be performed at a low dose, provided a sufficiently large ROI (> 1 cm) is selected, I The attenuation difference (ACT) between' the two measurements is roughly proportional to the iron or bone mineral content of the tissue, Generally, this technique is satisfactory for the assessment of hepatic iron, Calibration phantoms are used in quantitative CT to improve accuracy and further reduce measurement errors due to fat Phantoms may consist of solid calibration standards composed of calcium hydroxyapatite or aqueous solutions ofk2HP04, To correct for fat-related errors, the j absorption difference in fatty tissue must be ! estimated by direct measurement in the subcutaneous or abdominal fatty tissue or in a standard ethanol solution, The ethanol solution is F less accurate, because its atomic composition is not identical to that of fat The following equation is used to calculate the mineral concentration Cm in the ROI from the concentration c of the calibration material and the attenuation i differences ACT for the ROl, ACTM for the calibration material, and ACTp for the fatty tissue: em
=
c x fACT-ACTFXACTM-ACT
A
This type of analysis requires an accurate determinationofACTp,otherwisethemeasurement errors would negate the advantage of the technique, Generally, the results are slightly less reproducible than conventional (single-energy) CT measurements, because the negative expressions in the equation tend to amplify the effect ofstatistical variations, For this reason, the results of the method are seldom worth the increased effort
5 Radiation Dose and Image Quality M Prokop
Imponant Dose Parameters
_
Volume CT Dose Index (CTDlvoi) (p. 132) Dose-Length Product (DLP) (p. 134) Effective Dose (E) (p. 134) Other Dose Measures (p. 136) Guidelines (p. 136) Scanner
Technology
_
Scanner Geometry (p. 137) Focal Spot Tracking (p. 138) Geometric Efficiency (p. 138) Detector Efficiency (p. 139) Electronic Noise (p. 139) Noise Filtering (p. 140) Tube Current Modulation (p. 141) I-Axis Tube Current Modulation (p. 142) Dose Traps in Multislice CT (p. 142) Dose and Image Quality
_
Image Noise (p. 143) mAs Settings (p. 144) Tube Voltage (p. 144) Reconstruction Filter Kernel (p. 145) z-Filter (p. 146) Section Collimation (p. 146) Section Width and Pitch (p. 147) Body Diameter and Body Region (p. 149) Field-of-View(p.150) Window Setting and Perceptible Image Noise (p. 151) Dose Reduction
Negligible Reduction in Image Quality (p. 152) Trade-offs in Image Quality (p. 152) low-Dose
Scanning
Prerequisites for Low-dose Scanning (p. 153) HRCT(p.154) Screening CT (p. 154) CT Urography and Urolithiasis (p. 155) Children (p. 155)
_
5 Radiation Dose and Image Quality
Radiation Protection
156
Patient Selection (p. 156) Dose n Children (p. 156) Selection of Parameters (p. 157) Radiation Shielding (p. 158) Exposure of the Examiner to Radiation (p. 160)
The absolute values for radiation exposure in CT depend strongly on the scan parameters, scanner characteristics, and the patient. They may be 5 to 100 times higher than with conventional radiographs of the same anatomic region. This emphasizes the importance of radiation dose and tailoring the scan parameters to the individual patient.
Important
Dose
There are numerous parameters that de- ( scribe the radiation dose to the patient but the I following three are most important and have gained international acceptance to describe the local dose (volume CT dose index, CTDlvoi),Ihe total scan dose (dose-length product, DLP) and the radiation risk (effective dose, E).
Parameters
Volume CT Dose Index (CTDlvoi) The volume CT dose index (CTDlvoi) is a tool to indicate the average local dose to a patient within the scan volume. Its unit is the mGy (milli Gray) and there are separate data provided for body applications (measured in a 32cm PMMA phantom) or head scans (l6cm PMMA phantom). The CTDlvoi is a measure of the average local dose delivered by CT (conventional slice-by slice, single-slice or multislice spiral scanning) to a cross-section of such a phantom. The CIDlvoi is the most accessible dose indicator because it can be directly displayed on the user interface of modern CT scanners. It proivides immediate feedback about the dose delivered to the patient. Such a display is required by EU regulations, but outside the EU some manufacturers only provide display of the CTDlvoi when requested by the customer. The CTDlvoi allows for direct comparison of the radiation dose from different scan parameter settings, even for scanners of different manufacturers. It already corrects for the influence of pitch, mAs, kVp, filtering and other scanner-specific parameters. It does not, however, indicate the precise dose for any in-
dividual patient, but is rather an index of dose for a particular scanner and examination. The CTDlvoi is derived from the primary direct measurement of dose in CT, the CT dose index (CTDl) , which is defined as the integral under the dose profile D (z) of a CT section: 1 D(z)dz N.SC CTDl measurements are made at various positions within a body (32-cm diameter) or head (16-cm diameter) PMMA phantom. The phantom provides scattering media similar (though of course not identical) to a patient. The dose profile is markedly wider than the section collimation SC or section profile because of beam divergence and scattered radiation (Fig. 5.1 a). As a result, contributions from adjacent sections increase the local radiation dose during scanningofwhole body regions (Fig. 5.1 b). In theory, CTDl measurements must include all dose contributions from the tails of the dose profile. As this is often not practical, finite measurement limits are used. The subscript 100 in the term CTDhoo denotes that the measurement was made over lOOmm. The term CTDlw (weighted CTDl) represents the average radiation dose across the diameter of the phantom (in body scanning, the periphery and center measureCIDI
rs
133
Volume CT Dose Index (CTDlyoi)
1.0 0.8
2.0 Dose profile
CTDlvo\
Slice profile
1.5 (])
Vl
0.6
0
U (])
>
~rc
0.4
1.0
(])
e:::
0.5
0.2 0.0 -5 a
-4
-3
-2
-1
0
2
em
4
5
0.0 -10
o
-5
em
5
10
b
Fig. 5.1 Dose profile of a single CT section (a). The area under the curve describes the CT dose index (CTDI). Acquisition of multiple contiguous sections increases the local dose due to contributions from adjacent sections
(b). The maximum dose in the center of the scan volume corresponds to the volume CT dose index (CTDlvoi) if the scanned volume is large enough and a relative plateau is reached.
ment can vary by a factor of 2). The CTDIwis factor P [= TF / (NxSC) 1 and is then termed measured using a 100mm ionization chamber volume CTDI (CTDIvoi): that is placed at various positions within a CTDIvoi= CTDIw/ P 32cm (body) phantom (Fig.5.2a). The CTDIw is then calculated from the measurements in the 0::::::::--:: 1 em center and the average of the four measurements in the periphery, 1 cm below the phantom surface: CTDlw =- CTDIc+ -j- CTDlp The CTDIwis a very good measure of the average radiation dose within the scan volume for contiguous CT scans. To extend this to noncontiguous axial scans or spiral scanning with a pitch + 1, the CTDIw has to be corrected by the pitch
32em
o
16 em
o
o o
a
250% 200 % ..-
Fig. 5.2 Measurements of the CT dose index (CTDI) are performed in the center (CTDlc) and 1 cm below the surface (CTDlp) of a cylindrical Plexiglas phantom of 32 cm diameter for the body and 16cm for the head (a). The resulting CTDlw decreases with larger phantom diameters (b), which implies that the real average dose is underestimated for children and slim patients.
_
.
o b
(])
>
~
~
e<::
100% 50%
O%+---------~-------,---------,--------, o 30 10 20 Phantom diameter (ern)
40
Radiation Dose and Image Quality Hence, CTDIvoi represents an average over the scan plane (x and y dimensions) and the length of the scan (z dimension), It must be understood, however, that the actual local dose to the patient differs from the body CTDIvoi because attenuation properties differ from a 32 cm body phantom (Fig.5.2b), In a larger patient more dose is absorbed in the periphery so that the local dose in the center is less than it would be for a slimmer patient (assuming all scan parameters are unchanged), Dose to the periphery also decreases for larger patients, This is because the peripheral dose is accumulated from all x-ray tube positions and, in a larger patient, the x-rays that have traveled through the patient have been more strongly attenuated, As a consequence, the actual average dose within the scan volume (for the same scan parameters) may be lower for very large patients than the CTDIvoi indicates, In smaller patients or in organ regions with less attenuation (neck, chest) the opposite holds true, While the CTDlvoiis a valuable tool for protocol comparison, it underestimates the average dose within the scan volume for children and slim patients, and overestimates it in obese patients, Some manufactures correct the CTDlvoi tor children, depending on the chosen protocol or field-of-view, This makes comparison between manufactures difficult
Dose-Length Product (DI Pl The dose-length product DLP is a measure ofthe cumulative dose (total energy) delivered to a patient Its unit is mGy • ctn. It not only takes into account the average dose within the scan volume (CTDlvoi)but also the scan length L
DLP~CTDlvoiXL The DLP thus bears witness to the intuitive' fact that, for example, the dose must be less for a limited CT examination of the abdomen as compared with a CT examination of the chest, abdomen and pelvis, even if the same CTDlvoiis used, For conventional (non-spiral) scanning the scan length L is the sum of all section collimations
(e,g" 25xl mm for HRCT=25mm), For spiral scanning, one can use the difference in table positions of the first and last section as a good estimate for the scan length L However, spiral and multislice CT oversample data at the beginning and the end of the scan range because this data is needed for raw data interpolation of the first and last sections, There are differences between the various manufacturers, but approximately one half a rotation at the beginning and another half rotation at the end have to be added to the radiation exposure of the patient Thus, the scan length, as provided by the scanner, should be expanded by at least one table feed, • There is a considerable dose increase due to oversampling at the beginning and the end of the scan when short spiral acquisitions are employed, • The dose increase is particularly high in 16-slice scanners and short scan ranges (children),
Effective Dose (E) The radiation risk to the patient can be estimated from the effective dose E (unit: m Sv-rn llll Sievert), The effective dose can be estimated for a standard male or standard female using mathematical modeling, which computes organ conversion factors that vary along the z-axis depending on the anatomic region. Various computer programs are available that perform this task for a wide range of scanner types and organ systems, These programs can calculate dose for individual organs, which, when appropriately weighted using the organ weighting factors from lCRP-60, can be summed to calculate E. The mathematical modeling done to compute the organ conversion factors used to calculate E are performed on standard sized "mathematical" human phantoms, Separate data are available for males and females of various ages (infants through adults), However, in clinical practice, the data for a standard adult (70kg) is most frequently used, Such estimates of E will underestimate the radiation risk for children and slim patients and overestimates it for obese patients,
Effective Dose (E) Table 5.1 The effective dose E can be estimated as a percentage of the CTDlvoi or using a conversion factorfor DLPw for each scan phase. The estimates have been derived using data from a German survey of over 800 institutions (Galanski and Stamm) and from data for standard patients as described in the ED report 16262 EN. The estimates strictly hold true only for standard patients. For variation of the scan length, the DLP yields better estimates for the effective dose, while the CTDlvoi is probably superior for patients of different body height Organ
Neck Chest Liver Pelvis Abdomen + pelvis Chest + abdomen
Scan length
Conversion E/CTDlvoi
20cm 30cm 20cm 20cm 40cm 60cm
20% 42% 32% 41% 73% 96%
factors for estimation EIDLP
of
0.0090 0.015 0.018 0.017 0.017 0.016
L=scan length, E= effective dose (mSv), CTDlvoi = volume CTdose index (mGy), DLP = dose-length product (mGy -crn). EUAdata from EUR 16262; all other data from Stamm, 2001 (personal communication)
In addition to this size dependence, the in-' creased radiation sensitivity of children causes them to have an age-dependent radiation risk that is 2-3 times larger than that of the general population. A less computationally
intensive but still reasonably robust method of estimating E has been proposed that relies on the reported values of DLP. Individual organ doses are not computed, but rather for a given scan region (head, chest, etc.) a conversion factor from DLP to E is used. Since the calculations are based on standardsize patients, the DLP for very tall or very short patients (each having the same values of CTDIvoi) may over or underestimate E. For these cases, it may be helpful to redefine these conversion coefficient in terms of CTDIvoi (Table 5.1). For any method used to calculate a value of E, it is essential to remember that such values are estimates. The many assumptions used in the calculation process are likely not accurate for any given patient. As with CTDIvoi estimates of E do not indicate the precise radiation risk for any individual patient, but are rather an index of risk for a particular scanner and examination. The risk of death from radiation- induced cancer has been calculated from BEIR V and ICRP-60 data, which extrapolate the risk estimates from
accidentally or occupationally exposed groups (e.g. the atomic bomb survivors). Much of this data reflects high doses and high-dose rates and hence risks from diagnostic radiology procedures (typically low doses and low dose rates) must be extrapolated. It is extremely hard to detect an additional risk in the diagnostic dose range because all other environmental and hereditary risk factors appear to have a much larger influence. That is, the radiation-induced risk per mSv is approximately 0.5/10.000 persons, while the risk of a fatal cancer is about 3.000/10.000 persons. Because radiation risk is markedly age-dependent (Table 5.2), in-utero risks associated with the exposure of a pregnant woman should be considered. These are discussed in Chapter 20 (Female Pelvis).
Table 5.2 Calculated radiation-induced risk (ICRP 60) of dying from cancer. However, such risks can not yet be statistically proven for an effective dose below 20 mSv
Age
Risk (death per mSv)
Child (0-10 years)
14 /100000
Adolescent
18
(10-20 years)
/100000
Adult (20-30 years)
7.5 /100000
Adult (30-40 years)
3.5 /100000
Adult (60 years)
2.0 /100000
Adult (80 years)
1.0/100000
Average
5.0)100000
135
5 Radiation Dose and Image Quality
The entrance skin exposure (£S£) is ameasure of the skin dose but is not a good measure of Apart from the three most practically useful patient dose because it is strongly dependent on dose measures described above, a vast variety of scanner geometry and the energy spectrum ofthe other dose measures are available, x-ray beam, Identical ESE may lead to vastly The CTDlpoA was the first well-defined and different CTDlvoi (by factors of two or more), broadly used measure of dose in CT and is still The energy imparted is the most physically commonly used by manufacturers, It is identical correct measure ofthe radiation exposure, It deto the CTDlwo definition described above but is scribes the energy of the radiation that remains based on measurements over a length of 14 conwithin the body or an organ, It is more complex tiguous sections instead of a fixed length of to determine than the CTDl and therefore is not IOOmm The difference between the CTDlpDA in widespread use in clinical practice, and the CTDlioc is negligible for large collimaOther measures like air kerma (K) are important for dosimetry, One has to keep in mind tion (=27 mm): but the CTDlpDAmarkedly underestimates the dose for narrow collimation that correction factors have to be employed to (A 5 mm) such as in thin-section spiral or multitranslate air kerma into absorbed dose to a slice CT If in doubt whether the CTDlpDAor specific material (also in mGy) , although this CTDhoo is provided by a manufacturer, compare may be done automatically by some measurethe num bers for thick and thin collimation: the ment equipment CT01pDAfalls as the collimation decreases, while the CTDlloo remains constant or even increases, The multiple slice average dose (MSAD) is Guidelines the average dose delivered within a scan region of multiple contiguous scans (see FigS lb), Apart from various local regulations, few guidelines have been established yet, although both Conceptually, the CTDlvoi is identical to the MSAD ifthe 100-mm ion chamber collects all of the European Union and the United States work the dose in the tails ofthe dose profile, on such issues. The CJDlair (dose free in air) is the dose The European Guidelines for Quality in measured in the center of the gantry with all abComputed Tomography EUR 16262 suggests sorbers removed, Because no phantom is re- maximum dose levels for CTDlvoi and DLP that should not be exceeded unless there is a clinical quired it is easier to determine than the head or body CTDL It was used for dose estimates and reason (Table 5,3), These dose levels have been some local European regulations, However, it established from a British survey in the late cannot take into account the influence of the 1980s and represent the 75th percentile of the dose levels applied at the participating institubeam quality (energy spectrum of the x-ray tions, In the late 1990s similar studies were perbeam) and scanner geometry (distance from formed in Germany, Switzerland, and Austria tube focus to gantry center) on the attenuation by the body, The relation between the dose in and found markedly lower dose levels based on spiral CT scanners (Table 5,3), There are now efair and the actual body CTDl varies strongly with scanner type and manufacturer. It thereforts to further refine the suggested dose levels fore is not particularly useful for clinical purwithin the EU, and to start similar projects in poses, It is very important, however, as an input the US, parameter to the mathematical modeling softThe suggested dose levels (CT01voi) proware programs that estimate organ doses and E. vided by the authors of this book (see Table 5,7) The organ dose is a measure that describes are the product of vast personal experience but the average dose to an organ (in nrSv). Depending may have to be adapted to local requirements on the size ofthe body cross-section and the loca(for dose and image quality) and to the intion ofan organ relative to the surface, it can be dividual patient The suggested CT01voirefer to a estimated from CTDlc, CTDlp, and CTDlw (after cor170cm170kg "standard patient", who is obrecting for the influence of the pitch factor P), viously quite slim, The CTDl will have to be inComputer programs are available that can calcucreased substantially in obese patients (somelate the dose more accurately for standard times up to a factor of3 or higher) because x-ray patients (male, female, children of various ages), attenuation doubles for every 4-6 cm in additional soft tissue diameter (see below), Other Dose Measures
Scanner Geometry Table 5.3
Average
-a
exposure dose found in surveys CTDlw[mGy]
in various
countries as compared to the EU guidelines
DLP [mGy*cm]
E [mSv]
EU Larynx Chest Chest (HRCT) Liver Kidney Abdomen
38 18 21 21 21
EU 003
33 15 30 16 17 15
30 35
327 327 748
45
G = German survey of 830 institutions
(Galanski,
A = Austrian
(Novotny,
survey of 15 institutions
SSs Swiss survey of 6 institutions
638 326 96 321 383
415
4ffi
300 500
2.4 6.4
650 280
700 277 580
5.9 5.9 12.9
780
2.4
3.4
4.7 1.5 5.9 6.2 8.6
9 13 4.8 10
2001) 2002 personal
communication)
(Aroua, 2000)
EU= EUR 16262 guidelines
Guidelines for (accidental) radiation exposure in pregnant women are discussed in Chapter 20 (Female Pelvis).
Scanner Technology
Scanner technology markedly influences image noise and image quality at a given radiation exposure to the patient. The following paragraphs discuss the most important factors that influence patient dose, as well as techniques that can be used to reduce patient exposure. Scanner Geometry
Scanner geometry strongly influences the patient dose (CTDI) at a given mAs setting. With
more rapid tube rotation, the centrifugal forces on the tube increase with greater distance between the tube and isocenter of rotation. For this reason, most newer scanners have moved the tube closer to the isocenter (Fig. 5.3). At a given mAs setting, this will increase patient dose, especially skin entrance dose, substantially. This phenomenon causes a potential dose trap: if identical mAs settings are used on a new scanner as compared to an old scanner (even from the same manufacturer), this maysubstan-
X-ray tube
/
,
, Fig. 5.3 Scanner geometry. A large distance (a) between the focal spot of the x-ray beam and the isocenter of rotation suffers from increased centrifugal forces on the x-ray tube. A short distance (b) increases patient dose and detector signal at constant mA settings. Surface dose is increased most.
I I
,
/
/
X-ray tube
,
,, ,
/
, \
/ / /
._- .. Scan field-of-view ---
,,
,
, \
137
5 Radiation Dose and Image Quality
tially increase patient dose if the scanner geometry has changed (e.g., a GE LightSpeed compared with a GE HiSpeed scanner). For this reason, manufacturers recommend adjusting the mAs setting downward so that CTDIvoi remains similar. Focal Spot Tracking
With fast rotational speed, the centrifugal forces on the electron beam within the x-ray tube cause small alterations in the position of the focal spot during a spiral scan. This can be solved by slightly widening the pre-patient collimation ("overbeaming"). Such a technique was used in the very first versions of multislice CT scanners and caused a substantial increase in patient exposure, especially for thin collimation, relative to similar single-slice scanners. Tracking the focal spot by slightly adjusting the collimator setting (focal spot tracking) will keep the dose on the detector constant without having to increase patient dose. Such techniques are now standard with multislice CT. Geometric Efficiency
The geometric efficiency of a detector is determined by the amount of radiation that actually hits the detector elements relative to the amount of radiation that leaves the patient. The geometric efficiency on the one hand depends on the width, spatial orientation, and absorption of the septa that separate single detector
elements. Such septa are necessary to avoid cross-talk between neighboring detector elements and to absorb scattered radiation. Thirdgeneration scanners have an advantage over fourth-generation scanners in that their collimators are centered on the tube focus and not on the center ofthe scan. In multislice CT, radio-opaque septa will slightly reduce geometric efficiency when matrix detectors with a wide total collimation are used. Radiolucent septa will keep geometric efficiency high but will be more vulnerable to scattered radiation. Additionally, geometric efficiency depends on the width of the dose profile (in z-direction) relative to the total width of the detector elements. In single and dual slice CT, there usually is no post -patient collimation, which means that almost all radiation that leaves the patient also hits the detector (Fig. 5.4). In multislice CT, the detector is subdivided into four or more active detector channels. Most manufacturers try to make sure that each active acquisition channel (detector row) "sees" the same amount ofradiation. This makes it necessary to widen the dose profile by opening up the pre-patient collimation (Fig. 5.4) and therefore reducing dose efficiency. In fact, the slope of the dose profile due to the penumbra is not utilized. Since the absolute width of the penumbra and thus additional radiation exposure due to the penumbra is almost independent of the collimation, exclusion of the penumbra will have the greatest relative effect if the total width of collimation (N x SC) is small. It will cause a markedly reduced geometric dose efficiency and thus increased CTDI (Fig. 5.5). As can be seen. the effect is greatest for a 2 x 0.5 mm
Focal spot '. The penumbra contributes to the image and to patient dose
Beam collimator
The penumbra contributes to patient dose but not to the image
Beam collimator
20 cm detector width Single slice CT
Multislice CT
Fig. 5.4 The penumbra of the x-ray beam is caused by the finite size of the focal spot of the x-ray tube (a). It is fully utilized for single and dual detector CT scanners (b) but is excluded for multislice CT scanners (c).
Electronic
detector configuration and improves with a 4 x 0.5 mm collimation. With 16-slice scanners, the effect is almost negligible.
Detector Efficiency Scintillation (solid-state) detectors generally have a 20% to 30% higher quantum yield than Xenon gas detectors. This implies that using scintillation detectors, the same image nose can be achieved with considerably lower dose, relative to Xenon gas systems. The total quantum yield also depends on radiation prefiltering, detector design, detector electronics, and scanner geometry, which tends to reduce the differences between different types of detector.
139
Noise
The smaller the detector signal, the more pronounced the role of electronic becomes. The effect is particularly significant for obese patients, low-dose protocols and thin-section imaging. In multi slice CT, where thin collimation is becoming routine, it is very important to
•
Strong effect with thin collimation
Penumbra -
1111111111111111116x1.25mm 20 cm detector width Effect decreases with adaptive array detectors
.-.
I III
Electronic Noise
t"
I
I 4x 1 mm
20 cm detector width
The electronics of the detector system (signal amplifiers) cause a constant noise level that is independent on the incident radiation dose (Fig.5.6, see also Fig. 7.44). Quantum noise, on the other hand, increases with the detector dose. This increase, however, is substantially smaller than the simultaneous increase in detector signal (square root function). With a high detector signal, the electronic noise is negligible relative to the quantum noise but as the detector signal decreases, the electronic noise may assume a similar order of magnitude to the quantum noise and then contribute substantially to the noise levels on the final CT image.
Effect decreases with thicker collimation 11111
I I I
111114x2.5mm 20 cm detector width Effect decreases with more detector rows
1111111111111111116x1.25mm 20cm detector width
240%
a
reI. CTDI vol
220% 4-slice scanner
200% Fig. 5.5 The geometric efficiency of the detector system depends on the amount of radiation excluded (penumbra) relative to the radiation collected by the x-ray detectors (a). The geometric efficiency decreases with thinner sections but increases if more active detector rows are used. As a consequence, the (TDI of 4-slice scanners grows markedly for smaller collimation but less so for 16-slice scanners (b).
180% 160% 140% 120% 100% ~
16-slice scanner
J-~~'8--,-~::===;===~===;:==-'---+ o
2
3
4
7 5 6 8 Section collimation S( (mm) ~ b
140
5 Radiation Dose and Image Quality Fig. 5.6 Electronic noise limits lowdose applications and thin-section imaging. It dominates the image as soon as it is larger than the quantum noise, New detector electronics can substantially reduce electronic noise for 16-row scanners making thinner sections possible even at reduced dose,
100 ~,
'''''',~~
" "'.Signal
10
-u
'"
,,-,
'0,1
0.01 0.001
'<,
Electronic noise (4-slice scanners)
-TI~-R~-d~-~~-;~-~-~~~-~~~~i~~~~~~;~~~--~--~--~-·~~----,
0.0001 --+----,------,--------,-----,-----------, 100.000
10.000
1000
1 00
10
Rei, detector dose
individually adjust exposure settings, especially for obese patients (see Fig. 7.33). Newer detector electronics with 16-slice scanners substantially decrease electronic noise relative to single slice CT or four-slice scanners (Fig. 5.6) and thus allow for thinner sections and lower dose settings than older scanner models. Noise Filtering
Various manufacturers have already included digital filtering options in their scanners that reduce the effect of low signal (and thus high noise levels) on individual detector elements. Such filters work on the raw data and average the signal from neighboring detector elements if the signal in these detector elements becomes too low. Averaging n detector elements has the
same effect as "locally" increasing the detector dose by the factor n. Because there are often only a few detector elements for each projection angle that have too Iowa signal (e.g., the central elements that cover the spinal column), the averaging influences only a small portion of the total projectional data. As a consequence, inplane spatial resolution is affected only minimally. Most recently, three-dimensional filtering of raw data has been suggested that uses a three-dimensional weighting function to locally average the data from adjacent detector elements, from similar projection angles, and along the z-axis if the detector signal for a particular projection becomes too low. Such a technique can substantially reduce noise-related streak artifacts, for example in the shoulder or pelvic regions (Fig. 5.7).
Fig, 5.7 Noise filters selectively average raw data from projections with low detector signal and, thus, high image noise, Compared to an image without noise filter (a), the image with noise filter displays less noise and less noise-related streak artifacts (b).
a
b
Tube Current Modulation
both noise reduction and improvement of image quality. By increasing the dose selectively for Tube current modulation takes advantage of the those projections in which there is a substantial fact that there is often a substantial difference in increase in attenuation (usually lateral projecradiation attenuation between the lateral and tions) but at the same time decreasing the dose anteroposterior (AP) diameter of a body cross- for projections with little noise (most other prosection. Because attenuation follows an ex- jections), the overall image noise may be ponential function, small changes in diameter decreased while radiation exposure to the will result in major differences in attenuation patient also is reduced. and, thus, related quantum noise at the detector. Various systems are presently on the For an oval body, the exponential nature of at- market that use either sinusoidal dose modulatenuation will yield only a few projection angles tion (for example SmartScan, GE) or on-line with markedly increased attenuation and a adaptive dose modulation (for example C.A.R.E. larger range of angles with a relatively low at- dose, Siemens). Other manufacturers offer simitenuation (Fig. 5.8). lar products and the user should ask the manuWhen looking at how noise propagates facturer to provide a specific description of the from single projections to the final image, one rnA modulation algorithm. The sinusoidal dose modulation technique finds that the projections with maximum noise have a disproportionate contribution to the re- modulates the rnA in a sinusoidal fashion acsulting image noise. This holds the potential for cording to the maximum and minimum patient Tube Current Modulation
Fig. 5.8
Adaptive dose modulation
relies on mA reduction for those projections with less x-ray attenuation (usually ap views) (a). With adaptive dose modulation (right side of image) there is substantially less noise and fewer noise-related streak artifacts (b) even if no noise filtering is employed. As a result, the quality of coronal reformations is improved (right side of image, c).
Absorption, mAs
40 35 30 25 50%
20 15 10 5 0 +-----~~~~~--~~==~----~----+O% 0.4 0 0.2 0.6 0.8 1
1.2
Tube rotations
b
c
141
142
5 Radiation Dose and Image Quality size (within the scan range) as determined by an anteroposterior and lateral scanogram. Some manufacturer's algorithms use only one localization image. Since the minimum and maximum diameters vary along the z-axis, the amplitude of the sinusoidal variation (both the minimum and maximum values) must be adjusted to account for variations in patient size (along the z axis). It is important to determine if a specific algorithm implemented on one's scanner incorporates such z-axis modulation. Finally, because the actual attenuation varies much more than the diameter (Fig. 5.8), a sinusoidal modulation may not fully exploit the dose saving potential of mA modulation. The on-line adaptive dose modulation technique uses the attenuation information from every 1800 rotation during a spiral scan to modulate the mA setting in the following rotation to compensate for the measured differences in detector signal (Fig. 5.8). This technique takes full advantage of the potential of dose modulation. However, the early implementation of this technique keeps the maximum mA per rotation constant.ln the region of the neck and shoulders, for example, this will lead to an increased exposure to the neck if the modulation is to work properly for the shoulders. For multislice CT with wide total collimation (N x SC), large table feeds are used. This will lead to problems in regions with rapid changes in patient size along the z-axis (e.g., in the shoulder region). Adaptive dose modulation allows for dose reduction by some 10-30% without sacrificing image quality, depending on the shape of the examined body area. Noise-related streak artifacts can be reduced substantially. For low-dose applications, however, electronic noise may render the techniques less useful than the noise filtering methods described above. More advanced algorithms can be expected soon.
Z-Axis Tube Current Modulation As noted above, longitudinal dose modulation along the z-axis adapts the dose to the local requirements in the body (Fig. 5.9). Such a technique (e.g., Smart mA by GE) lowers the mA settings for the neck and chest, and increases it for the shoulders or the abdomen. It is a further advance over adaptive modulation and uses the attenuation information of either an AP or lateral
rnA Longitudinal modulation ConstantmA legs z-axis Image noise
, ~Ongitudinal
DV
modulation ~onstantmA
z-axis ~
Too much dose
c=::J
Too much noise
Fig. 5.9 Longitudinal dose modulation relies on mA reduction for those positions along the z-axis that have less x-ray attenuation (neck, lungs) and rnA increase for highly attenuating regions (shoulders, pelvis). Local dose varies while image quality remains more constant.
scanogram, the APjPA or lateral projection during a spiral scan or the continuous information from each spiral rotation to decide upon the overall mA requirements for a particular rotation. It is an important step towards "phototirning" for CT. However, there is little experience yet in how to correctly choose exposure settings using this technique.
Dose Traps in Multislice
CT
Multislice CT suffers from a number of "dose traps" that can substantially increase patient dose if mA settings are used identical to those for conventional spiral CT, even if a scanner from the same manufacturer is employed. Changes in scanner geometry (closer distance between tube and isocenter) will cause an increase in CTDI at identical mAs settings (e.g., GE LightSpeed versus GE HiSpeed scanners). Using narrow collimation (e.g. 4 x 11.25 mm) decreases geometric efficiency with four-slice scanners and leads to an increase in CTDI (see Fig.5.5b). This increase is approxi-
Image Noise
rnately 30-60% for 4xl mm or 4x1.25mm collimation (30% is still acceptable compared to the typical dose variations between sites), but may be as high as 150% with 2xO.5 mm or 2 x 0.625 mm collimation. For this reason. a collimation of 2xO.5-0.625 mm should be avoided for most body applications. This increase is no longer relevant for 16-slice scanners (see Fig. 5.5). The term effective mAs, which is used on Siemens scanners, refers to the values of the true mAs divided by the pitch factor. Some users find that this term makes it easier to choose an given level of noise, as pitch is already taken into account. However, confusion of the two terms (effective mAs and mAs) may lead to a
substantial increase in dose. This can occur if a user chooses to employ the same mAs settings that he previously used on a single-slice scanner of the same manufacturer. For example, a 200 mAs setting at a pitch of 2 (single slice) will correspond to IOOmAseff (multislice). Choosing 200 "mAs" on the user interface of the multislice scanner actually means choosing 200 mAseff, which will cause a twofold increase in patient dose compared to the single-slice system (all other determinants of dose, such as gantry dimensions, being equal). As a consequence, users should not look primarily at the rnA settings but choose the rnA in a way that ensures appropriate patient dose as indicated by the CTDlvoi (see Table 5.3).
SiBahdT mageQttft The minimum dose requirement in a CT examination is determined by the acceptable amount of image noise at the required spatial resolution. Acceptable image noise depends in turn on the necessary contrast resolution, the window width that will be used, the attenuation characteristics of the scanned region, and the diameter of the patient.
Image Noise Image noise (known also as pixel noise) is a measure of the statistical fluctuation ofCTnumbers and is measured as the standard deviation in a homogeneous region of interest. Image noise a is theoretically related to the section collimation SC (in mm): tube current 1 (in rnA), and section acquisition time t (in s) by the equation:
The acquisition time t per slice may equal several rotation periods RT in conventional CT, but in spiral scanning it is always equal to RT. Cp is a constant that depends on the convolution kernel, and k is a normalizing factor. The factor Cz(P) depends on the raw-data interpolation algorithm in spiral CT or on the z- filtering in multis lice CT. In spiral CT, it is usually a constant Cz, independent of the pitch. In multislice CT,
Cz(P) is strongly pitch-dependent (compare Fig. 1.28 b). From this it can be seen that, • The dose must be increased four times to reduce noise to one-half of the original value. • Using 2 mm collimation instead of 8 mm collimation leads to a doubling of image noise. • The convolution kernel has a major effect , on noise (Table 5.4).
Table 5.4 Influence of filter kernel on image noise and the dose required for constant image noise (measured for a Siemens Somatom Plus 4 scanner, Prokop, 1998)
Filter kernel
Relative
image noise
(%)
Relative dose require-
Spatial I resolution (Ip/em)-
ment(%)3 ABIO AB20 AB30 AB40 ABSO AB60 AB70 AB82 AB91
56 71 78
85 100 117 136 281 280
32 50 60 72 100 137 184 789 786
4.1 4.7 5.0 5.2 5.8 6.2 6.6 7.6 7.7
a measured in center of a 30 em water phantom b spatial resolution at 20% MTF (better reflects
clinical reality than cut-off frequency at 4% MTF)]
143
144
5 Radiation Dose and Image Quality The following additional quantities theoretically affect image noise by influencing the normalizing factor k: k = C.~ where M is the x-ray attenuation in the object and w is the sampling aperture. The factor C is an equipment-dependent measure of dose effectiveness (a small C denotes high effectiveness), which is most strongly influenced by the scanner geometry and the detective quantum efficiency of the detector system. The sampling aperture w is determined by the width of one detector element in the x-y plane and by the number of data samples per rotation. It is an important factor in the above equation, indicating that a smaller aperture and thus potentially higher spatial resolution leads to markedly increased noise.
mAs Settings Radiation exposure of the patient as well as the detector dose are proportional to the mAs settings. For a given rnA setting and rotation speed, the detector "sees" the same dose per rotation, independent of the pitch. Only with multislice CT interpolation schemes does the image noise vary with pitch at a constant mAs setting. The patient dose, however, is always proportional to the effective mAs (mAs per tube rotation divided by the pitch). Any measure to reduce patient dose will usually be coupled to a reduction in effective mAs.
As mAs settings per rotation are reduced, image noise grows. This has a major effect on contrast resolution but no effect on spatial resolution as long as the contrast of a small structure exceeds the noise levels by a factor of 2 to 3. Excessive reduction of mAs settings, especially in very obese patients and in combination with thin collimation, may reduce detector dose so much that electronic noise will dominate image quality. Reconstructing thicker sections or using smoothing filter kernels then can no longer recover reductions in image quality (see Fig. 7.33). Such an excessive mAs reduction, therefore, should be avoided by all means.
Tube Voltage For a constant mAs setting, the radiation output of the x-ray tube increases with higher x-ray energy. When the tube voltage is increased from 120 kVp to 140 kVp, the CTD! increases approximately by a factor of 1.4, while decreasing the tube voltage to 80 kVp decreases the CTD! by a factor of approximately 2.2 (Fig. 5.10). This effect may be useful for scanning the newborn and infants where the lowest possible tube output may be desired (the use of 80 kVp allows for further dose reduction below the minimum mAs settings available at 120 kVp). Conversely, the x-ray attenuation in the patient decreases with higher x-ray energy. This effect exceeds the increase in tube output with higher kVp, and results in a better dose utilization at higher tube voltages. This effect is especially important in the obese patient and in
160% 140% 120%
15 Iu Q)
100%
> ;::; ro
80%
c.:
60%
Qj
80
100
120
Tube voltage (kVp)
140
Fig.S.l0 At constant mA settings, the CTDI varies substantially with kVp. There are marked variations between scanners, depending on prefiltering of the x-ray beam.
Reconstruction Filter Kernel regions with high x-ray attenuation (such as the shoulders, abdomen, and pelvis). In such a situation, the highest possible tube voltage should be chosen (e.g., 140kVp). Finally, the relative difference in x-ray attenuation (contrast) between the various structures in the body increases with lower x-ray energy. At lower kVp settings, the CT numbers of metal, iodine, and calcium increase substantially, while the CT number offat decreases. The effect is especially pronounced for any contrast -enhanced (iodine-containing) structure: the enhancement at a given concentration of iodine is about twice as large at 80 kVp as compared to 140kVp (see Table 7.2). The effect can be used to optimize contrast-to-noise ratios at a given radiation exposure, or to decrease the radiation exposure at a given contrast -to-noise ratio. However, the positive effect of contrast is counterbalanced by the increased attenuation described above. When changing the tube voltage from the typical value, it is essential to increase mAs settings to compensate for the reduced tube output at lower kVp (which would increase image noise).
For contrast -enhanced examinations W the neck, chest, and in very slim patients, low kVp settings with properly increased mAs settings may be used (e.g., 80kVp, 150 mAs instead of 120 kVp, 100 mAs). This decreases radiation exposure (CTDI) and improves contrast enhancement. . For obese patients, especially for examination of the shoulders, abdomen, and pelvis, high kVp settings (e.g., 140 kVp) should be employed, Al For most other applications, a standard setting of 120 kVp suffices.
Reconstruction Filter Kernel
The filter kernel used for image reconstruction from the raw data has a substantial influence on image noise (Table 5.4). A slightly smoothing filter kernel will cause only minor reductions in spatial resolution but will require substantially less dose for constant image noise. On the contrary, increasing the spatial resolution by using a sharper filter kernel will substantially increase noise, often to a degree that can no longer be compensated for by raising patient dose (Fig. 5.11).
Although there are no firmly established rules yet, the following suggestions can be given: • For the newborn and infants, low kVp and mAs settings should be used.
300%-]' ReI. Dose Requirements 250%
200%'·······
150%
Fig. 5.11 Image noise depends strongly on the chosen filter kernel for image reconstruction from the CT raw data set. With growing spatial resolution there is a substantial increase in image noise, while a moderate reduction of spatial resolution (smoother filter kernels) allows for substantial dose savings.
50%
.
-
.
--'
O%+-----~------~----~------~----~ 60%
80%
100%
120%
140%
160%
ReI.spatial resolution
5 Radiation Dose and Image Quality
• In-plane spatial in for noise.
resolution
can be traded
z-Filter In spiral CT, the use of a 360' LI algorithm reduces image noise by 18% compared with conventional scanning, while a 180' LI algorithm increases image noise by 15% (Table 5,5), In multislice CT, image noise is strongly pitch-dependent (compare Fig,L28b), At present, only Siemens has chosen an implementation where image noise becomes independent of pitch if radiation exposure to the patient remains constant (see Fig, L29a), In this implementation (as with other vendors for a pitch of 6), the noise is similar to the noise seen with a 360' LI algorithm in conventional spiral CT Similarly it comes with an increase in section width of some 30%, Since multislice CT usually employs thinner sections, such a broadening of the section profile can be accepted easily because it leads to substantially lower dose requirements for identical image noise,
Table 5.5 Relative dose requiremerts raw data interpolation and z-filtering Z-filter
Pitch
algorithm
Conventional
Rel.section width (SW)
CT
depending
ReI. dose requirement
100%
1000/0
100% 66% 128% 110% 110%
133% 128% 66% 117% 93%
100% 128% 100% 128%
133% 66% 133% 66%
SpiralCT 180-LI 360" LI SmartHelical"
Multislice
CT
z-filtering HQ mode" HS mode" a
GE4-slice scanners
0-2 0-2 0.75 1.5
Use of a 180' LI algorithm increases noise while 360' LI decreases noise. A 360' LI algorithm may be used in very obese patients if mAs setting cannot be further increased. A 360' LI algorithm with thinner sections can be used at P = 1 as a standard in dualslice scanners for applications with high image noise (e.g., abdomen and pelvis). Use of an adaptive z-filter interpolation decreases noise.
Section
Collimation
Image noise increases as a thinner section collimation is chosen. This behavior parallels that of the mAs settings: reducing the collimation by a factor of 4 could be compensated for only by an increase in mAs by the same factor. Image noise increases by a factor of 2 if no such compensation is performed. The effect of section collimation SC on radiation exposure (CTDI) varies greatly between manufacturers. For most single-slice scanners, the CTDI remains constant for SC 5: 3 mm, and increases only slightly for smaller collimation (see Fig. 5.5). When a single detector row is used, not only the radiation from the ideal section profile but also the penumbra (see Fig. 5.4) of scattered and extrafocal radiation can be utilized. A dose increase may be seen when manufacturers use a post-patient collimation to improve the section sensitivity profile and open up the pre-patient collimation to compensate for the loss in detector dose. In some single-slice scanners employing post-patient collimation, the CTDI voi doubles when going from a 2-mm scan to a 1 -mm scan. With multislice CT, most manufacturers try to make sure that each active acquisition channel (detector row) "sees" the same amount of radiation. This makes it necessary to widen the dose profile by opening up the pre-patient collimation (see Fig. 5.4) and therefore reducing dose efficiency. In fact, the slope of the dose profile due to the penumbra is not utilized. Since the absolute width of the penumbra, and thus additional radiation exposure due to the penumbra, is fairly independent of the collimation, exclusion of the penumbra will have the greatest relative effect if the total width of colli-
Section Width and Pitch
mation (NxSC) is small. It will cause a markedly reduced geometric dose efficiency and thus increased CTDl (Figs. 5.4 and 5.5). As can be seen. the effect is greatest for a 2 x 0.5 mm detector configuration and improves with a 4xO.5 mm collimation. With l6-slice scanners, the effect is almost negligible. • Z-axis resolution can be traded in for noise. • Image noise doubles when 2 mm collimation is used instead of 8 mm collimation. • Reducing the collimation by a factor of 4. could be compensated by raising the mAsI settings by the same factor, but should be avoided in clinical practice. • With single-slice spiral CT, patient dose is nearly independent of the section collima1 tion • With thin-section multislice CT (four detector rows), patient dose increases by 30% for 4 x 1 mm collimation, up to 60% for 4 x 1.25mm collimation and by up to 250% for 2 x 0.5 mm collimation. • With thin-section multislice CT on l6-row is
Section Width and Pitch The section width determines image noise in a similar way as section collimation. In fact, section width is directly dependent on section collimation (and pitch) in single-slice spiral CT, but it becomes an independent parameter in multislice CTscanning. Section width in single-slice spiral CT is identical to section collimation for pitch of 1 (180 LI) and gradually increases by 28% up to a pitch of2 (180 LI). Such a widening by 28% also occurs with 360 LI at pitch 1. With increasing pitch factor P, the spiral is stretched and patient dose falls proportionally. This leads to the definition of effective mAs (=mAs/P) and CTDIvoi (=CTDlw/P), both of which compensate for the influence of pitch on dose. Consequently, a moderate widening of the section width by choosing a large pitch factor will cause a marked decrease in patient exposure. A pitch of 1 with a 180 LI interpolation is highly dose-inefficient and should be avoided wherever possible. It is the main "dose trap" for single-slice spiral CT 0
0
0
0
scanning. A pitch close to 2 is preferable (Fig. 5.12). Note that Philips is so far the only manufacturer that indicates the section width, not the section collimation, on the user interface. When scanning parameters are chosen to optimize z-resolution (i.e., constant table feed but reduced collimation), image noise will rise due to the smaller collimation. Choosing higher mAs setting can compensate for this. For example, patient dose and image noise with scrrF = 5110 at 200 mAs are identical to the patient dose and noise with scrrF = 10/10 and 100 As, but section width is 6.4 mm instead of 10 mm. • A pitch factor of 1 should be avoided with spiral scanning and 180 LI because of increased patient exposure at a given noise level and section width (Fig. 5.12). • Patient exposure remains constant if mAs are increased proportional to the pitch. • Z-Axis resolution can be improved at identical patient exposure (CTDlvoi) by reducing the section collimation and keeping the effective mAs constant. Image noise will remain similar. • Choosing dose settings according to the CTDlvoi, not according to the mAs, helps avoid dose traps. 0
Section width in multislice CT follows a more complex behavior with varying pitch, but it also ranges between 100% and 128% of the section collimation. Image noise shows a similarly complex behavior. At identical patient exposure, it is markedly higher if the section width is identical to the collimation, and is substantially reduced if the section width is increased by 28%. On GEmultislice CTscanners, dose requirements at identical image noise are about 50% higher for the HQ. mode than for the HS mode. While the HQ. mode allows for a section width that is identical to the collimation, the HS mode has a 30% wider section width. Note that the section width indicated on the user interface is only a rough estimate of the real values (see Table 1.8). With 8- and l6-row scanners the rule remains that higher pitch factors require less dose at a given image noise. Siemens uses z-filtering (four-slice scanners) and cone beam reconstruction algorithms (16-slice scanners) in such a way to keep noise
148
5 Radiation Dose and Image Quality
1000%
100%
sw=S(
10% 0
5
10
15
Section width SW(mm) .~ 1000% 0
c Q)
01
ro
E
.•.....
c
ru .•.....
Vl
C
0
U l-
cE 100% Vl .•.....
C Q)
E Q)
':::J
0Q)
Doselower than conventional (T
'-
Q) V1 "0
0 Q)
oc
10% 0
5
10
15
Section width SW (mm)
A.
1000%
100%
5mm Doselowerthan conventional (T
10mm
10%~-------------'-------------T-------------'
o
5
10
15
Section width SW (mm)
Fig. 5.12 Relative dose requirements for constant image noise on single-slice spiral CT images acquired with varying section thickness. Noise on a 5 mm conventional CT section is normalized to 100% (a). Dose requirements in conventional CT grow substantially with thinner section width. Note that section width and section collimation is identical in this case (b). Dose requirements in single-slice spiral CT are halved as the pitch increases from 1 to 2, but that the resulting section width increases by 30%. Note that for a given noise and section width, dose requirements with spiral CT are actually higherthan with conventional CT if low pitch factors are employed (c). For any given section width, thinner collimation and high pitch factors require the lowest patient dose. Note that the dose requirements for a section width of 10 mm are higher for SC =10 mm and P=1 than for conventional CT (SC= 1Omm) or spiral CT withSC= 7 mm and P=1.7.
Body Diameter and Body Region constant independent of the pitch. The default setting uses an increased section width (e.g .. SW= 1.25mm for 4x 1 mm collimation. and 3 mm for 4 x 2.5 mm collimation) and therefore has a reduced noise (as compared to standard CT sections) independent of the pitch. For the 4x 1 mm collimation. a setting is also available that reduces the section width to 1 nun, independent of the pitch. Because of a markedly increased noise level. this setting should be reserved only for few high-contrast applications. in which maximum z-resolution is required. Toshiba has a similar approach. As the user chooses thicker sections for a given collimation. the z-filters vary and the image noise decreases. The minimum section width is identical to the collimation. but suffers from a substantially increased noise as compared to the next wider section width (e.g .. 1.5mm instead of 1 mm). Philips uses the 360 MLI forlowpitch factors P ss 1. and 180 MLI for P > 1 on their four-slice scanners and an advanced cone beam algorithms (COBRA) for their 16-slice units. With the latter algorithm arbitrary pitches A 2 are available that have a comparatively low noise. 0
0
!. On GE four- and 8-slice scanners. the HS t and UF modes yield less noise for a conH stant patient exposure (CTDlvoi). 1 • The HQ. and UQ. modes should only be 1 used in applications in which artifact bel havior is crucial (e.g.. cervical spine, hands). • On Siemens and Toshiba scanners, reconI structing a section width that is identical 11 to the collimation should be avoided for H standard applications because of a substantial increase in image noise. • With four-slice scanners, high pitch factors can be chosen (p* = 5.5 for Toshiba, 6-8 for ( A Siemens and Philips) for standard applications, and low (P*=3) for applications, inwhich artifact behavior is crucial (e.g., cervical spine, hands). • Choosing dose settings according to the CTDlvoi, not according to the mAs, helps avoid dose traps. Reconstructing thicker section width from thinsection multislice CT raw data reduces image noise substantially, similar to using thicker collimation to begin with. The exact extent of the re-
duction in noise is greatly dependent on the implementation of z- filtering for this purpose. However, because of the reduced geometric efficiency at thinner collimations and the larger amount of electronic noise on four-slice scanners (see Figs.5.5b and 5.6), the image noise on thick sections reconstructed from 1 -2 mm multislice scans may be higher than in images of the same thickness reconstructed from thicker collimations. This holds true especially for obese patients. If a thick section is reconstructed from a secondary raw data set by multiplanar reformations, the noise reduction depends on the degree of overlap of these sections and the algorithm used for creating thick MPR. The effect of reducing noise by reconstructing thicker sections is more efficient for MPR perpendicular to the scan plane. This is because image noise on consecutive axial sections is correlated (depending on the reconstructed section width of the axial images that are the basis of the 3D data set used for MPR) while it is nearly independent for adjacent pixels within the scan plane (depending_on the reconstruction filter kernel and the FOV). Because of the increased attenuation in lateral projections, the noise on sagittal MPR is higher than on coronal MPR, especially for highly eccentric body regions such as the shoulders or pelvis (Fig. 5.13). Recommendations for MPR widths are given in Table 4.3. • Multislice CT gives a unique opportunity to maintain z-axis resolution and trading through-plane resolution (MPR thickness) for noise. • Reconstructing thick sections from a thinsection secondary raw data set is most effective if there is sufficient overlap (at least 50% of SW) of the axial images comprising this secondary raw data set. • Image noise is reduced most effectively on coronal reformations. Such MPR require less thickness than axial or sagittal MPR.
Body Diameter and Body Region Noise doubles in the center of a cylindrical water phantoms with every 4-6 cm increase in diameter d. depending on kVp settings, x-ray filtration, beam geometry and software noise fil-
149
150
5 Radiation Dose and Image Quality
a
c
b
Fig. 5.13 Comparison of image noise on 3 m m wide axial (a), coronal (b) and sagittal images (c), reconstructed from a volumetric data acquisition with 4 x 1 mm collimation. This patient could not elevate the
arms above the head. coronal image, and is sagittal image that are the posterior areas of
ters. In humans, the body cross-section is more elliptical. The minimum diameter (di) and maximum diameter (d2) of this cross-section determine an effective diameter d', which should result in identical noise as in a cylindrical body of this diameter. Ideally, d' should be equal to /\ di x ds' but in many scanners, noise filtering will reduce the influence of the larger (lateral) diameter, thus making a linear approximation de Cl xdi + C2xdz more appropriate. Noise ina human body, therefore, will increase less than in a water phantom, especially when noise filtering is employed by the scanner software. In addition, the subjective assessment of image noise does show that noise is less disturbing in obese patients because of a larger amount of body fat that serves as an "intrinsic" contrast material, and because of the larger FOV (see below) that must be employed with such patients.
There is less attenuation in the chest due to the high proportion of air in the lungs. Therefore, compared to the abdomen, noise increases less quickly in the chest as the patient diameter increases. Consequently, dose requirements for the chest are substantially lower than for other regions in the trunk.
X-ray attenuation in the patient rises exponentially with body diameter, so obesity leads to a disproportionate increase in noise. A useful rule of thumb is that the noise level doubles with every 4-8 cm increase in effective body diameter (except for the chest), depending on kVp settings and noise filtering.
Note that noise is least for the highest for those regions on the at the level of the shoulders and the chest (arrowheads).
Field of View
The FOV not only changes the pixel size but also the character of the image noise. As the FOV becomes smaller, the noise appears coarser and becomes more troublesome. Depending on the convolutionkernelandothernoise-determining factors, there is a minimum size for the FOV (approximately 25cm for the trunk, otherwise 15cm) below which the image will only have a noisier, grainier appearance with no improvement in spatial resolution (see also Table 4.4). • A large FOV creates a more finely speckled noise pattern but also reduces spatial resolution. • A small FOV in children makes noise appear coarser and also increases the effect of image blur, especially when smoothing filter kernels are used.
Window Setting and Perceived Image Noise
a
b
Fig. 5.14 The perceptible image noise greatly depends on the window setting. An unenhanced liver scan requires a narrow window setting (a, 200 HU) to optimize contrast between the liver parenchyma and lesion. This also doubles the perceptible image noise compared
Window Setting and Perceived Image
Noise
The perceived image noise 0* denotes the noise level that is perceived visually on the film or monitor. It depends critically on the window width setting W: o*=fxo/W,
where f is a suitable normalizing factor that also depends on the gradation of the display medium and the sensitivity ofthe eye. Thus, the window width has a much greater impact on the visual perception of noise than the dose or collimation: changing from a window width of 400 (standard soft-tissue window) to a width of 200 (used for unenhanced liver studies) causes a two-fold increase in perceptible noise (Fig. 5.14). For this reason, highresolution convolution kernels should be used only with a window width> WOO (e.g., in the lung or skeleton). Conversely, the use of a soft kernel should be considered in cases where a narrow window width is required (e.g., in the liver); otherwise noise can be reduced only by increasing the patient dose.
c
d
with a normal window width (b. 400 HU). The use of a high-resolution (MR) kernel provides excellent definition of the lung parenchyma (c. window width of 1500 HU), but the mediastinum suffers from increased noise (d. 400 HU).
Increasing the window width setting can greatly reduce perceived image noise. Doubling the window width reduces perceived noise by 50%. Doubling the window width also reduces image contrast by 50%. Consequently, structures with a high contrast require a significantly lower radiation dose (doubling the window width reduces the dose requirement by 75%).
151
5 Radiation Dose and Image Quality Dose Reduction
All dose reduction relies on the reduction of CTDIvoi or scan length. Negligible Reduction in Image Quality
On-line adaptive dose modulation will yield excellent results for most body regions where there is a marked difference in attenuation between AP and lateral diameters. This holds true especially for the shoulders but also for the chest, abdomen, and pelvis in slim patients. Dose reduction by some 10-30% will become possible, depending on the shape of the examined body area. Image quality is retained and even improved for the shoulders or the pelvis. Current implementations, however, suffer from a number of problems (see above). Longitudinal dose modulation adapts the dose to the local requirements in the body (e.g., less for the neck and chest, higher for the abdomen). This will become available soon (SmartrnA, GE; DoseRight, Philips; AEC, Siemens; Real EC, Toshiba) and will be an important step towards more constant image quality ("phototiming") in CT. New three-dimensional multislice CT interpolation schemes (noise filtering) substantially reduce noise without causing much smoothing of the data. This technique is not yet commercially available.
a
b
Fig. 5.15 The dose recommended by the manufacturer (a, 300 mAs, standard kernel) can be significantly reduced by the use of a soft kernel (b. 100 mAs). Note that the noise level is identical but the noise appears coarser.
Trade-ofts in Image Quality
The most well known trade-off is the increase in image noise if mAs settings are reduced and all other parameters are kept constant. Image noise can also be reduced by using more smoothing filter kernels for image reconstruction ("soft" instead of "standard", or "standard" instead of "high resolution" or "bone"). Thus there is a trade-off between dose and inplane spatial resolution if a smoothing kernel is employed together with appropriately reduced mAs (Fig. 5.15). Choosing a smoothing kernel for the soft tissues and a higher resolution kernel for the lungs or bones yields excellent results and allows for substantial dose savings (Fig. 5.16).
a
b
Fig. 5.16 The radiation dose in thoracic examinations can be significantly reduced (here: I = 40 mA, RT = 1.0s, pitch =1.6 => 25mASeff; CTDlvoi = 2.3 mGy) by using different kernels (HR and soft) to display the lung parenchyma (a) and mediastinum (b).
Image noise can also be reduced by choosing a thicker collimation or by reconstructing thicker sections (effective section width, SW) from thin axial data. This will reduce z-axis resolution but may be employed to reduce mAs settings. The multiplanar capabilities of multislice CT are best utilized by thin-collimation scan-
Low-Dose Scanning ning. However, patient dose (CTDIvoi) increases with such thinner sections, and image noise also grows substantially. Still, low-dose scanning is feasible with the following approach. An overlapping data set of thin sections (secondary raw data) is reconstructed first. This data set consists of noisy images (see Figs. 2.5, 2.14, and 5.18-19) but has high resolution in all three spatial directions. Thick multiplanar reformations are then performed from these data in any desired direction (axial, sagittal, coronal, oblique). These MPR will have good in-plane resolution as well as sufficient signal-to-noise (see also Fig. 5.13). The only drawback is the reduced through-plane resolution (resolution in the direction perpendicular to the MPR plane) due to the increase in MPRthickness. This is very similar to multiplanar imaging in MRI, where in-plane resolution is high, and a section thickness of 3-8 mm is employed. Image quality with this approach is excellent but-with four-slice scanners-the patient dose will be somewhat increased as compared to scanning with thicker sections because of the higher CTD] at thinner collimation (Fig.5.5a). With 16-slice scanners, geometric efficiency is substantially improved, and this negative effect is almost completely eliminated.
a
b
Fig. 5.17 Low-dose examination for colon cancer screening. Acquisition parameters: 4x 118 at 80kVp and CTDlvoi = 1.88 mGy, resulting in an effective dose of 1.3 mSv. Image noise is excessive with a soft tissue window setting (a) but is acceptable with a wide window
Low-Dose Scanning Prerequisites for Low-dose Scanning
Under the following four conditions, low-dose scanning becomes feasible: •
Reduced image quality requirements
Image quality is a critical feature in radiographic examinations. Reductions in quality are possible if there is pre-existing information about the disease (e.g., follow-up scans), if there is only one specific finding that is of importance in a certain clinical setting (e.g., presence of a polypoid lesion in the colon in colonic cancer screening) , or if only one organ system with high contrast has to be evaluated (e.g., lungs in lung cancer screening). Thick sections have been used as a standard in the past. If thick sections suffice (such as for the search for abdominal abscesses or hematoma) , dose may be reduced compared to protocols with a thinner section and similar signalto-noise ratio (e.g., by a factor of 2 if 10 mm instead of 5 mm sections may be employed). In multislice CT such thick sections may be reconstructed retrospectively as a means to improve in-plane image quality. Thin sections reduce partial volume effects and therefore increase the contrast of small structures. For this reason, an increase in image noise is acceptable on thin sections used for diagnostic re-
c setting ofW/L= 200010 (b) and is sufficient for virtual colonoscopy (c). In more obese patients, higher kVp settings should be used, and dose has to be increased to maintain image quality.
5 Radiation Dose and Image Quality view. The dose therefore needs not be proportionally increased as compared to thicker sections. •
Discontinuous
evaluation
of target organ
If the structure of interest can be evaluated in a discontinuous fashion. such as in HRCT for diffuse lung disease, effective radiation dose is reduced by a factor that is roughly identical to the relation between the width of the section and the table feed (e.g., factor 10 for 1 mm collimation and l Omm table feed in HRCT with a single-section technique). In multislice CT four or more sections are acquired simultaneously. Thus, the advantage of discontinuous scanning is reduced by that factor if no single-slice mode is available on such scanners. As a result, exposure dose for HRCT using such multislice scanners is still lower than with continuous CT but substantially higher than with single-slice CT. In addition, thin sections usually increase the CTDl somewhat, and will cause an increase in image noise that has to be compensated for by an increase in rnA settings. •
High contrast between the structures interest and their surroundings
of
High contrast makes it possible to employ wide window settings. Such wide window settings efficiently suppress noise (which is immediately evident when comparing lung and soft tissue window settings from HRCT of the lung, see Figs.5.14c, d). Thus, major dose reductions become feasible only if a wide window setting is required. This holds true for the lungs and the skeletal system. Improved contrast in CTA (by appropriate contrast injection protocols or scanning protocols) may be used to reduce patient dose while maintaining a reasonable signal-tonoise ratio between vessels and surroundings. •
Examined body part has little x-ray attenuation
Slim patients or children have lower x-ray attenuation than obese patients. This translates into lower dose requirements. The same holds true for organ regions with intrinsically low attenuation such as the lungs, the neck, and the extremities. For roughly every additional 4cm of soft tissue diameter, the required radiation dose for identical image noise doubles in phantom experiments. In clinical practice this increase is probably less (first results suggest some 8 cm for
doubling the dose) because obese patients contain more fat as an intrinsic contrast agent and thus are less vulnerable to noise, while children require somewhat sharper kernels for image reconstruction and thus require more dose than such a calculation would suggest. In practice, however, every 4-8 cm more soft tissue diameter requires double the dose, while for slim patients, the dose can be halved if the diameter is smaller by 4-8 em.
HRCT
High resolution CT (HRCT) of the chest is performed in a discontinuous fashion: 1-2 mm thick sections are acquired every 10-20 mm. This substantially reduces radiation exposure compared to spiral scanning. In standard size patients, the DLP should remain below 50mGyxcm, and the effective dose E should remain well below 1 mSv. Follow-up examinations or expiratory scans may be performed with further reduced dose.
Screening CT CT is excellent for screening of bronchogenic cancer patients since primary thin-
Multislice
section scanning allows for reconstruction of suspicious areas with a high spatial resolution and thus avoids additional scans in patients in whom a nodule was detected. In addition, it becomes much simpler to follow up patients with multiple nodules, especially when automated detection algorithms are used (Computer Aided Diagnosis or CAD). The mAs settings have to be adjusted to the size of the patient so that the CTDIvoi varies between 0.6 and 2mGy. Note, however, that image quality may deteriorate substantially if the patient is too obese and too Iowan exposure setting has been used (see Fig. 7.32, electronic noise). With single-slice spiral CT, a 5 mm collimation with a pitch of 2 is employed for lung cancer screening. Because of less noise on such thick sections than with thin-section multislice CT. an even lower CTDlvoi can be chosen, in the range of 0.3-1 mGy. However, additional thinsection scans are required if a nodule is detected.
Children Fig.5.18 CT Urographywith reduced dose. Acquisition parameters: 4x-1/8at120kVpand CTDlvoi= 1.88 mGy, resulting in an effective dose of 1.6 mSv. I mage noise on the thin-section images from the secondary raw data set (SWIRl = 1.25/0.7) is excessive (a) but can be substantially reduced when 20 mm thick curved planar reformations a re used to demonstrate the course of the ureter and the renal pelvis (b). Note the decrease contrast excretion by the tumor -bearing left kidney. a
In colon cancer screening there is a high contrast between the air-filled lumen and the bowel. Dependent on patient size, a CTDIvoi between 2 and 5 mGy can be used (Fig. 5.16). Smoothing reconstruction is mandatory and a section with of some 2mm is sufficient (and suffers from less noise) even for the secondary raw data set of overlapping axial images. CT Urography and Urolithiasis
CT urography after injection of IV contrast has an intrinsic high contrast. Thus, low-dose scans may be sufficient but mAs settings must be adapted to patient size (CTDIvoi =2-5 mGy). Thin-slab MIP or thick curved planar reformations (10-20 mm) parallel to the course of the ureters are used to evaluate the images (Fig. 5.1S) For the detection of urolithiasis, low-dose scanning is feasible. We found that even with CTDlvoi =2 mGy all therapeutically relevant stones > 2.5 mm could be seen in a 20x30 cm water phantom when thin-slab MIP were employed. However, noise increases quickly with more obese patients. A substantial increase in dose (5 mGy) may be necessary for this patient group.
b
Children
For the newborn and small infants it is advisable to reduce the tube voltage to SO-100 kVp because the CTDI falls with lower kVp. This allows for further dose reduction below the minimum mAs settings available with a standard of 120 kVp. Often as little as 0.3-1.5 mGy at SOkVp is sufficient (Fig. 5.19). The chest requires less dose than the abdomen. Contrast administration improves tissue differentiation and signal-to-noise, and therefore should be considered for examinations of the neck, mediastinum, and abdomen. For children between 4 and IOyears of age, 120 kVp with CTDIvoi= 2-5 mGy can be recommended (see Table 5.6). For all indications, a thin secondary raw data set should be obtained from which thicker sections can be reconstructed in arbitrary planes. The width of the MPR should remain as narrow as possible but as wide as necessary for a good signal-to-noise ratio (chest: 2-5 mm, abdomen 3-S mm).
155
156
5 Radiation Dose and Image Quality
a Fig. 5.19 Low dose examination in a 5-year old boy with a right-sided thoracic mass. Acquisition parameters: 4 x 1 18 at 80 kVp and CTDlvoi = 0.88 mGy, resulting in an effective dose of 0.35 mSv. Excessive noise on a 1 pixel thick coronal reformation from a secondary raw
data set that was reconstructed with a high-resolution kernel (a). The 5-mm thick coronal section from the same data set yields excellent image quality and demonstrates a right-sided interlobar effusion (b).
Radiation Protection
CT accounts for a large percentage (>30%, depending on the country) of the collective effective radiation from medical sources to which the general population is exposed. Patient selection, proper choice of scanning parameters, and radiation shielding can contribute to lowering overall radiation exposure. Patient Selection
The most important step is patient selection based on rigorous criteria. The question that is to be addressed by CT should be defined as clearly as possible, and the imaging modality itself must be capable of answering the question. The radiologist should also consider whether alternative modalities such as MRI or ultrasound could provide comparable information with less invasiveness. MRI should be considered as a primary imaging study mainly in young patients or in patients with non-neoplastic diseases that require frequent follow-up. Dose in Children
Radiation-induced risk in children can be substantially higher than the risk in the adult population. For this reason, the radiologist should
double-check the indication for CT, and should adapt the radiation dose to the body cross-section. Longitudinal and adaptive dose modulation would be ideal techniques, but presently only rough suggestions based on the body weight are available (Table 5.6). One has to keep in mind, however, that the body weight may underestimate the dose requirements in obese children because their body cross-section can be larger than that of a taller child of identical weight. Table 5.6 Suggested exposure settings depending patient size (CTDlvoiderived from data by Donnely, 2001 for aGE CT Ii). The pitch factor should always be 1.5 or larger (120 kVp) Weight(kg)
Chest
Abdomen/pelvis
4.5-8.9 9.0-17.9 18-26.9 27 -35.9 36-44.9 45-70 >70
1.7 mGy 2.1mGy 2.5 mGy
2.5 mGy 2.9 mGy 3.3mGy 4.1mGy 5.0mGy 5.8-6.2 mGy /\7.0mGy
2.9 mGy
3.3mGy 4.1-5mGy > 5.8mGy
Selection of Scanning Parameters Selection of Scanning Parameters
Table 5.7 gives recommendations for selecting the convolution kernel, slice collimation, kVp, and CTDIvoi for various clinical indications. The numbers indicated are based on a "standard patient" of 170cm and 70kg body weight (20x30 cm body cross-section), and have to be increased for obese patients. The following considerations are important in the selection of imaging parameters.
For objects with high intrinsic contrast (lung parenchyma and bone), spatial resolution is almost independent of image noise. Thus, the wide window setting used for skeletal imaging and HRCT of the lung permits the use of narrow collimation, a high-resolution kernel, and low mAs values (Fig. 5.16). Contrasts are lower in CT angiography, so a soft kernel must be used to compensate for the narrow collimation that is required (Table 5.7).
Table 5.7 Suggested exposure parameters for various clinical indications for standard patients (70 kg/170 ern). These exposure levels should be adjusted to individual requirements because dose efficiency and filter kernels vary between scannertypes and manufacturers
w- _ ,1'
_1
I
Iflndication
r:-t'-!!r--,.....,,~
Standard patients (70 kg/170 em) Neck (benign disease) Neck (tumor staging) I Chest (HRCT) Chest (benign disease) Chest (pulmonary embolism) Chest (tumor staging) Liver (pre-contrast) Liver (post contrast) Pancreas (tumor staging) Urolithiasis Abdomen (benign disease)
I
CTAchest CTAabdomen Obese patients (100kg/170cm) Neck (benign disease) I Neck (tumor staging) I Chest (HRCT) jfefChest (benign disease) Chest (pulmonary embolism) Chest (tumor staging) Liver (pre-contrast) Liver (post contrast) Pancreas (tumor staging) Urolithiasis Abdomen (benign disease) CTAchest CTAabdomen a b
C
smoothing kernel for the mediastinum, high resolution kernel (or display filter) for the lungs 5 mm sections for the mediastinum and lungs; additional 1.5 mm sections for peripheral tumors (multislice CT) thinner sections and higher dose are chosen for preoperative evaluation of tumor patients
157
Radiation Dose and Image Quality
For structures with low intrinsic contrast, detection of small objects (as described by a contrast-detail curve) depends on image noise, and small structures may be completely obscured by noise. As a general rule, one should use the convolution kernel that provides the best trade-off between low noise (smoothing effect) and adequate spatial resolution (acceptable unsharpness). While this trade-off is always based on a subjective appraisal, in most cases it can substantially reduce the dose requirement below manufacturers' recommendations. Contrast resolution in liver studies is critical, because there is often very little attenuation difference between a hepatic lesion and the surrounding parenchyma. Lesions smaller than 3 mm can almost never be detected. A narrow window setting is required for noncontrast examinations, and therefore soft kernels should be used to avoid an excessive increase in dose (Fig. 5.15). Intravenous contrast medium generally increases the attenuation difference between tumor and liver, so the dose can be reduced when a soft kernel is used (Table 5.7). In abdominal examinations, similar principles to contrast-enhanced liver studies apply. In CT examinations of the lungs and neck, relatively little attenuation occurs within the tissues. Since many neck examinations require a detailed structural analysis, a somewhat higherresolution kernel should be used than for mediastinal studies. The dose in thoracic examinations can be reduced by using different display kernels for the lung parenchyma and the mediastinum (Fig. 5.16). Intervertebral disk studies have the highest dose requirement: the attenuation difference between the disk and its surroundings is small, thin collimation is used, and abdominal attenuation is frequently high. Again, the dose requirement can be reduced by using a soft convolutionkerne1. The following parameters are critical: • Number of scans: are precontrast scans, biphasic scans, or delayed scans really necessary? • Scan length (preferably limited to the region of interest). • CTDlvoi (choose mAs judiciously, using a lower setting for thin patients and children) .
• Pitch (should be "1.5 or more whenever possible). Slice width (thicker sections, whether axial or multiplanar. reduce noise and allow for CTDl reduction). Convolution kernel (soft kernels reduce noise and allow for CTDl reduction).
Radiation Shielding Outside the scan volume lead shielding can be used to reduce scattered radiation. It has to be noted, however, that most of the scattered radiation arises from within the patient and therefore cannot be shielded. Lead shields must not be included in the scan range because they cause substantial streak artifacts that warrant repeat scanning of the affected region. Shielding of direct radiation is only in its infancy. Bismuth-containing shields have been suggested that are radio-transparent, but still reduce surface exposure substantially. Such shields are especially helpful for reducing dose to superficial organs such as the breasts, the testes, and the eye lenses. The more superficial the organs, the higher the radiation doses to which they are exposed. This particularly applies to organs that are directly exposed in the scanning field, but it also pertains to those receiving scattered radiation. Testes that are unprotected in the scanning field may receive organ doses higher than 20m Gy per examination. In women, the breasts are exposed to particularly high dose levels (up to 25mSv) if they are within the scanning field. The dose is several times that received in grid mammography, so particularly stringent criteria should be used in selecting young women for CT examinations. It should be noted that in women with large breasts, CT scanning of the liver will incur some radiation exposure to the mammary gland. This can increase the organ dose to more than 50mGy, especially in biphasic examinations. Consequently, the breasts should be positioned to minimize their exposure in liver examinations.
Radiation Shielding Table 5.8
159
Oose and image quality
Examination parameters
t i
mAs-settings Tube voltage
Remarks
radiation exposure radiation exposure
i
Section collimation Pitch facto r
Effect
i
radiation
image noise image noise
1800-lnterpolation
image noise
z-Filtering: SW = SC
image noise
z-Filtering: SW> SC
image noise
t
image noise image noise
High resolution
filter kernel
image noise
Window width
t
noise perception
t
Body cross-section
image noise
t
t = increase; t
i tt
image noise
it;
(t)
t
preferentially
t
between 1.5 and 2
adapt to indication
(see Table 2.2)
for P = 1 for P > 1 increased dose requirements decreased dose requirements
=
standard with multislice
CT
for mediastinum/abdomen
xy-resolution t t t ; contrast t
t ; dose t
as low as possible couple with mAs
only in obese patients
t ; z-resolution t t ; z-resolution t t ; z-resolution t t ; z-resolution = t ; z-resolution ( t ) t ; in-plane-resolution t t ; xy-resolution t
Smoothing filter kernel
Size of lungs
noise
t ; z-resolution t exposure t ; z-resolution
i
MPR-thickness
noise
image noise
3600 -Interpolation
Section width
t ; image i ;image
for lungs/bone for lungs/bone/CTA
requirements
t
reduce mAs chest only; predictable gram
from scano-
= decrease; = = constant
Table 5.9 Relation of mAs settings (at 120 kV) to CTOlvol (single slice data calculated from data from Nagel 2000. multislice data from www.impacscan.org). The multislice data hold true only for 5 mm section collimation. For smaller sections. the CTOI increases (see Fig. 5.11)
Single slice scanner GE9800/HiLight/HiSpeed
series
Philips SR series/AV series/Secura PhilipS/Picker PQ series Philips/Elscint/Picker
Twin Flash
Siemens AR series/Emotion/Balance Siemens Plus 4 series Toshiba Xpeed/Xpress/XVision
CTOlvol at 100 mAs
mAs at CTOlvol = 10 mGy
Pitch = 1
Pitch=2
Pitch = 1
Pitch=2
5.4
2.7
185
370
8.0
4.0
125
250
11.2 4.3
5.6
89
178
233
466
14.4 8.3 8.8
2.15 7.2 4.15 4.4
69
138
120
240
114
228
89
178
Shimadzu 7000 T
11.2
5.6
Multislice scanner
Pitch = 1
Pitch =2
GE LightSpeed series (SC = 5 mm)
10.1
5.1
99
198
7.0 7.3
7.0a 7.3a
143
143a
137
137a
11.2
5.6
89
Philips/Picker MX 8000 Siemens VolumeZoom Toshiba Aquilion
Multi
Pitch = 1
178
Numbers for 120 kV. For 140 kV the CTOI is larger by a factor of approximately 1.36. The mAs settings have to be reduced by 26% to keep the CTOI constant. a Siemens and Philips multislice scanners display the effective mAs-settings. Thus. there is no pitchdependence.
•
160
5 Radiation Dose and Image Quality Exposure of the Examiner to Radiation Only individuals who remain in the scanning room are subject to radiation exposure (interventional procedures, ICU patients, persons accompanying the patient) as structural shielding can easily be designed and installed that essentially eliminates exposure outside of the scan roo 111. Every scanner has its own isodose map with contour lines indicating the spatial distribution of scattered radiation (Fig. 5.20). As always, scattered radiation depends on the applied dose (mAs), tube voltage, and patient cross section. Dose to persons within the scan room can be minimized using the radiation protection principles of shielding (use of a lead apron), time (limit time within the scan room), and distance (move as far away from the scan plane and patient as is possible). Since the radiation emanates from the gantry aperture, a bystander will receive a minimum dose by standing at an angle next to the gantry or close to the narrow side of the gantry. Persons involved in interventional procedures should not leave their hands in the scan plane during imaging (especially during CT fluoroscopy). If necessary, this can be avoided by using suitable holding devices for the interventional instruments.
4
3 2
o -1
-2-
-2
-1
o
2
3
m--
Fig. 5.20 Typical isodose map showing radiation emissions from a modern CT scanner. The dose figures are stated per 100 mAs for 120 kV and a 10 mm slice thickness (CTD1w~ 8 mGy).
6 CT-Guided Interventional Procedures M. Prokop, A.J. van der Molen
CT-Guided Biopsies
....
General Principles (p. 162) Biopsies in the Chest (p. 167) Biopsies in the Abdomen (p. 169) Musculoskeletal Biopsies (p. 171) CT-Guided
Drainage Procedures
General Principles (p. 174) Specific Techniques in the Chest and Abdomen Specialized Interventions
<, (p. 177)
.............................•...
Liver Tumor Ablation (p. 178) Treatment of Osteoid Osteoma (p. 181) Spinal Interventions (p. 183) I Injection Therapy for Vertebrogenic Pain (p. 183) 6 Nerve Blocks (Neurolysis) for Pain Management in Cancer (p. 184) P Percutaneous Laser Disk Decompression (p. 185) ~ Percutaneous Vertebroplasty (p. 185)
j;
~-~
162
6 CT-Cuided Interventions! Procedures CT -guided interventions are common procedures. They can be diagnostic, such as biopsies or nerve root injections, or minimally invasive therapeutic procedures for drainage of abscesses, ablation ofliver lesions, nerve blocks for pain management, or as therapy of bone lesions.
CT-Guided
Biopsies
CT is an established guidance technique for collecting samples for cytological, histological, or bacteriological evaluation. It is particularly suited for sites that would be difficult to access with ultrasoundor fluoroscopy-guided procedures. The diagnostic accuracy and complication rate of CT-guided biopsies depend on the size and location of the lesion and the gauge of the percutaneous needle. Negative results must be viewed with caution and should be interpreted within the context ofCTmorphology.
General
Principles
Indications
Biopsies and abscess drainage still make up the majority of these procedures. CT serves as an important modality for guidance, but increasingly CT is complemented by ultrasound and fluoroscopy, sometimes in dedicated interventional suites.
and Contraindications
An important consideration in patient selection is whether other guidance modalities might be simpler or more effective for tissue sampling. CT offers advantages in cases that involve difficult routes of approach, small focal lesions, and bony or aerated structures. There are no absolute contraindications for CT-guided biopsies. Biopsy is relatively contraindicated in patients with deranged coagulation states (e.g., due to coagulation disorders or drug therapy) unsafe route of approach, uncertainity about the vascula nature of the focus, or other factors specified further on pp. 167. In general. adequately informed patient consent, preparation, post- interventional surveillance and aftercare are hallmarks of good practice and are as important for a successful outcome and a representative biopsy result as patient compliance and cooperation.
Risks
CT-guided biopsies generally have a low complication rate. Biopsy risks include nerve or vessel injuries with bleeding or paralysis, infection, and the dissemination of tumor cells (Table 6.1). The risk of bleeding is substantially increased Kn patients with abnormal blood coagulation (i.e., low platelet counts, anticoagulation), as well as for a transabdominal approach in patients with ascites. Other risks relating to the specific location of the biopsy site must be considered on an individual basis. The risk of tumor cell dissemination is low «0.5%). It depends on the puncture technique and appears to increase in biopsies of hypervascular tumors (e.g., hepatocellular and renal cell carcinomas (Fig. 6.1).
Table
6.1
CT-guided
needle
procedures
Risks Hemorrhage (up to 2%) Hemoptysis (2-5% after lung biopsy) Pneumothorax (10-60%, 10-30% of which require treatment) Pancreatitis (up to 2% after pancreatic biopsy) Abscess, empyema, peritonitis Renal injuries (very rare) Tumor cell dissemination (very Air embolism (rare) Risks
are dependent
(very
rare)
rare)
on:
Route of approach Number of biopsies Needle diameter Lesion vascularity
and puncture
Coagulation status Patient cooperation
(especially
technique platelet
counts)
General Principles
•
Position
A supine, prone, lateral, or oblique position may be necessary, depending on the approach. In all cases the patient should be placed in a stable, comfortable position with at least one arm extended above the head to avoid streak artifacts. Ample padding should be used. The table height should be set as low as possible to provide adequate space for the needle insertion and other interventional maneuvers within the gantry aperture. • Fig. 6.1 Metastasis caused by tumor seeding along the track of a fine-needle aspiration biopsy of hepatocellular carcinoma (HCC) in the left lobe of the liver. There is a coexisting contralateral adrenal metastasis.
•
Preparation
At the time a biopsy is scheduled, all possible approaches have to be considered on the basis of preliminary studies to assess the risks and costs of the procedure and determine the requirement of additional imaging. An up-to-date coagulation status, including platelet counts, must be available before the procedure. Reference values for safe biopsy are: platelets
> 70000/mm3,PPT <50s,Quick PT>50%
These values should be checked against the reference values of the reader's own laboratory. If possible, there should be timely termination of anticoagulation therapy before biopsy to allow a safety window for the procedure. Warfarin is terminated and IV heparin instituted instead. Patients with deranged clotting parameters may require substitution of their blood coagulation factors (e.g., fresh frozen plasma). Patients who are under direct threat of pulmonary embolism from deep venous thrombosis may require implantation of a temporary inferior vena cava filter before anticoagulation can be interrupted. If coagulopathy is suspected, extended clotting tests are obtained before planning the biopsy and special clearance by the hematology department may be required in particular cases.
Planning the Approach
Planning of the biopsy requires review of relevant previous cross-sectional and other imaging. A recent (contrast-enhanced) study of the biopsy region should be available. Alternatively, a diagnostic examination can precede the biopsy. However, in many cases, a limited non-enhanced study directly before the biopsy will suffice for safe evaluation of the access path. Additionally, this study serves as a test to assess if the position is comfortable for the patient. For abdominal lesions it is advantageous to have oral or rectal contrast medium on board to avoid bowel puncture. In doubtful cases an additional quick reference can be obtained with abdominal ultrasound to complement the assessment of bowel peristalsis. When selecting a representative area for tissue sampling, the operator should consider that the central part of a lesion often contains necrotic areas devoid of active tumor cells. In the case of pulmonary lesions, perifocal inflammatory reactions may occur at the periphery of the lesion (near the pleura), so the sample should be obtained as close to the hilum as possible. The route of approach should be as safe as possible (and the patient forewarned of any potential risks when consent is sought). Hence, it is essential to avoid nerves and vessels, the pleura and uninvolved organs if at all possible. In some cases the approach can be substantially facilitated by changing the depth of respiration, moving the patient's arm, or repositioning the patient. Approaches through the liver or lung parenchyma, or small bowel are acceptable, provided a sufficiently thin needle (18 G or smaller) is used and there is no significant amount of pleural or intraperitoneal fluid.
163
164
6 CT-Guided Interventlonal Procedures
b
a
c
Fig. 6.2 Needle biopsy of a small lesion in a lung transplantat patient. Planning (a), local anesthesia (b), fine core biopsy with a 20 G needle (c).
It is a sound general principle to take the simplest route of approach, bearing in mind that changing from a vertical or horizontal path to an oblique path, or even angling the gantry to place the needle path within the scan plane, increases the complexity of the approach. Only experienced examiners should select an approach that cannot be imaged in one plane of section. Multiplanar reformations of spiral or multislice CT data are useful in planning this type of approach. By viewing a suitable reference image, the operator can monitor the depth of infiltration anesthesia, the minimum distance between the needle insertion site and the lesion, and the maximum safe depth of needle penetration on the CT console. The cutaneous puncture site is marked on the z-axis with the laser-positioning device of the scanner (table position of the reference image). The location within the scan plane can be determined from anatomic landmarks (e.g., a vertebral spinous process), or from radiopaque markers attached to the skin (Fig. 6.2). •
Single scans will usually suffice for needle localization, but short spirals may be required for more complex approaches. With multislice CT, multiple sections are Simultaneously available, and the various manufacturers offer different solutions to take advantage of these sections. A good approach is to fuse the innermost sections to obtain a thicker target plane. The sections above and below this target plane serve as a control that easily demonstrates whether the needle deviates in a caudal or cranial direction and requires repositioning (Fig. 6.3).
Needle Localization Scans
For needle localization, it is generally sufficient to use the lowest scanning dose available (the smallest mAs and, if desired, the lowest kV setting). A foot switch and a display monitor in the examination room can vastly speed up the procedure, because the needle position can be quickly checked by the radiologist without having to leave the examination table.
Fig. 6.3 Multislice CT provides simultaneous sections. The inner ones can be used to define the target plane. Note that the needle is seen in the cranial image, indicating in which direction the needle deviates.
General Principles Guiding Systems
Various guiding systems are available to choose the correct needle path in a quicker and more reproducible way. Since less control scans are required. such guiding systems may also allow for a substantial reduction in radiation dose applied to the patient. The less experienced a user. and the more complex the access route. the more will the procedure profit from a guiding system. Simple guidtng devices help to define the correct angulation by providing an angular measure that can be directly attached to the needle or to the patient. Laser guides may either be part of a CT unit or may be provided by a different manufacturer and can be added to an existing unit. After planning the approach the correct angulation is transferred to the laser guiding system that indicates it as a laser beam either outside the gantry or inside the gantry opening. For systems that remain outside the gantry opening. no direct control of the needle position is possible as the needle tip is advanced. This has to be kept in mind when organs that move with respiration are biopsied. Even more sophisticated guiding systems are based on electromagnetic tracking of the tip and angulation of the needle. This information may then be overlaid on a reference CT data set that was acquired prior to the procedure. ifnecessary using contrast-enhancement. Such systems allow for complex angulation and reduced radiation exposure and promise not to compromise biopsy accuracy. Vital organ structures can be avoided more easily. especially since such systems (e.g .. the UltraGuide system) are aiso designed to compensate for respiratory movements. CT Fluoroscopy
CT fluoroscopy is a procedure that allows for online tracking of the needle as it passes through the tissue (see Fig. 1.33). The technique was developed for complex approaches or for regions with substantial respiratory motion (e.g .. lung bases). in which guided biopsies by conventional CT may be cumbersome. CT fluoroscopy should use low-dose settings because radiation dose can quickly become excessive with longer fluoroscopy times. However. such low-dose scans may suffer from
reduced image quality. which is particularly disturbing in difficult needle approaches. For practical purposes. fiuoroscopy should only be used intermittently during critical phases of the procedure. Having the patient change the depth of inspiration during fiuoroscopy may help bring the target lesion into the scanning plane. The needle is then advanced under fiuoroscopic control. Fluoroscopy with multislice scanners can demonstrate in which direction the needle leaves the target plane and thus make repositioning simpler. It has to be noted. however. that there is a potential dose increase if four sections are exposed instead of one. Speciai needle holders are available that make it possible to direct the needle without getting direct radiation exposure to the hand of the radiologist. Biopsy Needles
Percutaneous biopsies may be performed with fine needles (20-22 G). small core needles (1618 G). or large-core needles (14G and up). depending on the type of sample required. the risk of the approach. the anticipated pathology. and the experience of the evaluating pathologist. The smaller the lesion and the riskier the approach. the thinner the needle should be. Generally the presence of a trained cytologist and on-site analysis of the salvaged material help to significantly reduce the number of biopsy passes and to decrease the gauge of the required aspiration and guidtng needles. In many cases the biopsy needle can be substituted by aspiration cytology and unnecessary risk of hemorrhage and pneumothorax is avoided. This is especially important as the choice of the needle type (Fig. 6.4) or biopsy mechanism seems to have less effect on the diagnostic yield than the experience of examiner and cytologist. Aspiration needles can be of several types: Chiba needles (flat bevel tip). Franseen needles (serrated tip) or Westcott needles (with specimen side notch). Core biopsy needles generaily employ the Tru-Cut principle. There is a current trend toward semiautomated or automated high-speed biopsy instruments in which the trocar and core needle are individually or successively fired with a spring-loaded mechanism (Blopty, Temno, Gallini, etc.). There are severai coaxial biopsy systems in which multiple samples can be retrieved
165
6 CT-Cuided Interventional Procedures
Chiba
Greene
Westcott
Rotex
~
~
22,23
G
20,22
G
•
((E
18,20,22
18,19,5,21
~
G
G
22G
~
Fig. 6.4 Various types core biopsy needles.
G
22 G
~
Tru-Cut
G
16,18,20,22
~
Franseen
Otto
22,23
~
Turner
Madayag
making certain that the pleura, peritoneum, or periosteum are adequately infiltrated. For lung biopsies, the pleura should be anesthetized without piercing the pleural tissue with the needle. Striking a nerve will elicit a sharp, shooting pain, and the needle should be repositioned. If desired, localization of the inserted anesthesia needle can be performed to recheck the direction of the approach. Injections in children may be easier if a topical anesthetic cream (e.g., EMLA) is applied to the skin \hour before the procedure.
Size
Type
: ~
and sizes
14-22G.
of fine needles
and
through an indwelling outer cannula. Needles with graduated markings simplify the biopsy procedure, as do sleeve-type depth indicators that are mounted on the needle shaft. Some questions cannot be adequately resolved by fine-needle aspiration biopsy and cytology, and core biopsy is needed to furnish material for histological evaluation (Table 6.2). •
Local Anesthesia
The usual technique is to inject 5 to 20ml of local anesthetic (e.g., lidocaine, mepivacaine), Table 6.2 Techniques opsy indications
recommended
for common
Fine-needle aspiration adequate for: Pulmonary lesions: bronchial carcinoma, metastasis Hepatic lesions: HCC, metastasis Suspected pancreatic carcinoma Suspected lymph node metastases Core biopsy necessary for: Hepatic lesions: exclusion of benign disease Mediastinal tumors Retroperitoneal lesions Pleural lesions
bi-
Medication
for Sedation
and Analgesia
For some biopsies as well as for more complex interventions, conscious sedation may be needed for optimal results. In these cases adequate monitoring of pulse oximetry, ECG, heart rate and blood pressure is needed. A number of medications are frequently used, often in combination. Benzodiazepines give anxiolysis and sedation but do not provide analgesia. In older patients, and patients with cardiovascular disease, hypotonia and depression of breathing may be seen. Patients often like the short -term amnesia that may be associated with these drugs. They are contraindicated in patients with allergy, muscular diseases, and in pregnancy. Diazepam is the best known with good anxiolytic properties, but has a long elimination-time. For this reason, midazolam is more often used. Benzodiazepines can be antagonized byflumazenil. Propofoi in a selected dose can be used effectively for sedation in patients who breathe spontaneously. It is short acting and elimination is fast, but it has no analgesic effects. Dosing should be done carefully as overdosing may induce depression of breathing and there is no antidote available. Also, patients should be adequately monitored. In children, ketamine can be used for sedation in short procedures. It has both sedative and anesthetic effects. It is usually combined with atropine sulfate to counteract hypersalivation that might compromise the airway. Emergent reactions preclude the use of this agent in adults, but in children these are rare. For longer procedures it can be combined with benzodiazepines like midazolam. Opiates are mainly used for analgesia, but they may have some sedative effects in higher doses. Side effects include nausea, vomitting,
Biopsies in the Chest
breathing depression and hypotension. Frequently used examples are the strong opiates piritramid and the short-acting alfentanil and remifentanil. These opiates can be combined with benzodiazepines or propofol (dose adaptations'). Naloxone can be used as an antagonist for opiates. Non-steroid anti-inflammatory drugs can also be used for analgesia and also have antipyretic and anti-inflammatory profiles. They can be used to supplement opiates for pain reduction. They should be used carefully in patients with ulcer disease. asthma, and reduced renal function. Parenteral administration is possible with diclofenac or metamizole, but the latter can have serious side effects. Needless to say, adequate training should be sought before these drugs are administered by radiologists and their nursing staff. Hence, conscious sedation is often supervised by the referring clinician or anesthesiologist. Puncture Technique
A small stab incision is made in the skin, and the biopsy needle is introduced and advanced along the pre-selected path. It is easier to control the needle alignment when an assistant stands at the head or foot of the table. The laser guide of the scanner or external biopsy aids can be used to improve orientation. Depending on the risk of the approach, the needle may be inserted directly into the lesion or it may be advanced in stages while each stage is checked with a localization scan. The needle tip appears as a hypodense streak (hardening effect) in continuity with the axis of the needle shaft. In structures that move with respirations, the needle should not be fixed but should be allowed to move with the excursions, as otherwise it might lacerate the liver, lung, or other organs during respiration. The safest technique is to advance the needle while respiration is suspended. Sudden shooting pains signal that a nerve has been struck, and the needle should be repositioned. For a fine-needle aspiration cytology (FNAC), the needle is advanced to the surface of the lesion orjust inside it. The stylet is removed, a syringe is attached (20ml Luer-Lock), and the needle is moved in a fan-shaped pattern ("needling"). hence piercing the lesion several times, while suction is applied. The suction is released
before the needle is withdrawn to avoid seeding potentially malignant cells into the needle tract. When a core biopsy is performed with a core biopsy needle (or its automated or semiautomated variants), the biopsy needle is again advanced up to orjust inside the lesion. It should be noted that the core biopsy needle is surmounted by about a 5 mm-long trocar point that does not collect tissue. Advancing the trocar into the lesion loads tissue into the 5-20 mm-long sample notch of the needle, then the sharp-edged outer sleeve is rapidly advanced over the trocar to collect the tissue core. A spring-loaded mechanism may be used to accomplish one or both of these steps, usually providing a less hemorrhagic sample that is easier to evaluate. With larger core biopsy specimens simultaneous cytology "roll" smears can be made, or when coaxial technique is performed, the core biopsy can be preceded by fine needle aspiration cytology (FNAC). Final Check
After the sample has been obtained, a control CT scan should be taken to exclude a pneumothorax or acute hemorrhage. If a small pneumothorax is detected with CT, a conventional chest radiograph should be taken at 1 -4 hours after the biopsy and the clinicians informed. Depending on the needle size and route of approach, several hours' bed rest may be advised to reduce the risk of post biopsy hemorrhage.
Biopsies in the Chest Lung The location of the lesion is an important factor in deciding whether to perform a trans bronchial biopsy (central bronchial lesion, "positive bronchus sign" of a bronchus directly entering the lesion on CT) or a percutaneous transthoracic fine needle aspiration cytology (FNAC) or small core needle biopsy (peripheral lesion not related to the bronchi). Transthoracic biopsy is contraindicated in patients with a contralateral pneumonectomy, pulmonary hypertension, or pulmonary arteriovenous malformation and severe chronic obstructive pulmonary disease (COPD). FNAC of the lung is most often performed for the diagnosis of carcinoma, while core biopsies are more important for benign lesions,
167
6
CT-Guided Interventional Procedures
sclerotic lesions, lymphomas, and other noncarcinomatous malignancies, For diagnosing malignancy, transthoracic biopsy has reported sensitivities of 70-100% and high specificities, For benign processes, the yield is 85-95% but the specificity is much lower and varies between 15% and 65%, Prerequisites for a higher yield are repeated sampling from different parts of the lesion and expert on-site cytopathological advice with histological confirmation ff needed. Pulmonary lesions should be biopsied with the patient prone (if access is short enough) in shallow breathing and suspended respiration during needle movements, It is sound practice, especially with small lesions, to advance the needle into the lesion before the stiletto is withdrawn, In biopsies of extensive lesions, material should be sampled from perihilar portions of the lesion and not from its necrotic center. The risk of pneumothorax in lung biopsies may be as high as 40%, and patients must have adequate pulmonary function (Oz partial pressure > 60 mm Hg) and should not have serious emphysema or advanced pulmonary hypertension, The risk of pneumothorax is increased in patients with obstructive airway disease, emphysema, multiple needle passes, and in deep-seated or small lesions, Between 10 and 30% of all pneumothoraces have been reported to need chest drainage, The number of air leaks may be reduced by sealing the needle tract with Gelfoam or throm bin at the end of the procedure, Another method that has been successful for avoiding pneumothorax is a coaxial technique using a guiding canula with a sharp and blunt trocar. The sharp trocar is only used to advance the needle close to the pleura, which is then penetrated slowly using the blunt trocar. A sharp biopsy needle is only inserted close to the target lesion, This technique is less prone to causing pleural tears as the needle is advanced into the lung, The risk of a delayed pneumothorax can be reduced by positioning the patient with the biopsied side down, Small intrapulmonary hemorrhages along the needle track (see Fig,9,30a) are common, However, hemoptysis occurs only in 2% to 5% of cases, and more often at biopsy of vascularized
lesions, It is generally self-limiting, Rare complications like air embolism or needle tract seeding occur in 0.1 % and fatal complications have been reported in 0,2% of cases, Mediastinum
Percutaneous biopsy is appropriate for lesions that are inaccessible by the transbronchial route or by mediastinoscopy, It is also used in cases where a previous transbronchial biopsy has been attempted without success, Hilar lesions should be biopsied at a central site to avoid false-negative results from areas of posts ten otic pneumonia or atelectatic lung, The safe biopsy of mediastinal lesions requires that the location of vascular structures be known precisely, and contrast medium must be administered for vascular opacification, The risk of pneumothorax can be reduced By performing the biopsy at full expiration (increases contact area on anterior chest wall) and by injecting 20-50 ml of physiologic saline solution to displace the lungs laterally (Fig, 6,5), The chosen approach will depend on the location of the lesion, and is either anterior parasternal or posterior paravertebral. A direct mediastinal route is preferred to a trans pulmonary one to avoid the risk of pneumothorax, However, if a trans pulmonary approach cannot be avoided, FNAC should be preferred, Another option is the transsternal approach (Fig, 6,6), in which a hollow needle (e.g. a 2mm Ostycut needle (Bard) for biopsy with an 18 gauge cutting needle) is passed just through the sternum to establish access for the aspiration cytology or core biopsy, While FNAC has a high sensitivity and specificity for metastatic carcinoma, core biopsies are needed for non carcinomatous malignancies, lymphoma, and in benign lesions, For the diagnosis of thymoma, a surgical technique is generally suggested and may directly precede tumor resection, The success rate of mediastinal biopsies is Significantly higher with core biopsies (approximately 90%) than with FNAC, Pneumothorax is reported in some 10% of cases and hemorrhage in 3%,
Biopsies in the Abdomen
Fig.6.5 The route of approach can be expanded by the injection of physiologic salinesolution.
• Pleura CT is a good imaging modality in the identification of and guidance for aspiration and tap of loculated collections of the pleura, and biopsy of associated pleural thickening. A supra costal approach is mandatory to avoid damage to the intercostal neurovascular bundle. In most cases, the biopsy path is tangential to the thoracic wall. 18-20 G core biopsies give the best results in pleural masses, because the lesions are often inhomogeneous and immunohistochemical methods are needed for differentiation. For malignant disease, a sensitivity of >80% and specificity near 100% has been reported. In benign disease, the main role is identification of tuberculous pleuritis if CT morphology is equivocal. As with most biopsies, multiple passes are required for optimal accuracy. In experienced hands, major complications occur only in less than 1 % and include pneumothorax or hematothorax. There is a risk of tumor seeding, especially in mesothelioma. For mesothelioma and lymphoma, however, thoracoscopic (VATS) biopsy remain the reference standard. Biopsies in the Abdomen •
a
b
Fig. 6.6 Transsternai approach using a 2 mm Ostycut needle (a) followed by 18 gauge core biopsy (b) of a suspected recurrence ofT -cell non-Hodg kin's lymphoma.
Liver
In selecting patients for a liver biopsy, it should be decided whether the diagnosis could be adequately established by imaging procedures (typical findings in two separate procedures) rather than percutaneous biopsy. If hepatocellular carcinoma (HCC) is suspected, the risk of peritoneal tumor spillage would contraindicate liver biopsy in patients who require surgical treatment regardless of the biopsy result. An ultrasound-guided biopsy is preferred in many cases, because the biopsy can be monitored in real time and can often be completed during one inspiration. Care should be taken in patients with free intraperitoneal fluid as any liver biopsy can induce severe hemorrhage in these patients. Therefore, in patients with diffuse liver disease transjugular biopsy via a liver vein should be preferred, and in patients with focal liver pathology pre- interventional ascites tap may be helpful to reduce the risk of hemorrhage.
169
CT-Guided
imervenuone;
Procedures
A liver core biopsy is preferred over FNAC because of its greater accuracy, especially in the differentiation of benign hepatic lesions, The accuracy rate of core biopsy is greater than 90%, and its complication rate (bleeding, pneumothorax, puncture of other organs) is low, The individual route of approach to a hepatic lesion depends on its location and size, The shortest transhepatic route is generally preferable, Hemangiomas and other hypervascular tumors, should be biopsied through a longer, self-tamponading approach through normal liver tissue to reduce the risk of hemorrhage, If a subphrenic lesion cannot be biopsied in full inspiration via the extrapleural route, the biopsy should be performed in expiration or in the right lateral decubitus position to reduce the volume of lung tissue in the pulmonary recess, Biopsy employing gantry angulation can be beneficial and even trans pleural approaches have been used successfully with low complication rates in selected cases, Gallbladder,
Spleen
Percutaneous biopsy of the gallbladder is rarely indicated, because generally there is no difficulty in diagnosing a lesion that has expanded past the gallbladder to involve other tissues, Splenic masses are usually associated with systemic lymphomatous diseases that are more easily biopsied at other sites, Another reason to avoid gallbladder and splenic biopsies is their high rate of complications (biliary peritonitis and hemorrhage), FNAC should be performed only in highly selected cases, Pancreas
CT (fluoroscopy) and ultrasound are prerequisites for guidance of pancreatic biopsies, as they allow for real-time monitoring of the procedure, Using routine CT (non-fluoroscopy) guidance it is difficult to obtain a representative sample from small lesions, and negative cytology is generally unreliable, The sensitivity of FNAC is variable (70-100%) but is comparable to the accuracy of an intraoperative biopsy, The complication rate of pancreatic biopsy is low when a small-gauge needle is used, Many pancreatic lesions are difficult to detect without contrast medium, hence the planning scan should routinely include contrast CT Pancreatic biopsies can be performed through an anterior or posterior approach, A transgastric
or trans intestinal approach requires 6 hours' fasting to reduce the risk of contamination and infection, For safety reasons FNAC is generally preferred, The biopsy should be performed with a breath-hold in an intermediate position because respiratory excursions cause movement of the liver and other intraperitoneal organs relative to the retroperitoneal pancreas and deep abdominal respiration could withdraw an unfixed anterior needle from the pancreas, Kidney
Renal biopsy is generally not indicated in patients with renal cell carcinoma (RCC), because the biopsy sample cannot often differentiate the lesion from adenoma, the rich vascular supply increases the risk of hemorrhage, and tumors tend to disseminate into the needle tract or the lung, In doubtful cases the diagnosis should be established by surgical enucleation, which can be combined with intraoperative frozen section analysis, Renal biopsy for focal lesions is indicated only in patients with a suspected isolated renal lymphoma or metastasis, Exceptions are patients with tuberous sclerosis and equivocal findings in cross-sectional imaging, in whom repeated laparotomies cannot be performed, More frequent indications are core biopsies of native kidneys for diffuse disease or transplant kidneys for rejection, For best results, "14G core biopsies are taken from the lower-pole cortex, If ultrasound guidance is not preferred, the biopsy can be performed under CT-guidance by a posterior approach, and FNAC will generally furnish an adequate sample, Complications in the form of retroperitoneal or intraperitoneal hemorrhage or gross hematuria are rare, Adrenal Gland
Adrenal biopsy is appropriate in cases where a suspected metastatic lesion or indeterminate adrenal mass cannot be adequately diagnosed with CT MRI or other imaging procedures, In patients with a known primary tumor, FNAC is satisfactory for the detection of adrenal metastasis, In patients with a suspected primary adrenal tumor, an effort should be made to establish a histological diagnosis, The sensitivity and specificity are close to 100% with core biopsy and 90-100% with FNAC, If a pheochromocytoma cannot be excluded, appropriate pharmacological prepara-
Musculoskeletal
tion with' 0,1 and 0.2 blockade is indicated because of the risk of a hypertensive crisis. For this purpose the expertise of the local endocrinology department should be enlisted. Adrenal biopsy is best performed through a posterior approach with the patient positioned prone or with the affected side down. The adrenals can also be biopsied through a lateral approach or through an anterior approach (transhepatic, if necessary). Transrenal and transsplenic approaches should be avoided. Saline injection can expand the posterior paravertebral space and facilitate a posterior extrapleural approach (see Fig. 6.5). Retroperitoneal
and Mesenteric
Masses
A posterior paravertebral approach is most suitable for the biopsy of retroperitoneal lymph nodes and indeterminate retroperitoneal masses. Core biopsies are preferred, although injury to the lumbar arteries can produce extensive retroperitoneal hemorrhage. For lesions located between the vena cava and aorta, FNAC should be preferred using an anterior (trans intestinal) approach avoiding the colon. Nodular mesenteric masses in patients with an unknown primary tumor should be investigated by core biopsy from an anterior approach. Pelvic Region
The difficulty of differentiating scar tissue from recurrent tumor on cross-sectional irnagingespecially in rectal carcinoma-is perhaps the most typical indication for biopsy in this region. FNAC is usually adequate, especially since core biopsies can be very painful and, when scarring is present, often do not yield a diagnostic specimen. The complication rate of FNAC is extremely low, and its accuracy rate is higher than 90%. Structures cranial to the linea terminalis are biopsied via a trans peritoneal approach. Structures caudal to the linea terminalis are accessible through the abdominal wall or sciatic foramen. In selected cases specialized routes may be needed which can give good results (see also section on CT-guided drainage).
Biopsies
Musculoskeletal Biopsies Biopsy of Bone Lesions
In recent years, percutaneous biopsy of bone lesions has become a routine procedure in centers specializing in their treatment. For guidance, CT as well as fluoroscopy or ultrasound can be used. CT-guidance is especially valuable in deep-seated lesions and lesions that lie adjacent to vital structures. Biopsy is planned in close communication with the referring specialist (especially for malignant lesions). Bone scintigraphy is used to select the most accessible site for biopsy. Osteolytic regions are preferable for biopsy and thinslice CT is indispensable before biopsy to guide the appropriate access, as it can better differentiate osteolytic from sclerotic regions within the lesion. It is also used for selecting the proper needle type. Additional MRI may have a supplementary advantage in differentiating vital tumor regions from cysts or necrosis as well as gutding the needle pass in highly vascular lesions. All bone biopsies must be performed under strict aseptic conditions due to the high risk of infection. The periosteum should be thoroughly anesthetized. Sedation and general anesthesia should not be used for biopsies in proximity to neural structures. Bone biopsy can be performed as aspiration or trephine (core) biopsy. In general, trephine biopsy is more efficient than the alternative, both for neoplasms and infections, especially tuberculosis. However, aspiration can be useful in lytic lesions. A multitude of special needles are available for bone biopsy. Short trephine needles (e.g., Yamshidi, Tanzer) have large calibers of 8-11 G and are useful for superficial lesions or the penetration of cortical shells. They have the disadvantage that reinsertion for multiple biopsies is required. Longer trocar systems (e.g., Ackerman, Craig, Laredo-Bard) are placed coaxially and allow for multiple passes in deeper-seated lesions. Another coaxial system is the Bonopty needle which has an eccentric drill and is useful for penetrating intact cortex in peripheral bones. It can be used in combination with softtissue needles for lytic lesions. Pneumatic motor-drill systems (e.g., Ostycut) may be useful in locations where the cortex is thick or in sclerotic lesions. Recommended needle types are summarized in Table 6.3.
171
6 CT-Cuided Interventional Procedures
a
b
Fig. 6.7
c
Sample approaches for vertebral biopsies in the cervical spine (a), thoracic spine (b), and lumbar spine (c).
Table 6.3
Recommended
needle types (after Laredo,
1999) Bone tumors
Superficial bone lessions
Yamshidi, Tanzer
Deep bone lesions
Ackerman, Craig, Laredo-Bard
Sclerotic
Ostycut. Bonopty
Lytic
Tru-Cut core biopsy needle
iytic
+
bony shell
Yamshidi
+
Tru-Cut
Lesions in the thoracolumbar spine are approached via a posterolateral route with the patient in a lateral or prone position (Fig. 6.7). Lesions of the posterolateral body or pedicles are best reached via a transpedicular route, while lesions in the more anterior part require a steeper lateromedial angle (40°-60°) directly to the body. Needle tilt can be important to avoid damage to exiting nerve structures, and may require tilting the gantry. In the cervical spine an anterolateral or lateral approach should be chosen for lesions in the vertebral bodies. Neural arch lesions can be reached via posterior routes with tangential approaches. In peripheral bones the approach may be variable but especially in primary tumors should be chosen in close consultation with the referring surgeon, because the needle path should be resected at surgery. The path should be perpendicular to the cortex (Fig. 6.8) and aspiration of the bone marrow can be performed simultaneously. Penetration of the marrow may be painful, requiring general anesthesia. For some malignant lesions, such as osteosarcomas, open surgical biopsy may be
a
b
Fig. 6.8 Percutaneous biopsy of a lytic lesion in the talus that proved histologically to be a giant cell tumor.
preferable for better control of bleeding complications. In flat bones a tangential approach is better for sampling the maximum amount of pathologic bone and to avoid injuries to underlying structures. It is advisable to sample different sites of the lesion. Both the sampled bone tissue and any blood aspirated from the bone should be submitted to the pathologist. Close communication with the pathologist is needed for proper fixation, e.g., when electron microscopy is contemplated. If an inflammatory process is suspected, the bulk of the biopsy material should also be prepared for microbiological testing. In experienced hands complications are few and can be less than 1 %. Complications may adversely affect surgical treatment and outcome, especially in malignant bone tumors, in a significant number of patients. Therefore, proper
Musculoskeletal technique is important and these biopsies are best performed in specialized centers. In general, percutaneous biopsies are less invasive and associated with a lower morbidity and cost than surgical biopsies. The diagnostic accuracy depends on many factors, like the nature of the lesion, quality of preoperative work-up, selection of biopsy site, biopsy material, and technique and experience of the pathologist. The accuracy cited in the literature ranges from 80% to 95% for bone metastases and round cell tumors (Ewing' sarcoma, lymphoma) and 60-80% for primary bone tumors. In infection, success rates of 90% for skeletal tuberculosis and 60% for pyogenic infections have been reported. •
Biopsy of Malignant Soft Tissue Tumors
For many years, biopsy of malignant soft-tissue tumors has been the terrain of the oncologic surgeon because of the risk of compartmental contamination as well as the histopathological heterogeneity of most soft-tissue tumors. Recently, percutaneous biopsies have been more frequently used for the primary diagnosis with good results. Close interaction with the surgical oncologist is necessary and therefore these biopsies are best done in centers specializing in surgery of the tumors. CT is most commonly used for deep-seated lesions, while ultrasound is a strong competitor for large and more superficial lesions. CT has the added benefit that it can document the needle track accurately. Before biopsy, close communication with the operating surgeon is important as the needle track should be resected during subsequent surgery. General pre-biopsy principles should be followed as outlined above, and generally a preoperative gadolinium-enhanced MRI should be available to select the viable tumor regions. Biopsy can be performed as needle-core biopsy or fine needle aspiration biopsy (FNAC).Core biopsy is usually preferred in larger tumors and has the advantage of allowing specific subtyping of sarcomas. FNACis more difficult in mesenchymal tumors and is often reserved for small lesions.The preference of the surgeon or radiologist as well as the experience of the cytologist and pathologist will also influence the choice of the type of biopsy in the individual patient. Core biopsy is usually performed with 14-18 G side-cutting (e.g., Tru-Cut, Quick-Core, Gallini) or tip-cutting (Surecut, IBI) needles. To
Biopsies
prevent seeding of the biopsy track a coaxial approach may be taken through an outer sheath or Yamshidi needle, which is placed into the periphery of the tumor. Often the side-cutting needles are connected to a biopsy gun for ease of the procedure. Fine-needle biopsy employs 20-23 G side-cutting (Westcott, Burney) or tipcutting (Chiba) needles connected to a syringe. It is best to have a cytopathologic technician onsite for direct evaluation of the biopsy material. A suggested algorithm from the literature is presented in Table 6.4. Complications from percutaneous biopsies are similar to complications for other CT-guided biopsies. They occur in less than 1-4% of biopsies, and are less frequent than with open surgical biopsy. This is important since any complication may alter the timing and the type of subsequent surgical procedures, and thus may considerably influence patient outcome. It has been demonstrated that the number of clinically relevant complications in specialist centers is significantly smaller. Few studies have evaluated the accuracy of this technique. Core biopsies have an accuracy of 84-90%, while the accuracy of fine needle aspiration lies in the 90-96% range. Major reasons for inaccurate results are failures to obtain diagnostic samples and to sample parts of the tumor that are not representative for the whole lesion. The latter is more important in lowgrade malignancies and malignancies with necrotic and cystic areas.
Table 6.4 al.,1996)
Recommended
biopsy algorithm
(Logan et
Soft tissue tumors < 3 cm
20-22 G FNAC
;;. 3 cm
18 G core biopsy
Bone tumors Lytic + soft tissue mass <3cm
20-22 G FNAC
Lytic + soft tissue mass ;;.3cm
18 G core biopsy
Lytic + bony shell
Combination
Sclerotic
14 G Bone biopsy
technique
173
CT-Guided
Drainage
Procedures
CT-guided drainage procedures are indicated for pathologic fluid collections that require drainage (abscess, biloma, hematoma, etc.) and that are not amenable to or do not respond to ultrasound-guided aspiration, If a sufficient access to the lesion can be established, percutaneous drainage is preferable to the open surgical alternative owing to the significantly lower complication rate and shorter hospital stay, The puncture technique of CT-guided drainage is similar to that of CT-guided biopsy (see above),
General Principles Preparation
A brief period of hospitalization is generally required, If there is a high index of suspicion for abscess, prophylactic systemic antibiotics should be administered just before the procedure to prevent septic complications, Approach
The safest and most effective route of approach should be selected, As many non- involved organ structures as possible should be avoided, In the upper abdomen, a transhepatic approach is safe enough in most patients, or a trans peritoneal or transgastric approach may be considered, especially in patients with pancreatitis, The pleural recess should be avoided due to the risk of inciting pleurisy or pleural empyema, Injury to the spleen and intestinal structures must be scrupulously avoided, For pelvic abscesses, a transgluteal approach through the greater sciatic notch is generally employed, To avoid the sacral plexus and vessels the tract should be close to the sacrum, preferably above the sacrospinous ligament Specialized
Approaches
There are a number of new developments and approaches in CT-guided abscess drainage, Contrary to classical convention, subphrenic collections can be drained via a transpleural route if no safe transabdominal route is available, However, higher complication rates are associated with this technique, which is usually performed via a trocar puncture,
Deep pelvic collections may be difficult to approach via transgluteal or transabdominal access, Studies have demonstrated that in these cases transvaginal, transrectal, or transperineal routes can be safe alternatives for catheter placement Transrectal approaches can be especially useful in presacral lesions, Transperineal access can be obtained under CT-guidance, while transvaginal and transrectal access is guided more commonly by ultrasound, In many cases abscesses can be managed with a combination of systemic antibiotic therapyand (repeated) image-guided aspiration and lavage only, Even without placement of any indwelling catheters, similar results for pelvic abscesses, compared to hepatic ones, can be achieved, Diagnostic Aspiration
The site should be infiltrated with copious local anesthetic (10-20 ml): because often a large drainage catheter is used and its insertion can be quite painful, It is a sound general principle to precede drain insertion with a diagnostic fine-needle aspiration, as this allows sampling of fluid for gross inspection, and for bacteriological and cytological work-up, Purulent or turbid fluid is suggestive of an infection, greenish-brown fluid usually indicates biloma, and clear yellowish fluid is found in ascites or a seroma Catheters
Percutaneous drainage is performed with a pigtail or J-catheter whose gauge (French size) is matched to the viscosity of the fluid, While a 6-8 F catheter is usually sufficient for draining serous fluids (bilorna, seroma. low-viscosity abscesses), highly viscous (abscess) fluid will require a 9-12 F catheter. A suction- irrigation system (e.g. sump) may even require catheter diameters up to 16 F. Catheters for hematoma drainage are usually larger than 9 F. Necrotizing or suppurative pancreatitis often requires the use of multiple catheters and generally requires drainage by very large systems up to 28 F. The catheter should have side holes of adequate size, It is unsafe to leave side holes outside the abscess, because they facilitate dissemination of abscess, biliary and other hazardous
General Principles material to other sites in the body (especially with a trans peritoneal approach). •
Direct Puncture (Trocar Technique)
This technique employs catheter systems mounted on a long gUiding cannula with a rigid trocar. The system is introduced through a stab incision of the skin and passed directly into the collection. The trocar is then replaced by a stiff wire (e.g., Amplatz Super Stiff) and the catheter is then advanced further from the guiding cannula, which is held in a stable position to maximally enhance the guiding effect of the stiff wire. Alternatively the procedure can be performed without the additional help of a guidewire; the catheter is then directly inserted from cannula and trocar. Once the drainage catheter is in a safe position, the cannula is removed and material aspirated. Subsequently the catheter is locked and secured in place (Fig. 6.9). If no aspirate can be obtained, the system must be repositioned. The trocar technique is suitable for lesions that are easy to reach. It is fast and holds less risk of disseminating (infectious) material because the advancing catheter immediately seals the puncture hole in the fluid collection. It is
therefore ideally suited for drainage of pleural fluid or ascites and abscesses when spillage of material along the puncture tract should be avoided. •
Coaxial Technique (Seldinger Technique)
The coaxial technique is used in more difficult anatomic situations with a higher risk of puncture-related complications. The procedure is performed with a diagnostic aspiration with a fine needle, which serves as a guide to advance a sufficiently stiff guide wire into the lesion. This is often followed by insertion of a trocar/dilator combination (e.g., Acustix) to allow for guide wire exchange to standard size (0.035 inch) and the tract is subsequently dilated to increasing diameter before the final drainage catheter can be introduced (Fig. 6.10). If several wire and catheter changes are necessary, care must be taken not to dislodge the catheters, because this would make a new puncture necessary. •
Abscess Drainage
When draining an abscess, care should be taken to avoid spillage of material along the drainage tract that could contaminate neighboring organs
b Fig. 6.9 Abscess drainage by direct puncture, (a) The puncture set (drainage catheter containing a hollow needle with trocar) is inserted into the abscess, (b) The
puncture needle is withdrawn while the catheter is advanced into the abscess, (c) When the catheter has coiled into its definitive shape, the abscess is aspirated.
175
6 CT-Cuided Interventional Procedures
b Fig. 6.10 Abscess drainage by the coaxial technique. (a) The abscess is punctured with a hollow needle, (b) A guidewire is passed into the abscess and the puncture needle is removed, (c) A drainage catheter is threaded into the abscess over the guidewire. Depending on the
caliber of the puncture needle and the gauge of the drainage catheter, an intermediate step may be required in which a dilator is passed over a thin guidewire to enlarge the needle tract before inserting the thicker guidewire and drainage catheter.
or body cavities. After the drainage catheter has • Drainage Catheter Aftercare been successfully placed, the operator should attempt good evacuation of the abscess cavity. Catheters should be fixed to the skin with a Irrigation of an abscess cavity may induce tolerance slag to compensate for breathing sepsis and should always be performed under an- movements in transabdominal or transpleural tibiotic cover. Irrigation effectively liquefies positions. They should be directly fixed to the viscous material and is an important aid in re- skin if the catheter is not affected by significant moving cellular debris and other solid com- movement. Large bore drainage catheters ponents to prevent or clear catheter blockage. An should be fixed by suture. Several techniques irreversibly occluded catheter should be ex- can be performed such as multifilar Mersilene changed for a larger size drainage system or re- suture through an adhesive tape strip that is moved. Continuous irrigation systems such as the wrapped around the catheter. To avoid clogging van Sonnenberg sump catheter can work well and to facilitate drainage of viscous fluids, the using only a single access tract; however, double catheter should be flushed 3-4 times per day catheter irrigation is often more effective. with sterile normal saline. Control CT after inVarious compounds such as N-acetylcys- jection of water-soluble contrast media may be teine (ACC) and urokinase can be added to irri- beneficial to assess the drainage success. gating solutions to enhance the liquefaction efThe drainage should be maintained until fect. The dosage of urokinase can follow a the retrieved fluid is less than 10 ml/day. Ususimple, recently published protocol (Haaga ally, the catheter can be removed after 2000): abscess size <3cm: 12500IU, 3-5 cm: 7-10 days. In pancreatitis and abscesses com25000IU, 5-10 cm: 50000IU and >10cm municating with small or large bowel longer du100 000 IU followed by 10ml saline and ration (2-6 weeks) is generally required. Catheters with special suture-locking mechanisms are 15 minutes contact time, three times per day. removed under imaging guidance.
Success Rates
The results of percutaneous drainages depend on the location and nature of the lesions and on the techniques employed. Success rates may vary between less than 30% in central pancreatic abscess and more than 90% in the chest.
Specific Techniques in the Chest and Abdomen •
Drainage of Complicated Pleural Fluid Collections and Empyemas
CT-guidance for placement of thoracostomy tubes in empyemas. superinfected hematomas and other complicated fluid collections is usually reserved for those cases in which US-guided methods are not successful. This is often the case in fibrinopurulent and organizing loculated collections. Multiple loculations will need more than one drainage access and CT is optimal for planning the most appropriate approach of each one. Once again. the choice of catheter will depend on the viscosity of the fluid. Usually 8-16 F catheters are placed using a supracostal Seldinger technique or direct puncture. The entry should be in the most dependent part of the collection. If possible. posterior approaches or entry sites medial of the scapula should be avoided because of the risk of catheter displacement. Lung atelectasis should be re-expanded gradually to avoid pulmonary edema. Limited CT is the follow-up modality of choice. For catheter repositioning and exchange. however. CT alone can provide only suboptimal guidance and should be complemented by thoracic ultrasound. Guided drainage systems generally are smaller caliber than the non-guided alternative. This will in many cases result in a longer drainage duration as well as a higher susceptibility for catheter occlusion in most cases. However. the technical success rate of nonguided drainage will be improved from 50% to approximately 65-75% if catheter placement is aided by imaging guidance. The combination of drainage with intracavitary fibrinolysis (ICFT) using urokinase can further improve results (6595%). ICFT (technique see p. 176) should be instituted as early as possible in the fibrinopurulent phase and is not beneficial in effusions that are more than six weeks old.
Drainage of Malignant Pleural Effusions
Over 75% of malignant effusions are caused by carcinomas of breast. lung. and ovaries. and by lymphomas. Image-guided drainage insertion using small-bore catheters and sclerotherapy is equally successful as a nonguided large-bore access. Ultrasound is the first choice for guidance, whereas CT is usually reserved for difficult cases. Catheters (14-16 F) are placed in the sixth or seventh intercostal space and up to 1000 ml is slowly aspirated. The drainage catheters are put on 20-30 mm H20 suction and are irrigated with saline twice daily. Due to ex vacua phenomena postprocedure chest films show pneumothoraces in 30% of cases that should resolve spontaneously. After complete drainage palliative sclerotherapy is performed using talcum, antibiotics, or chemotherapeutics. Loculation may be treated first by ICFT (e.g., 250000 IU urokinase in 100ml saline). With this technique ambulatory palliation becomes possible. Drainage in Pancreatitis
Severe forms of pancreatitis are associated with life-threatening complications such as pancreatic abscesses and pancreatic necrosis with or without superinfection. In particular, infected necrosing pancreatitis is associated with a very high morbidity and mortality (up to 80%). Parenchymal necrosis can occur at any time in the course of pancreatitis and is detected on CT as areas of reduced or absent enhancement. Central organ necrosis is likely to affect the pancreatic duct integrity with the complication of fluid collections or pseudocysts refractory to fine needle aspiration. Pancreatic abscesses occur in less than 5% of acute pancreatitis cases and usually require 4-6 weeks to form. The infection is usually introduced by bacterial translocation from the colon. Unfortunately, CT is not very reliable in the differentiation of infected necrosis from abscess. Abscesses can appear as more or less well circumscribed collections in the vicinity of the pancreatic bed, the mesentery or even the psoas muscle, with evidence of gas inclusion in only 30-40% Abscess drainage requires large bore catheters sizes (12-28 F) for optimal results and multiple catheter changes are often indicated because average drainage times range between four and eight weeks. Infected necrosis is usually
178
6 CT-Cuided Interventionaf Procedures treated surgically, but in selected non-surgical candidates percutaneous drainage can be attempted using very large catheters over a long time period that are exchanged in timely fashion. Close cooperation between the interventional radiologist, surgeon, and lCU specialist is a prerequisite for the interventional management of patients with these complex conditions, The catheter placement follows the general principles outlined above, Multiple catheters
ASpecialized
may have to be placed within one collection. Post-drainage protocols should be strictly obeyed, With these techniques the success rate of abscesses and infected necrosis for drainage is 35-50%, Abscesses located centrally within the pancreatic parenchyma are associated with significantly poorer results and are better treated surgically,
Interventions
An increasing number of organ-specific interventions are now performed with the help of CT guidance, These include interventional techniques in the liver, spine, and the peripheral musculoskeletal system, as well as imageguided injection techniques and blockades for pain management This section provides a brief overview of selected procedures, For more information on this topic, please refer to the literature on interventional radiology listed in Suggested Reading atthe end of this book.
Liver Tumor Ablation Only few patients with primary or secondary hepatic neoplasms are good candidates for liver resection (10-15%), A number of minimally invasive therapies have been introduced recently that may are expected to improve and complement the otherwise poor outcome of conventional nonsurgical treatment such as radiation and chemotherapy, Radiofrequency
(RF) Ablation
RF ablation is currently one of the fastest progressing nonvascular interventional treatments in hepatic malignancy and is being investigated by many groups in Europe and the US, lt can be used exclusively or in combination with other ablative and transcatheter therapies and with hepatic resection, Patients should have less than five lesions, none measuring more than Scm, and be unfit for resection or have nonresectable disease, For primary hepatic malignancy such as HCC, the Child- Pugh classification and other functional data are also taken into account
Electric currents in the RF range (460480kHz) are used to heat tumor cells to 90-100°C via shielded needle electrodes and produce coagulative necrosis, For adequate (complete) tissue ablation, a perilesional rim of liver tissue should be included and subjected to cytotoxic temperatures (> 50°C), Vessels around the lesion may cause rapid cooling of tissue resulting in less optimum coagulation (heat sink effect), It has been demonstrated that overlapping treatment areas can increase the efficacy of thermo ablation in larger lesions, Ideally 6-14 overlapping ablations of 3cm in a lesion of 2-3 cm diameter should be used but for reasons of clinical practicality the number of overlapping areas is usually reduced to below six, Ablation of a cuff of 5-10 mm of normal liver tissue should always be attempted, Ablation needles with a variety of features are available, These include retractable prongs and internal cooling of the tip, Newer devices optimize the amount of tissue that is coagulated in a Single session, Simultaneous intratumoral saline or doxorubicin injections and an occlusion of the local (parenchymal) blood flow (Pringle maneuver) have been used to enhance the effect oftherrnoablation for lesions larger than 3-5 ern. The procedure can be performed intraoperatively as well as percutaneously under guidance ofCT or ultrasound or both, Before the procedure is scheduled, recent CT and MRI scans should be available or additional imaging may have to be obtained prior to intervention. Large-bore venous access is made available and supportive therapy with IV fluids and peri-interventional antibiosis is initiated, One or two ground pads are positioned on the
Liver Tumor Ablation patient's back or thigh and the patient is anesthetized (general anesthesia or local anesthesia including the liver capsule under conscious sedation). The vital parameters are closely monitored at all stages of the procedure. The needle is then advanced into the target area and the electrode is connected to a 100200 W RF generator. After deployment of the electrode prongs, the lesion is ablated with continuous or pulsed protocols at the desired temperature for 8-25 minutes, depending on the equipment and individual lesion size. Temperature sensors in the electrode tip control the temperature delivered to the tissue. Depending on its design, the electrode generates a sphere of tissue coagulation 3-7 cm in diameter (Fig. 6.11). If multiple overlapping ablations are required, the needle can be repositioned and the procedure repeated. Individualized strategies are employed for complex situations with subcapsular tumors and central lesions in close proximity to large vessels. Multiple ablations with only partial deployment of the prongs and the use of different designs of needle may be required to safely enhance efficacy. During and after the procedure, most patients experience pain and nausea that subsides within a week. Therefore good practice should include the best available management by the local pain team including the use of
a
b
Fig. 6.11 Radiofrequency (RF) ablation of a liver metastasis from colorectal cancer. The hypovascular lesion (a) is treated via an anterior approach using a needle with extendable prongs (b). After ablation (c),
epidural on-demand analgesia with morphine analogs. Major complications can include capsular necrosis, intraperitoneal hemorrhage, pleural (hemorrhagic) effusion, perforation of bowel or bile ducts, and cholangitis. A baseline CT to document therapy success is performed within a week after the procedure. Multiphasic technique is employed, preferably with portal venous and early delayed phases for hypovascular (metastases) and arterial and portal phases for hypervascular lesions (HCC). Follow-up with CT, MRI, or contrast -enhanced ultrasound every 3-4 months is recommended, ideally within a standardized surveillance program. As thermal ablation is a new, rapidly changing therapy, no long-term studies or studies with surgical correlation are yet available. Initial studies in HCC reported complete necrosis rates of 48-95% and a tumor -free interval of 12-24 months in 64-71% of patients. For small tumors, survival rates of approximately 60% after 3 years and about one-third after 5 years can be expected. For metastases, complete necrosis was achieved in 52-93% of lesions and tumor-free rate was 11-52% at 9-18 months follow-up. Long-term data indicate a 46% survival after 3 years. Local tumor control appears significantly better for small metastases with patient survival rates of78% for lesions below 2.5 cm diameter and 32% for lesions larger than 4.0cm.
c there is coagulation necrosis with lack of contrast uptake but a small remnant of tumor tissue in the periphery (arrow).
179
6 CT-Cuided
Interventionaf
Procedures
Laser Ablation
A second type of energy that can induce tissue coagulation in the liver is the laser. Laser ablation is commonly referred to as interstitial laser photocoagulation (ILP) or laser-induced thernnotherapy (UTI), and is most frequently used for the treatment of metastasis, Thermal ablation with laser uses neodymium: yttrium-aluminum-garnet (Nd: YAG) or solid-state laser types delivering their energy through 400 urn fibers, Different fiber geometries result in different areas of necrosis, Bare tip fibers will create spherical lesions and cooled-tip diffuser fibers will result in elliptical lesions <2cm, Multiple fibers are needed for creating larger areas of confluent necrosis, Currently coagulation lesions with a diameter of up to 6-7 cm can be created, The procedure is done percutaneously under general anesthesia or conscious sedation, CT or ultrasound are most commonly used as guidance modalities, although MRI is also well suited (LITT) because it can use thermometry sequences to monitor the coagulation process, Multiple (2-8) 18-19 G needles are placed deep in the tumor and exchanged for sterile laser fibers, The lesion is heated at 2W for 50 seconds, and the fibers are gradually withdrawn more proximally for additional ablation cycles, Total procedure time is in the order of 6O-90minutes, The procedure is generally well tolerated and no major complications have been reported, Minor complications include pain, pleural effusions, and hematoma formation, Ultrasound can be used for follow-up, but usually multiphasic contrast-enhanced CT or MRI are preferred, In metastases, complete necrosis has been reported in approximately 50% of lesions, Reported survival rates are some 75% after 2 years, and some 25%-30% after 5 years, Microwave
Ablation
Microwave energy is an alternative means to generate tissue coagulation for ablation of liver tissue, It is used most often in patients with unresectable HCC. Most published experience comes from centers in Iapan. Microwaves generate an elliptical zone of thermal coagulation around the tip of the electrode needle (maximum diameter of 2 em). The
technique is effective against fibrous tumors and metastases that are resistant to chemo-emboltzation. Candidates for this method are inoperable patients who are not candidates for chemo-embolizatlon. Patients should have four or less HCC lesions, each preferably smaller than 3 em, The procedure is usually performed percutaneously under local anesthesia or conscious sedation, Under CT-guidance, a 14 G access needle is placed at the periphery of the tumor. Subsequently a 18 G microwave needle is advanced through the guiding needle into the tumor and connected to a 2.450 MHz microwave generator. Single ablations are performed for 60-120 seconds at 60 W, which can be repeated in multiple (up to 12) sessions until total tumor ablation has been achieved, Complications are only minor and consist of pain, fever, pleural effusion, or hemorrhage, Success of therapy is evaluated with multiphasic CT or MRI follow-up, Complete tumor necrosis has been reported in over 70% of cases with survival rates of approximately 70-80% after 2 years, Percutaneous
Ethanol Injection
(PEl)
Ablation by percutaneous ethanol injection (PEl) is one of the most frequently used techniques for the therapy of primary liver malignancies in cirrhotic patients, Administration of 95% ethanol results in cell dehydration and protein denaturation and induces coagulation necrosis followed by fibrosis (Fig, 6.12), The amount of necrosis induced depends on the amount of alcohol injected, Necrotic lesions as large as 8 cm can be created with multiple sessions, PEl is used in patients with liver cirrhosis and HCC with a tumor load below 30% liver volume, It is most effective in lesions smaller than 3cm. The procedure can be performed either on an outpatient basis under local anesthesia with multiple (4-12) sessions, or on an inpatient basis under general anesthesia with a 'singleshot" approach, In each case 20-22 G needles are positioned in the tumors under ultrasound or CT guidance, A few ml of ethanol are injected in small aliquots during each session and necrosis is monitored with ultrasound or CT Usually the needle is advanced through the tumor and ethanol is injected as the needle is withdrawn
Treatment ofOsteoid Osteoma
a
c
b
Fig. 6.12 Percutaneous ethanol injection (PEl) of an inhomogenously hypervascular hepatocellular carcinoma (a). The needle is advanced after injecting small aliquots
of ethanol (b). Tumor necrosis corresponds to the hypoattenuating region after therapy (c).
under imaging guidance. Coagulation necrosis will cause a bright echogenic blush on ultrasound that makes the evaluation of regions distal to the injection site impossible. The patient will experience pain as soon as the alcohol flows back through the needle tract and reaches the liver capsule and peritoneum. This pain will limit the amount of ethanol that can be injected in a single session in a non-anesthetized patient. As a technical variant that is possible only with CT-guidance, the needle tip can be placed in the proximal portion of the tumor and then be advanced while small ethanol aliquots are injected. With this technique larger volumes of ethanol can be used because the proximal needle tract is sealed by the coagulation necrosis. This will delay the back-flow of ethanol to the liver capsule and consequently the onset of symptoms. PEl is relatively safe in experienced hands. However, patients usually present symptoms including pain and fever during and after the procedure. Major complications like abscess, pleural effusions, ascites, or hemorrhage are more frequently observed with "single-shot" than with multisession treatment. All patients require follow-up by contrast-enhanced CT or MRI. Complete tumor ablation can be achieved in 70-75% of cases. Depending on patient characteristics (Child-Pugh classification), and size and multiplicity of the lesions, the survival rates in HCC that can be expected are 60-80%
after 3 years and 30-60% after 5 years. Results for metastases are generally poorer with complete necrosis only in 50-55% of lesions and 3-year survival in the order of35-40%. Treatment of Osteoid Osteoma
Osteoid osteoma is a benign osteoblastic bone tumor with no risk of malignant transformation. It usually occurs in the weight -bearing bones before the age of 25 and is more frequent in men. Osteoid osteoma induces local pain that worsens at night and responds well to aspirin. Conventional treatment is surgical excision, which is a major intervention. Recently a number of minimal invasive techniques have been used to treat osteoid osteomas with very promising results. For successful treatment it is necessary to destroy the central nidus of the tumor. Various ablative techniques have been used, most commonly radiofrequency (RF) ablation (thermocoagulation), interstitial laser photocoagulation (ILP) , or-mechanical-percutaneous drilling. •
RF Ablation
All patients with osteoid osteoma are amenable to treatment by RF ablation. Pre-procedural imaging includes plain films, CT, and MRI for precise localization. The intervention is generally performed under general or epidural anesthesia because the
181
6 CT-Cuided Interventionai Procedures penetration or drilling of the osteoma is extremely painful. The lesion is localized with CT and the approach carefully planned to avoid all vital structures. First, a biopsy of the nidus is taken using an 18-20 G bone trephine needle with or without drill (e.g., Ackerman, Bonopty), and a hole with a diameter of 1 mm or more created for placement of the RF electrode. This is always performed under CT guidance (Fig. 6.13). In selected cases, it can be advantageous to enter the bone from the opposite cortex. The electrode is connected to the RF generator and the grounding pad is placed close to the entry site of the needle. Within 1-2 minutes the electrode is heated to a temperature of 90°C , which is maintained thereafter for 4-6 minutes, depending on lesion size. This gives tissue coagulation 5 mm around the needle tip. For lesions > "10mm, multiple electrodes may have to be placed for optimal results. Post -procedural pain should be alleviated with injection of a long-acting local anesthetic in the operation area before anesthesia is terminated. The procedure is safe and there usually are no major complications if performed under strict asepsis. Successful RF ablation of osteoid osteoma has been reported in 90-95% of cases (longterm follow-up). Rapid pain relief is seen within 24-72 hours. Recurrences can be treated with the same procedure, and late recurrences after successful therapy are rare. •
Interstitial Laser Photocoagulation (lLP)
As with RF ablation, all patients with osteoid osteoma are amenable to treatment by ILP. With the patient anesthetized, the lesion is localized with CT and the approach is planned.
Subperiosteal lesions can be targeted with a 18 G spinal needle, while cortical lesions may require 14 G bone biopsy needles with or without drilling (e.g., Bonopty, Osticut). When the tip is in the center of the nidus, a 400 u,m laser fiber is inserted. ILP is performed by using a diode laser operating at 805 nm. In continuous wave mode, a power of 2 W is applied for 200-500 seconds, depending on the maximal size of the nidus. The procedure is safe, and complications are rare. They may be associated with the biopsy procedure, but reflex sympathetic dystrophy following the procedure has also been reported. Results from studies with 2 year follow-up show success rates over 95% with rapid pain relief within 24-72 hours. Recurrences can be treated by the same procedure. •
Percutaneous Drilling
Instead of ablation by heat, percutaneous drilling can also be employed for treatment. As the motor-driven equipment is a standard ORtool for the orthopedic surgeon, the procedure can also be organized within a multidisciplinary team. The procedure is usually performed under general or epidural anesthesia and with CT guidance. After positioning, plain CT is performed to evaluate the lesion and choose the optimum approach and materials. Drilling is performed with motor-driven Kohler trephine or Auger drill sets. The latter are available with different drill sizes, but usually 3-7 mm is adequate. However, many alternative approaches have also been described. When using a Kohler system, the introducing trocar is anchored in the cortex and a K-wire
Fig.6.13 Radiofrequency (RF) ablation of an osteoid osteoma in the femur. After localization of the subperiosteal nidus (a), the needle is advanced into the lesion under CT guidance low-dose fluoroscopy and RF ablation is performed (b). Note the reduced quality of the low-dose image obtained during the intervention.
a
b
Spino'interventions is drilled through this trocar towards the center of the nidus. Subsequently, a serrated sheath is hammered in the cortex and the first trocar is removed. The bone is then drilled to the edge of the lesion and the lesion is subsequently removed with a 7 mm trephine needle. With the auger set, the procedure is simpler: after anchoring the trocar in the cortex, the entire path and nidus of the lesion is drilled out with the hollow auger drill. If smaller drills are used and the nidus is incompletely removed, the drilling path may subsequently be sclerosed with 96% ethanol. All removed tissue is submitted to pathologic analysis to demonstrate the nidus, which may be an advantage over the ablative therapies as outlined above. Patients infrequently complain of persisting pain, which can be treated by analgesics. When long paths are drilled through weight-bearing bones, the risk of fracture may be increased. Thus, RF or laser ablation are the preferred technique in these regions. Other major complications are few, but osteomyelitis is well recognized. After the procedure patients should be mobilized non-weight-bearing for six weeks and should refrain from sports for three months. Results from small studies with "1-4year follow-up show success rates of 84-100% with rapid pain relief within 24-72 hours. Treatment of recurrence can be attempted with a similar subsequent procedure. Comparable CT-guiding techniques can also be used to facilitate complex percutaneous screw fixations of pelvic fractures.
Spinal Interventions In recent years there has been a marked increase in the number of image-guided diagnostic and therapeutic interventions in the spine. Although C-arm fluoroscopy was initially used as the main modality for guidance, CT is used increasingly for more complex situations and a number of institutions have even installed combined CT-fluoroscopy interventional suites. It should be noted that performing most of these procedures require special expertise and are best done by interventional, musculoskeletal, or neuroradiologists after subspecialized training.
Injection
Therapy
for Vertebrogenic
Pain
Local spinal pain and radiculopathy are frequent conditions affecting many individuals during their lives. Percutaneous injection therapies for pain management have been used for a long time, mainly by specialized anesthesiologists without imaging guidance. Recently, imagingguidance with CT or fluoroscopy has increased in popularity and can increase precision. Based on clinical symptoms, patients may be selected for a variety of injection procedures. Indications are localized pain syndromes not responding to conservative therapy. Indications include: facet joint syndromes, radicular syndromes associated with disk herniation or spinal stenosis, inflammatory sacroiliitis, and spondylosis. The procedure is performed on an outpatient basis. Relative contraindications are hemorrhagic diathesis and skin infections. Neurological disorders causing pain may be masked by the procedure. In practice, the technique for facet joint injection is comparable with techniques for selective nerve root, epidural, sacroiliac joint, and spondylolysis injections. The procedure can be performed under local anesthesia or with conscious sedation. Following local anesthesia 20-22 G spinal needles are placed under imaging guidance. The type of guidance is operator-dependent, but usually fluoroscopy, CT, or CT-fluoroscopy are employed. The location of the needle will depend on the procedure and can be inside a facet or another joint, adjacent to the dorsal root ganglion, or along the medial branch of the dorsal root of the spinal nerve (Fig. 6.14). When the needle is placed correctly, a combination of long-acting local anesthetics (e.g., bupivacaine) and/orlong-actingsteroids (e.g., triamcinolone) are injected in small aliquots. After the procedure the patient is observed for another 2 hours in a short-stay unit or recovery room. The aim of most therapeutic injections is to provide longer-term (i.e., three months) pain relief. The published success rates of the procedures vary and strongly depend on patient populations, techniques and outcome definitions. There remains a considerable controversy about the (long-term) usefulness of interventional spinal pain relief procedures.
183
6 CT-Cuided Interventional Procedures Table 6.5 Sympatheticchain blocks and indications in oncology (Gangi, 1996)
Fig. 6.14 For spinal pain treatment, CT allows precise positioning of the needle tip either inside or close to a facet joint, adjacent to the dorsal root ganglion or along the medial branch of the dorsal root of the spinal nerve.
•
Nerve Blocks (Neurolysis)
for Pain
Management in Cancer
Tumor infiltration or compression of nerve roots or plexus may be a cause of excruciating pain in cancer patients. In a significant number of patients symptoms of causalgia (burning pain, hyperalgesia, hyperesthesia, hyperpathia) develop that can be associated with vasomotor changes. These patients often benefit from sympathetic chain blocks, the most important of which are summarized in Table 6.5. For reasons of brevity, we will only discuss the technique of celiac plexus blockade, which is technically similar to nerve blockade in other regions. Contrast -enhanced CT is preformed before the intervention to locate the celiac trunk and mesenteric artery. The celiac plexus is located
Block type
Indications
Cervicothoracic sympathetic (stellate ganglia)
tumors of the lung apex
Thoracic sympathetic
tumors of the posterior mediastinum, esophagus
Lumbar/sacral sympathetic
tumors of the uterine body and cervix
Celiac plexus
tumors of the pancreas, adrenals, stomach, liver and gallbladder
Splanchnic nerves
tumors of the lower mediastinum and upper abdomen
(Pancoast)
immediately lateral to the celiac trunk in front of the anterior wall of the aorta. Intervention may be impossible if the celiac trunk is entirely engulfed by tumor. Celiac blockade can be performed via anterior or posterior approaches. For an anterior approach the patient is in supine position and a long 22 G needle is directed to the plexus via a transhepatic route. For a posterior approach the patient is in prone position and a paravertebral route is selected (Fig. 6.15). For this access, bilateral injections are usually necessary and the gantry may have to be angled to avoid the posterior pleural sinus. Once the needle is in the correct position, a few ml of diluted iodinated contrast are injected to confirm the extravascular location of the needle tip and to anticipate the diffusion of ethanol to be injected. Then, 10ml local anes-
Fig. 6.15 Sympathetic nerve blockade. After positioning the needle tip in the region of the sympathetic plexus, contrast material is injected to check the position of the affected region. Finally ethanol! contrast mixture is injected for nerve ablation. The procedure can be done in the thoracic (a) or lumbar spine (b). a
b
Sp/'nol inrervenrions
thetic (e.g., lidocaine) is commonly injected followed by a mixture of 20-25 ml of 96% ethanol and 5ml contrast medium for neurolysis. The added contrast medium allows for CT documentation of the ethanol deposit. The procedure is usually very well tolerated and major complications are few. These can include orthostatic hypotension, intravascular injection, and pneumothorax. With this technique notable immediate pain relief can be achieved in a high percentage of patients. Reported studies show variable success rates of 40-90%, with long-term pain relief in 15-70% of patients. •
Percutaneous Laser Disk Decompression
Suboptimum results of open surgery for degenerative disk disease have led to a number of minimally invasive alternatives such as chemonucleolysis, percutaneous diskectomy, and percutaneous laser disk decompression (PLDD). In PLDD the aim is to reduce intradiskal pressure by evaporation of disk material using Nd:YAG or Ho:YAG laser energy. Patient selection is crucial and indications for laser disk decompression include: • contained disk herniation as demonstrated by CT or MRI, • disk protrusions and extrusions with positive neurological findings, and • failure of 6 weeks of conservative therapy. Contraindications include hemorrhagic diathesis, previous surgery at involved segment, spondylolisthesis, and spinal stenosis. PLDD is performed on an outpatient basis and is guided by a combination of CT and (lateral) fluoroscopy. The patient is placed prone on the table with elevation support of the abdo-
men to open up the disk spaces. The exact pathway is determined with CT to avoid nerve roots (Fig. 6.16). Under imaging guidance the skin, subcutaneous tissues, and articular process are infiltrated with a local anesthetic. It is important not to anesthetize the exiting nerve root. From a posterolateral approach an 18 G curved tip needle is subsequently introduced into the disk under (CT) fluoroscopy and its position is checked with diskography. This check can also be used to evaluate the patient's complaints. If the needle is in a correct position, the stiletto is removed and the optical laser fiber is inserted into the disk 5 mm beyond the needle tip. The laser procedure entails 15 W pulses of 0.5-1.0 s at intervals of 4-10 s. For the lumbar levels an energy of 1200-1500J is usually administered, except for L4-L5 where 1500 -2000J is more appropriate. CT is performed every 200250J to visualize and control the extent of the evaporated area. At all times patients should be able to communicate and respond to pain during the procedure. Complications of the procedure are rare. The most common is septic (spondylo)diskitis. Good to excellent results in the range of 60-76% have been published in the literature. The average follow-up was 2-3 years. However, larger prospective comparative trials are still lacking. •
Percutaneous Vertebroplasty
Osteoporosis is an increasingly important disease with more than 1.5 million fractures occurring annually in the US, especially in white women. The most frequently involved levels in the spine are T8, T12, LI and L4. The technique of vertebral vertebroplasty (PVP) was pioneered in France in the mid 1980s by Galibert and Dera-
Fig. 6.16 Percutaneous laser disk decompression (PLOD). Placement of the needle (a). Control scan during the evaporation procedure
(b). a
(PVP)
b
185
6 CT-Guided Interventional Procedures mond and was popularized throughout Europe and North America during the second half of the 1990s. Persistent. painful. or unstable vertebral collapse not responding to conservative therapies form the main indication. and therapy is geared toward pain relief. improvement of stability. and improved patient mobility. Apart from osteoporotic fractures. the technique can also be used for destructive vertebral hemangiomas. hematologic malignancies like multiple myeloma and for osteolytic metastatic disease (limited experience). Relative contraindications include significant compromise of the spinal canal. advanced posterior cortex destruction. collapse of more than 90% of the vertebral body (vertebra plana) and long-standing pain for more than one year. Patients are selected after close consultation with the referring physician. including selected screening examinations and review of all old examinations available. The procedure is nowadays most often performed on an outpatient basis. It can be performed under general anesthesia or conscious sedation with continuous monitoring of vital signs in cooperation with an anesthesiologist andlor an anesthesiology nurse. With the patient prone. CT or (biplane) fluoroscopy is used to guide needle placement in the vertebral body via a lateromedial transpedicular route. Although fluoroscopy is more frequently used. CT with or without CT-fluoroscopy may offer more control for ideal needle placement. Preferably the needle tip should be positioned near the midline in the anterior part of the vertebral body. When using CT guidance. needle placement can be performed under local anesthesia before moving the patient to the interventional suite. A variety of needles are available for this technique: generally bone biopsy needles of 11-14G are used. but more specialized needles like the Osteosite or Accuthread needles are currently available. When the needle is in the desired position, vertebrography with a small
amount of diluted iodinated contrast medium is performed to avoid needle placement in the basivertebral venous plexus. The actual vertebroplasty is performed using polymethylmethacrylate (PMMA) bone cement (e.g., Codman cranioplastic, Simplex P, Osteobond): which is mixed with manganese or tantalum for opacification. Some 6-7 ml of this paste is injected under fluoroscopic control in O.Sml aliquots using either small syringes or commercially available high pressure cement delivery systems. During impaction of the injected cement, the needle is pulled back and the injection is terminated when the cement reaches the posterior one-fourth of the vertebral body. If the cement remains unilateral, needle placement from the contralateral side is usually necessary. After the procedure the patient is closely monitored for 2 hours in recumbent and sitting positions and is re-evaluated clinically before discharge. Complication rates are low and vary from '1-10%. However, they include nerve root and spinal cord injury, epidural abscess, and pulmonary embolization ofPMMA particles via the epidural venous plexus. With rapid diffusion of the method, these rates can be higher than reported during the learning curve of the procedure. Based on the largest series in the literature, the results of PVP were good to excellent with prolonged analgesia in 90% of patients. PVP for vertebral hemangiomas has a success rate of over 90%. This rate can be further enhanced by concomitant alcohol or Ethibloc embolization of the remainder of the lesion. In metastatic lesions, experience is limited. PVP has a stabilizing and analgesic effect and should precede radiotherapy to which it is complementary. Improvement in quality of life and analgesia can be achieved in over 80% of patients. Standard vertebroplasty differs from the newer technique of balloon kyphoplasty in which a balloon is inflated in the vertebral body for restoration of height and reduction of kyphosis.
Image Analysis M. Prokop
Basic Rules Image Analysis and Scanning Strategy (p. 188) Interpretation Strategy (p. 189) Anatomy CT Numbers Measurement of CT Numbers (p. 191) Air and Gas Collections (p. 193) Fat(p.194) Mixed Tissues (p. 194) FtJKJs (p. 194) Blood (p. 195) Hemorrhage (p. 195) Abscess(p.196) Calcifications (p. 197) Hemosiderin (p. 197) Foreign Bodies (p. 197) Contrast Media (p. 198) CT Morphology
.
Size Determination (p. 199) Boundary Surfaces (p. 200) Lesion Localization (p. 201) Tubular versus Round Structure (p. 202) Displacement versus Infiltration (p. 202) Inflammation versus Tumor (p. 203) Infiltration versus Retraction (p. 203) Enhancement Characteristics Normal Enhancement Characteristics Focal Lesions (p. 206) Artifacts
. (p. 204)
and Pitfalls
Anatomy (p. 209) Previous Surgery (p. 209) Partial Volume Effects (p. 210) Window Setting (p. 211) Beam-Hardening Artifacts (p. 212) Image Noise (p. 213) Motion Artifacts (p. 214) Artifacts Due to Contrast Media (p. 217) Scanner-Related Artifacts (p. 219) Spiral and Multislice Artifacts (p. 220)
.
7 I mage Ana lysis The goal of image analysis in CT is to detect and then evaluate pathologic structures. CT morphology is rarely unequivocal, but patient history and clinical parameters usually allow for narrowing down the differential diagnosis or
making a definite diagnosis. When using CT to evaluate the extent of a disease process, it is important to know the pitfalls and artifacts that may simulate or hide real disease.
Basic Rules
The basic rule of CT image analysis is to focus istration can facilitate the differential diagnosis not on just a single section but also to view ad- of certain hepatic lesions and can differentiate jacent sections to trace all structures until they cysts from solid hypovascular lesions. They are helpful for evaluating the contrast-filled renal are identified. The recognition of suspicious findings requires a precise knowledge of collecting system and lower urinary tract. sectional anatomy. All features that cannot be However, they are only essential in patients positively identified as anatomic structures re- with a suspected urinoma or with trauma to the renal pelvis or ureter (Fig. 7.2). Often they are quire further investigation. First the CT number (attenuation value) of the suspicious area the only means of positively distinguishing a should be analyzed. Attention is then turned to poorly opacified bowel loop from a soft-tissue the morphology of the suspicious finding, and, mass. (Additional oral contrast may have to be where applicable, its enhancement characteris- administered in these cases to obtain sufficient bowel opacification.) tics are assessed. Artifacts and common pitfalls An additional scan with narrow section should be considered during each step of the width reduces partial volume effects, improves analytical process (Fig. 7.1). resolution along the z-axis, and makes it easier to appreciate fine image details. However, many lesions or anatomic structures can best be Image Analysis and Scanning Strategy appreciated during a short time period, so that Uncertainties of image interpretation may ne- delayed scans are no longer useful. In such cessitate additional scans. Ideally, a gross image cases, reconstructing images from the original analysis should be performed immediately after data set with a greater degree of overlap may the examination so that action can be taken if help to resolve uncertainties. the patient is still available for additional scans. With the increased number of examinations that is possible with modern equipment, however, this may become impossible. In fact, only in a limited number of cases benefit from additional late scans because most CT examinations use contrast enhancement, which has an optimal time window for most applications. Such late scans after contrast admin-
Morphology
Fig. 7.1
Enhancement characteristics
The foundations of CT image analysis.
a
b
Fig. 7.2 Renal trauma. Scans obtained during the parenchyma! phase of enhancement demonstrate a perirenal mass (a), which can be identified as a urinoma only on delayed scans (b). Note the nonenhancing parapelvic cysts.
Anatomy Consequently, it is more important to decide beforehand which examination technique to use, If in doubt, multislice CT offers the option to reconstruct thicker sections as a standard for primary image interpretation from a thinsection data set If questions arise, they can be resolved by reconstructing thin overlapping sections (a "secondary raw data set") and creating problem-adapted multiplanar reformations, With respect to delayed scans, one could either perform them routinely in appropriate patients or try to save radiation exposure by selectively reviewing the image data in such patients and deciding whether late scans are useful in a particular case or not Interpretation Strategy Image analysis starts with the precontrast examination, if available, Then the scans acquired after intravascular contrast administration are reviewed and compared with the precontrast images, The radiologist should give attention to "incidental findings" that could have life-threatening implications but may not be causing symptoms at the time of the examination, Prime examples are thrombi in the femoral or pelvic veins and signs of pulmonary embolism, In immunocompromised patients, the radiologist
should be alert for potential signs of infection, Asymptomatic hemorrhages are rare, Most other incidental findings do not have immediate implications, If suspicious findings are noted during image analysis, a search should be made for concomitant findings that might permit an etiologic classification, At the same time, close scrutiny should be given to sections that are apparently uninvolved, since focusing on the principal finding may cause other findings to be missed, With spiral and especially multislice CT, the axial sections can be reviewed interactively on a monitor using the cine mode or interactive multiplanar reformations, This makes it easier to identify boundary surfaces, curved structures that cross the scan plane multiple times, and small structures that are difficult to trace in single slices, An interactive image analysis is even recommended as a routine measure for certain indications (search for pulmonary metastases, hepatic studies, bowel studies, CTA) , Further image manipulation with 3D surface rendering or MIP views is reserved for the analysis of complex findings, such as in orthopedics, traumatology, and CT angiography, With the advent of multi slice CT and faster processing workstations, interactive multiplanar evaluations using volume rendering techniques may become the primary evaluation tool (see Fig, 2,27),
Anatomy CT scans of the various organ systems demonstrate typical anatomic patterns, which are described in the chapters dealing with specific body regions, An extensive review of crosssectional anatomy would exceed our scope, but anatomic atlases may be consulted for a detailed account CT scans in some body regions permit a side-to-side comparison (neck, trunk muscles, pelvic muscles, extremities) and can disclose asymmetries that, especially in the neck region, may be an important indicator of pathology, Partial volume effects appear in CT sections that cut the cranial or caudal portions of an anatomic structure or simultaneously cut structures with different CT densities (Fig, 7,3), The effects are most pronounced in the region of the thoracic
inlet, the diaphragm and adjacent structures, the upper and lower poles of the kidney, and the boweL Partial volume effects are also caused by anatomic structures that are small in relation to the selected section thickness, This can occur in bronchi, pulmonary vessels, the adrenals, and many abdominal vessels, causing a spurious change in attenuation values (Fig, 7.4), Tubular structures may appear round, oval, or elongated depending on their course relative to the plane of the CT section, Whenever an indeterminate structure is found that cannot readily be identified or assigned to a specific organ, the structure should be traced through adjacent sections until a positive anatomic identification can be made, Features that are not identified as anatomic should
189
7 Image Analysis Fig. 7.3 Partial volume effects. Changing from a thick slice collimation to a thinner setting can reduce the effect of surrounding structures on the CT number of a lesion.
00 Boundary surface
Lesion
Tubular structure
be considered pathologic, though one should bear in mind the possibility of anatomic variants, artifacts, and other potential sources of error (see Artifacts and Pitfalls, p. 209).
Fig. 7.4 Effect of slice thickness on the appearance of small lesions. With an 8mm section thickness (a), the CT numbers of nodules decrease from the center to the periphery. In addition, the maximum CT number of small nodules is reduced. Only larger nodules retain soft-tissue attenuation in their center. With a 2 mm slice thickness (b), almost all the lesions show similar attenuation values and well-defined margins.
CTNumbers
CT numbers (CT attenuation) are of limited use for tissue discrimination. Air, fat, protein-poor fluids, and fresh hemorrhage are the only substances in the body that display characteristics attenuation values (Fig. 7.5). Calcified structures are clearly distinguishable from noncalcified soft tissues, but areas that enhance intensely after iodinated contrast administration display similar attenuation values. Metallic foreign bodies or implants have extremely high CT numbers that in some cases may greatly exceed the maximum values that can be displayed by most scanners (approximately 3000 HU). Lower CT numbers may be seen with surgical clips and other small structures that are subject to partial volume effects.
The relationship between the CT number of a structure and its physical density p is defined by the equation: p=CT + 1000 (p in rug/em', CT in HU) This equation holds true as long as the effective atomic number of the substance (i.e., its atomic composition) in the displayed voxel is similar to that of water. This applies with reasonable accuracy to fluids, soft tissues, and less to fat but not to iodinated contrast media or calcified structures. For this reason, the CT number of a soft-tissue structure is often used interchangeably with "density." A lesion may be described as hyperattenuating (hyperdense), isoattenuating (isodense), or hypoattenuating (hypodense), de-
Measurement of CT Numbers
3000
100
Hemorha
Ie----+--
Liver ----+---
,
Soft tissue --t-~ Blood ----f-Exudate-----+Transsudate -+--Water ---t---~~ Mixed tissues
50
o
-+--------lic~
-50
Fat -------t-100
Air Fig. 7.5 Scale of CT attenuation numbers. The only physiologic media with fairly unique CT numbers are air, fat, protein-poor fluids, fresh hemorrhages, and calcifications.
Measurement of CT Numbers
The quantitative measurement of CT attenuation values is necessary only in doubtful cases. Air, fat, hemorrhage, and calcifications can usually be identified visually by their characteristic attenuation. This visual interpretation can be aided by comparing the feature of interest with "internal standards" such as extracorporeal air, subcutaneous fat, or bone (Fig. 7.6). The gallbladder, urinary bladder, and CSF provide good internal standards for identifying fluids, although errors can result from sludge or contrast medium in the gallbladder, contrast medium in the urinary bladder, or spinal cord tissue in the vertebral canal. In interpreting a low-density hemorrhage or a suspected liquid lesion, it is often necessary to measure the CT numbers directly on the monitor. When CT attenuation is measured in a suspicious structure, consideration should be given to local variations caused by image noise. Thus, measurements should not be limited to one pixel but should cover a sufficiently large re~ion of interest (ROI) , preferably larger than 5 mm . If doubt exists, the measurement should be repeated in adjacent sections or in other parts of the suspicious region. The ROI should not be located in an area that is subject to motion artifacts, respiratory misregistration, or beam-hardening artifacts. When CT numbers on pre- and postcontrast scans are to be compared, identical kVp settings, reconstruction filter kernels, and section width
pending on whether its CT number is higher than, equivalent to, or lower than its surroundings. Thus, a hyperattenuating lesion appears "brighter" than the surrounding tissues, an isoattenuating lesion is not delineated, and a hypoattenuating lesion appears "darker." Note that these terms are always used in reference to surrounding structures. For example, a lesion with a CT number of 30 HU may be hyperattenuating to the fluid- filled renal pelvis but hypo attenuating to the renal parenchyma. Fig. 7.6 Hypoattenuating lesion of the right kidney. Comparison with the subcutaneous fat shows a similar CT number, indicating that the mass is an angiomyolipoma and not a cyst.
191
7 Image Analysis
a
c
b
d
Fig. 7.7 Comparison of CT numbers on various phases of contrast enhancement: arterial phase image (a) and delayed phase image (b) with standard widow setting (300/60) reveals no apparent difference in lesion en-
hancement (arrow). With a binary window setting (0/85) adjusted to the CT number of the lesion in the arterial phase (c), there is clear washout with more "black" pixels on the delayed phase scan (d).
should be chosen. If there is marked enhancement of the surrounding tissue (such as the renal parenchyma), a smaller section width may be preferred to reduce pseudoenhancement due to partial volume effects (see Figs. 7.24 and 18.16). A simple technique to compare contrast enhancement on CT scans in various phases is to reduce the window width as much as possible to obtain a binary (black and white) image, and then to chose the window level in such a way that on the first phase there are approximately the same number of black and white pixels within the lesion. If the number of black pixels increases in the corresponding section on another phase, enhancement has decreased while a larger number of white pixels indicates contrast uptake (Fig. 7.7).
relative volume within the voxel. Thus, when a ROI is selected, it must be considered that spurious densities will be measured in the periphery of a suspicious region, especially the caudal and cranial portions (Fig. 7.8). Spiral and multislice CTwith reconstruction of overlapping sections provides seamless coverage of a structure of interest. But because of the widened slice profile, the CT numbers measured in spiral
•
Determination Structures
of CT Numbers
in Small
For small structures less than 10 mm in diameter, the smoothing or edge-enhancing effect of the convolution kernel will alter the maximum CT density that can be measured in the structure. In particular, the use of high-resolution kernels can mimic the presence of calcifications in small lesions. It is best, therefore, to use a "standard kernel" for measuring CT attenuation values in small lesions. Partial volume effects can distort attenuation measurements because the CT number in a pixel is always affected by the CT numbers of all structures contained within the associated voxel. These structures contribute proportionately to the CT number according to their
Fig. 7.8 Partial volume effects can distort the attenuation values measured in the cranial or caudal portions of a lesion and in lesions smaller than the selected slice thickness.
Air and Gas Collections 193 CT are theoretically correct only if the structure is larger than the sum of the table feed (TF) and slice collimation (SC). A good rule of thumb is that the structure of interest should be one and a half times to twice as large as the slice width to ensure that accurate density measurements can be made even at eccentric sites within the section. Air and Gas Collections
Air and gas collections are best appreciated with a lung window setting. In doubtful cases this setting should also be used in extrathoracic regions (e.g., to exclude free intra-abdominal air) while the image is viewed interactively on the monitor. Window settings in which fat appears too dark are not suitable for the detection of gas (Fig. 7.9). Particular care should be taken when a liver or soft-tissue window setting is used, since gas collections may be missed or areas of very low attenuation (fat with scant connective tissue) may be misinterpreted as air. Extrapulmonary air or gas is usually easy to identify despite partial volume effects. CT numbers that are significantly below the attenuation of fat mean that air is present in the corresponding voxel. An area containing less than 10% air by volume may show fat -equivalent attenuation, however. A CT density <-150HU in soft tissue is pathognomonic for the presence of gas. Pathologic gas collections should be sought in trauma patients (cutaneous emphysema, pulmonary laceration), in cases of suspected gastrointestinal perforation or infarction, and in patients with inflammatory processes (gas-
forming bacteria). Because free air rises within a closed cavity system (pleural cavity, peritoneal cavity), air collections in the supine patient (pneumothorax, free intra-abdominal air) are most likely to be found in the anterior portions of body cavities. Air within a fluid -filled cavity (small bowel, pleural cavity, aspirated cysts, etc.) forms a horizontal air- fluid level on CT images (Fig. 7.10). Air collections near the center of a presumed liquid mass suggest the presence of internal septations. This finding is highly suspicious for an abscess when corroborative symptoms are present (Fig. 7.10). The colon contents have a characteristic CT morphology due to gas-forming bacteria in the stool, enabling the colon to be distinguished from small bowel. Diarrhea or enemas can alter the appearance of the colon by creating fluidfilled bowel loops with fluid levels.
000
00
0
0
a
b
0
0
0
0
Q
c
Fig.7.10 Free air within a liquid mass (e.g., postoperative seroma, small bowel contents) rises toward the anterior side of the mass, forming an air-fluid level (a). If the contents of the mass are viscous or have solid elements (e.g., abscess, colon contents), intralesional gas inclusions are seen (b). If the mass is loculated or has internal septa (e.g., abscess, fibrinous seropneumothorax), multiple air-fluid levels are observed (c).
Fig. 7.9 A narrow window setting may prevent the detection offree air (a), which requires the use of a wider window (b).
a
0
b
7 Image Analysis Fat Fat is another tissue that can be identified by its characteristic CT density of approximately -IOOHU (±20HU, depending on kVp settings). Body fat provides a natural negative contrast medium that delineates different organ systems from one another. As a result, the neck in most patients cannot be adequately evaluated without intravenous contrast administration. Even in the abdomen, noncontrast examination can be quite difficult in thin patients who have a paucity of body fat. Fat can generally be identified even when partial volume effects are present. Whenever CT numbers measure in the range from 0 to -!OOHU, it may be assumed that fat is present within the measured region. When fat occurs in adipose tissue cells, there will usually be at least some small areas with CT densities less than -50 HU (given sufficiently thin sections). But when other cells (liver, pancreas, muscle) acquire fat as a result of fatty infiltration, it is rare to find densities lower than 0 HLJ. A decrease in the attenuation of such an organ by 10 HU or more is considered evidence that fatty infiltration has occurred. The attenuation decrease is proportional to the fat content of the tissue: in the liver, which has a normal attenuation of about 60 HL], CT density decreases at a rate of approximately 16 HU per 10"10 fat content ([CTliver-CTfat[ X 10%). The detection of fat in a mass is an important differentiating sign. If the entire mass consists of fatty tissue, this is diagnostic of a lipoma. But a predominantly fatty lesion that also contains streak-like or solid components of softtissue density is most likely a liposarcoma. Predominantly solid tumors with fatty elements can have numerous etiologies and are discussed in the chapters dealing with specific body regions. Notable examples are intrathoracic and intra-abdominal teratomas, adrenal myelolipoma, and renal angiomyolipoma.
Mixed Tissues Mixed tissues occur ubiquitously in the human body. Their CT densities are intermediate between the densities of their individual components, and this can occasionally lead to misinterpretations.
CT attenuation ofOHU almost always signifies fluid (water) but also may be caused by mixed tissue in which fat is combined with other soft tissues (e.g., adrenal adenoma). Another typical mixed tissue is cancellous bone. which is composed of calcium salts, bony matrix, connective tissue, hematopoietic marrow, and fatty marrow. For a constant calcium salt content, the measured CT attenuation varies with the amount of fat contained in the trabecular bone. This is one of the basic problems of bone densitometry using quantitative CT This technique assumes that trabecular bone is composed of a standard soft-tissue component (with a water-equivalent CT number of OHU) and a variable amount of bone mineral. Thus, variations in the fat content of the vertebrae can cause distortion of the CT measurements. But because the calcium salt content of the bone declines with progressive osteoporosis while the fat content increases, these effects tend to be mutually reinforcing, so that it is possible to detect smaller changes in bone composition. Another mixed tissue is the lung parenchyma, which consists of aerated alveoli, capillary vessels, and interstitial connective tissue. An increase in lung attenuation is always due to a reduction in aerated lung parenchyma. This may result from deficient alveolar expansion, or a material of soft-tissue density (fluid, protein, cellular material) may have infiltrated the alveoli or interstitium. Thus, both alveolar and interstitial changes can lead to a "ground glass" pattern of increased lung attenuation on CT scans. Fluids Water, by definition, has a CT attenuation of o HLJ. In theory, the density of fluids can be determined from CT scans based on the close relationship between the CT number and physical density of a fluid. In particular, it should be possible to differentiate exudative fluids (> 10l8mgiml) from transudative fluids (
Hemorrhage
numbers higher than 25 HU suggest an exudative or hemorrhagic collection. In the range from 10 to 25 HU, no differentiation can be made. Very protein-rich fluids and hemorrhagic fluids may have a CT density higher than 35 HU and may even appear hyperattenuating to parenchymal organs (chocolate cysts in the kidney or ovary, fresh hemorrhage). When an indeterminate structure is found that is less attenuating than ordinary soft tissues, the likelihood that the structure is a liquid area increases as its CT number approaches OHU. While a CT number of approximately 0 is highly suspicious for a protein-poor fluid, this is far from conclusive a similar values may be found in soft -tissue structures that have undergone fatty infiltration. For example, adrenal adenomas have a variable fat content and typically have a CT density of approximately 0 HU. Densities in the range of about 20-40 HU are nonspecific and may be seen in protein-rich or hemorrhagic fluids, areas of fatty infiltration, or solid tumors. In some cases a cystic mass can be reliably differentiated from a solid lesion (e.g., in the kidneys or adrenals) only after intravenous contrast administration. Cysts should show no apparent alteration in CT attenuation by more than 10HU after contrast administration (see also Pseudoenhancement, Figs. 7.24 and 18.16). Pleural exudate may show enhancement during the interstitial phase because contrast material can diffuse through the damaged pleura and enter the effusion.
20
40
60
80
Blood
The CT number of blood plasma is approximately 28 HU. The CT attenuation of blood increases with the relative amount of corpuscular elements and hemoglobin (iron) present. The CT number of blood is proportional to the hematocrit (Fig. 7.11a). When the hematocrit is normal, blood has approximately the same attenuation as muscle and vessel walls. A simple sign of anemia is hypoattenuation of the cardiac ventricles relative to the cardiac septum (Fig. 7.11b). If very severe anemia is present, the walls of the aorta and great vessels may show higher attenuation than their lumina. In a liver that is free of fatty infiltration, severe anemia will increase the hypoattenuation of the hepatic vessels (and hemangiomas) relative to the parenchyma and can mimic the appearance of a hyperattenuating liver (as in the setting of hemochromatosis) . Hemorrhage
The CT attenuation of extravasated blood depends greatly on its age, location, and the presence of clots (Fig. 7.12). Fresh hemorrhage is typically associated with clot formation and the rapid reabsorption of serum elements. Both processes increase the hematocrit of the collection, leading to a corresponding increase in CT numbers (up to 80 HU). Fresh hemorrhage may have a high overall density or may contain hy-
100Het(%)
100
HE
80 60 Soft tissue
40 20 8
16
24
32 Hb (g%)
b Fig. 7.11 The CT density of blood is proportional to the hematocrit but is also influenced by kilovoltage (a). When anemia is present, the cardiac ventricles become
hypoattenuating to the cardiac septum (b). Note the density of the hemorrhagic pericardial effusion.
195
7 Image Analysis
a
b
d
c
Fig. 7.12 CT features of hemorrhage. Fresh hemorrhage with typical hyperattenuating clots (a). Hyperattenuating perisplenicnen-iorrhage after the reabsorption of serum elements (b). Sedimentation effect in a
fresh hemorrhage due to active bleeding, showing typical linear contrast accumulation at the interface (c). Acute hemorrhage with fresh extravasation of isoattenuating blood (d).
perattenuating, coagulated elements that are pathognomonic for hemorrhage. The age of an intracranial hemorrhage can be estimated with reasonable accuracy, but extracranial hemorrhages are more difficult to assess. As a hematoma ages, it becomes less attenuating due to the breakdown of blood and the absorption of hemoglobin. The hemorrhage initially appears iso- or hyperattenuating to the soft tissues and later becomes hypoattenuating. The rate at which the blood is broken down depends on numerous factors, including the site of the hemorrhage: with intrahepatic bleeding, hyperattenuating areas may persist in the liver for several weeks. An old liquefied hemorrhage is indistinguishable from a cyst.
A very fresh hemorrhage in which serum reabsorption has not yet occurred is isoattenuating to intravascular blood (Fig. 7.12d). Hematomas do not enhance after intravenous contrast administration unless active bleeding is present (Fig. 7.12c). Abscess
Abscesses have nonspecific CT densities that can range from OHU (infected cyst, seroma, or biloma) to 80 HU (infected hematoma). Pure pus has a CT density of approximately 30 HU. Abscesses may mimic old hematomas (Fig.7.13a, compare with Fig.l6.20b, c), both with respect to CT attenuation as well as enhancement in the membrane around the fluid collection (abscess
Fig. 7.13 (a) Localized, encapsulated pleural empyema with rim enhancement and a hazy increase in CT attenuation in the center of the collection (arrow) similar to an old hematoma. Some 500 ml of pus were removed by CT-guided drainage, (b) Conversely, the air inclusions in this patient with recurrent bleeding after surgical removal of a psoas hematoma were due to fibrin sponges placed at surgery. a
b
Foreign Bodies
membrane, granulation tissue due to resorption of the hematoma). Air inclusions are suggestive of gas-producing bacteria but may also be a consequence of surgery (Fig. 7.13b). Calcifications
A rule ofthumb for detecting calcifications m small round lesions is that noncalcified foci should have approximately the same CT density as horizontal vascular segments of the same diameter. Local beam hardening and other phenomena cause characteristic hypo attenuating areas near small foci of calcification, which can be detected using lung-window settings (Fig. 7.14).
A CT density higher than 100 HU in soft tissue signifies a calcium-containing structure, a metallic object, or contrast medium. Pure calcific plaques often have CT densities higher than 1000 HU when imaged in a sufficiently thin section. Lower densities may be caused by partial volume effects or diffuse calcification. Calcifications may form as a consequence of hematoma, abscess, fat necrosis, in granulomas (tuberculosis, histoplasmosis), in hydatid cysts, and less commonly in noninfectious cysts. Calcifications may also develop after tumor therapy (e.g., metastases from thyroid carcinoma after radioiodine therapy, malignant lymphomas). Calcifications have been described in a variety of tumors and can provide a useful differential diagnostic criterion. The detection of calcium in pulmonary nodules is generally a sign ofabenigne lesion. In larger lesions (» Smm), the morphology of the calcification must also be considered (see Chapter 9). The attenuation values in small lesions depend on numerous factors, most notably the slice thickness and convolution kernel. A high-resolution kernel tends to overestimate the density of small lesions due to its edge-enhancing effect.
The CT attenuation of foreign bodies depends on their physical density and effective atomic number (Table 7.1). The attenuation of substances with a high atomic number (leaded glass,
a
c
b
Fig. 7.14 Calcified granulomas. Subtle local beamhardening artifacts are typically visible with a lung window setting (a) and are even seen in association with
Hemosiderin
The presence of iron in the liver causes a proportionate increase in CT attenuation. MRI is better for the detection of iron storage, however, because concomitant fatty infiltration of the liver creates a mixed attenuation pattern in which hepatic density may be normal, or even decreased, despite an iron overload. Dual-energy CT can be an\ effective technique for reducing the effect'Toffat and increasing densitometric accuracy (see Chapter 4, p.a). With tube voltages of 80 and 140 kVp, the attenuation difference in the liver should not exceed SH. An attenuation difference greater than 10 HU suggests increased iron storage. Foreign Bodies
d
very small lesions (c). With a soft-tissue window, calcium can be detected only in larger lesions (b) while small foci may display soft-tissue attenuation (d).
197
7 Image Analysis Table 7.1
CT characteristics of various foreign bodies
Safety glass
Hyperattenuating
Ordinary glass
Very slight hyperattenuation
Polyurethane
Soft-tissue
Teflon
Very
Polyethylene
Near-fat density
Metal
Extremely hyperattenuating, streak artifacts
Wood
Air inclusions
Table 7.2
F.
Contrast enhancement
Tube voltage
density
hyperattenuating
metals) is strongly influenced by the energy of the x-ray beam (kilovoltage, filtering). Metals have CT numbers outside the range that can be displayed with a 12-bit encoding ofCT numbers (i.e., > 3000 HLJ). They also cause significant beam hardening and may cause almost complete absorption, which will lead to artifacts due to photon starvation. This consistently leads to problems of image reconstruction due to hyperdense and hypodense streak artifacts. Some scanners have 16-bit encoding available (maximum CT numbers> 65,000 HLJ), or expand the CT number scale by a factor of 10, thus permitting the direct evaluation of metallic structures (such as a defective implants). New techniques can correct for beam-hardening artifacts as well. Contrast Media
The following rule of thumb applies to the effects of iodinated contrast media: at 120 kVp, the CT attenuation in a voxel increases by approximately 25 HU for each 1 mg/ml of iodine that is administered. The precise attenuation is determined by the local x-ray energy, which depends numerous factors that include tube voltage, prefiltering, and beam hardening in the
J-
140kVp 120 kVp IOOkVp SOkVp
per rnq/crn"
iodine
; Enhancement* 20 HU 25 HU 32 HU 40 HU
numbers vary between scanners
patient (Table 7.2). As the kilovoltage and iodine concentration are increased, less contrast enhancement occurs within the scanned region. Low kVP scanning can be used to increase the CT numbers of contrast-enhanced structures substantially, thus improving signal-to-noise and creating a potential for dose savings (Fig. 7.15). There are just a few situations in which the attenuation values after contrast administration are useful for identifying the underlying pathology. During the arterial phase of enhancement, for example, when the attenuation in the aorta exceeds the attenuation in the veins, areas that are isoattenuating to the contrast-filled arteries are most likely aneurysms, pseudoaneurysms, or active hemorrhage (e.g., Figs. 11.33and 14.17). In another example, the late phase after intravenous contrast administration is marked by urinary excretion of the contrast material. If highly concentrated contrast medium is found within the abdomen and outside the urinary tract, there must be a perforation or rupture of the renal pelvis, ureter, or urinary bladder (see Fig. 7.2). Cavernous hemangioma (e.g., in the liver) should appear isoattenuating to the vessels in the late phase of enhancement due to the pooling of blood within the lesion ("blood pool effect", see Figs. 11.9and 11.15).
Fig.7.15 Lower kVp settings improve contrast enhancement. Despite identical scan and contrast injection parameters average CT numbers are some 90 HU higher at 100 kVp (a) than at 140 kVp (b). Effective patient dose with 100 kVp could be lowered substantially (E < 2.5 mSv). Emboli (arrows) are harder to distinguish from partial volume effects (arrowhead) on 140kVp scans.
a
b
Size Determination CT Morphology
Morphologic criteria are hel pful in differentiating and localizing pathologic processes and evaluating their spread. Frequently, however, they are based on relatively imprecise subjective assessments, and in some cases a confident morphologic evaluation cannot be made. Size
Determination
Many factors influence the accuracy of dimensional measurements in CT. Accuracy in the image plane is determined by image sharpness, i.e., it depends on the pixel size (image matrix and field of view) and the convolution kernel and ranges from 0.6 to 1.5mm. The accuracy of measurements along the z-axis is influenced by the scan parameters, most notably the slice thickness. Spiral CT can provide an accuracy range of 2-3 mm along the z-axis, and submillimeter accuracy can be achieved when thin slices are used with spiral or multislice CT. For volume measurements, a constant breath-holding depth is essential in order not to over- or underestimate the size of a lesion. With regard to image display, care should be taken to select a window setting of sufficient width. The structure to be measured and its immediate surroundings should be displayed in gray levels and should not contain any black or
white areas. The caliper of the digital measuring tool should not be placed at the outmost margin of a lesion but in the center of the slope at the lesion border (Fig. 7.16). The goal is to measure the "full width at half maximum" of this lesion, which accurately represents the true diameter as long as the lesion is large enough. Alternatively, it has been shown that a relatively narrow window width, combined with a window level that is adapted to the contrast of the structure of interest, can yield very accurate measurements. For small lesions, the size of the lesion, and its orientation relative to the scan plane has also to be taken into account. Thus, such an approach has little practical value. Digital measurements on the monitor are necessary for optimum results, but for most practical applications it is sufficient to perform measurements on the CT film, using the reference scale on the film to calibrate the measurements. The size of very small structures (e.g., small pulmonary nodules) can only be estimated unless the region of interest is magnified by selecting a smaller FOV. Focal lesions, especially tumors, are measured by drawing two perpendicular lines that mark the greatest diameters of the lesion (Fig. 7.16).The longitudinal extent ofa large lesion can be estimated by determining the table positions for the first and last sections on which the
Line profile
o FWHM 8.0 mIT' -500
-1000
F?;;:::O""'''T"''I'''......,........-r-T
o
10
20
.•.•....••...••••.•
30
mm
a
b
Fig. 7.16 The size of a focal lesion can be estimated by drawing a line representing the major axis of the lesion and drawing a second, perpendicular line representing its largest minor axis (a). The size of smaller lesions can
be most accurately determined by the "full width at half maximum" of a line profile through the major and minor axis of this lesion (b).
199
7 Image Analysis
lesion can still be seen; the difference will give the approximate longitudinal extent of the lesion. Greater accuracy can be achieved by performing the measurement on mul tiplanar reformatted images. The volume V of an ellipsoid mass can be estimated from the three greatest mutually perpendicular diameters a, b, and c as follows: V=(axbxc)/2
In parenchyma! organs, the contrast of the lesion with its surroundings is a limiting factor. This is particularly a problem in the liver and pancreas, where unfavorable contrast parameters can result in significant underestimation of lesion size. A hypervascular margin may appear isoattenuating to normal liver parenchyma on portal phase scans and thus will lead to underestimation of the true lesion size. For this reason, measurements on precontrast or arterial phase scans may be more accurate. CT during arterial portography (CTAP) will generally define a hepatic lesion in its entirety. The occlusion of small portal venous branches, however, may produce a peripheral wedge-shaped perfusion defect leading to the overestimation of lesion size with CTAP. There is an increasing demand for more accurate volume measurements, e.g., for determining the volume of a tumor or a whole organ or part of an organ prior to and after therapy or transplantation. When performing such measurements manually, the organ or lesion has to be isolated form the data volume by proper editing techniques (see Chapter 2, p. 45). At present, numerous programs are being developed that aid in volume measurements and may even perform those measurements automatically (Fig. 7.17). Automated volume measurements are critical to distinguishing be-
tween benign and malignant disease, especially for the detection and follow-up of lung nodules. Boundary Surfaces
Most relevant tissue planes and boundary surfaces in the neck and trunk are oriented parallel to the longitudinal body axis. As a result, the boundary surfaces between the principal organs are well depicted by CT even when a relatively large slice thickness (7-10mm) is used. Problems with axial scanning arise at boundary surfaces that are parallel to the scan plane, such as blood vessels, bronchi, diaphragm, and the superior and inferior borders of the liver, kidneys, spleen, stomach, and bowel. Partial volume effects at these interfaces blur the boundaries of the structures, causing a gradual rather than a sharp transition in CT attenuation from the structure to its surroundings. Most uncertainties can be resolved by using a thinner sections and generating reformatted images perpendicular to the boundary surface. Multislice CT with thin sections offers substantial advantages over spiral CT in this regard because it can cover whole body regions with such a high spatial resolution (Fig. 7.18). If the contrast between the structure and its surroundings is low (e.g., a right -sided renal tumor abutting the liver parenchyma), often even thin sections do not show the distinction between a well-defined boundary (no tumor infiltration) and an indistinct boundary (invasion by tumor). Fine bony fissures, thin fascial planes, and interlobar fissures in the lung require thin sections and multiplanar reformations to be reliably detected.
Fig. 7.17 Automated size measurements are available for the vasculature (a) and for the determination of the volume of pulmonary nodules (b).
a
b
Organ of Origin
a
b
Fig. 7.18 Transfissural infiltration of this bronchogenic carcinoma is hard to evaluate on a 5 mm axial section (a) but is easy to detect on sagittal reformations from a multislice CT data set (b).
Organ of Origin
Attri bu ting a mass lesion to a specific organ is often the key to diagnosis. If the lesion is sur-
rounded by fat and has broad-based contact with just one organ, this question is easy to resolve. Difficulties arise when a lesion is found to be in contact with at least two organs. This is illustrated by the problem of differentiating empyema from a subpleural lung abscess, a subcapsular hepatic lesion from a peritoneal lesion, or a large renal tumor from a large adrenal tumor. An obtuse angle or bulge between the lesion and the organ suggests that the lesion arises from that organ, whereas an acute angle or indentation at the lesion periphery suggests that the lesion is extrinsic to the organ (Fig. 7.19). Typical routes of spread are usually oriented along pre-existing fascia-defined spaces. With a tumor that has invaded an adjacent organ, intravenous contrast administration may be necessary to heighten the attenuation difference between the lesion and adjacent organs and thus distinguish indentation from invasion. In all but a few cases, this differentiation can be successfully accomplished (Fig. 7.20).
Fig. 7.19 An acute angle between a lesion and contiguous organ suggests impression on the organ by an extrinsic lesion (a, adrenal tumor), while an obtuse angle suggests that the lesion arises from the organ itself (b, liver metastasis).
a
b
a
b
Fig. 7.20 Peritoneal metastasis from ovarian cancer that infiltrates the liver and simulates a liver lesion (a). Note, however, the obtuse angle between the mass and the liver surface (arrows) that could only be appreciated on the postcontrast scan. Bronchogenic carcinoma with retraction and thickening of the pleura (arrow) (b). The contrast-enhanced scan demonstrates that there is no invasion of the chest wall yet.
7 Image Analysis Tubular versus Round Structure
Displacement versus Infiltration
The differentiation of a round or oval structure (lesion, lymph node) from a tubular structure (vessel, bronchus, esophagus, bowel) generally requires an analysis of contiguous CT slices. Tubular structures can be traced through numerous sections, whereas round or oval structures are depicted in only a few sections. The cine mode is very useful in these cases, as it enables contiguous or overlapping sections to be viewed in a movie-like sequence. Another useful criterion is the attenuation in the periphery of a lesion, which is particularly helpful for identification of lymph nodes and pulmonary nodules. Tubular structures that are cut at right angles by the sectional plane show a sharp decline in contrast at the periphery of the structure. In the case of true round structures, this occurs only with a relatively large lesion whose central portion has been cut by the scan. Sections near the cranial or caudal end of the structure show a peripheral fading of contrast due to partial volume effects. This also applies to lesions that are smaller than the section width. Thus, for example, a pulmonary nodule is more easily distinguished from blood vessels on thick sections than on thin sections that pass through its center (Fig. 7.21).
Expansile lesions tend to displace or compress adjacent anatomic structures (blood vessels, parenchymal structures, bronchi, bowel). Typically this is manifested by an acute angle between the edge of the mass lesion and the border of the displaced organ. The organ capsule is intact, so the abutting mass usually forms a lens-shaped indentation. The boundary surfaces should be clearly depicted on perpendicular sections (axial or MPR). With an infiltmtive lesion, the contours of the lesion and infiltrated organ are blurred due to partial volume effects occurring along the infiltrative margin of the lesion. Again, an adequate evaluation requires images perpendicular to the boundary surface and obtaining sufficient contrast between the lesion and its surroundings. Unenhanced CT can resolve the question of fatty infiltration with an organ-transgressing tumor, but scans after contrast administration are generally required to evaluate the infiltration of other adjacent soft-tissues. Most organs are separated from one another by fa t planes. Expansile lesions cause thinning of these planes but generally do not disrupt them. Obliteration of the intervening fat plane suggests infiltration by an organ-transgressing tumor or inflammatory process. This sign is highly sensitive but only moderately specific. MRI with out -of-phase imaging is often more accurate for detecting remnants of a fat plane. Vascular invasion is an important prognostic criterion for many tumors. Because of their high intraluminal pressure, arteries present a circular cross section on imaging (unless this is prevented by atheromatous plaque). The only definitive signs of vascular invasion are a contour abnormality or the tumor-induced stenosis or occlusion of an artery. Contact with a tumor in itself is not a sufficient sign of arterial invasion, although the likelihood of invasion increases with the area of contact. If tumor covers more than 25% (one-fourth) of the circumference of an artery, it is likely that the tumor has invaded the vessel. With a more than 50% wide contact area, tumor invasion is almost always present. Venous cross sections depend on the fullness of the vein, its size, and its contact with surrounding structures. Because a venous wall is
Fig. 7.21 Decreasing attenuation due to partial volume effects in the periphery of a small nodule can help distinguish it from a tubular anatomic structure. The small nodule is difficult to detect with a slice 2 mm thick (a), but in one 8 mm thick it is clearly demarcated from surrounding vessels (b).
Infiltration versus Retraction thinner than an arterial wall, there is a greater likelihood of invasion when tumor contact is present. In particular, hepatic veins showing even a small area of tumor contact have almost certainly been invaded, or cannot be surgically separated from the tumor. Stenoses and occlusions are reliable indicators of venous infiltration. While it is rare for tumors to invade arterial vessels (this usually occurs with highly malignant tumors such as sarcomas and angiosarcomas), venous invasion by tumors is far more common. This is often associated with the formation of appositional thrombi, which can be distinguished from the actual tumor thrombus by their lack of contrast enhancement (no blood flow). The presence of a renal cell carcinoma, hepatocellular carcinoma, or sarcoma should elicit a search for venous invasion. If venous vessels are found to pass undisturbed through a suspicious mass with no displacement or constriction, it may be assumed that the mass represents a diffuse, nontumorous process. A typical example is focal fatty infiltration of the liver, which does not distort the architecture of the organ. Inflammation versus Tumor
Inflammatory processes and tumors may be differentiated in borderline cases only by taking into account clinical parameters of infection (bearing in mind that inflammatory signs may be less in immunosuppressed patients). Problems arise in the differentiation of:
• • • • •
focal lesion rounded shape mass effect lymph node> 2 mm in diameter absence of clinical signs of infection active protein =CRP, leukocytosis) • lack of response to antibiotics
(C-re-
Areas of liquefaction may occur in tumors as well as infectious processes. Generally tumors are distinguished by a larger soft-tissue component, a thicker wall, and less regular margins. In doubtful cases, an effort should be made to establish a tissue diagnosis by fine-needle aspiration or core biopsy.
Infiltration versus Retraction
The margins of a lesion can provide clues to its etiology. Smooth margins are typical of expansile lesions and most metastases. Spiculated margins may indicate an infiltrative lesion or may signify retraction effects due to scarring (Fig. 7.22). Infiltration may have a tumorous or nontumorous etiology. Tumorous infiltration is most commonly seen with primary organ tumors (bronchogenic carcinoma, hepatocellular carcinoma, pancreatic carcinoma) but is occasionally caused by heavily vascularized metastases (Kaposi sarcoma, angiosarcoma). N ontumorous infiltration may be caused by infectious processes with a cellulitic component.
• alveolar cell carcinoma from pneumonia (compare Fig. 9.31 to 9.39 and 9.40) • recurrent tumor from chronic infection following surgery (see Fig. 15.19) • chronic pancreatitis from pancreatitis in pancreatic-head carcinoma • inflammatory bowel disease from intestinal lymphoma • gastric cancer (Tl) or early lymphoma from gastritis • diverticulitis versus colon cancer (compare Fig.15. 21 toI5.15c). While there are no generally valid rules, the following clinical and radiologic features are suggestive of a tumor:
Fig. 7.22 Spicules radiating from a lesion may be the result of an infiltrative process or scarring: bronchial carcinoma (a), scar tissue in silicosis (b).
203
Image
Analysis
Retraction effects are typically caused by scarring as a result of trauma, surgery, radiotherapy, infection, or successful tumor therapy, lt can be difficult to differentiate the various causes of spiculated margins in any given lesion, but the following guidelines should be helpful, Radiating strands of scar tissue generally have well-defined margins and are less tightly clustered than the spicules of an infiltrative lesion, Perifocal edema is rare, Scars may distort the parenchymal architecture, which seldom occurs with infiltrative processes,
Within soft tissues (neck, mediastinum, abdomen), ill-defined spiculated margins on CT scans suggest an infiltrative process (inflammation or high-grade malignancy, especially nonHodgkin lymphoma), Again, well-defined spicules are more consistent with retractile processes due to scarring, Perifocal edema can obscure fat planes and is often indistinguishable from infiltration; its presence signifies an active process (tumor, infection, recent surgery, lymph stasis, venous stasis),
Enhancement Characteristics Normal Enhancement Characteristics Pulmonary Vessels
Opacified blood and nonopacified blood are generally well mixed within the right ventricle, As a result, the maximum enhancement of the pulmonary vessels is identical to peak aortic enhancement Maximum opacification depends strongly on the cardiac output (dilution effect) and the injection parameters (contrast concentration, flow rate), Enhancement of the pulmonary arteries is some 5 s earlier than the enhancement of the pulmonary veins, If there is a temporary enhancement of the aorta before contrast arrives in the pulmonary vessels, a persistent oval foramen can be suspected, Such a foramen opens temporarily as there is a rise in right atrial pressure due to contrast material injection and allows blood to enter the systemic circulation first Systemic Vessels
The aoJta and systemic etteties are defined with exquisite clarity during the arterial phase of enhancement The enhancement of the aorta (and its branching arteries) reaches a relative plateau rapidly after contrast material arrives at a particular level in the aorta, As contrast material is injected the level of this "plateau" increases further (see Fig, 24.3) until shortly after the injection (contrast material and saline flush) stops, The coverage of a large imaging volume with high spatial resolution during this phase requires the use of multi slice scanning, About 20 to 30 seconds after the onset of arterial enhancement, contrast material is
delivered to the portal system, with inflow of contrast occurring first through the splenic vein and later through the mesenteric veins, This delay can lead to artifacts in the portal vein, where non opacified blood can mimic intraluminal thrombosis (see Figs, H.53 and 15,34), The time at which systemic veins enhance after intravenous contrast injection depends on their location, The renal veins enhance very rapidly because of the high renal blood flow, accompanied by enhancement of the intrahepatic segment of the inferior vena cava, A similar pattern is seen in the jugular veins, Venous drainage of the lower extremity and azygos system proceeds more slowly, but even in these regions nonopacified blood at confluence sites can mimic thrombosis, If the contrast material is injected at a very slow rate, the arterial and venous vessels will show similar patterns of enhancement determined solely by the total volume of contrast medium and the patient's circulating blood volume, Various enhancement phases are best separated if the contrast material is injected as a short bolus (limited volume, high flow rates), Arterial Phase
The onset of aortic enhancement after contrast injection depends largely on the circulation time and can range from less than 10 seconds to more than 60 seconds, It correlates with the heart rate only in patients who have a healthy circulatory system. For this reason, tailoring the examination to the individual patient by bolus triggering or test bolus injection is superior to a standardized examination technique,
Normal
The arterial phase denotes the period in which the enhancement of an organ is caused predominantly by contrast medium delivered to the organ by arterial perfusion. This phase continues as long as a high concentration of contrast material is delivered through the aorta. But the purely arterial phase lasts only a short time (15 to 30 seconds) before other effects. such as contrast diffusion into the interstitium. contrast delivery through the portal vessels. and renal parenchymal enhancement. are superimposed. Typical patterns of arterial-phase enhancement are observed in the spleen (serpentine hypo- and hyperattenuating regions due to the differential perfusion of the red and white pulp). the kidney ("corticomedullary phase" with intense cortical enhancement and faint medullary enhancement). the liver (minimal enhancement of normal parenchyma. since 75% of the hepatic blood supply is from the portal vein). the pancreas (strongest enhancement toward the end of the arterial phase). and the thyroid gland (intense enhancement of the entire gland). Mucosal enhancement in the bowel is strongest at the end of the arterial phase and beginning of the parenchymal phase.
Enhancement
Characteristics
portal phase approximately 60 to 90 seconds p.i. In the kidney, the renal cortex shows intense initial enhancement, and shortly thereafter contrast is delivered to the collecting ducts in the renal medulla. Thus the renal cortex shows earlier and more intense peak enhancement than the medulla. The true parenchymal phase starts when the cortex and medulla show approximately equal levels of attenuation (at about 80 to WO seconds p.i.). The normal intestinal mucosa shows the strongest enhancement at the beginning of the parenchymal phase and then loses enhancement over time. The muscles do not show significant enhancement in either the arterial or parenchymal phase. Interstitial
Phase
The interstitial phase is determined by the diffusion of contrast medium into and out of the interstitium; perfusion effects are of minor importance. Little organ enhancement persists during this phase, and the vessels in most organs appear isodense to the parenchyma. Contrast
Excretion
Contrast material is normally excreted almost entirely by the kidneys. At 3 to 5 minutes after Parenchymal Phase the injection, highly concentrated contrast meThe parenchymal phase denotes the period in dium appears in the excretory system of the uriwhich the parenchyma of an organ shows its nary tract. Because contrast medium is denser than urine, it forms a dependent layer in the urihighest level of enhancement. The time required to reach the parenchymal phase depends on the nary bladder. It is very rare for intravenously inorgan (Table 7.3). and this phase shows varying jected contrast to fill the bladder completely; degrees of overlap with the arterial phase or in- generally this requires retrograde contrast administration. terstitial phase of enhancement. The liver excretes approximately 2% of the In the pancreas. for example. maximum parenchymal enhancement is produced by arte- contrast medium into the bile. This percentage rial perfusion and occurs only about 40 to may be substantially higher in patients with im60 seconds after the start of the injection (p.t.). paired renal function. Opacification of the gallIn the liver. however. enhancement results pre- bladder occurs approximately 6 to 24 hours dominantly from the portal delivery of contrast after intravenous contrast administration. medium, so peak attenuation occurs during the
Table 7.3 , Organ
Phases
of enhancement Arterial
phase
Liver
Minimal
Spleen
Tiger-striping
Homogenous
enhancement
Homogenous
Pancreas
Slowly increasing
enhancement
Maximum
Kidneys
Corticomedullary
differentiation
Homogenous
Bowel
Increasing
mucosal
enhancement
Decreasing
after after
50-
120s
after musocal
p.i.
60 s p.i.
enhancement 80-100
at 40-60 s p.i.
enhancement
s p.
7 Image Analysis
Focal Lesions
•
Peripherally Enhancing Tumors
Focal lesions in parenchymal organs have approximately the same CT attenuation as their surroundings, but their vascular supply (perfusion) can differ markedly from the surrounding parenchyma. As a result, contrast administration can improve lesion detection and narrow the differential diagnosis. The best phase of organ perfusion for lesion detection is the phase that provides a maximum attenuation difference between the lesion and its surroundings. The pattern of contrast delivery and clearance relative to the surrounding parenchyma can be used for lesion characterization.
Tumors with a rich peripheral vascular supply are most common in the liver. Most such tumors are metastases (usually from adenocarcinoma), but cholangiocarcinoma (Fig. 7.23 b) may also occur. Rim enhancement is seen during the arterial phase, and may persist through later phases. In some lesions, however, scans during the portal phase may show only the central hypovascular portion of the tumor, resulting in underestimation of tumor size. A similar enhancement pattern may be seen outside the liver (e.g., in neck tumors) but does not have particular diagnostic significance.
•
U
Hypervascular Tumors
Hypervascular tumors show greater enhancement than the surrounding parenchyma during the arterial phase, generally appearing as a focus with transient hyperattenuation. If the scan is performed too late, many hypervascular tumors are no longer detectable. As a result, small lesions can often be detected only with (biphasic) spiral or multislice CT scanning. Examples of hypervascular tumors are renal cell carcinoma (Fig.7.23a), hepatocellular carcinoma, sarcomas, and most malignant tumors with endocrine activity. Benign hypervascular tumors include renal adenomas, hepatic adenomas, focal nodular hyperplasia, pheochromocytoma, and benign endocrine pancreatic tumors.
a
Hypovascular tumors are less perfused than the surrounding parenchyma, so they are most conspicuous in the phase of maximum organ enhancement. Some hypervascular tumors are also hypo attenuating to their surroundings in this phase (e.g., all renal tumors), provided the attenuation produced by arterial perfusion is below the level of parenchymal enhancement. Almost any tumor histology may present as a hypovascular lesion. A large percentage of all hepatic metastases are hypovascular, and hypovascular lesions are typical of pancreatic carcinoma (Fig. 7.23c). The maximum enhancement of an organ during the parenchymal phase is strongly dependent on the contrast injection parameters.
c
b
Fig. 7.23 Lesions are optimally displayed times, depending on the vascularity of the organ: pancreatic metastasis from renal cell in the arterial phase (a), cholangiocarcinoma
Hypovascular Tumors
at different tumor and carcinoma in the early
portal phase (b), pancreatic carcinoma in the parenchymal phase (c). Note the primary tumor and the periarterial invasion (arrows) as well as the atrophic tail of the pancreas with a dilated pancreatic duct (arrowheads).
Focal Lesions
Maximum attenuation in the pancreas occurs about 20 to 30 seconds earlier than in the liver, while maximum attenuation in the renal nephrogram occurs 20 to 30 seconds later than maximum liver enhancement. As a result, the timing of optimum lesion detection varies from organ to organ, and contrast administration always involves a trade-off with regard to imaging goals. •
Tumors Outside Parenchymal Organs
Tumors outside parenchymal organs, such as softtissue tumors and head and neck tumors, must be delineated from muscle tissue. Except for sarcomas, they are frequently hypovascular or slightly hypervascular. Tumor enhancement is based on arterial perfusion and increased contrast diffusion into the interstitium due to altered vascular permeability. Depending on the vascularization, tumors can best be appreciated towards the end of the arterial phase (arterial vascularization) or later, up until the interstitial phase (diffusion effects). Optimum detection of these lesions requires scanning with a long delay (60 s or more) and a narrow window setting. •
Cysts
Cysts do not enhance after contrast administration. A diagnostic criterion for cysts is the absence of a significant attenuation increase (> 10 HU) during any perfusion phase. Attenuation changes less than 10 HU are still within normal accuracy limits or may result from random fluctuations. Scans during a late hepatic phase or during the excretion phase in the kidneys often improve differentiation between tumors and cysts because even small tumors tend to
8-mm Sections
pre-contrast
Fig. 7.24
Pseudoenhancement
post contrast
of small cysts.
have taken up contrast during these phases and become hard to visualize while cysts remain distinctly hypo attenuating. Pseudoenhancement is a problem in areas where there is marked contrast uptake in normal parenchyma outside a cyst. It is most common in the kidneys and is the more pronounced the smaller the lesion. Pseudo enhancement is a consequence of partial volume effects, which change in magnitude as the extracystic parenchyma enhances (Fig. 7.24). For small cysts, a less stringent criterion of > 15 HU enhancement may be used, but it is better to reduce the section width for such an examination and even perform a late scan with thin sections to keep partial volume effects low. Accurate measurements can be expected if the cyst is larger than twice the section width. There are no enhancing (solid) elements found in the periphery of a true cyst. The presence of such elements, even when very small, should raise suspicion of a cystic tumor, or, when membrane-like, of infection. •
Inflammation and Abscesses
Inflammatory diseases of the mucosae (e.g., in the gastrointestinal tract) are characterized by intense mucosal enhancement during the arterial phase and early parenchymal phase. The inflammatory involvement of deeper wall layers leads to enhancement of the entire wall, in some cases obscuring the adjacent fatty tissue. A focal inflammation often produces no detectable changes within a parenchymal organ, but unenhanced scans may occasionally show a hypo dense area (edema), and a slightly hyperdense area may be seen after contrast adininis-
2-mm Sections
pre-contrast
post contrast
207
7 Image Analysis
(ration (arterial hyperperfusion or impaired interstitial back-diffusion). The predominant finding in fatty tissue is a diffuse or streak-like attenuation increase caused by edema and an inflammatory reaction. Abscesses may develop from a region of focal inflammation or from the infection of an existing fluid collection or hematoma. Liquefaction and abscess formation are marked by a central, nonenhancing cystic region with an attenuation of approximately 30HL] (pus) surrounded by an enhancing rim of relatively uniform thickness (Fig.7 .25a) . Usually this granulomatous rim is best appreciated during the interstitial phase of enhancement. Hepatic abscesses may be rimless or display a hypodense rim due to the intense enhancement of the surrounding parenchyma (Fig.7.25b). Because of this often invisible ring of granulation tissue, hepatic abscesses usually appear to have indistinct margins. Superiniected fluids (seroma, biloma, hematoma) have nonspecific CT densities that can range from 0 to 80 H. They do not show central enhancement, and it takes several days for a typical granulomatous rim to develop at the periphery (Fig.7 .25c). Generally this rim is narrow and is most conspicuous during the interstitial phase. A superinfected fluid collection is indistinguishable from a noninfected collection in its early stage. Diagnostic needle aspiration should be considered if laboratory findings are consistent with infection. An enhancing rim may also form in association with resorptive processes such as hem a-
a
b
toma breakdown, so this sign is highly suggestive of, though not specific for, an infectious process. •
Hemangiomas
Cavernous hemangiomas of the liver are loculated, blood-filled cavities with a delayed, centripetal pel-fusion pattern and very thin walls. This accounts for their enhancement characteristics on CT scans: On noncontrast images, hemangiomas are isodense to blood (even in anemic patients). Following contrast administration, the enhancing areas in the lesion show somewhat delayed but approximately equal attenuation to the aorta. Contrast -filled lacunae appear first at the periphery of the lesion and subsequently spread toward its center. By the portal phase at the latest, all the lacunae are isoattenuating to the blood vessels (blood pool effect). The time required for the complete opacification of a hemangioma can range from seconds to more than 30 minutes, the lesion displaying a characteristics pattern of peripheral-to-central ("centripetal") enhancement. If the blood and liver parenchyma are isoattenuating on delayed scans, the hemangioma can no longer be discerned. Note that there are a few atypical hemangiomas that may appear as a homogeneous hypervascular tumor (small lesions up to 2cm), as a heterogeneous tumor (large lesions greater than 3cm), or as a hypovascular lesion (thrombosed hemangiomas).
c
Fig. 7.25 Loculated hepatic abscess with an enhancing rim (a), hepatic abscess with a hypodense rim and air inclusion (b), superinfected biloma (c).
Artifacts and Pitfalls Anatomy
Previous Surgery
Anatomic variants or normal structures scanned on an unusual plane should not be misinterpreted as mass lesions (Fig. 7.26). Differentiation is accomplished by comparing the adjacent slices and by taking enhancement characteristics into consideration (Fig. 7.27). Table 7.4 lists some of the anatomic structures that are most commonly misidentified on CT scans.
Scars can be quite difficult to distinguish from tumors if the scar tissue forms a mass lesion. This is particularly true after proctectomy. Pulmonary metastases after therapy may resemble bronchial carcinoma. Dynamic contrast studies can sometimes differentiate tumors from scar tissue, since tumors enhance during the arterial or parenchymal phase of perfusion, while scar
b
a
Fig. 7.26 An accessory left superior vena cava mimics a mediastinal lymph node (a). The cisterna chyli mimics a retrocrural node (b). Bowel simulates lymph nodes (c),
a
b
c
d
and the fatty infiltration of a lymph femoral vein thrombosis (d).
c
Fig. 7.27 Cervical veins simulate lymph nodes (a, b). Compression (c). Perisplenic varices simulate small bowel loops (d).
node
mimics
d by a synovial cyst mimics femoral vein thrombus
7 Image Analysis Table 7.4 some anatomic identified on CT scans
structures
that may be mis-
Structure
May be mistaken for
Geniohyoid muscle
Lymph node
Scalene muscles
Lymph nodes
Sartorius muscle
Lymph node
Supra-aortic
Lymph nodes
arteries
Left superior vena cava
Lymph node
Aberrant vein
Lymph node
pulmonary
Pericardiac recess
Lymph node
Right atrial appendage
Tumor, lymph node
Atypical renal veins
Lymph nodes
Paravertebral veins
Lymph nodes
Pelvic venous plexus
Infiltrative growth
Fatty infiltration lymph node
Venous thrombosis
of
tumor
Cisterna chyli
Azygos vein, lymph node
Crus of diaphragm
Lymph node, esophagus
Accessory spleen
Lymph node, renal or adrenal tumor, peritoneal tumor
Bowel
Intra-abdominal tumor, lymph node, varices
Splanchnic varices
bowel loops
tissue does not enhance until the interstitial phase. Bowel loops after reconstructive surgery of the biliary system often collapse and may easily
be mistaken for a local mass. Oral cholegraphic contrast agent may be used to opacify such loops and allow for reliable differentiation from a real tumor (see Chapters 12 and 15). M ul tislice scanning may improve anatomic delineation as long as there is sufficient local abdominal fat between various soft tissue structures. Bladder wall edema following a transurethral resection (TUR) can mimic a residual tumor (see Fig. 19.7). "Cysts" can be seen followingTUR of the prostate. Partial Volume Effects
The influence of partial volume effects depends on the size of the structure relative to the section width and the position of the structure in relation to the scan plane. Thus, structures that are approximately parallel to the scan plane are far more subject to partial volume effects than structures perpendicular to the scan plane, i.e., structures oriented along the z-axis. In studies using relatively thick sections and nonoverlapping reconstructions, partial volume effects can reduce the contrast of small pulmonary or hepatic lesions located in an unfavorable position (between two adjacent slices), causing them to be missed. Scans that cut a portion of an adjacent structure may simulate lesions where none exist. For example, a section traversing an ossified, posteriorly protuberant attachment of the first rib to the sternum can mimic a solitary pulmonary nodule (Fig.7.28a). Similarly, a sec-
c Fig. 7.28 The sternal attachment of the first rib mimics an intrapulmonary hepatic lesion (b).
d nodule (a). The air-filled
colon mimics a
Window Setting Window Setting
a
b
Fig. 7.29 Suspected low attenuation mass in the left adrenal gland on a 5 mm section (a) that can be attributed to partial volume effects due an almost horizontal course of a normal size gland (b) if coronal MPR are reconstructed from an overlapping secondary raw data set in multislice CT (4 x 1/6).
tion cutting the upper pole of the right kidney, the gallbladder, or adjacent air-filled bowel segments can mimic an intrahepatic lesion (Fig.7.28b). The lower part of the caudate lobe may be mistaken for a lymph node at the porta hepatis. The portal venous confluence can simulate a lesion of the pancreas. Thick sections through an obliquely positioned adrenal gland may mimic a mass (Fig. 7.29). Unsharp tumor interfaces in oblique scans may prompt a spurious diagnosis of infiltration.
a
b
Fig. 7.30 Dependence of CT measurements on window width. When a narrow window is used (a), the bronchial wall thickness is overestimated compared with a wide window setting (b). The degree of calcified
CT window settings affect image contrast and noise. If the window is set too wide, image contrast is reduced, with the result that low-contrast lesions in the liver or pancreas may be missed or the enhancement of an abscess membrane, for example, may go undetected. Conversely, too narrow a window setting can significantly increase image noise and abolish grayscale differentiation in fatty tissue areas. This can hamper the evaluation of fine structural details and cause pathologic changes in fatty tissues (edema, infiltration) or free air to escape detection (see Fig. 7.9). The accurate estimation of lengths and diameters on CT scans requires a sufficiently large window width. Both the structure of interest and its surroundings should still contain gray levels and should not appear totally black or white. A typical example is the evaluation of bronchial wall thickness. This requires a window width > 1OOOHU (preferably 1500 HU). Other window settings lead to the overestimation of bronchial wall thickness (Fig.7.30a, b). CTA of calcified stenoses or stents requires wide window settings to correctly assess the degree of narrowing (Fig. 7.30c, d).
c
d
stenoses on CTA of the carotids varies with window settings. While a setting ofW/L=500/150 overestimates the stenosis (c), a setting of 1500/300 displays the lumen correctly (d).
211
7 Image Analysis
Beam-Hardening Artifacts
The radiation emitted from the x-ray tube of a CT scanner represents a spectrum of high and low energies. As the thickness of the scanned object increases, the low-energy spectral components are absorbed more than the higherenergy components. This phenomenon is called beam hardening. In defining CT numbers, a certain "intermediate" energy must be assumed for the absorption of x-rays by water. Since absorption is reduced at higher energies, increased beam hardening would cause a decline in CT numbers. The technical solution to this problem is to estimate the thickness of the object based on the x-ray absorption for each individual projection and mathematically correct for beam hardening based on the estimated thickness. One assumption in this computer correction is that the entire object is composed of just one substance (usually water). If the local atomic composition of the object differs markedly from that of water (bone, metal, concentrated iodinated contrast medium), beam-hardening artifact will continue to occur. Newer techniques are available for further reducing beam-hardening artifacts. This can be done either by using complex computer algorithms to determine beam hardening from the density distribution in the CT image (e.g., RASP, Toshiba) or by acquiring thinner slices that are then assembled to make a larger slice thickness (this can reduce beam hardening by structures that pass obliquely through the slice).
a
b
•
Ribs
Low-attenuation beam-hardening artifacts that occur behind ribs can mimic intrahepatic lesions, but they can usually be correctly identified by their characteristic location (Fig. 1.31 a). •
Extremities
If both arms are left in the scanning field in an adducted position at the patient's sides disturbing horizontal streak artifacts will appear between the bony structures due to increased xray absorption and beam hardening. For this reason, at least one arm should be removed from the scanning field and extended above the patient's head. Similar beam hardening effects can be seen between the shoulders (Fig. 7.31 b) and between the femurs. The denser the skeletal structures, the more pronounced the artifacts. •
Metal Implants
Metallic objects cause significant beam hardening ranging and may also cause almost complete x-ray absorption (photon starvation at the detector), resulting in the appearance of hyperdense and hypodense streak artifacts (Fig. 7.31c). These artifacts generally do not occur with small objects (staples, vascular clips) because of partial volume averaging. Titanium implants absorb less radiation than other metals, so they cause fewer artifacts. Beamhardening artifacts caused by bilateral total hip implants cause decreased attenuation in the lesser pelvis (bladder, obturator compart-
c
d
Fig. 7.31 Beam-hardening artifacts may occur behind ribs (a) and may be caused by the shoulder girdle (b), metal implants (c), and highly concentrated contrast medium in the urinary bladder (d).
Image Noise
merit), making it difficult to evaluate this region. CT is not useful for evaluating metal objects (e.g., checking for implant fracture) unless the range of measurable CT densities is expanded upward (> 3,000 HLJ). This is possible if the scanner already employs 16bits per pixel (maximum CT number> 65,000) or if special software is available that extends the CT number scale. Algorithms for metal artifact suppression are available from some manufacturers. Such algorithms often use iterative correction. In a first step they determine the location of dense objects based on a reconstructed CT image. They then use this information for predicting beam hardening effects and photon starvation effects, and correcting for them. This procedure can be repeated in an iterative manner, thus further improving image quality. Such algorithms, however, require substantially longer for reconstructing the final image. •
Contrast Medium in the Bladder
A contrast-filled urinary bladder typically causes areas of very low attenuation to appear between the lateral bladder wall and the bony pelvis (Fig. 7.31d). •
Gastrointestinal Contrast Media
Barium residues in the gastrointestinal tract can give rise to very intense local streak artifacts. Undiluted iodinated contrast material can produce a similar effect. For this reason, CT examination of the abdomen should be postponed if gastrointestinal contrast residues can be seen on the scanogram. Contrast residues in diverticula may persist for days or weeks, but usually are responsible only for minor artifacts.
Image noise is particularly disturbing shaded surface displays because it will "flying pixels" and irregularities of the contour (Fig. 7.32). Similar effects can countered on volume rendered images. •
on 3D cause object be en-
Electronic Noise and Photon Starvation
If radiation dose to the detector is reduced substantially, then the electronic noise of the detector system (amplifier noise) becomes a major issue. Electronic noise adds random fluctuations of the detector signal and is negligible for standard radiation exposure. The absolute amount of noise is independent of the incident radiation. As the detector signal decreases, the relative effect of electronic noise becomes more apparent. Such a decrease in detector signal can be due to high absorption in the patient (e.g., obese patients, metallic objects, large cross-section such as in the shoulders) or due to low-dose scanning. As there is a continuing trend towards lower exposure dose, electronic noise will become more important. If the number of photons that hit the detector becomes too low (photon starvation), electronic noise becomes the predominant feature (Fig. 7.33). Reconstructing thicker sections or using smoothing kernels to reconstruct the raw data set can no longer compensate for this sudden increase in image noise. For this reason, a photon starvation effect should be avoided by all means because there is no way of compensating for the reduced image quality.
Image Noise
Image noise is a normal effect on CT sections that were not obtained with an excessive amount of radiation exposure. It represents random fluctuations of the measured CT number and is mainly caused by quantum noise (see also Chapter 5). Image noise may reduce the conspicuity of low-contrast detail but can be reduced by increasing the section thickness (axial images or MPR) or using a smoothing filter kernel (see Fig. 2.5).
Fig. 7.32 Image noise will lead to flying pixels on 3D shaded surface displays (a). Using a smoothing kernel for image reconstruction can reduce this effect (b).
213
214
7 Image Analysis
100 Signal
~
,-
'" i
.~
Quantum
noise
.~ ~
001
Electronic
norse
00001+-----,----.----.----.----, 10,000 1000 10n 10 Detector dose
Fig. 7.33 As exposure dose is decreased, image noise exceeds the numbers predicted from quantum noise alone. The reason is the growing contribution of electronic noise. which becomes the predominant feature as the detector signal drops below a certain threshold
and photon starvation occurs (a). Reconstructing thicker sections from multislice CT data sets can no longer sufficiently compensate for the reduction in image quality (b, c).
Motion Artifacts
•
Motion in the scanned section during one rotation of the x-ray tube will result in non-consistent projection data because of different configurations of the scanned objects in the various x-ray projections. As a result, varying degrees of motion artifact will appear throughout the reconstructed image, being most pronounced in the region of the moving structure.
Pulsation of the heart, aorta, and pulmonary vessels may cause double or multiple contours to appear on vascular and organ boundaries. When this effect occurs in the ascending aorta, it can simulate an intimal nap (see Fig.23.13). Pulsation of the heart may cause streak artifacts in adjacent organs and may simulate dissection in the descending aorta (Fig.7.34a). Vascular pulsation may lead to displacement of the vessel as such and may cause double contours
Fig.7.34 Pulsation of the heart may cause streak artifacts in the descending aorta simulating dissection (a). Pulsation results in displacement of the renal arteries
leading to serrated contours in this patient after renal transplantation (b). Pulsation artifacts can produce kymographic effects on MPRs (c).
Pulsation
Motion Artifacts
(Fig.7.34b) or serration of the vessel wall (see Fig. 23.13). Pulsating structures in multiplanar reformations cause serration artifacts to appealalong the z-axis (kymographic effect, Fig. 7.34c). Most pulsation effects decrease with faster scanning (multislice CT, e.g., with 4x2. 5 mm detector configuration) because one heart cycle spreads out over a larger range along the z-axis. Pulsation of the heart and great vessels causes the trachea to appear serrated on 3D or multiplanar displays (Fig.7.35a). Pulmonary vascular pulsations or transmitted lung pulsations most commonly occur behind the heart in the left lower lobe, causing herringbone-like distortion of vascular bifurcations in the lung periphery. Hypodense areas appear near the bifurcation sites and can mimic focal emphysema (Fig.7.35b). Pulsation may cause local areas of increased CT density that present as "pseudocalcifications" in the lung hilum, the heart, or the liver (Fig. 7.35 c). Fast ECG gating or prospective triggering can almost completely suppress pulsation effects in the aorta or pulmonary vessels. However, pulsation is non-uniform because pulsation waves travel from the heart to the periphery. Thus, there are different points in time that are optimum for the aorta and the pulmonary vessels. •
Respiratory Artifacts
Insufficient breath holding can cause double contours to appear on organ boundaries that
move with respiration. These artifacts can obscure pathology and are rarely mistaken for true pathology, although double contours within the lung can mimic bronchiectasis or pleural disease (Fig. 7.36a). These effects are caused by vigorous respiratory motion, such as a sudden deep inhalation or a coughing fit (Fig. 1.36 b). At the upper or lower border of large ovalshaped structures such as the kidneys, spiral artifacts may occur if the patient breathes during the scan (Fig.7.36c). Such artifacts do not occur if the patient holds his or her breath properly. With spiral and multislice scanning, acceptable results can be achieved even during shallow respiration. This applies only to axial scans, however. Multiplanar reformations are highly sensitive to any kind of patient movement, and even small respiratory excursions will cause undulating surface contours (Fig. 7.37a). Organs, vessels, or tumors may appear elongated or compressed, depending whether the patient breathes in or out during the scan (Fig. 7.37b). Deeper breathing leads to discontinuities in the moving structures (Fig. 7.37 c). Patient instruction is therefore critical if any type of 3D processing or 3D measurement is to be performed, such as in CT angiography or CT volumetry. •
Other Motion-Related
Effects
Motion artifacts also result from gross movements such as those that occur in patients with acute fractures or unrest due to pain (Fig. 7.38 a). While spiral CT has greatly reduced the motion problem on axial images, conspicuous motion
a Fig. 7.35 Pulsation causes the trachea to appear serrated on multiplanar reformations (a). Pulsation artifacts produce emphysema-like hypodensities at the bi-
furcations of pulmonary calcifications (c).
vessels
(b) and can mimic hilar
215
216
7 Image Analysis Fig. 7.36 Gross respiratory excursions during the scan (e.g., due to coughing) can mimic areas of bronchiectasis (a) or produce a double margin on organ boundaries (b). At the upper or lower border of the kidney, breathing may cause spiral artifacts.
a
b
a
b
Fig. 7.37 On multi planar reformations or volume-rendered displays motion artifacts can be appreciated as undulating contours of the body surface (a) or elonga-
tion or shortening of an organ or lesion (b). Marked breathing may cause discontinuities in vascular structures (c).
Fig. 7.38 Gross movements (here due to swallowing) lead to structural discontinuities that even can simulate fractures (a). A sagittal reformation of the dens in Parkinson's disease shows pronounced motion artifacts (b) that were not visible in the axial sections.
a
b
Artifacts Due to Contrast Media
artifacts can still appear in m ul tiplanar reformations and shaded surface displays (Fig. 7.38 b).
sequent blending of opacified and nonopacified blood from the two vessels can create a spurious intraluminal filling defect, especially when there is laminar flow. Pseudothrombi can occur at numerous sites of venous confluence (Table 7.5). Generally this artifact is easy to recognize, but occasionally it can be mistaken for throm bosis (Fig. 7.39, compare also Figs. H.53c and 15.34c). In the case of a pseudothrombus, scans acquired 30 to 60 seconds later will show complete, uniform opacification of the region in question.
Artifacts Due to Contrast Media
The short scan duration with spiral and multislice CT can increase artifacts caused by differential enhancement of various vascular territories. Venous structures are most commonly affected. •
Pseudothrombus
Artifact
A pseudothrombus artifact occurs most commonly at the confluence of veins that carry blood from anatomic regions with different circulation times, such as the infrarenal vena cava (slow return from pelvic and leg veins) and the renal veins (fast return though the kidneys). The artifact occurs if scanning is performed during a phase that is too early for complete opacification of both venous systems. The con-
Table 7.5
Sites of venous confluence
Affected vein
pseudothrombus
Inferior vena cava
Renal veins
Portal vein
Superior vein
Pseudothrombi
in Pulmonary
Vessels
During early scanning (particularly with multislice CT), there may be differential enhancement between various portions of the pulmonary arteries that rarely may simulate thrombus (Fig. 7.40 a). The differential enhancement of pulmonary arteries and veins during the early phase of
Causative vein,
Superior mesenteric
artifacts
may occur (listed in order of frequency) Mechanism Inflow of opacified
mesenteric
vein
blood
Inflow of nonopacified Reflux of opacified
Splenic vein
blood
blood
Mesenteric vein
Mesenteric
Inflow of nonopacified
blood
Superior vena cava
Azygos vein
Inflow of nonopacified
blood
Common
Internal iliac vein
Inflow of nonopacified
blood
Profunda femoris vein
Inflow of opacified
Brachiocephalic
Inflow of nonopacified
iliac vein
Common femoral
vein
Superior vena cava
a
where
•
veins
b
Fig. 7.39 Pseudothrombus artifacts due to differential portal vein (c), common femoral vein (d).
vein
c venous opacification:
blood blood
d azygos vein (a), inferior vena cava (b),
217
218
7 Image Analysis
a
Fig. 7.40 Pseudothrombus in the pulmonary arteries is a rare finding and is due to differential opacification of various pulmonary artery branches (a). Non-opacified pulmonary veins may simulate thrombi but are easily
scanning is more common, which may lead to misinterpretation of a non-opacified vein as an arterial em bol us (Fig. 7.40b). Because of the shorter scan duration, this effect is seen more frequently with multislice CT. Differentiation of arteries and veins is simple using a cine display or by checking for the bronchus that is adjacent to pulmonary arteries, not veins. •
c
b
Pseudolesions due to Hepatic Veins
Nonopacified hepatic veins appear hypoattenuating to the enhanced liver parenchyma and can appear as pseudolesions in some sections (Fig.7 .40c). This is most apparent during the early portal phase (about 50-60 seconds p.i.), when the portal vein is brightly enhanced but the hepatic veins are not yet opacified. It is better to delay scanning of the liver or to choose a caudocranial scanning direction during this phase, as this will delay imaging of the major hepatic veins until they are more adequately opacified. Pseudolesions can be correctly identified by following a suspected lesion through adjacent sections and comparing it to pre contrast or later contrast -enhanced scans. An identical effect can occur in arterial-portal CT (CTAP), in which contrast medium is injected into the superior mesenteric artery and the liver parenchyma is enhanced purely by contrast that is delivered to it through the portal vein. The hepatic veins are not opacified during the early phase of CTAP (about 20-40 seconds p.i.) and may appear as pseudolesions.
distinguished based on anatomical considerations (b). Non-opacified liver veins may simulate hypovascular lesions, especially during the early portal venous phase (c).
•
High-Contrast
Artifacts
If the concentration of contrast material is too high, the signal at the detector may be so weak (photon starvation) that typical streak artifacts are produced. This effect is accompanied by beam hardening, which also produces low-attenuation artifacts. After the administration of intravenous contrast, artifacts may appear in areas where the venous inflow carries a high concentration of contrast medium, most commonly the brachiocephalic vein or superior vena cava. These artifacts can completely obscure underlying signals, mimicking the CT appearance of an aortic intimal flap or pneumothorax (Fig. 7.41; see also Fig. 24.5). This problem can be reduced by lowering the contrast concentration, or by scanning in a caudo-cranial direction
Fig. 7.41 High-contrast artifact caused by contrast medium in the subclavian vein mimics a pneumothorax.
Scanner-Related
Artifacts
while using a saline flush to wash out the injection veins (see Chapter 3, p. 99). Scanner-Related Artifacts
CT scanners can cause a variety of equipmentrelated artifacts, most of which are due to errors of application, adjustment problems, or scanner defects. •
a
Inaccurate CT Numbers
Improper calibration of the scanner can lead to significant CT number errors (Fig. 7.42a). On multislice CT, improper calibration of the detectors will cause hypo attenuating rings that rotate around the patient and cause a pattern of undulating densities (zebra stripe artifact) on MPR (Fig. 7.42b). The scanner should be calibrated at least once a day, depending on the manufacturer's recommendations. If inaccuracies are noted, the scanner will require recalibration or service. In some scanners, a low radiation dose or certain convolution kernels can lead to inaccuracies in the measured CT numbers. In particular, high-resolution kernels should not be used for CT number measurements (Fig. 7.43). The effect of kilovoltage on CT numbers should also be considered. The more the effective atomic number of the scanned region differs from that of water, the greater the kilovoltage effect. Lowering the kVp will increase the CT numbers of structures with high atomic number (e.g., calcium- or iodine-containing compounds, see also Fig. 7.15), while it will reduce the CT number of fat.
b
Fig. 7.43 A high-resolution kernel for image reconstruction will simulate the presence of calcifications in small nodules (a). Correct assessment requires the use of a standard kernel (b).
•
Tube Malfunction
Breakdown of the x-ray beam during a spiral acquisition yields specific artifacts (Fig. 7.44) that are due to a complete lack of data from some projection angles. For these angles, only electronic noise contributes to the image. In some scanners such raw data cannot be reconstructed at all, others allow for image reconstruction and thus may avoid re-exposing the patient. •
Detector Malfunction
If a single detector malfunctions during one single projection, filtered back-projection along that ray will cause a single black line to appear on the CT image (Fig. 7.45a). The malfunction of several detectors in one projection will be represented as multiple lines centered on the corresponding tube positions. These artifacts
Fig. 7.42 Incorrect scanner calibration: the calibration processes was interrupted and led to a circular band of a low attenuation that simulates perihepatic fluid (a). On multislice CT, alternating bands of low and normal attenuation may occur that present as "zebra stripe artifacts" on MPR (b).
a
b
220
7
Image Analysis
Fig.7.44 Tube malfunction with complete interruption of the x-ray beam leads to purely electronic noise at the detector. Resulting artifacts with partial detector signal (a) and no detector signal (electronic noise only) (b).
b
a
Fig.7.45 Failure ofa single detector during one projection (a). Ring artifacts caused by faulty calibration of the detectors relative to one another (b). Out-of-field arti-
occur sporadically and reflect problems with the detector electronics. If a single detector malfunctions for all the projections in one tube rotation, all the CT numbers will be underestimated, and the image will show a hypodense ring the position of which correlates with the angular position of the faulty detector. This type of artifact may occur sporadically or may signify a permanent detector defect. •
c
b
a
Ring Artifacts
Multiple ring artifacts appear when several detectors malfunction in one tube rotation or, more commonly, when detector sensitivities are poorly calibrated relative to one another (Fig. 7.45 b; see also Fig. 7.41 b). The latter problem can usually be corrected by recalibrating the scanner.
facts in an obese patient, caused by tissues located outside the scanned field of view (c).
•
Out-of-Field
Artifacts
Out -of- field artifacts can occur in very obese patients or if the selected scanned field of view is too small. Areas of the body cross section located outside the scanned field appear hyperattenuating (Fig. 7.45 c, see also 25.14c). Spiral and Multislice Artifacts •
Spiral Artifacts
Spiral artifacts may occur in spiral or multislice CT at interfaces that are slightly angulated relative to the scan plane. They are due to data inconsistencies during one rotation that cannot be corrected by interpolation. They increase with pitch factor and present as an object distortion (Fig. 7.46) or spiral indentation of the object sur-
Splrol and Multislice Artifacts
Fig. 7.46 Spiral artifacts in the dome of the acetabulum from a multislice CT scan (4 x 2.5/15) with 3 mm section width.
face on 3D renderings. They are exaggerated if additional motion occurs (see Fig. 7.35c). •
Cone Beam Artifacts
Cone beam artifacts are due to the geometry of the x-ray beam and are more pronounced the more detector rows present. Only objects in the rotational axis (center of the gantry) are "seen" by the same detector row. Objects localized in the periphery of the scan field will be detected by different detector rows as the tube rotates around the patient (see Fig. 1.22). This effect leads to marked undulations and serrations on MPR (see Fig. 2.9) because the apparent position of the object cannot be precisely localized. New multislice cone-beam interpolation techniques promise to reduce this effect.
The artifacts can be seen in almost all objects with high contrast, such as bone, vessels, and intraluminal air in the bowel. Artifacts are more pronounced with a higher pitch factor and may even cause slight streak artifacts at object interfaces (Fig. 7.47). When comparing the image quality of thin sections with a high pitch factor to thick sections with a low pitch factor, such cone beam artifacts are less with the lowpitch images. However, if images of similar thickness (section width) are reconstructed from a thin-collimation data set, artifacts are further reduced. Thus, scanning with thin collimation yields the best image quality as long as image noise does not become a limiting factor. •
Noise and Spatial Resolution
Noise and spatial resolution are not uniformly distributed throughout a set of spiral CT images. Each table position along the z-axis corresponds to a particular angular position of the x-ray tube. At this position, little raw data interpolation is required, which will yield a locally improved spatial resolution. At an opposite position in the image, maximum interpolation is necessary, which yields less image noise but a somewhat lower spatial resolution (Fig. 7.48). Image noise is increased in thin-section multislice CT images because of the narrower collimation and the reduced detector dose. This makes such a technique more vulnerable to electronic noise as well.
Fig. 7.47 Cone-beam artifacts at high-contrast interfaces in the cervical spine (arrows) are hard to appreciate on axial sections (a) but are obvious on coronal reformations (b) from a data set acquired with N x SCITF=4 x 115 and reconstructed with SWIRl =1.25/0.7.
a
b
221
222
7 Image Analysis Fig. 7.48 Image noise and spatial resolution rotate in conjunction with the angular position of the x-ray tube. Close to the position of the tube, image noise is highest and spatial resolution is best (arrow) while opposite, spatial resolution as well as noise are reduced (arrow heads) (a). Image from a 5/10/5 data set. (b) Image 5 mm further distally. Note that now the areas with high and low noise are shifted by 180°.
a
b
o
IVt-LK A. J. van der Molen, M. Prokop
Anatomy
...........•.
225
Oral Cavity (p. 225) Pharynx (p. 225) Larynx (p. 226) Lymph Nodes (p. 227) Lymphoid Tissue (p. 227) Salivary Glands (p. 230) Thyroid and Parathyroid Glands (p. 230) Spatial Approach to Anatomy and Pathology
232
Fasciae (p. 232) Approach to Radiographic Evaluation of Masses of the Deep Face and Neck (p. 232) Spaces of the Suprahyoid Neck (p. 232) Spaces of the Infrahyoid Neck (p. 240) Masses Involving Multiple Cervical Spaces (p. 241) Examination Technique Cystic Lesions Thyroglossal
242 .
245
Duct Cyst (p. 245)
Branchial Cleft Cysts (p. 245) Thymic Cyst (p. 246) Cystic Hygroma/Lymphangioma (p. 246) Dermoid and Epidermoid Cyst (p. 247) Laryngocele (p. 247) Tornwaldt's Cyst (p. 247) Benign Tumors
248
Pseudotumors and Tumor-Like Lesions (p. 248) Neurogenic Tumors (p. 248) Mesenchymal Tumors (p. 249) Juvenile Angiofibroma (p. 250) Benign Tumors of the Larynx and Hypopharynx (p. 250) Malignant Tumors
.
251
Lymph Node Staging (p. 251) Nasopharyngeal Carcinomas (p. 252) Oropharyngeal Carcinomas (p. 253) Oral Cavity Carcinomas (p. 254) Hypopharyngeal Carcinomas (p. 256) Laryngeal Carcinomas (p. 258) Continue>
Lymphoma (p. 261) Post-Transplantation
Lymphoproliferative Disorder (PTLD) (p. 262)
Malignant Mesenchymal
and IMeurogenic Tumors
Sarcoma (p. 262) Salivary Gland Pathology Cysts (p. 263) Benign Tumors (p. 264) Malignant Tumors (p. 265) Inflammatory and Autoimmune Lesions (p. 266) Granulomatous Diseases (p. 267) Thyroid and Parathyroid Gland Pathology Goiter (p. 268) Thyroid Cyst (p. 268) Benign Thyroid Tumors (p. 268) Malignant Thyroid Tumors (p. 269) Parathyroid Tumors (p. 270) Inflammatory Thyroid Lesions (p. 271) Infectious
Diseases
Cellulitis and Abscess (p. 271) Tuberculous Adenitis (Scrofula) (p. 272) Necrotizing Fasciitis (p. 272) Miscellaneous Diseases with Adenopathy (p. 272) Fungal Infections (p. 273) Vascular lesions
.
Jugular Venous Thrombosis (p. 273) Arterial Thrombosis and Dissection (p. 273) Trauma Edema and Hematoma (p. 274) Vascular Injuries (p. 274) Laryngotracheal Injuries (p. 274) Post-Treatment
Neck
.. ,
Larynx Surgery (p. 275) Neck Dissections (p. 277) Radiation Therapy (p. 277)
Pharynx CT is an established imaging modality for examination of the head and neck (Table 8.1). but is increasingly being superseded by magnetic resonance imaging, especially in cooperative patients. A simple rule of thumb is that CT is preferable for inflammatory pathology. while MRI is preferred for tumor imaging. For tumor imaging, MRI is currently more important in the higher part of the neck while helical cr is still the preferred modality in the lower part, where more moving structures are present. This rule is, of course, prone to change with new developments in technology, cost and availability, and radiologist experience. The development of short bore magnets, fast imaging sequences, phased array neck coils, and dedicated protocols have led to an increased use of MRI, while with the advent of multislice helical CT some of the indications may shift back to cr. CT has advantages and should be used as the primary modality in several circumstances: • intensive care patients: shorter examination time, better monitoring; • severely debilitated patients: shorter examination time, less motion artifact; • patients with suspected destructive bone lesions or bone erosion.
Table 8.1 neck
Major indications for CT examination of the Detection, characterization, TNM-staging, bone erosion
I \lnfeGtion
Extent of infectious processes, abscess formation, differentiation of infection from tumor
{Ongenita""diseases
Evaluation of extent, characterization
lraUJT1a
Laryngotracheal injury, vascular injury
All in all, in many patients CT and MRI are complementary imaging modalities rather than competitive ones, with MRI the primary modality for tumor imaging in the upper neck. Modern ultrasonography also has a major role in imaging the neck. This is a primary modality for evaluating the superficial parotid and tile thyroid glands, Combined with ultrasoundguided fine needle aspiration it is probably the most reliable technique for evaluation of lymph node staging, although the retropharyngeal nodes are still difficult to assess with this technique.
Anatomy Most cervical structures are arranged symmetrically with respect to the midline. This facilitates side-to-side comparison, and soft-tissue asymmetry may provide the only radiological clue to the presence of a pathologic neck mass.
The anterior tongue consists of three intrinsic muscles (genioglossus. styloglossus. hyoglossus) that are supported by the floor of the mouth. The floor of the mouth is U-shaped and lined with mucosa and is formed by the mylohyoid and the anterior belly of the digastric muscles.
Oral Cavity The oral cavity is the anterior part of the oral aerodigestive tract and is separated from the oropharynx by a ring formed by the junction of the hard and soft palates, circumvallate papillae, and anterior tonsillar pillars. It contains the anterior two-thirds of the tongue, lips, hard palate, buccal mucosa, gingiva, mandible, maxilla, and the floor of the mouth. The major types of soft tissue found here are muscle, fat, and lymphoid tissue.
225
Pharynx The air-containing structures of the pharynx are traditionally divided into three levels, which are still routinely used in the description of squamous cell carcinomas and other tumors (Fig.8.1).
•
8 Neck
middle constrictor muscles. The oropharynx contains the base of the tongue and lingual tonsils in the anterior part and the faucial tonsils laterally, which are part of the Waldeyer's ring (see Fig. 8.3 b,c; Fig.8.4a,b). Hypopharynx
Fig. 8.1 Sagittal anatomy of the pharynx. The neck is commonly subdivided in suprahyoid (SHN) and infrahyoid parts (IHN). NC= nasal cavity; NP= nasopharynx; OC=oral cavity; OP=oropharynx; H = hyoid bone; HP = hypopharynx; L= larynx.
The hypopharynx stretches from the valleculae superiorly to the lower edge of the cricopharyngeal muscle inferiorly. There are three important subsites of the hypopharynx. Piriform sinus. This is formed by invaginations between the aryepiglottic folds medially and the thyroid cartilage laterally and anteriorly (see Fig. 8.3d). Its inferior tip reaches to the level of the cricoarytenoidjoints (true vocal cords). Postcricoid area or pharyngoesophageal junction. This is the anterior wall of the hypopharynx at the level of the cricoid. This site is difficul t to evaluate with cross-sectional imaging techniques. Posterior hypopharyngeal wall. This is composed of mucosa and the constrictor muscle.
Nasopharynx
The nasopharynx is the upper part of the pharynx. It is bordered superiorly by the sphenoid sinus, inferiorly by a horizontal line through the soft palate (see Fig.8.4c), anteriorly by the nasal choanae and posteriorly by the posterior pharyngeal wall. A landmark on the lateral wall is the cartilaginous end of the eustachian tube (torus tubarius). The pharyngeal recess (fossa of Rosenmuller) is located posterior and superior to the torus, which is difficult to evaluate clinically and may harbor malignancy (see Fig.8.3a). The adenoids are situated on the posterosuperior wall (roof) and are the most superior part of the Wakleyer's ring of lymphoid tissue. Oropharynx
The oropharynx stretches from the soft palate superiorly to the valleculae inferiorly. Anteriorly, it is separated from the oral cavity by a ring formed of the junction of the hard and soft palate, circumvallate papillae and anterior tonsillar pillars, and posteriorly by the superior and
Larynx
The soft tissue structures of the larynx are arranged around a cartilaginous framework of the cricoid, thyroid, and arytenoids. The epiglottic cartilage serves as "lid" for the laryngeal "box" (Fig.8.2b) . The thyroid cartilage has protective laminae that meet at an angle anteriorly (Fig. 8.3 e). From the posterior part of the laminae, inferior horns extend downward to articulate with the cricoid and superior horns stretch toward the hyoid bone providing attachment for the thyrohyoid ligaments. The cricoid cartilage is a complete ring. Its inferior border represents the junction of the larynx with the trachea. The greatest part or lamina is located posteriorly and its superior part lies at the level of the vocal cords. The paired, pyramidal arytenoid cartilages sit on top of the cricoid lamina (Fig.8.3e) and serve to move the vocal cords during phonation. All are made of hyaline or elastic cartilage and start to OSSifyaround the age of 20. The true vocal cords span the airway from anterior to posterior and are the inferior part of the vocal cord-ventricle complex. Anteriorly,
Lymphoid Tissue
,----
Epiglottis +---
Fig. 8.2 Corona] (a) and sagittal (b) anatomy of the larynx.
Hyoid bone
Supraglottis
.-+-f,Z-If---
Cricoid cartilage
Subglottis
a
o
Vocal cords and ventricle
Trachea
--tlt-----
b
they converge into the anterior commissure. The false cords are slightly more superior and are the inferior continuation of the aryepiglottic folds. The ventricles are clefts that also run anteroposterior and that curve upwards to a level above the false vocal cords (Fig. 8.2, Fig. 8.4b). The leaf-shaped epiglottis is the flexible anterosuperior part of the larynx and is separated from the hypo pharyngeal piriform sinuses by the aryepiglottic folds (Fig.8.3d). It is made of fibrocartilage and seldom shows any significant calcification. Its function is to protect the airway during swallowing. It is connected to the oropharynx by the anterosuperior pharyngoepiglottic folds and is also attached to the tongue by the glossoepiglottic fold in the midline. Anterior to this "lid" is the preepiglottic fat behind the anterior thyrohyoid membrane. The larynx is routinely subdivided into three major subsites. • The supraglottis extends from the tip of the epiglottis to the laryngeal ventricles. It contains the epiglottis, pre-epiglottic fat, aryepiglottic folds, false cords, arytenoids, and ventricles. • The glottis contains only the true vocal cords and anterior-posterior commissure. • The subglottis extends from the undersurface of the true vocal cords to the inferior edge of the cricoid. It contains only the mucosa lining the cricoid ring. Lymph Nodes
Approximately 300 of the 800 lymph nodes of the body are located in the neck. Although there
are many classifications, the most widely used are the ones of the Union Internationale Contre Ie Cancer/American Joint Committee on Cancer (LJICC/ AJCC) and the American Academy of Otolaryngology, Head and Neck Surgery. Very recently, an imaging-based variant based on these has been introduced-see Chapter 18. In the present chapter, the original regional classification of the American Academy of Otolaryngology, Head Neck Surgery (1991) is used. In this classification nodal groups, like the facial, OCCipital, retropharyngeal, and parotid nodes, are not included and should always be mentioned separately. The major nodal groups are classified into six levels: Level I: Level Level Level Level Level
Submental and submandibular groups II: Upper jugular group-skull base to hyoid II I: Middlej ugular group-hyoid to omohyoid muscle IV: Lower jugular group-omohyoid muscle to clavicle V: Posterior triangle group VI: Parapharyngeal and paratracheal groups
Lymphoid Tissue
Lymphoid tissue is primarily located in Waldeyer's ring, which is composed of the palatine and lingual tonsils, faucial tonsils, and the adenoid. The adenoid lies high in the nasopharynx and may also extend into the lateral fossae (Rosenmuller). The adenoid normally involutes
227
8 Neck
Maxilla Maxillary sinus -----Pharynqealorifice eustachian' tube
~---
_
[Medial [pterygoid process
/ ~-zygomaticbone
of ~
/"
~Temporali.slmuscle
Massetenrnuscle ~
<,
/
Eustachian tube ~
~
~ »<:>:
Lateral [pterygoid [process
/
Lateral pterygoid imuscle ~ Pharyngealrrecess lnternal.carotid.artery
~
------
[Parotid gland
------------
Styloid process
------""--+~'--7/~
Glossopharynqeajnerve
h~~~(
r-------
Mastoid process --------( Rectus capitis anteriorunuscle Sigmoid sinus----------./ Erector spinae muscles
Iinternal jugular vein (bulb)
'---------
~
Hypoqlossal nerve
'------------.------
---------=~~~~~~7r'
~===--
Maxilla --------------------..
Clivus
'----------Splenius
Perioralmuscles
~-----------
[Hard palate
~---------
Medial pterygoid muscle ------,
[Parotid duct
Iinferior alveolar nerve. lingual nerve
Masseter I muscle . ---------~
---------
Cranial nerves IX and X -----~
-----..---------
_
artery I
(posterior 'belly)
----\:-1'-+-.11-
[Mandible
Styloid process and styloid muscles
---______.
Internal jugular vein -----
Digastric imuscle
'capitis muscle
-----r-----------
Palatine tonsil, superior -------, pharyngeal, constrictor imuscle
Vertebral
Mandibularcondyle
--------
------_J
Vagus nerve -.--------~-/
Retromandibularvein
lLongus capitis unuscle
--
[Parotid gland
0f~--'2I-~-+f-----
[Internal carotid artery and cranialnerve Xll
'-----
Lonqus.capitis muscle
Longissimus, capitis muscle -.------\,.,\ Obliguus capitis i inferior .and :superior unuscle
Sternocleidomastoid
Deep cervica Ivein ----_-_ -_-_-_ -_-_-_-_ -_-_-_-•••-.,. -§-~-§, -~ -§.\~~\~~~~~~~~ Erector spinae muscles
Depressor anguli 'Orisimuscle -------
_
'---------
Atias(lateral
'--------
..---------------------------..-------
muscle mass)
[Dens of axis
[Mandible (body) Genioqlossusrmuscle [Mylohyoid muscle [Hyoglossus muscle
Masseter muscle -------------.... Submandibular gland -- ••••• ---------...,. Digastric Imuscle I (posterior [belly) ----I, Auriculotemporal nerve (branch) --~ and retromandibularvein Internal carotid artery --------/·---'!1L--F
4---'-."""'-+-1------
Internal jugularvein Levator scapulaemuscle
Trapezius I muscle
•••••
----------\'\\\'
r---f----
Spleniuscervicis
Sternocleidomastoid
muscle muscle
'------------
Vertebral artery
'------------
Axis (body)
7------------
Fig. 8.3 Axial cross-sectional anatomy of the neck. (a) nasopharynx; (b) oropharynx; popharynx-sinus piriformis; (e) larynx-vocal cords; (f) trachea-thyroid gland.
Spinous process
(c) floor of the mouth; (d) hy-
Lymphoid Tissue
~------
Epiglottis ----------______.
Thyrohyoid I membrane and preepiglottic space
~-----~nfrahyoid
muscles (sternohyoid and sternothyroid)
Platysma ----------______. ~------------
Piriform sinus
Sympathetic trunk ----------.... Carotid artery (bifurcation) Internal jugular vein
-:>.c-~-",y-::::::,,~
------~f----t,::/
..A-~~-\--------
Vagus nerve ----------f---7',\.,..__ Sternocleidomastoid
Inferior pharyngeal constrictor muscle Longuscolli muscle
----~.J
muscle
Anterior scalene muscle
l------
------""'''-\-""\t\
------------
Levator scapulae muscle ------\----t-',----tr-
Vertebral artery
--~;;;;;;:;;;;~~~~;t~~~~~~Z& -----------
Erector spinae muscles
facet joint
C4/C5
'--------
Body of C4 vertebra
d Infrahoid muscles (sternohyoid, -------... omohyoid, sternothyroid)
?~-------r-~-------~----~
Vocalis muscle ----------------.,L~~L/-!-'-I-IJ Cricothyroid joint ----------~~~+:.!..I---1 Common carotid artery --------,.(§ jugularvein -------------(--/--..... Inferior pharyngeall constrictor muscle Sternocleidomastoid
Anteriorlaryngeallcommissure Thyroid cartilage
X'--';p~,------'",,------
~(g'=-----""'~"""..---------Thyroid
gland
~~~~~~~~~~~~~-------~-
--t-t-----t--r--rn..-?
Esophagus --n;i-----~
Longus colli muscle
\'--1-----
Anterior scalene muscle
_
"'--------~
muscle
Vertebral artery
"----C5/C6 facet joint
Levatorscapulae muscle
----...d.~~~~~~~~s5f~~~:z "------
Erector spinae muscles
Cricoidlcartilage(lamina)
~~~-:::;:?!7/
L_--------
Body of C5 vertebra Nuchal ligament
Trapezius muscle
Sternohyoicd muscle ---------
_
SternothyroidJ muscle ----------_...... Cricoid cartilage
---------------;;~»<
Platysma .-------------
Esophagus
'x--------
Cornrnoru carotid artery
Thyroid cartilage, inferior cornu ---~~~...:, Thyroid glan~ Intern alj ugu Iarv ein -------f/;(-f---r:;;;-w-: Vagus nerve ------------I'!ffl--+--"7'---1 Vertebral artery
Sternocleidomastoid
~~~~~~;-T~~~~
Inferior pharyngeall constrictor muscle ~-----
----------r-'-r--T-:-'"ff
Longuscolli muscle
Brachial plexus Anterior scalene muscle
------/---:;?'-=-.J'-f11
Facetjoint ----------<--r----:r--Irw C7 vertebrae -------+--'~'<::_----T'"'II\
'-------
/~--
Fig.8.3d-f
muscle
Medial and posterior scalene muscles Levatorscapulae muscle
Splenius cervicis and capitis muscles
229
8 Neck
with age and in most adults no tissue is found by the age of 30. Incomplete involution may lead to subtle asymmetries that can mimic superficial tumors. The tonsils are usually symmetric soft tissue masses that may be prominent in children and young adults. Tonsils also involute with age. Asymmetry. especially when associated with lymphadenopathy, is suggestive for pathology. After infection dystrophic calcifications can be seen, especially in the faucial tonsils. After contrast injection, lymphoid tissue enhances more strongly than muscle but is hard to differentiate from other soft tissue structures. Gadolinium-enhanced MRI with fat suppression is the imaging technique of choice to image lymphoid tissue and to differentiate lymphoid hyperplasia from tumor.
deep tongue and ends in the sublingual papilla, at the level of the fatty sublingual septum. Usually this duct cannot be seen on CT. The sublingual gland is the smallest gland and lies on top of the mylohyoid muscle against the surface of the mandible in the sublingual space (SLS). Sometimes intraglandular ducts fuse into the sublingual (Bartholin's) duct that joins the submandibular duct just before the sublingual papilla. More than 750 minor salivary glands are situated within the mucosa of the oral cavity and upper aerodigestive tract. They are especially numerous in the buccal, palatal and lingual regions and have the same structure as the major salivary glands. These minor glands can host the same pathologic processes as their major counterparts. On plain or contrast-enhanced CT these glands cannot be distinguished from other surrounding soft tissues.
Salivary Glands The parotid is the largest salivary gland (Fig.8.3a, b). Its superficial lobe makes up 80% of the gland and overlies the ramus of the mandible and masseter muscle. The remaining 20% of the deep lobe extends through the stylomandibular tunnel (see Fig.8.7) adjacent to the parapharyngeal space (PPS). The gland has an equal amount of fat and acinar tissue, resulting in a hypodense aspect at non-contrast CT. With aging, the gland will undergo fatty change. The posterior facial (retromandibular) vein is the CT landmark that defines the course of the facial nerve which runs laterally to the vein. It marks the division of the deep and superficial lobes of the parotid. The parotid (Stensen's) duct (Fig.8.3b) emerges from the anterior part of the gland as a thin, soft-tissue dense band, courses over the masseter and buccal fat pad and pierces the buccinator muscle at the level of the second upper molar. The gland has 20-30 intraparotid lymph nodes which drain into the internaljugular chain of nodes. The submandibular gland is located in the floor of the mouth and curves around the free posterior edge of the mylohyoid muscle (Fig. 8.3 c, 8.4 b). In comparison to the parotid, the submandibular gland has less fat and a higher density on non-contrast CT. The submandibular (Wharton's) duct exits from the deep part of the gland, courses obliquely through the
Thyroid and Parathyroid Glands The thyroid gland is horseshoe shaped and located anterolateral to the larynx and trachea (Fig. 8.3 f). Its size varies, but on average each lobe measures approximately 6x4x2 em. A minority of patients have a central thyroid isthmus. From it the midline pyramidal lobe may arise that ascends anteriorly to the thyroid cartilage. Because of its high iodine content, the thyroid is relatively hyperdense (60-80 HU) to the surrounding soft tissues on non-contrast CT. The parathyroid glands are located posteriorly to the upper and lower poles of the thyroid lobes. Usually there are four glands, but a minority of patients has more. They are frequently not visible with CT, but when visible appear relatively hypodense to the thyroid. The superior and inferior thyroid vessels run posterior to the glands and may be used as markers on the axial CT slices when the glands are not visible.
Thyroid and Parathyroid Glands
Superior orbital fissure (with optic, trochlear" oculomotor" ophthalmic and abducent nerves)
Temporalis muscle
-f-j--------
Pterygopalatine fossa ------_ Lateral pterygoid muscle -----_ Med ial pterygoid muscle ------Parotid gland ----------\ Soft pa Iate ----------~ Tongue
-----------.---l~-rr,~~
Lingual nerve ------------,--m~~\ Mylohyoidl muscle ---------~
~~LC:~Cl3f7L---------Submandibular (S
Platysma ---------------'
Hyoid bone
Geniohyoid muscle ------------~ '--------------
Thyroid cartilage
~---~---
Medial pterygoid muscie -~Mandible(ramus)
gland
Digastric. muscle (tendon)
Y----------
Zygomatic arch Lateral pter-ygoid muscle
-------
--------
Masseter muscle
'?~~*_~:____h;H~/----------
Mylohyoid muscle --------
Soft palate
--~~-~H+-+~-------------Tongue Submandibular gland ------Omohyoid muscle ------Thyrohyoid muscle --------
~LL/----------
Hyoglossus muscle
gL-----------
Hyoid bone
~~-I-~---------Laryngeal ventricle -------------\--+''--1+Sternohyoid muscle
Sternocleidomastoid
Vestibular ligament Vocal ligament
~1_t_-I-------------
-------------/--+
muscle ---------/+
b
Temporalis muscle
--------1-~
'111iI---f-,.-.,-\------------
Lateral pterygoid muscle ------\---',,--,,~\Medial pterygoid! muscle Mandible (ramus)
Mandible (condyle)
-Ih-+--I--------
-------\-f------I---fl-H--/
----------\-\--\-\1
Uvula ------~-~~-----------_____\-~= Submandibular gland --------~-\
Hyoid bone (greater cornu) Vestibularfold
++!-rt-fofN---/-------
7Jrl11-.r~+-I--------
------------f
--------------------Common carotid artery
Soft palate
.----,'---j-----------------F---1-------
Epiglottis---===------------\--"--+1\ Laryngeal vestibule
Sphenoid sinus
-----------<Je
Vocal cord Hyoid cartilage
Omohyoid, sternohyoid and thyrohyoid muscles
\\"-----------
Fig.8.4
Coronal cross-sectional
anatomy of the neck (a) mid-tongue;
(b) tongue base; (c) pharynx.
Thyroid gland
231
8 Neck
Spatial Approach
to Anatomy
and Pathology
With the increasing use of cross-sectional imaging, the spatial approach to anatomy of the suprahyoid and infra hyoid neck became popular in the late 1980s and early 1990s and is commonly used today in the description of the extent and differentiation of neck masses,
Fasciae
For the description of the fascial spaces, a basic knowledge of the cervical fasciae is needed, There are two main cervical fasciae, the superficial and deep cervical fasciae, The superficial cervical fascia (SCF) is really a fat-filled layer of connective tissue that surrounds the neck and which also contains the platysma, superficial lymph nodes, nerves, and vessels, The deep cetvtcsl Iescis (DCF) consists of three parts: the superficial, middle, and deep layers (Fig, 8.5), The superficial layer of the deep cervical fascia (SLDCF) or the investing fascia also completely encircles the neck and runs from the skull base superiorly to sternum, clavicles, and scapulae inferiorly, It encloses the trapezius and sternocleidomastoid muscles, encircles the parotid space, forms a fascial sling for the inferior belly of the omohyoid muscle and contributes anterolaterally to the carotid sheath, The middle or buccopharyngeal (visceral) layer ofthe deep cervical fascia (MLDCF) is more variable, It also runs from skull base and is inferiorly continuous with the pericardium, The muscular layer envelops the infrahyoid strap muscles and the visceral part encloses the thyroid and parathyroid glands, larynx, and the trachea and so forms the visceral space, It is the anterior boundary of the retropharyngeal space and is also part of the carotid sheath laterally, The deep or prevertebral layer of the deep cervical fascia (DLDCF) runs from the base of the skull and merges inferiorly with the MLDCF and then continues into the mediastinum, It encircles the prevertebral and paraspinous muscles as well as the trunks of the brachial plexus, phrenic nerve, and vertebral vessels, It attaches to the transverse processes of the vertebrae and forms the anterior boundary of the prevertebral space, Laterally it contributes
to the margins of the retropharyngeal space and carotid sheath, Based on these fascial boundaries, a number of spaces can be distinguished, divided into spaces of the suprahyoid neck and infra hyoid neck, Four spaces are common to both the supra- and infra hyoid region, these are the carotid space (sheath) and the retropharyngeal, prevertebral, and posterior cervical spaces (Fig, 8.5),
Approach to Radiographic Evaluation of Masses of the Deep Face and Neck After identification ofa mass in the neck, a stepwise approach is taken: L Is the lesion a real mass or a pseudomass? 2, What is the space of origin of the mass: • Where is the epicenter of the mass? • In what directions are the surrounding structures or spaces displaced? 3. Evaluate the radiographic features of the mass, including: • Extension of the mass; • Is there erosion or invasion of the surrounding bony and cartilaginous structures? • Is there perineural extension of the mass? 4, Combine clinical and imaging features with space-specific differential diagnosis; 5, Provide the clinician with the most likely differentials with reasons (maximum three),
Spaces of the Suprahyoid Neck Parapharyngeal Space (PPS) The PPS has an inverted pyramidal form that extends from the base of the skull to the hyoid bone and lies anterior to the styloid process, Anteroinferiorly it is continuous with the submandibular space, Primarily composed of fat, it is the key space in the suprahyoid neck because PPS displacement can be used in assigning a lesion to a space of origin,
Spaces ofthe Suprahyoid Neck Fig. 8.5 Layers of the deep cervical fascia and spatial anatomy of the neck. (a) axial: suprahyoid neckoropharynx; (b) axial: infrahyoid neck-thyroid gland; (c) coronal: mid-tongue.
ACS = anterior cervical space BS= buccalspace CS= carotid space MS = masticator space PS= parotid space PPS = parapharyngeal space PMS = pharyangeal mucosal space PCS = posterior cervical space PVS= prevertal space AC= anterior compartment PC = posterior compartment RPS = retropharyngeal space SZMS = suprazygomatic partof masticator space
Sl.Ss sublinqual space SMS = submandibular space VS= visceral space
BS PMS MS RPS
BS MS ACPVS PS PPS CS
PPS PS
PCS
CS PCS
PC PVS *
PC PVS
:+:
:+:
a
ACS
VS ACS
CS
CS ACPVS PCS
PCS
PC PVS
b
I----SZMS
1=-1-----
MS-----\
--='-¥I:-----;l+--IHI------
MS PMS
I-----PPS SLS-----\~~~~ SMS-----~-~
I------SLS ,/-------
-
c
'I~ -
"*"
SMS
Superficial Layer Deep Cervical Fascia Middle Layer Fascia Deep Cervical Deep Layer Deep Cervical Fascia Dangel- Space
----_.
--
233
The contents of the PPS are:
The PPS is sandwiched between many other spaces: medially it is separated from the pharyngeal mucosal space, laterally it is adjacent to the masticator and parotid spaces and posteriorly it is bounded by the carotid sheath. Intrinsic masses from the PPS like neurinomas of the submandibular nerve (Vs) or minor salivary gland carcinomas are rare, but this space can act as an elevator shaft for tumors or infections either to the submandibular space or the skull base. Most common lesions result from a secondary extension of pathology originating in neighboring spaces (Figs. 8.6, 8.7; Table 8.2).
fat
ascending pharyngeal artery (from the external carotid artery) branches of the submandibular nerve (Vs) pharyngeal (pterygoid) venous plexus internal maxillary artery (from the external carotid artery) minor salivary gland rests salpingopharyngeus muscle.
PMS-----I-T~~ MS-----...
~---MS I-I~---RPS -I-----Ps
PS----\
PPS CS
I--J----~---I-----
CS ------./
\.!.......-+-------I----
AC
rvs
PC
rvs
.'::.L-I---.I------
Fig. 8.6 Patterns of spread of masses in suprahyoid neck spaces. Left: masticator space mass. Right: retropharyngeal space mass. Legends see Fig. 8.5
PPS--~ ~---MS
MS----------
-i-----PS PS----I
cs-----, RPS -----\--'\---(
PPS
1-1-----
~-I-----cs ~/--'/----.. ~L_/_-----
rvs PC pvs
AC
Fig. 8.7 Patterns of spread of masses in suprahyoid neck spaces. Left: parotid space mass. Right: pharyngeal mucosal space mass. Legends see Fig. 8.5
Spaces
Table 8.2 Differential space lesions
diagnosis
Pseudotumor:
Asymmetric plexus
Cystic lesions:
Atypical
Inflammatory lesions:
Malignant tumors:
parapharyngeal
pterygoid
second
venous
branchal
cleft cyst
Abscess spread from adjacent deep cervical spaces: adenoidsltonsils (PMS), odontogenic
Benign tumors:
of
(MS),
parotid
Pseudotumor:
Cystic lesions: Inflammatory lesions:
from
Mucoepidermoid and adenoid cystic carcinoma of salivary gland rests Direct spread of tumor from adjacent spaces: squamous cell carcinoma, nonHodgkin's lymphoma (PMS) or adenoid (PMS, PS)
diagnosis
of
Neck
pharyngeal
mu-
Asymmetric fossa of Rosenmuller Infectious or radiation-induced pharyngitis T omwaldt's Adenoidal abscess
cyst or faucial
Postinflammatory
(PS)
Lipoma Pleomorphic adenoma savilary gland rests
sarcoma (MS) mucoepidermoid cystic carcinoma
Table 8.3 Differential cosal space lesions
of the Supra hyoid
tonsillitis retention
or cyst
Benign tumors:
Pleomorphic adenoma salivary glands
Malignant tumors:
Squamous cell carcinoma (SCCa) of the pharyngeal mucosa Non-Hodgkin's Mucoepidermoid
from minor
lymphoma or adenoid
cystic
The most important intrinsic masses in the PMS are the squamous cell carcinomas arising from the mucosal surfaces of the nasopharynx, oropharynx, and hypopharynx (Fig, 8,7: Table8,3),
Pharyngeal Mucosal Space (PMS) The PMS includes the mucosal surfaces and submucosa of the nasopharynx, oral cavity, oropharynx and hypopharynx, This mucosa is lined with pseudostratified columnar or stratified squamous epithelium which can be the origin of squamous head and neck cancers, The PMS is surrounded posterolaterally by the middle layer of the DCF, which superiorly also encloses the posterior aspect of the pharyngobasilar fascia (PBF), This fascia attaches the pharynx and superior constrictor muscle to the skull base, The levator veli palatini muscle and the eustachian tube pass through the PBF superolaterally via the sinus of MorgagnL This provides a route of spread for nasopharyngeal tumors to the skull base, The contents of the PMS are: pharyngeal mucosa lymphoid tissue: tonsils, adenoids minor salivary glands torus tubarius superior and middle constrictor muscles palatoglossus muscle palatopharyngeus muscle levator veli palatini muscle salpingopharyngeus muscle pharyngobasilar fascia
Parotid Space (PS) The PS is located posterior to the masseter muscle and mandible and is lateral to the PPS, It is enclosed completely by the split layers of the superficial layer of the DCF, The posterior belly of the digastric muscle can indent the parotid from a posteromedial direction and can help in differentiating superficial lobe lesions from deep lobe lesions of the parotid, The contents of the PS are: • parotid gland • facial nerve (lateral to the retromandibular vein) retromandibular vein • intra parotid lymph nodes (20-30) • external carotid artery (medial) • parotid duct (Stensen's duct), Parotid gland masses are most important here and are termed intra parotid masses if> 50% of the circumference of the lesion is surrounded by normal parotid gland tissue, These lesions often extend through the stylomandibular tunnel, which can be widened by the mass (Fig, 8,7; Table 8.4),
8 Neck Table 8.4 sions
Differential
diagnosis
of paratoid
space
le-
Table 8.5 Differential diagnosis sions (suprahyoid and infrahyoid) Pseudotumors:
Pseudotumors:
Masseter hypertrophy Accessory parotid gland
Cystic lesions:
First branchial Benign
Lymphangioma
(children)
Parotitis
lesions:
Reactive adenopathy Sjogren's syndrome
Benign tumors:
or parotid
gland
Malignant
Hemangioma (children) Pleomorphic adenoma Warthin's tumor (cystadenoma
Carotid artery thrombosis, dissection, (pseudo)aneurysm IJV thrombosis or thrombophle-
node metastases from skin, breast, or lung
cancers Perineural tumor facial nerve
spread
along
Benign
Iymphoepithelial
cleft
bitis Cellulitis
Reactive or suppurative nopathy Tuberculous adenitis
or abscess
Benm
Paraganglioma: carotid body, glomus vagale or glomus jugulare Cranial or sympathetic nerve schwannoma, neurofibroma Neuroblastoma Meningeoma (via jugular
Malignant tumors:
ade-
foramen)
Encasement from direct infiltration by SCCa Lymph node metastases from SCCa, thyroid, melanoma Hodgkin's and non-Hodgkin's lymphoma Lymph node metatases and abdominal
lesions
Cysticlesions:
branchial
Inflammatory
cystic carcinoma
lymphoma-primary and metastatic Lymph SCCa,
second
lesions:
tumors:
Adenocarcinoma Non-Hodgkin's
Iftnultifocal
Atypical
Mucoepidermoidcarcinoma Adenoid
Asymmetric carotid bulb or ectatic carotid artery Asymmetricjugular vein
Cystic lesions:
abscess
Iymphomatosurn) Facial nerve neuroma
tumors:
le-
cyst cyst
(AIDS) Inflammatory
space
Vascular lesions:
cleft cyst
Iymphoepithelial
of carotid
from lung
carcinoma
cyst
(AIDS) Inflammatory lesions:
Sjogren'ssyndrome
Benign solid
Warthin's
lesions: Malignant lesions:
tumor (cystadenoma
Iymphomatosum) solid
Non-Hodgkin's
lymphoma
-
metastatic Lymph node metastases from SCCa, skin, breast, or lung cancers
Carotid Space (CS) The two carotid spaces run from the jugular foramen at the skull base to the aortic arch and are thus both suprahyoid and infrahyoid neck spaces. They are located posteromedial to the PPS and lateral to the retropharyngeal space (RPS). All three layers of the DCF condense into the thick carotid sheath. which prevents transgression of disease. The contents of the CS are: • common/internal • internaljugular
carotid artery vein (IJV)
• cranial nerve IX, XI, XII (upper part) • cranial nerve X (upper and lower parts) • sympathetic plexus • deep cervical lymph nodes. Lesions in the CS typically abut the carotid artery or jugular vein and are tubular or fusiform in shape. The most important lesions are the lymph node metastases of the deep cervical chain (levels II-IV) in adults and neurogenic tumors in children (Table 8.5). The lymph nodes of the neck are more extensively addressed in Chapter 22.
Masticator Space (MS) The masticator space is anterior to the PS, anterolateral to the PPS, and posterior to the buccal space (8S). It is most extensive in the craniocaudad direction and reaches from the superior border of the temporalis muscle to the inferior edge of the mandible. It can be thought of as
Spaces of the Supra hyoid Neck
two continuous parts, the suprazygomatic MS and the naso-oropharyngeal MS, This is important for imaging and axial slices should always continue above the zygomatic arch when delineating this space (Fig, 8,9), The MS is surrounded by two slips of the superficial layer of the DCF, which ends superiorly at the skull base medial to the foramen ovale. Through this elevator shaft, MS lesions can communicate with the intracranial structures (cavernous sinus), There is also a possibility of communicating with the pterygopalatine fossa through the pterygomaxillary fissure superomedially. The contents of the MS are: ramus and posterior body of mandible masseter muscle medial and lateral pterygoid muscles temporalis muscle inferior alveolar artery and vein masticator nerve branches (Vs) inferior alveolar nerve branches (Vs).
Differential
of masticator space
diagnosis
Pseudotumors:
Accessory parotid gland Hypertrophic masseter muscle Atrophy of mastication muscles
Inflammatory lesions:
Odontogenic abscess
Benign tumors:
Hemangioma Lymphangioma Leiomyoma Myositis ossificans Schwannoma, neurofibroma
from
Malignant tumors:
V3 de
nervation
Osteomyelitis
Malignant
of the mandible
fibrous
rhabdomyosarcoma,
histiocytoma, leiomyosar-
Chondrosarcoma, osteosarcoma Malignant schwan noma Perineural tumor spread along VB nerve Non-Hodgkin's lymphoma Squamous cell carcinoma of oropharynx
(retromolartrigone)
Differential
Pseudotumors:
diagnosis
Accessory
Cystic lesions: Inflammatory lesions:
Sebaceous
of buccal
parotid
space
lesions
gland
cyst
Spread of infection from adjacent spaces (MS) Reactive lymphadenopathy
Benign tumors:
Hemangioma Lipoma Fibromatosis
Malignant tumors:
SCCa-direct invasion or lymph mode metastases Non-Hodgkin's lymphoma Mucoepidermoid or adenoid cystic carcinoma from minor glands Soft tissue sarcomas
Buccal Space(BS)
Sarcomas (soft tissue and chondroand osteosarcomas) are the most dangerous lesions here and they can spread perineurally to the skull base via the masticator and inferior alveolar nerves (Table 8,6),
Table 8.6
Table 8.7
le-
The BS is a horizontal elongated space along the anterior part of the parotid duct, and is not a true fascially defined space, It lies between the buccinator muscle medially and the superficial muscles of facial expression laterally (Fig,8,5a), This space is almost entirely filled with fat that has some finger-like projections: laterally along the parotid gland, medially along the medial side of the mandible and superiorly along the both sides of the temporalis muscle, The contents of the BS are: buccal minor parotid facial buccal nerves buccal
fat pad salivary glands duct (Stensen's duct) artery and vein branches of facial and mandibular lymph
nodes,
The commonest lesions in the space are squamous cell carcinomas in buccal lymph nodes or that have spread from other spaces such as the MS and submandibular space (SMS) (Table 8,7), Minor salivary gland tumors occur infrequently,
Retropharyngeal
Space
(RPS)
The RPS is a midline space posterior to the airway that runs from the base of the skull to the upper mediastinum, It is sandwiched between
237
8 Neck
the pharyngeal mucosal space and the prevertebral space and is bounded by the middle layer of the DCF anteriorly and the deep layer of the DCF posteriorly. Its lateral walls are made of slips of the DCF called the alar fasciae. The anterior extension of the alar fascia divides this space in two parallel subspaces. the posterior of which has been referred to as the "danger space" (Fig.8.4c). This danger space continues even further downward. to the diaphragm. Both these spaces form natural conduits for spread of infection or tumors from the neck region into the (upper) mediastinum or skull base. The contents oftheRPSare: • lymph nodes (only suprahyoid) e fat.
The retropharyngeal nodes are important in staging SCCa. The lateral chain (nasopharynx and high oropharynx) can be visualized on imaging and contains the nodes of Rouviere, while the medial chain (nasopharynx to hyoid bone) is normally not visible on imaging studies. Besides lymph node metastases, frequent lesions are edema following radiation therapy or trauma, and direct invasion of the RPS from posterior extension of nasopharyngeal SCCa (Fig. 8.6; TableS.S).
Table space
Differential diagnosis of retropharyngeal lesions (suprahyoid and infra hyoid)
Prevertebral Space (PVS)
The PVS is also common to the suprahyoid and infrahyoid neck. It is a midline space from skull base to coccyx that lies posterior to the RPS. It is bounded anteriorly and posteriorly by the DLDCF which is attached to the transverse and spinous processes of the cervical vertebrae forming two compartments: the anterior compartment (AC) or PVS proper and the posterior compartment (PC) or paraspinal PVS. The corpus and pedicles of the vertebrae as well as the intervertebral disks are part of the anterior compartment, while the spinous process and laminae are part of the posterior one. The contents ofthe PVS are: prevertebral muscles vertebral artery and vein scalene muscles proximal brachial plexus cervical vertebrae and disks (posterior) paraspinous muscles (posterior) phrenic nerve (posterior) muscle motor nerve branches (posterior).
Table 8.9 Differential diagnosis of prevertebral lesions (suprahyoid and infra hyoid) Pseudotumors:
Degenerative lesions of the vertebrae and facet joints Cervical disk herniation
Inflammatory lesions:
Vertebral body osteomyelitis diskitis (pyogenic, TB)
8.8
Pseudotumors:
Ectatic carotid artery Edema secondary to radiation or (internal jugular) venous obstruc-
Inflammatory lesions:
Cellulitis or abscess and adenoids)
tion
Reactive pathy Benign tumors: Malignant tumors:
extension
Benign tumors:
of SCCa
Post-traumatic
Air
lesions:
Edema,
Malignant
Malignant
vical spine Vertebral body or epidural
bone tumors
of the cer-
metastasis: Lung, breast, prostate, thyroid and renal cell primaries Direct extension of SCCa into the
from
PVS Non-Hodgkin's
lymphoma-sec-
ondary Rhabdomyosarcoma
leukemia
hematoma
Chordoma Osteoblastoma, osteochondroma, ABC, giant cell tumor
tumors:
neighbouring spaces (PMS) Lymph node metastases from (nasopharyngeal) SCCa Lymph node metastases from melanoma, thyroid carcinoma Lymphoma,
or
colli tendinitis
Brachial plexus schwannoma/neurofibroma
adeno-
Hernangioma Lipoma Direct
Longus
(from tonsils
or suppurative
space
Post-traumatic lesions:
Edema,
hematoma
(paediatric)
Spaces of the Suprahyoid Neck
In the anterior compartment, lesions typically displace the pre vertebral muscles anteriorly, In the posterior compartment the fat of the posterior cervical space is displaced posterolaterally (Table 8,9), Bony lesions of this space are clinically the most important.
Sublingual Space (SLS) The teacup-shaped SLS is located deep in the oral tongue, superomedial to the mylohyoid muscle (the floor of the mouth) and lateral to the genioglossus-geniohyoid muscles, The space is not bounded by fascial structures and is posteroinferiorly continuous with the posterosuperior part of the submandibular space, For this reason lesions can easily spread out to the submandibular region, The contents of the SLS are: anterior hyoglossus muscle lingual artery and vein cranialnerves IX, XII lingual nerve(Vs) deep lobe of submandibular gland submandibular gland duct (Wharton's su blingual glands and ducts siiblinguallymph nodes,
tongue base from posterior to anterior ble 8.10), Spreading odontogenic infections commonly involve this space,
Submandibular
(Taalso
Space (SMS)
The horseshoe-shaped SMS lies above the hyoid bone and inferolateral to the mylohyoid muscle, The anterior part is bounded by fascial slips from the SLDCF. In the posterior part are no fascial boundaries, and here the SMS is continuous with the SLS and more superiorly with the inferior PPS,This creates a potential route from the submandibular region to the skull base, At the level of the hyoid bone the SMS is continuous with the anterior cervical spaces, The contents of the SMS are: anterior belly of the digastricus muscle facial artery and vein inferior loop of hypoglossal nerve superficial lobe of the submandibular gland submandibular and submental lymph nodes fat
duct)
The space is important in SCCa of the tongue, The tumors of the anterior tongue invade the SLS from superior to inferior and those of the
Congenital cystic lesions, submandibular gland lesions, and nodal metastases represent the commonest lesions in this area, depending on the age of the patient (Table 8,11), It is rare for lesions in the SMS to extend up into the SLS,
Table 8.11 Differential diagnosis of submandibular space lesions Table 8.10 lesions
Differential
diagnosis
of sublingual
space
Pseudotumors:
Pseudo-
Tongue
tumors:
poglossal nerve injury Lingual thyroid tissue
muscle
atrophy
from
hy-
Cystic lesions: Lymphangioma- cystic hygroma Ranula or diving ranula Dermoid-epidermoid Inflammatory lesions:
Cellulitis or abscess genie focus
from
odonto-
from calculi
Hemangioma
tumors:
Pleomorphic
adenoma
lesions:
Cellulitis
or abscess
Reactive adenopathy
sec
Invading a from anterior tongue or tongue base
Mucoepidermoid or adenoid cystic carcinoma (sublingual gland)
gland
inflamma-
tion duct from calculi Benign
Hemangioma
tumors:
Lipoma Pleomorphic adenoma (submandibular gland, parotid tail)
Malignant tumors:
Lymph node metastasis of face or oral cavity
(sublingual
gland) Malignant tumors:
Inflammatory
Second branchial cleft cyst Suprahyoid thyroglossal duct cyst Cystic hygroma-lymphangioma Diving ranula Epidermoid-dermoid
Submandibular
Ludwig's angina Dilated submandibular gland duct Benign
Cystic lesions:
Atrophy of digastric and mylohyoid muscle (V3 motor injury)
from SCCa
Lymphoma Mucoepidermoid or adenoid cystic carcinoma (submandibular)
239
8 Neck Table 8.12
Spaces of the Infrahyoid Neck Visceral Space (VS)
The miclline VS is the only space that is unique to the infrahyoid neck and is completely enclosed by parts of the MLDCF or the visceral fascia. It lies anterior to the RPS and PVS proper and superomedial to both carotid spaces. For practical purposes, the space can be further subdivided in four areas: laryngeal, thyroid, parathyroid, and esophageal areas. The contents of the VS are: thyroid and parathyroid glands paratracheal lymph nodes recurrent laryngeal nerve hypopharynx larynx cervical esophagus trachea. The visceral space is clinically most important thyroid and parathyroid lesions and squamous cell carcinomas of the larynx and popharynx, and lymph nodes (Fig. 8.8; ble8.12).
for for hyTa-
Anterior Cervical Space (ACS)
These small paired spaces lie in the anterolateral part of the infrahyoid neck. They are located lateral to the central visceral space, medial to the lateral strap and sternocleidomastoid muscles, and anterior to the carotid spaces. Their fascial boundaries are complex and all three layers of the DCF contribute to these. Su-
Differential
diagnosis
of visceral
space le-
Pseudotumors:
Pyramidal lobe of thyroid gland Prominent thyroid isthmus Patulous esophagus in tracheoesophagealgroove
Inflammatory lesions:
Cellulitis or abscess Reactive adenopathy
Laryngeal lesions:
Laryngocele Squamous cell carcioma Chondrosarcoma
Thyroid lesions:
Infrahyoid thyroglossal duct cyst Colloid cyst Goiter Acute/chronic thyroiditis Thyroid adenoma Thyroid carcinoma: papillary, follicular, medullary etc. Extranodal non-Hodgkin's lymphoma-primary or metastatic
Parathyroid lesions:
Parathyroid Adenoma
Esophageal lesions:
Zenker's diverticulum Esophageal carcinoma
Other malignant lesions:
Nodal metatasis from SCCa, thyroid Nodal lymphoma
cyst
periorly they are continuous with the SMS and, as no fascia separates the two, SMS lesions can spill into the fat-filled ACS as well. The contents of the ACS are: - fat.
The most important lesions in the ACS are transspatial diseases from the neighboring spaces, such as tumors or infections (Table 8.13). -------VS -------ACS
VS------------~-~
-------
ACS-------r~~"
~~
~b~~\"--==::s:.:-
CS RPS
Fig. 8.8 Pattern of spread of visceral space masses in the infrahyoid neck. (Legends see p. 233)
Spaces
Table 8.13 Differential space lesions
diagnosis
of anterior
cervical
Table 8.14 Differential space lesions
of the Infrahyoid
diagnosis
of posterior
Pseudomass:
Hypertrophied muscle
Second branchial cleft cyst Cystic hygroma-lymphangioma
Cystic lesions:
Cystic hygroma-lymphangioma Third branchial cleft cyst
Inflammatory lesions:
Cellulitis or abscess
Inflammatory lesions:
Benign tumors:
Lipoma
Malignant tumors:
Spread of thyroid carcinoma, phoma from adjacent spaces
Pseudotumors:
Enlarged
Cystic lesions:
anterior jugular
vein
Iym-
Benign tumors:
The contents
of the
pes
are:
fat preaxillary brachial plexus spinal accessory nerve (XI) dorsal scapular nerve spinal accessory lymph nodes. Lesions specific to the PCS should be centered in the fat and have a thin fat stripe that separate them from the CS. This can be difficult in the suprahyoid region where the PCS is only very small. Most important in this area are lymphoma or metastases from SCCa and other tumors in the spinal accessory lymph nodes (Table8.14).
Masses Involving Multiple Cervical Spaces There are few diseases that do not follow the spatial anatomic confinement or involve multiple spaces, either contiguously or non-contiguously (Table 8.15).
Cellulitis or abscess Reactive or suppurative nopathy Tuberculous adenitis
phoma Liposarcoma
Table 8.15 multispatial
ade-
neurofibroma
Nodal metastasis from SCCa (nasopharynx) Hodgkin's or non-Hodgkin's lym-
Posterior Cervical Space (PCS) The pes are also paired and lie in the posterolateral part of the suprahyoid and infrahyoid neck and completely encircle the paraspinal muscles in the paraspinal PVS. They have a posteriorly tilted. triangular shape, corresponding well with the occipital and subclavian triangles behind the sternocleidomastoid muscle. The fascial boundaries of the pes are complex. The DLDCF separates it from the perivertebral space and superficially the SLDCF forms a boundary to the subcutis and sternocleidomastoid muscles.
cervical
levator scapulae
Lipoma Hemangioma Schwannoma,
Malignant tumors:
Neck
(rare)
Differential diagnosis diseases of the neck
Transspatial Cystic lesions:
Inflammatory lesions:
diseases
of transpatial
(contiguous
and
spaces)
~
Cystic hygroma-lymphangioma Branchial cleft cyst Thyroglossal duct cyst Transspatial cellulitis Diving ranula
or abscess
Benign tumors:
Congenital hemangioma Lipoma Juvenile angiofibroma Schwannoma, neurofibroma
Malignant tumors:
SCCa-primary or metastatic Non-Hodgkin's lymphoma Rhabdomyosarcoma, thyroid carcinoma, minor salivary gland malignancies,
Multispatial Inflammatory nodal:
lesions
melanoma
(noncontiguous
spaces)
Reactive adenopathy from respiratory tract infection Idem, from mononucleosis, scratch fever, TBC Sarcoidosis
Malignant nodal:
Squamous
Non-nodal
Neurofibromatosis Hematogeneous metastases
disease:
cell carcinoma
Non-Hodgkin's lymphoma Other metastatic tumors
-;-'!ssSs~s~
upper cat-
241
Examination Technique
The CT examination relies on contrast-enhanced spiral or multislice scanning with thin sections. Care must be taken to choose the proper contrast injection protocol because scanning too early may yield suboptimum enhancement of malignant lesions. Scan Preparation and Patient Instruction
Patients are examined in the supine position with the neck slightly hyperextended and the shoulders lowered as much as possible. A proper alignment of the head in the cephalocaudal axis to prevent any asymmetry is important. Patients are instructed to breathe shallowly (to keep the vocal cords open) and to avoid coughing. Scanning during suspended inspiration is possible, but this may lead to artifactual narrowing of the airway. Suppression of swallowing is important because swallowing can cause major artifacts and may render an examination useless. There are two approaches: not to mention the word "swallowing" at all and to instruct the patient to breathe normally but keep their mouths open, or to ask the patients to swallow multiple times immediately prior to the examination to "exhaust" the swallowing reflex. A digital lateral scout radiograph is made to adequately plan the examination. Proper gantry angulation is important. In general the gantry is angulated parallel to the vocal cords (laryngeal ventricle) for the neck and larynx region and parallel to the hard palate for the face. If the vocal cords are not distinguishable on the scout view, angulation parallel to the hyoid bone or midcervical disk spaces are good approximations. The presence of large amounts of dental amalgam may require different neck positions and angulations to avoid metallic artifacts. Often this can be accomplished by using separate, overlapping acquisitions with different angulations for the face and neck (Fig. 8.9). Another source of artifacts is the shoulder region. In this region spiral scanning may be replaced by incremental technique using higher mAs and l
Fig. 8.9 Separate, overlapping acquisitions with different angulations forthe face and neck avoid metallic artifacts in patients with large amounts of dental fillings.
which provides high mA in lateral projections and markedly reduced mA in AP projections. For optimum results, an increase of the maximum mA to about 2-3 times that of a standard neck examination should be chosen. Despite this, the local patient exposure (effective mAs) is often less than with conventional scanning because of the marked dose reduction in the non-lateral projections. Radiologists are advised to check with the manufacturer whether the particular dose modulation program is able to reduce the mA settings to normal as the scan progresses to the neck. If not, there is a su bstantial increase of the patient's radiation exposure. Acquisition
Parameters
Spiral CT is performed with a collimation of 2-3 mm, a pitch of r. 1.5, and image reconstruction every 1-3 mm (Table 8.16). To optimize spatial resolution a field of view of 14-18cm and a 512 x 512 matrix are employed. For lesions of the nasopharynx, tongue, palate, and skull base, the axial images may be complemented by direct coronal scans with the patient positioned prone. This technique, however, is becoming obsolete with thin-section spiral and multislice scanning because high-quality coronal and sagittal sections can be reformatted from the axial data set. This requires overlapping image reconstruction for optimum quality.
Acquisition Parameters Table 8.16
Protocols for
CT
243
scanning of the neck
General Oral Contrast
None
Preparation
Remove dentures,
Patient position
Axial sections:
let patient swallow repeatedly
Coronal sections: Scan range Respiratory
prior to scan
Supine, axially symmetric with arms at the sides and shoulders low (may require strap) Prone, head reclined
lateral scanogram: skull base to jugular fossa parathyroid tumor, abscess, etc. may require scanning to the tracheal bifurcation phase
Quiet, shallow respiration; additional scans may be acquired during E phonation Noncontrast CT: Contrast-enhanced
Seen parameters
1-slice SC/TF/RI
a-suceSC
16-slicea SC
axial SW/RI
Standard (e.g. lymph nodes)
3/5/3 "
2-3 "
1-1.5 "
3/2
Tumor staging
24/2 "
1-l.25"
0.75-1.25
larynx (focused exam)
1/2/1!
0.5-1"
Contrast injection
t-suce V/F/D
4-slice V+N/F/D
Standard
70/2/20
70+ 50/2/20
Tumor staging Biphasic injection Biphasic scanning
100/2/60 70/2+50/1/80' 150 + 50/4/30 + 120d
100 + 50/2/60 70/3 + 60/1/80' 150 + 50/5/30 + 120
CT:
•
or a Valsalva maneuver
Windowing
W/L = 300/40 W/l=300/60
0.5-0.75
" "
MPRb SW/RI
3/2
2/2 cor
2/2
2/1 cor
50 + 30/2/30
d
100 + 50/2/70 70/3+50/1/80' 150 + 50/5/40+
120
d
SC= slice collimation (mm), TF = table feed (rnmjrotatlon}, RI'" recon. increment (mm), i! '" scanning direction. SW = effective section width (mm), MPR::: multiplanar reformats, axial= axial sections, cor= coronal, seq= sagittal eM = contrast material, Contrast concentration = 300 mg/ml iodine V=volume of CM (ml), N =volume of saline flush (ml), F= flow rate (mils), 0 =start delay, X '" trigger region a Pitch P=TF/(NxSC) should be chosen close to 1.5 with a-slice a scanners, 1.3-1.5 with 16-slice scanners b MPRare reconstructed from a "secondary raw data set" with SWIRl'" 1-1.510.6 or 0.5-0.810.4 ( read: 70 ml CM at 2-3 mils flow. followed by 50-60 ml CM at 1 mils, overall start delay of 80 s dread: 150 ml CM followed by 50 ml of normal saline, both at4-5 mils flow, start delay of 30-40s for l ", 120s for 2nd scan
The larynx should be scanned with thin (05-2 rnrn) sections and an imaging plane that is parallel to the vocal cords. Optimum alignment is possible using axial MPR. Focused spiral scans can also be performed separately, thus combining a survey examination of the neck with high resolution larynx imaging. E-phonation, modified Valsalva maneuvers, or puffed cheeks maneuvers can be employed to improve the evaluation of the vocal cords, the piriform sinuses. and the oral cavity, respectively. For a high success rate in these
functional scans, good patient instruction prior to the examination is needed. Multislice IT can either be performed with a spatial resolution similar to that of spiral CT (e.g., 4 x 2.5 mm
collimation)
or it can
be op-
timized to obtain near isotropic resolution (0.75 x 1.25 mm section collimation with 4- to 16-s1ice
scanners)
which
allows
for high-quality
reformations in any required imaging plane. The higher resolution scanning protocol is essential for optimum tumor staging: the lower resolu-
8 Neck
tion should be reserved for less critical indications such as lymph node staging. For evaluation of the larynx. 0.5-0.75 mm collimation may be used to further improve spatial resolution. For clinical evaluation. 2-3 mm thick axial sections should be reconstructed from a multislice CT data set. This can be done primarily from the raw data or by reformatting a thin-section overlapping "secondary raw data set". From these secondary raw data. 2-3 mm axial sections that may even be tilted retrospectively to match the anatomy (e.g .. parallel to the vocal cords) and 1.5-2.5 mm coronal or 2-3 mm sagittal sections can be reformatted. Different protocols for specific clinical questions should be devised. with adjustments to the parameters mentioned above. as needed. Complex protocols have been proposed for screening the entire neck using multiple spiral scans with optimized angulation and technique in the face and upper neck. lower neck. and shoulder regions. With multislice scanning and dose modulation many of these protocols will become obsolete. Scans should be reconstructed with a standard algorithm and displayed using soft tissue and bone window settings. For the soft tissues it is advantageous to keep the window fairly narrow (250-350 HU). With pathology in bone or cartilage. the raw data should be reconstructed with a higher resolution bone algorithm as well.
Contrast Administration The injection of intravenous contrast is mandatory for all but a few indications (e.g. dislocation of the arytenoid cartilages). If contrast material injection is contraindicated. other imaging techniques (MRl) should be preferred. Nonionic contrast medium is injected with a power injector using a monophasic or biphasic injection protocol. For good vascular opacification and delineation of lymph nodes. scanning can start after 20-25 seconds. and 80-100 ml of contrast material suffice. With multislice CT, the amount of contrast material can be further reduced for this specific purpose. Good enhancement of tumors requires a longer waiting period of more than 60 seconds because it relies on tumor hypervascularization as well as increased leakage of contrast material
into the interstitium. Most institutions will use a monophasic injection of 100-150 ml of contrast material with a flow rate of 2-3 mlls and a long enough scan delay of 60-70 s to obtain good tumor and vascular opacification. More sophisticated injection schemes use a biphasic injection that allows for continuing contrast injection well into the scan to ensure good opacification of the whole neck despite a relatively late scan phase. Such biphasic protocols rely on the injection of 50-70 ml at 2-3 ml/s, followed by 50-70 ml at 1 mlls. They allow for an increase in scan delay to 80-100 s. Another alternative is to inject a high volume of contrast material (e.g .. 150ml. followed by a 50ml saline flush) at a high flow rate of 4-5 mlls and perform biphasic scanning with a scan delay of 30s and 120s. This will allow for separation of the effects of tumor hypervascularization and interstitial enhancement.
Image Processing For complex clinical questions and for tumor imaging. near isotropic data acquisition should be performed. A "secondary raw data set" with 0.5-1.5 mm section width and 0.4-0.6 mm reconstruction increment is then used for further image processing. Standard processing of such data includes optimally angulated axial sections (e.g .. MPR parallel to the vocal cords or the hard palate). A symmetric display ofthe neck can be gained by appropriately tilting the reconstructed plane and correcting for a suboptimum alignment of the patient. Sagittal and coronal planes are excellent for displaying the full extent of the tumor and its relation to neighboring structures. Coronal planes may be tilted to be parallel to the anatomic structure of interest, e.g., the larynx and trachea. To keep the signal-to-noise levels acceptable and the partial volume effects low, the width of the reconstructed axial and multiplanar sections should be in the range of 1.5-2 mm. Thin-slab maximum intensity projections of 5-10 mm thickness are helpful for evaluating vascular structures, in particular carotid abnormalities and vascular involvement in tumor patients. Casts of the air-containing structures of the neck may provide an overview of asymmetries and gross deformation (see Fig. 2. 26a). Tfssuetransition projections provide a better under-
Contrast Administration standing of wall irregularities (see Fig.2.26b). Care has to be taken not to mistake a collapsed portion of the hypopharynx with tumor infiltration. In cases of doubt. additional scans using a modified Valsalva maneuver should be performed.
Virtual laryngoscopy can demonstrate subtle irregularities and asymmetries of the larynx and hypopharynx and can provide a roadmap to sites that are difficult to assess by conventional laryngoscopy (see Figs. 2.41 d and 8.21 b).
Cystic Lesions There is a large variety of cystic lesions in the neck that are best characterized by their location in a particular cervical space (Table 8.17).
Thyroglossal
Duct Cyst
The thyroid descends from the tongue base to its normal position during the first 7 weeks of gestation. During migration it is connected to the tongue via the thyroglossal duct. This duct normally involutes during the 8- Wthweek. If parts persist. they can give rise to midline (75%) or paramedian (
diagnosis are given
of cystic lesions in brackets)
Parotid space
Cystic hygroma (Warthin tumor) (Mucoepidermal
Mandibular
Branchial
(Neurogenic angle
regan
carcinoma)
tumors)
cleft cyst
CT Morphology The cyst is an unilocular, well-defined mass with fluid attenuation along the course of the thyroglossal duct in the anterior neck. It has a thin wall and septations can be seen infrequently. Normally it has a fluid-like attenuation and thin peripheral rim-enhancement may be seen on contrast-enhanced CT. After infection, the attenuation can increase, septations can occur and the enhancing rim becomes thicker and irregular. Sinus tracts/fistulae can be seen in the subcutaneous tissues as irregular tube-like fluid collections. Modularity or a softtissue mass within the cyst is suggestive for malignancy.
Cystic hygroma Dermoid cyst External
laryngocele
Branchial Cleft Cysts
Ranula Abscess (Malignant Sublingual
space
Epidermoid
lymphoma) cyst
Renna Mucocele
of submandibular
gland Median reqcn
of
anterior to the strap muscles infrahyoidally (65%). lt is the commonest congenital mass and the second commonest benign mass, usually presenting before the age of 20 and 4-5 times more frequently in females than males. The mass is 15-30 mm in size and can move upward with tongue protrusion. Very rarely, thyroglossal duct abnormalities can be associated with thyroid carcinoma, usually of the papillary type.
cervical
Thyroglossal duct cyst Submental dermoid cyst Lymphangioma Necrotic metastases
Lymphoma
Carotid space
(Lymphadenitis
and liquefac-
tion) (Cyst
thyroid
in ectopic
Aneurysm (Malignant
tissue)
of carotid artery lymphoma)
These rests sinus lateral adults. ratory
cysts probably arise from epithelial cell and can manifest themselves as cysts, tracts, or fistulae. They are usually uniand occur in older children or young The cyst can distend during upper respitract infections. First branchial cleft cysts: These manifest as recurrent abscesses around the ear or mandibular angle. A recurrent parotid abscess not responsive to antibiotics is typical. The cyst may be superficial, deep or located in the parotid gland. Sinus tracts can extend to the ear or skin. Second branchial cleft cysts: These are by far the commonest branchial cleft anomaly and
245
8 Neck
several subtypes exist. They are most frequently found lateral to the carotid sheath and posterior to the submandibular gland (type II) or extending medially in the carotid bifurcation (type III). The mass is painless, slow-growing and measures 1-10 cm. It can become painful when infected. Atypical forms can lie in the parapharyngeal space and project upward from the faucial tonsil to the skull base. Third and fourth branchial cleft cysts: These are rare. Third cleft sinus tracts or cysts arise from the apex of the piriform sinus and extend posterior to the sternocleidomastoid muscle and carotid space into the posterior cervical space. Anomalies of the fourth cleft are usually sinus tracts extending from the piriform sinus down as far the aortic arch or right subclavian artery. Most tracts are short and sometimes there is a cutaneous fistula at the base of the neck. •
CT Morphology
These cysts are well-circumscribed and homogeneously hypoattenuating. The cyst wall is thin, but may thicken after infection, and enhances after administration of contrast (Fig. 8.10). If large, they can displace the surrounding structures from their characteristic locations. Thymic Cyst
These uncommon cysts probably arise from thymopharyngeal duct remnants. They are found immediately adjacent to the carotid sheath and
can occur anywhere from the mandibular angle to the mediastinum. In 50% of cases there are connections with mediastinal thymic tissue. Patients usually present with a painless, slowly growing mass in the lateral neck near the upper mediastinum and in close relation with the sternocleidomastoid muscle. In 65% of patients they present during the first decade of life, are more common in males, and occur most frequently on the left side. When cysts are large, patients can complain of hoarseness, dysphagia, or stridor. •
CT shows an unilocular or multilocular, well-defined cystic mass with fluid attenuation. The cyst size can vary between 1 cm and 20cm. The mass is often elongated and located adjacent to the carotid space and may extend into the mediastinum. Thymic cysts may enlarge during Valsalva's maneuvers. Cystic Hygroma (Cystic Lymphangioma)
This is the commonest form of lymphangioma in infancy and childhood and is usually detected when the patient is 2 years old. It involves the lower face and neck and is most commonly found in the posterior cervical space or oral cavity. The lesion does not respect fascial boundaries and may extend inferiorly into the axilla and mediastinum or anteriorly in the floor of the mouth and tongue. Most are asymptomatic. Cystic hygromas manifest as painless soft masses of variable size. They grow slowly, but may suddenly increase in size after trauma or infection. Resection is difficult and many recurrences are seen after surgery. Other types of lymphangioma include the smaller cavernous (tongue, salivary glands) and capillary (skin) forms that are significantly smaller. •
Fig. 8.10 Superinfected lateral neck cyst (second branchial cleft cyst). The cyst contents have perforated into the surrounding soft tissues (arrow).
CT Morphology
CT Morphology
Cystic hygromas show multiple, poorly circumscribed cystic spaces with a homogenous fluid attenuation. Infected lesions show an increased attenuation. The spaces in the hygroma can vary in size from a few mm to more than 8 cm. They are commonly centered in the posterior cervical or submandibular spaces and show transspatial growth.
Contrast Ad ministration
Dermoid and Epidermoid Cyst
These cysts are manifestations of benign teratomas and are covered with squamous epithelium. Unlike dermoids, epidermoids do not contain skin appendages. Epidermoids present during infancy, while dermoids manifest during the second or third decade. In the neck, dermoids are more frequent than epidermoids. Both cysts are mobile, soft masses in the suprahyoidal midline that grow slowly and are of variable size. The most frequent location in the head and neck is the oral cavity. •
CT Morphology
These cysts are thin-walled, unilocular masses in the submandibular or sublingual space that have a fluid attenuation. The dermoid cyst may show "marbling": presence of fat nodules in the fluid matrix. The location of the mass with respect to the mylohyoid muscle is crucial for therapy planning, which requires coronal images or reformations. Laryngocele
A laryngocele is a dilation of a small pouch in the anterior roof of the laryngeal ventricle, the laryngeal saccule. Internal laryngoceles (40%) are limited to the larynx, while external ones (25%) extend through the thyrohyoid mem-
a
b
Fig. 8.11 (a) Mixed laryngocele with intratralaryngeal components (scan at the level of bone). There is a mass effect with displacement rowing of the hypopharyngeal lumen, (b)
brane and only have an external dilatation. Mixed laryngoceles (35%) have a dilatation both in the larynx and outside. A laryngocele develops due to a combination of a congenital long saccule and longstanding increased supraglottic pressure (shouting, coughing, glass blowing, playing wind instruments). Rarely they are congenital but are usually acquired and manifest in adult life with dysphagia, hoarseness, and stridor. In 1 % of cases they are associated with laryngeal cancer. •
CT Morphology
Laryngoceles appear as smooth, well-defined masses in the paralaryngeal space. They may cross the thyrohyoid membrane at the level of the superior laryngeal neurovascular bundle. Depending on the contents (air, fluid, mucoid secretions) the attenuation will vary and airfluid interfaces can occur (Fig. 8.11). T ornwaldt's
Cyst
This cyst is derived from embryological rests of the notochord and is found in 4% of the normal population. It measures 1-5 mm and is characteristically located in the midline of the nasopharyngeal roof between the longus capitis muscles. Infection may lead to postnasal discharge, prevertebral muscle spasm, or even abscess formation.
c and exthe hyoid and narAir-filled
right-sided laryngocele in a typical supraglottic location. (c) Volume-rendered coronal cut through the larynx in the same patient.
247
8 Neck •
CT Morphology
The cyst is usually smooth and well defined. It may have a fluid attenuation but is more often
isoattenuating to muscle due to a high protein content of the fluid.
Benign Tumors Pseudotumors and Tumor-Like Lesions Lingual Thyroid
This is the commonest form of ectopic thyroid tissue and is seen more frequently in women. It usually present during puberty. It represents an arrest of migration of thyroid tissue within the tongue, usually between the circumvallate papillae and epiglottis. This arrest may be incomplete and in 70-80% of patients there is no thyroid tissue in the lower neck. •
CT Morphology
This is seen as a high-density soft tissue mass in the posterior tongue with vivid enhancement after contrast administration (Fig. 8.12). Frequently, no thyroid tissue is seen in the lower neck. Vascular Pseudomasses
Normal variants can resemble masses of vascular origin. Often there is a marked asymmetry of
the internal jugular veins, with the right one bigger than the left. In elderly patients the carotid arteries can have a tortuous, more medial course. The pterygoid plexus can also be asymmetric, appearing as an enhancing area along the medial border of the lateral pterygoid. Neurogenic Tumors Schwan noma and Neurofibroma Schwannomas are small, encapsulated
tumors of the nerve sheath arising from Schwann cells in which the tumor grows eccentrically from the nerve. Patients are between 20-50 years and present with a painless, slow-growing mass in the anterolateral neck. The cervical nerve roots, vagus nerve, or sympathetic plexus are most frequently involved. They are usually solitary, but can be multiple in patients with neurofibromatosis 2 (5 % of cases). Rarely, schwannomas can show malignant degeneration (malignant peripheral nerve sheath tumor =MPNST, or neurosarcoma) . Neurofibromas are nonencapsulated and grow concentrically, giving an enlargement of the nerve of origin. They occur in patients between 20-40 years and in 90% are of the localized variety. Infiltrative growth can occur with plexiform subtypes. They can be multiple and are in 10% part of neurofibromatosis syndromes. •
Fig. 8.12 Lingual thyroid (noncontrast scan). The dystopic thyroid tissue shows nonhomogeneous CT densities. Note the resemblance to the CT morphology of a carcinoma of the tongue base. Two regressive punctate calcifications (arrows) suggest the possible etiology of the mass.
CT Morphology
Fusiform, well defined tumors with low density, frequently in the carotid space. Paraspinal tumors can have a characteristic dumb-bell shape. The density of schwannomas is related to lipid-rich Schwann cells and fat. Neurofi bromas tend to be more fatty, while schwannomas can have cystic or necrotic components (in 20% of cases). After contrast administration, there is a moderate, heterogeneous enhancement.
Mesenchymal Tumors quires angiography for exact delineation tumor vessels and feeding arteries. •
Fig. 8.13 Glomus tumor (paraganglioma) in the right carotid bifurcation. CT shows an intensely enhancing mass with splaying of the internal carotid artery (ICA) and external carotid artery (ECA).
of
CT Morphology
The lesions are usually well defined and show intense enhancement in the carotid bifurcation, along the vagus nerve or in the jugular bulb (Fig. 8.13). This enhancement is homogenous but shows a rapid washout. The tumors can distort the normal course of adjacent vessels. In large tumors hypodense, necrotic regions are visible. Lesions adjacent to the skull base can show permeative bone erosion (glomus jugulare). A CT angiography technique with thin sections, preferably using multislice CT, yields the best results. Mesenchymal Tumors
Paragangliomas
Paragangliomas (chemodectomas or glomus tumors) are rare, slow growing, hypervascular tumors arising from neural crest cells and are classified by location and origin. The commonest types are: glomus caroticum in the carotid bifurcation, glomus jugulare from the nerves in the bulbus jugularis, glomus vagale from the ganglion nodosum of the vagal nerve, located between carotid bifurcation andjugular bulb, and glomus tympanicum from the tympanic plexus in the cochlear promontory. Symptoms depend on location of the tumor: glomus caroticum and vagale present as slow-growing masses with no other symptoms, while glomus jugulare and tympanicum may have associated pulsatile tinnitus, hearing loss and cranial nerve palsy. There is a familial history in 30% of patients. There is an increased incidence in females and in patients living at high altitudes. The lesions are multiple in 30% of patients with familial paragangliomas but less than 5% in non- familial types. Paragangliomas can occur in association with other paragangliomas or tumors in other body parts, but these combinations are rare. Head and neck paragangliomas are hormonally active in only "170. The lesions can be malignant in 10% of cases. Therapy of carotid and solitary vagal body tumors is surgical, whereas forjugular and tympanic glomus tumors a wait-and-see policy is advocated. Further preoperative work-up re-
Lipoma
Lipomas are frequent tumors of the neck, usually presenting at advanced age. They are soft, mobile, and painless masses. They can arise in any neck space, but are most frequent in the posterior cervical space. The retropharyngeal ones can become large before clinically apparent. Hibernomas are lipoma variants that contain brown fat. •
CT Morphology
Well-defined masses with a homogenous fat density and thin septa. There is no clearly identifiable capsule or only a very thin capsule that does not enhance after contrast administration. Hemangioma
These are the commonest cervical tumors in children. Girls are more often affected than boys. The two most frequent varieties are capillary and cavernous hemangiomas. Capillary hemaugiomas usually present directly after birth and show proliferative growth during the first 12 months. They may be located in nearly all spaces, but have a preference for the masticator or parotid spaces. These lesions are more frequently superficial and may show a bluish discoloration of the skin. They show fatty replacement and involution in adolescence.
249
Neck
Cavernous hemangiomas and other vascular malformations occur in adolescence and do not involute at later ages. Often they are deeply located and can grow large involving multiple neck spaces. Phleboliths may arise in these hemangiomas. Large AVMs with high flow can show hypertrophic feeding vessels. •
CT Morphology
Hemangiomas may be well defined or grow in an infiltrative pattern. Most lesions are isoattenuating with muscle and show intense enhancement after contrast administration. Phleboliths may be present in cavernous hemangiomas. Rhabdomyoma
These are very rare lesions of striated muscle cells with a predilection for the head and neck. They usually affect the oropharynx. submandibular region. or larynx. On CT the lesions are circumscribed. smoothly marginated and have muscle density. After contrast. enhancement equal to that of muscles is seen. Aggressive Fibromatosis (Desmoid Tumor)
This is a well-differentiated fibrous lesion with locally infiltrative growth. originating from the musculoaponeurotic junctions. The lesions occur most often in the face or supraclavicular region of young adult patients and are usually painless and slow-growing. The CT findings are nonspecific and may resemble malignancy. Due to the high fibrous content. there is persistent interstitial enhancement of the tumor. Benign Fibrous Histiocytoma
Rare benign tumors consisting of fibroblastic cells. They are common in the subcutaneous tissues and rarely found deep to the superficial cervical fascia. Young adults are most frequently involved. The CT findings are nonspecific and may resemble MFH.
Juvenile Angiofibroma
Angiofibromas occur most often in teenage boys and originate from the superior wall of the nasal cavity or nasopharynx. The lesion is highly vascular and. although benign. is locally infiltrative. Extension can be to the pterygopalatine fossa laterally. sphenoid sinus or orbit superiorly. and even into the middle cranial fossa. It is often accompanied by local bone destruction. CT Morphology
Juvenile angiofibroma presents as a soft tissue mass with infiltrative growth pattern and often bone destruction. The tumor enhances strongly after contrast administration. Benign Tumors of the Larynx and Hypopharynx
Benign laryngeal tumors account for approximately 10% of all laryngeal masses. Benign tumors of the larynx and hypopharynx include fibromas, angiofibromas, and fibromyxomas (sessile or pedunculated "laryngeal polyps"), which most commonly involve the anterior part of the vocal folds. Adenomas are rare tumors that may involve the aryepiglottic fold, ventricular fold, or subglottic area. Myogenic tumors, lipomas, hemangiomas, chondromas (most arise from the cricoid cartilage), and fibroneuromas are also very rare. They are diagnosed by visual inspection and are not an indication for CT, although they are occasionally noted as incidental findings. Papillomavirus-induced laryngeal papillomas occur predominantly in young males and most commonly affect the anterior vocal cords. The multiple superficial growths of papillomatosis are common in younger patients « 10years), whereas most lesions in older patients are solitary. Malignant transformation may occur. CT is indicated only if there is suspicion of malignancy (stage > TI). CT Morphology
Benign tumors generally have smooth margins. Laryngeal or hypopharyngeal polyps may present as small nodular protrusions at the anterior commissure of the vocal cords, or as a large, smoothly marginated mass within the larynx or hypopharynx (Fig. 8.14). Pedunculation is in-
Lymph Node Staging Fig. 8.14 Small calcified fibroma of the posterior wall of the hypopharynx (a). The relationship to the surrounding structures is best appreciated on a volume-rendered coronal view (b).
a
b
frequently detected at CT. Absence of tissue invasion may suggest the diagnosis. Papillomas may present a cauliflower-like surface when extensive. Chondromas have a lower attenuation than most soft tissue tumors, typically in the 20-30 HU range (Fig. 8.15). They frequently lead to displacement of the smaller cartilages. From CT morphology alone there is no secure differentiation between benign and malignant disease, although of course, infiltrative growth suggest malignancies. Fig. 8.15 Laryngeal chondroma causing subtotal obstruction of the larynx. The noncontrast scan shows a markedly hypodense mass with splaying of the arytenoid cartilages (arrows).
Malignant Tumors Lymph Node Staging
The presence oflymph node metastasis is an important prognostic factorin head and neck cancer. Asinglenodalmetastasisreducestheprognosis by 50%, and the presence ofacontralateral metastasis byanother 50 %. Ifextranodalspreadis detected the prognosis is further reduced by an additional 50%. Evenin palpable normal necks the riskofoccult metastases can be as high as 20-50%, depending on the type of the primary tumor. In evaluating the lymph nodes of the neck it is important to discriminate between two clinical scenarios, patients with a palpable NO neck and patients with a palpable positive neck (N411).
NO patients are often electively treated by neck dissection if the expected risk of occult metastasis exceeds 20%, based on characteristics of the primary tumor. If the risk is lower, a wait -andsee policy under strict clinical and imaging control is advocated. Negative imaging studies may thus reduce the number of dissections while positive studies (upstaging) can guide appropriate therapeutic procedures. In palpable N + necks, finding additional positive nodes at imaging may guide the type of neck dissection and adjuvant therapy, depending on the lymph node levels involved. Lymph node invasion can be evaluated with CT and is usually staged according to the UICC (TNM) cri-
251
8 Nec!<
Table 8.18 Lymph node staging of cervical tumors (excluding the thyroid gland and proximal esophagus) NO
No regional lymph node metastases
Nl
Metastases in a single ipsilateral node, 3 cm or less in greatest dimension
N2a
Metastases in single ipsilateral node, greater than 3 cm but no more than 6 cm in greatest dimension
N2b
Metastases in multiple ipsilateral nodes, none more than 6 cm in greatest dimension
N2c
Metastases in bilateral or contralateral nodes, none more than 6cm
N3
Metastases in a node more than 6 cm in greatest dimension
teria for head and neck tumors (Table 8.18). Criteria used for differentiation are size, shape, grouping, necrosis, and extracapsular spread. The criteria frequently used are: minimal axial diameter of 11 mm for level II, minimal axial diameter of 10mm for other levels, grouping of three or more borderline (8-10 mm) nodes and irregular enhancement. These criteria are based on neck dissection studies and probably lack sufficient sensitivity for patients with clinically NO necks. Because CT is relatively inaccurate in preoperative diagnosis, ultrasound combined with fine needle aspiration biopsy is most often used as a first line modality. Other methods in current development like (CT)-PET, thallium-SPECT or USPIO-enhanced MRI may improve the accuracy of preoperative lymph node imaging in the near future. A more detailed overview of lymph node levels and criteria for malignancy can be found in Chapter 22.
The cervical nodes can also harbor metastases from primaries other than squamous cell carcinomas of the neck (Fig. 8.16). Thyroid cancers, salivary gland cancers, renal cell carcinoma, breast carcinoma, lung carcinoma, and melanoma are all known to metastasize to this region. Nasopharyngeal Carcinomas
Up to 80% of nasopharyngeal malignancies are squamous cell carcinomas. A number of risk factors have been established including EpsteinBarr virus, genetic predisposition in ethnic Chinese, poor living conditions, and chronic sinonasal infection. Males are 2.5 times more often affected than females. S
CT Morphology
The main tumor subsites are the posterosuperior, anteroinferior, and lateral walls. The tumor is based most often in the fossa of Rosenm uller (lateral wall). It may spread submucosally in all directions along fascial planes, with early infiltration of palatal muscles (Fig. 8.17). The pharyngobasilar fascia represents a relative boundary so that posterior extension is infrequent. Deep extension often involves the PPS and MS. Advanced lesions may erode bone, spread perineurally, and reach the skull base or the cavernous sinus via the foramen lacerum (intracranial in 30%). Nodal metastases are the rule: 80-90% of patients develop metastases in lateral retropharyngeal, level II, III, or V nodes. In 35-50% these nodes are bilateral (depending on stage of primary tumor). There is also a high
Fig. 8.16
(a) Lymph node metastases with central necrosis (arrow) from squamous cell carcinoma of the oropharynx. (b) The coronal MPRin a patient with NHL demonstrates the rounded shape (ratio of longitudinal: transverse diameter < 2) even of small pathologic nodes.
a
b
Oropharyngeal Carcinomas Oropharyngeal Carcinomas
Fig. 8.17 Nasopharyngeal carcinoma of the torus tubarius on the right side. Note the asymmetry and the polypoid protrusions. In this scan after 100 s delay, there is only minimum contrast uptake by the tumor.
frequency (5-40%) of distal metastases in bone, lung, and liver. The T-stages are given in Table 8.19.
In the oropharynx, 90-95% of malignant tumors are squamous cell carcinomas. Tumors in the oropharynx tend to be more aggressive and more poorly differentiated than in the oral cavity. The commonest risk factors are a long history of tobacco and alcohol use, but increased risk is also associated with Plummer-Vinson syndrome and syphilis. The tumors are seen twice as frequently in men as in women. Because of their aggressive nature and mild symptoms, these tumors are often advanced at presentation. Symptoms include sore throat, foul breath, dysphagia, or ear pain. The tumors tend to extend submucosally and along fascial planes. Four separate tumor subsites are of interest: tongue base, tonsillar region, posterior wall, and soft palate. Tonsillar and tongue base carcinomas are most frequent. Each tumor subsite has a specific pattern of spread and lymphatic drainage. •
Important imaging features for evaluation of nasopharyngeal carcinomas: • assessment of tumor spread in surrounding deep spaces: PPS, MS, CS • skull base erosion (requires coronal thin section CT) • perineural tumor spread via the mandibular nerve • intracranial (cavernous sinus) extension.
Table 8.19 T-staging (UICC, 1997)
of
nasopharyngeal
carcinoma
Tis
Carcinoma in situ
T1
Tumor confined to the nasopharynx
T2
Tumor extends to soft tissues of oropharynx and/or nasal fossa
T2a
Without para pharyngeal
T2b
With para pharyngeal
T3 T4
extension
CT Morphology
The T-stages of Oropharyngeal carcinomas are given in Table 8.20. Tongue base: the tumor remains unilateral until it is quite extensive. It can spread anteriorly into the oral tongue, inferiorly in the valleculae or pre epiglottic fat and superiorly to the tonsillar fossae (Fig. 8."18).Frequently the tumors initially present with cervical node metastases. Adenopathy is present in 50-80% of patients, and is bilateral in 20-30%. Usually, this involves levels II, III, and 1.Surgical treatment options depend on the involvement of the neurovascular bundle, submucosal extension in adjacent areas and crossing of the midline. Table 8.20 T-staging (UICC, 1997)
of
Oropharyngeal
carcinomas
Tis
Carcinoma
in situ
T1
Tumor 2 cm or less in greatest dimension
T2
Tumor invades bony structures and/or paranasal sinuses
Tumor between 2 and 4 cm in greatest dimension
T3
Tumor with intracranial extension and/or involvement of infratemporal fossa
Tumor greater than 4 cm in greatest dimension
T4
Invasion of adjacent structures: pterygoid muscles, mandible, hard palate, deep tongue muscles, larynx
extension
253
8 Neck
a
b
c
Fig. 8.18 Tongue base carcinoma seen on axial (a) and coronal sections (b) as well as sagittal volume-rendered ages (c) from a thin-section multislice CT data set.
Important imaging features carcinoma:
in tongue base
• extension to floor of the mouth and surrounding structures • relationship to ipsilateral lingual neurovascular bundle • extension across midline and relation to contralateral neurovascular bundle. Tonsillar region: the tumor may spread inferiorly into the tongue base or infiltrate the surrounding structures like the tongue or pterygoid muscles. Because of the asymmetry of the normal tonsils, small tumors are difficult to detect. Tonsillar fossa tumors show early lymphatic metastasis, which is seen in 60-75% of patients at the time of presentation. In tumors of the anterior tonsillar pillar this figure is somewhat lower at 40-55%. Bilateral adenopathy is seen frequently and lymph node levels II, I or less commonly 111 orIVareaffected. Soft palate: these tumors are often well differentiated and have the best prognosis. They usually affect the oropharyngeal side of the palate and extend in all directions, but most often to the hard palate, tonsillar pillars, and PPS. Lymphatic metastasis is present in 40-60% of patients at time of diagnosis, which is frequently bilateral and involves usually level I, II,III,orretropharyngealnodes. Posterior wall: tumors lead to a thickening of the pharyngeal wall and can extend along the constrictor muscles superiorly to the skull base
im-
or inferiorly to the hypopharynx. Deeper invasion can be seen in the tongue base, retropharyngeal space or prevertebral space. Cervical adenopathy is common and seen in 50-70% of patients at diagnosis. The retropharyngeal nodes are first affected, followed by nodes in levels II, I II, or V. Important imaging features oropharyngeal carcinomas'.
in
other
• submucosal extension into surrounding soft tissues of the neck (PPS, CS, nasopharynx) • invasion of the tongue base • bone erosion (for skull base thin section coronal CT needed) • encasement of the carotid artery • prevertebral muscle invasion.
Oral Cavity Carcinomas
As in the oropharynx, the great majority of malignancies are squamous cell carcinomas. There is a histological gradient with the more aggressive tumors located posteriorly and most tumors arise in the dependent positions, such as the floor of the mouth. The most important risk factors are (chewing) tobacco and alcohol use. Pipe smoking and exposure to sun are specifically associated with carcinomas of the lip. Superficial lesions are normally asymptomatic, but larger lesions can be associated with
Oral Cavity Carcinomas
ear pain, foul breath or ill fitting dentures. ~ive separate tumor subsites are of interest: Ilps, floor of the mouth, oral tongue, buccal mucosa, and gingiva and hard palate. The lip and the floor of the mouth are most freq uently affected. Each tumor location has a specific pattern of spread and lymphatic drainage. •
CT Morphology
The T-stages of oral cavity carcinomas are given in Table 8.21. Lips: Imaging is only required in extensive lesions. Most often, lip tumors arise on the vermilion border of the lower lip and can extend in the orbicularis oris muscle, skin, and mandibula. Perineural extension along the mental nerve i? rare. Lymphatic spread is seen in 10-20% of patients (depending on stage prima~y tumo~) and usually involves levels I and II. ThIS tumor IS frequently seen in younger women.
Table 8.21 1997)
T-staging
of oral cavity carcinomas
(UICC,
Tis
Carcinoma
in situ
T1
Tumor 2 cm or less in greatest dimension
T2
Tumor between dimension
T3
Tumor greater than 4 cm in greatest sion
T4 lip
Invasion of adjacent structures: skin of the face, inferior alveolar nerve, floor of the mouth, cortical bone
T4 rest
Invasion of adjacent structures: extrinsic muscles of the tongue, cortical bone, maxillary sinus, skin
2 and 4 cm in greatest dimen-
Important imaging features in lip carcinomas: • soft tissue invasion • bone erosion (buccal surface of mandible). Floor of the mouth: Most tumors
arise within 2cm of the anterior midline and can extend in the extrinsic tongue muscles, sublingual or submandibular spaces. Encasement of the ipsilateral neurovascular bundle is seen frequently and may lead to denervation atrophy. Crossing of the midline with infiltration of the contralateral neurovascular structures is important therapeutically as this requires a total 910ssectomy. Although infiltration of the ma~dIbular periosteum occurs early, frank destru~tlOn of the mandible is a late phenomenon (FIg. 8.19). At the time of diagnosis, 30-50% of patients show nodal metastases (depending on stage primary tumor), usually in level I and I I nodes. Important
imaging
in floor of the
• deep invasion along mylohyoid and hyoglossus muscles • relation to ipsilateral neurovascular bundle • extension across midline and in contralateral lingual neurovascular bundle • tongue base invasion • invasion of soft tissues of the neck • extent of bone erosion. Oral tongue: Most tumors arise from the under-
surface and lateral borders. They infiltrate
Fig. 8.19 Small squamous cell carci noma of the floor of the mouth on the right side (a). The air marks the oral cavity, (b) Large floor of the mouth carcinoma with encasement of the mandible. There is central tumor necrosis. Similar enhancement is seen in an ipsilateral nodal metastasis anterior to the sternocleidomastoid muscle (arrow).
a
features
mouth carcinomas:
b
the
255
8 Neck
deep intrinsic and extrinsic muscles and can extend submucosally to the floor of the mouth, tonsils, and mandible, Crossing of the midline has important therapeutic implications, Bone erosion of the mandible is only seen in very advanced lesions, Tumors metastasize early to level I. 11, and III lymph nodes and 35-65% of patients (depending on stage primary tumor) present with (bilateral) adenopathy, Important
cinomas:
imaging features
in oral tongue car-
• relation to ipsilateral neurovascular bundle • extension across midline and in contralaterallingual neurovascular bundle invasion of floor of the mouth and associated bone erosion.
Buccal mucosa: These tumors commonly arise in the lateral walls, They spread laterally along the buccinator muscle to the masticator space or the pterygomandibular raphe and can erode underlying bone from the maxilla or mandible, There are often aggressive with early lymphatic spread to level I. 11, and parotid nodes, Important
nomas:
imaging
features
in buccal carci-
• submucosal extension • bone erosion.
Gingiva and hard palate: The extent of these tumors is often underestimated by crosssectional imaging, Primary hard palate SCCa is rare and usually an extension from an upper gingival tumor. Via the hard palate the tumor can break into the nasal cavity, Lower gingival tumors can extend into the floor of the mouth or erode the mandible, The amount of bone erosion will determine the type of surgical therapy, Up to 50% of patients present with lymph node spread to the level I (gingiva) or retropharyngeal and level II (palate) nodes, Retromolar tligone carcinomas are a special kind of gingival tumor with a complex spread pattern. They can extend anteriorly into the buccal region, posteriorly in the tonsillar fossa, superiorly along the pterygomandibular raphe into the nasopharynx and skull base, and inferiorly into
the floor of the mouth, Bone erosion can occur early and can only be evaluated accurately by thin section imaging, It is typically located at the ventral side of the ramus mandtbulae. Important
imaging
features
hard-palate carcinomas:
in gingiva and
• bone erosion • submucosal extension • perineural extension (e.g. incisive and greater/lesser palatine foramen),
canal
Hypopharyngeal Carcinomas The vast majority of hypo pharyngeal tumors are squamous cell carcinomas, Risk factors for hypopharyngeal carcinomas are alcohol and tobacco use and for postcricoid lesions there is an association with Plummer-Vinson syndrome, H ypopharyngeal tumors are clinically silent and often present in advanced (T3-T4) stages, When symptomatic, patients will complain of sore throat, (referred) ear pain, or dysphagia, All tumors tend to grow submucosally, so endoscopic diagnosis frequently underestimates the extent of the tumor. Tumor volume evaluation is important as low-volume lesions «6,5cm3) without apex involvement can be managed effectively by radiotherapy, Early forms of hypopharyngeal carcinoma can be difficult to diagnose endoscopically because of their site of occurrence or submucosal growth characteristics, A typical situation is a cervical lymph node metastasis in which a negative endoscopy leads to classification as a carcinoma of unknown primary (C UP), Given an adequate examination technique with expanded piriform sinuses, early tumor stages can be detected with CT CT Morphology The T-stages of hypopharynx carcinomas are given in Table 8,22, Suspected early lesions should be interpreted with reference to endoscopic findings or oral contrast examination, A careful side-to-side comparison for asymmetries as an early sign of tumor is mandatory, Collapsed, asymmetric sinuses or unilateral mucus retention, however, can mimic tumors, When in doubt, another scan
HypopharyngeaJ Carcinomas Table 8.22 T-staging (UICC, 1997)
of
hypopharynx
carcinomas
Tis
Carcinoma in situ
T1
Tumor is limited to one subsite and 2 cm or less in greatest dimension
T2
Tumor involves more than one subsite or measures between 2 and 4cm in greatest dimension, without fixation of hemilarynx (vocal cord)
T3
Tumor greater than 4 cm in greatest dimension or with fixation of hemilarynx
T4
Tumor invades adjacent structures: thyroidl cricoid cartilage, carotid artery, soft tissues of neck, prevertebral muscles, thyroid gland, esophagus
with Valsalva maneuver should be performed to yield optimum expansion of the piriform sinuses. More advanced hypopharyngeal carcinomas show an asymmetric intraluminal growth pattern with associated mass effects
(displacement, compression, or deformation of adjacent structures) and possible signs of cartilage destruction or fatty infiltration. Hypopharyngeal tumors may compress the internal jugular vein, and tumors invading the vessel wall can lead to thrombosis. Invasion of the carotid artery is manifested by asymmetric luminal narrowing or occlusion. Invasion is also indicated by tumor encasing more than a 90 segment of the vessel wall and by circumferentialluminal narrowing. There are three important subsites of the hypopharynx: piriform sinus, postcricoid area, and posterior wall. Each subsite has its specific pattern of spread and lymphatic drainage. Piriform sinus: small tumors often involve one wall and can easily be missed at endoscopy. More often these tumors will infiltrate all walls of the sinus (Figs. 8.20, 8.21). Extension can be endolaryngeal in the paraglottic fat, via the posterior wall or postcricoid area to the esophageal junction, or into the apex of the sinus. Cartilage 0
Fig. 8.20 Hypopharyngeal carcinoma. Postcontrast CT demonstrates a collapsed piriform sinus with an enhancing, thickened wall (a). Virtual endoscopy (b) only demonstrates an asymmetry of the piriform sinuses. P= Piriform sinus, V = valieculae, E = epiglottis.
a
b
a
b
Fig. 8.21 Piriform sinus carcinoma. Postcontrast multislice CT shows a polypoid, moderately enhancing mass projecting far into the pharyngeal lumen (a). Note the multiple hyperenhancing bilateral lymph node metastases (arrows), some of which display central necrosis. Virtual laryngoscopy (b) demonstrates a mass in the piriform sinus. Anatomic orientation on such endoluminal views may be difficult.
257
8 Neck
invasion (thyroid, arytenoid) is seen frequently and can lead to extralaryngeal spread, Lymph node metastases occur in 55-75% of patients at presentation and involve mainly level 11, III and (less frequently) V nodes, Posterior wall: tumors here commonly extend into the oropharynx as well. The tumor may extend upward into the piriform sinus, caudal to the esophagus, or in unusual cases dorsally in the prevertebral muscles, Cartilage invasion
is uncommon
Supraglottic Tumors
Oncologically, the supraglottis is divided in five subsites: suprahyoid epiglottis. aryepiglottic folds (laryngeal aspect), arytenoids, infrahyoid epiglortts and false vocal cords (UICC),
for tumors in this loca-
tion, Lymph node invasion is seen early with retropharyngeal, level 11and III nodes affected, Postcricoid area: primary tumors in this location are rare, They will grow circumferentially in the submucosa, thereby narrowing the hypopharynx, The tumors can extend in the esophagus or in the posterior larynx with associated invasion of the cricoid cartilage, Lymph node metastases occur primarily in level III and (less frequently) 11or IV nodes, Important imaging features in hypopharyngeal tumors: • tumor volume
midline extension involvement of the apex of the piriform sinus
cartilage invasion • extension into esophageal verge or paraglotttc space • extralaryngeal spread • prevertebral
(contra-)indications are addressed in the section on the post-treatment neck,
muscle invasion.
CT Morphology
The T-stages of supraglottic larynx carcinomas are given in Table 8,23. Ventral tumors usually extend anteriorly in the preepiglottic fat Low lying and lateral tumors ofthe false cords, ventricles, or aryepiglottic folds primarily extend into the fat of the paraglottic space, Because of their submucosal growth, they are usually diagnosed in advanced stages, Posterior tumors of the (inter)arytenoid region infiltrate posteriorly in the postcricoid hypopharynx, Inferior extension is most important to decide whether supraglotttc laryngectomy is feasible and high quality coronal images with thin sections are important Tumor volume is important for nonsurgical management and low-volume « 6,Ocm3) tumors can be managed effectively by radiotherapy, Cartilage invasion, although infrequent, leads to an upgrade of the tumor and increases the risk of therapy failure, Lymph node spread is commonest in lesions involving the aryepiglottic folds and epiglottis. Spread is to the level II-IV nodes and nodal metastasis is seen frequently, in 30-50% of patients at the
Laryngeal Carcinomas
Almost all tumors are squamous cell carcinomas that are etiologically related to alcohol and tobacco use, Endoscopy and imaging are complementary in evaluating the extent of the tumors as they often extend submucosally Because these tumors may be surrounded by a peri tumoral inflammation zone, imaging with CT or MRI may lead to overestimation of tumor volume, The tumors are traditionally subdivided into supraglottic, glottic and subglotttc tumors, each with their own TNM-staging, pattern of spread and lymph node involvement The types of conservational surgery of the larynx and their
Table 8.23 (UICC,1997)
T-staging
of supraglottic
larynx carcinomas
Tis
Carcinoma
T1
Tumor confined to one subsite vocal cord mobility
T2
T umcr invades more than one adjacent subsite or supraglottis, glottis, or region outside supraglottis (mucosa of tongue base, vallecula, medial piriform sinus)
T3
Tumor
limited to
tion and/or
pre epiglottic T4
in situ
larynx with
with normal
vocal
invasion of postcricoid space tissues
cord fixaarea
or
Tumor extends through thyroid cartilage and/or extends into the soft tissues ofthe neck, thyroid and/or esophagus
Laryngeal Carcinomas time of presentation the primary).
(depending on the stage of
Important imaging features larynx tumors:
Table 8.24
T-staging of glottic larynx carcinomas
(UICC,1997)
in supraglottic
• tumor volume • distance of caudal edge of tumor to commissura anterior • trans glottic extension • invasion of piriform sinus or pre-epiglottic fat • exolaryngeal spread • cartilage invasion.
Tis
Carcinoma in situ
T1
Tumor limited to vocal cords (and commissures) with normal mobility
T1 a
Tumor limited to one vocal cord
T1 b
Tumor limited to both vocal cords
T2
Tumor extends to supraglottis and/or subglottis and/or with impaired cord mobility
T3
Tumor limited to larynx with vocal cord fixation
T4
Tumor invades through thyroid cartilage and/or to other tissue beyond the larynx: trachea, soft tissues of neck, thyroid, pharynx
Glottic Tumors
The subsites of the glottis are the true vocal cords and the anterior and posterior commissures. •
CT Morphology
The T-stages of glottic larynx carcinomas are given in Table 8.24. Vocal cord paralysis can be detected by CT scanning during shallow respiration. The affected vocal cord is not abducted and causes partial luminal obstruction. Glottic tumors typically arise in anterior part of the vocal cords and extend in the anterior commissure (Figs. 8.22, 8.23). From there they may spread into the supraglottic, subglottic or prelaryngeal regions. Lateral extension in the thyroarytenoid muscle may lead to vocal cord fixation, destruction of cartilage, and perilaryngeal spread (Fig. 8.24). Inferior spread to the
subglottis is also common and difficult to detect endoscopically. These extensions determine the type of larynx surgery to be performed. Tumor volume and cartilage status are important for nonsurgical therapy. Low volume tumors «3.5cm3) without cartilage sclerosis can be managed effectively by radiotherapy as well. Lymph node metastases are rare as long as the tumor is confined to the larynx, but become more frequent with extralaryngeal extension. Important imaging features in glottic larynx tumors: • tumor volume • craniocaudal extension (transglottic or subglottic) • involvement of anterior commissure • invasion of the deep paraglottic fat
Fig. 8.22 Small glottic carcinoma (T1) with nodular enhancement of the tumor on the vocal cord. Axial section (a) and coronal volumerendered image (b). a
b
8 Neck
Fig.8.23 Epiglottic carcinoma. The tumor has crossed the midline, has invaded the body and stem of the epiglottis, and is occupying most of the ipsilateral vallecula (arrow).
involvement of cricoarytenoid joints extension to the posterior commissure cartilage invasion.
Subglottic Tumors
True subglottic tumors are infrequent, and most often these represent inferior extension from a glottic lesion. •
Fig. 8.24 Advanced transglottic laryngeal carcinoma causing high-grade laryngeal obstruction and extensive destruction of the thyroid cartilage on the right side.
vanced stages. Cartilage invasion is less frequent than in other laryngeal tumors and may involve the cricoid or thyroid. Lymph node metastases are quite common and involve the level VI (Del phi an) , IV and upper mediastinal nodes. Important imaging larynx tumors '.
1
features
in subglottic
• inferior extent • cartilage invasion.
CT Morphology
The T-stages of subglottic larynx carcinomas are given in Table 8.25. True tumors extend inferiorly to the trachea and thyroid gland or posteriorly to the cervical esophagus. Direct extension through the cricothyroid membrane is seen only in more ad-
Table 8.25 T-staging of subglottic larynx carcinomas (UICC, 1997) Tis
Carcinoma in situ
T1
Tumor limited to the subglottis
T2
Tumor extends to vocal cord with normal or impaired mobility
T3
Tumor limited to larynx with vocal cord fixation
T4
Tumor invades through cricoid/thyroid cartilage and/or extends to other tissues beyond the larynx: trachea, soft tissues of neck, thyroid, pharynx
Cartilage Invasion
Invasion of the cartilage tends to occur at those sites where the perichondrium is interrupted by attachment of collagen fibers. These are located, for instance, at the anterior commissure, cricoarytenoid joint, cricothyroid membrane attachment, and along the thyroid laminae. Nonossified cartilage is relatively resistant to tumor infiltration compared to ossified cartilage. Ossification may precede actual tumor involvement through erosion via osteoclast activity. The diagnosis of tumor invasion with CT is difficult as the ossification pattern of laryngeal cartilage is very variable and tumor tissue has an attenuation similar to nonossified cartilage. Studies by Becker (1998)
Lymphoma have revealed four diagnostic signs of neoplastic invasion using CT (Fig. 8.25): • • • •
extralaryngeal tumor spread sclerosis of cartilage erosion of cartilage lysis of cartilage.
Sclerosis is a sensitive sign for tumor involvement but has limited specificity, especially in the thyroid cartilage. Erosion and lysis of cartilage adjacent to a tumor have a specificity of over 90% but a low sensitivity. This also holds true for extralaryngeal tumor, which is only seen in advanced stages. By combining these signs, a sensitivity and specificity of 80% can be reached with proper technique. State-of-the-art MRI however, may be the preferred modality due to a higher specificity and negative predictive value than CT. Lymphoma
Lymphomas can be classified as either Hodgkin's disease (HD) or non- Hodgkin's lymphoma (NHL) and constitute the second most common malignancy in the neck. In children HD is the most frequent, but overall 75% of cases are associated with NHL. Hodgkin's disease usually involves the lymph nodes, whereas non- Hodgkin's lymphoma frequently involves extranodal sites like the lymphatic tissue of Waldeyer's ring, salivary glands or thyroid.
Nodal Lymphoma
Most patients present with enlarged nodes at the level III -V nodal stations, but superficial nodes or higher level nodes can also be involved. The disease can be either unilateral or bilateral. HD shows contiguous spread between nodal stations, whereas in NHL the spread is often non-contiguous. Necrosis and extracapsular extension in primary disease is infrequent, but can be seen following therapy. It is more frequent in NHL. The nodes rarely calcify unless patients had previous radiation or chemotherapy and are most prominent with sclerosing forms of Hodgkin's disease. Nodal lymphoma with associated nasopharyngeal lymphoid hyperplasia or lymphoepithelial cysts in the parotid is suggestive of HIV infection. In these cases extranodal NHL is often present. •
CT Morphology
Unilateral or bilateral enlarged lymph nodes with a homogenous density are seen. The nodes are often round and clustered. After contrast administration slow and homogeneous enhancement is present. Necrosis or calcification can be seen occasionally in patients with a history of radiation or chemotherapy. Differentiation from nodal metastasis is often impossible. Extranodal Lymphatic Lymphoma
In the neck this form is the most common presentation of extranodal NHL. It may be present in the lymphatic tissue of the adenoids and tonsils as an asymmetrical bulky mass with no infiltration of the surrounding structures or bone ero-
I a Extralaryngeal tumor spread Fig. 8.25 Criteria for CT diagnosis of neoplastic cartilage invasion (Becker, 1997) (a) extra laryngeal tumor spread; (b) sclerosis; (c) lysis (top) and erosion.
b Sclerosis
lll = tumor
o
=
nonossified cartilage
c Lysis and Erosion
0 •
= =
ossified cartilage sclerosis
261
8 Neck
sion. Often. concomitant nodal disease is present. Contrast enhancement is moderate and necrosis is absent. Differentiation of the mass from squamous cell carcinoma is usually not possible. but affected nodes in uncommon locations for SCCa may suggest the diagnosis. In young patients it is also indistinguishable from infectious mononucleosis. In the nasopharynx. NHL masses lack septa that are present in lymphoid hyperplasia {best seen on fat-suppressed Cd-enhanced MRI).
Extranodal Extralymphatic Lymphoma
Parotid gland lymphoma is rare. The gland is involved with either Hodgkin's disease or nonHodgkin lymphoma. There are associations with Sjogren's syndrome, HIV and Waldenstrom's disease. At CT, the disease is most often confined to homogeneous intraparotid nodes that enhance moderately. True extranodal disease is exceedingly rare and can show an ill-defined infiltration of (part of) the gland. Concomitant nodal disease is frequent. Thyroid lymphoma is seen in up to 10% of patients, usually middle-aged or elderly women. Over 80% of patients have a history of Hashimoto thyroiditis, but the two conditions cannot be separated by imaging. Regional adenopathy is seen frequently. Other locations for NHL include the nasal cavity, paranasal sinus, orbit and skin. The presence of multiple masses in extralymphatic sites is highly suggestive ofNHL.
Malignant
Mesenchymal
and Neurogenic
Sarcoma
All sarcomas are rare in the head and neck and require surgical biopsy for definitive diagnosis.
Post-transplantation Disorder (PTLD)
Lymphoproliferative
Post-transplantation lymphoproliferative disorder is a serious complication in patients after solid organ transplantation. It encompasses a spectrum of disorders that range from lymphoid hyperplasia to (non-Hodgkin's) lymphoma. It represents an unregulated B-cell proliferation and the development is associated with EpsteinBarr virus infection. The disease occurs in up to \0% of transplant recipients and depends on the organ transplanted and the immunosuppressive regimens employed. Most often it involves the lung and gastrointestinal system. The neck is an uncommon site for PTLD. It can manifest as focal masses in Waldeyer's Ring or as lymphadenopathy. cr Morphology In the lymphoid tissue of the nasopharynx and oropharynx focal soft tissue masses can be seen that can extend in the parapharyngeal space submucosally. Central necrosis is often present, and there is a variable amount of contrast enhancement. Adenopathy can manifest as a large soft tissue mass or as excessive numbers of normal-sized nodes that appear in clusters. CT is valuable in diagnosis and for monitoring response to treatment.
Tumors On CT the density of the fat is often higher than subcutaneous fat and enhancement may be seen in the soft tissue parts.
Fibrosarcoma
Liposarcoma The neck is a rare location for liposarcomas, but they may arise in the anterior and posterior cervical spaces in patients 40-60 years old. They are often infiltrative and inhomogeneous with fat and varying amounts of soft tissue elements.
The neck is a common location for fibrosarcoma. It presents in adults with a slow-growing painless mass that may reach a considerable size before showing symptoms. Metastases are very common (>60%). The infantile form is much more benign and has a low incidence of
metastasis. CTshows a homogeneous soft-tissue mass that may erode bone. Calcifications. necrosis. and hemorrhage can be present.
Chondrosarcoma In the neck. chondrosarcoma most often originates in the chondroid larynx skeleton. The thyroid and cricoid are most commonly involved. Other locations are the masticator space and skull base. It is usually well defined with submucosal growth. but shows irregular. ring-like calcifications on CT. The airway may be compromised in laryngeal tumors.
Rhabdomyosarcoma This is the commonest sarcoma in children. usually occurring before the 10th year. The nasopharynx. parapharyngeal space. or infratemporal fossa are involved in 30% of cases. The tumor frequently extends through the skull base into the cavernous sinus. Lymph node. lung. and bone metastases are seen in up to 50% of patients. At CT. the tumors present as homogeneous. infiltrative bulky masses with soft tissue density and variable enhancement. There is often bony erosion or remodeling of the skull base or maxillary sinus.
Synovial Sarcoma
in the retropharyngeal space and patients present with hoarseness and dysphagia. The tumors show slow, infiltrative growth and frequently metastasize to the lung. CT shows homogeneous or heterogeneous lesions that often have welldefined margins. Fluid - fluid levels, hemorrhage, and calcifications are common. Malignant Peripheral Nerve Sheath Tumor (MPNST) MPNST is the unified name for malignant neurogenic tumors and describes a highly malignant sarcoma. These sarcomas may arise de novo or from malignant transformation of a neurofibroma or schwannoma, usually a plexiform neurofibroma. Some 65% show distant metastases. They are commonest in patients with neurofibromatosis I and following radiation (latency period 10-20years), but are uncommon in the head and neck. CT is nonspecific and shows a large, fusiform, invasive tumor with necrosis and hemorrhage. Enhancement is heterogeneous.
Malignant Fibrous Histiocytoma The neck is a rare site for this tumor. CT findings are a circumscribed mass of homogeneous softtissue density, often within muscle or along fasciae. Small foci of calcification, necrosis, or hemorrhage may be present.
The neck is the most usual site of this tumor outside the extremities. Lesions are commonest
Salivary Gland
Pathology
Cysts
Retention cysts develop secondarily to an incomplete or intermittent obstruction caused by inflammatory stricture, calculus, surgery, or trauma to the salivary ducts. They have an epithelial lining and most often occur in the submandibular gland. A ranula is a mucous retention cyst of the sublingual gland. The simple form is a true cyst limited to the sublingual space. The plunging or
diving ranula develops from a rupture of a simple ranula and is in fact a mucocele or pseudocyst without an epithelial lining. It extends over the posterior margin of the mylohyoid muscle into the submandibular space. Lymphoepithelial cysts are associated with AIDS and often multicentric and bilateral. They are usually located in the parotid glands and associated with cervical adenopathy.
8 Neck
and have and lack more illtail into
rhage, or calcification, When left untreated, 25% of the tumors will develop malignant transformation (usually adenocarcinoma), When incompletely resected, the tumor can recur in up to 50% of cases within lOyears after surgery, MRI is the preferred modality in these cases,
Between 70% and 80% of parotid, 40-60% of submandibular, 15-30% of sublingual and 20-40% of minor gland tumors are benign, Imaging is unreliable in giving a definite diagnosis but should be used to select therapy and serve as a roadmap for surgery, Table 8,26 gives an overview of solid lesions in the parotid gland,
Small tumors are homogeneously hyperattenuating and well defined on non enhanced studies, With increasing size the masses are more inhomogeneous with hyperattenuating areas of hemorrhage or calcification andlor more hypodense regions of necrosis and cystic degeneration, Enhancement after contrast injection increases (Fig, 8,26) and becomes more homogeneous over time, and is best evaluated on delayed scans,
CT Morphology
These cysts are smooth, well defined fluid attenuation, The walls are thin calcifications, A diving ranula can be defined and often has a characteristic the sublingual space,
CT Morphology
Benign Tumors
Pleomorphic Adenoma Pleomorphic adenoma or benign mixed tumor is the commonest and represents 70-80% of all benign solid masses in the salivary glands, The majority occurs in the parotid, usually lateral to the plane of the facial nerve (superficial lobe), They are usually solitary, well-defined and ovoidshaped, Larger tumors may have a lobulated contour and may show areas of necrosis, hemor-
Table 8.26
Differential diagnosis of parotid gland
tumors Benign
tumors
Pleomorphic adenoma Warth in's tumor"
Oncocytoma Lipoma Facial nerve Malignant tumors
schwannoma
Mucoepidermoid
lymphoma"
Acinar cell carcinoma Malignant mixed tumor Squamous cell carcinoma Metastases
Melanoma' Squamous
Inflammatory
cell carcinoma
Abscess Lymphoepithelial
"9 Sarcoidosis' .• May
CT Morphology
These are small, ovoid, well-defined masses often arising in the tail of the superficial lobe, Small tumors are homogeneous, well-delineated masses that can be multifocal or bilateral. In larger tumors, cyst formation is seen frequently, These cysts may have focal tumor nodules in their walls, Contrast enhancement is inhomogeneous, The differential diagnosis for noncystic lesions includes lymphoma, adenopathy, and granulomatous disease, and, for cystic lesions, cavitating metastases, and lymphoepithelial cysts,
carcinoma
Adenoid cystic carcinoma Non-Hodgkin
Warth in's Tumor Warthin's tumor or cystadenoma lymphomatosum is a monomorphic adenoma of the parotid gland accounting for 2-10% of benign tumors, It is the commonest lesion to occur as a multifocal or bilateral (10%) disease, The lesion is rare in non-whites,
be multifocal
cysts
of skin' (In AIDS)';
Oncocytomas
n1g~~del'n
Oncocyto~~~gl:FedU~~~~ymln~ patients over 55 years old, They occur preferentially in the parotid gland, are slow growing, and are well circumscribed,
Malignant Tumors
a
c
b
Fig. 8.26 Pleomorphic adenoma of the deep lobe of the parotid gland ("iceberg tumor") (a). Hypervascularized mass medial to the right mandible displacing the origin of the pterygoid muscles and the soft palate. Bilateral fatty infiltration of the parotid glands (arrows) is
a common finding. A pleomorphic adenoma at the lower pole of the right parotid gland (b, c) simulates a lymph node. There was a typical echo-free appearance on ultrasound with posterior echo enhancement. Note the dental artifacts on the coronal view (c, arrowheads).
•
Table 8.27 1997)
CT Morphology
The imaging characteristics are nonspecific and mimic both pleomorphic adenomas and Warthin's tumor. Malignant Tumors
Malignant tumors are more frequent in smaller salivary glands and occur more often in Inuit Eskimos and in children. Pain, nodal metastases, and facial nerve paralysis are associated with poor prognosis. Staging is based upon the TNMclassification (TableS.27). Mucoepidermoid Carcinomas
Mucoepidermoid carcinomas are the commonest salivary malignancies in adults and children. They represent 30% of parotid malignancies and occur most often in the parotid or minor salivary glands. They can be oflow, intermediate, or high grade. •
CT Morphology
Low-grade tumors are often well described while the high-grade tumors are diffusely infiltrative. Cyst formation is frequent in lower grade tumors, and hemorrhage or necrosis may be present. At imaging the morphology varies with the histological grade and as such may mimic pleomorphic adenoma or squamous cell carcinoma.
T-staging of salivary
gland tumors
(UICC,
T1
Tumor 2cm or less (greatest dimension) withoutextraparenchymal extension
T2
Tumor greater than 2 cm but less than 4 cm without extra parenchymal extension
T3
Tumor greater than 4 cm but less than 6 cm ortumorwith extra parenchymal extension without facial nerve involvement
T4
Tumor greater than 6 cm or tumor invades facial nerve or skull base
Adenoid Cystic Carcinomas
This tumor occurs most often in the minor salivary glands of the palate, in the parotid or submandibular gland of patients in the fifth or sixth decade. It is the commonest malignancy in the submandibular gland. These tumors grow slowly and have a tendency to show perineural invasion that can lead to pain as well as spread along the facial or mandibular nerve. Cd-enhanced MRI is preferable to CT for evaluating cell ularity and perineural extension. •
CT Morphology
Tumors show a more aggressive growth in the minor salivary glands than in other glands where their aspect is more benign, depending on the subtype and cellularity. Tumors in the minor salivary glands arise in the mucosa and
265
8 Neck
therefore cannot be morphologically differentiated from pharyngeal carcinomas. The imaging characteristics are compatible with their histological grade.
Carcinoma ex Pleomorphic Adenoma Carcinoma ex pleomorphic adenoma is a malignant transformation of a (resected) pleomorphic adenoma.
cr
Morphology
Small tumors resemble pleomorphic adenoma with benign characteristics. but larger tumors are more inhomogeneous and infiltrative with areas of necrosis. The CT morphology directly reflects this variation in macroscopy.
Acinic Cell Carcinoma Acinic cell carcinoma occurs almost exclusively in the parotid gland and can be bilateral in 5% of cases. They usually occur after the 40th year but also are the second commonest salivary malignancy in children. CT Morphology
The tumors can either be solid or cystic. The imaging characteristics are nonspecific. The tumors generally have a benign appearance.
Metastases Metastases occur almost exclusively in (the intraglandular nodes of) the parotid gland. The commonest primary is melanoma of the temporal skin or other parts of the neck. Less frequent are head and neck. renal, lung, breast, and Gl carcinomas.
cr
Morphology
Unilateral or bilateral, enlarged intraglandular lymph nodes with a homogeneous density. After contrast administration homogeneous enhancement is seen. Necrosis can be seen occasionally, and is suggestive for malignancy. Differentiation from nodal lymphoma is often impossible.
Inflammatory and Autoimmune Lesions Sial adenitis Acute bacterial and viral sialadenitis commonest pathologies of the salivary Bacterial infections usually ascend from cavity when saliva flow is decreased and pathogens like Staphylococcus aureus,
are the glands. the oral involve
Streptococcus pyogenes. Streptococcus pneumoniae and Haemophilus influenzae. Viral infections can be
caused by mumps, Coxsackie or (Para)influenza viruses and preferentially involve the parotid glands. Intra- and periglandular lymph nodes may be involved in the process. The main role for imaging is to differentiate adenitis from abscess formation. Abscesses are often formed when acute adenitis is undiagnosed or incompletely treated. These abscesses can spread quickly in the para pharyngeal space or other parts of the neck. CT Morphology
The involved gland is dense and somewhat enlarged. It enhances slightly. Abscesses are demarcated regions of low attenuation.
Sialolithiasis Stones can be located in the salivary gland ducts or the extraglandular ductal system. From 80-90% occur in the submandibular gland, W-20% in the parotid and <5% in the sublingual glands. The stones are usually solitary and most often located in the extraglandular ducts. Complete obstruction will lead to gland atrophy while incomplete obstructions are often associated with recurrent infections and swelling. The main role of CT lies in differentiating obstructive from nonobstructive diseases. This directly affects the choice of management. CT Morphology
Sialoliths are readily identified at CT, and even noncalcified stones can be detected. Some 80% of submandibular stones and 60% of parotid stones are calcified. The gland is often enlarged and more ill defined. After contrast there is increased enhancement of the gland on the affected side. In case of obstruction, hypoattenuating dilated ducts can be visualized (Fig. 8.27).
Granulomatous Diseases Fig. 8.27 Inflammation leads to hypervascularization of the affected left submandibular gland (arrows). Note the dilated duct (arrowhead) on the coronal volume-rendered image (a). A curved planar reformation is able to display the whole length of the dilated duct (b). Note that there is no stone visible, which suggests a stricture, e.g., after spontaneous passage of a sialolith.
a
Sialadenosis
This is a noninflammatory and nontender chronic enlargement of the parotid glands that can lead to xerostomia. It seems to be caused by a degeneration of the autonomic nervous system and is associated with a variety of endocrine and nutritional diseases, the most important being diabetes mellitus. It can also be associated with medications, in which case also the submandibular glands can be involved. •
CT Morphology
The CT appearance is nonspecific. The glands are enlarged and appear either dense or infiltrated by fat. Sjogren's syndrome
This systemic autoimmune disorder can occur alone or in association with connective tissue diseases like rheumatoid arthritis. Classically, symptoms are keratoconjunctivitis sicca, xerostomia and connective tissue disease. The adult form is much commoner than the childhood form and often proceeds to more advanced stages. It is mainly seen in women 40-60 years old and is associated with parotid enlargement. The peripheral ductuli and acini are involved first and superinfection may lead to gland destruction and multiple abscesses. There is a 40-fold enhanced risk of subsequent development of lymphoma in the gland.
b
•
CT Morphology
In early stages of the disease the gland appears normal. With progression of disease, the gland enlarges and becomes denser. Later, a honeycomb glandular pattern may develop, which can also be seen in granulomatous diseases or chroni csialadenitis. Contrast can be retained in small punctate or globular collections. Rarely, a macrocystic change can occur resembling HIV -associated lymphoepithelial cysts. Granulomatous Diseases
These systemic diseases can also affect the intraparotid or periglandular lymph nodes. CT usually shows multiple benign appearing masses throughout the gland that may mimic lymphoma or other granulomatous diseases. Diffuse adenopathy is frequent. Sarcoidosis affects the parotids in 10-30% of patients leading to a chronic bilateral nontender multinodular enlargement. Often there is associated adenopathy. When combined with uveitis and facial nerve paralysis this condition is known as Heetiottit's syndrome. Tuberculosis or atypical mycobacterial infections are rare and usually arise from a focus in the tonsils. The parotid and submandibular glands may be affected. The presentation can be acute or more insidious and both clinical and CT findings are nonspecific and can mimic infectious sialadenitis or tumor. Other causes include cat-scratch fever, toxoplasmosis and actinomycosis. In the last two
261
S Neck
forms the disease can be associated with more diffuse lymphadenopathy. In actinomyces infec-
tions the soft tissues (masticator space) are also infected and sinus tracts are formed.
Thyroid and Parathyroid Gland Pathology
Primary diagnostic methods in the thyroid are serology, scintigraphy, and ultrasound. However, on CT examinations for other indications one may come across coincidental findings in the thyroid. CT is also indicated in the evaluation oflarge thyroid tumors, especially when malignancy is suspected. Iodinated contrast material should only be given after TSH-values have been assessed, as hyperthyroid patients may develop long-standing iodine-induced hyperthyroidism (IIHT). The use ofiodinated contrast agents is also contraindicated when patients are scheduled for ablative therapy with radioactive iodine (see also Chapter 3). In these cases, MRI is the modality of choice. Goiter
Goiter is a benign enlargement of the thyroid that is often asymmetric. It can be simple, associated with iodine deficiency, or multinodular, associated with autoimmune disorders or dietary factors. The nodules in multinodular goiter represent either colloid nodules or true follicular adenomas of variable sizes. Patients can be hypothyreotic, euthyreotic, or hyperthyreotic. Multinodular forms can become bulky, extending in the anterior mediastinum. The frequency of carcinoma developing in m ul tinodular goiter is 4-7%. •
Fig. 8.28 Unilateral goiter causing enlargement of the left lobe of the thyroid gland and displacing the trachea to the right side. Hypodense areas within the enlarged left lobe and a small calcification are present, signifying regressive changes.
H
CT Morphology
These cysts are smooth and well defined. Their attenuation depends on the cyst content and varies from hypodense to isodense. The walls are thin and lack calcifications. Large cysts may be hard to identify as arising from the thyroid but hyperattenuating (compressed) thyroid tissue in the wall of these large cysts is typical (Fig. 8.29).
CT Morphology
Mixed solid and cystic zones are present in a nodular and enlarged gland. Hemorrhages and calcifications may be present (Fig. 8.28). Enhancement is patchy and inhomogeneous. The trachea, esophagus or neck vessels can be displaced or compressed. Thyroid Cyst
These are the result of degeneration of adenomas and are usually well-defined lesions, often with fluid attenuation. With elevated protein content, hemorrhage, or infection, these lesions have a higher attenuation.
Benign Thyroid Tumors
Thyroid nodules are very common and occur in 4-7% of the adult population. Colloid {adenomatous} nodules are composed of hyperplastic epithelium and are not true neoplasms. They are usually non- functioning and may be cystic (colloid cyst). Follicular adenomas present as slow growing well-defined encapsulated nodules that can be functioning or non-functioning. They are usually solitary and measure up to 3 cm. They may undergo cystic degeneration, calcification, or hemorrhage. Although benign adenomas do not undergo malignant transformation, "10-'15% of all single (non-functioning)
Malignant
b
a
Thyroid Tumors
c
Fig. 8.29 Large thyroid cyst that leads to displacement and extrinsic compression of the larynx (compare to Fig. 8.11 a). It can be identified due to the hyperenhancing rim that is isoattenuating to thyroid tissue and is
contiguous with the thyroid gland. Axial coronal (b) and sagittal volume-rendered from a thin-section multislice CT data set.
adenomas are malignant. The Hiuthle cell subtype is more variable in size and shape and has less well-defined borders.
Papillcny carcinoma is the commonest thyroid malignancy accounting for 55-75% of cases, especially in young or middle-aged adult females. They may be multifocal or bilateral and can have cystic, hemorrhagic, or calcified components. Although these tumors grow slowly, 50% of patients have metastases in the 10coregional lymph nodes at diagnosis (Table 8.28). Follicular carcinoma is most frequently seen in young female adults and may be either well encapsulated or diffusely infiltrative. It frequently invades vessels. Cystic degeneration is rare. Spread to nodes is infrequent « 10%) but hematogenous dissemination to lung or bones occurs early. Anaplastic carcinoma occurs more frequently in the elderly, especially in people with longstanding goiter. These tumors are often densely calcified and show necrosis. The
•
CT Morphology
Follicular adenomas are usually well defined, relatively hypodense solid masses on noncontrast scans. Cyst formation and calcification can often be seen. Some may show a marked enhancement with iodinated contrast. It is virtually impossible to differentiate benign tumorous nodules from a nodular goiter. Malignant Thyroid Tumors
The commoner malignancies of the thyroid are usually well differentiated. Risk factors include exposure to ionizing radiation during childhood and prolonged increased levels ofTSH. Table 8.28
image (a), images (c)
Thyroid carcinomas
PTC
FTC
ATC
MTe
Incidence
55-75%
15-20%
8-15%
2-8%
Iodine-storing
+ +
Metastases
Regional LN Lungs
Lungs Bone
Regional
Regional LN Lungs Liver Bone
PTC = papillary thyroid carcinoma, FTC = follicular thyroid carcinoma, ATC MTC = medullary thyroid carcinoma, LN = lymph nodes
= anaplastic
thyroid carcinoma,
269
8 Neck Table 8.29 1997)
TN-staging
of thyroid
carcinoma
(UICC, •
CT Morphology
T1
Tumor 1 cm or less in greatest dimension, limited to thyroid
T2
Tumor between 1 cm and 4 cm in greatest dimension, limited to thyroid
T3
Tumor more than 4 cm in greatest dimension, limited to thyroid
T4
Tumor of any size extending beyond the capsule
NO
No regional lymph node metastasis
N1a
Regional metastasis in ipsilateral cervical lymph nodes
The T-stages of thyroid carcinomas are given in Table 8.29. The pattern of spread of the various tumor types is provided in Table 8.28. Imaging findings are nonspecific. There is much overlap between benign and malignant tumors. Thyroid carcinomas are well- to ill-defined lesions, often with inhomogeneous contrast enhancement. Areas of hemorrhage, necrosis, calcification, and cystic degeneration are often present. Aggressive carcinomas show extrathyroid spread or vascular invasion.
N1 b
Regional metastasis in bilateral, midline or contralateral cervical or mediastinal lymph nodes
Parathyroid Tumors
tumor is aggressive and invades neighboring structures commonly. It metastasizes early, with 70-80% of patients having affected nodes, which are often necrotic. Medullary carcinoma is rare. It has a familial incidence and can be part of MEN -syndromes: MEN IIa associated with pheochromocytoma and parathyroid adenoma and MEN lib associated with mucosal neuroma and marfanoid facies. About 50% of patients have locoregional metastasis in the lymph nodes and distant metastases are common. Metastases to the thyroid are rare and often clinically occult. The commonest primaries are bronchogeniC and renal cell carcinomas. The lesions are usually multiple and sometimes hemorrhagic.
Adenomas of the parathyroid can be found along the tracheoesophageal groove anywhere from the upper neck to the mediastinum, but frequently lie posterior to the thyroid poles. In most cases they are associated with hyperparathyroidism. Multiple adenomas can be associated with MEN I and MEN IIa syndromes. Carcinomas are very rare and usually cause hyperparathyroidism. They are slowly growing tumors that may mimic adenomas but grow locally invasive. They spread via lymphatics and hematogenously in 30% of cases. •
CT Morphology
An adenoma is a well-defined soft tissue lesion that is hypodense relative to the thyroid (Fig. 8.30). Some 25% of adenomas enhance. Sensitivity ofCT is however only 60-70%. The diagnosis using ""Tc-sestamibi scintigraphy is much more sensitive. Carcinomas can only be differentiated if local invasion or metastases are visible.
Fig. 8.30 Parathyroid adenomas may appear as a hypodense expansile mass on the posterolateral aspect of the thyroid gland (a), showing a marked attenuation difference to the thyroid tissue (after IV contrast administration). Ectopic parathyroid adenoma lateral to the aortic arch (b).
a
b
Cellulitis
Inflammatory
Thyroid
Lesions
Acute infective (suppurative) thyroiditis is a rare cause for swelling with pain, fever and dysphagia. Commonest pathogens are Streptococcus hemolyticus, Staphylococcus aureus and Streptococcus pneumoniae. It can be associated with a fourth branchial cleft anomaly and may be the result of piriform sinus or thyroglossal duct fistulae. Hashimoto's [lymphocytic) thyroiditis is a chronic autoimmune disorder with inflammation that leads to a symmetric nodular enlargement of the gland. It occurs in middle-aged women with hypothyroidism and predisposes to non- Hodgkin's lymphoma. It is associated with Graves' disease and other autoimmune disorders.
and Abscess
Riedel's thyroiditis is a variant chronic flammation giving extensive fibrosis leading hypodense, enlarged gland at CT. It may press the esophagus and is associated with diastinal and retroperitoneal fibrosis. Slight trast enhancement may be seen.
into a commecon-
De Quervain :s {granuloma taus} thyroiditis is a
subacute disease of middle-aged women. It is associated with viral infections ofthe upper respiratory tract. Patients may progress from hyperthyroidism through hypothyroidism to euthyroidism in 1-2 months. At CT the gland is symmetrically enlarged or may show a dominant mass.
Infectious Diseases Cellulitis
and Abscess
Most localized infections are treated by antibiotics. CT is only required in nonresponsive cases, and can differentiate cellulitis from suppurative adenitis with edema and abscess formation. Dental infections: Infections from dental origin are common and may spread to the sublingual/submandibular, parapharyngeal, or masticator spaces. Abscess formation is seen frequently and there may be cortical erosion of adjacent bone. Miicosal lesions: Infections of the nasopharynx, oropharynx, tonsils, and middle ear can extend into the retropharyngeal space via spread to the retropharyngeallymph nodes. Cellulitis or suppurative adenitis may occur and associated swelling may cause bowing of the pharyngeal wall. Abscesses can occur locally ([perij tonsillar) or as a complication of an abscessed adenitis (retropharyngeal). In this age of broad-spectrum antibiotics, retropharyngeal abscess formation is more often caused by penetrating trauma, diskitis, or surgery. In the larynx, acute supraglottitis can spread locally to involve the glottic or subglottic larynx and hypopharynx and may cause life-threatening swelling of these structures. Ludwig's angina is an extensive, acute cellulitis of the floor of the mouth that can be life threatening. It is generally seen in middle-aged men with infected molars. Streptococci or
staphylococci are the causative agents and the infection involves the sublingual and submandibular spaces bilaterally. Abscess formation is infrequent. If not controlled, rapid spread to the suprahyoid soft tissues can ensue. Cellulitis can easily spread to adjacent facial spaces or extend along fascial planes, upward or downward. It may extend into the mediastinum via the visceral and posterior cervical spaces. Abscesses conform to fascial spaces and can occur anywhere in the neck. Localized abscess collections may give a clue to their origin by their location. Extensive infections may be complicated by venous thrombosis, osteomyelitis, and airway narrowing from mass effect. CT Morphology
Cellulitis may present
with a soft tissue mass showing marked contrast enhancement. There is infiltration and edema with thickening of the skin, obliteration of normal fat planes and irregular thickening offasciae (Fig. 8.31 a). Gas inclusions are indicative of infection with gas-producing bacteria (Fig. 8.31 b). Myositis may cause enlargement of muscles. Infective (suppurative) adenitis is characterized by enlarged lymph nodes with a low-attenuation center, suggestive of liquefaction necrosis. Surrounding edema leads to a smooth expansion of the structures within the space with displacement of the pharyngeal wall anteriorly. Abscesses are masses with low attenua
271
8 Neck
a
c
b
Fig. 8.31 (a) Cervical cellulitis appears as an inflammatory mass that obscures the sternocleidomastoid muscle and the intrinsic neck muscles, (b) Gas-forming cellulitic inflammation of the cervical soft tissues on the right side, with significant soft-tissue swelling and mycotic aneurysm of the common carotid artery due to salmonella sepsis. The inflammatory process shows pe-
ripheral enhancement (arrow), (c) Incipient parapharyngeal abscess (arrow) contains a very small area of liquefaction partially surrounded by an enhancing wall. There is significant concomitant soft-tissue swelling with slight displacement and narrowing of the pharyngeal lumen.
tion that show enhancing peripheral capsules (Fig. 8.31 c). Abscesses may contain air and often show surrounding edema. Severe complications of neck infections include septic jugular vein thrombosis with or without distant embolization and mediastinitis (see Fig. 8.32). Arterial rupture is a very severe complication of disseminated neck infection. It is preceded by small hemorrhages with hematoma formation and has a protracted course. It usually affects the internal carotid artery.
changes. Infections with atypical mycobacteria are more often seen in children and usually involve a more localized, unilateral lymphadenopathy, commonly in the upper cervical nodes.
Tuberculous
Adenitis (Scrofula)
This accounts for 5% of cases of cervical lymphadenopathy. The disease is caused by Mycobacterium tuberculosis or atypical mycobacteria, is usually painless and seen especially in young adults. In tuberculosis, it mayor may not be a manifestation of a systemic disease. In AIDS-patients there may be co-existence of tuberculous adenitis and Kaposi's sarcoma. •
CT Morphology
Nodes can be homogeneous or show central necrosis with rim enhancement or calcification. A multilocular complex of nodes is a specific sign. The posterior triangle or jugular nodes are involved without surrounding inflammatory
Necrotizing Fasciitis
Necrotizing fasciitis causes a rapidly progressive infection of subcutaneous tissue and deep fascia. It can be caused by anaerobes or streptococci and leads to widespread necrosis, especially in immunocompromised patients. •
CT Morphology
Gas may be present, but otherwise distinction from cellulitis is difficult. Miscellaneous Diseases with Adenopathy
Infectious mononucleosis is primarily caused by the Epstein-Barr virus. It is associated with lymphadenopathy, and most often the posterior cervical nodes are involved. Hyperplasia of the lymphatic tissue ofWaldeyer's ring is common. Castleman's disease (angiofollicular lymphnode hyperplasia) originates in the mediastinum but can be multicentric in W% of cases. Most often this affects male patients in their
Arterial Thrombosis and Dissection 50 s or 60 s. The nodes show an early dense uniform enhancement following contrast administration. Sarcoidosis gives rise to diffuse adenopathy. The nodes are homogenous and do not show necrosis. There is often involvement of Waldcyer's ring and the salivary glands. Cat-scratch disease is a granulomatous infection in young patients caused by Bartonella benselae. It can be associated with painful cervical adenopathy. Usually one node group is involved, but multiple sites can be found in disseminated disease (in 5-'10%). Kimura and Klkucbi disease are rare causes for cervical adenopathy in oriental patients. Kimura occurs in young males and shows enhancing nodes and salivary gland infiltration.
Kikuchi is more frequent in young females, showing a necrotizing lymphadenopathy. Fungal Infections
Histoplasmosis is ubiquitous in the central US. Rarely, it may show granulomatous masses in the larynx, oropharynx, or trachea. Cryptococcosis is usually seen in immunocomprornised patients that in rare cases may show a neck mass originating from suppurative adenitis. In South-America, acute or chronic paracoccidioidomycosis (South American blastomycosis) can be encountered which shows mucosal infection of the oral cavity or lymphadenopathy.
Vascular Lesions Jugular Venous Thrombosis
Thrombosis of the internal jugular vein can be the result of infection, obstruction, advanced tumors, IV drug abuse, or sinus thrombosis, postoperative or after insertion of central venous catheters. •
CT Morphology
A fresh thrombus is hyperattenuating. The acute thrombophlebitic phase shows an enlarged thrombus- filled vein with loss of adjacent soft tissue planes and enhancing vasa vasorum (Fig. 8.32). This is often associated with fluid (venous transudate) in the retropharyngeal space. In the thrombosed phase perivascular changes have resolved, but enlarged collaterals are shown.
•
CT Morphology
Thrombosis: The imaging features are comparable to venous thrombosis Dissections: CTA can show an endoluminal membrane separating the false (often bigger) and true lumen (often smaller). If one lumen is thrombosed, the only sign of dissection may be a long narrowing of the artery (Fig. 8.33).
Arterial Thrombosis and Dissection
Thrombosis is often associated with pseudoaneurysm formation and leads to ischemic symptoms or Homer's syndrome. Dissections of the large arteries of the neck can be post-traumatic, part of an aortic dissection, or spontaneous. They are an important cause of ischemic cerebral lesions in young patients.
Fig. 8.32 Fresh thrombosis of the internal jugular vein (arrow). Note the inflammation in the anterior parapharyngealspace.
273
274
8 Neck Fig. 8.33 Traumatic dissection of the carotid artery in a young man following a motor vehicle accident. The false channel is completely thrombosed (a) making a diagnosis from axial sections almost impossible. There is, however, occlusion of the medial cerebral artery on the right side, seen on coronal thin-slab MIP from a multislice CT data set (b). The distal portion of the internal carotid artery (arrow) and the A 1 segment of the anterior cerebral artery are stenosed. Note the excellent collateral supply of the right-sided cortex.
b
a
Trauma
Spiral CT has become the primary investigation for important soft tissue trauma and can be used for soft tissue, vascular, pharyngeal, and laryngeal lesions.
aneurysms. Due to a higher incidence of spinal lesions, vertebral artery lesions are more frequent than carotid lesions, with most lesions occurring in the C1-C2 and C6-C7 segments. These patients are often asymptomatic.
Edema and Hematoma
•
In traumatic lesions of the cervical spine or associated ligaments, prevertebral space edema or hematomas may develop that can extend up or downward. Retropharyngeal hematomas are much less frequent, but can result in significant airway compromise. Air can dissect retropharyngeally after laryngeal trauma or barotrauma from assisted ventilation after intubation. Vascular Injuries
Vascular injuries are most frequently caused by penetrating trauma such as stab or shot wounds and are preferentially diagnosed by CT angiography. The source images can also visualize the trajectory of the penetrating injury in the soft tissues. Penetrating injuries lead most frequently to carotid occlusion (35%), pseudoaneurysm formation (35%) and much less frequently to arterial dissections. Vertebral artery in] ury is usually associated with spinal fractures. Blunt trauma leads to stretching of vessels and is more often associated with occlusions (35%) and dissections (35%) than pseudo-
CT Morphology
Direct findings of vessel injury include irregular margins, contrast extravasation, caliber changes, or lack of enhancement. Indirect signs like carotid space hematoma or bone or metal fragments close to vessels can be evaluated at the same time. In blunt trauma, there are frequently signs of vertebral fractures. Laryngotracheal
Injuries
These injuries are frequently a complication of penetrating trauma and seen only seldom after blunt trauma. The mortality rate of these injuries, especially major dislocations, is high. CT provides the best evaluation of the extent of fractures, dislocations and hematomas. •
CT Morphology
Fractures: cricoid ring fractures are often m ul tipIe with the anterior fragment collapsed posteriorly. There may be associated tears at the petiole of the epiglottis due to shearing forces. Exposed fragments in the airway are prone to infection and must be removed. Thyroid fractures are either horizontal (more often in adults), vertical (more in children), or com-
Larynx Surgery pound. It can sometimes be difficult to detect horizontal ones by CT. There is associated soft tissue swelling and hematoma formation, all leading to airway compromise. Dislocations: cricoarytenoid dislocations can be caused by relative minor trauma such as intubation. There is usually a ventral dislocation. Vocal cord paralysis and aryepiglottic fold hematoma can serve as indicators of such dislocations. Cricothyroid dislocations require much
greater forces and are associated with cricoid or thyroid fractures in most cases. The distance between the cricoid and thyroid is enlarged and the cricoid can be rotated relative to the thyroid. There is a large risk of recurrent laryngeal nerve injury in these dislocations. Hematomas: hemorrhage in the soft tissues is associated with fractures, but can also occur independently. The paraglottic fat and subglottic region are the most likely regions involved.
Post-Treatment Neck
Because of the changed morphology of the treated neck, a thorough knowledge of the therapy undertaken and its expected changes is very important in the evaluation of patients after surgical or radiation therapy. Even so, posttreatment imaging remains a significant challenge and is often associated with a limited sensitivity for diagnosing local recurrence. LarynxSurgery
The goal of oncologic surgery in the larynxpharynx is complete resection of tumor with preservation of speech and deglutition. In the larynx, the most frequent conservational procedures and their resulting morphology are reviewed below (Fig. 8.34; Table 8.30).
Vertical Hemjiaryngectomy
In this procedure the ipsilateral vocal cord, paraglottic space and part of the thyroid cartilage are resected. Variations include resection of one arytenoid for posterior spreading lesions or the anterior aspect of the opposite true cord in anterior extensions. These procedures are indicated for limited glottic carcinomas with normal cord mobility and no supra/subglottic extension. Contraindications are extension to the false cord or in the subglottic as well as paraglottic or cartilage invasion. •
CT Morphology
The laryngeal axis is tilted and the thyroid on the side of resection is shortened, often with a paramedian plane of section. The adjacent car-
b VH Fig. 8.34 Lines of resection of procedures used in conservation surgery of the larynx, (a) horizontal supraglottic laryngectomy (HSL), (b) vertical hemilaryngectomy
(VH), (c) supracricoid laryngectomy with cricohyoidopexy (SL + CHP), and (d) supracricoid laryngectomy with cricohyoidoepiglottopexy (SL + CHEP).
275
276
8 Neck Table 8.30
CT landmarks
after conservative Hyoid
Horizontal supraglottic laryngectomy
bone
Present
laryngeal
surgery
Present
Supracricoid larynqectorny- CUP
Present
1997)
Thyroid
Cricoid
Arythenoid'
Arytenol
Resected
Upper third resected
Not modified
Present
Present
Ipsilateral
Not modified
(resected)
Vertical hemilaryngectomy
(Maroldi,
Epiglottis
Present
ala resected
Resected
Resected
(resected) Resected
Present
Shifted
Resected
Present
closer to
(present)
hyoid bone Supracricoid
qectomy-Cf-lf
laryn-
l>
Present
Presert
Resected
Suprahyoid
Shifted
Resected
closer to
(present)
Present
hyoid bone a
Ipsilateral,
"Contralateral,
Cf-H'ecricohyoidopexy,
CHEP=cricohyoidoepiglottopexy
tilage often becomes irregular and sclerotic. Glottic symmetry is usually distorted from scar tissue between the thyroid and arytenoid and often a laryngocele is present.
Supracricoid Laryngectomy Supracricoid laryngectomy with cricohyoidoepiglottopexy entails resection of the thyroid cartilage (except the cornu inferior). one arytenoid. paraglottic tissue. both true and false vocal cords and the lower portion of the epiglottis ineluding the pre epiglottic space. The epiglottic tip and hyoid are sutured to the anterior arch of the cricoid. This is the procedure chosen for more advanced glottic carcinomas. Contraindications are arytenoid fixation. invasion of preepiglottic fat. or massive extralaryngeal growth. Supracricoid laryngectomy with cricohyoidopexy differs in the fact that the entire epiglottis and its preepiglottic fat are resected. followed by pexis of the hyoid to the anterior cricoid. lt is performed for advanced supraglottic carcinomas extending into the ventricle as well as for transglottic carcinomas with limited cartilage invasion. Arytenoid fixation. invasion of cricoid or extension beyond the larynx (tongue base or valleculae) are contraindications.
CT Morphology
The thyroid and (part of) the epiglottis are absent while the hyoid and cricoid are adjacent in the midline. The redundant mucosa around the apex of the preserved arytenoid is regular, but a soft tissue pseudocord is seen extending downwards into the subglottis. The axis of the asymmetric larynx is often oriented in the transverse plane and pharyngeal pouches with air or fluid can be observed lateral to the preserved arytenoid.
Horizontal Supraglottic Laryngectomy In this procedure. the false cords. epiglottis and pre epiglottic space and upper part of the thyroid are resected. although variations exist in which part of the tongue base. piriform sinus or one arytenoid are also resected. lt is performed for selected supraglottic carcinomas with normal cord mobility and without extension to the ventricle. Contraindications are lesions that involve both arytenoids. postcricoid area. tongue base. cartilage or glottic region (crossing of vent riel e). CT Morphology
The anterior glottis is elevated just below the tongue base with hyoid and thyroid remnant close together and often irregular. The residual mucosa around the preserved arytenoids used for reconstruction is often thickened and redundant resulting in asymmetry of this neovestibule.
Radiation
Total
Laryngectomy
Total laryngectomy has to be performed if lesions cannot be treated conservatively as in subglottic extension or extensive cartilage invasion by the tumor. This is also the procedure for local recurrences or if radiotherapy has failed or produced osteoradionecrosis. All laryngeal structures. the hyoid bone. the piriform sinus. and strap muscles are resected and the defect is reconstructed in a tube-like neopharynx connecting the base of the tongue with esophagus. Contraindications are synchronous tumors or hematogenous distant metastases. •
CT Morphology
The resected structures
are absent resulting in a collapsed. irregular aspect of the mucosa. The neopharynx is a tubular structure with a regular wall thickness and smooth external margins. Also, the walls of the tracheostoma are thin and regular.
Neck Dissections
Surgery to the head and neck is often associated with some form of neck dissection for resection of (possible) positive lymph nodes. The terminology has recently been standardized and several forms are practiced. Radical neck dissection: In the classical radical dissection, all lymph nodes at levels I-V are resected, and the spinal accessory nerve, the sternocleidomastoid muscle and the internal jugular vein removed. Sometimes the superior belly of the omohyoid muscle is also removed. Modified radical neck dissection: In this procedure all lymph nodes of the radical dissection are removed but one or more major non-lymphatic structures are preserved. Selective neck dissections: Beside preservation of major non-lymphatic structures, selected lymph node levels are also preserved. There are four subtypes:
• supraomohyoid:
resection of all lymph nodes at levels I - III located above the superior belly of the omohyoid and anterior to the posterior margin of the sternocleidomastoid; • posterolateral: resection of all lymph nodes at levels II-Vas well as suboccipital and
Therapy
retroauricular lymph nodes (especially used for melanoma of the scalp); • lateral: removal of all lymph nodes at levels II-IV along the internaljugularvein; anterior compartment: resection of all lymph nodes at level VI from the hyoid bone level down to the suprasternal notch level.
Extended radical neck dissection: In this procedure one or more additional lymph node groups andlor nonlymphatic structures are resected compared to the radical neck dissection. Examples can include level VI, suboccipital or periparotid nodes and nonlymphatic structures like the carotid artery, vagus or hypoglossal nerve, or paraspinous muscles. Because the post-surgical neck is usually evaluated by a combination of clinical and endoscopic means, the role of CT in diagnosing recurrence is not clear. It is probably best reserved for patients with a high risk of recurrence based on pretreatment findings and clinical follow-up and for evaluation of late complications. It can evaluate deep extension or guide biopsy. For these high-risk patients, an early post-treatment baseline study may be beneficial. Tumor recurrence is suspected by the presence of an irregular, enhancing soft tissue mass > 1 cm, thickened commissure or lysis of residual cartilage. CT Morphology
With radical dissections the neck is markedly asymmetric with absence of lymph nodes and characteristic nonlymphatic structures. The trapezius muscle is atrophied with concomitant hypertrophy of the levator scapulae muscle. In selective dissections only minor differences may be present, especially with regard to lymph nodes and surrounding fat.
Radiation
Therapy
Primary radiation therapy is generally possible for Tl and T2 tumors as well as selected T3 tumors of the neck. It is the primary modality for nasopharyngeal and small glottic tumors but for other tumors there is an institutional variability based on local experience. In general. large volume tumors or tumors with massive invasion of the pre-epiglottic fat are a contraindication
8 Neck
for primary radiation therapy. Secondary radiotherapy is often performed in T3 and T4 tumors as an adjunct to conservation or radical surgery to get a better local control. Successful radiation therapy will lead to considerable shrinkage of the tumor within months after its institution. Failure of radiation therapy is suspicious when at 4 months after therapy 50% or more of the tumor mass is still visible. However. the high doses (>60Gy) used in radiation therapy lead to diffuse changes in the tissues in the neck. mainly as a result of inflammatory reaction with edema. fat infiltration. and fibrosis. The high doses can also be associated with complications like soft-tissue necrosis with ulceration or fistulae and osteochondronecrosis. Radiation treatment can cause a progressive inflammatory mucositis with edema. Severity varies with the type of treatment. The process is most severe 2-4 weeks after treatment and often involves the pharynx or larynx. There may be associated swelling of the retropharyngeal structures and airway narrowing. In closes of 45 Cy and larger. the salivary gland is affected too. resulting in a chronic sialadenitis with hypoplastic glands and xerostomia. Postradiation surveillance by palpation or endoscopy is often difficult. Therefore. it would be preferable to rely on imaging. Cost-effectiveness issues limit routine follow-up with imaging studies in high-risk patients. Pretreatment fac-
tors like tumor volume. cTNM-staging and cartilage invasion can identify patients at high risk of treatment failure and in these patients a baseline study at 4 months is advisable. Patients at low risk or those where there is clinical suspicion of complications can be imaged as needed. Early results of FDC-PET and especially 'Thallium-SPECT are promising. resulting in an improved accuracy for distinguishing between recurrence and radiation changes. •
CT Morphology
Expected changes after radiation therapy can be manifold and depend on the dose given. type of therapy and irradiated portals. The changes are often symmetric. start within weeks. and may resolve within 12-24 months or persist indefinitely. Usually, the skin and platysma are thickened with reticulation of subcutaneous and deep fat structures. The mucosal structures and pharyngeal wall may thicken also, with surrounding edema and increased enhancement. Salivary glands show a chronic adenitis with increased enhancement and atrophy and normal lymph nodes will atrophy. In the larynx, specific changes include infiltration of pre epiglottic and paraglottic fat, thickening of vocal cords and epiglottis, and mucosal thickening at the glottic commissures and subglottic region. Cartilage is usually unaffected by the radiation.
Acknowledgments We are grateful for the advice of Jonas A. Castelijns. MD PhD and Suresh K. Mukherji. MD who have helped to organize and write this
chapter and who have given incorporate their material.
us permission
to
9 Lungs and Tracheobronchial System c. Schaefer-Prokop,
M. Prokop
Anatomy Examination Congenital
282 Technique Disorders
286 ........................................................
292
Tracheal Bronchus, Cardiac Bronchus (p. 292) Bronchial Atresia (p. 292) Tracheobronchomegaly (p. 292) Bronchogenic Cyst (p. 293) Pulmonary Sequestration (p. 293) Venolobar Syndrome, Scimitar Syndrome, Congenital Hypogenetic Lung Syndrome (p.294) Anomalous Pulmonary Venous Drainage (p. 294) Acquired Disorders of the Tracheobronchial System Nonneoplastic Tracheal Stenosis (p. 295) Bronchiectasis (p. 296) Bronchiolitis(p.298) Allergic BronchopulmonaryAspergiliosis (ABPA)
295
(p. 302)
Tracheobronchial Tumors
303
Benign Tumors (p. 303) Malignant Tracheal Tumors (p. 304) Carcinoid Tumor (p. 304) Bronchial Carcinoma (p. 305) Secondary Tracheobronchial Neoplasms (p. 313) Pulmonary
Nodules
314
Metastases(p.315) Granuloma (p. 317) Pulmonary Pseudotumor (p. 317) Hamartoma (p. 317) Carcinoid Tumor (p. 317) Septic Emboli (p. 318) Peripheral Bronchial Carcinoma (p. 318) Predominantly Focal Pulmonary Opacities Atelectasis(p.321 ) Rounded Atelectasis (p. 321) Bronchiolitis Obliterans with Organizing Hemorrhage, Hemosiderosis (p. 323)
Pneumonia
.
319
(BOOP) (p. 321)
Continue
/\
Alveolar Cell Carcinoma (p. 324) Lymphoma (p. 324) Kaposi Sarcoma (p. 327) Pulmonary Blastema (p. 328) Pulmonary Infarction (p. 328) Wegener's Disease (p. 328) Churg-Strauss Syndrome (p. 329) Vascular Malformation (AVM) (p.330) Infectious
Diseases
330
Bacterial Pneumonia in the Normal Host (p. 332) Atypical (Nonviral) Pneumonia in the Normal Host (p. 333) Viral Pneumonia in the Normal Host (p. 334) Tuberculosis and Atypical Mycobacterioses (p. 335) Bacterial Pneumonia in Immunocompromised Patients (p. 336) Viral and Pneumocystis carinii Infections in the Immunocompromised Opportunistic Fungal Infections (p. 338) Lung Abscess, Cavitation and Pneumatocele (p. 340) Differentiation of Infectious from Noninfectious Disease (p. 341)
Patient (p. 337)
Diffuse Interstitial Reticular or linear lung Changes Pulmonary Edema (p. 341) Idiopathic Interstitial Pneumonia Asbestosis (p. 347) Diffuse Nodular lung
341
(liP), Collagen Diseases (p. 344)
Changes
. . . . . .. . . . . . . . .. . . . . . . . . ..
348
Pulmonary Lymphangitis Carcinomatosa (PLC) (p. 348) Sarcoidosis (p. 350) Silicosis and Coal Worker's Pneumoconiosis (CWP) (p. 352) Diffuse lung Changes with Increased Parenchymal Density.. Extrinsic Allergic Alveolitis, Chronic Allergic Pneumonitis Pulmonary Alveolar Proteinosis (PAP) (p. 356) Lipoid Pneumonia (p.357) Eosinophilic Pneumonia (p. 358) Drug-Induced Lung Disease (p. 358)
354
(p. 355)
359
Parenchymal Diseaseswith Decreased Density Emphysema (p. 359) Congenital Lobar Emphysema (p. 361) Swyer-James Syndrome (p. 362) Pulmonary Diseaseswith Cystic Changes Langerhans Cell Histiocytosis (p. 362) Pulmonary Lymphangioleiomyomatosis
.
(LAM) and Tuberous Sclerosis
Acquired Disorders of Pulmonary Vasculature Chronic thromboembolic pulmonary hypertension Pulmonary Veno-occlusive Disease (p. 366) Hepatopulmonary Syndrome (p. 366)
. (CTEPH)
362
(p. 364)
365
(p. 365)
Continue
>
Trauma
and
Postoperative
Chest
Trauma
Adult
Respiratory
Pneumonectomy lung
Chronic Airway
Distress and
Syndrome
Lobectomy
Transplantation
Reimplantation Acute
367
Changes
(p.367)
Rejection
(p. 368)
(p. 370)
371
.
Response
(p. 371)
(p. 371)
Rejection Complications
(p. 371) (p. 371)
Plain chest radiographs are an indispensable part of the diagnostic work-up of suspected thoracic disease. However. projection effects and limited density resolution limit the value of plain films as a diagnostic tool. Computed tomography. with its excellent contrast resolution and nonsuperimposed views. is the most sensitive modality for evaluating the lung parenchyma. CT is the method of choice for tumor staging. the diagnosis and follow-up of pulmonary and mediastinal lesions. and the detection of occult thoracic lesions that cannot be diagnosed with conventional x-ray films (Table 9.1). High-resolution thin-slice CT scanning (HRCT) improves our ability to define and evaluate fine structural details in the lung parenchyma. HRCT is used for the localization. characterization. and quantification offocal and diffuse parenchyma! lung diseases. The pathomorphologic response pattern of the lung parenchyma is fairly uniform for different etiologies. however. so clinical information is essential for the interpretation offindings. CT is an important adjunct to bronchoscopy for evaluating the airways. By defining the precise location of an abnormality, CT can accurately direct bronchoalveolar lavage (BAL) or bronchoscopic biopsy. CT can visualize surrounding structures as well as bronchial segments that are inaccessible to bronchoscopy.
Table 9.1 Spiral
Indications and Multislice
for CT examination CT
Tumor diagnosis • Pulmonary and mediastinal • Metastases • Tumor characterization • Tumor Detection • (Occult)
tumors
(solitary
nodule)
staging and quantification of infectious processes
• Cavitation Asbestosis, silicosis • Emphysema • Bronchiectasis Localization prior to • Biopsy, bronchoscopy, bronchoalveolar lavage • Differentiation of pulmonary from pleural lesions 1
High-resolution
CT(HRCT)
Diffuse lung diseases • Detection of subtle parenchymal • Morphologic characterization • Ouanification • Localization choalveolar
changes
of parenchymal changes for open lung biopsy or bronlavage
Anatomy •
Trachea
The trachea starts at the inferior border of the cricoid cartilage and is 9 to 12 cm long. In 50% of patients the membranous posterior wall forms a convex bulge projecting into the tracheal lumen. The average transverse diameter of the trachea is 15mm in women and 18mm in men. The upper limits for coronal and sagittal diameters are 26mm in men and 22mm in women; the lower limit of normal is 13 mm in men and 10 mm in women. The extrathoracic tracheal diameter varies markedly during forced inspiration and expiration (by up to 35%), and this should not be misinterpreted as wall instability. The intrathoracic part of the trachea remains fairly constant during the normal respiratory cycle. •
Central Bronchial System
The right main bronchus is shorter than main bronchus and appears adjacent upper lobe bronchus in longitudinal (Figs. 9.1, 9.2). The bronchus intermedius to 4cm distally and presents an elliptical
the left to the section runs 3 section
on the CT scan. The apical segmental bronchus of the lower lobe (B6) arises at approximately the same level as the middle lobe bronchus. The latter then bifurcates into a lateral and medial segment. The left upper lobe bronchus divides into only two segmental bronchi (apicoposterior, B1I2, and anterior, B3). Further distally, the lingular bronchus branches anteriorly from the inferior aspect of the upper lobe bronchus. Its division into a superior (B4) and inferior segmental bronchus (B5) is poorly visualized on axial CT scans because of its oblique course. It originates 1 to 2 cm cephalad to the right middle lobe bronchus. Past the origin of the B6 segmental bronchus, the right lower lobe bronchus divides into four segmental bronchi while the left lower lobe bronchus usually divides into three segmental bronchi (there is no B7). The branching pattern is subject to a considerable range of normal variation.
Fig. 9.1 Anterior view of the central tracheobronchial system. Upper Lobe 1 apical (R) 112 apicoposterior 2 posterior (R) 3 anterior (R/L)
(L)
Middle Lobe 4 lateral (R) 5 medial (R)
,-.:1'----1/2 1Mi!t----3
----5
4
---6 8 ~--9 -T-"'r~~-1
0
Lingula 4 superior (L) 5 inferior (L) Lower Lobe 6 superior (R/L) 7 medial basal (R) 8 anterior basal (R) 8 anteromedial basal 9 lateral basal (R/L) 10 posterior basal (R/L)
Anatomy
Fig. 9.2 Axial anatomy of the pulmonary segments and central tracheobronchial system. The major and minor fissures can be clearly identified (as fine lines or
avascular zones). The intralobar segmental boundaries (broken lines) can only be estimated from the course of veins and segmental bronchi.
283
9 Lungs and Tracheobronchial System
Thickening of the posterior wall of the intermediate bronchus that borders directly on the lung parenchyma is a sensitive sign of a mass lesion (except in the approximately 2 % of individuals with an aberrant superior pulmonary vein). Pulmonary
Vessels
The main pulmonary trunk bifurcates into a short left and longer right pulmonary artery. The right pulmonary artery runs between the superior vena cava and the right main stem bronchus and divides into an upper branch (truncus anterior) and the descending interlobar branch. The left pulmonary artery bridges over the left main bronchus. In some cases. the left upper lobe artery directly originates from the main artery. More frequently. the main pulmonary artery continues into a vertical interlobar or descending pulmonary artery running posteriorly to the lower lobe bronchus. from which the segmental arteries to the upper and lower lobe originate. The upper limit for the vascular diameter is 29mm for the pulmonary trunk. 28 mm for the left pulmonary artery and 24 mm for the right pulmonary artery. The right inferior pulmonary artery presents a circular cross section in the angle between the middle lobe bronchus and the origin of the 86 segmental bronchus (Fig.9.2d). Increased soft-tissue structures or scalloped borders in this area are a sign ofhilar lymphadenopathy. On the right, there are three pulmonary veins: the descending superior pulmonary vein and the middle lobe vein, which both run into the superior venous confluence at the posterior aspect of the left atrium, and the inferior pulmonary vein running more horizontally into the inferior pulmonary venous confluence. On the left there are two veins: the superior pulmonary vein draining the upper lobe and the lingula and the inferior pulmonary vein draining the lower lobe. They run into the superior and inferior venous confluences, respectively. Pulmonary
of each lung is easily traced on axial CT scans in its
posterosuperior-to-anteroinferior
Table 9.2 Azygos
Accessory fissures fissure
caused by downward invagination of azygos vein
Inferior accessory fissure
between medial segment (7) and the remaining lower lobe
Superior fissure
between upper segment and the basal segments t..pper lobe
accessory
Segments
There are ten pulmonary segments in the right lung and eight in the left lung. The major fissure
course
(Fig. 9.2). The minor fissure between the upper lobe and middle lobe of the right lung runs approximately parallel to the scan plane and is identifiable as a nonvascular zone on one to three axial slices. The anatomic relationship between the segmental bronchus and the concomitant artery is helpful to differentiate the middle from the upper lobe: in the upper lobe the artery runs medially to the segmental bronchus while in the middle lobe the artery is located laterally to the segmental bronchus. Thin slices and multiplanar reformations based on multislice CT data sets better demonstrate the course of the lobar fissures than axial slices alone. Additional fissures can be identified more frequently than with any other technique. The commonest accessory fissures include the azygos fissure, the inferior and superior accessory fissure, and the left minor fissure (Table 9.2). Incomplete fissures are frequent findings that allow for collateral air drift and spread of disease. The bilateral inferior pulmonary ligaments represent parietal pleural reflections extending from the inferior aspect of the hilum to the posterior diaphragm. On axial CT images they appear as broad bands between the posterior mediastinum (region of esophagus) to the diaphragm. CT cannot define the segmental boundaries in normally aerated lung, but the segments can be identified indirectly by tracing the central segmental bronchi and the interlobar veins. Substantial variations may be encountered in segmental pulmonary anatomy. A slice collimation of 5 mm or less is necessary for the positive identification of segmental bronchi.
Left minor fissue
between lirwla upper lobe
and left
(6) ofthe
Anatomy •
Peripheral Bronchi
The airways undergo approximately 23 divisions between the trachea and the respiratory bronchioles (Fig. 9.3). Standard CT can define fourth-order bronchi (3 mrn), while HRCT can resolve eighth-order branches (1-2 mrn). The CT detection of normal bronchi relies on visualization of the bronchial wall. The thickness of the bronchial wall is approximately one-tenth the diameter of the bronchus. Normal intralobular bronchi are not visible, as their wall thickness is less than 0.15mm. The presence of visible bronchial structures in the lung periphery (within 2 to 3cm subpleural space) signifies pathologic bronchial wall thickening or ectasia of the small airways. The small airways of the lung « 2m m in diameter) have a total cross-sectional area greatly exceeding that of the larger airways but account for only about 25% of the total airway
resistance. The small airways must undergo severe destructive changes before pulmonary function is measurably affected. •
Lung Parenchyma
The smallest structural unit visible on CT lung scans is the acinus, which is about 6-8 mm in diameter and consists of alveolar ducts, alveolar sacs and approximately 4000 alveoli distal to the terminal bronchiole (Figs. 9.3, 9.4). The secondary lobule is the smallest discrete portion of the lung that is surrounded by connective tissue. It is the smallest anatomical unit that can be clearly identified on HRCT, and assessment of the distribution of abnormalities within it can be helpful in the differential diagnosis of parenchymal disease. It consists of a maximum of 12 acini and has a characteristic polygonal shape with an edge length of 1-2.5 ern, a septal boundary, and a central core consisting of a bronchiolus and arteriole (Fig. 9.4).
Fig. 9.3 Schematic representation of the bronchial divisions, showing the limits of spatial resolution of CT and HRCT.
Sekundary lobe (1-2,5 ~ cm) Bronchi (2-4)
Acinus (6-8
Q
mm)
Primary lobe Bronchioles
Terminal :Respiratory' bronchiole:,bronchiole: , :(17-19) :
, , ,
Alveolar ducts (20-22)
Alveolar saccules (23) 23 generations
I
Visible with CT
Visiblewith HRCTwhen norma
j
Visible with HRCTwhen diseased ,
Bronchiectasis:
Bronchiolectasis
I Emphysema
I
285
286
9 Lungs and Tracheobronchial System
_-----------
Interstiti u m: a) central b) septal c) peripheral
Viszeral pleura
c~ _""f-'--
b
\
)7
~
----
Secondary
Acinus lobule
*
Centrilobular artery -
Interlobular
septum
Terminal ::;', ~\'T-----
bronchiole
ILobular artery
Terminal bronchiole
Respiratory I bronchiole· Central acinar artery *----''
*
Lymph vessels" Pulmonary vein
* Visible only when diseased
• Never visible
Fig. 9.4 Schematic representation ing veins, and interstitium.
of two adjacent
secondary
The arterioles lie 5-10 mm deep to the pleura. The veins take an interlobular course and are visible 1-2 cm below the pleura. Normally the bronchioles cannot be seen on CT. This means that normal Iung contains an immediate subpleural zone 5-10 mm wide that is devoid of visible vascular and bronchial structures. The interstitium is the connective-tissue framework of the lung. Its fiber tracts support the bronchi, the pulmonary vessels, and the membranes for alveolar gas exchange. The interstiti urn consists of the central or axial interstitium, which surrounds the central airways and vascular structures, the septal interstitium, which supports the alveolar walls, and the peripheral interstitium, which forms the basic
lobules with afferent arteries
and bronchioles,
drain-
framework of the lobar and lobular septa. The interlobular septa normally are not visible on CT or may appear only as faint linear structures in the basal lung zones. In the presence of thickened lymphatic vessels or interstitial disease, however, they may become very prominent and define the architecture of the lung parenchyma. The CT density of the lung parenchyma normally shows an anteroposterior and an inspiratory-expiratory gradient. The anteroposterior density gradient is 20 ± IOHU in inspiration and 150±20HU in expiration. The average pulmonary density in inspiration ranges from -900 to -850 HLJ.
Examination Technique
The choice of examination technique for the Iung depends on whether the whole lung is evaluated (usually in cases offocallung disease) or whether a disseminated process can be examined by discontinuous scans with high spatial resol utian (HRCT) . Only multislice scanning allows for a combination of these two different requirements.
Oral contrast is usually not indicated for CT of the lungs unless tumor involvement of the esophagus is suspected (see Chapter 15, p. 548). Intravenous contrast material should always be used for the investigation of mediastinal and pleural abnormalities. Contrast administration is not only mandatory for the evaluation of focal intrapulmonary lesions (tumor, rounded atelec-
Routine Chest CT 287 tasis). Enhancement characteristics are not important for the characterization of diffuse parenchymal changes. Contrast administration is essential for all vascular investigations (pulmonary embolism. aortic studies) and for staging examinations.
Table 9.3
Suggested
protocols
for CT scanning
Routine Chest CT Spiral CT is the standard for examinations of the lung (Table 9.3). Most patients will be scanned in inspiration. Patients with marked dyspnea who are not be able to hold their breath during the scan should breathe shallowly instead.
of the thorax
General Preparation
not necessary
Patient
Supine with arms raised (standard) Prone position with elevated arms: • Differentiating hypostasis from subpleural • Differentiating edema from infiltration
position
Scan range
Respiratory
From thoracic From thoracic phase
inlet to posterior pleural recess (standard) inlet to below the renal veins (suspected bronchogenic
carcinoma)
End-inspiratory scans (standard) End-espiratory scans: • Air trapping: early detection of peripheral airway obstruction • Differentiation of air trapping from ground-glass opacity • Compliance testing (fibrosis versus edema) • Bronchogenic carcinoma: chest wall fixation
Windowing
Lung (standard): Emphysema: Mediastinum (noncontrast): Mediastinum (post contrast):
Scan parameters
l-slice SC/TF/RI
4-slice' SC
t
2-3.75
Standard
5/10/5
Tumor staging, CT-bronchography
3/5/2 t or 2/4/2 i
Screening
fibrosis
5/10/5 i
t'
W/L ~ W/L~ W/L~ W/L~
1500/- 650 800/-800 400/40 400/60
16-slice' SC
1-1.25
t t
1-1.25
i
axial SW/RI
1-1.5 i
t'
MPRb SW/RI
5/4
0.75-1.25
r
0.75-1.25
t
3/3
3/3 cor' 4/4 sag'
5/4
t'
HRCT(discontinuous) (spiral scan)
1-1.5/10-20
Contrast
l-slice V/F/D
4-slice V+N/F/D
16-slice V+N/F/DX
Standard
70/2/20
70+50/2/20
50+50/2/30
Tumor staging
90/3/30
90+50/3/30
90+50/3/25A
Pulmonary vessels e
140/3/20
120+60/3/20
90+50/3/5pd
injection
1-1.25/10-20 1-1.25i
0.5-1/10-20 0.5-0.75i
1-2/10-20 1-1.5/10
1.5/10 cor 1.5/10 sag
d
SC= slice collimation (mm), TF = table feed (mm/rotation), RI = recon. increment (mm), i t = scanning direction, SW = effective section width (mm), MPR = multi planar reformats, axial = axial sections. cor= coronal, sag = sagittal CM= contrast material, Contrast concentration = 300 mg/ml iodine V = volume of CM (ml), N = volume of saline flush (ml), F = flow rate (rnlts), D = start delay. X = trigger region a Pitch P = TF/(NxSC) should be chosen close to 1.5 with 4-slice CT scanners, 1.3-1.5 with 16-slice scanners b MPRare reconstructed from a "secondary raw data set" with SW/RI = 1-1.5/0.7 or 0.5-0.8/0.6 c read: single section mode HRCTwith 0.5-1.5 mm collimation and 10-20 mm increment between sections dread: 25A (5P) = start delay 25 s (5 s) after reaching 50 HU enhancement in the desc. aorta (pulmonary artery) e choose lower dose and kVp setting (CTDI", = 3-6 mGy; 80-100 kVp, depending on patient size and CT scanner)
II
9 Lungs and Tracheobronchial
System
When such a patient is scanned during inspiration, marked artifacts will occur as soon as the patient begins breathing again, Breathing artifacts decrease with caudocranial scanning, The technician, however, has to take care that there is a sufficiently long waiting period (2-4 seconds suffice) between the inspiration command and the beginning of the scan, Otherwise, the lung bases will suffer from marked artifacts due to the final inspiration of the patient. Patient compliance can be improved if they are instructed prior to the scan and asked to hyperventilate (2-5 deep breaths) immediately prior to the spiral examination, With spiral CT, a 5 mm collimation and a l Ornm table feed (pitch = 2) will give a good compromise between spatial resolution along the z-axts, scan duration (usually some 30 seconds), and patient exposure, Reducing the pitch to 1.5-1.8 will slightly improve image quality at the base of the lung because lower pitches are somewhat less susceptible to pulsation artifacts, For multislice CT with 4-slice scanners in a routine setting, a 4 x 2.5 mm (4 x 2-4 x 3 mm) collimation usually suffices unless there is a high likelihood that coronal or sagittal reformations will have to be reconstructed from the data set Under such conditions, volumetric imaging using 4xl-l.25 mm is advocated, With 8- and 16-slice scanners, 1-1.5 mm section collimation can be used on a routine basis, Reconstructing axial sections of 5-7.5 mm section width is sufficient for rou tine interpretation, Reconstruction of an overlapping secondary raw data set with SWIRl = 3/1.5 (for 4 x 2.5 mm collimation) or SWIRl =l.25-2/0,7 (for 1-1.5 mm collimation on 4- to 16-slice scanners) are required only if multiplanar reformations were initially planned or become necessary to solve a particular question on the axial sections, For optimum image quality, the images of the mediastinum should be reconstructed using a soft kernel, while images of the lung require a high resolution kernel. Instead of directly reconstructing two sets of images from the raw data set (larger amount of images to be processed and archived), display filters may be used to smooth images from a single data set reconstructed with a high resolution kernel. For opacification of the mediastinal vessels, relatively small amounts of contrast material are required for spiral CT or multislice CT acquisitions: the injection of some 50-70ml of contrast
material followed by 50ml of a saline flush (2 mlls) is usually sufficient. Because of the substantially reduced absorption in the lungs the radiation exposure (mAs settings) for the chest can be approximately halved as compared to abdominal scanning, Usually a CTDlvol = 5 mCy (more for obese patients, less for slim tndtviduals) yields good results, One has to keep in mind, however, that image noise in the cranial portions of the liver as well as in the shoulders may then be substantially increased, This usually does not affect diagnostic accuracy because liver imaging using a chest protocol is far from optimum (and should be optimized separately anyway), and the spectrum of neck pathology in chest examinations is very limited (lymph nodes, masses extending into the neck), Image quality in these regions may be improved even at reduced exposure settings with the use of longitudinal dose modula-
tion techniques.
Tumor Staging The highest possible image quality should be employed for tumor staging during the initial evaluation of primary tumors or prior to surgery. Since high space resolution is possible only for a short scan range with spiral CT, a composite approach with thin and thick sections has to be taken. We suggest focusing primarily on the pulmonary lesion including the pulmonary hila first, and then scanning the rest of the lung or repeating a scan of the whole lung using the protocol described above. For the initial high resolution scan, a (SWI TFIRI =) 214/2 protocol can be recommended if the scan range can be covered within one breath-hold phase. It is advisable to test the breath-holding capacity prior to the scan because breathing artifacts substantially impair the quality of reformatted images. For larger ranges, a 3/5/2 protocol can be recommended. When only the primary lesion (peripheral nodule) has to be scanned (for in-detail analysis of morphologic features or for nodule volumetry), a 11211 protocol can be employed. With a 11311 protocol, a 50% longer scan range can be achieved at only slightly reduced image quality (if the scanner supports a pitch of 3). With 1 mm sections, the mAs settings have to be increased to ensure sufficient image quality.
High resolution CT (HRCT)
From this thin-section scan multiplanar reformation should be reconstructed as described below for multislice CTscans. For multislice CT, volumetric imaging with a 0.5-1.25 mm collimation (P ':1.5) yields excellent results for 4- to 16-slice scanners. Only in very dyspneic or obese patients should thicker sections be used. The acquired data are reconstructed as a "secondary raw data set" with SW IRl= 0.7 -1.5/0.7. From this secondary raw data set, axial sections of 4-5 mm width are reconstructed, followed by coronal and sagittal thin sections across an intrapulmonary tumor (SW = 1. 5mm). If lymph node, mediastinal, or chest wall involvement is suspected, additional 3-4 mm coronal and 4-5 mm sagittal reformations are reconstructed across the central portions of the mediastinum (including the hila), or across the affected portions of the chest wall. Additional expiratory scans can determine whether a tumor adheres to the parietal pleura (no movement on expiration). Scans after inducing a pneumothorax have been suggested for the same purpose but are substantially more invasive and have the same problems with differentiation between mere adhesion and chest wall infiltration. For best results, a high resolution kernel should be used for the lungs and a soft kernel for the mediastinum and chest wall. However, this will create up to WOO images per patient (two secondary raw data sets). As a compromise, a Single secondary raw data set can be reconstructed with a high resolution kernel. and only the multiplanar images that are reconstructed from this data set are subsequently subjected to softening display filters. For tumor staging, a volume of 70-100 ml of contrast material should be injected at 2-3mlls, followed by a saline flush of 50ml using the same injection rate. A scan delay of 30 seconds yields good results in most patients. Larger volumes and higher flow rates improve contrast enhancement but add little to the diagnostic yield of the scan. Radiation exposure is no limiting factor for tumor staging. Instead, exposure setting should be chosen so that image quality is optimized. If thick axial (and coronal or sagittal) sections are sufftctent, a CT01voi of 5-10 mGy is recommended, but thin-section imaging requires a CTDlYoiof 10-25 mCy, depending on the size of the patient.
High Resolution CT (HRCT) High resolution CT is indicated for diffuse lung disease. If focal lesions are suspected, a sole examination by HRCT is not indicated. In such cases, an (additional) spiral CT should be performed. When volumetric imaging using multislice CT is performed, thick conventional as well as thin HRCT sections can be reconstructed form the same data set. HRCT examination of the lung parenchyma is usually performed with a slice thickness of 1 -2 mm, a high-resolution kernel and discontinuous acquisition with a 10 to 20mm gap between sections. Scanning is generally performed at end inspiration with the patient in a supine position. Expiratory scans are necessary to detect air trapping and thus early (peripheral) airwayobstruction, and to differentiate air trapping from ground-glass opacities. They have also been suggested for testing lung compliance (fibrosis versus edema) or bronchial collapse. It is possible to use a wider gap of > 20 mm for this purpose. Scans in prone position help differentiate hypostatic edema from subpleural fibrosis, and edema from infiltration. Only the areas that were suspicious on the supine scans need to be examined again (usually the lung bases). Scanning primarily in a prone position may be useful for patient with asbestos exposure (HRCT screening examination) because the initial findings with this disease predominantly affect the posterior portions of the lung. For screening purposes, the number of sections may be reduced to keep radiation exposure low. The mAs settings for HRCT of the lungs should be about 50-100% higher than for routine chest CT because of the higher noise on thin sections. Despite the fact that the CTDlw (dose per section, displayed on the user interface for incremental CT) may be quite high (1020 mGy), the dose to the patient remains low because of the large interslice gap (CTDlvol =1-4 mGy) Multislice CT scanners should be used in a Single slice mode. If available, only a Single section should be acquired at each table position, because a set of adjacent sections rarely adds to the diagnostic information but substantially increases patient dose. Acquiring Single sections is not yet possible with most 4-slice scanners but
9 Lungs and Tracheobronchial System can be clone with most 8- and 16-slice scanners. If no single-section mode is available, dedicated HRCT examinations should preferably be performed on single-slice systems. When 4-slice scanners have to be used, discontinuous sections of 4 x 1 mm to 4 x 1.25 mm width (or 4xO.5 mm width on Toshiba units) should be acquired every 10-15 mm, preferably every 20mm. In order to take full advantage of the increased number of sections per scan, one should not only view the axial sections but create thinslab MIP and mIP across the (four) simultaneously acquired sections. This may improve the detection of initial disease in some patients. Multislice HRCT of the whole lung is based on volumetric imaging with 1-"1.25mm section collimation for 4-to 8-slice scanners and 0.5-0.75 mm collimation for 16-slice scanners. In general. a pitch P s 1.5 (P" '6 for 4-slice scanners) can be recommended. A sufficiently high dose (CTOIvoi 'IOmCy) is required for good results. Image noise is reduced when 1.5 mm thick axial as well as coronal or sagittal reformations are reconstructed from a secondary raw data set. Multislice HRCT of the whole lung may add findings unsuspected from conventional discontinuous HRCT. Because very few of these findings appear to hold therapeutic consequences, HRCT scanning of the whole lung cannot yet be recommended on a routine basis (radiation exposure is increased by a factor of 5-10). Because scanning of the whole lung improves side-by-side comparison of findings, it may be indicated for patients in whom even minor changes during follow-up warrant a change in therapy. In such a follow-up situation, scans with reduced dose may be feasible.
Low-dose
CT
Because of the low absorption of the lungs, radiation exposure can be substantially reduced ifC T scanning can focus on the lungs only. With spiral CT. a pitch of 2 should be employed to get the best image at a given (low) exposure dose (see Fig. 5.11). For multislice CT, the same holds true: high pitches yield better quality (in terms of noise) at a given low exposure dose, especially when thin sections are employed (less effect of electronic noise, see pp.120, 139, and 213: also compare Fig. 4.10). The CTDIvoi can be reduced to 2 mGy for most patients.
U sing a less edge-enhancing standard filter substantially improves image quality, especially for ultra-low dose applications, such as bronchogenic cancer screening (see below). For HRCT, a reduced dose (CTDIw= 2-6 mGy) is feasible for follow-up examinations (depending on the therapeutic implications and the subtlety of disease). Increasing the gap between sections to 2cm or higher also reduces patient dose but is only feasible for relatively evenly distributed disease processes. If in doubt, additional sections may be added in the gap between the original sections. With multislice CT and thin-section lowdose scanning, the noise on the original axial sections may be unacceptably high, especially in the mediastinum. Reconstructing thicker sections either primarily from the raw data or from a secondary raw data set will substantially improve image quality (see Fig. 5.18). Since low-dose scanning is at the lower edge of acceptable image quality, any further decrease in quality may reduce diagnostic yield. Obese patients require more dose, otherwise image quality may deteriorate substantially, especially in the lung apex (shoulder regions). Also, a higher kVp setting (usually 140 kVp) should be employed to improve radiation penetration. If in doubt, a single section test scan (which is often performed anyway because of the selection of the reconstruction field-ofview) with the same mAs and kVp settings (not CTDIvoi) as compared to the planned spiral or multislice scan can be used to judge if image quality is sufficient. For this test scan, the CTOIvoi provided on the user interface will be 1.5-2 times (= pitch) higher than the CTDIvoi of the actual scan because (per definition) a single section has a pitch of 1. If available, adaptive noise filtering (see Fig.5.7) substantially improves image quality in critical regions (lung apices). For children, the dose should be adapted to their size (see Table 5.6). In small children, 80 kVp and minimum mAs settings may be employed. The test scan used for determining the reconstruction field-of-view can be used to decide whether image quality is too high (lower dose possible) or too low (higher dose or thicker sections necessary). Low kVp settings can also be recommended for imaging of the pulmonary vessels (see also p.837)
kernel for image reconstruction
vtnuet Low-Dose
Screening
of Bronchogenic
Virtual
Bronchoscopy
Bronchoscopy
and CT Bronchography
and
CT
Cancer
Bronchography
For low-dose screening of bronchogenic cancer, even lower dose may be employed, With conventional spiral CT, a 5110/5 protocol at 120 kVp with a CTDlvoi < 0,8 mGy may be feasible in normal size patients (depending on the minimum mAs setting of the scanner), This corresponds to an effective dose of approximately OAmSv for females and 0,3 mSv for males, It has to be compared to conventional chest radiography, which yields some 0,05-0.1 mSv for a single PA projection, and 0.12-0,2 mSv for PA and lateral radiographs, Again, the dose has to be increased in obese patients, A standard filter kernel for image reconstruction substantially improves image quality, If a suspicious nodule is detected on a screening scan, the patient is brought back and a thin-section CT with moderately increased dose (CIDlYol= 2-4 mGy) is performed with a 112/0,5 protocol and a field-of-view (FOV) that is targeted on the nodule of interest This scan is used for in-detail morphologic analysis (presence of calcifications) and for nodule volumetry. Depending on the size of the nodule (usually < 1 cm) and the screening algorithm, the scan is repeated after 6 weeks to 6 months, Changes in nodule volume by 20%-100%, depending on its size, indicate Significant growth and thus malignantpotentiaL With multislice CT, scanning can be performed immediately with thin sections (0,751,25mm collimation for 4- to 16-slice scanners), Thick sections of some 5 mm width are usually used for image review unless an automated detection program is available, If a nodule is detected, thin sections can be reconstructed around this nodule using a small FOV, and morphologtc and volumetric analysis can be performed, Automated programs substantially simplify nodule analysis because local multiplanar reformations centered around the nodule, volume rendered displays, and volumetric analysis become almost instantly available (see Fig, 9.18), No second scan is required, and the detection of nodule growth may become feasible at an earlier point in time (6-12 weeks follow-up),
Virtual bronchoscopy is a three-dimensional (3D) visualization technique that requires perspective rendering and various interactive tools to obtain the visual impression that the observer is moving within the data volume, The technique can be used to assess anomalies of the central tracheobronchial system noninvasively but cannot substitute for fiberoptic bronchoscopy in patients in whom biopsy has to be performed, lis most useful indications are for the planning of complex transbronchial biopsies (peripheral bronchi or central masses next to major vessels), lt is potentially more sensitive for detecting small irregularities of the tracheobronchial wall but such irregularities are often caused by mucous retention, especially in smokers, Other rendering techniques (compare Fig, 2.16, 2.18, and 2,36) are often superior to virtual bronchoscopy for the display of the tracheobronchial system (CT bronchography). Such techniques present an "intraluminal cast" (using SSD or VRT, see Fig.2.36a) or display the tracheobronchial anatomy as three-dimensional structure with semitransparent walls (tissue transition projection, see Fig.2.36b). Virtual bronchoscopy and CT bronchography require a highly resolved data set. Thus a protocol should be used as described above for tumor staging. M ultislice CT has the advantage that there are fewer pulsation or breathing artifacts because of the more rapid scanning procedure. In addition, it not only provides good results for the central tracheobronchial system but also for more distal bronchial segments. Due to the perspective distortion, measurements should not be performed on the endoscopic views but on the axial or reformatted images.
9 Lungs and Tracheobronchial System Congenital Disorders
Congenital disorders of the lungs and airways are very rare and may cause recurrent infections. CT is indicated if chest radiographs and bronchoscopic findings are equivocal. Tracheal Bronchus, Cardiac Bronchus
Tracheal bronchus refers to a bronchus (upper lobar, apical segmental, or middle lobar) arising from the trachea at an atypical site or a supernumerary bronchus arising directly from the trachea. In most cases the anomalous airway arises from the right lateral wall of the distal third of the trachea. Cardiac bronchus is a supernumerary bronchus, usually rudimentary, that most commonly arises from the medial wall of the intermediate bronchus and passes downward and medially toward the heart (Fig. 9.5). These anomalies are clinically significant in that they predispose to recurrent infections and bronchiectasis, frequently isolated to the bronchopulmonary territory of this bronchus. After endobronchial intubation, the balloon may inadvertently obstruct an atypical bronchus causing right upper lobe atelectasis. Bronchial Atresia
The apicoposterior segment of the left upper lobe is most commonly involved. The lesion may be associated with a bronchogenic cyst.
•
CT Morphology
While there is atresia of the lobar or segmental part of the bronchus, the more distal structures are well preserved. Mucus secreted within the airways distal to the obliterated lumen cannot pass the bronchial stenosis and subsequently causes a mucous plug and a mucocele. Collateral air drifts keep the involved segment inflated. It becomes hyperinflated due to expiratory air trapping. CT reveals the mucocele as perihilar branching structure and hyperinflation of the involved segment. Tracheobronchomegaly
Tracheomegaly can have various causes (Table 9.4), the commonest being destruction of the cricoid cartilage. •
CT Morphology
Tracheobronchomegaly is present when the tracheal diameter exceeds 2.5 cm and the diameter of the main bronchus exceeds 2.0 and 2.3 ern, respectively. Absence or hypotrophy of elastic fibers and hypotrophy of the muscles result in pathologic airway dynamics with dilatation in inspiration and collapse in expiration. Intrapulmonary bronchiectasis is a frequent associated finding.
Fig.9.5 VRT(a)and mIP(b)of the central tracheobronchial system in a newborn: two bronchial buds arise from the intermediate bronchus. Images were derived from a reduced dose CT scan (4 x 1/6, CTDlvoi = 1.8 mGy). Note the small size of the bronchial system with a tracheal diameter of 4mm.
a
b
Pulmonary Sequestration Table
9.4
Etiology of tracheobronchomegaly
Traumatic • Long-term
irtubation
Congenital syndromes • Ehlers-Danlos (cartilage defect) • Mounier-Kuhn (tracheobronchial Chronic inflammation • Recurrent childhood • Immune defect • Cystic fibrosis
diverticulosis)
Pulmonary
Bronchogenic
upper lobar or diffuse pulmonary sarcoidosis, idiopathic pulmonary
Cyst
Bronchogenic cysts are the commonest bronchopulmonary anomaly and result from abnormal bronchial arborization during embryologic development. They are classified by location as intrapulmonary (70%) or mediastinal (30%). •
Sequestration
infections
Traction • Predominantly fibrosis (e.g., fibrosis)
area. They communicate only very rarely with the tracheobronchial system. Infection is extremely rare. The mediastinal bronchogenic cyst is usually larger as compared to the pulmonary site (c- 20 em).
CT Morphology
The intrapulmonary type occurs most frequently in the lower lobe. The cyst may reach a diameter of several centimeters (central type) or may be a multilocular mass composed of multiplesmallcysts (peripheral type). Cysts with very proteinaceous contents may have a CT density as high as 50 HU. causing the "cystic" mass to exhibit soft-tissue density. The uninfected lesion does not enhance after intravenous contrast administration. Characteristically the cysts do not communicate with the tracheobronchial tree until they become infected. which occurs in about 75% of cases. When communication is established. the cysts contain air or show an air-fluid level. With infection of the cyst. the usually sharp definition of the cyst border becomes obscured by consolidation in the adjacent lung parenchyma. the cyst wall becomes thickened and contrast-enhancing, in that way simulating the appearance of a lung abscess. Differentiation between a bronchogenic cyst and an acquired cyst (e.g., after adult respiratory distress syndrome (ARDS) or a healed abscess) may be difficult. Mediastinal cysts are most commonly seen in the right subcarinal area but can also be located paratracheally or in the paraesophageal
A pulmonary sequestration is a nonfunctioning mass of aberrant lung tissue that lacks a normal communication with the bronchial system and has a systemic arterial blood supply. Sequestrations are found in the posterior left lower lobe segment in approximately 65% of cases and in the right lower lobe segment in the remaining cases. Two forms are distinguished according to their relationship to the pleura: • intralobar sequestration (75-85%) which is located within the normal pleural covering of the lung and drains into the pulmonary veins, it typically manifes ts in ad ults; • the extralobar form (15-25%) which has its own separate pleural covering and drains into the vena cava or azygos vein, and is diagnosed in the newborn or early infant. The diagnosis should be considered in any patient with persistent or recurrent lower lobe pneumonia. CT Morphology
An intralobar sequestration may be aerated (by collateral air drift through the pores of Kohn) and frequently shows areas of hyperinflation due to air trapping phenomena. In this case it can be identified as a sequestration only by its atypical arterial supply from the descending aorta. Sites of mucoid impaction in the rudimentary bronchi are a typical finding. If the sequestration has become infected or has perforated into the bronchial tree, CT may show uni- or multilocular cysts containing variable amounts of air and fluid. Extralobar sequestrations generally do not communicate with the bronchi and are of softtissue density throughout (Fig. 9.6). Homogeneous enhancement of the mass coinciding with aortic enhancement suggests the correct diagnosis on postcontrast scans. Some patients show emphysematous changes in the adjacent lung parenchyma and cystic areas within the sequestration.
293
294
9 Lungs and Tracheobronchial
System
Fig. 9.6 Extralobar sequestration in a newborn child. Left lower lobe consolidation (a) with an systemic arterial blood supplyfrom the infradiaphragmatic descending aorta (arrows) seen on a 3 mm thick MIP (b). The arrowheads on (b) indicate a pneumothorax treated by a pleural drain (arrowhead in a). The black arrowheads point to a gastric tube. Ultralow dose scan with 0.6 mGy CTDlvoi after intravenous injection of 2 ml contrast material (4 x 1/6).
a
b
The anomaly is frequently associated with other congenital abnormalities. In rare cases the sequestered segment may fistulate to the esophagus or stomach. Venolobar Syndrome, Scimitar Syndrome, Congenital Hypogenetic Lung Syndrome
Venolobar syndrome is a collective term for various congenital anomalies of the thorax, which frequently coexist in the same patient (Table 9.5). •
CT Morphology
CT demonstrates a small, hypoplastic lung with anomalies in the course and branching of the airways. Bronchial diverticula and bronchiectatic areas are common associated findings. The examiner should be alert to anomalous pulmonary venous return to the vena cava, right atrium, or atrial appendage. Most commonly there is a single vein draining the entire right lung, that runs inferiorly parallel to the right Table 9.5
Venolobar
cardiac border into the inferior vena cava below the diaphragm. (On a frontal radiographs this vein has the appearance of a Turkish sword or scimitar.) Atypical pulmonary veins may be found even in the absence of detectable anomalies of the lung parenchyma. The lung usually receives a systemic blood supply from the descending or abdominal aorta, and sequestrations may be present. Anomalous
Pulmonary Venous Drainage
Anomalous venous drainage is characterized by venous drainage directly into the right heart or the systemic veins (Table 9.6). It may be partial or complete, involving some or all veins, and in any case producing an extracardiac left -to-right shunt. The anatomy is highly variable with approximately 30 different patterns. It is frequently associated with other cardiovascular abnormalities (e.g., atrial septal defect). One diagnostic clue is the identical oxygen saturation in all four heart chambers because pulmonary venous blood mixes with systemic venous blood. The long-term effect of the increased
syndrome
Hypoplastic right lung, lobar agenesis, aplasia or hypoplasia
69%
Partial or complete anomalous veins
pulmonary
31 %
of the right puimo-
14%
Absence of hypoplasia nary artery
Table 9.6
Pulmonary sequestration
24%
Partial or complete systemic arterial supply to the lung without sequestration
10%
Absence or disruption of the inferior vena cava Duplication of the diaphragm right hemithorax
separating
7% the
Patterns of anomalous venous drainage
Supracardiac (50%) • Venous return into superior vena cava, persistent left vena cava or the left innominate vein Cardiac (30%) • Venous return into right atrium or right coronary sinus Infradiaphragmatic (15%) • Venous return into the portal vein or its branches
7% Mixed (5%)
Nonneoplastic Tracheal Stenosis
a
b
c
Fig. 9.7 Partially anomalous pulmonary venous return (PAPVR). Axial sections demonstrate a large superior vena cava (SVC) and a symmetrical vein on the left side (a). There is direct drainage of the right upper lobe vein into the SVC (b). The volume-rendered image demon-
strates that the left-sided mediastinal vein drains the almost the whole left lung into the left braehioeephalie vein (c). The right upper lobe vein drains into the superior vena cava (4 x 1 16).
blood flow in the involved lung is pulmonary hypertension.
left superior vena cava lateral of the aortic arch (Fig. 9.7), and a lack of veins that drain into the left atrium. The anomalies, however, are best described and detected on volume-rendered displays of the pulmonary vasculature because such a display gives the best overview over the -often complex-anatomy. In the presence of arterial septal defects there may be a jet of highly concentrated (undiluted) contrast material entering the left atrium which even can be seen on volume-rendered images (see Fig.23.19b).
•
CT Morphology
CT angiography using spiral or multislice CT allows for complete description of such anomalies, including hypoplastic native pulmonary veins. Without prior clinical suspicion, the anomalies can be missed easily on axial sections. Typical features involve a pulmonary vein that directly enters into the right atri urn or the superior vena cava (Fig.9.7a), an aberrant left pulmonary vein in the position of an accessory
Acquired Disorders of the Tracheobronchial Nonneoplastic
Tracheal Stenosis
Tracheal narrowing may affect the entire trachea or portions of it and may result from pulmonary disease, systemic disease, or a process confined to the trachea itself. CT can be used to distinguish an intrinsic lesion from an extrinsic process (compression) and exclude malignant disease. A "saber-sheath" trachea occurs in patients with COPD and is related to an abnormally high intrathoracic transmural pressure.
•
System
CT Morphology
Tracheal stenosis mayor may not be associated with wall thickening (Table 9.7). CT morphology is often nonspecific, so histological evaluation is frequently necessary when wall thickening is present. CT is performed to document the affected tracheal segment, define the degree of stenosis, and visualize involved bronchial segments. A "saber-sheath" trachea is present when the tracheal area is reduced by 70% and the transverse (coronal) tracheal diameter is reduced to less than one-half of its sagittal diame-
295
9 Lungs and Tracheobronchial Table 9.7 Nonneoplastic Tracheal Intrinsic
stenosis
tracheal
System stenosis:
etiology and CT features
without wall thickening
Rare, 2-4 cm in length, smooth or irregular
Idiopathic Saber-sheath
Intrathoracic trachea, long segment involved, elliptical to triangular cross-section, coronal diameter < half of sagittal diameter
trachea
After tracheostomy
Region of stoma or lower edge of tube, hourglass configuration
After intubation
At cuff site or tube tip, hourglass
Post-ischemic
after lung transplantation
Smooth,
Post-traumatic
configuration
Region of anastomosis hourglass
configuration
Extrinsic Vascular
Detection of vessel (aberrant right subclavain duplicated aortic arch, pulmonary sling)
Goiter
Relation to thyroid gland, intense contrast ment, cysts, calcium
Tracheal
stenosis
Osteochondroplastic
with wall thickening t rac he ob ron ch op ath ia
Calcified tracheal
Polychondritis
Submucosal
Scleroderma
Diffuse thickening,
Wegener's
granulomatosis
cartilage,
irregular tracheal
nodularity of anterolateral
central and peripheral stenoses
Rare, smooth or irregular wall thickening, calcified cartilage, focal or diffuse wall thickening, laryngeal involvement Submucosal
Sarcoidosis
Mural granulomas, smooth or irregular, laryngeal involvement (tuberculosis,
fungal, scleroma)
nodules, diffuse>
local
Rare, smooth or irregular, tuberculous
cavities
Smooth or irregular
Postirradiation
ter. Typically ber beginning ing extending
contour
walls
Amyloidosis
Postinfectious
artery,
enhance-
there is abrupt change of the caliat the thoracic inlet, the narrowfor the entire intrathoracic length.
Bronchiectasis Bronchiectasis is a localized dilatation of the bronchial tree. The changes may be very subtle and may be regional or widespread. Today, bronchiectasis is commonest in patients with cystic fibrosis, mucociliary dysfunction, or an immunodeficiency disease (Table 9.S). Reversible bronchiectasis may occur in the setting of pneumonia and may resolve completely in four to six months. HRCT is superior to standard CT
in the detection of bronchiectasis, providing sensitivity and specificity greater than 90%.
a
CT Morphology
One criterion for bronchiectasis is an internal bronchial diameter that is enlarged in relation to the corresponding artery (normally the artery has a slightly larger caliber than the bronchus). But since the ratio of bronchial to arterial diameter shows a relatively large range of physiologic variation (values found in a normal population range from 0.5 to 1.4), two additional CT criteria for bronchiectasis have been introduced:
Bronchiectasis
Table 9.8
Etiology of bronchiectasis
Congenital diseases • Structural wall defect: Kartagener's syndrome • Mucociliary dysfunction: Williams-Campbell syndrome • Cystic fibrosis • Alpha-l-antitrypsin deficiency • Congenital or acquired immune deficiency
Infections • • • • • •
Allergic bronchopulmonary aspergillosis Chronic granulomatous infection (tuberculosis) Measles Whooping cough Swyer james syndrome Chronic aspiration or inhalation
Bronchial obstruction • Tumor • Foreign body • Retained secretions • Lymphadenopathy
or compression
(mucoid
impaction)
Pulmonary fibrosis (traction bronchiectasis)
• failure of the bronchi to taper toward the periphery (a bronchus has the same diameter as its parent branch for a distance of more than 2 cm); and • visualization of bronchi within "1 cm of the costal pleura and visualization of bronchi abutting the mediastinal pleura. A bronchoarterial diameter ratio can be considered a reliable sign of bronchiectasis by itself only when it is at least 1.5. When it is between 1 and 1.5, it must be present in several airways or be associated with other findings such as bronchial wall thickening and lack oftapering, or both. Care should be taken to assess the smallest crosssectional airway diameter, to avoid comparisons near bifurcations of bronchi and vessels and not to overcall bronchiectasis in areas with decreased vessel size as a result oflocal vasoconstriction. On 10 mm collimated scans normal bronchi are seen only in the inner third of the lung, on HRCTimages Table 9.9
normal bronchi can be seen in the inner twothirds of the lung. Optimal HRCT assessment is obtained using a window level of -700 HU and a width of /\lOOOHU; smaller window width leads to artificial bronchial wall thickening. An extended window width (1500 HU) may be used routinely. However, it reduces the contrast between lung parenchymal structures and makes slight attenuation differences less conspicuous. Vessels may have double contours due to pulsations or respiratory motion, and this can mImIC bronchiectasis (see Fig.7.36a). Bronchiectasis may be missed in areas of dense consolidation or atelectasis. Three morphologic types of bronchiectasis are identified in the Reid classification: cylindrical (tubular, fusiform), varicose, and cystic (saccular) (Table 9.9). Depending on the relation of the bronchi to the plane of section, CT scans of cylindrical bronchiectasis may show signet -ring configurations in cross sections or "tram lines" when the bronchus is scanned in-plane. Tram lines often can be traced far into the periphery (Fig.9.8a, d). The rare varicose form is recognized on CT by marked variations in the calibers of airways that are imaged in longitudinal section (Fig. 9.8 b). Areas of cystic bronchiectasis often show retained secretions with fl ui d levels (Fig.9.8c). Air-filled cystic lesions in atelectatic lung areas have a cluster-of-grapes appearance. Dilated bronchi that are filled with secretions appear as Y- or V-shaped structures of soft-tissue density [mucoid impaction, Fig.9.8c). Thickening of the bronchial walls indicates a bronchitic component or mucosal edema (Fig.9.8b). Mucoid impaction in bronchiectasis is particularly common in patients with cystic fibrosis and allergic bronchopulmonary aspergillosis (ABPA).
Bronchiectasis
Cylindrical
Varicose
Cystic
Mildest form Tram lines (horizontal) Signet ring sign (vertical)
Rare (Swyer -james) String-of-beads contours
Most severe form Transverse diameter> 1 em Frequent wall instability Clustered cysts Air-fluid levels
297
298
9 Lungs and Tracheobronchial System
b
a
c
d
Fig. 9.8 Bronchiectasis. (a) Cylindrical bronchiectases (signet ring sign, arrow), (b) Extensive mucous plugging in right lower lobe bronchiectases (arrowheads). The bronchi are larger than the accompanying arteries. (c) Varicose and cystic bronchiectasis with fluid levels. (d) Semicoronal reformat from a low dose multislice CT data set (4x1/6, 1.2mGy) demonstrates cylindrical
and varicose bronchiectasis in a patient with cystic fibrosis. There is bronchial wall thickening (sign of chronic inflammation) and retained secretions (mucoid impaction). The secretion-filled bronchi produce typical Y- or V-shaped structures. The subpleural "tree in bud" pattern is also evident.
Bronchiolectasis is detectable only if the dilated bronchioles are filled with mucus. CT shows thin, branching structures of soft-tissue density in the subpleural lung zone (5-10 mm) that have small bulbous terminations ("tree in bud" pattern, Fig. 9.8d).
Table 9.10
The isolated circumscribed dilatation of a mucus-filled bronchus may be an indirect sign of a small intraluminal tumor (e.g., carcinoid, bronchialcarcinoma). Differentiation of mucoid impaction from arteriovenous malformations is easily achieved after intravenous administration of contrast material.
Bronchiolitis
A variety of pulmonary diseases are characterized predominantly by inflammation of membranous and respiratory bronchioles. Various classifications schemes are found in the literature that differ with respect to whether they are based on etiologic, histologic, clinical or radiologic features (Table 9.10). From a clinical point of view it is most useful to differentiate infectious bronchiolitis from non-infectious, the latter being related to immunologic reactions.
Different forms of bronchiolitis
Etiology
Morphology
Acute infectious
Tree-in-bud, nodules
Chronic inflammatory (asthma, chronic bronchitis)
Bronchial wall thickening, tree-in-bud
Panbronchiolitis
Marked tree in bud, air trapping, bronchial wall thickening
Respiratory
Centrilobular nodules (predominantly in lower lobes)
bronchiolitis
Respiratory bronchiolitis with interstitial disease
(RBIO)
acinar
Centrilobular nodules, emphysema, bronchial wall thickening
Follicular bronchiolitis
1-3 mm (occasionally 1-2 ern) subpleural illdefined nodules
Obliterative bronchiolitis (bronchiolitis obliterans)
Air trapping, pattern
mosaic
Acute Infectious Bronchiolitis
Acute infectious bronchiolitis is most commonly the result of infection by viruses (respiratory syncytial virus, adenovirus), Mycoplasma pneumoniae and Chlamydia species. Immunocompromised patients develop acute Aspergillus bronchiolitis or endobronchial spread of tuber-
Bronchiolitis
culosis. In principle, each bacterial infection can cause acute bronchiolitis if confined to the small airways. •
CT Morphology
The hallmarks of infection-related bronchiolitis in HRCT are small centrilobular nodules with branching lines (tree- in-bud) that represent thickening of the bronchiolar wall or filling of dilated bronchioles with granulation tissue, mucus, or pus (Fig. 9.9 a, b). The tree-in-bud sign is most strikingly appreciated in the peripheral one-third of the lung parenchyma. In some cases it is accompanied by scattered areas of ground-glass attenuation or consolidation. Acute infectious bronchiolitis has to be differentiated from chronic inflammatory conditions (asthma, chronic bronchitis), diffuse panbronchiolitis and respiratory bronchiolitis, which present with centrilobular small nodules (see Fig. 9.9 c): The centrilobular nodules in infectious bronchiolitis are usually well-defined and have sharp margins whereas those seen in patients with respiratory bronchiolitis typically have poorly defined margins. In infectious bronchiolitis nodular opacities are usually combined with branching linear opacities (tree- in-bud). In infectious bronchiolitis centrilobular nodules may enlarge then resembling acinar nodules that can come together to form rosettes.
a
b
Fig.9.9 Bronchiolitis. (a) Infectious bronchiolitis with tree-in-bud-pattern and air-trapping (expiration) indicating disease of the small airways, (b) Infectious bronchiolitis with small nodules and tree-in-bud. (c) Respira-
Respiratory Bronchiolitis and Related Disease
Respiratory bronchiolitis and respiratory bronchiolitis associated interstitial lung disease (RBILD) have recently been grouped together with desquamative interstitial pneumonia (D IP) as part of a single spectrum of diffuse interstitial infiltrative pulmonary disease (see Table 9.34) since they have clinical similarities and significant overlapping of histological and HRCT features. Both diseases are strongly associated with cigarette smoking, underlining the concept that all three diseases represent different degrees of severity of small airway and parenchyma! reaction to cigarette smoking. Radiological as well as histological features of respiratory bronchiolitis and extrinsic allergic alveolitis are also in a continuum, since bronchiolitis is a prominent component of extrinsic allergic alveolitis. Inhalational injury caused by S02, oxides of nitrogen, or a variety of other gases or fumes is characterized by identical CT morphology. •
CT Morphology
Respiratory bronchiolitis is characterized by illdefined centrilobular nodular opacities (Fig.9.9c) producing an image that may be virtually indistinguishable from acute extrinsic alveolitis (see Fig.9 .56a). In most patients, the upper lobes are involved predominantly or even exclusively (as opposed to extrinsic allergic alveolitis) .
c
d
tory bronchiolitis with centrilobular nodules and bronchial wall thickening, (d) Follicular bronchiolitis with multiple ill-defined centrilobular nodules, predominantly in subpleural location.
299
9 Lungs and Tracheobronchial System In addition to small centrilobular nodules, patients with RB-ILD show more prominent bilateral areas of ground-glass attenuation mostly in a patchy distribution and subtle findings offibros is (honeycombing, intralobular lines) predominantly in the lower lung zones, An upper lobe centrilobular emphysema is frequently present; bronchial wall thickening may be an associated finding, In contrast to DIP, lung volumes are usually normal, Similar findings as seen in RB and RB-ILD are also seen in asymptomatic smokers, but the findings are usually less extensive, Clinical and radiological findings improve promptly when the patient stops smoking and is treated with corticosteroids,
Diffuse Panbronchiolitis (DPB) Diffuse pan bronchiolitis is a disease ofunknown etiology, It is characterized by luminal obliteration through accumulation of macro phages, The disease has been recognized almost exclusively in East Asia, The patients present with a chronic productive cough and progressive dyspnea, The long term prognosis is poOL CT Morphology
The hallmarks are extensive diffuse centrilobular nodules with branching lines (tree-in-bud), marked bronchial wall thickening, bronchiectasis, and localized areas of decreased attenuation (air trapping) and perfusion,
gates (Fig,9,9d), The nodules are usually small (1-3 mm). and only occasionally larger (1-2 em). Tree-in-bud, bronchial wall thickening, and patchy areas of low attenuation are associated findings,
Bronchiolitis Obliterans (Constrictive Bronchiolitis) Bronchiolitis obliterans (BO) -also termed constrictive bronchiolitis-is characterized by proliferation of submucosal and peri bronchiolar fibrous tissue resulting in a more or less concentric narrowing of the airway lumen (without intraluminal granulation) through to complete obliteration in extreme cases, The idiopathic form is rare, and the condition can have various causes (Table 9.11), Patients present clinically with progressive obstructive lung disease that frequently shows no abnormalities on plain chest radiographs, Bronchiolitis obliterans is a separate entity from bronchiolitis obliterans with organizing pneumonia (BOOP, see P 321), Table
9.11
Etiology
of obliterative
Inhalation of toxic gases • (e.g., nitrogen dioxide, chorine) Infectious
bronchiolitis
nitric oxide,
ammonia,
agents
• Mycoplasma (in children) • Viruses (in adults) Drug reactions • Penicillamine • Lomustine
Collagen diseases
Follicular Bronchiolitis Follicular bronchiolitis is characterized by the presence of abundant lymphoid tissue within the walls of bronchioles and to some extent of bronchi. The finding is not specific and can be found in association with connective tissue disease (Sjogren's disease and rheumatoid arthritis), imm unodeficiency syndrome, and systemic hypersensitivity reactions, CT Morphology
CT shows nodular opacities mainly in a peribronchovascular or subpleural distribution, consistent with the presence of lymphoid aggre-
• Rheumatoid • Scleroderma
arthritis
-SLE Transplantation
(lung, heart,
• Graft-versus-host • Chronic rejection In setting
of other
lung diseases
• BOOP • Chronic
bronchitis • Cystic fibrosis • Bronchiectasis • Extrinsic allergic alveolitis Ulcerative Idiopathic
colitis
bone marrow)
reaction
•
CT Morphology
The hallmark of bronchiolitis obliterans is air ~rapping: the density of the affected lung areas IS normal or reduced on inspiratory scans, and on expiratory scans the normal parenchyma shows increased density while affected areas show little or no increase in attenuation (due to check-valve obstruction, Fig.9.10a, b). Regions of air trapping may show decreased vascularity with redistribution and consequent vasodilatation in other lung areas, giving rise to a "mosaic" pattern (Fig. 9.10c). . Peribro?chiol.ar fibrosis is rarely detected directly but ISmamfested by centrilobular punctate or branching structures that are most conspicuous in the peripheral third of the lung. Bronchiectasis (mainly central), bronchial wall thickening, and signs of interstitial fibrosis are common associated findings but are not primaryfeatures ofbronchiolitis obliterans. It can be difficult to distinguish the mosaic pattern of bronchiolitis obliterans from a geographic pattern of ground-glass opacities. Ground-glass areas of increased density are pathologic and are seen in the setting of acute alveolar diseases and interstitial inflammatory proc~s:es leading to fibrosis. Ground-glass opacities show a very slight, homogeneous density increase that does not obscure pulmonary vessels or parenchyma! structures.
Mosaic pattern is seen in bronchiolitis obliterans (1), an acute inflammatory process (2) or in chronic thromboembolic disease (3), which can be differentiated from each other by expiratory CT scanning (Table 9.12): 1. In the case of air trapping, the areas of increased attenuation are normal while the are~s of decreased attenuation are patholOgIC. They do not show an increased attenuation on expiratory scans but become even more pronounced in stead. 2. In infiltrative disease, ground-glass opacities represent areas of pathologically increased density. In expiratory views, they show an attenuation increase like that observed in normal lung parenchyma. 3. Pulmonary vessels are decreased in size in the low attenuation areas when air trapping or vascular disease are present, but are normal in acute infiltrative disease. Expiratory scans help to exclude air trapping in hypo attenuating areas.
r
a
b
Fig. 9.10. Bronchiolitis obliterans is characterized by air trapping due to check-valve obstruction, (a) Inspiratory CT shows sharply demarcated areas of different lung density with vascu lar rarification in the hyperlucent areas (mosaic pattern), (b) Expiratory CT shows marked increase of lung density in the normal parenchyma
c while lung areas with air trapping show no change in volume or density, (c) Obliterative bronchiolitis secondary to a drug reaction. CT reveals a patchwork of sharply marginated areas of increased lucency (air trapping) adjacent to areas of normal or apparently decreased lucency with dilated blood vessels ("mosaic pattern").
9
Lungs and Tracheobronchial System
Table 9.12
Differential
diagnosis
of mosaic pattern
Bronchiolitisobliterans
Infiltrativedisease
Chronicpulmonary embolism
Small air ways disease
Ground-glass
Vascular
Airtrapping
Density of normal and opacified areas increases with expiration
No air trapping
Normal vessel
Vessel diameter is decreased areas of hypoattenuation
in expiration
Vessel diameter is (slightly) decreased in areas of hypoattenuation
opacification
diameters
obliteration
in
--
Allergic Bronchopulmonary Aspergillosis (ABPA)
Allergic bronchopulmonary aspergillosis is caused by a pathologic immune response to colonization of the bronchi by Aspergillus species (not to be confused with invasive aspergillosis). Patients with asthma or cystic fibrosis are predisposed. Other rarer forms of hypersensitivity reaction to Aspergillus are extrinsic allergic alveolitis and a Loffler-like syndrome. 1
•
CT Morphology
The initial finding is that of migrating alveolar infil trates in a setting of pneumonitis. The infiltrates are usually bilateral and show a predilection for the upper lobes. The hallmarks of ABPA are bronchiectasis and mucoid impaction involving predominantly the segmental and subsegmental airways of the upper lobes (Fig. 9.11). Retained secretions in segmental bronchi lead to V- or Y-shaped opacities that may persist for months. Focal areas of mucoid impaction and bronchiectasis may mimic a nodule or mass lesion on CT. CT typically shows branching tubular opacities that emanate from the hilum (finger-in-glove appearance). Bronchiectasis may also be widespread and involve more peripheral bronchi; involvement of main and lobar bronchi, however, is very uncommon. The combination of bronchiectasis on CT and blood eosinophilia is typical of allergic bronchopulmonary aspergillosis. Segmental atelectasis and peripheral subsegmental atelectasis give rise to patchy areas of consolidation. Postobstructive cavitation may occur. Late changes consist of hyperinflation and fibrosis. Involvement of small bronchioles leads to a "tree- in-bud" pattern and air trapping on expiratory scans.
255 137 1
o
a
b
Fig. 9.11 Allergic bronchopulmonary aspergillosis (ABPA) with upper lobe tubular bronchiectasis.
Bronchiectasis is seen more frequently in asthmatic patients with ABPA than in asthmatic patients without ABPA. However, bronchiectasis may also occur in uncomplicated asthma, and may not be present in all patients with ABPA.
Benign Tumors Tracheobronchial Tumors Benign Tumors
Benign neoplasms of the trachea and bronchi are very rare and are always intraluminal. Commonest are mesenchymal tumors (hemangioma, hamartoma), but occasional paragangliomas, neurofibromas, and papillomas can occur. The detection of extraluminal tumor growth on CT excludes a benign lesion. Tracheobronchial papillomatosis is caused by a viral infection (papilloma virus). It is usually confined to the larynx, rarely spread of papillomas to the trachea, bronchi and lung occurs. Papillomas may be solitary or multiple. Histologically they consist of central fibrovascular bundles covered by layers of well-differentiated squamous cells. Transformation oflesions into invasive squamous cell carcinoma is known. •
a
Fig. 9.12 Tracheal lesions, (a) Shaded surface display of a benign intraluminal tracheal papilloma (2/4/2). (b) Coronal reformatted image of a malignant esophageal tumor invading the trachea (3/5/2). Note the intraluminal and extraluminal components of the tumor.
CT Morphology
They have a pathognomonic CT morphology with nodules that frequently cavitate (Fig. 9.13).
Benign neoplasms usually have smooth, welldefined margins and are less than 2cm in diameter. They do not display specific morphologic CT features, however, and biopsy is necessary to establish a diagnosis (Fig. 9.12). Polypoid and sessile masses protrude into the lumen and may cause subtotal obstruction. Tracheobroncbial papillomatosis may present as polypoid lesions on the surface of the tracheobronchial system, but frequently intrapulmonary papillomas are the only findings on CT.
a
b
b
Fig.9.13 Intrapulmonary papillomatosis. (a) Intrapulmonary cavitating, irregular nodules (HRCT). (b. c) Lowdose multislice CT (4x1/6, CTDlvoi=1.8mGy) of a
Particularly in elderly patients with an obtunded cough reflex, secretions may accumulate in the central airways and create a pseudomass appearance. Usually these can be identified by their teardrop-shaped configuration, and coughing tends to displace the collection or cause it to disappear.
c
~ d
young girl with multiple small cavitating nodules due to papillomatosis. (d) Compare to cavitating intrapulmonary metastases.
303
9 Lungs and Tracheobronchial
System
Malignant Tracheal Tumors
Carcinoid Tumor
Primary malignant tumors of the trachea are extremely rare (180 times rarer than bronchial tumors). Clinical symptoms are a late finding, so generally these tumors are in an advanced stage when diagnosed (significant luminal narrowing in more than 75%, mediastinal invasion in 40%). The three commonest tracheal malignancies are squamous cell carcinoma, adenoid cystic carcinoma, and mucoepidermoid carcinoma. Secondary malignant tumors involving the trachea include the thyroid and the esophagus.
Carcinoids are classified as neuroendocrine tumors. They account for less than 4% of all bronchial neoplasms. They are locally invasive and metastasize to regional lymph nodes, bone, and liver. Approximately 90% are endobronchial tumors that involve the central airways and may become obstructive. Only 10% involve the lung parenchyma. According to their histology and their biological behavior, typical, carcinoids are differentiated from the more aggressive atypical carcinoid (metastasizes into the lymph nodes) and the small-cell carcinomas (mostly associated with bulky mediastinal disease). Radion uclidecoupled somatostatin analogs (octreotide scan) are used to visualize somatostatin-receptorbearing tumors.
•
CT Morphology
Primary tracheal tumors grow exophytically and by infiltration. They may spread over a distance of several centimeters and sometimes up to 10cm. The full extent of intramural (submucosal) spread tends to be underestimated with CT. While squamous cell carcinoma causes irregular sometimes nodular wall thickening, adenoid cystic carcinoma leads to smooth and long-standing wall thickening and consecutive lumen narrowing. For surgical planning, it is crucial to assess the involvement of the infraglottic larynx (indication for laryngectomy) and of the principal bronchi.
a
b
Fig. 9.14 Axial section (a) and shaded surface display (b) of an intraluminal carcinoid at the level of the carina (3/5/2). It is typical to find the smooth-bordered mass located in proximity to bronchial divisions. A carcinoid
•
CT Morphology
Endobronchial carcinoids appear on CT as wellcircumscribed nodular masses that occur predominantly at bronchial divisions (Fig. 9.14). They frequently cause airway obstruction and may be associated with air trapping, bronchiectasis, mucoid impaction, or pneumonia. Calcifications are seen in 30% and due to their good vascularization tumors show a marked enhancement after application of contrast.
c
d tumor (arrowheads) in another patient (c) has caused atelectasis of the anterior segment of the right upper lobe (d). Carcinoid tumors are hypervascular and thus may be hard to distinguish from adjacent atelectasis.
Bronchial Carcinoma 305 sis, idiopathic pulmonary fibrosis, scleroderma), The therapeutic approach depends on the local tumor extent, the invasion of mediastinal and hilar lymph nodes, and the presence or absence of distant metastases, Bronchial carcinoma is characterized by early lymphogenous and hematogenous spread, Lymph node metastases are commonest with small-cell carcinomas, The highest rates of metastasis are to the liver (up to 40%), adrenals (up to 40%), brain (up to 43%), bone (up to 33%), kidneys (up to 25%), and abdominal lymph nodes (up to 30%), The incidence of metastases depends on histology and primary tumor stage,
Bronchial Carcinoma Bronchial carcinoma is the most frequent tumor in the statistics of neoplasm-related death rates and the commonest tumor of the central tracheobronchial system. The five-year survival rate is only 8 to 13% lower as compared to that of other neoplasms (colorectal carcinoma=62%, breast neoplasm=85%), Active cigarette smoking increases the risk for developing a bronchial carcinoma by factor 10, passive smoking by factor 2, The histological classification was revised in 1999 and differentiates the small-cell carcinoma (SCLC=20%), the adenocarcinoma (30%), the squamous cell carcinoma (25%) and the large cell carcinoma (10%) as most frequent tumor types (Table 9.13), Certain occupational agents (asbestos, coal, arsenic, chromium, chloromethyl ether, mustard gas) are known to lead to increased risk of lung cancer, the most important of these being asbestos, A combination of asbestos exposure and Cigarette smoking is multiplicative, Other diseases that have an increased prevalence of lung cancer are characterized by diffuse or focal pulmonary fibrosis (e,g" tuberculosis, asbesto-
Table 9.13
Staging The staging of bronchial carcinoma (Tables 9.149.16) is useful in selecting patients for surgical resection, planning radiotherapy, and monitoring the response to chemotherapy, Stages 1 and 11 are amenable to resection, Stage III was divided into two classes: lllA includes tumors with limited mediastinal or chest wall invasion and are considered potentially re-
Bronchiogenic carcinoma Frequency
Histology
Etiology and
Region of
Squamous
cell carci-
30-35%
noma (subtype:superiorsulcus tumor)
Special features
predilection
Prognosis
Best prognosis; local
2/3 central
Central: atelectasis
metastases nodes
endobronchial 1/3 peripheral
strucnve pneumonitis Peripheral: thick walled
to lymph
or postobcavi-
tation (30%), solitary nodule Small-cell
carcinoma
20-25%
(sci c)
Strong association with smoking; most aggres-
Central (85%)
sive with early lymph node Large-cell
(undifferen-
<5%
tiated) carcinoma Large-cell neuroendocrine carcinoma
<5%
Grows along pre-existing tissue spaces; hilar or perihilar mass (massive nopathy)
metastasis
Strong association with smoking, poor progno-
Peripheral
Early lymph node
Central
(70%)
Large tumors
lymphade-
(>
4cm)
Extended intratumoral (30-7fP/O)
necro-
sis
metastases
(LCNEC) Adenocarcinoma
33-50%
Association
with fibroearly
sis; metastasize
Peripheral (55%)
Slowly growing, solitary nodule or mass; often large
and spiculated Bronchioalveolar cell carcinoma (special from of adenocarcinoma)
2-5%
Good small
prognosis tumors
for
Parenchyma
May imitate
inflammatory
infiltrate, hazy ill-defined ground gram,
glass;
air broncho-
consolidation, nodule (solitary or multiple)
306
9 Lungs and Tracheobronchial Table 9.14 1997)
System
TNM staging of bronchial carcinoma (UICC,
TO
No direct evidence of a primary tumor
T1
Intrapulmonary tumor :s 3 ern, surrounded by lung parenchyma or visceral pleura, no infiltration of the intermediate or main bronchi (except for tumors confined to bronchial wall)
T2
T3
T4
Tumor> 3 cm, or tumor involving the intermediate or main bronchus but at least 2 cm distal to the carina, or tumor involving the visceral pleura; associated atelectasis or postobstructlve pneumonia must involve less than an entire lung Tumor of any size involving the main bronchus within 2 cm of the carina, or tumor directly invading the parietal or mediastinal pleura (including superior sulcus tumors) diaphragm, chest wall, or pericardium, associated atelectasis or postobstructive pneumonitis involving the entire lung Tumor of any size invading the heart, great vessels, trachea, carina, esophagus, or spinal column, or the presence of malignant pleural effusion, satellite lesions within the some lobe as primary tumor
NO
No metastasis to regional lymph nodes
N1
Metastasis to ipsilateral bronchial or hilar lymph nodes
N2
Metastasis to lpsilateral mediastinal and subcarinallymph nodes
N3
Metastasis to scalene or supraclavicular lymph nodes, or metastasis to contralateral mediastinal or hilar lymph nodes
M1
Distant metastases, satellite lesions in different lobe as primary tumor (ipsi- or contralateral)
sectable, while IIIB is considered non-resectable. T4 means infiltration of vital mediastinal structures (e.g .. heart, trachea, esophagus, great vessels). N1 alters the prognosis but not the surgical approach. N2 diseases are considered resectable but usually patients undergo neoadjuvant chemotherapy sometimes combined with radiation therapy before operation. N3 disease precludes resection. It is therapeutically important to differentiate between stage II and iliA tumors (lobectomy versus pneumonectomy) and between stage lilA andlllB disease (resectable versus unresectable) (Fig. 9.15 b). Transfissural tumor spread, pulmonary vascular invasion, the invasion of a main bronchus, or the involvement
Table 9.15 Size criteria for mediastinal lymph nodes (Glazer 1985 adopted to AJCC/UICC classification 1996) Station
Description
Normal size
2R
Right upper para tracheal nodes
:s7mm
2L
Left upper paratracheal nodes
:s7mm
4 R
Right lower paratracheal nodes
:s 10 mm
4 L
Left lower paratracheal nodes
:s 10 mm
5
Subaortic aortopulmonary nodes
:59mm
6
Para-aortic (anterior mediastinal) nodes
::=;8mm
7
Subcarinal nodes
8R
Right para-esophageal nodes
:s 11 mm :s 10 mm
8L
Left para-esophageal nodes
:s;7mm
10 R
Right hilar nodes
:s 10mm
10 L
Left hilar nodes
:s7mm
Table 9.16 Stages of bronchial from Mountain, 1998)
T1 T2 T3 T4 M1
carcinoma
(adapted
NO
N1
N2
N3
M1
IA IB liB IIIB IV
IIA liB lilA 1116 IV
IliA lilA IliA IIiB IV
IIIB IIiB 1116 IIIB IV
IV IV IV IV IV
of upper and lower lobe bronchi, preclude a lobectomy and necessitate pneumonectomy (iliA). Invasion of a main bronchus less than 2 ern from the carina signifies a T3 tumor (Fig.9.15c). Involvement of the carina or involvement of contralateral lymph nodes indicate advancement beyond stage iliA and preclude surgical intervention. The non-small-cell neoplasms are staged according to the TNM classification (last modified in 1997). A small-cell carcinoma is classified as "limited disease" (stages I through iliA). Stage IIIB, denoting "local extensive disease," is still considered to be localized from a radiotherapeutic standpoint. Stage IV denotes "extensive disease" with detectable distant (ex-
Bronchial Carcinoma
a
b
Fig. 9.15 Central bronchial carcinoma, (a) Tumor has encased the intermediate bronchus (stage I) without invading the pulmonary artery. Small lymph nodes, not considered significant, are visible on the contralateral hilum (arrow), (b) Stage IliA bronchial carcinoma has obliterated the right lower lobe bronchus (arrow) and invaded the pulmonary artery, necessitating a pneumonectomy rather than lobectomy. (c) Tumor has en-
trathoracic) metastases (cervical or axillary lymph nodes or other distant metastases). •
CT Morphology-
T Staging
CT has an accuracy of only 50% to 70% in evaluating the local invasiveness (T staging) of bronchial carcinoma. CT and MRI are equally limited in their ability to assess involvement of a main bronchus, the carina, or the trachea, due to a lack of precision in the delineation of submucosal tumor spread. Intrabronchial structures of soft-tissue density, bronchial wall thickening, and an irregular reduction in bronchial caliber are indicators of bronchial tumor involvement (Fig.9.15). Contrast-enhanced CT is essential, though not always successful, in differentiating bronchogeniC tumor from areas of poststenotic atelectasis and pneumonia. Atelectasis shows more intense and more uniform enhancement than tumor-involved lung tissue on postcontrast scans. Pneumonia shows similar characteristics, though to a considerably lesser degree. The following findings are more suggestive of an occult central tumor with postobstructive consolidation rather than of simple pulmonary consolidation due to pneumonia: • S-shaped distortion of the lung fissure due to tumor-induced lobar atelectasis;
c
d
cased a long segment of the anterior upper lobe bronchus and the right main bronchus to within 2 cm of the carina (stage IliA disease), (d) Inoperable tumor (stage IV) has invaded the mediastinum and encased the aorta. It has also invaded the pulmonary artery and partially obstructed the left main bronchus. The atelectasis shows greater postcontrast enhancement than the tumor itself (arrowheads).
• Loss of volume due to lobar atelectasis, or the presence of an air bronchogram (in patients over age 35); • 'Irregular stenosis in the central bronchial tree; • Hilar lymphadenopathy; • Insufficient regression of pneumonia in response to antibiotics. Reconstruction of thin axial sections and MPR are particularly worthwhile in patients with focal lung lesions that are in the vicinity of pleural interfaces, because they frequently provide additional diagnostic information with respect to segment localization, pleural dissemination, and transfissural tumor growth (Fig. 9.16). It was demonstrated that the sensitivity of CT scanning for the presence of extension through the major fissure was 57% for thick 10mm sections, but could be increased to 87% with thin 2 mm sections. Additional multiplanar reformations (MPR) could further improve sensitivity to 100%. For the minor fissure, 6/51 cases were inconclusive on axial section but only one remained so on MPRs. Three-dimensional images based on shaded surface displays were found to facilitate the differentiation of simple pleural tags (i.e., bands extending from the lesion to the visceral pleura; T-stage =s T1) from pleural puckering associated
308
9 Lungs and Tracheobronchial System
a
b
c
d
Fig. 9.16 Multiplanar reformations improve staging of bronchial carcinoma (4x1/6, 7.2 mGy). (a) Transfissural tumor growth and broad tumor adherence along the superior vena cava and the right atrium, (b) Incomplete lobar fissures support translobar spread of carcinomatous lymphangitis, (c) Large tumor infiltrating the pulmonary artery (arrow) with pathologic sub-
carinal lymph nodes (T4N2, stage IIIB disease) and a contralateral acute pulmonary embolus (arrowhead). (d) Bilateral metastatic lymph nodes in the right perihilar region, pretracheal, and in the aortopulmonary window. These nodes are round but smallerthan 10 mm in short axis diameter.
with an indrawn locally thickened pleura seen with visceral pleural invasion (> T2). While the sensitivity of pleural puckering was high, specificity was only 76% because it was not only seen with tumor invasion but also with reactive fibrotic changes.
sidered a reliable indicator for intrapericardial tumor extension, while infiltration of the inferior pulmonary vein does not necessarily mean that the pericardium is infiltrated. Although chest wall and limited mediastinal invasion of lung cancer are no longer contraindications to surgical treatment, the mortality rate associated with en bloc resection of the tumor and the contiguous structures is still substantial. Therefore, information about mediastinal and chest wall invasion is still one of the important factors in the clinical decision to perform surgery. Gross extrapulmonary chest wall invasion does not impose problems. However, accurate distinction between tumor contiguity with adjacent extrapulmonary structures and actual invasion is often difficult on static images, especially when there is no obvious soft tissue mass in the mediastinum or chest wall. Contiguity of the tumor with parietal pleura was found not to be equivalent to definite invasion of the chest wall, even when associated with a pleural thickening contiguous to the tumor. Apart from gross invasion, focal rib destruction is the only sign that is "100% sensitive for chest wall invasion. All other findings such as pleural thickening, an obliterated epipleural fat plane, an obtuse angle between the tumor and chest wall, more than 3 cm of contact between the tumor and chest wall
Presence of a pleural effusion requires the exclusion of pleural involvement (pleural tap, if negative PET). Pleural involvement corresponds to stage IV and makes the tumor irresectable. Mediastinal invasion represents stage T4 and in almost all cases makes the tumor irresectable. It should be diagnosed if CT shows tumor encasement or constriction of vital mediastinal structures, e.g., of the large mediastinal vessels, the heart, the esophagus, spine, or the trachea, or if the tumor is transgressing the midline (Fig. 9.15d). Local infiltration of the mediastinal fat after transgression of the mediastinal pleura is not considered a criterion for irresectability. MRI is superior to CT in differentiating tumor from fat and will reveal even minimal changes, but it is no better than CT for distinguishing tumor infiltration from an inflammatory reaction of the epipleural or mediastinal fat. Infiltration of the superior pulmonary vein is con-
Bronchial Carcinoma and a ratio between tumor-pleura contact and tumor diameter of > 0.9 showed sensitivities between approximately 60% and more than 90%. Obliteration ofthe epipleural fat plane may be due to tumor invasion or inflammatory reaction and is equally unreliably differentiated with MRI and CT (sensitivity/specificity of 85%/87%withCTand90%/86%withMRI) . In general, there is a superiority of \ mm slices as compared to 10mm slices for the delineation of anatomic chest wall structures and thus for evaluating the presence of chest wall involvement. For the delineation of the extrapleural fat plane and the assessment of soft tissue structures a smoothing standard reconstruction algorithm was superior to the high resolution algorithm due to its lower noise level. Both, CT and MRI have been employed to use dynamic imaging during different phases of respiration to assess the respiratory shift that is defined as a change in the relative location between the peripheral lung tumor and the chest wall with deep inspiration and expiration. Expiratory dynamic CT was found to provide accurate information but only for tumors located in the middle and lower lobes. It has also to be noted that only absence of tumor invasion can be confidentially diagnosed in those patients in which free movement of the tumor directly proves lack of attachment between the visceral and parietal pleura. Benign fibrous adhesion may cause false positive results by simulating tumor infiltration. Such limitation holds true for
a
both CT and MR imaging as well as alternative approaches using ultrasound or pneumothorax CT. The latter is based on the idea that tumor infiltration can be differentiated from the parietal pleura on the basis of air in the absence of tumor infiltration. However, this technique is invasive, it may cause symptomatic pneumothorax (described in 4/43 patients) and therefore has never found broad application. Owing to its superior contrast resolution, MRI used to be the method of choice for imaging of superior sulcus tumors. The use of surface coils provides high resolution images and chest wall invasion is best depicted as a disruption of the extrapleural fat line (bright signal of fat in Tl-weighted images). MRI was also thought to be more accurate than CT in depicting chest wall involvement from superior sulcus tumors with respect to involvement of the subclavian vessels, the brachial plexus and the vertebra. However, high quality spiral CT (thin sections, bolus injection of contrast agents, and sagittal and coronal reformats) provide the same anatomic information of the anatomical environment of the plexus, and MRI may be reserved for patients with extension into the neural foramina and the epidural space (Fig. 9.17). •
Staging
The evaluation of mediastinal lymph nodes is an important aspect of staging in patients with non-smaIl-cell lung cancer. The nodal status is an integral part of the TNM staging system and
b
Fig. 9.17 Superior sulcus (Pancoast) tumor (4 x 1/6). On 5 mm axial sections (a) there is no clear demonstration of chest wall invasion, which is clearly demonstrated on 3 mm coronal reformations from the same
CT Morphology-N
c multislice CT data set (b). Coronal MPR in another patient with small cell lung cancer extending into the cervical soft tissues and invading the superior vena cava (c).
309
9 Lungs and Tracheobronchial
System
provides critical information for planning treatment and determining prognosis. Traditionally imaging evaluation of mediastinal lymph node involvement has relied on anatomic features of the nodes. most notably nodal size. CT has been shown to be the best predictor of nodule volume. With regard to the determination of the nodal status. however. its limitation with relatively low sensitivity (4167%) and low specificity (79-86%) are well documented. based on the fact that enlarged lymph nodes may be hyperplastic rather than neoplastic. especially in patients with post-obstructive pneumonia. and normal-sized nodes may contain neoplastic cells. Microscopic tumor involvement was seen in S-64% of normal-sized lymph nodes. while IS-30% of enlarged lymph nodes (IO-IS mm) were free of tumor. Patients with COPD. recurrent cardiac failure. or previous granulomatous disease also tend to have enlarged lymph nodes. Various approaches have been taken to refine the size criterion by modulating the threshold size for various nodal locations within the mediastinum. or taking into account not only the smallest diameter as measured on the axial scan but also the shape of the lymph node (round versus oval). An increased diagnostic accuracy was described for morphologic evaluation of thin section imaging ofhilar lymph nodes. In 9S% oflocations with normal lymph nodes there was a straight or concave pleural interface between hilar structures and lung parenchyma, while in 9S% of locations with malignant lymph nodes there was a convex interface. Application of this criterion led to an improvement in sensitivity from SO% for the 10 mm size criterion to 87% for the morphologic criterion, and of specificity from 80% to 83 %. The approach of considering size and shape of lymph nodes takes advantage of the availability of thin and high quality multiplanar images provided by multislice CT. No quantitative data of diagnostic accuracy are available up to now. However, even though it is likely that refinement of morphological criteria in evaluating mediastinallymph nodes will improve the diagnostic accuracy, it will most likely never be able to achieve the same diagnostic accuracy as functional imaging using PET or (still investigational) lymph node-specific MR contrast agents (USPlO). Lymph node metastases may arise through a continuous mode of spread (intrapulmonary
=> bronchopulmonary or hilar => ipsilateral mediastinal => contralateral mediastinal). or a group of nodes may be skipped (in up to 30% of cases). Tumors of the right lung tend to metastasize to ipsilateral nodes, with lower lobe tumors usually causing more extensive metastasis. Contralateral nodal metastasis is more common with tumors of the left lung {> 30%). CT serves not only as basis for selecting those patients who have to undergo further invasive procedures (mediastinoscopy, mediastinotomy, thoracoscopy, trans bronchial biopsy) for lymph node staging but also to select the appropriate procedure depending on the location of enlarged lymph nodes to be sampled. Though the inability of CT to detect microscopic metastases within normal sized lymph nodes is known, it is generally agreed that preclusion of enlarged lymph nodes is sufficient to obviate further evaluation and patients directly undergo thoracotomy. However, to avoid over staging it is common understanding that enlarged lymph nodes seen in CT must be sampled by biopsy. The paratracheal, subcarinal, and proximal bronchopulmonary lymph nodes are accessible to mediastinoscopic biopsy, but not the commonly affected aortopulmonary nodes.
Therapeutically significant lymph nodes that appear enlarged on CT (Tables 9.1S, 9.16) should be biopsied due to the limited specificity ofCT. A mediastinoscopy is recommended in all patients with T3 tumors, with centrally located adenocarcinoma because of their high incidence of lymph node metastases, and in patients with chest wall infiltration or with Pancoast tumors, because of their poor prognosis with advanced lymph node staging.
Screening with Low Dose CT Previous studies using chest radiography and sputum cytology had not been able to statistically prove reduction of mortality rates, most likely due to the too low sensitivity of these tests for early tumor stages. Spiral CT is acknowledged as the most sensitive diagnostic method for the detection of small intrapulmonary nodules. The option for Significant reduction of acquisition dose
Bronchial Carcinoma
without comprormsmg the conspicuity of intrapulmonary nodules makes this technique feasible for screening (Fig. 9.18). Such CT examinations have a sensitivity >90% for nodules larger than 5 mm in diameter and an acquisition dose that is only about 2-3 time larger than chest radiographs in two projections. If a nodule is detected, the patient has to be rescanned with thinner sections unless thin-section multislice CT has been employed (see p.291). Multislice CT, in addition, facilitates automated detection, morphologic, and volumetric analysis of nodules. Depending on the geographic area and the frequency of histoplasmosis or tuberculosis, there may be a high rate of nodules with benign histology (granuloma, hamartoma, focal infection, lymph nodes). Thus, the screening protocols may vary locally. Biopsy has been suggested for all nodules> A cm that are not definitively benign (e.g., because of benign calcification, see p.314, or a known long-term course). Nodules between 5 and W mm in size are either biopsied or followed, nodules below 5 mm are followed with a larger time interval for rescanning. •
CT Morphology
Nodule detection is performed on axial sections with a lung window setting and a section width of some 5 mm. Overlapping reconstruction is mandatory, and viewing in a cine display speed up the evaluation and improves the detection rate. Automated detection algorithms work best on thin-section multislice CT data sets. It is not yet clear whether these algorithms are good
a
b
Fig. 9.18 Lung cancer screening Low-dose scans (CTDlvoi=2.3 mGy) demonstrate a small solid nodule (adenocarcinoma, a) and a ground-glass nodule (alveolar cell carcinoma, b). The volumetric assessment of
enough to be used exclusively or whether visual assessment of the whole data set is still necessary. Detection failures in a Japanese trial were due to the misinterpretation of lesions in close vicinity to pulmonary vessels or associated with pre-existing disease due to old tuberculosis. Three types of nodules have been identified: ground-glass nodules (focal areas of ground-glass opacities), solid nodules (no ground-glass associated), and mixed nodules (both features present). For nodules below "I em, patients with ground-glass and mixed nodules may be put on a course of antibiotics and are rescanned after 3-6 weeks. Solid nodules are biopsied or removed if thin-section scanning reveals morphologic criteria that are suggestive of malignancy (see Table 9.18). All other nodules are followed for 6 weeks to 6 months, depending on the local protocols and the available evaluation software. Computer-based evaluation algorithms were found to be able to reliably assess volumetric changes already after 4 weeks. Morphologic analysis is improved if volume rendered displays of the nodule surface are available and if multiplanar reformations centered on the nodules can be performed. •
Whom to Screen and When?
All studies published so far have in common that CT was found to be highly superior to radiography in terms of sensitivity. Bronchial carcinoma was detected in 1.1-2.7% of screened patients. Most (62-93%) tumors were determined to be stage 1. There is no general consen-
c
d
another nodule at the lung base (c) demonstrates no increase in tumor volume in follow-up scans after up to 6 months (d).
311
9 Lungs and Tracheobronchial
System
sus yet, however, as to whether screening can be generally recommended, and which risk groups are eligible. There are numerous potential sources of bias in the present trials (e.g., ELCAP, various Japanese and European studies). A lead time bias mainly affects the initial (prevalence) screening and is caused by tumors that are detected earlier but without an effect on overall life expectancy. A length time bias is caused by tumors, such as adenocarcinomas, that are slow growing and are thus more likely to be detected during follow-up (incidence screening) while more aggressive tumor types, such as small-cell cancers, may be diagnosed based on clinical symptoms in the time intervals between two screening procedures. Peripheral tumors are more likely to be detected early than central tumors {location bias}. In addition, there is the danger of overdiagnosis in those individuals who normally would die from other causes: small tumors detected in patients with large other risk factors (such as coronary artery disease, previous heart attacks, severe emphysema, or other smokingrelated diseases) may induce unnecessary treatment. Most present evidence, however, suggest that lung cancer screening programs in a selected risk population may be even more effective than mammography screening. There are no general rules yet but patients should be over 50 or 60 years old and have a major history of smoking (20-40 pack years). Obviously, the older the patient and the longer the smoking history, the higher is the likelihood that screening would be effective in detecting tumors and reducing overall mortality. Control after Treatment/Tumor Recurrence
CT remains the procedure of choice for routine follow-up or in suspected tumor recurrence in the post-pneumonectomy patient. The superiority of MRI for the detection of soft tissue recurrence is shown by the combination of mass effect and altered MR signal intensity. In fact any inhomogeneous signal in the pneumonectomy space should be suspicious for tumor recurrence (Fig. 9.19). A problematic topic is the reevaluation of the residual tumor and tumor activity after
a
b
Fig. 9.19 Tumor recurrence after lobectomy (4 x 1/6). Axial section (a) and coronal reformation (b) demonstrate a large recurrent tumor that invades the chest wall, the pulmonary arterial trunk, and thejugular vein and superior vena cava, leading to extensive venous collaterals (c).
neoadjuvant chemotherapy to be followed by surgery. CT criteria tend to underestimate the therapeutic effect demonstrated by pathologic examination. PET is likely to be the superior diagnostic method for evaluation of residual tumor or tumor recurrence. •
CT Morphology
There is a considerable variation in the appearance of the post-pneumonectomy space ranging from complete obliteration (in 27%) to a persistent fluid collection (in 73%). Filling in of previously patent ectatic bronchi is suggestive for recurrent disease. A 50% size reduction, a change in tumor morphology (round versus irregular), and the disappearance of density enhancement after application of intravenous contrast have been suggested as criteria for complete tumor response. Virtual Bronchoscopy and CT Bronchography
Virtual bronchoscopy yields excellent displays of the central tracheobronchial system and has various advantages over fiberoptic bronchoscopy. It is less invasive and therefore well tolerated by the patients. It can display the bronchial system distal to a narrow stenosis. Inversion of the viewing direction of the virtual
Secondary Tracheobronchial
bronchoscope is possible and one is able to look towards the proximal portions of the tracheobronchial system. Most important, however, is that all the additional information from the surrounding structures is available from the scans. Thus, there is information of the transmural extent of a tumor, the presence and location of lymph nodes, and the location of suspicious areas wi thin the 1ung parenchyma. In a clinical trial it was found that virtual bronchography improves the recognition of mild changes in airway caliber that otherwise are difficult to appreciate. It also helps in the understanding of complex tracheobronchial anomalies. The technique can be used to noninvasively assess diseases of the central tracheobronchial system (Fig. 9.20) but cannot substitute for fiberoptic bronchoscopy in patients in whom biopsy has to be performed or in whom visual assessment of the mucosal surface (e.g., color changes) is necessary. For pretherapeutic staging, virtual bronchoscopy serves as adjunct to bronchoscopy. Virtual bronchoscopy with transparent walls or colored extramural pathology may be used for guiding and planning trans bronchial biopsy procedures. In post-therapeutic imaging, non- invasive virtual bronchoscopy may be used alternately with fiberoptic bronchoscopy, e.g., in patients after bronchoplastic surgery with complex anatomic situations and a susceptibility for developing bronchial strictures. However, other visualization techniques based on MPR or volume rendering often provide equivalent or superior information (Fig.9.20).
Secondary Tracheobronchial Neoplasms
Neoplasms of the larynx, thyroid gland, or esophagus can involve the trachea and central bronchi by direct invasion or distant metastasis. Breast cancers, renal carcinomas, colorectal carcinomas, and melanoma have a propensity for endobronchial metastasis, which may be palliable by laser ablation and stent insertion, depending on the extent of the extrabronchial lesion and the severity of the obstruction. •
CT Morphology
CT has proven more sensitive than bronchoscopy for evaluation of the subglottic larynx. The subglottic spread of a laryngeal tumor into the trachea or, conversely, the spread of a trachea! tumor to the undersurface of the vocal cords will dictate the preferred therapeutic approach (laryngectomy versus partial tracheotomy). When a thyroid tumor is present, the ill-defined boundary that is often found between the thyroid gland and tracheal wall should not be misinterpreted as a sign of tumor invasion. Tracheoesophageal fistulas are most likely to develop in patients with advanced esophageal tumors, and CT can often locate the fistula more easily than endoscopy. Endobronchial metastases display nonspecific CT features and should be biopsied if doubt exists. Hodgkin's disease can lead to the formation of obstructive endobronchial lesions with atelectasis. In very rare cases, non-Hodgkin's lymphoma (NHL) may infiltrate the central airways and cause diffuse submucosal nodules.
Fig. 9.20 Virtual bronchoscopy (a) and CT bronchography (b) in a patient with a stenosis following sleeve resection of bronchogenic cancer(2/4/1 ).
a
Neoplasms
b
313
9 Lungs and Tracheobronchial System Pulmonary
Nodules
Pulmonary nodules are among the commonest focal pulmonary lesions detected at imaging (Table 9.17). Vascular processes or atelectatic lung areas can have features that resemble pulmonary nodules. For many diseases. however. the presence of a nodule is simply one variant in a range of possible radiologic presentations of the disease.
Table dules
9.17
Differential
Benign
neoplasms
• Hamartoma • Chondroma
Approximately 50% of surgically removed nodules turn out to be benign and noninvasive. Evaluation of the solitary pulmonary nodule remains a substantial challenge in modern medicine. Table9.18 reviews morphologic findings that may be helpful in differentiating benign from malignant lesions. In selected cases. ancillary findings such as pleural thickening. vascular relationships. cavitation. or enhancement characteristics may also aid in making the correctdiagnosis. Clusters of small intrapulmonary nodules with an internodular distance of less than Wmm and without associated infiltrates most frequently represent postinfectious granulomaFeatures
that help to differentiate
benign
Pseudotumor
Malignant neoplasms • Peripheral bronchial carcinoma • Alveolar cell carcinoma •
Metastases
• Carcinoid • Kaposi sarcoma •
Lymphoma
Septic
emboli
Intrapulmonary
lymph
nodes
AV malformation Round
atelectasis
and malignant
Benign Margins
pulmonary
• Tuberculosis • Histoplasmosis • Sarcoidosis
• Pulmonary
9.18
of
Granulomas
Differentiation of Pulmonary Nodules
Table
diagnosis
solitary
pulmonary
nodules
Malignant
Smooth
Stellate,
spiculated
Rigler notch sign Converging
vessels
Pleura! tail <Scm
Size Change
in size
Associated
>3cm
Size unchanged
blood vessels
for 2 or more years
No feeding artery, septic emboli
except
for
Feeding Draining
artery pulmonary
Calcifications
Central Dumbbell-shaped Peripheral lamination Calcification >10% of volume
Eccentric Stellate Disseminated Calcification
CTdensity
>150HU
<100HU
Enhancement
characteristics
Gradual
enhancement
<15HU Cavitation
Associated
Concentric,
findings
Rapid
vein
<1cr/oof
volume
enhancement
>15HU thin wall
Eccentric (squamous thick wall Lymphadenopathy
cell carcinoma),
no-
Metastases
tous disease (Aspergillus, mycobacteria, histoplasmosis), Intrapulmonary lymph nodes are typically less than 15 mm in size and locate within 15 mm distance from the pleura or interlobar fissures, They have a sharp contour, are round, oval or lobulated, and show contrast uptake of 35-85 HU, While preliminary work in vitro and in vivo had indicated that dual-kilovolt peak computed tomography may be useful in the identification of benign lesions this could not be confirmed in a recent multicenter trial. There was no significant increase of CT numbers with 80kVp compared to 140 kVp images based on a high calciumcontent. The value of enhancement after bolus injection of contrast medium was also assessed in a multicenter trial. Absence of significant nodule enhancement (A 15 HU) at CT was found to be strongly predictive of benignity, With a threshold of 15 HU the sensitivity was 98%, the specificity was 58% and the accuracy was 77%, The size of the nodules ranged between 14mm and 55 mm. A 15 mm spiral CT cluster of 3 mm slices was obtained L 2, 3, and 4 minutes after the onset of injection of 420 mg iodine/kg body weight with an injection rate of 2 ml/sec (the concentration was 300 mg iodine/ml). Data analysis was based on peak nodule enhancement and time attenuation curves, Malignant lesions enhanced (median of 38 HU) significantly more than benign lesions (median of 10 HU), This technique may also be used for follow-up of doubtful lesions, While malignant (42 HU) and acute inflammatory benign lesions (44 HU) both showed significantly higher peak enhancement levels than noninflammatory benign nodules (13 HU), in follow-up CTs after treatment (3 to 4 months later), the enhancement rates in nodules with active inflammatory reactions significantly decreased, making discrimination possible, Other discriminating features between malignant and inflammatory lesions are a retarded wash-out resulting in a slower decline of the density curves in malignant nodules, a lower attenuation on plain scans, and pronounced peripheral enhancement in inflamma tory n od ule s.
The low false positive rate of the results of the CT nodule enhancement technique is potentially valuable in the treatment of patients, A nonenhancing lung nodule can be managed with radiological surveillance alone even in a supportive clinical situation, An enhancing nodule either will be appropriately followed, if clinical suspicion for malignancy is low, or will undergo biopsy, Intratumoral necrosis and low cardiac output may contribute for low peak enhancement and thus false negative evaluation, It is important to note that PET achieves a comparable sensitivity for the assessment of malignancy in nodules as compared to the reported values for enhancement characteristics (94-100% versus 98%) but superior specificity values of over 90% (versus 60 to 75%), The high specificity ofFDG PET for the diagnosis of benign lesions has important clinical utility, Lesions with low FDG uptake may considered benign, However, they should be followed radiographically since false negative results were reported for carcinoid tumors and bronchioalveolar carcinomas if lesions were smaller than 1 ern. Observation of nodule growth is another valuable tool for monitoring pulmonary nodules, especially if they are too small for the assessment of enhancement characteristics and occur in patients at risk, Computer programs have been developed that allow for nodule segmentation and volume rendering and proved to be more precise and sensitive towards small volume changes than visual assessment on 2D axial follow-up scans or manual assessment oflesion diameter.
Metastases
Pulmonary metastases most commonly originate from carcinomas of the breast, kidneys, colon, stomach, and pancreas, They can also develop from seminomas and sarcomas, CT is the most sensitive modality for detecting metastases in the staging of tumors, Spiral CT can detect pulmonary nodules as small as 2-3 mm in diameter. Conventional CT has similar capabilities, but up to 30% of smaller lesions (A 10 mm in diameter) may be missed on
315
316
9 Lungs and Tracheobronchial System
sequential tion. •
monary lymph nodes, focal scars) are also being detected with increasing frequency. This leads to problems of differential diagnosis in preoperative tumor staging. Since a definitive histological diagnosis cannot be established for all of these findings, other differentiating strategies must be considered, such as scheduling followup examinations at 3 to 6 months or defining a minimum size of 3-5 mm for focal lesions that require further evaluation. At present, however, there are no adequate data to substantiate this approach. Also, some tumors such as thyroid carcinoma (Fig. 9.21 b) often produce miliary metastases that measure only a few millimeters in size. Calcifications are rare and generally suggest benignancy, although calcified metastases are known to arise from osteosarcoma and less commonly from chondrosarcoma. These lesions cannot be reliably distinguished from granulomas. Rarely, calcified metastases develop from the spread of mucous- forming adenocarcinomas (breast, colon, ovary), thyroid carcinomas, soft-tissue sarcomas, and after chemotherapy. Cavitation is typical of septic emboli but rarely occurs also in metastases (Fig. 9.22 a), e.g., in squamous cell carcinoma. Fatwithin a pulmonary nodule usually suggests the presence of a harmatoma, but may also be due to a metastasis from liposarcoma (Fig.9.22b) .
scans due to respiratory misregistra-
CT Morphology
All focal lesions in the lung that are not attributable to vessels are suspicious for pulmonary nodules. Linear densities without a central mass are not considered to be nodules and generally represent small areas of scarring (e.g., post infection). Small nodules that are imaged in thick CT slices show fading peripheral density due to partial volume effects (see chapter 7 on Image Analysis) . Metastases typically have smooth margins. A mass with ill-defined margins should raise suspicion of a peripheral bronchial carcinoma. Metastases may acquire ill-defined margins as a result of local lymphedema, intralesional hemorrhage (angiosarcoma, choriocarcinoma), or postchemotherapeutic scarring. Solitary metastases are rare, accounting for just 5% of all solitary nodules. The likelihood that pulmonary nodules are metastatic increases with the numbers of nodules that are seen. Metastases are commonest in the outer portion of the lung (90%) and near the pleura. They predominantly involve the lower lobes (66%). Approximately 40% of metastases are found to have associated blood vessels (feeding vessel sign, Fig. 9.21 a). Spiral CT can detect pulmonary nodules with such high sensitivity (range of 2-3 mm) that small benign lesions (granulomas, intrapul-
a
b
Fig. 9.21 Pulmonary nodules, (a) Predominantly subpleural hematogenous metastases with feeding vessel sign. (b) Multiple miliary metastases from thyroid carci-
c
d
noma, predominantly in the lower zones, (c) Tuberculoma with central calcification, (d) Pulmonary hamartoma with lobulated margins and fat deposits.
Carcinoid Tumor ganizing focal pneumonia in patients with subclinical infection, and by others to be a precursor of pulmonary lymphoma (see also p.324). They show a variety of histologic entities including xanthoma, histiocytoma, and mast-cell or plasma-cell granuloma. •
CT Morphology
They present as well-defined solitary nodules, calcifications are seen in about 20%. Airway involvement is extremely rare.
a
b
Fig. 9.22 Atypical presentation of pulmonary metastases: Cavitating pulmonary nodule on 1.5mm thick sagittal reformation (4 x 1/6) mimicking a pneumatocele (a). Nodule containing fatty tissue (-55 HU) in a patient with liposarcoma (b).
Hamartoma
Hamartomas constitute more than 5% of all solitary pulmonary nodules, and the great majority (> 95%) occur after 40 years of age. •
Granuloma
Tiiberculomas are focal lesions in which a balance is established between the activity of the infectious organism and the host defenses. Histoplasmosis is prevalent only in certain regions such as the southern and eastern United States; it is rare in Europe. Lung infection gives rise to calcifying granulomas. •
CT Morphology
A tuberculoma is a sharply circumscribed nodule 0.5-4 cm in diameter. Approximately 25% of tuberculomas have lobulated margins. A pathognomonic feature is calcification (Fig. 9.21 c), which usually develops at the center of the nodule or may fill it completely. The upper lobes are a preferred site. Cavitation is rare and, when present, is usually small and eccentric. The nodules are multiple in 80% of cases. Histoplasmosis is marked by the development of granulomatous nodules that often show central calcification like tuberculomas. Cavitation is very rare, and there is a predilection for the lower lobes. Pulmonary Pseudotumor
The etiology is unknown, though they are thought by some to represent sequelae of Of-
CT Morphology
Hamartomatous nodules are usually smaller than 4cm and are more often lobulated than round. They may be slow growing. Approximately 90% of hamartomas occur in the peripheral lung (within 2cm of the pleura). Approximately one-third show calcifications (popcornlike, stippled, central), and an equal percentage contain fatty areas, which are considered pathognomonic (Fig. 9.21 d). Identification of intrapulmonary fat deposits requires density measurements in thin (1-2 mm) sections to avoid misinterpretation of partial volume effects as negative density units. Rarely, a fat-containing nodule may be a metastasis from liposarcoma (Fig.9.22b).
Carcinoid Tumor
The lung parenchyma is an uncommon site of occurrence for carcinoid tumors (10% of all intrathoracic carcinoids). Carcinoids belong to the group of neuroendocrine tumors, and may be associated with ectopic hormone productions (e.g., ACTH). Peripheral carcinoids tend to be type 2 tumors, with metastases in up to 50% of patients. Typical carcinoids (type 1) rarely metastasize.
317
9 Lungs and Tracheobronchial
•
System
CT Morphology
Intrapulmonary carcinoids appear as well-circumscribed nodules. Isolated bronchial dilatation with mucoid impaction may provide the only indirect sign of an occluding intraluminal carcinoid. These lesions rarely show calcification or cavitation, and most enhance intensely with intravenous contrast. Dependent on the histologic subtype they are slow growing; up to 30% of carcinoids contain calcifications. Septic Emboli
Thromboembolic material can arise in septic conditions and cause the occlusion of small pulmonary arteries. Septic emboli may originate from an infected catheter, abscess, endocarditis, or pulmonary or urinary infection. •
CT Morphology
CT usually shows multiple round or wedgeshaped densities in various stages of cavitation (abscess formation in approximately 50%) (Fig. 9.23). The detection of a feeding vessel (feeding vessel sign) confirms the hematogenous origin of the lesion. The typical appearance is that of disseminated bilateral lesions that are mainly located in the lower lobes. There may be an accompanying pleural or pericardial effusion.
Fig. 9.23 Septic emboli. (a) Unsharply demarcated nodules, (b) From a patient with history of intravenous drug abuse showing multiple round opacities with cavitations and signs of bronchogenic spread of disease as indicated by the centrilobular densities.
Septic foci have less distinct margins than metastases and often appear as wedgeshaped opacities based on the pleura, some of which are cavitating.
Peripheral Bronchial Carcinoma
Approximately 40% of all bronchial carcinomas develop distal to the segmental bronchi. Peripheral bronchial carcinoma should always be included in the differential diagnosis of a solitary pulmonary nodule. Many bronchial carcinomas can be confidently identified by their CT morphology alone. •
CT Morphology
The CT appearance ranges from that of a solitary, well-circumscribed pulmonary nodule to lesions that show definite morphologic signs of malignancy (Table 9.14). Malignancy is suggested by signs of infiltrative growth with retraction of surrounding structures. The signs include ill-defined margins, radiating spicules (corona radiata), radiating strands with pleural retraction (pleural tail), and the distortion of small vessels by the nodular lesion. Other signs of bronchial carcinoma are visible intraluminal tumor growth, a notched indentation in the tumor border caused by the entry of a pulmonary arterial branch (the Rigler notch sign), and the detection of a pulmonary vein leaving the tumor (Fig. 9.24). Small air inclusions often represent opened bronchi [pseudoca vitation) or foci of intratumoral necrosis and are a common feature of alveolar cell carcinoma. When CT demonstrates a bronchus entering the tumor ("positive bronchus sign," Fig.9.24b), the accuracy rate of bronchoscopic histologic or cytologic sampling increases from less than \5% to 60%. This finding is best appreciated on thin sections and is helpful in diagnostic planning (bronchoscopy vs. transthoracic needle biopsy vs. surgery). Multiplanar imaging is advantageous for imaging morphologic detail, the relationship to pleural interfaces and the localization of lesions in close vicinity to anatomic interfaces (chest wall/interlobium) (Fig. 9.25).
Atelectasis
A
c
b Fig. 9.24 Peripheral bronchial carcinoma, (a) Axial scan demonstrates ill-defined, spiculated margins and intraluminal tumor growth, (b) Semicoronal reformatted image shows a bronchus passing into the spicu-
lated mass (positive bronchus sign), (c) Shaded surface display shows a pulmonary vein (arrow) draining the tumor.
-4 Fig. 9.25
Peripheral bronchial carcinoma (4x 1/6). The coronal MPR (b) better demonstrates the transfissural tumor growth of this mixed type nodule (adenocarcinoma) than the axial section (a).
·/'\../'\.Af'... __
»<:
Predominantly Focal Pulmonary Opacities
Atelectasis Lobar or segmental atelectasis generally presents no problems of differential diagnosis. It is defined as a volume loss of a lung or portions of
Upper lobes
Fig. 9.26
Lobar atelectasis.
a lung. Types of atelectasis are differentiated according to their etiology (Table 9.19). The CT findings correspond to those in conventional radiography (Fig. 9.26).
Middel lobe orlingula
Schematic
representation
of CT morphology.
Lower lobes
319
9
Lungs and Tracheobronchial System
Table 9.19
Types of atelectasis
Resorption atelectasis (endobronchial) • Tumor • Lymphoma • Mucoid impaction • Foreign body • Stricture (tuberculosis, trauma, postoperative) • Postinflammatory middle lobe syndrome Resorption atelectasis • Mediastinal mass • Lymphadenopathy • Enlarged left atrium
If there is a central tumor causing the lobar atelectasis, the lateral contours of the lobes may run convexly rather than concavely (similar to Golden's reverse S sign in posteroanterior radiographs in patients with upper lobe atelectasis due to a central tumor).
(extrabronchial)
Passive (compression) • Pleura! effusion • Pneumothorax • Rounded atelectasis • Adjacent space occupying disease (bulla, mass, ascites) • Decreased diaphragmatic excursions Cicatrization • Granulomatous disease • Pneumoconiosis • Interstitial fibrosis Adhesive • ARDS • Postoperative
•
dially. The lower lobes collapse posteromedially against the posterior mediastinum and spine.
Poststenotic (resorptive) atelectasis generally does not contain an air bronchogram, and mucus-filled bronchi appear as hyperdense tubular or ring-like structures. With compression atelectasis (e.g., secondary to basal effusion), a positive air bronchogram is usually observed. Cicatrization atelectasis occurs in areas of pulmonary fibrosis and is almost invariably associated with bronchiectasis. Adhesive nonobstructive atelectasis or microatelectasis is related to a decreased amount of surfactant in patients after surgery or ARDS and most commonly presents with quite extensive consolidations. After contrast administration, the atelectatic lung tissue shows stronger and more homogeneous enhancement than lung parenchyma involved by neoplastic or inflammatory infiltration (Fig. 9.27).
CT Morphology
Depending on the extent of the atelectasis, CT may show a homogeneous area of increased densitywith a typical location and configuration. Important indirect signs that help distinguish atelectasis from infiltration are the bowing (concave displacement) of the interlobar fissures and the crowding of vascular structures as a result of the reduced lung volume. There may be compensatory overinflation of the remaining lung with the vascular structures being spread apart. In addition to the location and configuration of the opacity itself, the displacement ofthe hilar structures is most helpful in identifying the atelectatic part of the lung. While the right upper lobe collapses against the mediastinum, the left upper lobe retains more contact with the anterior and left lateral chest wall (Fig. 9.26). The middle lobe is particularly susceptible to chronic postinflammatory /\collapse; CT is used to exclude a central endobronchial tumor in these patients and well demonstrates the patent origin of the middle lobe bronchus. The medial margin of the middle lobe abuts the right heart border and the posterior margin ofthe middle lobe is displaced anterome-
a
b
Fig. 9.27 Atelectasis. (a) Comparison of atelectasis (lower lobe) and pneumonia (lingula) after contrast administration. The atelectasis shows marked, homogeneous enhancement due to the rich vascular supply of the collapsed lung parenchyma. Pneumonia, like other types of consolidation, shows considerably less enhancement, (b) Left upper lobe atelectasis due to a central bronchial carcinoma.
Bronchiolitis Obliterans with Organizing Pneumonia (BOOP)
or cryptogenic organizing pneumonia, histologically characterized by:
COP) is
• granulation tissue polyps within the lumina of bronchioles and alveolar ducts; and • patchy areas of cellular infiltrates (mainly mononuclear cells and macro phages) in the surrounding air spaces.
b Fig. 9.28 Rounded atelectasis. Pulmonary window settings show a typical comet-tail configuration of the bronchovascular structures (a), with a broad area of pleural contact and pleural thickening. Soft tissue window settings in another patient (4 x 1/6, coronal MPR) demonstrate the homogeneous enhancement after contrast administration (b).
Rounded Atelectasis
Rounded atelectasis is a persistent compression atelectasis secondary to effusion. It may present a tumor-like morphology, but biopsy is rarely necessary owing to the characteristic CT features of the lesion. Rounded atelectasis can be seen with any cause of pleural thickening but is commonest in asbestos-related disease. •
CT Morphology
CT typically shows a solitary rounded or wedgeshaped mass located near the pleura, most commonly in the lower lobe, and showing homogeneous contrast enhancement. The affected lobe shows signs of volume loss. Diagnosis is aided by the comet tail sign caused by the distortion and displacement of bronchi and blood vessels in a whorled pattern leading into the atelectatic mass (Fig. 9.28). Pleural thickening is always an associated sign. Air bronchograms or focal air collections (pseudocavitation) may be present. Bronchiolitis Obliterans with Organizing Pneumonia (BOOP)
Bronchiolitis obliterans organizing pneumonia (BOOP) (also named organizing pneumonia (OP)
Most cases are idiopathic. Clinical manifestations include an unproductive cough, subfebrile temperatures, several months' history, and restriction of pulmonary function. The symptoms regress with steroid therapy. A form with identical pathomorphologic features, called the BOOP reaction, may occur as a sequel to viral, bacterial, or fungal pneumonia, chronic eosinophilic pneumonia, collagen diseases, radiotherapy, and after lung or bone marrow transplantation, Wegener's granulomatosis and after toxic inhalation (Table 9.20). A recent international consensus on standardizing classification of idiopathic interstitial pneumonias included idiopathic BOOP in the group of idiopathic interstitial pneumonias (see p. 344). This view has been the subject of debate since the lesion is caused by airspace organization and is not primarily interstitial in a histological sense. Clinical history may be helpful for differential diagnosis. BOOP usually appears as a subacute illness with duration of symptoms for about 2 to 6 months before diagnosis. In many patients, the clinical and radiographic signs of disease remit completely after systemic corticosteroid therapy. Only a subset of patients show a rapid progression and have a worse prognosis. In these patients BOOP is frequently associated with connective tissue disease or drug therapy. While infectious diseases may be excluded on the basis of clinical symptoms, lung biopsy is usually performed because of suspicion of a carcinoma.
Table 9.20
Causes of BOOP
Idiopathic Following infection Chronic eosinophilic pneumonia Collagen vascular diseases Radiotherapy Bone marrow or lung transplantation Wegener's granulomatosis Following toxic inhalation
9
•
Lungs and Tracheobronchial System
CT Morphology
The morphologic features are variable and nonspecific (Fig. 9.29). The commonest pattern consists of patchy (non-segmental) air space consolidations in a subpleural (cortical) or central peribronchiolar distribution in more than 50% of the patients. The patchy consolidations may be migratory, they may show pleural tags or spiculae (in approximately 30 to 40%). While consolidations may be the only finding in about 50% of the patients, ground-glass opacifications are mostly part of a mixed pattern. The size of the individual opacities ranges from 3 cm to nearly complete lobar consolidations. The margins of the consolidations are indistinct; they may contain air bronchograms. Bronchial wall thickening and dilatation are seen in most patients and are usually restricted to areas of consolidation or ground-glass opacifications. It is not yet clarified how well these bronchial abnormalities are reversible. In 15% to 50% of patients in the various study populations multiple peri bronchiolar (centrilobular) nodules were seen that mostly measured between 1 and 10mm. These tend to have mostly irregular margins (88%) and may also show air bronchograms (45%). Nodules are the only finding in about one-third of the
a
b
Fig. 9.29 Various morphologic appearances of BOOP. (a) Peripheral pleural-based consolidations, (b) Peribronchial consolidation with air-filled bronchi mimicking pneumonia, (c) Multiple peribronchial nodules.
patients showing nodules. Rarely bilateral areas of ground-glass opacities are seen with overlying thickened intralobular septal lines. Nodules and ground-glass attenuation are seen more frequently in the immunocompromised than in the immunocompetent patient. Additional findings are pleural thickening (33%), small pleural effusions (30%), and parenchyma! bands (25%). Dependent on the range of distinctive patterns, the differential diagnosis for BOOP is rather extensive (Table 9.21). In cases with solitary mass-like focal consolidations that have morphologic criteria for malignancy, such as spiculae, pleural tagging or regional pleural thickening, a bronchial carcinoma cannot be differentiated from BOOR Multiple patchy air space opacities may also be associated with alveolar cell carcinoma, vasculitis, or pulmonary hemorrhage. Numerous bilateral small nodules in BOOP have to be differentiated from sarcoidosis or acute infectious bronchiolitis (atypical and typical mycobacteria). Chronic eosinophilic- pneumonia has to be considered for differential diagnosis if the pattern is dominantly peripheral.
c
d
(d) Patchy, well-circumscribed consolidation with ectatic air-filled bronchi and combined with signs of architectural distortion and air trapping of adjacent lung.
Hemorrhage, Table 9.21
Differential diagnosis
of BOOP
Focal mass-like lesion
bronchial
carcinoma,
Multiple
vasculitis,
lymphoma
patchy air space opacifications
Numerous small nodules
sarcoid,
infectious
Subpleural
chronic
eosinophilic
patchy consolidation
Hemorrhage, Hemosiderosis
Table 9.22
b Fig. 9.30 Pulmonary hemorrhage, (a) Fresh pulmonary hemorrhage with ill-defined, confluent acinar opacities following transpulmonary needle biopsy of a central tumor, (b) Fresh hemorrhage in the posterior apical segmental bronchus of the upper lobe, manifested by ground-glass opacity of the affected sec-
carcinoma
bronchiolitis pneumonia
syndrome
Collagen diseases and vasculitis • Lupus erythematosus • Wegener's disease • Polyarteritis nodosa • Rheumatoid arthritis Hemorrhagic necrotizing pneumonia (e.g., leptospirosis, acute viral pneumonia)
CT Morphology
The CT appearance ranges from small, indistinct (acinar) focal densities (Fig. 9.30) to more diffuse ground-glass opacities and widespread homogeneous opacification with an air bronchogram. The opacities clear in two to three days. Some irregular patchy or linear opacities may persist for a longer time period. With chronic recurrent hemorrhage (increasing amount of hemosiderin deposits within the interstitium), irreversible interstitial changes are superimposed. The severity of
bronchoalveolar
Causes of pulmonary hemorrhage
Goodpasture's
Regardless of its etiology (Table 9.22), pulmonary hemorrhage is characterized by increased alveolar density in the acute stage. Idiopathic pulmonary hemosiderosis (IPH) is characterized by recurrent episodes of pulmonary hemorrhage without associated glomerulonephritis or serologic abnormalities. •
Hemosiderosis
Idiopathic
pulmonary
hemosiderosis
Coagulation disorders • Medications • Myelodysplastic diseases Trauma, biopsy
parenchymal changes is dependent on the time sequence and the number of hemorrhages that have occurred.
c
d
ondary lobules, (c) Intrapulmonary hemorrhage demonstrating confluent acinar opacities as well as dense consolidation in a patient with Goodpasture's syndrome. (d) Ground-glass opacities associated with thickened interlobular septa indicating resorptive processes in hemosiderosis.
323
324
9 Lungs and Tracheobronchial System
Alveolar Cell Carcinoma
Alveolar cell carcinoma (bronchiolar adenocarcinoma, bronchioloalveolar carcinoma) accounts for no more than about 5% of all lung cancers. Bronchogenic spread (e.g., by coughing) is common and leads to multiple lesions and even bilateral involvement in more than 40% of cases. The multifocal or diffuse form has a much poorer prognosis than the local form. •
CT Morphology
The local form of alveolar cell carcinoma presents as a solitary pulmonary lesion that is often peripheral or subpleural (Fig. 9.31 a). It may appear as a well-circumscribed nodule or may show the characteristic features of bronchial carcinoma including a "pleural tail" (streak-like densities extending to the pleura, >50%) and spiculated margins (> 70%). Most lesions exhibit a mixed density pattern. It is common to find intact air-filled bronchi within the tumor (pseudocavitation), and the embedded bronchi may appear stretched. Atelectasis generally does not occur. The multifocal or djffuse form resembles air space consolidation due to pneumonia. The spectrum ranges from patchy ground-glass opacities to areas of consolidation (Fig. 9.31 b, c). Involvement of both lungs may occur.
b Fig. 9.31 Variable morphology of alveolar cell carcinoma. (a) Solitary ill-defined lesion (ground glass nodule), (b) Pneumonia-like subpleural opacities with pseudocavitation (air-filled bronchi), (c) Disseminated confluent acinar opacities, (d) Positive "CT angiogram"
Whenever CT demonstrates multiple lesions that show the morphologic features of bronchial carcinoma or focal pneumonia, alveolar cell carcinoma should be considered in the differential diagnosis. Pleural effusion is seen in approximately onethird of cases, and mediastinal lymphadenopathy is seen in one-fifth. Cavitation is rare. The "CT angiogram" sign appears when pulmonary vessels coursing through the tumor show greater enhancement on postcontrast scans than the surrounding tumor. While a "CT angiogram" is commonly seen with alveolar cell carcinoma, it is not specific and may also occur in consolidation due to lymphoma, pneumococcal infection, or tuberculous infection (Fig. 9.31 d). Lymphoma
Pulmonary involvement by malignant lymphoma occurs in up to 40% of cases and is commoner in Hodgkin's lymphoma than in the nonHodgkin's type (6%) (Table 9.23). In Hodgkin's disease, more frequently, the lung is secondarily involved by widespread mediastinal disease, primarily involvement of the lung parenchyma is extremely uncommon. Recurrent Hodgkin's disease, however, is often
c
d
caused by greater vascular enhancement relative to the surrounding tumor. This sign is common but not specific for alveolar cell carcinoma. Note also the calcified pleural plaque secondary to asbestos exposure.
Lymphoma Table 9.23 Hodgkin's
Types of Lymphoma and lymphoid lymphoma
Lymphoma (WHO classification, listing only the most frequent subtypes)
Non-Hodgkin's B-Cell
Neoplasm
• Immature (precursor cell) 8-cell neoplasm Lymphoblastic leukemia/lymphoma • Mature (peripheral) 8-cell neoplasm Chronic lymphatic leukemia/smail cell B-cell lymphoma myeloma extranodal marginal cell B-cell lymphoma (MALT) follicular lymphoma mantiecell lymphoma diffuse large cell lymphoma (DLBCL) Burkitt lymphoma T-Cell
Neoplasm
• Immature (precursor cell) T-cell neoplasm Lymphoblastic leukemlallymphoma • Mature (peripheral) T-cell neoplasm Peripheral T-eelllymphoma (PTL) Anqioimmunoblestic T-cell lymphoma (see below) Anaplastic MycosIs funfIOJd,es/,;ezB'ry Angiocentric
immunoproliferative
lesions
• Grade 1: benign lymphocytic angiitis • Grade 2: lymphoid granulomatosis • Grade 3: lymphoma Lymphoid • Focal:
hyperplasia
nodular lymphoid hyperplasia, pseudolymphoma
• Diffuse
Lymphoid interstitial pneumonia (LIP) Follicular bronchiolitis
within the thorax along the radiation treatment port margins. Less than 50% of nonHodgkin's lymphoma initially develop a thoracic manifestation. The most frequent manifestation of thoracic involvement in non-Hodgkin's lymphoma (NHL) is mediastinal or hilar lymph node enlargement. If lymphoma is limited to the lungs (with or without mediastinal lymphadenopathy) it is classified as primary pulmonary lymphoma (no extrathoracic involvement for at least three months), while cases with previous or concomitant extrathoracic lymphoma are classified as secondary pulmonary lymphoma. The most recent working formulation subdivides B-cell non- Hodgkin's lymphoma into low, intermediate, and high grade, and predominantly T-cell angioimmunoproliferative lesions (lymphoid granulomatosis) .
Mucosa-associated lymphoid tissue (MALT) represent one type of intrapulmonary low grade B-cell lymphoma in the lung, also described as bronchus-associated lymplzoid tissue (BALT). Also most high grade NHL are B-cell lymphomas. Patients may suffer from underlying disease such as AIDS, or have undergone organ transplantation. In contrast to patients with low grade lymphoma, they suffer from respiratory or systemic symptoms. High grade non-Hodgkin lymphomas tend to recur at the initial sites of disease. Angiocentric immunoproliferative lesions represent a histologic spectrum and are therefore subdivided into three grades, dependent on the severity of cytologic atypia and polymorphism. While grade 1 describes a benign lymphocytic angiitis, grade 2 is defined as lymphoid granulomatosis, and grade 3 represents true lymphoma. From the literature it appears that there is also some overlap between pulmonary lymphoid hyperplasia and low grade lymphoma, and it remains unclear whether the first represents a truly benign (possibly reactive) process, a premalignant form, or a malignancy difficult to diagnose. Focal
lymphoid hyperplasia
(synonyms:
nodular lymphoid hyperplasia, pseudolymphoma, well-differentiated lymphocytic proliferation, monoclonal B-cell proliferation) can be differentiated from diffuse pulmonary lymphoid hyperplasia. For the latter, two varieties have been described: lymphoid interstitial pneumonia (LIP), predominantly affecting the parenchymal interstitium, and follicular bronchiolitis, predominantly affecting the peribronchiolar interstitium (see also p. 300). Secondary pleuropulmonary involvement with lymphoma in patients with extrathoracic disease is much commoner than the primary condition. Spread occurs either by direct involvement from mediastinal or hilar lymph nodes or by hematogenous dissemination. Post-transplant lymplioproliferative disorder (PTLD) is seen in about 3% of organ transplant recipients and is related to immunosuppression and the Epstein- Barr virus. CT
Morphology
Lymphomas have a broad, nonspecific spectrum of CT findings that range from miliary nodulation and discrete nodules to ground-glass opacities and confluent, pneumonia-like patches of
325
326
9
Lungs and Tracheobronchial System
c Fig. 9.32 CT appearances of pulmonary lymphoma. (a) Well-defined nodule, (b) Patchy nodules with nodular pleural thickening and pleural effusion, (c) Consolidation with perifocal ground-glass opacification in
consolidation with or without an air bronchogram (Fig. 9.32). The pattern of intrapulmonary Hodgkin and non-Hodgkin's manifestations can be identical. In Hodgkin's disease the commonest findings are ill-defined nodules or masses spreading out from the hila along the central bronchovascular bundle. The appearance may be similar to that of sarcoidosis or Kaposi sarcoma. Air bronchogram and cavitations are frequent findings. Recurrent disease also demonstrates intrapulmonary or perihilar nodules and mass lesions, sometimes along the radiation treatment port margins. Care should be paid to the diaphragmatic and pericardia! lymph nodes that are not included in the mantle radiation treatment. In non-Hodgkin's disease chest wall involvement is rather common and results from contiguous spread from the mediastinum. The pleura may be involved by intraparenchymal disease as well as by chest wall involvement. While plaque-like pleural thickening represents lymphomatous spread, pleural effusions are mostly due to lymphatic obstruction and do not represent direct pleural involvement. The lung parenchyma is a common site of involvement by extranodal NHL. Though there is morphologic overlap, some characteristic features or the different grades of NHL can be described: CT findings in low grade NHL (MALT or BALT) comprise solitary nodules or ill-defined focal opacities between 2 and Scm in size with air bronchograms in about 50% (Fig. 9.33). Bronchi
d
acute myeloic leukemia, resembling invasive aspergillosis. (d) Ground-glass opacifications sparing the subpleural space.
Fig. 9.33 MALT lymphoma. (a) III-defined focal opacity. (b) Diffuse micronodular opacities with tree-in-budpattern and bronchial wall thickening simulating acute endobronchial spread of infection.
within affected lung parenchyma may appear stretched and slightly narrowed. Less commonly there are multiple nodules or infiltrates. They slowly grow over months or years. Lymphadenopathy is primarily seen in less than 5%. CT findings of high grade NHL are nonspecific and range from focal to diffuse consolidations with or without air bronchogram and occasionally with a diffuse reticular pattern. Patients may show rapid progression of consolidations with respiratory failure. The commonest manifestations of englecentric immunoproliferative lesions are multiple
Kaposi Sarcoma
nodules or masses between 0.5 and 8 cm in size. They may start as ill-defined opacities, and develop to confluent nodules or masses. In subpleural location they mimic pulmonary infarcts. Cavitation is seen in up to 40%; air bronchograms are extremely rare. Since medium to small-sized vessels are involved it mimics the appearance ofWegener's disease in the lung but never involves the kidneys (see also p. 328). A focal lymphoid hyperplasia presents with a solitary nodule or a focal area of consolidation usually limited to one lobe, 2 to 5 cm in diameter and rarely exceeding to masses and infiltrates with W cm diameter. Virtually all lesions contain air bronchograms. Typically there is no associated lymphadenopathy; if there is, lymphoma should be suspected. UP is frequently associated with diseases with abnormal immune reaction (Table 9.24, see Fig. 9.65 b). Follicular bronchiolitis has been seen in up to 20% of patients with rheumatoid arthritis. The most frequent abnormalities consist of bilateral areas of ground- glass opacification, small nodules, and mostly centrally located cysts. In follicular bronchiolitis CT shows small nodular opacities in a peribronchiolar and centrilobular location (1-3 mm in diameter, rarely up to 10 mm). In contrast to primary pulmonary lymphoma, secondary involvement tends to affect the central airways resulting in atelectasis and poststenotic pneumonitis. Lymphatic congestion presents with thickened interlobular septa. The commonest intrapulmonary manifestations are solitary or multiple ill-defined nodules or masses, most frequently in the lower lobes, with linear stranding extending into the surrounding parenchyma. Cavitation is rare. Post-transplant lymplioproliferative disease presents with solitary or multiple nodules and occasionally hilar adenopathy. The lesions usually regress with decrease ofimmunosuppression. Table 9.24
Diseases associated with LIP
Sjogren's syndrome
AIDS Chronic hepatitis Renal tubular acidosis Myasthenia gravis Autoimmune thyroiditis Systemic lupus erythematosus (SLE) Primary biliary cirrhosis Allogenic bone marrow transplantation
Kaposi Sarcoma
Kaposi sarcoma in HIV -infected patients may involve the lung in addition to the skin and gastrointestinal tract. The overall incidence is on the decline (from 60% to the current 15-20%). •
CT Morphology
CT typically demonstrates cloudy nodular masses or patchy opacities that show a predominantly perihilar distribution (bilateral, often asymmetrical) and are oriented along the bronchovascular structures. This may be accompanied by nodular thickening of the interlobular septa, creating a pattern that resembles carcinomatous Iym phangitis. The lesions have ill-defined margins due to infiltration of the surrounding interstitium (local lymphangiosis), leading to spiculated, lanceolate, or flameshaped densities. Similar findings are seen with angiosarcoma (Fig. 9.34). Endobronchial tumor spread can lead to atelectasis. Being highly vascular, Kaposi sarcoma enhances intensely after intravenous contrast administration. Extensive pleural effusions and mediastinal lymphadenopathy are commonly found.
a
b
Fig. 9.34 Kaposi sarcoma (a) appearing as multiple perivascular focal opacities surrounded by halos caused by lesional hemorrhage. Perilesional hemorrhage is also seen with pulmonary angiosarcoma (b).
327
328
9 Lungs and Tracheobronchial System
Pulmonary Blastoma
Pulmonary blastoma is a rare primary lung tumor, histologically consisting of immature epithelial and mesenchymal components. There is a biphasic age distribution with peaks in the 15t and 7th decade. The prognosis is poor and appears to be dependent on the size of the primary tumor. A histologic variant resembling pulmonary blastoma without its sarcomatous stroma is named "well differentiated fetal adenocarcinoma" or "pulmonary endodermal tumor resembling fetal lung" . •
CT Morphology
CT demonstrates a large well-defined intrapulmonary nodule or mass with diameters ranging from 2.5 to 25cm. Occasionally metastases and pleural invasion are seen. Rarely tumors are cavitated, calcified or multiple. Well differentiated fetal adenocarcinoma tend to be smaller than the classic blastoma. Radiologic findings are nonspecific consisting of a peripheral nodule or mass. Pulmonary Infarction
The thromboembolic occlusion of a pulmonary artery restricts blood flow to the more distal lung segments, leading in 10-15% of cases to pulmonary infarction. The hypoperfused tissue is prone to secondary infection (postinfarction pneumonia) or abscess formation (postinfarction abscess) and may cavitate when a connection with the bronchial system is established (postinfarction cavitation). •
CT Morphology
Pulmonary infarction typically appears as a wedge-shaped density that is broadly based on the pleura and points to the hilum (Fig. 9.35). The lower lobes are more commonly affected than the upper lobes. Contrast-enhanced scans show a central hypodense area (necrosis) and peripheral rim enhancement (inflammatory reaction). An accompanying pleural or pericardial effusion may be found.
a Fig. 9.35 Pulmonary infarction presents as a pleuralbased opacification with reduced contrast enhancement that is typically wedge-shaped (a), but may be more rounded (b).
Wegener's Disease
Wegener's disease represents a systemic disease with necrotizing granulomatous vasculitis involving several organs. Approximately 90% of patients develop pulmonary manifestations (renal changes occur in 85%, joint changes in 50%), thus the lung has to be examined obligatorily once the diagnosis of Wegener's disease has been established. Approximately 90% of patients have an increased level of C-ANCA antibodies, which is also seen in patients with microscopic polyangiitis, progressive glomerulonephritis, Churg-Strauss, and some infectious diseases (aspergillosis, amebiasis, bacterial endocarditis, chromomycosis). Wegener's disease is characterized by a necrotizing granulomatous vasculitis of the upper respiratory tract, a disseminated vasculitis (small and medium-sized vessels) and a focal necrotizing glomerulonephritis. •
CT Morphology
Typical CTfindings are multiple intrapulmonary nodules or focal densities (0.5 to 10cm) predominantly in the lower zones and frequently subpleurally located. The nodules may show spiculae or feeding vessels that are also described for metastases or infarcts. Cavitation is seen in up to 50% of cases, resulting in thickwalled cavities with irregular inner margins (see Fig. 9.36 b) that are susceptible to secondary infection (e.g., with Aspergillus], an air broncho-
Churg-Strauss Syndrome
, a
b
c
d
Fig.9.36 Wegener's Disease, (a) Diffuse confluent acinar opacities due to pulmonary hemorrhage, (b) Chronic changes during resorption of pulmonary
hemorrhage, (c) Thin-walled nodule with cavitation an air-fluid level, (d) Thick-walled cavity.
gram in 25% and pleural effusion also in about 25%. Women frequently show a smooth subglottic tracheal stenosis typically over 3 to 4cm length. CT shows inflammatory reactions around the thickened tracheal ring. Strictures can also be located in the main or lobar bronchi. A pneumothorax or pneumoserothorax are the result of subpleural cavitations and the development of bronchopulmonary fistulae. Diffuse bilateral ground-glass opacifications or consolidations are caused by alveolar hemorrhage (Fig.9.36a). Recurrent hemorrhage results in fibrotic changes. Nodules and hemorrhage are seen in active disease, while fibrotic parenchymal bands and septal thickening indicate an inactive phase. Follow-up studies may show marked resolution of parenchymal changes. The lymphomatoid granulomatosis shows identical CT features to Wegener's disease (it does not affect the kidneys) but is considered a neoplasticlymphoproliferativedisease.
phase characterized by pulmonary or gastrointestinal eosinophilic infiltrates and peripheral eosinophilia, and the vasculitis phase with hemoptysis, purpura erythema urticare, glomerulonephritis and gastrointestinal ulcers, depending on the affected organs.
Churg-Strauss
•
and
CT Morphology
The most frequent HRCT findings are patchy ground-glass opacifications and consolidations predominantly in subpleural locations. Rarely, there are bronchiectases, bronchial wall thickening, centrilobular nodules, and septal thickening, the latter caused by cardiac insufficiency and interstitial edema (Fig. 9.37).
Syndrome
Churg-Strauss syndrome (allergic angiitis or allergic granulomatosis) is defined as a combination of systemic necrotizing vasculitis affecting the small and medium sized vessels in at least two organs (e.g., heart, lung, skin, nervous system, or kidneys), blood eosinophilia, and lung infiltrates. There are three clinically distinctive phases: the prodromal phase characterized by asthma and allergic rhinitis, the eosinophilic
a
b
Fig. 9.37 Churg-Strauss syndrome demonstrates areas of consolidation and ground glass opacification indicative of pulmonary hemorrhage. Thickening of interlobular (a) or intralobular septa (b) may be present.
329
9 Lungs and Tracheobronchial
System
The following diseases are associated with a peripheral eosinophilia: idiopathic pulmonary fibrosis, allergic bronchopulmonary aspergillosis, chronic eosinophilic pneumonia and Churg-Strauss syndrome.
Vascular
Malformation (AVM)
More often congenital than acquired (post-traumatic) , pulmonary vascular malformations are characterized by the abnormal shunting of blood between the pulmonary arteries and veins or, less commonly, between the bronchial and pulmonary arteries or between the bronchial arteries and pulmonary veins. Because there is no capillary network between the arterial and venous vessel, blood is shunted to the left atri urn without being oxygenated. From 40 to 60% of patients with pulmonary arteriovenous malformation (pA VM) suffer from Osler-Weber-Rendu disease with cutaneous and mucosal teleangiectasias and AVM also in other organs (e.g., liver). •
CT Morphology
Arteriovenous malformations appear as rounded opacities that measure several millimeters to several centimeters in diameter and typically have a feeding arterial vessel and a draining venous vessel (Fig. 9.38). Multiple feeding and draining vessels are seen in complex pAVMs. Most pAVMs are solitary (60-70%) and
a
b
Fig. 9.38 Arteriovenous malformation appearing as a group of peripheral nodules (a). The feeding and draining vessel in another AVM (4 x 1 /6) is demonstrated on a MIP display (b).
are located in the central part of the lung. Calcifications have been described. With spiral CT, an pAVM can be identified on unenhanced scans by its vascular anatomy, which is most clearly demonstrated by shaded surface displays and MIP images (Fig. 9.38). Delineation of the angioarchitecture is important for planning of interventional therapy (embolization). A sign, pathognomonic for pAVM, is its intense contrast enhancement synchronous with the arterial vascular opacification. Only rarely does throm bosis lead to a lack of enhancement.
Infectious Diseases
As a rule, pulmonary infections in themselves are not an indication for CT scanning. CT is used mainly to detect complications (cavitation, abscess formation); detect or exclude predisposing diseases (bronchiectasis, tumor); differentiate among infiltration, atelectasis, and effusion; monitor progression and therapeutic response; and localize lesions prior to bronchoscopy or lavage. Clinical features are important for differentiating various infectious agents. Communityacquired pneumonias occurring in previously healthy individuals have another spectrum of infectious agents as compared to nosocomial
pneumonias acquired in the hospital by patients who were already ill (Table 9.25). Immunocompromised patients are susceptible to a greater diversity of infections that also include atypical forms of pneumonia [Pneumocystis ceruiu, cytomegalovirus, mycobacterial infections, and fungal infections; Table 9.26). Experience has shown that chest radiographs are negative in approximately 10 of patients with pulmonary infection. This percentage is substantially higher in immunocompromised patients. Today, the clinical suspicion of pulmonary infection in a patient with a nega%
Infectious
Table 9.25
Common
Community-acqu
infections
in the normal
Diseases
host
ired pneumonia
Bacterial • Streptococcus
lower lobes,
• Staphylococcus
postviral,
• Haemophilus
lower lobes,
consolidation,
children>
rapid spread,
adults,
cavitation,
bronchopneumonia
round
(children)
or sublobar,
effusion
and cavitation
rarely lobar
empyema
(COPD),
rarely
Atypical diffuse
• Mycoplasma
tions,
centrilobular mosaic
nodules,
pattern
tree in bud, patchy
with air trapping,
• Legionella
upper lobes, segmental
• Chlamydia
similar to Mycoplasma
consolidation,
(lobular)
lymphadenopathy rapid spread
ground
glass
or consolida-
in up to 40%
to diffuse
disease
Vims diffuse acinar shadows, pathyor diffuse ground solidations due to bacterial superinfection
• Influenza
in small
• RSV' • Adenovirus • Varicella
zoster
Nosocomial
children,
bronchiolitis
consolidation
like in bacterial
diffuse acinar
opacities,
glass opacification
frequently
con-
with airtrapping infection
ground
glass,
nodular
calcifications
postinfectious
pneumonia
Bacterial • • • •
see above, septic emboli (catheters) lower lobes, consolidation or multiple nodules, upper lobes, expanding consolidation, necrosis lower lobes, consolidation, multiple cavitations
Staphylococcus Pseudomonas Klebsiella Escherichia coli Respiratory
Table
9.26
syncytial
Common
virus
infections
in immunocompromised
patchy
Bacterial Viral infections • Cytomegalovirus • Herpes
(CMV)
simplex
Pneumocystiscarinii
cavitation and cavitation
patients
or lobar consolidations
patchy or diffuse ground glass, consolidation post transplantation (first 4 months), under chemotherapy, HIV«20 c ells/mm") associ ated with mucocutaneous ground glass
consolidation
disease ->
fibrosis.
cysts
HIV Atypical Fungal
mycobacterial infections
acinar
or nodular
patchy opacities,
opacities,
consolidation
consolidations,
halo sign
• Candida • Aspergillus
HIV, under chemotherapy,
post transplantation
• Cryptococcus
tive chest radiograph is considered an indication for CT or HRCT. Given the nonspecific response pattern of the lung. it is only possible to reduce the number of causative organisms based on morphologic criteria. Clinical features (e.g .. CD4
cell count) can help to further reduce the differential diagnosis of etiologic agents (Tables 9.279.29).
331
9
Lungs
Table
and Tracheobronchial
9.27
(with respect After
Common
System
infections
after
bone
marrow
< 30 days
Bacterial Pneumonia
transplantation
Aspiration Gram-negative Aspergillus
Pneumocystis Cytomegalovirus (CMV) Idiopathic pneumonitis Graft-versus-host reaction
> 4 months
Streptococcus, Staphylococcus Varicella Graft-versus-host reaction
solid organ
transplantation
< 30 days
Aspiration Gram-negative bacteria Catheter infections
< 4 months
Aspergillus Nocardia,
Mycobacteria
Pneumocystis Viral infections > 4 months
A lobar pneumonia is characterized by diffuse alveolar wall injury accompanied by hemorrhagic edema. Bronchopneumonia (lobular pneumonia) results from infectious organisms in the epithelium of distal airways causing peribronchiolar exudates. Chronic (carnifying) pneumonia is considered to be present if the infiltrated alveoli do not clear within 15 to 30 days. This condition involves a proliferative fibroblastic inflammation with scarring and fibrous change. Depending on their etiology. bacterial infections are divided into community-acquired and nosocomial infections. The radiographic pattern is quite variable depending on the organisms. One organism may produce different patterns depending on the progression of disease. the patient's immune status, or preexisting or concomitant disease.
bacteria
< 4 months
After
Infectious Diseases in the Normal Host
transplantation
to time of occurrence)
(e.g., CMV)
Pneumocystis Cryptococcus
Table
9.28
Predominant
pattern
of opportunistic
infections
Consolidations
Nodular/patchy
Bacteria Legionelia Mycobacteria Nocardia
Aspergillus Cryptococcus Nocardia Bacterial abscess Septic emboli
Cryptococcus
Table 9.29
Incidence
of HIV-associated
pulmonary
opacities
infections
Diffuse
ground-glass
opacities
Pneumocystis Virus
related
to immune
status
(CD4 cell count)
Bacterial pneumonia Nocardia Typical tuberculosis
any cells/cm3 < 200 cells/cm3 any cells/em'
MAC
< 50 cells/cm3
Histoplasmosis Cryptococcus Aspergillus
< 100 cells/cm3 < 100 cells/cm3 < 50 cells/em'
Viral
Cytomegalovirus
<
Protozoan
Toxoplasmosis
<
100 cells/em'
<
200 cells/em'
Bacterial
Fungal
Pneumocystis
20 cells/em'
Infectious Diseases in the Normal Host
•
-Delayed resolution beyond two months requires exclusion of an underlying systemic disease and evaluation of the immune status it should also initiate the search for complications of infections such as atelectasis, empyema, or cavitation, or for a predisposing local factor, such as an obstructing neoplasm, or bronchiectasis.
CT Morphology
The typical pattern in bacterial pneumonia is focal or patchy areas of consolidation with a positive air bronchogram (Fig. 9.39). Broncho pneumonia features multiple peribronchial air space opacities that frequently involve several lobes. They may become confluent and produce. segmental or lobar air space consolidations. The\ opacities show decreased contrast enhance ment that is very nonhomogeneous in some cases. Central cavitation and lung abscesses may Atypical (Nonviral) Pneumonia develop, and fluid levels-may-be.seen in-cavities that communicate with airways (see Fig.9.46b). Atypical pneumonias refer to infections that do not respond to antimicrobial therapy or which A hypo dense border or "halo" is caused by perifocal hemorrhage (compare Fig. 9.41 d). have different clinical features compared to bacterial infection (fever without chills, nonproA lobar pneumonia (e.g., Streptococcus ductive cough, extrathoracic symptoms such as pneumoniae, Klebsiella, Legionella pneumophila right upper quadrant pain headache, myalgias and Mycoplasma pneumoniae) is characterized by a lobar infiltration that is sharply marginated or diarrhea). Atypical non-viral pneumonias are by the interlobar fissure (Fig.9.39b). A lobar caused by Mycoplasma pneumoniae, Legionella pneumonia may result in expansion of the lobe pneumophila and Chlamydia. due to extensive edema (most typical of Kleb• CT Morphology siella pneumoniae). Chronic (carnifying) pneuMycoplasma pneumonia is characterized by a monia is characterized by scarring and fibrotic change that may involve an entire lobe, loss of bronchiolitis progressing to a bronchopneuvolume, and bronchiectasis (Fig.9.39d). Most monia. Accordingly the dominant finding of pneumonias resolve radiologically within a acute Mycoplasma pneumoniae infections are month, often within 10 to 21 days. centrilobular nodules (tree-in-bud), signs of airway obstruction with air trapping, and mosaic perfusion in addition to ground-glass opacifications and patchy consolidations in a lobular dis-
a
b
Fig. 9.39 Bacterial pneumonia, (a) Incipient patchy bronchopneumonic infiltrates, (b) Upper lobe pneumonia with air bronchogram and opacifications of varying density. Note the volume loss under treatment (arrowheads). (c) Pneumococcal pneumonia (4x1/6,
c
d
2.4mGy, sagittal MPR) with complete consolidation of segment 5 of the middle lobe with a positive CT angiogram sign (arrow), (d) Chronic carnifying pneumonia with a sharply circumscribed area of consolidation, traction bronchiectasis and signs of volume loss.
333
9 Lungs and Tracheobronchial System •
a
b
Fig. 9.40 Atypical pneumonia, (a) Multiple centrilobular nodules with tree-in-bud pattern due to mycoplasma-induced bronchiolitis. (b) Segmental consolidation with positive air bronchogram in a patient with Legionella infection.
tribution (Fig. 9.40). In children, HRCT may show signs of airway obstruction after an interval of 1 to 2 years in those areas that had been infiltrated. In legionnaires' disease the typical pattern consists of segmental consolidation, predominantly in the upper lobes, with rapid spread to the entire lobe and the adjacent lobes. The parenchymal changes are extensive. A pleural effusion is rare. The radiologic findings of Chlamydia pneumonia may be identical to those of mycoplasma pneumonia with a predominance for patchy areas of consolidation in the middle or lower lobes.
CT Morphology
Viral respiratory infections progress in a predictable sequential fashion. After inhalation of the agent, the inflammatory process initially starts in the airways. CT shows patchy areas of heterogeneous parenchymal attenuation (Fig. 9.41 a, b) and a diffuse tree-in-bud sign indicating bronchiolar impaction. Scans in the subacute phase show the accentuation of interlobular septa by lymph vessels thickened due to the absorption of intra-alveolar exudate in addition to the thickened central bronchovascular interstitium and the bronchiolar abnormalities. With further progression, infection also involves the alveolar walls inciting intra-alveolar edema and hemorrhagic inflammation. The results of filling in of peribronchiolar air spaces are patchy consolidations resembling a bronchopneumonia pattern. Peri bronchiolar consolidation may also cause an acinar pattern. Perifocal hemorrhage produces a hypodense border, or halo (Fig. 9.41 c). Influenza pneumonia is most frequently complicated by bacterial pneumonia and may therefore develop more extensive consolidations (Fig. 9.41 c). Radiologic features of adenovirus pneumonia are very similar to bacterial infection (pneumococcal pneumonia). RSV occurs most frequently in children, predominantly involving the bronchioles. Varicella can cause severe pneumonia in adults. It presents with diffuse ill-defined acinar nodules (4 to 6mm). The findings persist over several weeks and may heal with discrete nodular calcifications. Hilar adenopathy is seen in up to 40% of cases.
Viral Pneumonia
Primary respiratory viruses are the parainfluenza and influenza group (most frequent agent in adults), the respiratory syncytial virus (RSV, most frequent in children), the adenovirus, and the picornavirus. The infection begins in the central large airways and only produces parenchymal changes when disease involves the peribronchial or peribronchiolar alveoli. Especially in children it predominantly affects the small airways leading to airway obstruction.
Anthrax (Bacillus anthracis)
Bacillus anthracis is a gram-positive spore-forming organism. After intra-alveolar inhalation the spores are phagocytosed and transported via lymphatics to hilar lymph nodes where they germinate into the vegetative form. The latter pass via efferent lymphatics into the systematic circulation. In addition to cutaneous and gastrointestinal manifestations, the thoracic manifestation represents the most dramatic form and is lethal in the vast majority of untreated patients. Pathologic organ manifesta-
Tuberculosis and Atypical Mycobacterioses
••••
a
b
~ d
c
Fig. 9.41 Viral pneumonia may show various appearances ranging from focal to diffuse ground-glass opacification or dense consolidation, (a) and (b) CMV pneumonia with patchy ill-defined peribronchial (a) or focal
(b) ground-glass opacification. (c) Diffuse ground-glass opacities in CMV infection, (d) Influenza pneumonia with patchy, confluent opacities, air bronchogram, and perifocal halos due to hemorrhage.
tions are thought to be caused by toxin overload duringsepti cemia.
ment of cavitation. As the immune status decreases further, consolidations are more widespread and cavitation is less common. Patients may develop large bihilar lymphadenopathy. Atypical mycobacterial infections occur in late stage AIDS «50 cells/em"). Patients may show the same features as in typical mycobacteriosis, lymphadenopathy with nodular consolidations, or even lymphadenopathy alone.
•
CT Morphology
Hemorrhagic and thus hyperdense enlargement of mediastinal and hilar lymph nodes is the most typical finding. Peribronchial thickening may be due to inflammation or lymphatic obstruction. Bilateral hemorrhagic pleural effusions are frequent. Intrapulmonary ground-glass opacifications or consolidations are due to intraalveolar edema and hemorrhage. Tuberculosis and Atypical Mycobacterioses
CT is more sensitive than conventional radiography for detecting and following cavitation, cavity superinfections (often with Aspergillus'), bronchopleural fistulas, pleural empyema, and mediastinal lymphadenopathy. Infection with HIV or any other condition of immunosuppression carries a risk of development of tuberculosis. While patients with HIV may suffer from tuberculosis at any CD 4 count, appearance of tuberculosis changes with the immune status. In patients with existent immune reactions (CD4>200 cells/em") tuberculosis has the typical reactivation pattern, with areas of consolidation in the apical and posterior segment of the upper lobes or the apical segment of the lower lobes, with the develop-
•
CT Morphology
CT morphology varies according to the host immune status and the course of the disease. 111defined acinar shadows are caused by endobronchial spread of the infection and always signify active disease. The shadows are 2-10 mm in size, show a peribronchial (centrilobular) distribution, and tend to coalesce to form rosettes or patchy opacities (Fig.9.42a). Miliary lesions result from hematogenous dissemination of the disease. They appear as discrete nodules 1-2 mm in diameter typically distributed throughout the lung parenchyma and showing a predominantly perivascular and paraseptal distribution. Patchy areas of consolidation with a positive air bronchogram signify a "caseous pneumonia" in which the bronchi often show irregular (varicose) dilatation. Differentiation from other forms of lobar pneumonia is aided by noting the small satellite lesions that are usually seen in tuberculosis.
335
336
9 Lungs and Tracheobronch'ial System
a
b
c
d
Fig. 9.42 Various appearances of mycobacterial tuberculosis. (a) Miliary disease, (b) Bronchogenic tuberculosis with ill-defined, confluent acinar nodules in a centrilobular (peri bronchiolar) distribution, (c) Peribron-
chovascular infiltrates, multiple acinar nodules bronchogenic spread.
(d) Lobar consolidation with in the other lobes indicating
Tuberculous cavities may be thick- or thinwalled. While there is no correlation between CT morphology and disease activity, the thickness of the cavity walls decreases in response to therapy and in cases that become chronic (Fig.9.4 2a). Enlarged mediastinal lymph nodes typically show central necrosis and intense rim enhancement. Atypical mycobacterioses [Mycobacterium aviumcomplex=MAC, Mycobacteriumfortiiitum, Mycobacterium kansasii, and others) have imaging features that resemble classic tuberculosis, including the presence of acinar, nodular, or miliary opacities and accompanying lymphadenopathy. There is no upper lobe predominance, however, and cavitation is uncommon. Bronchiectasis is also seen. When miliary opacities are found, the differential diagnosis should include pneumoconioses, fungal infections, and hematogenous metastases. Ill-defined acinar shadows are consistent with alveolar cell carcinoma, extrinsic allergic alveolitis, or viral infection (Table 9.22).
Infectious Diseases in Immunocompromised Patients
Bacterial Pneumonia
The most commonly encountered types of infection in the immunocompromised patient are bacterial in origin. Many of the bacteria that infect the immunocompromised patient are also capable of producing illness in the healthy person. In the immunocompromised patient, however, bacterial infections have the propensity to progress more rapidly, the pneumonia is more severe, and the recovery prolonged. Patients with a decreased humoral immunity (multiple myeloma, chronic lymphatic leukemia, lymphoma, graft -versus-host disease, HIV) are at increased risk of developing a pneumococcal pneumonia, an infection resulting from the encapsulated (more virulent) form of Haemophilus influenzae or Mycoplasma pneumoniae. Chronic obstructive lung disease is a major risk factor for developing H. influenzae pneumonia because many of these patients are colonized with these bacteria. Hospitalization results in an increased risk of developing pneumonia with gram-negative organisms [Pseudomonas, Klebsiella, Acinetobacter, Escherichia coli, Serratia, Enterobacter}. Staphylococcus aureus is second to gram-negative bacteria as a cause of nosocomial pneu-
Infectious Diseases in Immunocompromised Patients monta, especially in patients with granulocytopenia and chronic granulomatous disease, Immunosuppression is a predisposing factor in the development of nocardiosis, Solid organ transplantation, especially heart transplantation, is associated with an increased incidence of Nocardia infections, Patients with leukemia, or lymphoma are also at increased risk, In patients with HIV infection, nocardiosis often occurs simultaneously with other infections such as mycobacterial tuberculosis, or My-
cobacterium evium.
Immunocompromised patients are more susceptible to developing an infection with Legionella; other risk factors include chronic pulmonary disease, diabetes, alcohol abuse, smoking, and old age, Infected central venous catheters account for most nosocomial bacteremias [Staphylococ-
cus aureiis, Staphylococcus epidermidis), •
CT Morphology
The typical pattern in bacterial pneumonia is focal or patchy areas of consolidation with a positive air bronchogram, The pneumonia begins as an area of poorly defined opacity, The opacity increases in size to involve a segment or alobe. In infection with Haemophilus. pleural effusion is common (>50%), cavitation is rare and usually occurs in the lobar form, Cram-negative bacteria develop a patchy bronchopneumonic pattern, abscess formation may develop in any of the gram-negative pneumonias but is most frequently encountered with
Pseudomonas. The commonest manifestation of nocardiosis is air space opacities that may involve large areas of the lung. Mass-like densities are also seen. Small diffusely distributed nodules are a less frequent pattern. More than 60% of patients have pleural effusions. Cavitations occur in consolidations and nodules.
Viral
and
Pneumocystis
carinii
Infections
Viral and Pneumocystis caIinii pneumonia (PCP) are common pneumonias in patients with impaired cell-mediated immunity. Because of overlapping imaging features they are often considered simultaneously in the differential diagnosis.
Pneumocystis carinii is the commonest causative agent of pneumonia in HlV-infected patients (60-75% of all patients have at least one episode of PCP). The infection rate has been declining since the introduction of inhalational chemoprophylaxis. Cytomegalovirus (CMY) infections are common in patients during chemotherapy or in patients following organ transplantation within the first four post-transplant months (>50%, especially after renal or liver transplantation). Reactivation CMV infection (in seropositive patients) is more common. Since introduction of prophylactic anti-CMV therapy, the incidence of CMV pneumonitis in bone marrow transplant (BMT) recipients substantially decreased (usually 50th to 60th day after transplantation, more frequent after allogenic than after autologous BMT). If it occurs it is still associated with high mortality. CMV is less common in the AIDS population. CMV pneumonia in AIDS should be considered only in patients with severe immune compromise (<100 cells/em'). especially when extrathoracic CMV infections have been documented. Herpes simplex virus (HSY) pneumonia is rare and usually occurs after clinically evident mucocutaneous disease. Squamous metaplasia in the tracheobronchial tree (after radiation or chemotherapy) predisposes patients to lower respiratory HSV infections. CT Morphology
CT findings in CMV pneumonia include signs of airway disease including bronchial wall thickening, tree-in-bud, and bronchiectasis. In the stage of air space involvement, hazy granular opacities or-less commonly-dense consolidation are noted. Nodules or masses up to several cm in diameter are also reported (in about 60%). In fact, the combination of air space consolidation and interstitial pattern appears to be the commonest finding. Abnormalities tend to be bilateral and symmetric (Fig. 9.41 a). The initial appearance ofa PCP pneumonitis is that of a predominantly alveolar process with patchy ground-glass opacities showing a diffuse bilateral distribution with perihilar accentuation. In patients receiving prophylactic aerosol therapy, the changes show an upper zone predominance (Fig. 9.43). With further progression of the infection, immunocompromised patients develop extensive consolidation with an air
338
9 Lungs and Tracheobronchial System
a
b
Fig. 9.43 Pneumocystis carinii pneumonia, (a) Acute form of Pneumocysfo corini'i pneumonia with diffuse patchy ground-glass opacities, (b) Acute form of Pneumocyshs carinii pneumonia with diffuse ground-
c
d
glass (note dark bronchus sign) and thickened interlobular septa, (c) Fibrosing form of Pneumocystis carinii pneumonia with signs of parenchymal distortion. (d) Cystic changes.
bronchogram while immunocompetent patients Candida and Aspergillus species. The clinical and (receiving treatment) show a granulomatous re- radiologic features are highly variable and often action with interstitial fibrosis and nodular nonspecific. Because many of these organisms changes (Fig. 9.43). Pleural effusion and medi- can colonize the upper airways, sputum cultures astinallymphadenopathy (rarely with calcificaare considered diagnostically unreliable. Defitions) are somewhat unusual. The dynamics of nite diagnosis requires culture of the fungus the ground-glass opacities provide a criterion from infected tissue or demonstration of the orfor monitoring the patient's response to phar- ganism in microscopic examination. macologic therapy. Noninvasive (obstructing) endobronchial AIDS patients receiving prophylactic treat- aspergillosis is caused by massive intraluminal ment with aerosolized pentamidine are the sub- overgrowth of fungus without airway wall invagroup most likely to develop destructive cystic sion'and is commonest in patients with AIDS. lung changes (pneumatoceles or thick-walled Secondary colonlzatjonofa preexisting cavity cysts), which predominantly involve the upper by Aspergilus is also referred to as noninvasive lobes and the apical segments of the lower I aspergillosis.!nvasive aspergillosis denotes peri lobes. Initially, small cysts appear within con- bronchial infection of the parenchyma by transsoli dated areas and later coalesce to form larger bronchial spread of fungal mycelia from invacysts. Pneumothorax is a potential complication sion of the airway walls. It is most commonly' of subpleural cysts. The changes may resolve al- reportedlh patients with AIDS and in allogeneic most completely in response to therapy. Cysts bone marrow and lung transplant recipients, require differentiation from cavitating lesions IChronic necrotizing or semi-invasive aspergillosis caused by septic emboli or fungal infection. typically occurs in patients with chronic obstructive lung disease, sarcoidosis, or underlying malignancy, or in patients treated with corticosteroids. Opportunistic Fungal Infections Pulmonary candidiasis is uncommon and Opportunistic fungal infection is a common occurs only in severely immunosuppressed cause of serious morbidity and mortality in im- patients. It may result from aspiration of ormunocompromised patients (e.g., chemotherganisms of the upper airway, from hematogenous spread from the gastrointestinal tract, or apy- induced neutropenia, AIDS, immunosuppression after organ or bone marrow transplanfrom infected central venous catheters. Diffuse tation). The most important opportunistic fun- endobronchial dissemination of the fungi can gal pathogens include Cryptococcus neoiormans, precipitate the rare "miliary" form of candidiasis.
Infectious Diseases in Immunocompromised
Cryptococcus
neoformans
minimally pathogenic in immunocompetent patients but can cause severe, often disseminated infection in immunocompromised hosts. •
is
CT Morphology
Fungal infections with a propensity for vascular invasion (invasive aspergillosis, mucormycosis) are characterized by nodular opacities and features that resemble pulmonary infarction. The early signs of invasive aspergillosis consist of solitary or multiple pulmonary masses ranging from a few millimeters to several centimeters in size and showing a perifocal halo of ground glass density (Fig. 9.44 a). Although this "halo" sign is not specific (it is also seen in tuberculosis, CMVinfection,herpes, candidiasis, Legionella pneumonia, etc.) ,its presence in immunosuppressed patients should raise suspicion of a-fungal infection If sighs' Of-infection (fever, CRP) persist despite broad-spectrum antibiotic therapy. Mycotic vascular infiltration re'suIting in vascular occlusion leads to wedgeshaped infarcts (Fig. 9.44 c) that show peripheral rim enhancement after contrast administration , (granulation wall) but otherwise remain hy/\dehse7 It Is common to find a nondilated bronchus entering the infiltrated area and terminating abruptly within the infiltrate. Progression of the infection is_.initially characterized by small patchy opacities that resemble bronchopneumonia. Later the opacities coalesce and may form extensive areas of con-
a
b
Fig. 9.44 Pulmonary aspergillosis. (a) Angioinvasive Aspergillus: nodular opacity with a ground glass halo. (b) Airway-invasive Aspergillus: tree-in-bud sign with patchy peribronchiolar opacities, (c) The air crescent
Patients
solidation. Cavitation is a late finding, observed .after two to three weeks. The cavities typically" show an "air crescent sign that indicates a necrotizing process and Heralds an upturn in the patient's imm une status (Fig. 9.44 b.l.These findi ngs require differentiation from other types of superinfected cavities (tuberculosis, Wegener's disease, noninvasive aspergillosis). Lymphadenopathy, effusion, and the infiltration of soft-tissue structures are very rare. Intrapulmonary mycotic aneurysms pose an acutely life-threatening complication due to their potential for rupture after the pneumonia has resolved (Fig. 9.44(1). In obstructing tracheobronchial aspergillosis, CT demonstrates large branching mucoid impactions, usually in the lower lobes. Distal atelectasis or consolidation is common. Chronic necrotizing aspergillosis manifests with .slowly progressive cavitary consolidations_ that usually affect the upper lobes. Intracavitary mycetomas are seen in more than 50% of cases. Pulmonary mu'cormycoses like that ofAspergillus have a propensity for invasion of arteries. It mimics the pathologic and radiologic features of invasive aspergillosis showing consolidations that frequently cavitate; the air crescent sign is seen less commonly. Candidiasis is characterized by scattered patchy or nodular infiltrates (septic fungal emboli) predominantly in the lower lobes that also may have halos of ground-glass attenuation. Pleural effusion occur in 25%, adenopathy and
c
d
sign is virtually pajJToc)rio"rnpnjc;JQrasc)ergill.o.sJs-and signal"ITrcreasincfriost resistance, (d) Wedge-shaped subpleural infiltrate caused by a mycotically induced infarction.
339
9 Lungs and Tracheobronchial System
cavitation are rare. Most patients with pulmonary candidiasis suffer from superimposed infections. Radiologic manifestations of cryptococcosis are solitary or multiple nodules or masses. Diffuse small nodular opacities, cavitation, adenopathy and effusion are more frequent in the immunocompromised patient.
Table 9.30 cavitation
Pulmonary masses that are associated
Abscess
Lung Abscess, Cavitation and Pneumatocele
A lung abscess is a circumscribed intrapulmonary infection that is associated with parenchymal necrosis. If the abscess erodes into the Jxonchial system, the necrotic tissue is replaced .byair, and a cavity develops. The differential diagnosis of pulmonary cavities includes infec-tifius, inflammatory granulomatpus, neoplastic. and post-traumatic etiologies (Table 9.30; Fig. 9.45). A pneumatocele represents solitary or multiple focal air collections most commonly in a subpleural location resulting from alveolar rupture (e.g., Staphylococcus aureus).
cavity
Tumor
Bronchial carcinoma Metastases Hodgkin's disease Kaposi sarcoma
Granuloma
Wegener's disease Rheumatoid nodules Progressive massive fibrosis (silicosis) Sarcoidosis Histiocytosis
Trauma
Contusion or hematoma (secondarily infected) Traumatic pneumatocele
CT Morphology
A lung abscess appears as a mass of central low density within a pneumonic infiltrate or rounded opacity. Contrast administration more .clearly demarcates the hypodense-center from _the enhancing rim. Multiple small cavities or microabscesses are seen in necrotizing pneu-
a
Staphylococcus aureus Klebsiella pneumoniae Tuberculosis Fungal infections (invasive aspergillosis) Aspiration (gram-negative bacteria, anaerobes) Others: nocardiosis, actinomycosis, Pneumocystis carinii
Septic emboli Postinfarction
•
with
c
b
Fig. 9.45 Differential diagnosis of cavitating sions. (a) Infectious: thick-walled tuberculous the upper lobe. (b) Parasitic: air-fluid level and ed membrane in a ruptured hydatid cyst. (c)
monia (e.g., gram-negative bacilli, anaerobic bacteria, Staphylococcus aureus, Streptococci). If the abscess communicates with the bronchial system, the resulting air collection forms a cavitati on (Fig.9 .46a, see also Fig.lO.18c). Extensive necrosis results in vasculitis and thrombosis
lung lecavity in collapsNeopla-
d
stic: liquefying metastases from a teratocarcinoma. (d) Large cystic bronchiectasis mimicking a lung abscess. Note also the small peripheral foci of bronchiectasis, some of which are secretion-filled.
Pulmonary Edema
a
b
c
d
Fig. 9.46 Complications of pneumonia, (a) Sharply demarcated abscess within dense consolidation. (Note the inhomogeneous enhancement of the consolidation and the space occupying effect of the abscess), (b ) Cavitating pneumonia with abscess formation (arrowheads)
and air inclusions (arrows) due to connections between abscess and bronchi, (c) Cavitating pneumonia with air inclusions and pleural effusion, (d) Pneumatocele with airfluid level and signs of endobronchial spread of infection.
causing ischemic necrosis, pulmonary gangrene, and sloughed lung portions (e.g., Klebsiella pneumoniae).
Differentiation of Infectious from Noninfectious Disease
An abscess may have a similar appearance to a necrotic tumor (e.g., squamous cell carcinoma). A cavitation requires differentiation from air-filled (dilafed) bronchi, preexisting emphysema, and a pneumatocele. The cavitation wall is thicker than the pneumatocele wall and has an irregular inner margin.
Diffuse Interstitial
It can be quite difficult in immunocompromised patients to distinguish the pathologies listed in Table 9.18 from infectious pulmonary infiltrates on the basis of CT morphology.
Reticular or Linear Lung Changes
Interstitial reticular and linear lung changes (Tables 9.31, 9.32; Fig. 9.47) produce varying pathomorphologic patterns on CT scans. Foci of interstitial thickening may be uniform or irregular (beaded) depending on whether they are caused by fluid accumulation (edema), cellular infiltration (tumor, inflammation), or connective tissue proliferation (fibrosis). The goal of CT imaging is to detect subtle early changes and to locate, characterize, quantify, and follow diffuse diseases of the lung parenchyma. Individual morphologic findings are frequently nonspecific. It is necessary to analyze distribution patterns and consider mor-
phologic and clinical findings in conjunction in order to draw meaningful differential diagnostic concl usi ons. Pulmonary Edema
Pulmonary edema is not a primary indication for CT but is by far the commonest cause of increased interstitial markings and should always be considered in the differential diagnosis. Pulmonary edema may be due to either of two mechanisms: an elevation of the pulmonary venous pressure or increased permeability of
341
342
Lungs and Tracheobronchial System
9
Table 9.31
Differentia]
diagnosis
of noninfectious
parenchymal
densities
in immunocompromised
patients
Edema • Most common cause of interstitial changes in immunocompromised • Uniformly thickened interlobular septa, ground-glass opacities
patients
Drug reaction during chemotherapy • Bleomycin, • Mosaic-like Rejection
busulphan, methotrexate pattern of reticular and linear densities,
response,
• Patchy or diffuse Hemorrhage
• • • •
opacities,
and patchy consolidations
reaction after transplantation ground-glass
opacities,
opacities,
consolidations,
ground-glass
pleural effusion,
bronchial wall thickening
opacities
lymphomas
2% of all transplantation patients Weeks to years after transplantation Ground-glass opacities or solid, tumor-like infiltrates Often reversible with reduction of immunosuppressive
Table 9.32
opacities,
in myelodysplasticdiseases
• III-defined alveolar Secondary
graft-versus-host
alveolar
ground-glass
Linear and reticular changes
therapy
(see Fig. 9.47)
1 Peribronchovascular thickening Affects axial interstitium (bronchi) a Thickened bronchial walls (cuffing) b Irregular interfaces with lung parenchyma face sign)
(inter-
2 Prominent intralobular interstitium Affects septal interstitium (acinus, alveoli) • Fine reticular
pattern
3 Thickened Affects lobule)
interlobular septa peripheral interstitium
(secondary
• Lines 1-2 cm long, running perpendicular pleura ->, Peripheral arcades -> Coarse reticular pattern -» Impaired lymphatic drainage
£ 4 Parenchyma!
to the
bands
• Fibrotic parenchymal scars • Chain of thickened interlobular
septa
5 Subpleural lines • Lines running parallel to the pleura -> Reversibleatelectasis ->, Irreversible fibrosis -> Impaired lymphatic drainage 6
Fig.9.47 Schematic representation of interstitial reticular and linear lung changes (compare Table 9.32 for numbering) .
Honeycombing
• Multiple cysts (mm to ern) with thickened -> End stage ofi rreversi blefi brosis
walls
Pulmonary Edema
the alveolar capillary membrane. As the edema progresses, the tissue fluid tends to spill from the interstitium into the alveolar airspaces. The changes may be localized as a result of regional venous inflow stasis or the obstruction of lymphatic drainage (e.g., by a perihilar tumor). •
Table 9.33 densities
Peribronchovascular thickening
CT Morphology
An edematous interstitium is manifested by smoothly thickened interlobular septa, thickened bronchial walls ("cuffing"), and thickened bronchovascular interfaces (Fig. 9.48). A fine reticular pattern also may be present at the intralobular level. Subpleural lines result from obstructed lymphatic clearance due to interstitial edema. Mild cases show hazy ill-defined centrilobular opacities caused by increased prominence of the centrilobular artery and thickened perivascular interstitium. Alveolar edema is manifested by focal, patchy or diffuse areas of increased parenchyma! density that can range from ground-glass opacities to air space shadows. Frequently the density of the lung parenchyma is diffusely increased (dark bronchus sign). The subpleural space may be less involved. The vascular diameter is increased and vascular structures may be followed down to the subpleural space. All possible transitions between the two stages exist with various combinations of thickened septal lines and patchy ground-glass. Pleural effusion may be also present.
a
b
Fig. 9.48 Pulmonary edema, (a) Interstitial edema with smoothly thickened interlobular septa and an anteroposterior parenchymal density gradient. (b) Smoothly thickened interlobular septa with alveolar filling-in of exudates (4 x 1 /6, coronal MPR). (c) Alveolar
Differential diagnosis of linear and reticular
Smooth
Interstitial pulmonary edema Carcinomatous lymphangitis Pulmonary fibrosis
Irregular, nodular
Carcinomatous lymphangitis Sarcoidosis Silicosis Pulmonary fibrosis Chronic allergic pneumonitis
Prominent intralobular
interstitium
Common
Pulmonary fibrosis Asbestosis Chronic allergic pneumonitis Alveolar proteinosis Lipoid pneumonia
Rare
Carcinomatous lymphangitis Interstitial pulmonary edema Sarcoidosis, silicosis
Thickened interlobutar septa Smooth
Interstitial pulmonary edema Carcinomatous lymphangitis Alveolar proteinosis Lipoid pneumonia Lymphangioleiomyomatosis
Nodular
Sarcoidosis (active) Carcinomatous lymphangitis Silicosis (early stage) Continue
c
r-.
d
edema with patchy centrilobular opacities and sparing of the peripheral zone. (d) Atypical presentation in concomitant emphysema causing a reticular pattern with ground glass opacification.
343
9
Lungs and Tracheobronchial System
Table 'I I
9.33
ified the term BOOP and recognized two other forms; the acute interstitial pneumonia (AlP) most closely corresponding to the entity described in 1944 by Hamman and Rich and the nonspecific interstitial pneumonia and fibrosis (NSIP or NIPF, also called non-classifiable interstitial pneumonia). While the original classification implied that UlP and DIP have different etiologies and pathogenesis, other investigators have suggested thatthe patterns ofD IP and UIP represent the cellular and fibrotic spectrum of a single disease. This interpretation is supported by reports of patients who showed both patterns simultaneously in a single lung specimen and by reports of patients who showed UlP in a follow-up biopsy specimen obtained at a site originally showing DIP (Table 9.35). According to the ATS/ERS International Consensus Classification from 2000, the diagnosis of idiopathic pulmonary fibrosis (IPF) is limited to patients who have histologic findings ofUIP. According to this classification, idiopathic interstitial pneumonia with fibrosis (UlPIIPF) is separated from all other IlP such as DIP, AlP, LIP, OP (BOOP), and NSIP.
(Continue)
Thickened
interlobular
Irregular
Sarcoidosis (end stage) Silicosis, pneumoconiosis Pulmonary fibrosis Asbestosis Chronic
Parenchymal
septa
allergic
pneumonitis
bands Asbestosis Sarcoidosis Pulmonary fibrosis Silicosis, pneumoconiosis Tuberculosis
Subpleural
lines Asbestosis Pulmonary Pulmonary
fibrosis edema
Honeycombing Common
Pulmonary fibrosis Asbestosis Chronic allergic pneumonitis
Rare
Histiocytosis,
Sarcoidosis Silicosis,
pneumoconiosis
Usual Interstitial Pneumonia (DIP) Idiopathic Interstitial Pneumonia (liP), Collagen Diseases The idiopathic interstitial pneumonias (IlP) are a heterogeneous group of inflammatory and interstitial fibrosing lesions. The original classification by Liebow (1968) described five subtypes (Table 9.34); it was modified by Katzenstein and Myers (1992), who dropped LIP and GIP, modTable 9.34
Classification
Classification Usual
interstitial
Desquamative
of interstitial
by Liebow
interstitial
pneumonias Classification
1958
pneumonia pneumonia
UlP is the commonest type of idiopathic interstitial pneumonia. The heterogeneity of findings is the characteristic histological finding representing the evolution of fibrosis at early and late stages with a combination of old and active lesions, distributed in a patchy predominantly subpleural distribution with normal parenchymainterspersed.
UIP DIP
Idiopathic
Desquamative Respiratory disease Acute
Lymphocytic Giant
interstitial
cell interstitial
Bronchiolitis pneumonia
pneumonia
pneumonia
obliterans
interstitial
by the ARS/ERS
pulmonary
interstitial bronchiolitis
interstitial
Meeting
fibrosis
IPF
pneumonia interstitial
DIP lung
pneumonia
RB-ILD AI P
LIP
Lymphocytic
GIP
Hard metal
BIP
Idiopathic bronchiolitis obliterans organizing pneumonia = Cryptogenic (organizing) pneumonia
BOOP
Nonclassifiable
NSIP
pneumonia
interstitial
2000
pneumonia
lung disease
(nonspecific)
UP excluded
interstitial
COP orOP
Idiopath'ic Interstitial Pneumonia
Table 9.35
Features of idiopathic
interstitial
pneumonia
UIP/1PF
.p1P111:\
Age
40-70
Therapylprognosis Mean survival
(modified
(liP), Collagen Diseases
from Webb, Muller, Naidich
2001)
RB-ILD
AlP
NS1P
>40
>30
>60
around 50
poor
good
good
poor
good
5 years
n.a.
n.a.
1-2 months
n.a.
HRCT Ground-glass
+++
centrilobular
Consolidation
+
+++
++
Reticular lines
+++
++ late
++ late
Honeycombing
+++
+ (late)
+
n.a.
•
+++
diffuse
=
not available
CT Morphology
Typical HRCT features 0 fU I P are fine or irregular intralobular linear opacities (reticular pattern) often associated with traction bronchiectasis and bronchiolectasis, irregular interfaces and honeycombing consisting of thick walled cystic air spaces (2 to 20mm in diameter) predominantly in the basal and subpleural areas. Honeycomb cysts usually enlarge slowly over time. The hallmark of alveolitis is ground-glass opacities, which may be the dominant morphologic finding in patients with an acute exacerbation of the disease (Fig.9.49a). Ground-glass attenuation is common but usually less extensive than the reticular abnormality. Fibrotic change
a
is manifested by thickening of the intralobular interstitium (fine reticular structure), of the interlobular septa (coarse polygonal lines), and of the peribronchovascular interstitium (irregular boundaries, bronchial wall thickening) (Fig.9.49b). Honeycombing, characterized by numerous thick-walled cysts 5-20 mm in diameter and foci of traction bronchiectasis, is a late change that signifies advanced irreversible parenchymal destruction. Typical sites of occurrence of IPF are the peripheral subpleural lung and basal lung zones. Architectural distortion is present reflecting lung fibrosis with lobar volume loss. A patchy distribution is apparent in most cases with areas
b
Fig. 9.49 Various degrees of pulmonary fibrosis. (a) Acute alveolitis (here: AlP) indicated by patchy groundglass opacification even in prone position, (b) Groundglass opacification associated with traction bronchiectasis is indicative of fibrosis rather than active alveolitis.
(c) Initial fibrosis (UIP) with a subpleural band can be differentiated from hydrostatic changes by scanning in supine position, (d) Advanced fibrosis with honeycombing and traction brochiectases.
9
Lungs and Tracheobronchial System
of reticular pattern intermingled with areas of normal lung parenchyma. The predominantly subpleural distribution of the reticular pattern and the honeycombing is evident in 80-95% of patients and represents the most characteristic feature of IPF on HRCT; lack of this feature should suggest an alternative diagnosis. Mediastinal lymph node enlargement is evident in 70-90% of patients (10-15 mm in the short axis diameter, mostly right paratracheal nodes) and less severe in patients under steroid therapy. The significance of ground-glass opacification in patients with IPF has been the focus of considerable interest and controversy. Groundglass opacities, if a predominant feature (occurring in a minority of patients at about 10%) may be completely or partially reversible under steroid therapy while ground-glass opacities as part of a mixed pattern with reticular opacifications were found to precede and predict development of lung fibrosis (Fig.9.49c). Ground-glass attenuation in patients with histologically proven DIP was likely to improve under corticosteroids, while patients with DIP and ground-glass opacities (and those patients usually show a mixed pattern) showed a progress under steroids. The morphologic changes described for idiopathic pulmonary fibrosis are also seen in rheumatoid pulmonary fibrosis, systemic lupus erythematosus, scleroderma (Fig. 9.49 d) and other collagen diseases, and in histiocytosis (see Fig. 9.65 d). They are morphologically indistinguishable on CT scans. Although the rate of progression is slower in patients with collagen vascular disease.
Desquamative Interstitial Pneumonia (DIP)
Desquamative interstitial pneumonia (DIP) and respiratory bronchiolitis-associated interstitial disease (RB-ILD, see p. 299) have recently been grouped together as part of a Single disease spectrum that is strongly associated with cigarette smoking. RB-ILD is thought to represent an exaggerated respiratory bronchiolitis response.
•
CT Morphology
The dominant HRCT features of DIP are bilateral areas of ground-glass attenuation with predominance in the peripheral and lower lung zones. Irregular linear opacities (reticular pattern), which are seen in 50% of patients, are much more limited in extent compared to those in patients with LJIP. Only very mild honeycombing is seen in about 30% of patients. The differential diagnosis from RB-ILD and DIP includes hypersensitivity pneumonitis, sarcoidosis, NSf? and infections such as ?neumocystis calinii pneumonia. Centrilobular nodules represent the key for differential diagnosis since they are very uncommon in DIP and hypersensitivity pneumonitis.
Nonspecific Interstitial Pneumonia (NSIP) Nonspecific interstitial pneumonia (NSIP) resembles IPF but appears to be associated with a Significantly different course and outcome. While IPF is a recurrent and progressing process resulting in a mixture of signs of active inflammation and long-standing chronic fibrous changes, NSIP shows a more uniform histological picture. The parenchyma! changes appear to have occurred over a Single relatively narrow time span. Originally NSIP was not differentiated IPF (approximately 5-15% of patients with IPF turn out to have NSIP). However, clinical symptoms in NSIP are less severe and disease progression is considerably slower and results in a better prognosis than IPF. NSIP is divided into three histological subtypes with varying clinical outcome. Type 1 is predominantely inflammatory, type II is mixed inflammatory and fibrotic, while type III is predominantely fibrotic. CT Morphology
Typical HRCT features of NSIP are patchy areas of ground-glass opacification with intervening areas of unaffected lung, seen in about 80% of patients (Fig. 9.50). They involve mainly the middle and lower lung zones with a predominance in the subpleural areas. They are the sole abnormality in one-third of cases. A reticular pattern superimposed on the areas of ground-
Idiopathic Interstitial Pneumonia (liP), Collagen Diseases
Fig. 9.50 Various degrees of pulmonary fibrosis in NSIP. (a) Subtle subpleural ground glass opacification with cystic changes (arrow) and traction bronchiectases (arrowhead) in scleroderma. (b) Mixed pattern of ground glass, traction bronchiectases, and a faint reticular pattern.
glass opacification is seen in about 50%. Consolidations, seen in about one-third of patients, are generally bilateral, symmetric and subpleural. The differential diagnosis of NSIP depends on the CT patterns it exhibits. The broad variety of disease entities that have to be enconsidered such as UIP, hypersensitivity pneumonitis, or BOOP mirror the variable CT pattern of NSIP.
attenuation correlates with disease duration. Consolidations are also seen but less frequently than ground-glass attenuations. Ground-glass opacification is seen in all phases of AlP and its meaning depends on the stage: areas of increased attenuation without traction bronchiectasis are associated with the exudative or early proliferative phase ofAlP (reflecting the presence of alveolar edema and hyaline membranes), while areas of opacity with traction bronchiectasis are associated with the chronic fibrotic phase. The combination of ground-glass attenuation, air space consolidation, traction bronchiectasis, and architectural distortion is seen in the majority of patients with AlP. The differential diagnosis of AlP depends on the stage and includes hydrostatic edema, hemorrhage, alveolar proteinosis, bronchoalveolar carcinoma, DIP, and diffuse infectious infiltrations. Patients with an accelerated form of IFF also show multifocal areas of peripheral air space consolidations (mimicking BOOP, DIP, or eosinophilic pneumonia) .
Asbestosis
Acute interstitial pneumonia is a fulminant condition of unknown cause that occurs in previously healthy persons and produces histologic findings of organizing diffuse alveolar damage (DAD). The radiologic, clinical, and also histologic features of AlP are similar to ARDS (suggesting the term idiopathic ARDS). The prognosis is very poor.
Asbestos exposure can induce fibrotic changes in the lung. A causal relationship has been established between asbestosis and bronchial carcinoma. Smoking is a highly significant cofactor, as asbestos-exposed smokers have a 50 times higher risk of developing lung cancer than the normal population, whereas asbestosis or smoking alone is associated with a 5- to lO-fold increase in risk. Asbestos-related parenchymal changes can often be appreciated on high-resolution CT scans (HRCT) even when chest radiographs show no apparent parenchymal changes (an important consideration in workers' compensation evaluations). Scanning should be performed in the supine and prone positions.
•
•
Acute Interstitial Pneumonia (AlP)
CT Morphology
Typical HRCT features of AlP are extensive bilateral ground-glass attenuation in diffuse or patchy distribution with focal areas of spared almost normal lung parenchyma resulting in a geographic pattern. The extent of ground-glass
CT Morphology
Asbestos-related parenchymal changes predominantly involve the posterior portions of the lower lobes. Early changes consist of subpleural lines, punctate centrilobular densities located near the pleura (peri bronchiolar fibrosis), and thickening
347
9 Lungs and Tracheobronchial System
of the interlobular septa (Fig.9.49d). More advanced cases present with honeycombing, parenchyma! bands, an irregular interface between the peribronchovascular structures and the lung parenchyma (interface sign), and areas of traction bronchiectasis or round atelectasis. Pleural plaques in asbestosis are described in Chapter 10. It should be noted that the pulmonary findings are nonspecific in themselves and are referable to asbestos exposure only when interpreted in conjunction with pleural plaque formation and a known history of asbestos exposure. Bronchial carcinoma in asbestosis does not show a particular histology or site of predilection. A tumor is considered to be asbestos-related only when typical accompanying pleural changes are identified.
pecially their pattern of distri bution (Table 9.36; Fig. 9.51). The main diseases to be considered in the differential diagnosis of nodular lung changes are miliary tuberculosis, fungal infections, disseminated metastasis, carcinomatous lymphangitis, sarcoidosis, histiocytosis, and silicosis (Table 9.37). Concomitant fibrotic destruction of pulmonary architecture is seen in sarcoidosis, histiocytosis, and silicosis. The nodular lesions of histiocytosis become less prominent than fibrotic and cystic changes as the disease progresses. Pulmonary Lymphangitis Cardnomatosa (PLC)
Carcinomatous lymphangitis (pulmonary lymphangitis carcinomatosa, PLC) results from the permeation of pulmonary lymph vessels by neo-
Diffuse Nodular Lung Changes
Nodular densities in the lung can be characterized by their size, density, margination, and es-
Table 9.36 Distribution pattern of small nodular densities (see Fig. 9.51) 1 Prominence of centrilobular core structures!A!
la--------------~
~
,------------
..1.. .. t..
~1
•
Affects axial interstitium (bronchioles) a Centrilobuar: punctate, linear, or branched -> Prominence of arteriole -> Visualization of bronchiole b Branched pattern with small bulbous terminations ("tree-in-bud" pattern) -> Dilated mucus-filled bronchioles or peri bronchiolar fibrosis 12 Interstitial (perilymphatic)
nodules
• 1-2 mm in diameter • Usually well-circumscribed • Perilymphatic distribution: peribronchovascular, periseptal, and subpleural; also interlobar, interlobular, and centrilobular 3 Disseminated (perivascular) nodules
2
Fig. 9.51 Schematic representation of nodular and reticulonodular lung changes (see Tables 9.36, 9.37).
• Several mm in diameter • Usually well-circumscribed • Disseminated distribution with no sites of predilection (miliary pattern; perivascular, periseptal, and subpleural but not predominantly interstitial)
• • • •
Several mm to 1 cm in diameter Margins usually ill-defined Ground-glass opacity to soft-tissue density Disseminated distribution, but never subpleural or peri pleural
D i If use Nodular Lung Changes Table 9.37
Differential
diagnosis
Prominent centrilobular Prominent interstitium • • • •
of nodular densities
core structures
Carcinomatous lymphangitis Pulmonary edema Pulmonary fibrosis Asbestosis (early form)
Tree-in-bud pattern
I
• Bronchiolitis, smoker's bronchiolitis • Endobronchial tuberculosis, atypical rioses • Cystic fibrosis, bronchiectasis • Bronchopneumonia
mycobacte-
Centrilobular (acinar) nodules Histiocytosis Endobronchial rioses
tuberculosis,
atypical mycobacte-
BOOP Extrinsic allergic alveolitis Smoker's bronchiolitis Asbestosis Pulmonary edema Alveolar cell carcinoma Vasculitis Bronchopneumonia Viral infection, Pneumocystis Bronchiolitis obliterans (rare) Interstitial (perilymphatic) • Lymphangitis (mm) • Silicosis, pneumoconiosis • Sarcoidosis (2-10 mm)
infection
nodules
•
CT Morphology
HRCT features of PLC are smooth or beaded thickening of the central peribronchovascular interstitium and of the interlobular septa producing a characteristic reticular pattern peripheral arcades) and thickened septa outlining distinct pulmonary lobules (polygonal arcades). Thickening of the intralobular axial interstitium results in prominent vascular and bronchiolar structures (centrilobular core structures). Polygonal septal thickening in combination with prominent centrilobular core structures is one of the most distinctive features of PLC (Fig. 9.52). The centrilobular interstitial thickening predominates in only few patients. In approximately 50% of patients the abnormalities appear focal, unilateral, or asymmetric rather than diffuse. Axial, peripheral subpleural and central perihilar bronchovascular thickening may occur together, or one of these features predominates or occurs alone. Associated findings are intrapulmonary nodules, hilar or mediastinallymphadenopathy (38-54%), or pleural effusion. A preservation of normal lung architecture at the lobular level despite the presence of reticulonodular and linear opacities represents an important feature for differential diagnosis.
(1-5 mm)
Diffuse (perivascular) nodules Hematogenous metastases Miliary tuberculosis Fungal infection Silicosis, pneumoconiosis Histiocytosis (rare)
plastic cells. Tumors that spread in this way include carcinomas of the breast, lung, stomach, pancreas, cervix, prostate, thyroid, and metastases of adenocarcinomas of unknown primary site. PLC usually results from metastatic spread to the lung with subsequent interstitial and lymphatic invasion, but can also occur because of direct lymphatic spread of the tumor or secondary to distension of the vascular and lymphatic channels distal to tumor emboli or tumor obstruction.
Fig. 9.52 Carcinomatous lymphangitis, (a) Irregular beaded thickening of the interlobular interstitium (arrowhead) with accentuation of centrilobular core structures (arrow). (b) Thickened interlobular septa with marked thickening of the bronchovascular bundles (arrow) and pleural effusion.
350
9 Lungs and Tracheobronchial System clues for differentiation between PLC and interstitial edema, in that both are
Diagnostic
characterized by perihilar, peribronchovascular, and reticular septal thickening, are the following: • In PLC, the thickened interstitium is sharply marginated from the adjacent aerated lung, there is no filling-in of the alveoli, which remain well aerated. Interlobular thickening due to PLC is more nonuniform as compared to interstitial edema: different degrees of thickening may occur within one septum as well as alterations between smooth or beaded contours. PLC may occur in a focal and asymmetric distribution. The pulmonary arterial branches adjacent to the bronchi also appear larger than normal, that way maintaining the size relationship of the thick-walled bronchi and adjacent vessels. Interstitial edema is most frequently bilateral and symmetric. The interstitial edema is dependently distributed, various degrees of alveolar opacification represent air space edema. The left ventricle and atrium may be enlarged, and pleural effusions are frequently seen. Administration of diuretics will result in a decrease of interstitial edema within hours. clues for differentiation of PLCfrom sarcoid, pneumoconiosis and pulmonary fibrosis, which may all show nodular or beaded
Diagnostic
peribronchovascular
thickening,
are:
• Septal thickening is less extensive in sarcoidosis and pneumoconiosis than in patients with PLC, and reticular opacities are not the predominant feature. In sarcoidosis, pneumoconiosis, and pulmonary fibrosis cicatricial distortion of the lung architecture and secondary lobule anatomy is common, especially when septal thickening is present. In PLC however, lung architecture remains normal and the lobules preserve their size and shape.
Sarcoidosis Sarcoidosis is a systemic, non caseating epithelioid-cell granulomatosis that can affect all organs and is most commonly manifested in the
lung (90%). The granulomas typically show a perilymphatic distribution in the peribronchovascular (perihilar and centrilobular), interlobular, and subpleural interstitium. The pulmonary changes are accompanied in early (stages 1 and 11) by bilateral hilar and mediastinallymphadenopathy. CT Morphology
The CT appearance of pulmonary sarcoidosis varies greatly and is known to masquerade many other diffuse infiltrative lung diseases. The commonest pattern consists of small (2-10 mm): well-defined nodules in a characteristic "perilymphatic distribution" in relation to the subpleural surface, adjacent to the major fissures, along thickened interlobular septa and adjacent to vessels in the lobular core. As a result, pulmonary vessels may be irregularly enlarged. The nodules may be evenly distributed throughout both lungs though predominantly in the upper and middle lung zones; however, in many cases (50%) they are clustered in the perihilar and peribronchovascular region with relative sparing of the lung periphery, or they may be grouped in small areas uni- or bilaterally. A typical finding is irregular polynodular thickening of the bronchovascular bundle and interlobular septa (Fig. 9.53). Pseudoplaques formed by the coalescence of multiple subpleural granulomas are a characteristic feature. Nodular densities measuring between 1 and 4cm in diameter are seen in 15-25% of patients (so called nodular or nummular sarcoidosis). Confluence of granulomas results in large, mostly ill-defined, opacities or consolidations that are often located centrally around bronchovascular structures. There are patchy areas of ground-glass opacities, which may be superimposed over interstitial nodules or signs of fibrosis. They are rarely seen on the radiograph (0.6%) but are commonly present on HRCT (20-60%) mostly in association with small nodules. Pathological studies have shown that they represent interstitial granulomatous inflammation and occasionally microscopic foci of parenchyma! fibrosis. Approximately 20% of patients develop a pulmonary fibrosis with septal thickening, traction bronchiectasis and honeycombing (Fig. 9.53). Irreversible fibrosis is commonest in stage 3 (diffuse pulmonary disease un associated
Diffuse Nodular Lung Changes
a
b
c
d
Fig. 9.53 Sarcoidosis. (a) Very small diffuse nodular densities that partially coalesce to ground-glass opacification. (b) Grouped nodular opacities partially coalescent to patchy consolidations (arrow) with tiny subpleural nodules ("pseudoplaques", arrowheads), (c) Irregular beaded thickening of the central bronchovascular bundle with nodules in interstitial distribution. (d) Perihilar fibrotic mass with parenchymal distortion and interstitial nodules, (e) Coronal 5mm wide MIP demonstrating the apical predominance of nodular densities (4 x 1/6).
with lymph node enlargement). Conglomerate masses mostly in a perihilar location represent areas of fibrosis causing characteristic traction bronchiectasis. The posterior displacement of the upper lobe bronchi (and later also of the main bronchi) is considered to be an early sign oflung distortion in sarcoidosis, indicating loss of volume in the posterior segments of the upperlobes. Differential diagnosis: conditions that most closely mimic the HRCT appearance of sarcoidosis are pulmonary lymphangitis carcinomatosa (PLC), silicosis. and coal worker's pneumoconiosis (CWP). All of these show small perilymphatic nodules. However, differences of the predominant distribution and the combination with signs of fibrosis represent the diagnostic key factors.
In sarcoidosis, the nodules are predominantly located along the central bronchovascular bundle and in the subpleural area; in PLC nodules are mostly located septally and along the central bronchovascular bundle. Septal thickening in sarcoidosis is a much less dominant feature as compared to PLC and if present, it is usually combined with findings offibrosis and lung distortion, precluding the diagnosis of PLC. There are, however, descriptions of cases in which the parenchymal involvement of sarcoidosis and PLC are quite similar and indistinguishable. In silicosis and CWP, nodules are more evenly distributed throughout the whole lung with a predominantly centrilobular and subpleural location, a finding that is much less typical for sarcoidosis. Differen-
351
Lungs and Tracheobronchial System tiation of conglomerate masses of fibrosis in sarcoiclosis from those in stltcosis can be made by the more common presence of air bronchograms in sarcoidosis. As fibrosis develops over time, sarcoidosis and IPF increasingly share a number of findings. Both show irregular or nodular septal thickening, irregular interfaces and traction bronchiectasis. In sarcoidosis, honeycombing is a less frequent feature of irreversible fibrosis as compared to IPF. However, there appear to be two types of fibrotic progression in sarcoidosis. Most patients develop fibrotic changes with loss of volume, fibrotic bands and cysts predominately in the perihilar region and the upper lobes, that way suggesting the diagnosis of sarcoidosis. There are, however, descriptions of a few cases that developed a diffuse fibrosis with predominant basal and subpleural honeycombing very similar in appearance to UlP. HRCT is the method of choice for showing early it is important to know that HRCT cannot be used to rule out involvement of the lung parenchyma. Several authors report that intrapulmonary granulomas could be histologically shown in patients with normal HRCT appearance of the lung parenchyma. A lot has been written about the meaning of ground-glass opacities in patients with sarcoid. One study indicated that areas of ground-glass opacities are associated with disease activity as assessed by 'Ca-scintigraphy. Correlation with the pathologic specimen, however, could prove that ground-glass opacities reflect the presence of confluence of extensive interstitial sarcoid granulomas rather than active alveolitis. Ground-glass, nodular and irregular linear opacities and interlobular septal thickening were described as representing potentially reversible disease, while cystic air spaces and architectural distortion are irreversible findings. The extent of consolidations and nodular densities showed higher correlations with other indicators of disease activity (serum-angiotensin converting enzyme assay, Ga-scanning and bronchoalveolar lavage) than reticular opacities and ground-glass attenuation. Signs of fibrosis (such as lung distortion and reticular densities),
fibrosis in sarcoidosis. However
however, correlate better with impairment of lung function (obstruction) than extent and distribution of nodules and consolidations. Silicosis and Coal Worker's Pneumoconiosis (CWP) The WHO defines pneumoconioses as irreversible lung changes caused by the inhalation of mineral dust and its deposition in the lungs. Silicosis and CWP are pathologically distinct entities with differing histology, resulting from the inhalation of different inorganic dusts (dust containing crystallized silicon dioxide and coal dust, respectively). However, the radiographic and HRCT appearances are similar and cannot be reliably distinguished. The primary imaging modality is plain chest radiography, but HRCT is more sensitive in the detection of subtle changes. Pneumoconiosis may take an uncomplicated nodular course or a complicated course that is associated with progressive massive fibrosis (PMF). The complicated forms present clinically with severe impairment of pulmonary function and frequently with cor pulmonale. Silicoproteinosis occurs after brief, intense exposure to fine silica dust and is characterized by fulminant respiratory impairment. Silicosis can also predispose to the reactivation of tuberculosis, which may take an accelerated, fulminating course (silicotu berculosis). CT Morphology The nodular form is characterized by multiple 1 to 5 mm nodules, usually with sharp margins, that are scattered throughout the lungs (Fig. 9.54) but which show a predilection for the upper and middle zones especially posteriorly. Small nodules may coalesce to form macronodules (8-20 mm in diameter, with subpleural lesions forming pseudoplaques). In contrast to sarcoidosis, the nodules are typically uniformly distributed in centrilobular (peribronchiolar) and subpleural locations and rarely clustered or distributed along central bronchovascular structures. Calcification may occur within the nodules, and hilar lymph nodes may develop typical eggshell-like calcifications. Sometimes the centrilobular densities show tiny short branches instead of being round, corresponding to irregular fibrosis around the respiratory
Diffuse Nodular Lung Changes
haled stimulating an alveolar and subsequently extensive fibrous tissue (silicoproteinosis). olar ground-glass opacifications crete silica nodules, resembling alveolar proteinosis.
a
b
Fig. 9.54 Silicosis. (a) Diffuse rnicronodular densities (type q' in ILO classification), (b) Conglomerate mass (PMF) with parenchymal distortion, perifocal emphysema, flbrosls. reactive pleural thickening, and diffusely scattered nodules.
bronchioles surrounded by a small zone of focal emphysema. As simple silicosis and CWP progress, the number of nodules increase and increasingly coalesce resulting in distortion of the adjacent lung structure. Coalescence of small nodules to ~arger nodules (> 1 em) and eventually to large, Irregularly defined conglomerate masses refl,e~ts t?e transition from simple to complicated silicosis/pneumoconiosis (Fig. 9.54). The masses of progressive massive fibrosis (PMF) tend to develop in the mid portion or periphery of the upper lung zones and migrate towards the hila. They may cavitate due to ischemic necrosis although this seems to be more common with CWP. Calcifications may befound. Regardless of age and smoking habits, most pati,ent,s with silicosis or coal worker's pneumocomOSIS develop focal areas of centrilobular emphysema. Unlike classic centrilobular emphysema, these areas contain central micronodules and their extent correlates with the impairment of pulmonary function. The extent of emphysema correlates more closely with functional imp,airment than does the degree of nodular profusion. Functional impairment and p:~gn~sis are usually poorer for patients with sIlIcoSIS than for CWP. Accompanying paracicatri~ial emp?ys~ma reflect the process of progressive massive fibrosis, An acute form of silicosis has been described in which a large dose of silica was in-
lipoproteinosis production of CT shows alveinstead of disthe image of
The differential diagnosis of silicosis/CWP includes all diseases that may develop numerous small well-defined nodular opacities such as sarc~idosis, pulmonaly lymphangitic carcinomatosis, Langerhans cell histiocytosis but also infectious diseases such as miliary tuberculosis, fungus infection, and liematogenous metastases. • Dependent on the profusion of the nodules in a seemingly random distributi~n, it may be impossible to distinguish miliary tuberculosis (perivascular nodules) from silicosis (peri bronchiolar nodules). Also the differentiation between sarcoidosis and silicosis can be difficult when the sarcoid nodules are numerous a~d ~o n,ot show the typical perilymphatic distribution but a diffuse centrilobular location. The differentiation between silicosis and Langerhans cell histiocytosis is difficult if the latter is characterized solely by t~e presence of nodules while cystic leSIOns are completely missing. • Si,licosis/CWP and PLC can usually be differentiated quite easily by the different distribution of nodules. In silicosis and CW\, the nodules appear bilaterally, symmetnc and more uniformly distributed. Beaded septa or reticular densities are usually absent. • Perihilar mass-like fibrosis or upper lobe consolidations in combination with p~~en~hymal distortion may occur in both slhcoSlS/CWP and end-stage sarcoidosis. Both entities are characterized by an upp~~ lo~e predominance and by potential calcifications of the nodules. Differentiation of conglomerate masses of fibrosis in sarcoidosis from those in silicosis can be ~ade b~ the, presence of air bronchograms In sarcoidosis and by associated findings such as the distribution of nodules. • Conglomerate masses in PMF are occasionally indistinguishable from tuberculous cavities or bronchial carcinoma.
354
9
Lungs and Tracheobronchial
System
Diffuse Lung Changes with Increased Parenchymal Density
Parenchyma! changes that cause increased radiographic density are evaluated according to the degree of the attenuation increase and the shape and extent of the area of increased attenuation (Tables 9.38, 9.39; Fig. 9.55). The areas may be described as nodular, patchy, geographic, or lobular.
Table 9.38 Differential diagnosis enchymal density in the lung Ground-glass
of
increased
3
par-
opacity
Pulmonary edema Pulmonary fibrosis (acute) Sarcoidosis Extrinsic allergic alveolitis Alveolar proteinosis Lipoid pneumonia Pneumocystis or viral pneumonia Eosinophilic pneumonia
BOOP Smoker's bronchiolitis Hemorrhage Vasculitis Radiation pneumonitis Alveolarcell carcinoma Lymphoma Alveolar opacities (airspace shadowing) Infection (bacterial, atypical, fungal) Alveolar cell carcinoma Sarcoidosis Wegener's disease Eosinophilic pneumonia Lymphoma Pulmonary edema ARDS Alveolar proteinosis Radiation pneumonitis
Fig. 9.55 Patterns of increased (see Tables 9.38, 9.39).
Table 9.39
1 Ground-glass
Hemorrhage Contusion masses
Silicosis, pneumoconiosis Sarcoidosis (end stage)
BOOP
Carnifying pneumonia Wegener's disease
density
opacity
2 Alveolar opacities (PMF)
parenchymal
density
Patchy increase in parenchymal density Vascular structures not obscured Air bronchogram may be present (dark bronchus sign) Thickened alveolar walls, intraalveolar material Occurs with interstitial and alveolarprocesses-> active, treatable, potentially reversible process
BOOP
Conglomerate
Increased
parenchymal
• • • •
(airspace shadowing)
Profuse focal or patchy opacity Obliteration of vascular structures Positive air bronchogram Alveolar airspaces filled with cells or fluid
3 Conglomerate
masses
• Bunched, confluent strands • Traction bronchiectasis • Volume loss
of fibrosis
Extrinsic Allergic Alveolitis,
Extrinsic Allergic Alveolitis, Chronic Allergic Pneumonitis
Extrinsic allergic alveolitis (EAA) or hypersensitivity pneumonitis is an allergic response to inhaled organic antigens (farmer's lung, birdbreeder's lung). Known causal antigens include microorganisms (fungi, bacteria, actinomycetes) and animal and plant proteins. The radiographic and pathologic abnormalities are quite similar, independent of the causing antigen The clinical and CT features ofEAA vary with the duration of the exposure. The changes are classified as acute, subacute, or chronic, and combinations of these forms can be found in individuals who are repeatedly exposed. Mostly HRCT is performed in the subacute and chronic stage. Chronic changes are also referred to as chronic allergic pneumonitis. A careful clinical history, typical HRCT findings and concordant serologic findings confirm the diagnosis of EAA thus precluding the need for lung biopsy. In patients with discrepant or atypical HRCT findings diagnosis is mostly based on trans bronchial biopsy and abnormal Tlymphocytes in the bronchoalveolar lavage. •
CT Morphology
The acute form is generally diagnosed clinically, so little information is available on CT morphology.
Chronic Allergic Pneumonitis
In the subacute stage, HRCT typically (in 40-70%) shows small (2-3 mm, maximum 5 mm) centrilo bular acinar densities, usuall y ill- defined and showing low attenuation (Fig.9.56a). The acinar densities may coalesce to form diffuse ground-glass opacities (Fig. 9.56b). All lung zones are affected, but there is a slight middle and lower zone predominance that often spares the subpleural space (unlike the alveolitis in idiopathic pulmonary fibrosis). The changes tend to regress in 7-10 days after exposure to the antigen is discontinued. Bronchial walls may be slightly prominent. Recurrent EAA episodes and the chronic stage of allergic pneumonitis are characterized by signs of irreversible interstitial fibrosis and parenchyma! distortion (irregular reticular opacities, intralobular interstitial and interlobular septal thickening, visible intralobular bronchioles, traction bronchiectasis, and honeycombing) (Fig.9.56c). Frequently there is a mixture of chronic (fibrotic) and subacute (inflammatory) changes in patchy distribution and combined with signs of air trapping. The distribution of the fibrosis may be variable, sometimes predominantly subpleural, in others patchy or peribronchovascular. Honeycombing is usually localized subpleurally. Divergent findings are reported about the lobar predominance of chronic fibrotic changes. While one study reported a lower lobe predomi-
r
Fig.9.56 Extrinsic allergic alveolitis. (a) Diffuse, iii-defined acinar (alveolar) opacities in a centrilobular distribution. (b) The alveolar opacities may coalesce to form ground-glass opacities. Differentiation is required from viral pneumonia, acute Pneumocystis pneumonia, and acute exacerbation of an early form of interstitial fi-
brosis or sarcoidosis. (c) Chronic allergic pneumonitis with an mixed pattern of acute alveolitis (ground-glass opacities) and chronic fibrosis (thickened septa, thickened interstitium, bronchiectasis, subpleural opacities) associated with regional air trapping indicating a process involving the airways.
356
9
Lungs and Tracheobronchial System
nance in about one-third of patients, others described a mid zone predominance or even distribution for the majority of patients. In contrast to idiopathic pulmonary fibrosis, however, there is no subpleural or lower zone predominance. In rare cases chronic EAA may be indistinguishable from a late form of sarcoidosis. • In patients with bilateral nodular, patchy or diffuse ground-glass opacities, the differential diagnosis includes desquamative interstitial pneumonia (DIP) and pulmonary alveolar proteinosis (PAP). DIP, however, is very rare and usually shows a subpleural predominance of the groundglass opacification and is not associated with centrilobular nodules. • Faint acinar nodules, which may coalesce to a diffuse ground-glass opacity, are seen in various disorders including allergic alveolitis, viral infections, and vasculitis. Classic alveolar proteinosis shows a crazy paving appearance and is readily distinguished by bronchoalveolar lavage. The latter is also suited to rule out infectious diseases such as CMV or Pneumocystis cetinii as etiologies for diffuse groundglass opacities. The differential between an alveolar sarcoidosis and subacute EAA requires trans bronchial biopsy. • Patients with chronic EAA and UIP may show identical HRCT findings, and differential diagnosis may be possible only by the clinical history and laboratory findings. Only a predominant location of fibrotic changes in the upper and/or middle lung zones allow for a distinction between these entities.
tings (in children: lymphopenia, thymic aplasia, immunoglobulin deficiency; in adults: lymphoma and leukemia) and infectious diseases (CMV, mycobacteria, Pneumocystis, Histoplasma, Candida species, etc). Pulmonary proteinosis is very rare. In addition to the suggestive HRCT pattern, diagnosis is usually based on characteristic features of BAL fluid. •
CT Morphology
The classic CT finding is a reticulated parenchymal pattern formed by the juxtaposition of normal areas and pathologic areas of increased density (crazy paving). Opacifications range from ground-glass to consolidation, and may be patchy or geographic with sharp demarcation from the surrounding normal parenchyma. While some borders follow anatomic structures of lobar or lobular septa, others are independent of anatomic boundaries. The smoothly thickened inter- and intralobular septa often create a very fine polygonal pattern that is usually seen only in areas with groundglass opacification (Fig. 9.57). Resolution of air space consolidation can be monitored radiographically or by HRCT. The overall extent of disease and the degree of Opacifications correlate well with the impairment of pulmonary function and severity of hypoxemia.
Pulmonary Alveolar Proteinosis (PAP)
Pulmonary alveolar proteinosis is characterized pathologically by filling of the alveoli with a PAS-positive proteinaceous material rich in lipid. It is caused by a dysfunction of pneumocytes that desquamate in the alveolar spaces secondary to a defect of lipid metabolism at the intracellular level. An abnormality in surfactant production, metabolism, or clearance also has been strongly implicated, as well as an association with a variety of immunocompromised set-
a
b
Fig. 9.57 Pulmonary alveolar proteinosis. (a) Acute: ground glass opacification superimposed by smooth intralobular thickening (crazy paving) in a geographic sharply demarcated distribution, (b) Chronic: after treatment, irregularly thickened intralobular septa may persist.
Lipoid Pneumonia
Although the crazy paving appearance is highly suggestive for alveolar proteinosis with a 100% prevalence of this pattern, it is not specific. A recent publication listed 14 airspace and interstitial lung diseases that showed crazy paving with varying but mostly much lower prevalence of crazy paving. The main differential diagnoses include diffuse alveolar damage superimposed on UIP (prevalence of crazy paving in 67%), acute interstitial pneumonia (31 %) and ARDS (21 %). Other diseases with lower prevalence of crazy paving were drug induced pneumonitis (12%), pneumonias (bacterial 6%, tuberculosis 1%, mycoplasma 6%, Pneumocystis cetinii 7%), BOOP (8%), chronic eosinophilic pneumonia (8 %) , radiation pneumonitis ( 4 %) , and cardiogenic pulmonary edema (14 %) . The wide differential diagnosis is understandable considering that any kind of fluid or cellular filling of the air spaces and the interstitial spaces causes crazy paving. In alveolar proteinosis, the crazy paving appearance was found to be due to an accumulation of PAS-positive material in the air spaces adjacent to the interlobular septa rather than to thickening of the septa themselves. Most disease entities with similar HRCT features can be distinguished from alveolar proteinosis by clinical symptoms and associated HRCT findings. It has to be noted that superimposed infection- frequently by Nocardia asteroides-is a common complication of alveolar proteinosis,
and CT is rather limited in differentiating air space consolidation due to infection from underlying disease. Lipoid Pneumonia
Lipoid pneumonia is caused by the aspiration of oily substances (e.g., foods, laxatives, nasal drops) and is commonest in elderly patients with neuromuscular disorders or swallowing difficulties. A large amount of oily material must be aspirated before symptoms develop. It differs from alveolar proteinosis in that the alveoli are filled with a lipid-containing material rather than proteinaceous material. •
CT Morphology
Lipoid pneumonia presents with a homogeneous low attenuation (-35 to -75 HU) consolidation of the affected area (predominately in the middle or lower zones). It may cause alveolar opacities resembling those in alveolar proteinosis. There is also ground-glass opacity with a superimposed reticular interstitial pattern caused by thickened interlobular septa (Fig. 9.58). Fibrosis and inflammation accompany the presence of intra-alveolar lipid material (HU value lower than chest wall musculature but higher than subcutaneous fat). Sometimes cavitation and necrosis are seen. In rare cases, fat -equivalent CT densities are found in the affected bronchi.
Fig.9.58 Lipoid pneumonia. (a) Diffuse changes with dense ground glass opacification and accentuated intralobular septa, (b. c) Focal changes with a tumor-like presentation (b) but fat-attenuation on soft tissue windows (c) in an actor that uses oily nose drops.
a
b
c
357
358
9 Lungs and Tracheobronch'ial System
Eosinophilic Pneumonia
Drug-Induced Lung Disease
Eosinophilic pneumonia is typically associated with blood eosinophilia and responds promptly to steroid therapy (regression within days with complete recovery within months). An acute type (Loffler) is differentiated from the chronic type (disease duration longer than one month). Some 50% of patients have an atopic disposition, 40% have a history of asthma. It may be idiopathic or represent secondary immunologic reaction to drugs, parasites, or other agents.
Pulmonary drug toxicity is increasingly being diagnosed as a cause of acute or chronic lung disease and has to be considered in the differential diagnosis, especially in selected patient groups (e.g., those under chemotherapy or cardiogenic antiarrhythmic therapy). Affected patients typically present with progressive dyspnea, dry cough, and, occasionally, fever. The radiologic findings consist of a broad spectrum of morphology reflecting the underlying histopathologic process. Pulmonary pathology may start to decrease after cessation of drug therapy (e.g., BOOP, eosinophilic pneu-
•
CT Morphology
Dominating features in the acute eosinophilic lung disease (Loffler's syndrome) are groundglass opacities (" 100%) and consolidation (in 92%) in diffuse distribution without a preferred zone (Fig. 9.59). Infiltrates are transient and "migratory". Ill-defined nodules (in 54%), septal thickening (in 70%), and a thickening of the central bronchovascular bundle (in 60%) are other frequent findings. Bronchiectasis and lymph node enlargement are rare. Two-thirds have a pleural effusion. Dominating features in the chronic type are also ground-glass opacities (88%) and consolidations (100%) but now in a predominantly subpleural distribution in the cranial and middle lung areas. Nodules (38%), septal thickening (18 %), thickening of the bronchovascular bundle (38%), and pleural effusions (10%) are rarer than in the acute form.
Table 9.40
Drug-induced lung disease
DAD
• Early: scattered or diffuse ground-glass opacities • Later: fibrosis • Common: bleomycin*, busulphan, carmustine*, cyclophosphamide, melphalan, mitomycin, gold salts Nonspecific interstitial pneumonia (NSIP) • Scattered or diffuse ground-glass opacities, later: basal fibrosis • Common: amiodarone*, chlorambucil, gold salts, methotrexate, nitrofurantoin (chronic) • Less common: carmustine* (Bicnu), cyclophosphamide, bleomycin* Bronchiolitis obliterans with organizing pneumonia (BOOP) • Patchy opacities, tree-in-bud, bronchiectasis • Common: amiodarone, chlorambucil, nitrofurantoin, penicillamine, sulfalazine • Less common: amiodarone*, bleomycin*, cyclophosphamide, gold salts, methotrexate Eosinophitic pneumonia • Reverse edema pattern • Common: anti-inflammatory drugs, nitrofurantoin (acute), para-aminosalicylic acid, penicillamine, sulfalazine Pulmonary hemorrhage • Patchy opacities with air-space nodules, consolidation • Common: amphotericin B, anticoagulants, cyclophosphamide, cytarabine, penicillamine
a
b
Fig. 9.59 Eosinophilic pneumonia with subpleural filtrates of varying density.
in-
• relationship between lung injury and dose and duration of therapy
Emphysema monia). Some patients require the administration of corticosteroids. •
CT Morphology
Pulmonary drug reactions include findings caused by diffuse alveolar damage (DAD), nonspecific interstitial pneumonia (NSZP), obliterative bronchiolitis, bronchiolitis obliterans with organizing pneumonia (BOOP), eosinophilic lung disease, pulmonary hemorrhage, edema, hypertension or veno-occlusive disease (Table 9.40). DAD is divided into an acute exudative phase (first week after injury) and a late reparative or proliferative phase (second week), which is characterized by a varying degree of interstitial fibrosis, architectural distortion, and honeycombing. NSf? in its early phase shows scattered or diffuse ground-glass opacities, followed by findings of fibrosis (traction bronchiectasis, honeycombing) predominantly in the basal lung areas (Fig. 9.60). Patients with BOOP show nodular or patchy consolidations, diffuse tree- in- bud opacities and bronchiectasis.
Fig. 9.60 Drug-induced lung disease, (a) Nonspecific interstitial pneumonitis (NSIP) due to Glivec therapy. (b) Bronchiolitis with organizing pneumonia (BOOP) as a consequence ofAmiodarone therapy.
Eosinophilic pneumonia presents with peripherally located subsegmental consolidations (reverse pulmonary edema pattern), while pulmonary hemorrhage leads to bilateral patchy or diffuse ground-glass opacities or consolidations.
Parenchymal Diseases with Decreased Density
A regional or diffuse decrease in the attenuation of the lung parenchyma can have numerous
potential causes, several of which are distinguished by their typical morphologic features on HRCT (Tables 9.41, 9.42; Fig. 9.61).
Table 9.41 Differential parenchymal density
Emphysema
diagnosis
of
decreased
Emphysema • Chronic obstructive lung disease (COLD), smoker's emphysema • Alpha-1-antitrypsin deficiency • Check-valve mechanism (foreign body, tumor, stenosis) • Scarring or atelectasis -> hyperinflation • Postinfectious emphysema in early childhood (Swyer-James) Airtrapping • Bronchiolitis obliterans (causes see Table 9.8) • Focal: check-valve mechanism Mosaic perfusion • Chronic
pulmonary
embolism
Emphysema is an irreversible dilatation of the peripheral airways distal to the terminal bronchioles accompanied by destruction of the alveolar walls. Emphysema is classified into three types-centrilobular, panlobular, or paraseptal-according to the arrangement and distribution of the emphysematous areas (Fig. 9.62). Clinical impairment of pulmonary function is related to the extent of parenchymal destruction rather than to the type. The pathogenic mechanisms of peripheral airway enlargement may be based on hypoplasia, atrophy, hyperinflation, or destruction. Hypoplasia of the peripheral airways can occur as a sequel to bronchopulmonary diseases in in-
359
9 Lungs and Tracheobronchial System Table 9.42
Decreased parenchyma! density
1 Emphysema a Centrilobular, b panlobular, c paraseptal Air trapping - Hyperlucent areas with air trapping alternating with areas of normal density • Well-defined margins • On expiratory scans, little or no density increase in hyperlucent areas (air trapping) with a normal density increase in the rest of the lung parenchyma
- Hyperlucent areas with narrowed vessels alternating with areas of slightly increased density with expanded vessels • Margins frequently ill-defined • On expiratory scans, uniform density increase in all areas
5-Table 9.43 Cystic structures
in the lung (see Fig. 9.61)
3 Pulmonary cysts -<1 em • <3 mm thick with smooth (epithelialized) • Surrounded by normal lung tissue
4----"
Fig. 9.61 Patterns of decreased parenchymal density and cystic parenchyma! changes (numbering see Tables 9.41-9.43).
fants and children. Alveolar atrophy involves the secondary loss of alveoli in an initially normally developed lung (e.g., in patients with age-related emphysema). Hyperinflation may result from a check-valve mechanism, atelectasis, or scar traction. Emphysema due to destruction of the alveolar walls can occur in the end stage of inflammatory fibrous diseases (sarcoidosis, histiocytosis, etc.) or in patients with congenital alpha-I-antitrypsin deficiency.
4 Pneumatocele • Post-traumatic • Delicate wall
or postinfectious cavity
5 Emphysematous bullae • >1 ern, up to 20cm • Thin border «1 mm), no wall • Always associated with emphysema 6 Honeycombing • Multiple cysts (mm to cm) with thickened walls • End stage of irreversible parenchymal destruction 7 Cavities • Cavity formation within an opacity • Thick wall (mm to cm)
Terminal bronchiole
k---
-
Normal secondary lobule
wall
Respiratory bronchiole
Centrilobular emphysema
Panlobular emphysema
Fig. 9.62 Schematic representation of Centrilobular and panlobular emphy-
Congenital Lobar Emphysema •
A bulla is defined as an emphysematous space larger than 1 cm in diameter. It results from the expansion or confluence of several secondary lobules due to progressive septal destruction. Unlike a pulmonary cyst, the wall of the bulla is not lined with bronchial epitheli um. Paraseptal or subpleural emphysema involves alveoli adjacent to pleural structures and bronchovascular bundles (Fig.9.63a). It is clinically silent, causing no restriction of pulmonary function. Preferred sites are the costophrenic angle, apical lung, and bronchovascular branches. The CT appearance of emphysema can be graded based on a scoring system that determines the area of emphysema relative to the total area of the axial section. HRCT can also be used to detect and classify very early emphysematous changes that are still clinically occult (normal pulmonary function).
CT Morphology
The CT diagnosis of emphysema is based on the detection of well-defined, unwalled areas of decreased density or increased lucency. There is an accompanying reduction in the number and caliber of pulmonary vessels and their branches. With centrilobular (centriacinar) emphysema, the respiratory bronchioles are affected while the lobular periphery remains intact. Small, circumscribed areas of hyperinflation are found scattered within normal parenchymal (Fig. 9.63 a). The changes show an upper zone predilection and are associated with a history of smoking. Decreased vascularity is seen only in advanced stages. With panlobular (panacinar) emphysema, there is additional involvement of the alveolar ducts and alveoli. The emphysematous areas coalesce, normal parenchyma is sparse, and there is a more pronounced decrease in vascularity (Fig. 9.63 b). The previous distribution pattern is reversed, with the lower zones affected more than the apex. A special form of panlobular emphysema, also with basal predominance, occurs in patients with alpha-I-antitrypsin deficiency. In advanced stages of emphysema, the centrilobular and panlobular forms become indistinguishable.
a
b
Fig. 9.63 Forms of emphysema, (a) Centrilobular emphysema with small, circumscribed areas of hyperinflation within an intact lobular periphery (common in smokers). There is accompanying paraseptal emphysema with subpleural bullae. (b) Pan lobular emphysema
Congenital Lobar Emphysema
Lobar emphysema results from the congenital or postinflammatory stenosis of a lobar bronchus. The most commonly affected structures are, in descending order of frequency, the left upper lobe bronchus (50%), the middle lobe bronchus (25%), and the right upper lobe bronchus (20%).
c with destruction of all intralobular alveoli and decreased vascularity (alpha-1-antitrypsin deficiency), (c) Severe emphysematous parenchymal destruction with progressive loss of tissue (4 x 1 /6).
361
9 Lungs and Tracheobronchial
System
CT Morphology
CT Morphology
CT demonstrates creased lucency and mediastinal
a hyperinflated with diminished displacement.
area of invascularity
Swyer-James Syndrome Swyer-James syndrome (MacLeod's syndrome) is not a congenital disorder but results from injury of the immature lung (hypoplastic lung) caused by a persistent bronchiolitis obliterans in children who have recurrent viral infections during the first 8 years of life.
Pulmonary
Diseases with Cystic Changes
Pulmonary diseases that are manifested predominantly by cystic changes include Langerhans cell histiocytosis (eosinophilic granuloma) and lymphangioleiomyomatosis. They require differentiation from cystic changes with a bronchial etiology and from parenchymal diseases leading to solitary or multiple, nondiffuse cystic lung changes (Tables 9.43, 9.44; see Fig.9.G!).
Langerhans Cell Histiocytosis Langerhans cell histiocytosis (also described as eosinophilic granuloma of the lung, Langerhans cell granulomatosis of the lung, or histiocytosis X) accounts for approximately 4 % of all diffuse Table 9.44 Differential structures in the lung Pulmonary
The typical CT finding is a hyperlucent lung with markedly decreased vascularity (= unilaterallucent lung). A bronchial check-valve mechanism and collateral air drift lead to very pronounced unilateral air trapping. It is important to exclude (by C1) an acute central airway obstruction as the cause. Differentiation is required from pulmonary arterial aplasia, chronic pulmonary embolism, lobar emphysema, central obstruction by a tumor or foreign body, toxic inhalational injury, and bronchiolitis obliterans secondary to a graft-versus-host reaction following bone marrow transplantation.
diagnosis
of thin-walled
cystic
cysts
CT Morphology
• Lymphangioieiomyomatosis • Histiocytosis X • Honeycomb pattern (see Table
9.17)
Pneumatocele • Postinfectious after Pneumocystis
pulmonary diseases. It may occur in children as part of a multisystem disorder (Abt-LettererSiwe disease), and the disease in adults shows a peak incidence between 20 and 30 years of age. In its early stages it is characterized by a bronchiolocentric granulomatous reaction with proliferation and infiltration of the bronchiolar wall and the adjacent blood vessels by Langerhans histiocytes and eosinophils. Bronchiolar obliteration causes progressive alveolar wall fibrosis and cyst formation, eventually resulting in increasing fibrosis and parenchymal distortion. The HRCT findings closely mirror this stepwise histological progression. The etiology of histiocytosis is unknown. However, there is a striking relationship to smoking (>90% of patients with histiocytosis are active smokers). Histologic confirmation of the diagnosis is usually required. Thirty percent of patients are found to have coexisting granulomas in the bone andlor liver.
or Staphylococcus
monia • Post-traumatic (laceration) • After successful treatment
pneu-
of metastases
(rare)
CT reveals nodular and cystic lung changes. The nodular lesions are 1-5 mm in diameter (rarely several centimeters), usually have well-defined margins, and show a peribronchiolar centrilobular distribution (Fig.9.G4a). Larger nodules {> 1 Omm) also occur but are less common. Nodules can be few or numerous depending on disease activity. The nodules have irregular contours particularly when they are surrounded by
Langerhans Cell Histiocytosis
a
b
c
d
Fig. 9.64 CT manifestations of Langerhans cell histiocytosis. (a) Diffuse centrilobular nodules and microcystic spaces, (b) Multiple, predominantly small cysts, some confluent, accompanied by isolated subpleural nodules. The intervening parenchyma shows ground-
glass opacification. (c) Coronal MPR (4 x 1/6) demonstrates the upper lobe predominance of large confluent cysts with bizarre shapes, (d) Advanced parenchymal destruction leads to fibrotic changes.
cystic or reticular disease. They may be solid or show small lucent centers, these "cavitations" are thought to represent dilated bronchioles surrounded by granulomas and thickened interstitium. Development of cavitated nodules into cysts has been observed. The cysts are usually smaller than 1Dmm in diameter with walls that range from barely perceptible to being several millimeters thick. The cysts may be round, or have bizarre shapes; they may coalesce and then become larger than 2Dmm (Fig. 9.64b, c). Subpleural cysts predispose to recurrent episodes of pneumothorax. Cysts as well as nodules show an upper lobe predominance (in 57%); the lung bases and the basal pleurocostal angle are relatively spared. The patients are prone to pneumothorax, probably from cyst rupture. There may be rarely patchy or diffuse ground-glass opacification, but it is never a prominent feature. The lung volume is usually increased. With disease progression the presence of nodules decreases while the whole lung parenchyma consists increasingly of thin-walled cysts up to the extent that intact lung parenchyma will be present only in the basal areas (vanishing lung disease). The nodular lesions and cyst-wall thickness may regress in response to therapy.
Most patients show no evidence of fibrosis or septal thickening. Only few show an irregular interface sign or a fine reticular network correlating with intralobular fibrosis, which differs from idiopathic pulmonary fibrosis in the relative sparing of the subpleural lung (Fig. 9.64 d). The differential diagnosis includes lymphangiomyomatosis (LAM), cystic bronchiectasis, idiopathic pulmonary fibrosis (IPF) and lymphocytic interstitial pneumonia (LIP) . The cysts found in LAM may have a similar appearance to those of histiocytosis but are not associated with nodular changes or an upper lobe lung distribution. In addition, the cysts in histiocytosis are less uniform than in LAM. The round shaped cystic air spaces-especially when close to a vessel-may mimic the appearance of cystic bronchiectasis (signet -ring sign), close inspection of serial sections above and below help to quite readily differentiate tubular from spherical structures. Unlike IPF the cysts show no predominant subpleural distribution. The cysts in histiocytosis are usually discrete rather than clustered, and most importantly the lung volumes are usually preserved or even increased rather than decreased as frequently seen with IPF.
363
9 Lungs and Tracheobronchial
System
Both, histiocytosis and LIP, are characterized by centrilobular nodules and cystic air spaces. Distinguishing features of histiocytosis from LIP include the absence of interlobular thickening and lymphadenopathy, both of which were found in 82% and 70%, respectively, in patients with LIP. The combination of pulmonary nodules and cysts is virtually diagnostic for pulmonary histiocytosis X. It was shown that a first choice diagnosis of histiocytosis had a 60% likelihood of being correct when based on the chest radiograph but a 90% likelihood of correctness with HRCT. Although the HRCT findings are quite suggestive, there are no other specific laboratory tests to further underline the diagnosis. Langerhans cells can be identified in specimens obtained by BAL. However, they can also be seen in cases of pulmonary fibrosis of other etiology. A definite diagnosis is only established after examination of lung tissue. While trans bronchial biopsy has a very low yield due to the focal nature of the disease, open or video-assisted thoracoscopic Iung biopsy is recommended. The extent of lung abnormality as determined by CT correlates well with the lung diffusion capacity. Serial HRCT scans could show that regression of the nodules leaves multiple thin walled cysts because of check-valve obstruction by bronchiole-centered granulomas, and these small cysts could only be detected by HRCT and not by radiography.
•
CT Morphology
The CT features are pathognomonic and consist of predominantly small, thin-walled cysts of varying size (2 mm to 5 ern) that are distributed uniformly throughout the lungs (Fig. 9.65). The cysts range from 2 mm to 5 cm in diameter, but rarely can become even larger. Their size tends to increase with disease progression. The cysts are primarily round in shape; only some of them are confluent. In most patients the intervening parenchyma is normal. However, there are also reports of cases showing patchy ground-glass opacities or increased interstitial markings in the pericystic lung parenchyma and even rarely signs oflung distortion. The cysts are distributed diffusely throughout the lungs; no lung zone is spared and upper and lower lobes are involved to a similar degree. The wall thickness of the cysts ranges from rarely perceptible to 4mm in thickness. Very rarely, few nodules may be seen. Chylothorax is present in 60% of cases, and spontaneous pneumothorax develops in 40%. Mediastinal lymphadenopathy is common (50%). Without treatment, the cystic transformation of the lung parenchyma will progress for a period of years, accompanied by a relatively mild fibrotic component. It should be noted that normal-appearing CT lung scans do not exclude LAM.
Pulmonary Lymphangioleiomyomatosis (LAM) and Tuberous Sclerosis
Lymphangiomyomatosis (LAM) and pulmonary involvement in tuberous sclerosis (BournevillePringle disease, in A %) are radiologically and pathologically identical. Pulmonary LAM is a rare disease based on the proliferation of atypical spindle cells in the lung parenchyma and in the walls of lymphatic vessels. Ruptures of pulmonary venules lead to episodes of hemoptysis and pulmonary hemorrhage, obstruction of pulmonary lymphatics by smooth muscles cause chylous pleural effusions. It most commonly affects women of childbearing age. A hormonal cause (estrogen) has been postulated.
a
b
Fig. 9.65 Cystic parenchymal changes, (a) Lymphangioleiomyomatosis (LAM) with complete structural transformation of the lung parenchyma and relatively uniform cysts that are evenly distributed throughout the parenchyma, (b) Lymphoid interstitial pneumonia (LIP) with ground glass pattern and cystic airspaces.
Chronic Thromboembolic Pulmonary Hypertension (CTEPH) 365 The differential diagnosis of diffuse lung diseases with primarily cystic lesions include histiocytosis X, centrilobular emphysema, end-stage interstitial fibrosis and lymphocytic inters titial pneumonia (U P) (Fig, 9,65 b), LAM can be reliably differentiated from histiocytosis by the diffuse lung involvement (in LAM the lung bases or costophrenic suici are not spared), the infrequency of nodules, and the regular shape of the cysts, Additionally LAM is not associated with smoking history, while more than 90% of patients with histiocytosis are smokers, Emphysematous cystic air spaces have no perceptible walls, In LAM, vessels are typically seen at the margins of the cysts rather than at the centers of air spaces as is characteristically seen with emphysema, Differential diagnosis, however, may be difficult especially in early stages of LAM, and in advanced stages of paraseptal emphysema, Complete absence of fibrotic changes or of signs of distortion facilitate the differential from IPF, However, in accordance with reported pathological findings, interstitial changes may be present in LAM more frequently than previously reported, Interstitial changes may be explained by additional factors such as intercurrent relapsing infections, Lymphocytic interstitial pneumonia (LIP, as seen in patients with HIV infection, Sjogren syndrome or Castleman's disease) also shows
Acquired
Disorders
of Pulmonary
cystic air spaces that have perceptible walls, The cysts, however, are less uniform in size and reportedly in a predominant subpleural location, They may also be scattered throughout the whole lung but rarely represent the dominant feature, Other findings usually seen in UP include ill-defined centrilobular nodules (3 mm and 30 mm with most being small) and diffuse ground-glass opacification of the lung parenchyma, They represent the key features for differential diagnosis: the virtual absence of centrilobular nodules in LAM allows for ready distinction, HRCT is significantly superior to chest radiography and conventional CT in determining the extent and distribution of air cysts, It should, however, be noted that normal CT findings do not rule out parenchyma! disease in patients with LAM, TheCT appearance of LAM is diagnostic in the correct clinical context The presence of many thin-walled, round orgeographiccystic air spaces scattered throughout both lungs in a young woman isvirtually pathognomonic, Definitediagnosts, however, would require open lung biopsy, CT correlates better than radiography with clinical and functional impairment The extent of cystic disease was found to correlate well with impairment of gas exchange and severity of airways obstruction,
Vasculature
Chronic Thromboembolic Pulmonary Hypertension (CTEPH) Pulmonary embolism is a common finding in hospitalized patients, but more than 80% ofpulmonaryemboli produce no clinical symptoms, If adequate recanalization does not occur, a chronic stage with pulmonary arterial hypertension will supervene, CT can demonstrate mural thrombi in the central pulmonary vessels, so it can aid in selecting patients for pulmonary thromboendarterectomy (see Chapter 23), The CTA technique should be used for this type of investiga-
tton, although signs of mosaic perfusion can be appreciated on unenhanced CT and HRCT images, CT Morphology In more than 70% of patients with chronic recurring pulmonary emboli, CTshows a patchwork of sharply demarcated areas of increased and decreased parenchyma! density caused by regional variations in perfusion ("mosaic pattern" or "mosaic perfusion") (Fig, 9,66), Vascular calibers appear increased in the higher-density areas and decreased in the lower-density areas: this distinguishes the mosaic pattern from ground-
9 Lungs and Tracheobronchial System
b
a
Fig. 9.66 Chronic pulmonary embolism, (a) CT A of the pulmonary arterial trunk in chronic embolism shows thrombi adherent to the vessel walls (arrowhead) and demonstrates a recent embolus in the left lower lobar artery (arrow), (b) "Mosaic perfusion" pattern in chronic
c pulmonary embolism with areas of decreased density (narrow vessels) and areas of compensatory hyperperfusion (dilated vessels), (c) Coronal MPR (4 x 1/6) demonstrating mosaic perfusion and peripheral vascular rarification.
glass opacities. Mosaic patterns due to air trapping are accentuated on expiratory scans. Accompanying pulmonary infarcts are most commonly found in the basal zones. The signs of pulmonary hypertension (main trunk of pulmonary artery> 29 mm in diameter with rapid tapering of distal vessels) are accompanied by wall irregularities, hypodense deposits, or nonenhancement of smaller vessels. Pulmonary Veno-occlusive Disease
Pulmonary veno-occlusive disease (PVOD) is a rare abnormality characterized by the combination of pulmonary arterial hypertension and postcapillary hypertension, chiefly pulmonary edema. There is evidence of repeated pulmonary venous thrombosis. The etiology is unknown. Associations with viral infections, bone marrow transplantation, some medications, the use of oral contraceptives, and radiation therapy to the thorax have been described. •
CT Morphology
The combination of dilated pulmonary arteries, signs of increased right ventricle load, and pulmonary edema is virtually diagnostic. CT shows interlobular thickening and areas of groundglass attenuation consistent with interstitial edema (Fig. 9.67).
Fig. 9.67 Veno-occlusive disease with faint alveolar edema and bronchial wall thickening.
Hepatopulmonary
Syndrome
Patients with liver cirrhosis may develop hypoxemia without intrinsic pulmonary disorder. The fact that liver transplantation is able to reverse the signs underline the functional etiology.
Chest Trauma •
CT Morphology
Arising from decreased vascular tone there is dilatation especially of the small pulmonary vessels and capillaries. With increasing severity of cirrhosis there is increasing occurrence of small arteriovenoUs shunts that are best appreciated in the subpleural area (Fig. 9.68).
a
b
Fig. 9.68 Hepatopulmonary syndrome with increased subpleural vascularity (a) and small arteriovenous shunts are best appreciated on a thin slab MIP (b).
Trauma and Postoperative Changes Chest Trauma
Chest injuries are responsible for s:25% of all trauma-related deaths. A blunt trauma (with an intact chest wall) is commoner that a penetrating trauma and is-most frequently related to deceleration forces due to a motor vehicle accident. In most cases, the plain chest radiograph is the initial investigation and CT plays a secondary role, mostly to clarify an abnormality suspected on the plain radiograph. Indications for a CT examination in chest trauma are relatively limited (Table 9.45). An acute indication for CT exists in patients with suspected injury to mediastinal structures, particularly the great vessels. Bleeding from arteries or brachiocephalic veins (unenhanced CT) and aortic injuries (see Chapter 24) can be confidently diagnosed with optimal spiral CT, which has replaced preoperative angiography. Other life-threatening injuries include a tracheal rupture, tension pneumothorax, hemothorax, and cardiac tamponade. Other indications for CT in chest trauma are the differentiation of effusion, hemorrhage, atelectasis, and contusion; localizing an opacity to a specific anatomic structure (lung vs. pleura); and defining the extent of an indeterminate pulmonary density on the chest radiograph. A pulmonary contusion leads to intra-
Table 9.45 Recommended chest trauma
diagnostic
procedures
Type of injury/location of injury
Diagnostic procedure of choice
Chest wall/skeleton
Chest radiograph
Hematothorax/ pneumothorax
Chest radiograph/ultrasound
Pulmonary contusion
Chest radiograph, followed by CT
Pulmonary laceration
Chest radiograph, followed by CT
Mediastinal
CT angiography
hematoma
Aortic injury
CT angiography
Cardiac injury
Echocardiography
Bronchial rupture
Bronchoscopy
Diaphragmatic
Combination of procedures in most cases
Indeterminate
rupture opacity
in
CT
parenchymal hemorrhage and its presence alone does not indicate CT. A more severe trauma causes disruption of the parenchyma resulting in laceration. The radiological diagnosis of diaphragmatic rupture is problematic, and often delayed. Diaphragmatic rupture is commoner on the left side than the right, and only a few cases are associated with the acute herniation of abdominal contents into the chest.
361
9 Lungs and Tracheobronchial System •
CT Morphology
Pulmonary contusion is manifested by focal, confl uent alveolar shadowing that may progress to an area of consolidation. Free intraparenchymal air is suggestive of pulmonary laceration. A pneumatocele is a rounded intraparenchymal air collection caused by air leakage and lung retraction following a laceration (Fig. 9.69). Pneumatoceles are usually small (less than 5mm) but may also range from 2 to 5 cm in diameter, rarely larger (up to 14cm) dependent on the size of laceration and can be demonstrated several hours after the injury. They do not have well-defined walls, and they may be filled with fluid (e.g., blood). Pneumatoceles and hematomas generally resolve without treatment. Hematomas may persist for some months and may produce a solitary nodule or mass that should not be mistaken for a malignant tumor. A tracheal or bronchial rupture is mostly associated with an injury to the upper bony thorax (e.g., the first three ribs). The tear may be partial or complete. Bronchial rupture occurs within 2.5 cm of the carina in 80% of patients. A right -sided rupture is more common than left sided. Hallmarks of a tracheobronchial rupture are a (frequently large) persistent pneumothorax, a persistent pneumomediastinum, an extrabronchial air collection at the rupture site,
a
b
Fig. 9.69 Chest trauma, (a) Trauma-induced laceration of the lung parenchyma (absence of pleura, arrow) with a broad pneumothorax. There is extensive chestwall emphysema but no mediastinal emphysema. (b) Cystic pneumatocele secondary to a pulmonary laceration. Areas of intralesional hemorrhage appear as
peripheral atelectasis that persists despite treatment, or sagging of the lung away from the hilum to the inferior and lateral hemithorax ("fallen lung" sign with complete disruption). Torsion of the lung is extremely rare; it occurs almost exclusively in children, and represents a surgical emergency. Torsion of the whole lung or of only a lobe usually results in pulmonary infarction. It is manifested by anatomic malposition of the hilar structures, the bronchial tree and shows pulmonary opacities due to atelectasis or infarction. Multiplanar reformats in the coronal or sagittal plane are most hel pful for diagnosis of a diaphragmatic rupture. Gas-containing bowel structures within the hemithorax, displacement of the nasogastric tube, elevation of the hemidiaphragm, and a pleural effusion are suggesting or associated findings. Adult Respiratory Distress Syndrome
Adult respiratory distress syndrome (ARDS) is a generalized, but heterogeneous, response to pulmonary injury. A great variety of mediators can precipitate ARDS, including sepsis, shock, trauma, pneumonia, aspiration, transfusion, and micro emboli. Often more than one factor is present. The pathomorphologic changes initially
c consolidations and ground-glass opacities, (c) Extensive pulmonary contusion with consolidation of the lower lobe. The intra-alveolar blood shows less contrast enhancement than atelectasis, with clear delineation of the vessels (CTangiogram). The peripheral air inclusions suggest the presence of a laceration.
Adult Respiratory Distress Syndrome
consist of a capillary permeability defect producing a noncardiogenic interstitial and alveolar edema, secondary leading to fibrosis and vascular obstruction resulting in ischemia and necrosis. Initial radiological manifestations may not appear until 12 hours after the onset of clinical manifestations (Table 9.46). Radiographically, three phases are differentiated: the exudative, proliferative, andfibroticphase. •
CT Morphology
Stage 1 ofARDS
(first 24 hours) is dominated by signs of interstitial edema (ground-glass opacity, prominent interstitial septa) and pulmonary hypertension (dilated central pulmonary arteries, enlarged right heart). Scanning in stage 2 (2 ro 7 days) shows widespread, patchy or diffusely confluent consolidations with a lower zone predominance and preservation of normal parenchyma! subdivisions. The incidence of pneumonic infiltration is around 70% in this stage with an at least doubled mortality risk. The diagnosis of pneumonia is equally difficult with radiography and CT. Bronchiectasis within areas of ground-glass opacification are regarded as early sign offibrosis. Signs of pulmonary interstitial emphysema (PIE) due to advanced alveolar wall destruction can be seen in prolonged cases of ARDS that have received continuous positive-pressure ventilation. CT reveals air collections around the Table 9.46
Stages in the evolution of ARDS
Stage 1 (early exudative stage)
First
-24 hours • Vasodilatation • Interstitial and alveolar edema (alveolar filling) • Capillary stasis with thrombosis leads to intersti-
tial thickening Stage 2 (intermediate stage) • Alveolar
shadowing
proliferative
2-7 days
due to hemorrhage
and
fibrin deposition
• Early (day 2 to 4) consolidation with air bronchogram • Late (day 4 to 7) patchy consolidation Stage 3 (chronic fibrotic stage) • Coarse reticular distortion • Pneumothorax
pattern, is frequent
interstitial
>7 fibrosis,
days lung
pulmonary veins and lymphatics within the interstitium earlier than chest films, air cysts up to 5 mm in diameter can be seen in a subpleural or perihilar location. When arranged in peribronchovascular chains, these cysts form "interstitial air tracks' that predispose to pneumothorax (32%), formation of bullae (30%) or pneumomediastinum (13%), and imply a poor prognosis. Stage 3 (after 7 days] is dominated by a coarse reticular pattern, signs of interstitial fibrosis, and signs of parenchyma! distortion. Pneumothorax and mediastinal emphysema are common associated findings. Long-term, sequelae are fibrotic changes predominantly in the ventral lung areas suggesting that parenchyma! consolidation has a protective effect against mechanical injury through ventilation. CT is superior to radiography in demonstrating the inhomogeneous distribution of atelectatic lung areas. The goal of ventilation with positive end-expiratory pressure (PEEP) is the recruitment of atelectatic alveolar air spaces in the dorsobasal lung areas. It was found that subtotal atelectases with reduced but still existing ventilation (representing ground-glass opacification in C1) are more readily recruited by PEEP as opposed to areas with complete atelectasis and non existing ventilation (consolidation on C1). CT studies in patients with ARDS under zero and positive end-expiratory pressure (ZEEP and PEEP) showed that alveolar recruitment takes place with increasing PEEP along a craniocaudal and ventrodorsal axis. The dorsal and basal areas are recruited last at the expense of overinflation of the ventral and cranial lungs (which increases the risk of damage and subsequent fibrosis). CT was suggested as visual guide for individually optimizing the mechanical ventilation parameters with respect to ventilation pressure and tidal volume, in addition to respiratory function parameters (Fig. 9.70).
369
9 Lungs and Tracheobronchial
System
Fig. 9.70 Severe ARDS with consolidation and ground glass opacification. There is alveolar recruitment as positive end expiratory pressure (PEEP) is increased from 15 mm Hg (a) to 25 mmHg (b). Note the pleural effusion and ventrodorsal density gradient (low-dose scan in a very obese patient).
Pneumonectomy
and Lobectomy
CT is the modality of choice for evaluating the post -pneumonectomy chest. CT is useful for detecting tumor recurrence and diagnosing complications (e.g., empyema). MRI has not shown any definite advantages for these applications. •
CT Morphology
CT shows the anatomic distortion caused by resective surgery, marked particularly by the rotation and ipsilateral displacement of mediastinal structures. These changes may be sufficiently pronounced, especially after a right-sided pneumonectomy, to cause angulation or compression
of the distal trachea or left main bronchus. The formation of a fluid collection that may persist for years after the resection is more common than complete obliteration of the empty thoracic cavity (60% vs. 40%). Recurrent tumor usually presents as a perihilar or paramediastinal mass of soft-tissue density. If an empyema develops, CT shows the lesion as a hyperdense fluid collection (>20HU) surrounded by a thickened, enhancing rim. A lobectomy results in distortion of the central tracheobronchial system with compensatory expansion and displacement of the remaining pulmonary lobes. After an upper lobectomy, for example, the apical segment of the lower lobe (segment 6) shifts upward to occupy the apical zone. After a lower lobectomy, the middle lobe or lingula (anteriorly) and the posterobasal segment of the upper lobe (posteriorly) occupy the vacated basal zone. Bronchoplastic techniques are used in central bronchial carcinomas that affect only one pulmonary lobe but extend locally beyond the lobar bronchus into the main bronchus. Such tumors may still be resected by lobectomy if the infiltrated part of the main bronchus is resected either as circular cuff or as wedge-shaped patch and the remaining bronchial surfaces are anastomosed directly (bronchial sleeve resection). Short-term complications are bronchial dehiscence, long-term complications are tumor recurrence and fi brotic stenosis (Fig. 9.71, see also 9.72 c).
Fig. 9.71 Coronal thin MPR (a) and 10 mm thick MPR (b) in a patient after upper lobe lobectomy and bronchial sleeve resection (4 x 1 /6). High grade anastomotic stenosis with prestenotic pouch-like bronchial dilatation and severe poststenotic fibrotic changes due to recurrent infection.
a
b
Lung Transplantation Lung Transplantation
Pulmonary densities seen in the postoperative period may signify a reimplantation response, an infection, or allograft rejection (Table 9.47).
Reimplantation Response The reimplantation response occurs in 40% of lung transplant recipients and is manifested within 8-12 hours after the transplantation. It is an edematous response caused by increased capillary permeability due to ischemia and impaired lymph clearance due to the disruption of lymphatic pathways. •
CT Morphology
The interstitial and alveolar edema are manifested on CT by accentuation or thickening of bronchial walls and interlobular septa, increased reticular markings, and ground-glass opacities. The changes usually resolve in 1 to 2 weeks. In protracted cases where adequate resolution does not occur, the changes may progress to "mntaldysfuncnon" (transplantfailure).
Acute Rejection Almost all patients experience at least one episode of acute rejection during the first three weeks, and many have from one to six episodes during the first three months. Acute rejection is manifested clinically by hypoxia and histologically by peribronchial and perivascular monocytic infiltrates. •
CT Morphology
HRCT has proven to be a more sensitive but less specific indicator of pulmonary infiltrates in the
Table 9.47
Chronic Rejection The incidence of chronic rejection is approximately 10% during the first year after lung transplantation and 20% during the second year. Chronic rejection is manifested clinically by obstructive ventilatory impairment and histologically by intraluminal granulations in the small bronchi with involvement of the alveolar airspaces and interstitium (obliterative bronchiolitis, see also p. 300). CT Morphology
Sensitive but nonspecific HRCT findings are decreased vascularity, bronchiectasis, nonseptal linear densities with a basal predilection, and a mosaic pattern (Fig.9.72b). Regions of air trapping are seen on expiratory scans. Additional signs are cylindrical bronchiectasis and a progressive increase in pulmonary opacity. Advanced cases show increasing signs of pulmonary fibrosis, which in rare cases may reach the proportions of progressive massive fibrosis (PMF).
Airway Complications Tracheal combined
anastomoses heart-lung
are performed transplantations.
CT manifestations of lung allograft rejection
Acute rejection • • • • •
setting of a rejection response. Findings consist of thickened bronchial walls and interlobular septa, diffuse air space shadowing (typically but not always in a symmetrical perihilar distribution) , and occasional ground-glass opacities and pleural effusions (Fig.9.72a). Because the morphologic changes are nonspecific, CT is not helpful in differentiating acute rejection from infection (especially CMY or Pneumocystis carinii).
Thickened bronchialwalls Thickened interlobularsepta Diffuse airspace shadowing (symmetric,perihilar) Ground-glass opacities Pleural effusion
Chronic rejection Thickened bronchialwalls Thickened interlobularsepta Cylindricalbronchiectasis Decreasedvascularity Pleural and interstitialfibrosis Focal peribronchialopacities(BOOP)
only in Single
371
372
9 Lungs and Tracheobronchial System
a
b
Fig. 9.72 Complications after lung transplantation. (a) Acute rejection: with diffuse alveolar damage (DAD): ground-glass opacities, prominent intralobular septa and interlobar effusion, (b) Chronic rejection: tubular bronchiectases and mosaic pattern with air trapping due to bronchiolitis obliterans. (c) Chronic re-
and double lung transplantations employ anastomoses in the area of the distal main bronchi near the origin of the upper lobe bronchus. Bronchial dehiscence is a frequent complication during the early postoperative days. It is caused by ischemia, since the bronchial segments bordering the anastomosis are supplied only by retrograde flow through the bronchial arteries and by diffusion. As healing occurs, sites of high-grade fibrotic stenosis may develop in the central tracheobronchial system. The diagnostic modality of choice is spiral CT using a high-resolution technique (1-2 mm). Spiral CT is superior to bronchoscopy for detection and localization.
d
c
jection: bronchiolitis obliterans with subpleural fibrosis. (d) Bronchial anastomotic dehiscences: (1) broad bronchial wall defect at the inferior border of the main bronchus (arrow) and (2) small fingerlike protrusions (arrowhead).
•
CT Morphology
Two forms of anastomotic dehiscence are recognized on CT. One is manifested by small peribronchial air collections with small "punctate" bronchial wall defects that resemble pseudodiverticula. This type heals without sequelae. The second form is manifested by shallow, pouch-like protrusions with broad bronchial wall defects that occur chiefly in the inferomedial aspect of the right main bronchus and extend inferiorly in continuity with the main bronchus, medial to the bronchus intermedius (Fig. 9.72 c). This type predisposes to bronchial scarring and stenosis.
10 Mediastinum, Pleura, and Chest Wall C. Schaefer-Prokop
Anatomy Examination Technique Enhancement Characteristics Differential Diagnosis of Mediastinal Masses Masses Predominantly Involving the Anterior Mediastinum Thymic Lesions: Hyperplasia, Thymoma, Thymic Carcinoma Extragonadal Germ Cell Tumors (p. 382) Goiter (p. 383) Parathyroid Adenoma (p. 383) Hodgkin's Disease (p. 383) Non-Hodqkin's Lymphoma (p. 384) Lymphangioma (p. 385) Mesenchymal Tumors (p. 385)
(p. 380)
Masses Predominantly Involving the Middle Mediastinum Lymphoma (p. 386) Tumors of the Tracheobronchial System (p. 386) Bronchogenic Cyst, Pericardial Cyst (p. 386) Paraganglioma (p. 386) Masses Predominantly
Involving the Posterior Mediastinum
Neurogenic Tumors (p. 387) Extramedullary Hematopoiesis (p. 388) Esophageal Tumors (p. 389) Esophageal Diverticula and Duplication Cysts (p. 389) Esophageal Varices, Azygos Continuation (p. 389) Meningoceles and Neuroenteric Cysts (p. 389) Lymph Node Enlargement Pancoast Tumor Tumors Involving the Chest Wall
Pleural Tumors
.
394
Benign Pleural Tumors (p. 394) Pleural Metastases (p. 394) Malignant Mesothelioma (p. 395) Pleural Lymphoma (p. 395) Superior Vena Cava Syndrome (p. 396) Diffuse Mediastinal Diseases
.,
397
Acute Mediastinitis (p. 397) Fibrosing Mediastinitis (p. 397) Mediastinal Lipomatosis (p. 398) Diffuse Pleural Diseases
.
398
Pleural Effusion, Hemothorax, Chylothorax (p. 398) Pleural Empyema (p. 399) ,Pleural Fibrosis, Asbestos-Associated Changes (p. 400) Trauma and Postoperative Findings
401
Pneumothorax (p. 401) Mediastinal Hematoma (p.402) Mediastinal Emphysema (p. 402) Diaphragmatic Rupture (p. 403) Cardiac Surgery (p. 404) Heart Transplantation (p. 404)
Computed tomography is the standard imaging technique for the examination of the mediastinum. It is most commonly used for the evaluation of suspected masses and lymph node staging in chest tumors (Table W.I). The pleura and chest wall are generally examined as part of a chest C'I, CT used to be inferior to MRI with respect to the evaluation of the relationship of abnormalities (e.g .. tumor infiltration) to the various mediastinal structures, the chest wall, or the pleura. This was due to the excellent soft-tissue contrast and the availability of multiplanar im-
aging with MRI. With the advent of spiral CT and, more recently, multislice CT, this has changed. Multiplanar imaging is now also available with CT, and the spatial resolution of multislice CT is often superior to MRI. In addition, calcifications and associated findings in the lung parenchyma can be better-detected with CT, With proper timing of the injection of intravenous contrast agents, the contrast of the various anatomical structures is usually sufficient with CT, leaving only a few cases to MR where the clinical question cannot be solved with CT.
Anatomy
Table 10.1
Indications
for CT scanning
of the mediastinum
Tumor diagnosis Suspected tumor (on chest radiograph) Tumor characterization T staging N
staqnq
Detection
or exclusion
of a mediastinal
Differential diagnosis Bronchial carcinoma,
primary
Bronchial
lymphoma,
carcinoma,
mass
mediastinal
tumors
testiculartumors
Diffuse disease Mediastinitis Superior vena cava syndrome Vascular disease See Chapter 24 Pleural
and chest-wall
lesions
Suspected empyema Differentiation of effusion/infiltrate/atelectasisltumor Asbestos-associated disease Chest
trauma
Anatomy The mediastinum is conveniently divided into three compartments-anterior, middle, and posterior-despite the absence of anatomic boundaries that would limit the spread of disease between the compartments, The ant'enor compartment is located anterior to the pericardium, ascending aorta, and superior vena cava and contains the thymus and may contain portions of the thyroid, The middle compartment contains the heart and great vessels and the central tracheobronchial system including the pulmonary hila, The postertor compartment is composed of the retrocardiac and paravertebral spaces, lt contains the descending aorta, esophagus, thoracic duct, and the azygos-hemiazygos venous system The differentiation of vascular structures and lymph nodes in these regions requires a sound knowledge of cross-sectional anatomy (Fig, 10.1), The size of the normal thymus is highly variable until about 20 years of age, If the thy-
mus can be visualized in young adults, it usually appears as an arrowhead-shaped structure (62%), lt is less common to find two thymic lobes (32%) or one isolated lobe (6%), The thymus undergoes involution from about 20 to 60 years of age and is replaced by fatty tissue, Normally, cannot delineate the visceral and parietal layers of the pleura, The pleural layers, endothoracic fascia, and deepest intercostal muscles blend together on CT scans to form a well-defined "pleural line" of 1-2 mm in thickness (Fig, lO.2), Normally, there are no structures of soft-tissue density along the inner contour of the ribs, Epipleural fat may be visible in obese patients and along marginal rib segments, Paravertebral portions of the pleural line often cannot be visualized, but segments of the intercostal veins can be identified as linear streaks of soft-tissue density, Epipleural fat and intercostal veins should not be mismterpreted as pleural thickening,
cr
375
376
10 Mediastinum, Pleura, and Chest Wall
SVCfRA
Fig. 10.1
anatomy. Pericard
RA -----
Schematic
representation
of
mediastinal
Ao Aorta AUL, VUL, BUL Upper lobar arteries, veins, and bronchi ALL, VLL, BLL Lower lobar arteries, veins, and bronchi Ai-io, 81-10 Segmental arteries and bronchi Bi Bronchus intermedius AZV Azygos vein BCA, BCV Brachiocephalic artery and vein CCA Common carotid artery E Esophagus PA Pulmonary artery RA, LA Right atrium, left atrium RV, LV Right ventricle, left ventricle RPA, LPA Right pulmonary artery, left pulmonary artery SCA Subclavian artery SVC Superior vena cava T Trachea
Contrast Administration
nI IJ
Extrapleural
fat
--
-Intercostal muscles
Vein Artery Nerve
sharp line (1-2 mm wide)
Muscles of chest wall
a Fig. 10.2
Paravertebral muscles
Schematic
Examination
Endothoracic fascia --1---\0:
Internal ---+-intercosta I muscle
---
Visceral and parietal pleura
External intercostal muscle
Extrapleural
fat
Intercosta I intimus muscle Intercostal
finepleural line on CT
)
fat
~
representation
b of the anatomy
of the chest wall.
Technique
Scanning of the mediastinum, pleura, or chest wall is usually done as part of a general thoracic CT examination (see Chapter 9). A more focused examination of tumors, hilar lymph nodes, or the chest wall benefits from coronal or sagittal reformations, and therefore should rely on thinsection spiral or multislice CT. Contrast enhancement is of subordinate importance because it mainly serves as a means of differentiating between vessels and solid structures (lymph nodes, tumors). Patient Preparation
No special patient preparation is needed. Having the patient swallow contrast material (e.g., highly viscous barium preparations such as Esophocat®) on the examination table may improve the identification and evaluation of the esophageal lumen. However, luminal opacification is very often incomplete and the approach therefore cannot be generally recommended.
and should be employed whenever available. For most indications, however, it is sufficient to reconstruct axial sections of 5-7 mm width for evaluating mediastinal abnormalities. Retrospective ECG-gating can be employed to reduce pulsation artifacts in regions adjacent to the heart. Cardiac gating techniques with a four detector-row scanner typically employ a pitch factor P*=1-2 (corresponding P=0.250.5). This leads to a substantial increase in radiation exposure and reduced volume coverage per time. Cardiac retrospective gating algorithms, therefore, cannot be generally recommended. Instead, special gating techniques that exclude the projection data of the period of maximum movement (late systole) from the multislice raw data set are better suited because they allow for pitch factors P of 0.75-0.875 (p*= 3-3.5 with 4slice scanners) and only slightly increase radiation exposure. This makes it possible to use 0.75-1.25 mm sections and still cover the mediastinum within one breath hold. Not all manufactures, however, do have such algorithms available.
Acquisition Parameters
Thin collimation reportedly improves the evaluation ofthe pulmonary hilum, the chest wall at the lung apex, and the diaphragm. Thin-section multislice CT (SC= 0.75-1.5 mm with 4- to 16slice scanners) is best suited for this purpose
Contrast Administration
Noncontrast CT is mainly required for evaluating hemorrhage, indeterminate mediastinal masses, or pleural thickening. It also has been suggested as an appropriate scanning technique
10 Mediastinum, Pleura, and Chest Wall
and investigating vascular disorders (Tables 10.2, 10.3). Most indications require enhancement only of the mediastinal vessels to be able to differentiate them from lymph nodes or tumors. For this indication relatively small amounts of contrast medium «90ml) need to be injected at moderate flow rates (s. 2 mlls). Inflammatory processes or evaluation of chest wall invasion are best detected after injection of 90-120 ml contrast material with a start delay of 30-50 s. Evaluation of tumor vascularity for characterization of suspected mediastinal lesions requires injection of higher volumes (100150 ml) at higher How rates (3-5 ml).
for patients with peripheral bronchogenic carcinoma since lymph nodes that exceed the size criterion of 1 cm short axis diameter can be detected even on noncontrast scans. However, a more subtle evaluation of such nodes may not be possible, and with the advent of more sophisticated morphologic criteria for lymph node differentiation, multislice scanning with contrast administration can be highly recommended. Intravenous contrast is helpful for differentiating vessels, lymph nodes, and other structures of soft-tissue density, for evaluating lesions of the chest wall, characterizing lesions,
Table 10.2
Suggested
protocols
for CT scanning
of the mediastinum,
pleura and chest wall
General Oral contrast
none or 1 tbi spoon Esophocatw
Patient position
Supine with elevated
arms
Scan range
Chest: from posterior
lung recess to lung apex
Respiratory phase
Inspiration
Windowing
Noncontrast CT: Contrast-enhanced Bone:
Scan parameters Bf Standard Volume
imaging
W/L=400/40 W/L= 400/60 W/L= 1500/300
CT:
1-slice SCITF/RI
4-slice
(HR kernel)
SC
16-slicea SC
axial SWIRl
5/8/4 T
2-3 i
1-1.Sf
5/4
3/5/2 t
1-1.25 t
0.75-1.25T
3/3
a
[e
MPR1• SWIRl
3/4 cor
Asbestos screening
2/25 i
Contrast
1-slice V/F/D
4-slice V+IM/FID
16-slice V+IM/FIDX'
Standard
70/2/20
70+50/2/20
50+30/2/30
Tumor staging
90/3/30
90+50/3/30
90+50/3/25A
Abscess,
90/2/60
90+50/3/60
90+50/3/75
injection
empyema
SC = slice collimation
1-1.25/251
(rnrn). TF = table feed (rnrn/rotation).
width (rnrn). MPR = multiplanar CM = contrast
c
material.
reformats,
RI = recon. increment
axial = axial sections,
Contrast concentration
0.5-0.75/25
Pitch P=TF/(NxSC)
b
MPR are reconstructed
C
read: single section mode HRCT with 0.5-1.5
dread:
(rnrn). f i = scanning direction,
cor = coronal. sag = sagittal
= 300 mg/ml iodine
V = volume of CM (rnl). N = volume of saline flush (rnl) F = flow rate (mils), a
D = start delay, X = trigger
should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with from a "secondary
25A = start delay of 25 s after reaching
1-2/25
raw data set" with SWIRl ~ 1-1.5/0.7 mm collimation
or 0.5-0.8/0.5
and 10-20 mm increment
50 HU enhancement
region
16-slice scanners
in the desc. aorta
between sections
SW = effective section
Image Processing Table 10.3
Suggested
examination
technique Scan
Contrast
Injection
standard
standard
enhancement
standard/volumetric
standard/vascular
volumetric
vascular
standard
standard
standard/volumetric
standard/vascular
standard
late phase
DO of effusion/infiltrate/atelectasisltumor
standard
standard
Pleural calcifications
HRCT
Suspected Tumor
tumor
(on chest radiograph)
characterization
T staging N staging Pleural and chest-wall Suspected
Vascular
lesions
empyema (asbestosis)
Parameters
CTAsee
diseases
Image Processing Thick coronal reformations (3-5 mm) through the mediastinum improve the evaluation of mediastinal masses and lymph nodes. The central airways can best be evaluated with thicker semicoronal MPR (1 Omm section width) parallel to the course of the trachea (see Fig. 2.18). Tissue transition projections (TTP) of some 20 mm width are an alternative display tool for lesions of the
Enhancement
enhancement
enhancement enhancement enhancement
scanning enhancement
Chapter24
central airways (see Fig. 2.36). Thin-slab mlP (MinlP) also can also be used for display of the central airways but are usually inferior to thick MPTorTTP (see Figs. 2.16 and 2.18). Thin-slab MIP or volume rendering (VRT) can help to define vascular abnormalies (see Fig. 10.13). VRT may also be a helpful tool for preoperative visualization of mediastinal masses (see Figs. 10.10 and 10.12).
Characteristics
Ectopic thyroid tissue. parathyroid adenomas. and paragangliomas are examples of primary. hypervascular mediastinal masses that show intense enhancement following intravenous contrast administration. Foci of tumor necrosis (in germ cell tumors or lymphomas) appear as hypoattenuating areas after contrast administration. Cystic lesions (enteric. pleuropericardial. bronchogenic) do not show contrast enhancement (dOH attenuation difference between pre- and postcontrast scans).
Thickened pleural layers that enhance with intravenous contrast suggest a neoplastic or inflammatory processes. Tumors that have invaded the chest wall are more clearly delineated after contrast administration. depending on the vascularity of the tumor. Hypervascular chest wall lesions are rare but enhance earlier than hypovascular tumors in which the enhancement builds up in a later phase, often due to increased contrast leak into the interstitium.
Differential Diagnosis of Mediastinal Masses The main objective of CT is to evaluate the location and extent of a mediastinal mass. The presence of fat and cystic components together with size. shape. and precise location within the mediastinal compartments are the key features for differential diagnoses of mediastinal masses. Generally there is no difficulty in distinguishing tumor-like lesions from true neoplasms. When a
tumor is found, CT may help to narrow the differential diagnosis (Table 10.4). The patient's age, the presence of calcifications, and the degree of vascularity of the tumor provide additionalinformation.
379
380
70 Mediastinum, Table
10.4
Location
Pleura, and Chest Wall and CT morphology Anterior
Thymoma, thymic Goiter Lymph nodes Hodqkin's disease
Solid masses
Cystic
masses
Calcifications
Fatty
Vascular
lesions
Pseudomasses
Masses
carcinoma
masses
Middle Lymph
mediastinum node
Non-Hodqkin's
lymphoma
Castleman'stumor Goiter
Germ cell tumors Parathyroid adenoma Mesenchymal tumors
Paraqanqioma
Thymic cyst Teratoma Dermoid cyst Goiter
Bronchogenic cyst Pericardia I cyst
cell tumors
Lipoma Thymolipoma Teratoma Dermoid cyst Aortic
Predominantly
mediastinum
Neurogenictumor LYl"TlJh node Non-Hodgkin's lymphoma Goiter (rare) Esophageal tumors Extramedullary hemato-
nberculosis. silicosis
Neurogenic cyst Enterogenic cyst Pancreatic pseudocyst Meningocele Neuroblastoma
Goiter
Ganglioneuroma Neuroma Goiter
Lipomatosis
Lipoblastoma Neurogenictumors
aneurysm
Morgagni
Posterior
poiesis
Lymph node: sarcoidosis,
Teratoma Dermoid cyst Malignant germ Goiter
elements
of mediastinal
mediastinum
Aortic aneurysm Sinus of Val salva aneurysm Pulmonary artery aneurysm Anomalous pulmonary venous drainage Coronary artery aneurysm
Bochdaiek hernia Hiatal hernia Achalasia
hernia
Involving the Anterior
The commonest anteriormediastinal masses are thyroid or thymic lesions, teratomas, and lymphomas.Mesenchymaltumorsarerare.CTisused to investigate suspicious findings on chest radiographs and for the staging of known tumors. Thymic Lesions: Hyperplasia, Thymoma, Thymic Carcinoma
Thymic hyperplasia may be seen in thyrotoxicosis, Addison's disease, lupus erythematosus, Behcet disease, and Hashimoto thyroiditis. It frequently occurs as a rebound phenomenon in the period of recovery after stress reactions or chemotherapy, especially in young patients, following chemotherapy for various diseases (Hodgkin's disease, sarcoma, Wilms' tumor, ALL). The
Esophagealvarices Azygos vein anomalies
Mediastinum
"thymichyperplasia" seeninm yastheniagravisis actually a lymphoid response that is not always associated with enlargement ofthe gland. Benign and malignant thymomas represent a neoplastic proliferation ofthe thymic epithelium. The prognosis ofthymic epithelial tumors follows the classification of Levine and Rosai who differentiate between benign thymoma, invasive thymoma, and thymic carcinoma. Invasive thymoma is a malignant thymoma (category 1) with metastases and local invasiveness but it is cytoarchitecturally indistinguishable from benign thymomas. Thymiccarcinom aisamalignantthymoma (category 11) withdefinite cytologicalatypia. Thymomas are most commonly diagnosed in the fourth decade of life. The incidence of thymoma in patients with myasthenia gravis is ap-
Thymic Lesions: Hyperplasia, Thymoma, Thymic Carcinoma
proximately 15%. Benign tumors are indistinguishable from malignant lesions by their CT features alone, and all thymic tumors require surgical removal. In patients with myasthenia gravis nonionic contrast media should be used. When frank bulbar symptoms are present, care should be taken to be avoid a myasthenic crisis. The thymus is often involved in Hodgkin and non-Hodgkin lymphomas, mostly as combined thymic and mediastinal node disease and only rarely as isolated involvement. •
CT Morphology
Given the variability in thymic size, normal limits cannot be stated for patients under 20 years of age. After age 20, the maximum transverse diameter should be less than 15 mm. A diameter of 15 mm or more signifies thymic hyperplasia. Rebound hyperplasia after chemotherapy may resolve completely within a period of months, with 25% of patients showing persistent thymic enlargement by more than 50% of the original gland size. Benign thymomas typically appear as rounded or oval well-circumscribed structures of soft-tissue density located in front of the aorta in the upper anterior mediastinum (Fig.10. 3a). In rare cases it is located more caudally at the base of the heart. Calcifications occur in 25% of thymomas, regardless of whether they are benign or malignant. Primary cystic components are common. Approximately 30-40% of thymomas are classified as invasive or malignant, meaning that they show a locally invasive pattern of growth. The invasion of adjacent structures is manifested by pleural or pericardial thickening, effusion, or bony erosion of the sternum. The obliteration of
mediastinal fat planes is generally seen in invasive thymoma but may also be present in noninvasive tumors. Pleural extension as droplet spread without continuity with the primary tumor is seen in 15%. Thymic carcinomas do not show distinguishing CT features from invasive thymomas, yet they infiltrate more aggressively and mediastinal lymph nodes and extrathymic metastases are frequent (Fig. 10.3b). Thymic carcinoids present with an ectopic ACTH Cushing syndrome. They are indistinguishable from thymomas and diagnosed by open biopsy or surgical resection. Thymic cysts frequently develop after chemotherapy or after radiation therapy to the mediastinum. Intracystic hemorrhage is common, and thymic cysts show a highly variable CT density (from -10 to 80HLJ) depending on whether the cyst contains fresh or old blood, necrotic foci, or cholesterol. Involution of the thymus occurs between 20 and 60 years of age as the glandular tissue is replaced by fat. Thymolipomas are very rare tumors encountered in children and young adults. CT demonstrates a mass of predominantly fat density. Thymolipomas can reach a very large size and, being soft tumors, mold themselves to adjacent mediastinal structures without displacing them. The thymus is involved in approximately 30% of patients with Hodgkin 5 disease. Accompanying lymphadenopathy is almost invariably present. CT cannot distinguish rebound hyperplasia from normal thymic tissue or early invasion by Hodgkin's disease. There are no pleural implants in lymphomas.
Fig. 10.3 (a) Benign thymoma in a patient with myasthenia gravis. (b) Recurrence of malignant thymoma with a pleural drop metastasis. a
b
10 Mediastinum, Pleura, and Chest Wall
Extragonadal Germ Cell Tumors
Germ cell tumors (teratoid blastomas) arise from cells of the embryonic germ layers and comprise approximately 15% of all mediastinal masses. Germ cell tumors are commonly located in the thymus bed or within the thymus. While there is no sex predilection for the benign forms, the malignant forms predominantly affect young men (25% of cases). Germ cell tumors include benign and malignant teratoma, dermoid cysts, seminoma, the rare embryonal carcinoma, and choriocarcinoma. Teratomas contain elements of all three germ layers, whereas dermoid cysts have a (predominantly) epidermal origin. •
CT Morphology
Dermoid cysts can grow quite large. They have a rounded shape and are sharply demarcated by a capsule. They are composed of one or more cysts with well-defined walls that may show calcification. Chest radiographs and CT usually demonstrate an asymmetrical mass projecting into the upper anterior mediastinum on one side. Rarely it may extend into the posterior mediastinum or lung. The contents of the cyst show water or fat density. The detection of rudimentary teeth, foci of ossification, or a fat-fluid level is pathognomonic for a dermoid cyst. Teratomas originate from all three germ layer derivatives and show a mixture of CT densities ranging from fat and soft tissue to calcium. Dermoid cysts and cystic teratomas are on a continuum both histologically and in their CT features (Fig.lO.4a). Malignant teratoma is distinguished from benign forms by its ill-defined margins, large necrotic areas, and its compres-
sion and displacement of mediastinal vascular structures (Fig. 10.4b). Other malignant germ cell tumors are much rarer and present as solid masses that are more frequently lobulated than round. Seminomas characteristically are lobulated masses of homogeneous tissue density. This distinguishes them from other malignant germ cell tumors (chonocarcmoma, embryonal cell carcinoma, and mixed forms), which typically have a very heterogeneous appearance due to hemorrhage and necrosis but calcifications and fatty material are not seen. Criteria for malignancy are rapid growth, signs of local invasiveness, and the detection of distant metastases in lung, pleura, or bone. Invasion of mediastinal structures is difficult to assess, since a tumor may merely adhere to adjacent structures without invading them. The benign or malignant nature of the mass cannot be stated with certainty on CT criteria, and surgical removal is recommended. Larger tumors may displace or compress bronchial airways or the vena cava (10-30% of cases) . • In the case of extragonadal malignant germ cell tumors, the primary nature of the mediastinal mass is confirmed by noting the absence of gonadal and retroperitoneal tumor. • Residual masses of "mature" teratomatous tissue may persist following treatment, These masses may be cystic and, though benign, may enlarge over time.
Fig. 10.4 Cystic teratoma with an extensive solid component. causing posterior displacement of the mediastinal vessels (a). Large teratoma with extensive bone formation, a pleural effusion and a mediastinal shift to the left(b).
a
b
Hodgkin's Disease
Goiter
Parathyroid
Thyroid enlargement producing an intrathoracic mass may result from hyperplasia (iodine deficiency, hyperparathyroidism) or an inflammatory process (Hashimoto or Riedel thyroiditis) .
Orthotopic parathyroid glands are located posterior to the upper and lower poles of the thyroid gland. Ectopic adenomas may occur anywhere in the anterior upper mediastinum, and rarely also in the tracheoesophageal groove. MRI is often considered the imaging modality of choice owing to the distinctive signal characteristics of parathyroid tissue (high signal in T2weighted imaging with strong enhancement after contrast medium injection) but CT yields similar results in most cases.
•
CT Morphology
An intrathoracic goiter generally represents a downward extension of orthotopic thyroid tissue. Therefore, finding the pedicle connecting the intrathoracic mass with the thyroid gland will suggest the correct diagnosis. Rarely, the only connection is a narrow fibrous or vascular pedicle not visible on CT. Thyroid tissue enhances intensely after intravenous contrast administration (often by more than 40 HU). In 75-80% of cases, the intrathoracic goiter originates from the isthmus and lower poles of the thyroid and extends into the anterior mediastinum. In 20-25% of cases, the hyperplastic posterior portions of the thyroid gland pass lateral or posterior to the trachea or, very rarely, posterior to the esophagus, and extend into the posterior mediastinum (more common on the right side than the left). The intrathoracic thyroid mass commonly shows regressive changes with cystic elements, and circular, nodular, or eggshell patterns of calcification (Fig.W.5). Malignant transformation can be diagnosed only from secondary signs such as infiltrative changes and lymph node enlargement. The mass may cause significant displacement and narrowing of the trachea.
•
CT Morphology
Ectopic adenomas appear as soft-tissue attenuating or slightly hypoattenuating, round to oval masses of 0.5-3 cm diameter (mostly < 2 ern) in the thymic bed in the anterior upper mediastinum (see Fig.8.30b). Adenomas in the tracheoesophageal groove may be hard to differentiate from lymph nodes or veins on axial sections. Multiplanar reformations after contrast injection are helpful in these cases. Adenomas display a marked enhancement after contrast administration. Hodgkin's Disease
Hodgkin's disease (or Hodgkin's Lymphoma) has a bimodal age distribution, with peaks occurring at approximately 30 and 70 years of age. Most patients present with involvement of the cervical and supraclavicular lymph nodes. The prognosis depends on the histologic type and stage of the disease, and treatment is tailored to the extent of disease. CT is an essential part of tumor staging. •
Fig. 10.5 Goiter in the middle mediastinum. The mass shows regressive changes (cysts and calcifications) and has caused tracheal narrowing.
Adenoma
CT Morphology
Unlike non- Hodgkin lymphoma (NHL), Hodgkin's disease (HD) predominantly affects the anterior mediastinum and tends to spread from one group oflymph nodes to contiguous nodal groups. In the mediastinum, the prevascular and paratracheal lymph nodes of the anterior mediastinum are most commonly involved. Pulmonary involvement (approximately 10-"15%) is always associated with hilar lymphadenopathy. Concomitant enlargement of the thymus is common (30%). Pleural effusion and interstitial pulmonary edema may indicate venous and lymphogenous obstruction or direct tumor infiltration, while
384
10 Mediastinum, Pleura, and Chest Wall Fig. 10.6 (a) Hodgkin's disease in the anterior mediastinum. CT shows a large mass compressing the vena cava and invading the lung. (b) Hodgkin's lymphoma after therapy. There is a partially calcified residual mass in the region of the thymus.
a
b
pericardia! effusion usually signifies direct tumor invasion. In patients with pleural or pericardial involvement, the size of the anterior mediastinal mass usually exceeds 30% of the internal thoracic diameter (bulky disease, Fig. 10.6). Atelectasis may result from intrabronchial involvement (HD > NHL) or extrinsic bronchial compression (more common in children). Recurrences of Hodgkin's disease tend to occur at the original site after chemotherapy and outside the therapy port after irradiation. The absence of hilar adenopathy virtually excludes pulmonary involvement by Hodgkin's disease.
Non-Hodgkin's
Lymphoma
Non- Hodgkin lymphoma refers to a heterogeneous group of tumors with diverse histologi-
cal features. It is the third most common childhood malignancy after leukemia and eNS tumors butis unusual in children under 5 years of age. The adult form is most prevalent in the sixth and seventh decades. There is a 40 to 100 times greater risk for NHL in patients with immunodeficiency syndromes, after organ transplantation, and with collagen vascular diseases. Treatment depends on the histological classification. Frequently the disease is widespread at the time of diagnosis. •
CT Morphology
Non-Hodgkin lymphomas show a noncontiguous mode of spread with involvement of unusual lymph node groups. Involvement of the mammary and paracardiac nodes and isolated lymphomas in the posterior mediastinum are suggestive of non-Hodgkin lymphoma (Fig.lO. 7a). Extranodal disease is commoner than with Hodgkin's lymphoma (Figs.lO.7b, 10.8). Pulmonary lesions may appear as discrete
Fig. 10.7 (a) Extensive precardiac mass caused by non-Hodgkin lymphoma. (b) T-ALL with massive enlargement of the thymus and encasement of the superior vena cava (arrow).
a
b
Mesenchymal
a
c
b
Fig. 10.8 Non-Hodgkin's lymphoma in a 6-year-old child. Note the enlargement of the thymus (a), the axillary and supraclavicular lymph nodes (arrowheads). the
nodules, parenchymal infiltrates, glass pattern (see Chapter 9).
Tumors
or ground-
In up to 50% of cases, residual masses are seen after the successful treatment of malignant lymphomas. These masses are inactive and consist of sterile fibrous tissue. For distinguishing residual fibrotic masses from active tumor, CT relies on follow-ups showing lack of change in the size of the residual masses. An increase in size signifies an active lymphoma. MRI offers its own criteria of response to therapy: signal decreases on T2-weighted images with treatment, and residual fibrous masses showing low signals in Tl- and T2weighted images. However, so far MRI cannot reliably exclude focal vital lymphoma tissue within a residual mass.
invasion of the lower neck (b) and the stenosis of the trachea seen on a sagittal reformation of a low-dose (4 x 1 /6, CTDlvoi = 2.4 mGy) multislice CT data set (c).
•
CT shows a cystic mass, mostly unilocular, sometimes multilocular or septated, that molds to adjacent mediastinal vessels. The protein-rich content causes attenuation values of up to 40 HU. Unilateral or bilateral effusions may occur and may be chylous. Mesenchymal Tumors
Mesenchymal tumors may occur in all three mediastinal compartments. The benign forms most commonly occur in the anterior mediastinum, and the malignant forms (liposarcoma, fibrosarcoma) are commonest in the posterior mediastinum. •
Lymphangioma
Lymphangioma (cystic hygroma) is a rare congenital malformation of the lymphatic system consisting of lymph channels or cystic lymph spaces. In infants they are most commonly located in the anterior-superior mediastinum and frequently extend into the neck. In adults, lymphangiomas tend to occur in the lower anterior mediastinum and do not communicate with the neck.
CT Morphology
CT Morphology
Lipomas show characteristic fat attenuation on CT scans. They are soft tumors that model themselves to the mediastinal contours rather than displace adjacent organs. Other mesenchymal tumors such as fibroma, fibrosarcoma, and hemangiopericytoma can cause significant displacement. Hemangiomas are hypervascular masses that enhance intensely with intravenous contrast and often contain phleboliths.
385
10 Mediastinum, Pleura, and Chest Wall
Masses Predominantly
Involving the Middle Mediastinum
Lymphoma
Both Hodgkin's lymphoma and NHL may present as a mass in the middle mediastinum, although NHL is commoner than Hodgkin's disease at this location (see above). Tumors oftheTracheobronchial
System
See Chapter 9 (Lungs and Tracheobronchial tems, p. 303).
Sys-
Bronchogenic Cyst, Pericardial Cyst
Bronchogenic cysts are rare congenital anomalies of the tracheobronchial tree (see Chapter 9) that occur more frequently in the mediastinum than in the lungs. They have a fibrous capsule, frequently contain cartilage, are lined by respiratory epithelium, and contain mucoid material. Unlike intraparenchymal cysts they do not usually communicate with the bronchial tree. Pericardia! cysts are mesothelial anomalies (oLitpouching of the parietal pericardium without communication to the pericardial sac). MRI is often superior to CT in differentiating these cysts from mediastinal tumors. •
CT Morphology
Pericardia} cysts are most commonly found in the right cardiophrenic angle but may occur at any pericardial or pleuropericardial site (posterior cardiophrenic angle, superior pericardial recess). Most broncbogenic cysts are located adjacent to the carina on the right side (>50%). They may also occur adjacent to the trachea or esophageal wall or in the retrocardiac region (Fig.W.9). Both types of lesion typically range from 3 to Scm in size and display the features of a bland cyst with a round to oval shape, sharply defined margins, and homogeneous contents. The CT density is dose to that of water or may be higher, signifying mucus formation (pericardia! cyst: 20-50 HLJ; bronchogenic cyst: 0-50 HLJ). The lesions do not enhance after contrast administration.
Fig. 10.9 Bronchogenic cyst. The cyst. the upper border of which is below the carina, extends downward and to the right. terminating behind the left atrium.
Paraganglioma
Paragangliomas are neurogenic catecholamineproducing tumors that are malignant in 10% of cases. N on- functioning paragangliomas are called chemodectomas, functioning paragangliomas correspond to pheochromocytomas. Almost all chemodectomas are aortic body tumors. •
CT Morphology
Aortic body tumors are most often located at the angle of the ductus arteriosus, but can also occur lateral to the brachiocephalic artery, anterolateral to the aortic arch, or above and to the right of the right pulmonary artery. Pheochromocytomas are mostly found in the posterior mediastinum, in or adjacent to the heart or the pericardium, particularly in the wall of the left atrium or the interatrial septum. Paragangliomas are hypervascular and enhance intensely following intravenous contrast administration. Multicentric tumors are possible. Local invasion or metastases indicate malignancy (Fig.10 .10).
Neurogenic Tumors
a
b
c
Fig. 10.10 Malignant paraganglioma with encasement of the left coronary artery (4 x 1/6). There are two overlapping stents placed in the left main and left anterior descending coronary arteries (a). The tumor invades and displaces the pulmonary artery superiorly (b.
Masses
Predominantly
Involving
the Posterior
Neurogenic Tumors
•
With the exception of paragangliomas neurogenic tumors are found exclusively in the posterior mediastinum. Approximately 30% are malignant. Various histologic types are known (see Table 10.5). Neurogenic tumors can be divided into nerve sheath tumors (schwannoma, neurofibroma), ganglion cell tumors (malignant neuroblastoma and benign ganglioneuroma), and paraganglionic cell tumors.
Table 10.5
volume-rendered image). The extent of the tumor is excellently displayed on volume-rendered images of the heart (c).There is residual peripheral enhancement after radiation therapy indicating residual vital tumor tissue (arrowheads) .
Mediastinum
CT Morphology
CT typically shows a paravertebral mass (Fig.10.11) that is ellipsoidal or dumbbellshaped and may cause expansion of an intervertebral foramen. It is common to find sites of pressure erosion and deformation, usually involving the ribs. Foci of bone destruction without sclerotic borders indicate malignancy. Most neurogenic tumors have an oval shape and are of homogeneous soft -tissue attenuation.
Neurogenic tumors
Neuromas
Common
in neurofibromatosis,
show moderate
Neurofibroma
Frequently
expands
Schwannoma
Frequently
ellipsoidal
Paragangliomas
Very rare, show intense enhancement
the intervertebral
Chemodectoma
Frequently
near the aortic arch
Pheochromocytoma
Frequently
near the posterior
Sympathicoblastomas
Heterogeneous
Neuroblastoma
Malignant, age <8 years, or coarse calcifications
Ganglioneuroblastoma
Mixed form, age <10 years, traspinal component
contrast
enhancement
foramen
pericardium
pattern of enhancement large mass, usually metastasizes, dumbbell-shaped
contains
stippled
mass with an intra- and ex-
387
388
10 Mediastinum, Pleura, and Chest Wall Fig. 10.11 Neurofibroma with a typical dumbbell shape and expansion of the intervertebral foramen (a). Schwannoma presenting as a paravertebral mass compressing the adjacent lung (b). Multiple mediastinal masses in neurofibromatosis (c).
a
c
b
Some may show low attenuation areas that are due to lipid content, cystic degeneration, and entrapment of neural tissue. Central necrotic foci are rare. Except for paragangliomas that enhance strongly after contrast administration, most neurogenic tumors show moderate contrast enhancement and are virtually indistinguishable, based on their CT features. Relationship to vertebrae, ribs, and the spinal canal are essential for therapy planning and are best imaged by thin-section spiral and multislice CT, or by MRI. Extra medullary Hematopoiesis
Extramedullary hematopoiesis in the liver, spleen, and lymph nodes is common in various forms of anemia (sickle cell disease, spherocyto-
a
b
Fig. 10.12 Extramedullary hematopoiesis, manifested by paravertebral masses of soft-tissue density at the level of the lower thoracic spine (a). Coronal reforma-
sis, hemoglobinopathies) or bone marrow replacement disorders (leukemia, lymphoma, myelofibrosis). The collections of hematopoietic tissue may form extensive paravertebral masses, especially in patients with thalassemia or sickle cell disease. CT can differentiate the hematopoietic masses from malignant lesions. •
CT Morphology
Foci of extramedullary hematopoiesis typically appear on CT scans as convex paravertebral masses of soft-tissue density (Fig. 10.12). The masses usually extend downward for several vertebral segments below T7 and may have a subpleural component. Unlike neurogenic tumors, hematopoietic masses do not cause pressure erosion ofbone. Extramedullary hematopoiesis is accompanied by splenomegaly and skeletal changes due to bone marrow hyperplasia.
c tion (b) and volume-rendered image (4 x 1/6) of paraspinal masses due to extramedullary hematopoiesis in another patient (c).
Meningoceles and Neuroenteric Cysts Esophageal Tumors
See Chapter 15 (Gastrointestinal Tract).
be demonstrated by extending the examination to the upper abdomen. The azygos and hemiazygos veins are greatly dilated (equaling or surpassing the aortic diameter) and show homogeneous contrast enhancement (see Fig. 24.84).
Esophageal Diverticula and Duplication Cysts
See Chapter 15 (Gastrointestinal
Tract).
Esophageal Varices, Azygos Continuation
Esophageal varices in the lower half of the esophagus are caused by portal hypertension. Congenital or acquired occlusions ofthe inferior vena cava stimulate collateral circulation through the azygos venous system (azygos continuation). CT can positively distinguish the dilated veins from paravertebral masses.
Meningoceles and Neuroenteric Cysts
Cystic masses in the posterior mediastinum are very rare and may be caused by a meningocele, myelomeningocele, or neuroenteric cysts (which have pleural and alimentary components). Approximately 75% of intrathoracic meningoceles occur in association with neurofibromatosis. •
CT Morphology
CT demonstrates a well-circumscribed cystic mass in the posterior mediastinum. CSF -filled meningoceles may be unilateral or bilateral and • CT Morphology cause enlargement of the intervertebral foramina. Neurofibromatosis has a high associaEsophageal varices have a variable CT appearance depending on their size and the extent of tion with meningoceles as well as with solid neuthe underlying disease. Nonenhanced CT shows rofibromas, which can be distinguished by their nonspecific thickening of the esophageal wall CT attenuation. Thin-section spiral or multislice CT help detect the connection of the lesion with accompanied by nodular paraesophageal masses. On contrast-enhanced scans, the varices the subarachnoid space. MRI, however, is still the appear as tubular, tortuous structures that show technique of choice for this imaging task. Neuroenteric cysts are associated with bony marked pooling of contrast medium (Fig. 10.13). deformities of the thoracic spine (hemiverteEsophageal varices inside the esophageal lumen brae, butterfly vertebrae, scoliosis). Air incluhave to be distinguished from paraesophageal varices that are often more prominent and can- sions may be seen in neuroenteric cysts that not be detected by endoscopy. In patients with communicate with the esophagus. Differentiation is required from intramediunderlying liver disease, other venous collateral astinal pancreatic pseudocysts, which are rare vessels may be seen in the upper abdomen. In patients with azygos continuation, ab- (history), and from esophageal duplication sence or occlusion of the inferior vena cava can cysts.
Fig. 10.13 Varices appear as brightly enhancing tortuous veins in a submucosal location (a, esophageal varices, arrowheads) and adjacent to the esophageal wall (paraesophageal varices). The latter cannot be detected byendoscopy. Paraesophageal varices may present as a posterior mediastinal mass (b, volume-rendered image,4x2.5/15).
a
b
389
10 Mediastinum, Pleura, and Chest Wall
• For high specificity, a threshold of 15mm short axis diameter is suggested for subcarinal nodes and the azygos node, and 10 mm should be used for the other nodes. • Hilar (bronchopulmonary) lymph nodes should be regarded as suspicious if they bulge into the adjacent lung parenchyma (convex outer border; Fig. 10.15a). • With thin-section multislice CT, not only the cross-section but also the longitudinal shape of lymph nodes can be evaluated: round nodes should be considered suspicious even if they are between 5 and 10mm in size (Fig.10.15b). • The superior pericardial recess, which appears posterior to the ascending aorta, should not be misinterpreted as an enlarged lymph node (see Fig. 22.9). • Paracardiac and supradiaphragmatic lymph (Fig.10.15c) nodes are pathologic if their size exceeds 5 mm in diameter (except after previous upper abdominal surgery).
Lymph Node Enlargement
The anatomy and nomenclature of the mediastinal and hilar groups of lymph nodes are reviewed in Chapter 22 (see Fig. 22.2). CT is currently the technique of choice for evaluating the status of the mediastinal and hilar lymph nodes. MRI is not superior to CT for detecting enlarged nodes. In patients with extensive lymphoma, MRI or thin-section multislice CT provide images in any desired plane and are able to better define vascular structures. •
CT Morphology
Lymph nodes can be identified on CT when they are 2 to 5 mm in diameter. The differentiation between benign and malignant mediastinal lymph nodes is difficult (Fig. 10.14), because lymph node size is generally the only criterion available on CT scans. When a short axis nodal diameter of 1 cm is taken as the threshold, CT is approximately 60% sensitive and 70% specific in detecting malignant involvement. Similar outcomes are found for a long axis diameter of 15mm. A lower threshold will yield a higher sensitivity but at a reduced specificity. Neither CT nor MRI can provide a confident differentiation between benign and malignant nodes. Micrometastases are found in 5-15% of normal size lymph nodes, while 15-30% of moderately enlarged lymph nodes (10-15 mm) are free of metastases-an important possibility to consider in patients with accompanying pneumonia, COPD, or chronic heart failure.
a
b
Fig. 10.14 Various diseases can cause similar patterns of mediastinal lymph node enlargement: sarcoidosis (a), lymphoma (b), metastases from non-small cell lung
•
Differential Diagnosis
(Table 10.6) Between 80% and 90% of patients with sarcoidosis develop mediastinal lymphadenopathy (Fig. 1O.14a). The most frequent inflammatory causes of mediastinal lymph node enlargement are fungal infections and tuberculosis. Tuberculosis most often shows a characteristic peripheral enhancement of lymph nodes after contrast administration (Fig. 10.16a). Necrotic areas usually have a CT attenuation of below 30 HLJ.
c
d
cancer (c). Lymph node enlargement due to chronic left heart failure is more prominent in the perihilar region (d).
Lymph Node Enlargement
b
a
c
Fig. 10.15 Morphologic indicators of malignant lymph node involvement, (a) Bronchopulmonary nodes with a convex contour relative to the lungs suggest malignancy (arrow) while lymph nodes with a straight or concave interface to the lungs are very likely of benign eti-
Table 10.6
Differential
diagnosis
of mediastinal
ology. (b) Rounded shape of lymph nodes (long and short axis diameters nearly identical in all planes) makes malignant involvement more likely, (c) Supradiaphragmatic lymph nodes> 5 mm short axis diameter are considered pathologic.
lymph nodes (according
Calcified lymph nodes
Peripherally enhancing necrotic lymph nodes
Common:
Tuberculosis Sarcoidosis Silicosis Posttreatment Metastases (Hodgkin's disease, thyroid carcinoma)
Tuberculosis Fungal infection Lymphoma Metastases (seminoma, bronchial carcinoma)
Rare:
Histoplasmosis Amyloidosis Scleroderma Castleman's disease Metastases (mucinous carcinoma, osteosarcoma)
Most mediastinal lymph nodes are hypoattenuating relative to the mediastinal vessels when scanned with the start delay of20-40 s that is typical for most mediastinal CT examinations. At a later time some lymph nodes become almost isoattenuating to the vessels (indicating a relative hyperemiaduetoinflammationortumor), others remain hypoattenuating (indicating hypovascularization due to regressive changes or hypovasculartumors). Such a behavior is seen when the chest is scanned at a later phase, e.g., after having examined the upper abdomen first. There is no evidence yet as to whether the contrast enhancement patterns may be used for differentia-
to Naidich, Zerhouni, or
Hypervascular
and Siegelman) lymph nodes
Metastases (renal cell carcinoma, bronchial carcinoma, carcinoid tumor)
Sarcoidosis Castleman's disease Kaposi sarcoma
tion between benignormalignantdisease. Lymph nodes that already enhance after 20-30 s, however, are almost invariably malignant (Fig. W.1Gb; Table 10.5).Thepresence ofcalcific ations (Fig. W.1Gc) may help with the differential diagnosis only with additional clinical findings. Castleman disease (angiofollicular lymph node hyperplasia or giant lymph node hyperplasia) is characterized by well-defined masses in the mediastinum or proximal hilum, that are sometimes calcified. Mul ticentric forms present as multiple enlarged lymph nodes and have a poorer prognosis. Striking uniform contrast enhancement suggests the diagnosis.
391
10 Mediastinum, Pleura, and Chest Wall
b
a
Fig. 10.16 The etiology of mediastinal lymph nodes can be suspected from CT morphology only in conjunction with clinical data. (a) Rim enhancement of mediastinal lymph nodes in tuberculosis in a patient with
c low-grade fever, (b) Bilateral hilar and infracranial hypervascular lymph node metastases in a patient with Ewing's sarcoma, (c) Calcifications in pathologic lymph nodes in a patient with colorectal carcinoma.
Pancoast Tumor
ning, and coronal as well as sagittal reformations are mandatory for optimum evaluation of tumor extent (Fig.10.17b). When neurological symptoms are present, it should be possible to demonstrate a mass of soft-tissue attenuation lying adjacent to the vertebrae or in the supraclavicular fossa. Obliteration of the fat plate between the anterior and middle scalene muscles or lateral asymmetry of the scalene muscles • CT Morphology should raise suspicion of brachial plexus invaPancoast tumors show a characteristic location sion. Multislice CT can define the nerves only in in the superior pulmonary sulcus (Fig. 10.17). areas where there is sufficient perineural fat. They have a strong propensity for invading the Plexus evaluation is still the domain of MRI but ribs, vertebral column, esophagus, trachea, bra- multislice CT with sagittal reformations holds chiocephalic vessels, and cervical soft tissues. A the promise of similarly good results. thin-section technique should be used for scanPancoast tumor (superior sulcus tumor) is a bronchial carcinoma arising in the apex of the lung. The typical clinical manifestations are caused by invasion of the brachial plexus (shoulder and arm pain) or the sympathetic chain (Homer's syndrome). If multislice CT is not available, MRI is the technique of choice for images in the coronal and sagittal plane.
Fig. 10.17 (a) Pancoast tumor showing broad pleural adhesion and invasion of the mediastinal pleura (arrows). Scans at a higher level revealed encasement of the subclavian vessels, (b) Coronal MPR (4 x 1/6) of a small Pancoast tumor infiltrating the second rib (arrow).
a
b
Tumors Involving the Chest Wall
Tumors Involving the Chest Wall
Tumors of the chest wall are usually metastatic in origin. Most such tumors in adults are adenocarcinomas, while most childhood lesions are neuroblastomas or leukemic infiltrates. In patients with peripheral bronchogenic carcinoma, chest wall invasion marks a tumor as irresectable only if it is diffuse and associated with a malignant pleural effusion. Involvement of the chest wall by lymphoma (1.5%) generally arises from a mediastinal mass lesion. Primary tumors of the chest wall are rare (Table 10.7). Because of its superior soft-tissue contrast, MRI is often superior to CT in evaluating invasion of the soft tissues of the chest wall. •
Table 10.7 Primary the chest wall
tumors
and tumor-like
lesions
of
Cartilage or bone lesions Chondroma, chondrosarcoma Plasmacytoma Ewing's sarcoma Paget's disease Fibrous dysplasia Soft-tissue tumors Lipoma, liposarcoma Fibroma, fibrosarcoma, Desmoid tumors Rhabdomyosarcoma Hemangioma Neuroma
MFH
CT Morphology
Soft-tissue tumors cause a mass in the muscles of the chest wall with obliteration ofthe fat planes or the epipleural fat stripe. Usually these tumors are clearly delineated only on contrast-enhanced scans. Tumors of the bony structures cause local expansion of the bone or rib cartilage and may have extensive soft-tissue components. Invasion ofthe chest wall by a mediastinal or pulmonary tumor can be recognized by the obliteration of fat planes and by contrast enhancement. Loss ofthe narrow epipleural fat stripe is associated with chest wall invasion in 80% of patients with peripheral bronchogenic carci-
Intrapulmonary with pleural i
noma. Broad contact wi th the chest wall (> 3 ern) and an obtuse angle between the tumor and chest wall are not considered reliable indicators of invasion (Fig. 10.18). Unchanged position of a tumor on inspiratory and expiratory scans suggests adhesion to or invasion ofthe chest wall. Thickening ofepipleural (fatty) tissue at the site of contact between a lesion and the chest wall indicates pleural adhesion and a more long-standing process, although an initial chest wall invasion cannot be ruled out (Fig. 10.19).
I
Pleural
Intrapulmo
Fig. 10.18 Schematic comparison of the CT appearance of pleural lesions, peripleural intrapulmonary lesions, and lesions of the chest wall. The arrows indicate the "acute" pleural angles that are associated with intrapulmonary lesions.
Extra I with rit
··Ision
393
10 Mediastinum, Pleura, and Chest Wall
Fig. 10.19 Peripheral bronchogenic carcinoma with adhesion to the chest wall and mediastinal lymph node metastases (4 x 1/6). Note that there is a local thickening of the extrapleural soft tissue (arrow) in the region of contact, which has a distinctly different contrast enhancement pattern than the tumor. The arrowhead indicates encasement and infiltration of the upper lobe artery.
Pleural Tumors Benign Pleural Tumors
Approximately 70% of benign pleural tumors (benign mesothelioma, lipoma, fibroma) arise from the visceral pleura and only 30% arise from the parietal pleura. On the whole, benign pleural tumors are very rare. Most of them are fibrous tumors. •
CT Morphology
Benign pleural tumors form circumscribed nodular pleural thickenings (Fig. 10.20) that may reach considerable size and may even show locally invasive growth. They do not metastasize, however, and they have a good prognosis. Involvement of the mediastinal pleura can mimic a mediastinal tumor, and involvement of the interlobar fissure can mimic an intrapulmonary
tumor. Lipomas show fat density on CT. Benign fibrous tumors of the pleura t'benign mesotheliomas} most often arise from the visceral pleura and may be located within a fissure. They commonly form acute angles with the pleura with tiny amounts of adjacent pleural fluid mimicking focal pleural thickening. A relatively thin stalk may be present but usually cannot be seen on CT. Tumors tend to be hypervascular; non-homogeneities may be present only in large lesions. Tumors located within a chronic pleural empyema may be difficult to define. Pleural Metastases
Pleural metastases most commonly originate from the lung, breast, and ovary and are gen-
Fig. 10.20 Benign fibrous tumor of the pleura (4 x 1/6). The lesion is hard to evaluate on axial sections (a) but displays typical features (acute angle relative to the pleura) on sagittal reformations
(b). a
b
Pleural Lymphoma Fig. 10.21 (a) Pleural metastases from gastric carcinoma appear as fine nodular thickening that enhance after intravenous contrast administration. They also involve the mediastinal pleura, (b) Pleural metastases from a mucinous ovarian tumor appear as diffuse pleural nodularities with peripheral enhancement and hypoattenuating central areas containing mucinous material.
a
erally associated with malignant effusions. The diagnosis is established by the cytological analysis of aspirated pleural fluid. CT is useful for detecting solid tumor components. •
CT Morphology
CT typically demonstrates multiple foci ofnodular pleural thickening that show intense contrast enhancement (Fig. 10.21a). Involvement of the mediastinal pleura is highly suspicious for neoplastic disease (Fig.10.21b). Diffuse pleural thickening with contrast enhancement may occur but is entirely nonspecific. If no pleural calcifications are present, pleural metastases are morphologically indistinguishable from malignant mesothelioma by their CT features alone and only clinical features like a history of asbestos exposure may give clues to the diagnosis. Malignant Mesothelioma
The incidence of malignant mesothelioma is 300 times higher in asbestos-exposed patients than in the normal population. The latent period is 20 to 40 years, and the prognosis is poor. The main clinical symptoms are chest pain and shortness of breath. CT can define the extent of advanced tumors but is less reliable in patients with asbestosis and very early tumor stages. A definitive diagnosis requires core-biopsy. •
CT Morphology
All portions of the pleura may be involved. The tumor may cause extensive nodular thickening of the pleura or complete encasement of the
b
lung (Fig. 10.22). The tumor tissue usually shows heterogeneous enhancement on postcontrast scans. A hemorrhagic pleural effusion (>30HLJ) is often present. Advanced cases may present with pericardial effusion, rib destruction, contralateral involvement, thickening of the interlobar septa, and intrapulmonary metastases. Mesothelioma is distinguished from fibrothorax by the more irregular pleural thickening, which often shows a more intense contrast enhancement. In early cases the differentiation may be difficult (Fig. 10.22a). Pleural Lymphoma
Both Hodgkin' disease and non-Hodgkin lymphoma may involve the pleura. Pleural involvement is rarely the only initial manifestation of the disease. It occurs more frequently in conjunction with mediastinal nodes, thymic involvement, or parenchyma! pulmonary disease, or is due to recurrence of the lymphoma. A pleural effusion (30% patients with Hodgkin's disease) caused by lymphatic obstruction of mediastinal lymph nodes has to be differentiated from lymphomatous involvement of the pleural itself. Pleural involvement in patients with lymphoma alters radiation therapy and unrecognized disease increases the risk for treatment failure. Leukemia may also cause pleural thickening.
395
10 Mediastinum, Pleura, and Chest Wall
a
c
b
Fig. 10.22 Malignant mesothelioma. (a) CT initially demonstrated irregular posterior and mediastinal pleural thickening with calcified plaques that do not show significant enhancement after contrast administration. This resembles the findings seen in chronic effusion with incipient pleural thickening, (b) Scan two
months later shows marked progression of the neoplastic pleural thickening, which now is polynodular. Note the involvement of the mediastinal pleura, (c) Involvement of the major (arrowheads) and minor fissure (arrow) in another patient with extensive pleural mesothelioma (4 x 1/6, sagittal MPR).
Superior Vena Cava Syndrome
Superior vena cava syndrome may result from intravascular or extravascular obstruction of the superior vena cava. Approximately 80% of extrinsic obstructions are caused by malignant disease (bronchial carcinoma, primary mediastinal tumors, lymphoma), and only 20% have a benign cause (fibrosing mediastinitis, radiationinduced fibrosis, catheter-induced thrombosis). CT is useful for establishing the cause of the obstruction and, in acute cases, planning radiotherapy or vascular intervention.
•
CT Morphology
Postcontrast scans can differentiate extravascular causes (compression) from intravascular causes (thrombosis). The development of a collateral supply is manifested by dilatation of the azygos vein, superior intercostal veins, and small cutaneous veins that drain into the azygos-hemiazygos venous system, into pericardiophrenic veins, or rarely into the portal veins (Fig. 10.23).
Fig. 10.23 Superior vena cava syndrome with collateralization. Downhill varices in the esophagus (a). Collaterals via intercostal, pericardiophrenic, azygos and hemiazygos veins (b) seen on a lateral MIP(4X1/6).
a
b
Fibrosing Mediastinitis
Diffuse Mediastinal
Diseases
Acute Mediastinitis
In more than 90% of cases, acute mediastinitis is caused by an esophageal perforation that may be post-traumatic, spontaneous (Boerhaave syndrome), iatrogenic (after surgery or endoscopy), or neoplastic. The remaining 10% of cases are due to complications following sternotomy or an inflammatory process that has spread from the neck along fascial planes (visceral and posterior cervical space). Mediastinitis has a high mortality if untreated. Management include antibiotics, drainage or surgical debridement. •
CT Morphology
In mediastinitis there is a streaky or diffuse increase in CT attenuation in the mediastinal fat (Fig.lO .24a). Complete obliteration of normal mediastinal fat by material of water to softtissue density indicates a severe mediastinitis with edema and inflammatory infiltrates. Abscesses present as well-defined fluid collections with CT attenuation up to 30HL] that mayor may not have an hyperattenuating rim and may contain gas collections. Aspiration may be needed to determine the infection. Pleural effusion is common. Sternal osteomyelitis is characterized by demineralization and permeative destruction of the bone (see Fig. 23.58). It is not necessarily accompanied by increased attenuation or abscess formation in the surrounding fat and soft tissues. Following a median sternotomy, these findings are almost always associated with a
sternal dehiscence of more than 2 mm. Destruction of the sternum by malignant neoplasms is often accompanied by soft -tissue masses (e.g., in multiple myeloma or metastases from lung, breast, prostate, or renal cell carcinoma). Air and fluid collections may be seen for up to 20 days postoperatively, and rarely for up to 50 days, without signifying an inflammatory process. If follow-up examinations show new or enlarging fluid or air collections, the presence of an inflammatory process is highly probable (Fig.lO.24b). Normal CT findings exclude mediastinitis. False-positive CT diagnoses may occur.
Fibrosing Mediastinitis
A chronic granulomatous or fibrosing (sclerosing) mediastinitis may result from an infectious disease (tuberculosis, histoplasmosis, fungal infections, syphilis), a mediastinal hematoma, radiation therapy, or drug therapy (methysergide). Fibrosing mediastinitis is often associated with other fibrotic processes such as pseudotumors of the orbit, Riedel's struma, or retroperitoneal fibrosis (Ormond's disease). •
CT Morphology
The clinical and radiological features of fibrosing mediastinitis vary according to the location and the extent of the fibrosis. Postinfectious forms are marked by a self-perpetuating medi-
Fig. 10.24 Acute mediastinitis with diffuse increase in attenuation of the mediastinal fat in a patient after neck surgery (a). Mediastinal air and haziness of the mediastinal fat in a patient after liver surgery
(b). a
b
10 Mediastinum, Pleura, and Chest Wall
astinal fibrosis that may cause very little stranding or increase in attenuation of the perivascular fatty tissues, and may be difficult to identify. In other cases, however, the fibrosis forms solid tumor-like masses or ring-like zones of increased attenuation around the mediastinal structures. Approximately 10% of cases show obstruction of the vena cava (Fig. 10.25) or the central tracheobronchial system. Extensive calcifications may occur. Mediastinal Lipomatosis
The excessive deposition of mediastinal fat may occur as a nonpathologic process in obese individuals and in patients on cortisone therapy. CT can easily establish the cause of the mediastinal widening that is detected on chest radiographs. •
Fig. 10.25 CT in fibrosing mediastinitis (Ormond's disease) shows an indistinct peripheral increase in paratracheal and paracaval density and narrowing of the superior vena cava (arrow) by fibrotic encasement, leading to a superior vena caval syndrome. Note the intense opacification of the azygos vein due to the collateral circulation (arrowheads).
CT Morphology
CT shows a diffuse increase in mediastinal fat that is most pronounced in the upper mediastinum. Inhomogeneities, as opposed to uniform fat density, indicate a different etiology
such as fibrosis, hemorrhage, acute or chronic inflammation, tumor invasion, or postoperative or postirradiation changes.
Diffuse Pleural Diseases Pleural Effusion, Hemothorax, Chylothorax
Pleural effusion may consist oftransudate or exudate, depending on the protein content of the fluid. Chylothorax refers to an effusion of lymphatic fluid, hemothorax to an effusion that contains blood. Pneumonic infiltrates that border the pleura incite an parapneumonic effusion that progresses from an exudative to a fibrinopurulent stage as it becomes organized. This type of effusion is on a continuum with pleural empyema. A tuberculous pleural effusion results from the perforation of a subpleural pulmonary focus into the pleural space and usually develops 3 to 6 months after the primary infection. Malignant effusions are also exudative and are not" always associated with pleural thickening. Thus needle aspiration is often indicated. Hemorrhagic effusions usually have a neoplastic or traumatic etiology. The most frequent causes
of Chylothorax are lymphomas (> 50%) and trauma (25%). Effusions are a common associated finding in thoracic CT examinations but are not a primary indication for CT. •
CT Morphology
Pleural effusions with a volume of 15 ml or more can be detected with CT. Pleuritis sicca is not visible on CT scans. Free pleural effusions occur in the posterobasal region and form a crescentshaped collection that abuts the chest wall. Effusions that are loculated or encapsulated by fibrous septa may occur anywhere within the pleural cavity (Fig. 10.26). Fibrous septa or fibrin threads cannot be visualized with CT. Interlobar effusions have a typical biconvex shape and can present as a pseudotumor on chest radiographs. An air-fluid level can result from a bronchopleural fistula or prior therapeutic measures (surgery, aspiration biopsy, drainage).
Pleural Empyema
A cbylothorax has a CT density of approximately 0 HU and is usually quite extensive. Transudates are homogeneous and have nearwater attenuation « 15HLJ). Neither the pleura nor the effusion enhance with intravenous contrast. Attenuation values higher than 20 HU indicate an exudative, hemorrhagic, or purulent effusion. The effusion may enhance on late postcontrast scans due to diffusion of contrast medium into the effusion. There are borderline cases in which transudate and exudate are indistinguishable byCT. Withafi brinopurulenteffusion, the thickened pleura shows contrast enhancement in the absence of empyema (reactive pleura! thickening). It is common at this stage to find septations and local encapsulated collections (Fig. 10.26). Increased density and widening of the subpleural fat stripe signifies a long-standing process. In tuberculosis, pleural calcifications and extensive areas of pleural fibrosis are a common finding (d. Fig.10.28a). High CT numbers (>30HU) and heterogeneous attenuation are consistent with hemotborax. Sedimentation effects are seen if the blood in the pleural space does not coagulate. Pleural effusions are very commonly associated with compression atelectasis of adjacent lung areas. Compressed lung tissue floating in a free effusion appears on CT as a tag-like structure (see Fig. 9.13), a small portion of which may be fixed to the chest wall by adhesions. Multiple chest wall adhesions lead to corresponding taglike deformities of the atelectatic lung due to fibrous bands and indicate a septated effusion. Repositioning the patient (lateral decubitus) can be hel pful in distinguishing between free and encapsulated effusions, although ultrasound is the technique of choice for this indication. With a free effusion, atelectatic portions of the lower lobe in the posterior diaphragmatic recess may be mistaken for the diaphragm because of their curved shape, and anterior portions of the effusion may be misinterpreted as subphrenic fluid. A pleural effusion lies posterior (superior) to the diaphragm and extends medially adjacent to the spine. Ascites lies anterior (inferior) to the diaphragm and does not extend medially.
Fig. 10.26 borders.
Loculated
pleural effusion
with scalloped
Pleural Empyema
Pleural empyema is on a continuum with parapneumonic exudate (fibrinopurulent effusion). Most empyemas are accompanied by bronchopneumonic infiltrates (mostly Pneumococcus, and Staphylococcus pyogenes). Tuberculous pleural effusions progress to pleural empyema in approximately 20% of cases. CT can detect areas suspicious for empyema and can distinguish pleural empyema from a peripheral lung abscess. •
CT Morphology
The transition from exudative effusions to pleural empyema can be appreciated on CT scans. Empyema is associated with more intense contrast enhancement and greater thickening of the visceral and parietal pleural layers than exudates (the "split pleura sign", Fig.10.27a). With increasing duration of the empyema, there is increased thickening of the extrapleural fat stripe. Diffuse pleural thickening and calcification may be the end result after healing of a pleural empyema. Air inclusions (bubbles) signify infection by gas- forming bacteria or rarely an esophagopleural fistula, while an air-fluid level results from a bronchopleural fistula or previous needle procedure. Criteria for differentiating between pleural empyema and a subpleural lung abscess (Fig. 10.27 b) are listed in Table 10.8.
399
10 Mediastinum, Pleura, and Chest Wall Fig. 10.27 (a) Pleural empyema with thickened, markedly enhancing visceral (arrow) and parietal pleura ("split pleura sign"), (b) Peripheral lung abscess with an airfluid level. Note the acute pleural angles (arrows) compared with the pleural process in (a), confirming the intrapulmonary origin of the lesion.
a
b
Table 10.8 Features that distinguish pleural empyema from lung abscess Empyema
Peripheral lung abscess
Oval or crescent shape
Round or oval shape
Obtuse angle with chest wall
Acute angle with chest wall
Consistently associated with pleural thickening
Not consistently associated with pleural thickening
No air in lesion wall
Air in lesion wall
Bronchopleural fistula
Communicates with bronchus
Compresses contiguous lung
Distorts contiguous lung
Changes shape with position changes
Does not change shape with position changes
Common features: central air inclusions (depending on pathogen), peripheral contrast enhancement
Pleural Fibrosis, Asbestos-Associated Changes
Asbestos-associated pleural changes (pleural plaques) may not appear until several decades after the initial asbestos exposure. Approximately 3% of asbestos-exposed individuals develop an exudative pleural effusion during the first 10 years after exposure. The effusion may remain asymptomatic or it may cause pain and persist for more than a year. Many patients subsequently develop rounded atelectasis and diffuse pleural fibrosis with restrictive impairment of pulmonary function. Pleural mesothelioma, bronchial carcinoma, and tuberculosis show an increased incidence in asbestosexposed individuals. •
CT Morphology
Pleural plaques most commonly occur on the subcostal pleura and may be 1 to 15mm thick. Thickening of the pleura may be diffuse al-
though more commonly there is focal plaque formation. Plaques typically appear as sharpedged elevations ("mesa" sign) and tend to form calcifications that start at the center of the plaque and are therefore separated from the ribs by an non calcified zone (Fig.10.28b). The plaques spare the lung apices and costophrenic angles. Visceral pleural involvement, particularly within the fissures, and mediastinal pleural involvement are uncommon, and should raise the suspicion of a mesothelioma. Plaques may be associated with fibrosis of the subpleural fat, subpleural pulmonary fibrosis, and the formation of parenchymal bands and adjacent nodular opacities compatible with areas of early rounded atelectasis ("crows' feet" or "rounded atelectasis equivalents").
Pneumothorax Fig. 10.28 Pleural calcifications. (a) Extensive flocculent calcifications in pleuritis calcarea following tuberculosis, (b) Typical asbestosrelated pleural plaque, appearing as an isolated elevation ("mesa" sign) with central calcification (arrow). Note also the posterior pleural thickening.
a
The attachments of the serratus and external oblique muscles, epipleural fat, and intercostal vessels should not be mistaken for pleural plaques.
b
The term asbestosis is reserved for the pulmonary parenchyma! fibrosis caused by asbestos exposure and should not be used for asbestos-related pleural abnormalities.
Trauma and Postoperative Changes
CT is indicated in chest trauma for the evaluation of suspected injury to mediastinal structures, particularly the great vessels. CT angiography is a sensitive, noninvasive procedure that can replace arterial angiography for the diagnosis of aortic rupture (see p.915). In a postoperative setting CT is a problem-solving tool.
•
CT Morphology
CT can detect pneumothorax with high sensitivity by delineating the visceral pleura from the air- filled pleural space. In patients with recurrent pneumothorax, CT can be used to search for causative subpleural bullae, which may be congenital or may form in the setting of obstructive or fibrosing lung diseases (Fig. 10.29).
Pneumothorax
The presence of a pneumothorax is rarely a primary indication for CT. CT is used to evaluate recurrent spontaneous pneumothorax or insufficient pleural drainage (faulty tube placement, frequently in major fissures or in the lung parenchyma). It is also useful for differentiating a pneumatocele or a large bulla from a pneumothorax. Identification of even a small pneumothorax can be important in trauma patients about to undergo mechanical ventilation or general anesthesia for surgery. Therefore it appears useful always to evaluate the most cephalad scans of an abdominal CT examination in a trauma patient in lung window settings to identify otherwise occult pneumothorax.
Fig. 10.29 Patient with spontaneous pneumothorax. A coronal volume-rendered image (4x 1/6) demonstrates small apical bullae due to paraseptal emphy-
401
10 Mediastinum, Pleura, and Chest Wall
Mediastinal Hematoma
Mediastinal hematomas may result from central venous puncture, chest trauma, the spontaneous rupture of an aortic aneurysm, aortic dissection, or a coagulation disorder. If aortic injury is suspected, CT angiography should be performed as an adjunct to non contrast CT (see Chapter 24). •
CT Morphology
A fresh mediastinal
hemorrhage initially produces streak-like densities in the mediastinal fat. Only larger hematomas contain areas of increased attenuation. The approximate location of the bleeding site can be inferred from the site of the hematoma. Contrast extravasation into a fresh hemorrhage is rare and is usually an indication for immediate surgery. Bleeding into the pleural space signifies injury to the lung (Fig. 10.30) or intercostal arteries or an open or confined rupture of the aorta (ascending aorta on the right side, aortic arch or descending aorta on the left side). A (hemorrhagic) pericardia! effusion poses a risk of pericardial tamponade, especially if CT reveals a concomitant dissection of the ascending aorta. Rupture of the brachiocephalic veins due to blunt chest trauma leads to hematoma formation in the upper anterior mediastinum. Traumatic aortic aneurysms occur in up to 95% at the level of the ligamentum arteriosum and are associated with left-sided hematomas in the posterior mediastinum. Older hematomas show decreasing attenuation on CT scans and may have soft-tissue or water attenuation, making them difficult to distinguish from a solid mediastinal tumor (hematomas enhance by less than WHU after intravenous contrast administration).
Fig. 10.30 Massive chest trauma with injury of the right lung and acute intrapleural hemorrhage demonstrated after contrast injection.
surgery, intubation, or endoscopy). Mediastinal emphysema may be an indirect sign that a neoplasm has eroded the esophagus or tracheobronchial tree. Retroperitoneal and intraperitoneal free air can also track into the mediastinum. Mediastinal emphysema is rarely a primary indication for CT scanning. •
CT Morphology
CT can sensitively detect even the smallest mediastinal air collections, but the cause can be determined only when the CT scans are interpreted within the context of clinical findings (Fig. 10.31).
Mediastinal Emphysema
Mediastinal emphysema, known also as pneumomediastinum, usually signifies a tracheobronchial injury or subpleural alveolar rupture with air tracking along the pulmonary interstitium and lung root into the mediastinum. Possible causes are a sudden intra-alveolar pressure rise due to strenuous physical effort or mechanical ventilation (Macklin effect), bronchial rupture, esophageal rupture (Boerhaave syndrome), or therapeutic procedures (prior
A Fig. 10.31 Pneumomediastinum due to iatrogenic injury of the left main bronchus during difficult intubation. Soft tissue emphysema after insertion of a pleural drain.
Diaphragmatic
Rupture
~ a
b
Fig. 10.32 Traumatic diaphragmatic rupture on the left side with herniation of stomach and colon into the thoracic cavity, accompanied by extensive right-sided
pulmonary contusions. Note the exposed stomach wall and pneumothorax on the left side (5/8/4).
Air within the pulmonary interstitium associated with a pneumomediastinum is a sign of alveolar rupture. A traumatic rupture of the trachea or large bronchi occurs within 2.5cm of the carina in 85% of patients and leads to mediastinal emphysema. With rupture of a bronchus distal to the insertion of the pulmonary ligament a pneumothorax will form as well. A rupture is presumed to exist if a large pneumothorax persists despite adequate drainage, if the lung is completely collapsed, or if the lung collapses toward the chest wall rather than toward the hilum ("fallen lung" sign). Direct CT visualization of the rupture is rarely achieved and usually requires spiral or multislice scanning with thin slices (see Chapter 9).
angle. Tears of the central portion of the diaphragm are most difficult to evaluate. Multislice CT strongly improves evaluation of the diaphragm. Reformatted images in the coronal and sagittal plane from spiral and mul tislice CT data sets are extremely helpful to establish the diagnosis (Fig. 10.32). Indirect signs of a diaphragmatic tear are an apparent elevation of the diaphragm with pleural effusion, deformation of the hepatic or splenic contour by herniating viscera, and an unusually thin border separating the stomach, small intestine, or colon from the adjacent lung (lack of diaphragmatic coverage). In addition, thickening of the posterior portions of the diaphragm on the side of the rupture has been reported. This sign, however, must be distinguished from a relative hypertrophy of the diaphragm on the contralateral side, which may be due to phrenic nerve palsy.
Diaphragmatic Rupture
Diaphragmatic rupture is commoner on the left side than on the right. Visceral herniation into the chest is rarely seen in acute cases. The radiological diagnosis of diaphragmatic rupture is a difficult challenge, regardless of the modality used. Even with multiple imaging modalities, a definitive diagnosis must await operation in approximately 50% of cases. •
CT Morphology
A diaphragmatic tear can be directly visualized only on thin CT slices that cut the posterior part of the diaphragm almost at a perpendicular
Cardiac Surgery
CT can be used in postoperative cardiac patients to investigate a new occurrence of mediastinal widening, confirm or exclude suspected postoperative hemorrhage, or search for a focus of infection. •
CT Morphology
During the initial weeks after surgery, it may be difficult to distinguish infection-related changes from changes due to surgical manipulation.
403
10 Mediastinum, Pleura, and Chest Wall
However, a hypoattenuating mass that shows marked rim enhancement after contrast administration is suspicious for an abscess. Presternal or retrosternal edema during this period has no pathologic significance. It is manifested by stranding and increased density of the fat in the middle and anterior mediastinum. Fluid collections may be serous (lD20 HU). bloody (40-80 HU), or of mixed composition. Mediastinal and pericardial air inclusions may occur initially along with a small pericardial effusion. Enlarged mediastinal lymph nodes are a common finding. The abeve changes gradually resolve during the first three or four postoperative weeks and rarely persist for up to 50 days. In patients who have undergone a sternotomy, the osteotomy site should be fully reunioned, although small discontinuities are normal. A gap up to 2 mm wide may still be considered normal, but the risk of sternal instability increases with the width of the residual gap (see Fig. 23.58). Aortic pseudoaneuIJsms are a rare late complication « 1 % at 1 to 2 years after surgery) resulting from dehiscence of the suture line due to an acute or chronic infection. Typical sites of occurrence are the site of an aortotomy (for connection to the heart-lung machine) or an anastomosis between the aorta and a bypass (see Fig. 24.48). Constrictive pericarditis is another rare late complication (0.2% at months to years after surgery) manifested on CT by 3 to 20mm of pericardial thickening with manifestations of right heart failure: caval dilatation, flattening of the right ventricle with curvature of the septum to the left, pleural effusion, and ascites. Pericardial calcifications are rare in postoperative cases but occur in 50% of cases with other etiology (see Fig.23.52b).
Heart Transplantation Acute or chronic rejection after heart transplantation requires biopsy confirmation and is not an indication for CT. In cases of infection, which are responsible for up to 40% of postoperative mortality in heart transplant recipients, CT is a sensitive modality for the detection and localization of the focus (see Chapter 9). The findings are nonspecific, however, and cannot identify
the causative organism (most frequent problem organisms: gram-negative bacteria, viruses, fungi, protozoans). Long-term complications include lymphoma, carcinoma, and coronary sclerosis. CT Morphology Scanning of anastomotic sites in the ascending aorta or right atrium may reveal a constriction or an abrupt change in caliber. The lumen of the inferior vena cava is frequently enlarged. Unusual separation may be noted between the superior vena cava of the recipient and the ascending aorta of the donor or between the recipient pulmonary artery and the donor aorta. Pericardial effusion is frequently encapsulated and may persist for a period of weeks or months. Tricuspid insufficiency is a post-transplant complication that can be suspected from CT if an extreme dilatation of the right atrium including the valvular ring is present (see Fig. 23.57). Steroid therapy can lead to mediastinal lipomatosis and spinal osteoporosis, which is frequently associated with vertebral body compression.
i i 1lit, Liver M. Prokop, A. J. van der Molen
Anatomy Segmental Anatomy (p. 408) Vascular anatomy and variants Liver Attenuation (p. 411) Examination
(p. 409)
Technique
Patient Preparation (p. 412) Aquisition Technique (p.412) Intravenous Contrast Injection (p. 416) Intra-arterial Contrast Injection Techniques Perfusion CT(p.420)
(p. 419)
Enhancement Characteristics Hepatic Arterial Phase (HAP) (p.422) Portal Venous Phase (with CT AP) (p. 423) Portal Venous Phase (after i.v. Contrast Administration) Interstitial Phase (p. 424) Hepatic Excretion Phase (p. 425) Vascular Opacification (p. 425) Cystic Liver Lesions
(p. 423)
...........................••.......
Developmental (Simple) Hepatic Cyst (p. 425) Polycystic Liver Disease (p. 427) Biliary Hamartomas (p. 427) Mesenchymal Hamartoma (p. 427) Cystic Metastases (p. 428) Benign Tumors and Tumor-like Lesions Incidentally Found Lesions (p. 428) Focal Liver Steatosis (p. 428) Hemangioma (p. 430) Focal Nodular Hyperplasia (FNH) (p. 432) Hepatocellular Adenoma (HCA) (p. 434) Lipoma (p.435) Angiomyolipoma (p. 435) Inflammatory Pseudotumor (p. 435) Infantile Hemangioendothelioma (IHE) (p. 436) Hepatic Alveolar Echinococcosis (HAE) (p. 436)
.
Malignant
Tumors
Metastases (p. 437) Hepatocellular Carcinoma (HCe) (p. 441) Fibrolamellar Carcinoma (FLe) (p. 446) Cholangiocellular Carcinoma (CCC) (p. 446) Hepatoblastoma (p. 448) Undifferentiated Embryonal Sarcoma (p. 448) Epithelioid Hemangioendothelioma (EHE) (p. 449) Angiosarcoma (p. 450) Other Mesenchymal Tumors (p. 450) Lymphoma (p. 450) Post-transplant Lymphoproliferative Disease (PTLD) (p. 452) Diffuse liver Disease Fatty Infiltration (p. 452) Cirrhosis (p. 453) Primary Biliary Cirrhosis (p. 455) Hemochromatosis (p. 455) Radiation injury (p. 456) Miscellaneous diffuse liver diseases (p. 456) Infectious
Disorders
Hepatitis (p. 457) Abscesses (p. 457) Hydatid Disease (p. 459) Liver Tuberculosis (p. 459) Rare Infectious Diseases (p. 460) Vascular Disorders and Perfusion Abnormalities Perfusion Abnormalities (p. 461) Portal Hypertension (p. 462) Portal Venous Obstruction (p. 463) Hepatic Venous Outflow Obstruction (p. 465) Hepatic Arterial Thrombosis and Hepatic Infarction Rendu-Osler-Weber Disease (p. 467) Peliosis Hepatis (p. 469)
(p. 467)
Trauma Blunt Liver Trauma (p. 469) Penetrating Abdominal Trauma
(p.471)
Postoperative and Post-therapeutic Changes Liver Resection (p. 471) Liver Transplantation (p. 472) Transjugular Intrahepatic Portosystemic Minimal Invasive Therapies (p. 473)
Shunt (TIPS) (p.473)
11
The use of scintigraphy. such as " "Tc-sulfur colloid or IDA, is rapidly declining as these have been replaced by multiphasic spiral CT or MRI. Sulfur colloid scans can be used for focal fatty infiltration, regenerative nodules, and in characterizing primary tumors like focal nodular hyperplasia. Tc-red blood cell scan with SPECT is still a good technique for hemangioma. New techniques like 'In-octreotide imaging may have a niche in the work-up of neuroendocrine tumors, especially gastrinomas, or carcinoid tumors. The role for PET in clinical oncology is also increasing steadily, but the applications in the liver are still scarce. Image fusion of SPECT and PET data with CT data sets is still quite cumbersome but there are combined CTPET or CT-SPECT scanners-some even with multislice technology-becoming available that elegantly solve the problem of co-registration of data sets. While these scanners can provide excellent results, but still suffer from their substantial costs.
Computed tomography is a standard technique for imaging of the liver. It is mainly used for the evaluation of tumors but offers useful information also for a vast range of other diseases (Table 11.1). Ultrasound offers an economical. noninvasive alternative formany indications. but further evaluation by CT or MRI is frequently required in patients prior to surgery. Contrast-enhanced ultrasound may become a competitor in the diagnosis of focal lesions. but this technique is still difficult to perform for coverage of the entire liver. As compared to MRI. CT offers a consistently good image quality and requires only minimal patient cooperation. Complex invasive procedures such as CT hepatic arteriography (CTHA) or CT during arterial portography (CTAP) had been justified only if they had essential therapeutic implications (assessment of resect ability). At present. they are substituted for by biphasic multislice CT and dynamic Cd-enhanced MRI or MRI with liver-specific contrast media.
Table 11.1 Tumor
Indications
for CT examination
diagnosis:
of the
Indeterminate
liver
hapatic
lesion:
Differentiation lesion
Primary hepatic tumors: Hepatic
metastases:
Complications
" After surgery:
(hematoma,
resectability
Detection,
resectability
bilioma,
abscess)
Portal vein thrombosis Arterial anatomy Osiers disease
Other:
Quantitative
(before
intervention/surgery)
Abscess Trauma measurements:
Hemosiderosis (iron content) Liver and tumor volume
benign-malignant,
characterization
Detection,
Malperfusion Vascular:
The Liver
Anatomy
With a volume of 1400 to 1700ml, the liver is the largest parenchymal organ in the body. On average, it measures 13.5cm in its largest craniocauclal dimension. The size of a particular liver varies with the height and gender of the patient. Hepatomegaly is present in >80% of patients when the liver exceeds 15.5cm in its craniocaudal extent, or the volume of the liver is 2000 ml or more. The normal size ratio between the right and left hepatic lobes is 3 :2. •
Segmental
Anatomy
Knowledge of hepatic segmental anatomy is helpful to communicate CT findings to the surgeon. However, numerous approaches use the same names for different structures. The numbering system for the segmental anatomy is identical for most of these approaches. All approaches are only approximations of reality because there are marked variations in the individual anatomy of almost every patient. The liver is divided into four segments by the planes defined by the inferior vena cava (lVC) and the right, middle, and left liver veins. These define the anterior and posterior segment of the right hepatic lobe, and the medial and lateral segment of the left hepatic lobe. The right and left lobe are separated by the plane defined by the IVC, the middle liver vein, and the gallbladder. These vertically oriented segments are divided into subsegments by the horizontally oriented portal vein. In the most usual surgical classification (Bismuth) in North America, the portal venous Table 11.2
Segmental classification
Anatomic subsegment
Caudate lobe Left lateral superior subsegment Left lateral inferior subsegment Left medial subsegment Right Right Right Right
anterior inferior subsegment anterior superior subsegment posterior inferior subsegment posterior superior subsegment
Couninaud's
supply of the liver defines eight subsegments of the liver. The boundaries of these segments are grossly defined by the hepatic veins (in the craniocaudal direction) and by the plane of the portal vein (horizontal on the right, oblique on the left) (Fig. 11.1). Other numbering systems, such as the one by Couinaud, are more common in Europe. Table 11.2 gives an overview of the most important terminologies. Segment 1is the caudate lobe. Subsegments 11 and III are the portions of the left lobe that are 10-
Fig. 11.1 Bismuth classification of hepatic segments. The left, middle, and right hepatic veins divide the liver into four segments (Couinaud: sectors), which are further subdivided into cranial and caudal subsegments (Couinaud: segments) by the plane of the portal vein. Viewed from the front, the numbering of the (sub)segments proceeds in a clockwise direction (except for segment I and subsegment IVa).
of the liver (Soyer, 1993) Nomenclature Couinaud,
Bismuth,
Goldsmith and Woodburne,
1957
1982
1957
IV V VIII
III IV alb V VIII
Caudate lobe Left lateral segment Left lateral segment Left medial segment Right anterior segment Right anterior segment
VI VII
VI VII
Right posterior segment Right posterior segment
hemiliver, sector, and segment correspond to Bismuth's lobe, segment, and subsegment
Anatomy
cated to the left of the falciform ligament. Segment IV is the medial portion of the left hepatic lobe (the" quadrate lobe ") that is located between the falciform ligament and the vena cava-gallbladder line/middle hepatic vein. In the Bismuth classification segment IV is subdivided into subsegments IVa cranially and IVb caudally. The right hepatic lobe is divided into subsegments V through VlI 1, with a clockwise numbering system starting from the subsegment V inferiomedially next to the gallbladder (Figs. "11.1, 11.2). Common anatomic variants include a hypoplasia of segment IV b and a large left hepatic lobe that fills the left dome of the diaphragm. Complex congenital heart diseases may be associated with a symmetrical, centralized liver
(Fig. 11.3). A left-sided liver is part ofa complete or partial situs inversus. •
Vascular Anatomy and Variants
The common hepatic artery normally originates from the celiac trunk. After giving rise to the gastroduodenal artery behind the pylorus, it becomes the proper hepatic artery that passes through the hepatoduodenal ligament anteromedial to the portal vein and anterior to the common bile duct. In the liver hilum it bifurcates into the right (RHA) and left hepatic arteries (LHA). The middle hepatic artery that supplies segment IV originates with equal frequency from the right or left hepatic artery.
VII~ VII./.l
a
b
(
d
Fig. 11.2 Location of the hepatic segments on CT sections. In scans through the cranial portion of the liver, the segmental boundaries are defined by the vena cava and the main branches of the hepatic veins. At more caudal levels the cava-gallbladder line and the falciform
ligament define the central and left segmental planes. Often the plane of the portal vein can only be estimated. In case of doubt coronal reformations are helpful.
409
11 The Liver
patic ligament and reaches the left lobe via the fissure of the ligamentum venosum (>20% of all patients). Axial CT sections demonstrate such a vessel above the liver hil urn and posterior to the left lobe (Fig. T1.4a). Accessory and replaced arteries can be distinguished by searching for a normal left hepatic artery originating from the proper hepatic artery at the liver hil urn. Variations of the right hepatic artery are almost as common. They all present as an artery in the hepatoduodenal ligament posterior to the portal vein (Fig.l1.4b). By tracing this artery back to its origin, one can distinguish between a RHA arising from the aorta, the superior mesenFig. 11.3 Symmetric liver as an associated anomaly in a teric artery, or one directly from the celiac patient with Kartagener syndrome (immotile cilia synartery. By following it in both directions, it is drome) and incomplete situs inversus. Also note the just as easy to distinguish between an accessory right-sided location of the stomach and an aplasia of the or replaced artery, and to determine whether it intrahepatic segment of the inferior vena cava with azygos continuation. is just an aberrant RHA, or a common hepatic artery, or even the celiac artery that is displayed. Apart from the hepatic artery, there can be Variations in the arterial blood supply of other sources of arterial supply to the liver, which of periphthe liver occur in some 45% of individuals and are important in the vascularization are important for planning surgical or radiologieral hypervascular tumors (parasitic supply) and cal interventions. Various classifications have when the hepatic artery is occluded. The most been proposed. The one by Michels is most important is the right inferior phrenic artery, popular but is not very precise. Therefore, it is but also the left inferior phrenic, internal mammary, left gastric, or intercostal arteries may more advisable to properly describe variations give offbranches to the liver surface (Fig. "11.5). than to try to classify them. Aberrant arteries The main portal vein is formed by the concan either be a replaced (variation in origin) or accessory (additional vessel). Variants that in- fluence of the superior mesenteric vein and the splenic vein and runs posteriorly in the hepavolve more than one artery are common. The commonest variations involve an aber- toduodenal ligament. At the porta hepatis it branches in the right (RPV) and left portal veins rant left hepatic artery that runs in the gastrohe-
Fig. 11.4 The following patterns suggest an aberrant hepatic vascular supply, (a) An aberrant left hepatic artery (almost always from the left gastric artery) is present if there is artery entering the liver above the hilum and anterior to the caudate lobe. (b) An aberrant
right hepatic artery (most commonly from the superior mesenteric artery) is present ifthere is an artery supplying the liver posterior to the portal vein. Further analysis is required to determine the origin of the vessels and to distinguish accessory from displaced arteries.
Anatomy Fig. 11.5 Aberrant vascular supply to the liver in a patient with occlusion of the hepatic artery after embolization of a metastatic gastrinoma. Note the arterial revascularization (arrows) of the thrombosed of the portal vein (a) and collateralization (arrowheads) via phrenic artery branches (b).
a
(LPV). Most commonly, the RPV further divides in anterior and posterior segmental branches, while the LPV divides in superior and inferior segmental branches. Anatomical variants of the MPV (trifurcation, posterior from MPV, anterior from LPV) occur in up to 20% of patients, and variants of segmental branches are present in almost every individual. The anterior branch of the RPV is more constant than the posterior branch. The course of the main LPV almost always remains constant. There are usually three hepatic veins that course in the vertical fissures that separate the liver segments. The right hepatic vein (RHV) drains segments VI -VIII into the IVC. The middle hepatic vein (MHV) drains segments V and IV B, while the left hepatic vein (LHV) drains segments II -IV A. The MHV and LHV often form a common trunk before draining into the IVC. The caudate lobe (segment I) drains separately into the IVC via small veins at a more inferior level. This is the reason why it often hypertrophies in Budd-Chiari syndrome (occlusion of the large hepatic veins). There are numerous variations in the drainage pattern of the several segments to the hepatic veins. Other common variants include accessory right, middle, or left hepatic veins, and an accessory inferior RHV that may be important for atypical liver surgery because it separately drains segments V and VI. There are m ul tiple branching patterns ofthe LHV. •
b
liver attenuation decreases approximately 15 HU for each 10% increase in the proportion of fat. When the fat content reaches 10-15%, the intrahepatic vessels become isoattenuating to the parenchyma and cannot be seen on unenhanced scans (Fig.1l.6b). A further increase in fat content reverses the display of vascular structures, and the intrahepatic vessels appear hyperattenuating to the liver tissue on noncontrast scans (Fig. 11.6a) . In patients with hemochromatosis or hemosiderosis, liver attenuation increases in proportion to the iron load (Fig.11.6d). However, this attenuation increase depends strongly on the effective energy of the x-ray beam and varies with the tube voltage, scanner type, and patient diameter. A concomitant fatty infiltration of the liver may yield a falsely low CT attenuation. Table 11.3 lists other, rarer causes of increased liver attenuation. Table 11.3
of
increased
liver
attenuation
Iron:
Primary hemochromatosis Transfusion hemosiderosis Nutritional hemosiderosis
Copper:
Wilson's disease
Iodine:
Amiodarone
Gold:
Therapeutic thritis
Thorotrast:
Thorium-containing contrast medium (diffuse reticular distribution pattern)
Thallium:
Accidental
Liver Attenuation
Normal liver parenchyma has a CT attenuation of 55 to 65HU (Fig.H.Gc). This slightly higher attenuation as compared to other soft tissues is thought to be due to the glycogen content of the liver. When diffuse fatty infiltration is present,
Causes
(>70HU)
Arsenic: Acute massive
(antiarrhythmic
drug)
use in rheumatoid
or suicidal ingestion
Chronic poisoning protein storage
Glycogen storage diseases
ar-
411
77 The Liver
c
b
a
Fig. 11.6 CT densityofthe hepatic parenchyma: normal (c), hemochromatosis (d).
Examination
d
fatty liver (a, with hepatic metastases),
mild fatty infiltration (b),
Technique
CT scanning of the liver is usually a part of an upper abdominal or complete abdominal examination. With few exceptions (exclusion of hemorrhage, hemochromatosis), every CT of the liver should use intravenous contrast material (Table 11.4). A power injector is required. Table I1.S lists the scanning techniques that we recommend for specific clinical problems. Biphasic multislice CT and various MRI techniques increasingly substitute for procedures that require intra-arterial contrast application such as CT hepatic arteriography (CTHA) or CT during arterial portography (CTAP). Patient Preparation
Except in dire emergencies or CTAP studies, oral contrast medium should be administered. For upper abdominal studies, SOO-600ml of oral contrast medium given 30 minutes prior to the scan is sufficient, while whole abdominal examinations require administration of 1000IS00ml of oral contrast medium over 60-90 minutes before the examination. In trauma cases, SOOml given via nasogastric tube is beneficial, and extra 2S0ml portions may be added if the waiting time for CT is prolonged. Positive contrast material is still standard, but negative contrast material such as water or methylcellulose preparations are becoming in-
creasingly important for arterial phase liver examinations as part of a biphasic or multiphasic upper abdominal study, especially when multislice scanning is employed. Such negative contrast has the advantage that there is no superimposition of bowel and vessels for CTA, and there is a better display of the bowel walls. Acquisition Technique
Spiral scanning is the technique of choice for liver examinations, independent of the mode of contrast material application (Tables 11.4, 11.S). Spiral CT is the prerequisite for scanning the whole liver within the arterial phase of contrast enhancement, a phase that was inaccessible with older techniques such as dynamic incremental CT. Trade-offs and Parameter Optimization
Lesion detection in the liver depends on signalto-noise ratios and thus on the size of the lesion, its contrast, the amount of partial volume averaging, and the image noise. Large lesions usually do not pose a problem as long as there is an attenuation difference to normal liver parenchyma. Small lesions, however, suffer from partial volume effects that may further reduce
Acquisition Technique Table 11.4
Suggested parameters for examination
413
of the liver
General Oral contrast
Liver only: 0,51, administered 30 min prior to examination Abdomen: 1-1.51, administered 60 min prior to examination
Patient position
Supine with arms elevated
Scan range
Liver only: from diaphragm to inferior hepatic border Abdomen: from diaphragm to symphysis
Breath hold phase
Inspiration (Spiral-CT)
Windowing
Noncontrast or delayed CT: After I. v. contrast: CTHA, CTAP: Lipiodol-CT:
W/L=200/40 W/L=350/60 W/L= 500/100 (adjust as needed) W/L=300/40
Scan parameters
1-slice SCfTF/RI
Standard, precontrast
7/12/
Volumetric imaging
5/8/4
Perfusion CT
7-10/0
€ontrast injection
l-slice V/F/D
4-slice V/Ii/Ii)
16-slice 'if./F/f!)d
Abdominal screening
120/2/60
120 + 50/3/60
120 + 5012/205
Portal venous CT
150/4/70
150 + 50/5/60
150 + 50/5/205
Biphasic CT
150/4/25 + 70'
150 + 50/5/25 + 70'
150 + 50/5/10Aa 80 c.
Early delayed phase
3-5 min
3-5min
3-5 min
Late delayed phase
10-15 min
10-15 min
10-15min
Perfusion CT
50/5/10
50 + 50/5/10
50+ 50/5/10
CTHA
50/2/5 La.9
50/2/5 La.9
CTAP
120/2/30 La. 9
120/2/30 + 60
Lipiodol CT
5 ml Lipiodol i.a. h
5 ml Lipiodol i.a. h
Portal venous CT
150/4/70
150 + 50/5/60
4-slice' SC
t i
2-3
16-slice' SC
i
1-1.25
1-1.5
t
i
5-7.5/4-5
0.75-1.25
t
3/3 cor'
4/3
16 x 1.25/2/0'
4 x 5/0'
10-20/-
II f
50/2/15 i.a. 9 120/2/30 + 60 9
9
5 ml Lipiodol i.a. h 150 + 50/5/205
sc = slice collimation (mm), TF= table feed (rnrnjrotanon). RI = recon. increment (mm), T ~ scanning direction, width (mm), MPR = multiplanar reformats. axial = axial sections, cor coronal, sag = sagittal eM = contrast material, Contrast concentration = 300 mg/ml iodine (optional 400 mg/ml for 16-slice scanning) V= volume of eM (ml), N =volume of saline flush (ml), F" flow rate (mils). D = start delay, X = trigger region a Pitch P = $Cf(NxTF) should be chosen close to 1.5 with a-slice scanners. 1.3- 1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI"1-1.5/0.7 { single section (perfusion) mode with no table increment, scans every 2-5 s dread: 205 (lOA) '" start delay 20 s (lOs) after reaching 50 HU enhancement in the spleen (desc. aorta) e read: biphasic scan (arterial + portal phase) after 25 s (lOA) for the pi for the scan, 70-80s for the 2"d scan =::
SW= effective section
=::
f 9
h
when 400 rnqfml is used for 16-slice scanning, the contrast volume can be reduced to 120 ml injection into proper heaptic artery (CTHA) or superior mesenteric/splenic artery (crAP); prior injection of vasodilator (optional); reduced iodine concentration of 150 mqjml (optional) examination performed approximately 14 days after Upiodol injection inot the proper heaptic artery
their contrast to surrounding low contrast will make such a ceptible to image noise: lesions noise and no longer be visible.
parenchyma. A lesion more susmay "drown" in For this reason,
there are three ways of improving lesion detection (and possibly also characterization): reduce partial volume effects, improve contrast. and limit image noise.
414
11 The Liver Table 11.5
Which CTtechnique
Screening for metastases hypovascular hypervascular Prior to resection Lesion of unknown
dignity
for which clinical indication? Phase after i. v. contrast
injection
i.a. contrast
application
Precontrast
arterial
portal
delayed
GAP
lipiodol
(+)
++
++ ++ ++
(+)
++
++
+
(+)
++
++
+
Suspected
HCC
(+)
++
++
+
Suspected
CCC
(+)
(+)
++
+
Post transplantation
(+)
++
++
Trauma
+
(+)
++
The recommended techniques are printed in blue. The arterial phase is always combined with the portal venous phase (biphasic well suited + ~ may provide additional (+) ~ may provide additional - ~ not recommended
G
++
spiral CT).
++ ~
information information
in selected
cases
Reduction oj partial volwlle averaging relies on the choice of a narrow section width and overlapping image reconstruction. Overlapping reconstruction increases the number of images to be viewed. documented. networked. and archived. There are two other important tradeoffs: thin-section scanning reduces the coverage per time. and it increases image noise substantially. The longer time that is required to cover the liver may become problematic for arterial phase imaging. Because there is only a narrow window for purely arterial phase imaging (see below). there will be varying degrees of overlap between the arterial and the portal phase of contrast enhancement during a single spiral acquisition. Timing. therefore. becomes crucial. In addition. the longer scan duration with thinsection imaging will increase the likelihood of breathing artifacts during the scan. Thinner sections also increase image noise. A reduction of section width by a factor of four from Srnrn to 2 mill will lead to a two-fold ( 4) increase in image noise. Lesion detection therefore becomes dose-dependent. To take full advantage of the increased lesion detection with thin sections. one has to substantially increase patient dose or use other means of noise reduction. This is underlined by a recent study that shows that evaluation of 1 mm mulrislice CT images no longer improves lesion detection but actually reduces it as compared to 2-4mm sections at a dose of 15mGy (CTDlvol). At a dose of
20 mGy. another study shows that the number of detected lesions can be substantially increased with 2.5 mrn section width as compared to 5 rnm or more. Characterization becomes a problem with small lesions. Here. the thinner section width is better because it improves morphologic evaluation. Delayed scans help differentiate between small cysts and tumors. Still. a substantial number of lesions of less than 8 mm diameter will remain undiscovered. It has been shown that the vast majority of such small lesions are benign. even in tumor patients. At the same time. the larger lesions seen in a patient with metastatic disease are only the tip of the iceberg. Depending on the primary. 1.6 to 4 times more metastases smaller than 1 ern have been shown to be present for every lesion> 1 ern. There is no consensus yet as to which section width and radiation dose is appropriate for which indication. Given the results mentioned above. however. we suggest reconstructing a moderate section width (S-7.5mm) at a reduced patient dose (CTDlvol= 6-10 mGy) for benign diseases or indeterminate lesions. a moderate section width (S-7.5mm) and moderate patient dose (8-12mGy) for known malignancies. and a narrow section width (2-4mm) and high dose (12-20 mGy) only for those patients in whom major surgery is planned. Dependent on body diameter in more obese patients. the emlvol has to be increased by a fac-
Acquisition
tor of 2-4 (see Chapter 5). In the near future. automated exposure control will become available. which will make it much easier to individualize the required patient dose. In addition. contrast material application should be optimized so that high liver-to-lesion contrast is guaranteed.
Single-Slice
CT
The main limiting factor for single-slice CT is coverage. To optimize coverage and section width. a pitch factor> 1.5 should be used. For most situations. 5 mm collirnation and a table feed of 8 mm (5/8/4 protocol) is a good solution. For improved scanning speed in the arterial phase. the table feed can be increased to 10 mm. If the spatial resolution has to be optimized, a 3/512 scanning protocol should be considered if a subsecond scanner is available. A table feed of 6mm is also possible but image quality of 3D reconstructions is visibly lower. Such a protocol can be important for planning surgery, but the protocol suffers from increased noise and therefore cannot be recommended for obese patients. In very obese individuals, the section collirnation may even be increased to 7 mm to obtain a s ufficie nt signal-to-noise ratio. Overlapping image reconstruction is necessary to take full advantage ofthe spiral CTacquisition. The number oflesions increases substantially when images are reconstructed in an overlapping fashion. Ideally, there should be 50-60% overlap with respect to the effective section thickness (section width) but in clinical practice, a somewhat lower overlap may suffice.
Multislice
CT
Multislice CT has similar trade-offs as singleslice spiral CT. It can be optimized either to near-isotropic resolution or to more homogeneous liver enhancement. Unless extremely rapid scanning of the liver is necessary « 5 s), a 5mm collirnation should not be used, even with 4-slice scanners This gives the user the choice between a 0.75-1.5 mm collirnation for nearisotropic resolution and a 2.5-4 mm collirnation for rapid scanning. For optimum speed, a pitch P > 1.35 should be used with either technique. A test bolus injection or bolus triggering becomes
Technique
mandatory with contrast injection to avoid missing the optimum perfusion phase. H
With
4-slice Scanners
a narrow collirnation of 1-1.25 mm {volumetric imaging}, multislice CT provides excellent multi planar reformations that are especially helpful before hepatic surgery (e.g., Fig. 11.19). This technique allows for excellent evaluation of the arterial and portal venous anatomy and the relationship of tumors to the liver hilum, the hepatic vessels, and the bile ducts. lt also yields the bestCTangiographic results for the arterial as well as the portal venous supply. However, contrast enhancement of the liver and vascular structures changes over time during arterial phase scanning; the most cranial sections will represent a relatively early arterial phase while the caudal sections will represent a late arterial/early portal venous phase. This problem is less pronounced in portal venous imaging because the portal phase lasts for a longer period. With a collirnation of 4x2.5 mm or even 4x5 mm (fast spiral scanning), the liver can be scanned in some 8-12 s or 4-6 s, respectively. This makes it possible to perform scans in an early and late hepatic arterial phase (see below). Contrast enhancement of the liver and vessels is more homogenous but spatial resolution of multiplanar reformations is lower. In patients with hepatomegaly, higher pitch factors (P" up to 8) or thicker collirnation (e.g., 4x3.75 mm) may be used to cover the liver in a sufficiently short time. As mentioned above, we suggest reconstructing a moderate section width (5-7.5 mm) for benign diseases, indeterminate lesions, or follow-up of known malignancies, and a narrow section width (2-4 mm) for patients before major surgery. CE scanners at present are limited as to which section widths can be reconstructed for a specific collirnation setting. This may require the use of 4x2.5 mm collirnation for the first group of indications if one wants to reconstruct axial 5mm sections (maximum section width with 4xl.25 mm collirnation is 2.5mm). Alternatively, a 4x1.25 mm protocol can be used and 5mm thick MPR can be reformatted from the resulting secondary raw data set of overlapping thin axial sections. Other scanners can use the thin-section protocol and reconstruct images of arbitrary section width (even 7 mm or above).
11 The Liver
In obese patients, thin-section imaging will lead to suboptimum results because of excessive noise. Such patients should be scanned using at least a 4x2.5mm protocol (4x3 mm or 4 x 3.75 mm, depending on the scanner type, will be better) and a section width of7-7.5 mm. In very obese patients, the collimation and section width may be increased even more. 8- and 16-slice Scanners
Newer multislice scanners with 8 or 16 detector rows allow for covering the whole upper abdomen in less than lOs, even with 1-1.5 mm section collimation. This makes it possible to use such a thin collimation and a pitch P >-1.35 as the standard for liver imaging. Thinner collimation (0.5-0.75 mm) is rarely necessary for the liver unless a detailed analysis of the vasculature or biliary system is required. With such thin collimation, higher pitch factors (P , 1.5) should be used. Like with 4-slice scanners, the reconstructed section width (either from the original raw data or an secondary raw data set of overlapping thin sections) depends on the clinical indication, the exposure dose and the size of the patient. Thick coronal and sagittal MPR (3-5 mm Width). especially when aligned parallel and perpendicular to the portal vein, are helpful for the evaluation of the porta hepatis and for the planning of partial liver resection.
Intravenous Contrast Injection While non-contrast CT is easily performed even on non-spiral CT scanners, contrast-enhanced CT of the whole liver was limited to scanning in the portal venous or later phases because the arterial phase of enhancement was too transient for capturing by conventional scanners. Spiral scanning allowed for rapid acquisition of the data volume within one breath hold, but arterial phase imaging has become a standard technique for the liver only with the possibility of quick repeat scanning of the same volume (biphasic CT). With multislice scanners, even more rapid data acquisition becomes possible and an early and late arterial phase of enhancement can be captured separately. Timing of contrast application is a crucial factor for optimum results.
Noncontrast
CT
Noncontrast CT has limited indications in modern practice. It is generally indispensable for evaluating suspected hemorrhage in trauma and is the best phase for documenting confluent fibrosis in cirrhosis, calcifications, and hemochromatosis. Most often however, it is used in the evaluation of hyper vascular liver tumors as part ofa multiphasic approach. Some of these tumors can best be detected on non contrast CT. We especially recommend them for patients with carcinoid or other endocrine tumors because lesion size can be assessed to better advantage. For hypovascular lesions, such as metastases of colorectal cancer, noncontrast CT is not required. Because of the low contrast of many liver lesions, noncontrast CT requires sufficiently high mAs settings depending on the size of the patient (CTOIvol 'IOmGy) and narrow ("liver") window settings for image documentation (150250/40). Blood vessels and lesions generally appear hypoattenuating compared to normal liver parenchyma. If differentiation between blood vessels and liver parenchyma is not possible because of slight fatty infiltration, noncontrast CT is not helpful for identifying lesions. Portal
Phase CT
For most indications in which the liver is just part ofa general work-up, scanning in the portal venous phase is considered sufficient. This is also the case in patients with known malignancies that cause hypovasculcir metastases, such as colorectal cancer. In a pre-surgical work-up, however, biphasic spiral scans should be preferred. For spiral CT in the portal venous phase, 100 to 200ml of contrast medium is administered by power injector (at 3-5 mlls). For optimum lesion-to-liver contrast, liver enhancement should be as high as possible. Peak enhancement increases with larger volumes of contrast material. Best results require injection of 150 ml or more, but there is a strong dependence on patient weight. At present, there is a trend to individualize contrast material injection, with 1.5-2 ml contrast material per kg body weight being a good compromise. There is some contradictory evidence in the literature as to whether flow rates influence he-
Intravenous Contrast Injection
patic parenchymal enhancement. Work by Chambers et a1. (1994) suggests that using injection rates of 3 mlls lead to an earlier and higher hepatic enhancement than rates of 2 mlls. Model calculations by Bae et a1. (1998) could show that peak enhancement is almost independent of flow rates and only depends on the total amount of iodine injected. As a consequence, the volume of contrast material probably cannot be reduced, even for 16-row multislice CT scanning. The duration of the phase of optimum liver enhancement increases with a biphasic injection protocol. The optimum parameters for a biphasic injection vary from patient to patient (because of different cardiac output and variations in splanchnic perfusion) but a combination of 50 ml contrast at 5 mlls, followed by 100mI at 3 mlls yields good results. With rapid scanning using either spiral or multislice technique, however, there is little advantage to such a protocol. As a rule of the thumb, a relative plateau of parenchymal enhancement is reached some 20 seconds after the end of contrast injection (Fig. 11.7), with peak enhancement occurring about lOs later. This enhancement plateau lasts for more than 30s although parenchymal enhancement may vary by up to 15HLJ during this phase. Thus, scanning is initiated about 50 to 100 seconds after the start of the injection depending on the volume and flow rate of the contrast medium. Because there may be substantial variations between patients, results are superior
Enhancement(H
E)
3001-
250
Aorta
200 150
Portal vein ~
100
~Live~veins_
50 a
------~ _ Liver parenchyma
30
40
50
60
70
80
90 Time (5)
Fig. 11.7 Schematic representation of vascular and hepatic enhancement after injection of 150 ml of contrast material at 4 mils.
when the scan delay is tailored to the individual circulation time (bolus triggering). Bolus triggering in the portal venous phase may either use a region of interest in the liver, spleen, or aorta, or rely on enhancement of the liver veins as a visual clue. Portal venous triggering is not always easy because the maximum enhancement of the liver parenchyma may vary strongly between patients, and there are variations in the time delay between start of aortic and optimum portal venous enhancement. Some authors have suggested using an ROI in the liver (enhancement of 50 HU as a trigger level) for starting the scan but more than 30% of patients may never reach such a level. We also suggest a trigger level of 50 HU enhancement, but use either the spleen (15-20 s delay for the portal venous phase, 30-40 s for the hepatic venous phase, see below) or the aorta (40-45 s for the portal venous phase, 55-65 s for the hepatic venous phase). For longer scan duration (e.g., with singleslice CT), we recommend scanning in the caudocranial direction because it delays imaging of the hepatic veins until they are more fully opacified. •
Biphasic Spiral CT
Biphasic spiral CT includes scanning in the arterial phase of contrast enhancement in addition to the portal venous phase. A biphasic examination is the standard technique for planning partial liver resections or whenever hypervascular lesions are suspected (see Tables 11.5, Table 11.6). Some authors advocate additional noncontrast scans in patients with hypervascular metastases but the number of additionally detected lesions is generally small « 5%). Biphasic CT is often sufficient to characterize focal liver lesions (see Figs. 11.12, 11.14-11.17),but a third phase after contrast injection increases the confidence oflesion characterization. Biphasic spiral CT uses similar scanning parameters to monophasic examination, except that scans are performed during both the arterial phase (start delay about 25 seconds) and the portal phase of enhancement (start delay about 60-90 seconds), in each case using a short scan duration. High flow rates improve the conspicuity of hypervascular lesions and good results are obtained with uniphasic flow rates of 4-5 mlls, although flow rates as high as 8 mlls have been advocated and yield excellent
11 The Liver
lesion-to-liver contrast (see Fig. 11.30). A secure venous access route is necessary for such high flow rates to avoid extravasation. Biphasic contrast injection protocols with a flow rate of 5mlls over 10-15 s and a lower flow rate of 3mlls over 20-25 s improve both arterial enhancement and the duration and homogeneity of the portal venous phase. High dose techniques using 150-200ml of a 300-370 mg IIml contrast agent injected at 5mlls have shown very promising results. comparable with spiral CTAP in detection of liver lesions. but with improved characterization. Bolus triggering further optimizes contrast enhancement of hepatic lesions. The trigger level chosen should not be at the top of the scan range because diaphragmatic movements may cause considerable artifacts in the aorta. The aorta at the level of the liver hilum is more suitable for triggering the scans. Arterial phase scans should start 5-10 s (5s for 20s scan duration. lOs for < lOs scan duration) after the enhancement in the trigger ROl exceeds 50HLJ above baseline. while portal phase scans are initiated a fixed time after the arterial phase has finished. Ideally. the inters can delay should be adapted to the duration of the arterial phase scan. We suggest aortic triggering with a 40-45 s delay for the portal venous phase and a 55-65 s delay for the hepatic venous phase (relative to the aortic trigger level). Triphasic Spiral CT
The term "triphasic spiral CT' is used inconsistently in the literature. It may indicate a noncontrast scan followed by a biphasic CT, or a true triphasic CT with three distinct enhancement phases. In this book. only the latter definition is applied. With Single-slice spiral CT, a triphasic examination consists of a scan in the hepatic arterial phase. portal venous phase. and a scan in the interstitial phase after contrast administration. Such a triphasic examination is performed identically to a biphasic CT but adds a third scan afterward with a start delay of either 3-5 minutes (early delayed. vascular equilibrium phase) or 10-15 minutes (late delayed. parenchymal equilibrium phase). A triphasic CT is helpful for the characterization of known hepatic lesions (hemangiomas. HCC. cholangiocarcinomas) and can improve the detection rate of hepatocellular carcinomas, especially in cir-
rhotic patients. Hemangiomas tend to demonstrate a blood-pool effect with an enhancement of the perfused portions of the tumor that is similar to the vascular space (better seen on early delayed scans). HCC may demonstrate a capsule, or may appear as a hypoattenuating lesion (better on late delayed scans). Peripheral cholangiocarcinomas may only be seen on delayed scans as a hyperattenuating lesion spreading parallel to the portal branches. Cysts remain hypoattenuating in all scan phases. An early delayed scan 3-5 minutes after contrast injection helps to identify hemangiomas and small cysts. • A late delayed scan after 10-15 minutes helps to detect small cysts and tumors with a large fibrotic component (e.g., cholangiocarcinoma, HCC with capsule, some nodular NHL). With multislice CT, early and late arterial phase scanning became possible, and a triphasic protocol has been suggested that includes these two phases followed by a later portal (= hepatic venous) phase. Rapid scanning is necessary to capture distinct phases. With 4-slice scanners, 4 x 2.5 mm to 4 x 3.75 mm collimation should be used. A high pitch factor is to be preferred over a wide collimation (GE scanners: HS instead of HQ. mode). Aortic bolus triggering is beneficial: the early arterial phase is started with a 3-5 s delay, the late arterial phase is started 5 s after the first phase ends. The hepatic venous phase is initiated some 20 s after the late arterial phase has ended. Such a protocol yields the best separation of arterial and venous anatomy (early arterial and hepatic venous phases). Lesion detection, however, does not profit from the early arterial phase because most hypervascular lesions are better (or only) detected during the late arterial phase. The protocol may therefore be recommended for a pre-surgical work-up but, because of the improved differentiation of detected lesions, we usually prefer a triphasic examination with an arterial phase scan initiated some lOs after contrast arrival in the aorta, an hepatic venous phase, and a (usually 3-5 min) delayed phase.
Intra-arterial Contrast Injection Techniques An early arterial phase is helpful for pre-surgical 3D reconstruction of the arterial supply but rarely improves tumor detection. Due to the increased radiation exposure. one should consider MRI with dynamic sequences after gadolinium or with liverspecific contrast agents as the first choice technique in younger patients with suspected benign liver tumors.
Intra-arterial
Contrast
Injection
Techniques
Intra-arterial contrast injection techniques have been developed to improve detection of liver lesions. but such techniques at present only hold fringe indications. CT hepatic arteriography (CTHA) may be used during arterial chemoembolization or before surgery of hepatocellular carcinomas (H CC). and CT during arterial portography (CTAP) may be used in critical preoperative situations if more modern CT or MR techniques are not available. •
CT Hepatic Arteriography
(CTHA)
CT hepatic arteriography is an interventional procedure in which a catheter is placed under fluoroscopic guidance into the proper hepatic artery. The patient is then transferred to the CT room and a scan (spiral or multislice CT) is performed during direct contrast injection into the hepatic artery. This procedure yields excellent and purely arterial contrast enhancement of hypervascular hepatic tumors. It is mostly performed in surgical candidates with liver cirrhosis and suspected hepatocellular carcinoma. nowadays often in conjunction with spiral CTAP. With a catheter in the hepatic artery. 7S ml of 1: 1 diluted contrast medium (l.e .. ISOmgll ml) is injected at a rate of 2-3 mlls. Scanning is initiated after 3-10 seconds using a technique that covers the entire liver in one breath hold. Replaced or aberrant hepatic arteries and perfusion abnormalities (cirrhosis) may give problems in the interpretation of the scan. Addition of a delayed phase (biphasic CTHA) is used for improved differentiation of tumor versus perfusion abnormalities. Injection of vasodilators like papaverine. Priscoline or prostaglandin Et has been reported to increase portal flow and to enable more confident differentiation of true le-
sions from pseudolesions due to systemic venous inflow tracts. The advent of multislice scanning probably makes the technique obsolete, because proper timing of the multislice scan and use of increased flow rates can reach high and almost selective arterial enhancement of the liver. CT during Arterial
Portography
(CTAP)
CTAP utilizes the principle of indirect splenic and mesenteric portography. It can detect hepatic lesions with high sensitivity by the selective delivery of contrast medium to the portal venous system (see Fig. 11.14). CTAP is used in the planning of partial hepatic resections (to exclude tumors in residual liver segments), but cirrhosis and large tumors can limit its effectiveness by altering portal venous hemodynamics. Spiral CT during arterial portography offers a detection rate of well above 90% for lesions ss 1 cm in diameter. For lesions smaller than \ cm, results are poorer. Specificity of CTAP is very limited, however, because all types of lesions appear hypoattenuating on these scans. Apart from biphasic multislice CT, MRI with liver-specific contrast agents (SPIO) has been shown to be equally accurate and more specific, and may replace CTAP in the future. The technique involves injection of ISO200ml of normal or 1: 1 diluted (I.e., ISO300 mg lIml) contrast medium at 3-S mlls through catheters selectively placed in the superior mesenteric artery, splenic artery, or both. Some authors claim better results with splenic artery injections. After a delay of 20-60 s spiral or multislice CT of the entire liver is performed during a breath hold. For improved characterization of lesions, biphasic studies during the true portal phase after 20s and late (recirculation) phase after 60s can be performed. Its specificity is still limited, however. The late scan is useful for distinguishing perfusion-related artifacts from lesions. Small rounded lesions that are conspicuous during the portal phase but not in the late phase are interpreted as neoplasms. Combination with CTHA is often performed for improved characterization, but there is controversy about which technique should be used first. As in CTHA, injection of va so dilating agents can enhance portal venous flow, leading to better consptcutty of small lesions and differentiation from pseudolesions.
11 The Liver
CTAP is a completely nonspecific method of lesion detection (except for cysts). Low-attenuation artifacts further reduce its specificity (Fig. 11.8).These artifacts may be caused by laminar flow in the portal vein (poor mixing of splenic and mesenteric blood) or by the compression of portal venous branches. Aberrant portal vessels that drain directly into the liver (e.g., the left gastric vein, the gallbladder veins) can mimic malignant lesions in segment IV or in the gallbladder bed. Compression of the liver parenchyma by ribs may cause hypoperfusion artifacts as well (see Perfusion Abnormalities, below). •
Lipiodol CT
Lipiodol CT is a technique that offers high specificity but only moderate sensitivity in the detection of HCC. Best results are achieved with hypervascular HCC tumors, especially those that are larger than 2 cm. In other cases, sensitivity is poorer. Prior to the CT examination, approximately 5 to 10ml of Lipiodol, an oil-like iodinated substance, is injected into the proper hepatic artery through a catheter under angiographic guidance (taking care to include accessory or aberrant hepatic arteries). Some 2-4 weeks later, a CT examination of the liver is performed without further administration of intravenous contrast material. By this time the marked uptake of Lipiodol by HCC nodules will contrast sharply
a
b
Fig. 11.8 Perfusion artifacts in CTAP may have a typical wedge-shaped appearance distal to a tumor with compression of portal venous branches (a), or they may be
with the very low Lipiodollevels in normal liver parenchyma. Hyperattenuating nodules that show essentially homogeneous high density are considered pathognomonic for HCC (see Fig. 11.34). A homogeneous high density pattern is not seen with all HCCs. Other hypervascular tumors may show nonspecific Lipiodol uptake. Spotty uptake of Lipiodol in the periphery of a lesion does not positively identify the lesionasHCC. Note that Lipiodol is a contrast agent for lymphography that is not approved for intraarterial injection in many countries [off-label use}.
Perfusion CT
Perfusion CT or Single-slice dynamic CT has (re) gained increased popularity for studying liver perfusion. Single-level CT scanning is performed while a bolus of 40-80 ml of contrast medium is administered by fast intravenous bolus injection (5-10 mlls) followed by a saline chaser-bolus (e.g., 40ml). Scans are started at 10-15 s after the begin of the injection and are acquired every 1 to 4 seconds for 60-120 seconds, then at larger intervals for up to 5 minutes after the start of the injection. Perfusion quantification relies on using slope measurements or
c caused by veins that drain directly into the liver, e.g., gastric veins anteriorto the portal vein in segment IV (b) or gallbladder veins into the gallbladder bed (c).
Perfusion CJ compartmental modeling. Several indices (transit time. hepatic perfusion index. peak perfusion values. distribution volume) can be measured and may be displayed in parametric images. To minimize radiation exposure. an 801
(injection rate 4-5mlls, 40ml saline chaser bolus), and the scans are acquired every 2 to 3 seconds for 30 seconds, then every 30 seconds for up to 5 minutes after the start of the injection. Generally a small delay of 10 seconds is used before the first scan. The scans can be repeated at a different level if multiple lesions are present, but residual enhancement will make these results somewhat less accurate. Small lesions may be missed due to respiratory misregistration. This use has now been replaced completely by multiphasic spiral CT. However, characterization of focal lesions and differentiation of true lesions from perfusion abnormalities by evaluating perfusion parameters can still be useful in difficult cases. Currently 4-16, and in the future 256 simultaneous sections may be measured, enabling wider liver coverage with thin slices. There has been the suggestion of using routine biphasic scanning to calculate an "arterial perfusion index" for each voxel of the liver. Such an effort, however, should be viewed critically at present because enhancement of the liver parenchyma changes rapidly during a scan, and thus strongly influences the measured values. With further improvement of technique, however, parametric images of the arterial perfusion index may indicate parenchyma at risk of harboring metastases and may help differentiate small benign from malignant lesions.
Enhancement Characteristics The liver has a dual blood supply via the portal vein and hepatic artery. The portal vein delivers 75-80% of the blood, while the hepatic artery delivers 20-25%. These two systems have numerous communications such as the transsinusoidal, transvasal and transplexal routes. A decrease in portal venous flow is associated with an increase in hepatic arterial flow, but not vice versa. Hepatic venous obstruction is often associated with flow reversal in the portal vein (acting as a draining vein) and increased hepatic arterial flow to the occluded area. All tumors have an arterial blood supply, but with the exception of the very periphery of other tumors, only primary hepatic tumors (well differentiated HCC, few adenomas or FNH) additionally may have some portal supply, but it
is usually less than that of the surrounding liver parenchyma. Besides perfusion, tumor enhancement is determined by contrast diffusion into the interstitium, which increases with time after the contrast injection and is the more rapid the more "leaky" tumor capillaries are present. Direct enhancement of the liver cells is a very late phenomenon that occurs during the phase of hepatic contrast excretion. Only the intra-arterial injection of contrast medium can selectively demonstrate the delivery of portal venous contrast (indirectly via the superior mesenteric artery or splenic artery). Direct intra-arterial contrast delivery to the liver (via the hepatic artery) leads to a prolonged phase of arterial liver enhancement.
11 The Liver
Contrast material administered by intravenous injection initially produces arterial enhancement (the arterial phase) followed swtftly by portal opacification via the splanchnic vessels (the portal phase). The portal phase of enhancement always overlaps the arterial phase. During the early portal venous phase. CT numbers in the portal vein may be higher than in the aorta. However. this generally requires flow rates of s4mlls. During the late portal venous phase. there is strong enhancement of the liver veins. which has led some authors to call this phase a hepatic venousphase. Even later. a vascular equilibrium phase is established between the arterial. portal. and venous territories. Finally. diffusion of contrast medium into the interstitium produces an equilibrium between the liver parenchyma and blood vessels (the interstitial phase or parenchyma! equilibriumphase). In the postprandial period. portal venous flow is increased and leads to an improved enhancement of the hepatic parenchyma and portal vein. YRapid scanning and accurate timing are essential for optimum lesion detection and differentiation in the liver. The onset and duration of the arterial and portal phases of enhancement depend strongly on the contrast volume (iodine load). injection rate (flow rate). and individual circulation time. Ideally. a bolus triggering technique should be used that will automatically initiate scanning during the desired phase of enhancement. Hepatic Arterial Phase (HAP)
In the early arterial phase (£AP) immediately after the onset of aortic enhancement. only the hepatic arteries and extremely hypervascular lesions are opacified. Most tumor enhancement is seen S to 10 seconds later and persists through the late arterial phase [LAP) some IS-20 s after the arrival of contrast in the aorta. Recent studies with compact bolus administration have shown that the duration of the true arterial phase (no portal venous inflow) may be as short as 8 seconds. irrespective of the flow rate of contrast administration. This means that optimum tumor enhancement will occur when there is already initial portal vein enhancement.
Transient arterial hyperattenuation (also called transient hepatic parenchymal enhancement. transient hepatic attenuation difference. THAD) is a phenomenon that is a direct consequence of the dual blood supply of the liver. Any compromise in portal venous perfusion will cause a compensatory increase in arterial blood flow to the affected region to an extent that keeps the overall flow constant. Such areas will appear hyperattenuating in the arterial phase but isoattenuating in the portal phase (see Fig.I1.S0). This transient arterial hyperattenuation will sometimes persist through the early portal phase. Hypervascular tumors display an intense. often inhomogeneous enhancement during the arterial phase (Fig. 11.9; see also Figs. "!US. 11.17C, 1l.18, 11.20, 11.26, 11.28-11.32, 11.35). Such lesions may be benign (capillary hemangioma, adenoma, focal nodular hyperplasia) or malignant (HCC, metastases from hypervascular tumors). Markedly heterogeneous enhancement is highly suggestive of malignancy (with the exception of adenomas with intratumoral hemorrhage). Peripheral rim enhancement exists almost exclusively in malignant disease and is most frequently due to metastases or cholangiocellular carcinoma (see Figs. 11.2S, 11.36). Cavernous hemangiomas have an exclusively arterial blood supply and show a characteristic peripheral to central pattern of enhancement in the lacunar vessels (see Figs. 11.9, 11.16). Some capillary hemangiomas (Figs. 1l.l7, 11.SOC) or peliosis hepatis (Fig. 11.S9) will appear as a homogeneous, hypervascular lesion (enhancement almost identical to that of the arteries). Results from dynamtc perfusion studies indicate that metastases in the liver will lead to an increased arterial flow, even if the lesions themselves cannot yet be detected. A contrast enhancement index (= parenchymal enhancement during the arterial phase/peak parenchymal enhancement during the portal phase) of> 0.4 at 40s after the begin of contrast injection (lSOml at 4mlls) is highly predictive of hepatic micrometastases. Arterial phase images are useful not only for tumor diagnosis but also for the preoperative evaluation of arterial variants in the upper abdomen (CT angiography). No 3D renderings are required for this purpose because of typical patterns on axial sections (see Fig. 11.4) but negative oral contrast material retains the op-
Enhancement
Characteristics
1=Cyst 2=Hemangioma 3 = Hypervasculartumor
=
FNH Adenoma Capillary hemangioma
Metastasis HCC
4 Peri pherally hypervascu lar tumor Metastasis
CCC
5 = Hypovasculartumor Metastasis Well-differentiated
HCC
CCC Lymphoma
6=Transient arterial hyperenhancement Portal venous compression or occlusion Lokal increase of intrahepatic pressure
~
-
\
••••• /
• Arterial
Fig. 11.9
phase
Typical
enhancement
•
,I I
'-
'-
-
I
Portal phase
characteristics
of hepatic
tion of performing such 3D displays. The early arterial phase yields the best results, but also images acquired during the late arterial phase are sufficient for making the diagnosis. This technique obviates the need for conventional angiograms in most patients. CTA is also an appropriate study in patients with hepatic aneurysms. Image analysis relies on overlapping axial CT sections, preferably using an interactive" cine display". Thin-slab MIP in a caudocranial direction (matching the orientation of CT slices) is useful for the presentation of findings. It should be noted that the best results are obtained if a minimum number of sections are used for MIP reconstruction. Volume-rendered images are optim urn for display of complex vascular anatomy, especially when thin-section multislice CT scanning was performed.
, I I
'-
'--
I
Interstitial
lesions
•
following
intravenous
phase
contrast administration.
Portal Venous Phase (with CTAP)
CTAP is the only technique that can selectively deliver contrast material to the portal venous system. All lesions are hypoattenuating in CTAP owing to their relative lack of portal perfusion. The only exceptions are lesions that receive a substantial portal blood supply such as regenerative nodules, some small adenomas, and some well-differentiated HCC. •
Portal Venous Phase (after IV Contrast Administration)
The arterial phase after intravenous contrast injection is followed by a phase with predominance of portal enhancement [portal venous phase, PVPj. The level of enhancement of the portal vein may exceed the level of aortic enhancement when an optimum injection proto-
11 The Liver
col is used (see Fig. 11.9). This happens directly after the late arterial phase. some 20-30 s after contrast arrives in the aorta. A similar phenomenon can be seen during the late portal venous phase: some 20s after optimum portal vein enhancement. the liver veins enhance strongly (hepatic venous phase, HVP). Even later (3-5 min after contrast material injection), the differences in contrast enhancement between the various vascular territories are lost (vascular equilibrium phase). Such more subtle differences in portal venous enhancement, however, rely on rapid injection of contrast material (>4mlls). With slower injection rates (e.g., 2mlls) , the phases blend into each other. The hepatic parenchyma receives most of its blood supply from the portal vein while tumors do not. This makes hypovascular tumors appear hypoattenuating to the liver parenchyma. Problems result from contrast diffusion into the interstitium, which can cause small lesions to become isoattenuating to the parenchyma ("vanishing lesions"). This problem increases with time after injection, so rapid scanning is important to ensure that small lesions are not missed. In tumors that have a rich arterial supply but a scant portal supply (e.g., most HCCs and hypervascular metastases), the effects of arterial tumor enhancement and portal enhancement of the surrounding liver parenchyma may cancel out with the result that the tumor is not visualized during the portal phase (Fig. 11.9, see also Fig. 11.30). Except for sharply defined lesions with near-water attenuation (hepatic cysts), hypoattenuating lesions in the portal venous phase are frequently malignant (metastases from hypovascular tumors, CCC or some HCC). Heterogeneous enhancement (except for the typical pattern of hemangiomas) is characteristic for malignant disease. Lesions with near-water attenuation that have irregular borders, heterogeneous or peripheral contrast enhancement may be due to cystic metastases (usually from mucin-producing tumors). embryonal carcinomas, or abscesses. Target or bull's eye phenomena are characteristic for metastases or CCC, but may also occur with some hepatic abscesses (especially fungi). Compression of normal parenchyma surrounding a benign lesion (e.g., FNH) may cause localized arterial hyperperfusion of this area. If this transient hyperattenuation persists into the portal venous phase,
it may simulate the rim enhancement (target phenomenon) seen with metastases. Haemangiomas display a typical pattern of peripheral patchy (lacunar) enhancement that progresses from the arterial to the portal venous phase. Because of the slow arterial flow in their lacunar spaces, hemangiomas gradually become isoattenuating to the vessels (blood-pool effect). As long as vascular enhancement is higher than hepatic enhancement, hemangiomas should also remain hyperattenuating to the liver parenchyma (e.g., during the vascular equilibrium phase! delayed phase). Some small hemangiomas « 2 em) display atypical enhancement patterns. Some may remain hypoattenuating in the portal phase but demonstrate only a tiny enhancing dot in the periphery of the lesion ("bright dot sign"). Others are homogeneously hyperattenuating in the arterial phase, and remain so during the portal and delayed phases. Hemangiomas are isoattenuating to the surrounding liver only during the parenchyma! equilibrium phase. Interstitial Phase
The diffusion of contrast medium into the interstitium starts immediately after the delivery of iodinated contrast to the liver and increases during the portal phase. Finally, during the interstitial phase (parenchymal equilibrium phase), it becomes the main determinant of the lesionto-liver attenuation difference. This interstitial phase of hepatic enhancement occurs when equilibrium is established between the parenchymal and vascular levels of attenuation. Focal hepatic lesions are approximately isoattenuating to the liver parenchyma during this phase. As a result, the interstitial phase is not useful for tumor detection, but it can define lesions that do not have a blood supply (cysts, abscesses, dilated bile ducts). Exceptions are tumors that have a large extracellular space (lymphomas) or large fibrous components (cholangiocarcinoma): these lesions appear hyperattenuating due to the delayed diffusion of contrast medium back into the bloodstream (Fig. 11.9). This effect can be used to detect cholangiocarcinomas that would otherwise evade detection. In hypervascular tumors with a rapid wash-out (HCC). lesions may become hypoattenuating with respect to surrounding liver enabling better detection than during the portal venous phase. A tumor capsule may also be visible to better advantage.
Hepatic
Scanning is usually performed some 10-15 minutes after contrast administration. In some cases. however. there will be a residual enhancement difference between vessels and liver parenchyma. which makes it clear that a separation between the various phases is somewhat artificial. Hepatic Excretion Phase
The hepatic excretion phase occurs several hours after injection of the contrast medium. Hepatobiliary contrast excretion accounts for up to 2% of the total iodine load. depending on the contrast material used. and leads to an increase in the CT attenuation of the liver. This effect improves the attenuation difference between tumors and the hepatic parenchyma. Gallbladder opacification also may be noted. The use of this phase has been abandoned as it does not yield additional information compared to multiphasic helical or multislice CT.
Cyst
Vascular Opacification
Depending on the pel-fusion phase the hepatic arteries. the portal veins. or the liver veins show the highest level of enhancement. A biphasic spiral CT study provides optimum visualization of all vascular systems and can obviate the need for diagnostic arterial angiography. The early arterial phase does not suffer from overlap of venous structures but provides no additional information over later arterial phases. If scanning in the portal phase is performed too early. the hepatic veins may still be insufficiently enhanced. N onenhanced veins can even mimic hypoattenuating pathology. For this reason. scanning of the hepatic veins should be delayed somewhat by performing the scan in the caudocranial direction. or by scanning primarily in the late portal venous (= hepatic venous) phase.
Cystic Liver Lesions The characteristic feature of cystic liver lesions is that the fluid-filled portions of the lesions do not enhance with intravenous contrast. regardless of the phase in which imaging is performed. Enhancement may be seen at the periphery of the lesion. within a capsule. within septa. or in accompanying solid components. depending on the type oflesion (Table 11.6).
Developmental (Simple) Hepatic Cyst Simple hepatic cysts are common benign lesions that occur in up to 7% ofthe population. They can be solitary or multiple and 95% are unilocular. Cysts can vary in size from several mm to more than 10cm. Middle-aged women are predominantly affected and the cysts are typically asymptomatic. Large cysts can present with symptoms of abdominal fullness or obstructive jaundice from extrinsic compression of the biliary tree. Complications occur rarely and include intracystic hemorrhage. superinfection. or rupture. Peribiliary cysts result form cystic dilatation of periductal glands and occur adjacent to large intrahepatic bile ducts. There is an increased incidence in cirrhosis.
In all types of cysts. CT can be used preoperatively to evaluate space-occupying cysts and complications. For complicated cysts however. MRI is the preferred modality. CT Morphology The cysts appear as hypoattenuating lesions (0-30 HU) up to 20cm in diameter with welldefined outlines (caution: partial volume effects) and no internal septa (Fig. 11.lOa) . Larger cysts may have an accentuated "wall" formed by the compression of surrounding liver parenchyma. Small cysts « 1 cm) may be hard to distinguish from hypovascular hepatic metastases due to partial volume effects. Thinsection multislice CT usually solves such problems for lesions > 3-4 mm but may introduce new ones because it can detect even smaller lesions. Interstitial phase scanning. however. allows for secure distinction between a small cyst and a hypovascular tumor because the latter will appear isoattenuating to the liver in this phase (Fig.ll.lOb.c). Pseudoenhancement (10HU or less) can be seen in contrast-enhanced studies due to partial volume averaging from the surrounding enhanced liver tissue.
425
11 The Liver Table 11.6 Differential diagnosis of cystic liver lesions Cysts of hepatic origin Uncomplicated liver cyst
No septa, no contrast enhancement
Hematoma, traumatic cyst
Sometimes hyperattenuating, no contrast enhancement, patient history
Dysontogenetic cyst
Diffuse involvement, often pancreas and kidney cysts
Bilioma, bile leak
Localization: liver edge, hilum; history: postoperative, trauma
Choledochus cyst
Localization: liver hilum, connection with common bile duct
Caroli syndrome
Cystic formations along portal vein branches, string-of-beads bile ducts
Cysts of inflammatory origin Abscess Microabscesses
Unsharp margins, air enclosures, enhancing rim target)
+
edema possible (double
Diffuse spread of small hypoattenuating lesions (mm-cm), often Candidasepsis
Ecbinococcus cyst
Thin calcification of margin, contrast enhancement of capsule daughter cysts
+
septa,
Ii
I Cystic tumors
a
Cystic metastases
Enhancing, irregular-nodular rim; often in mucinous carcinomas or sarcomas
Biliary cystadenoma
Polycystic lesion near hilum, contrast enhancement of wall
Mesenchymal hamartoma
Age < 19Y, multiple cysts, solid often hypervascular tumor regions
b
+
septa
c
Fig. 11.10 Liver cysts. Large cysts usually pose no diagnostic problem (a). Small cysts (arrow) may be indistinguishable from metastases in the portal venous phase
(b) but, in contrast to metastases (arrowhead), remain hypoattenuating in the interstitial phase (c).
Peribiliary cysts are usually multiple and lo-
cases can often be resolved by contrast -enhanced spiral or multislice CT with reconstruction of thin overlapping sections. Infection is manifested by peripheral enhancement and the presence of gas bubbles that cannot be explained by a previous needle procedure.
cated in the hilum of the liver. Hemorrhagic cysts may have attenuation values higher than 30HL] but do not enhance with intravenous contrast. This should be confirmed by measuring CT attenuation. Doubtful
Mesenchymal Hamartoma POlycysticLiver Disease
•
CT Morphology
Multiple hepatic (dysontogenetic) cysts occur commonly in patients with autosomal dominant polycystic kidney disease (prevalence up to 75%). The cysts are variable in number and size and may lead to hepatic enlargement or vascular and biliary obstruction. Hepatic cysts may be seen without renal involvement. Hepatic cysts are not a feature in patients with autosomal recessive polycystic kidney disease. (Hepatic fibrosis is seen in this entity.)
CT shows either a single or multiple tiny (5-15 mm) cystoid lesions with irregular margins. If multiple lesions are present, they are usually scattered throughout the liver and often located in proximity to medium-sized portal branches. Hamartomas do not enhance, thus detection is best in the portal venous phase. A thin pseudo-enhancing rim can be caused by compression of surrounding normal liver tissue.
•
Mesenchymal Hamartoma
CT Morphology
Dysontogenetic liver cysts appear on CT as multiple hypoattenuating lesions (0-20 HU) that frequently show a clustered arrangement and do not always have smooth margins (Fig. 11.11). They do not enhance following contrast administration. Scattered amorphous wall calcifications may be present. Biliary Hamartomas
Mesenchymal hamartomas are very rare developmental liver tumors that occur almost exclusively in small children (peak incidence 15 to 22 months) but may occur as late as 19 years of age. The tumor consists of a mixture of mesenchymal tissue and bile ducts. It grows rapidly, contains cystic areas, and predominantly affects the right lobe of the liver (> 80%). •
CT Morphology
Biliary hamartomas (von Meyenburg complexes) are benign biliary malformations that consist of cystic components within a varying amount of fibrous stroma. These malformations do not communicate with normal biliary ducts. They can be solitary, but are most often multiple. The lesions are asymptomatic and the estimated prevalence from autopsy series is in the order of 2-3%.
Most hamartomas are large, hypoattenuating masses (5 to 30cm) that contain multiple cystic spaces of a few mm up to more than 10cm in size (Fig. 11.12). Cysts are grossly visible in 80% of the tumors and vary in density. Patchy contrast enhancement occurs only in hypervascular portions of the mass.
IFig. 11.11 IPolycystic liver clustered cystic lesions.
IFig.11.12 Mesenchymal Ihamartoma lin a small child. CT demonstrates a large mass with multiple small cysts and no evidence of intralesional hemorrhage.
disease
with
multiple
421
77 The Liver
Cystic Metastases
Necrotic metastases (especially from sarcomas) or metastases after successful chemotherapy may occasionally have a cystic appearance. Primary cystic components may occur in metastases from mucinous aclenocarcinomas such as mucinous ovarian or colon carcinoma. Less frequent causes are melanoma, lung carcinoma, and carcinoid tumor. The absence of inflammatory signs helps distinguish the lesions from abscesses. The superinfection of a necrotic metastasis may cause gas formation. •
CT Morphology
Cystic metastases usually have a rim of viable tumor tissue whose enhancement contrasts with that of the hepatic parenchyma. Scans typically show a broad, irregular wall with sites of nodular thickening (Fig. 11.13).Often larger
Fig. 11.13 Cystic metastasis noma of the ovaries.
from cystadenocarci-
solid tumor components are found at the periphery of the cystic lesion. Cystic metastases can mimic the CT appearance of an abscess.
Benign Tumors and Tumor-like Lesions
CT is part of the basic diagnostic work-up of focal hepatic diseases. Despite the variety of potential lesions, CT findings are pathognomonic only for cysts, steatosis, hemangiomas, and focal nodular hyperplasia (Tables 11.7, 11.14). Adenomas cannot be confidently distinguished from malignant tumors. The confident classification of a solid lesion relies on typical findings in two independent diagnostic procedures. Otherwise surgical or biopsy confirmation is required. Incidentally Found Lesions
With growing use of multiphasic helical CT an increasing number of small lesions are found in the liver that are often too small to characterize. With the expanding use of multislice CT, this number may increase even further. In populations evaluated for benign disease, incidental benign lesions can be found in normal individuals in up to 15% of cases. The likelihood ofa liver lesion being malignant is about 1% in patients without known primaries. Most lesions represent small cysts, hemangiomas, or other benign entities.
Even in patients with known primaries, the number of actual metastases is small among lesions < 2 cm in largest diameter. The incidence of metastases in a large follow-up study in tumor patients was less than 12% in the overall study group, and varied from 8% for malignant lymphoma to 15% for colorectal cancer, and 22% for breast cancer. Only 4 % of patients with small lesions had additional metastatic lesions larger than 2 cm. CT follow-up using an identical examination technique or ultrasound (if the lesion is visible with this technique) appears best suited for such patients unless the lesion can be confidently diagnosed as benign (cyst, hemangioma). There is no consensus yet as to which follow-up intervals should be recommended. Focal Liver Steatosis
Focal variations of fatty infiltration of the liver often result from the focal hypoxia of liver tissue and may have a vascular etiology (e.g., portal vein occlusion) or a toxic-metabolic cause (e.g., chemotherapy). The various causes can lead to regional fatty infiltration or to areas of near normal parenchyma within a fatty liver. Rapid change (days to months) may be observed.
Foeaf Liver Steatosis Table 11.7 I Solitary
Differential
diagnostic
characteristics
of solid liver lesions
lesion
Benign: Focal fatty
infiltration
Localization:
gallbladder
next to the ligamentum Hemangioma FNH
Strong
Hepatocelhfar Infantile
bed, segment teres,
IV anteriorto
right portal vein or
can be geographic
Peripheral nodular enhancement, iris diaphragm phenomenon, pool effect, central thrombosis, centripetal fill-in ofCM
adenoma
hemangioendothelioma
arterial
Usually arterial possible Age
eM
<6
hypervascularization,
central
hypervascularization,
scar typical,
hemorrhage,
mo, CHF, large, calcifications,
no capsule
central
hypervascular,
blood
necrosis
centripetal
fill-in of
Malignant: Metastasis
Portal phase hypovascular, arterial phase depending on primary tumor: either hypervascular, peripheral hypervascular or hypovascular
HCC
Usually arterial hypervascular, uncommonly forms) infiltrationlthrombosis of portal vein uptake
FLC
Central
CCC
Hypovascular or peripheral hypervascular retraction of the liver capsule possible
scar, stellate
calcification,
Hepatoblastoma
Age <15 y, large, hypervascular,
Embryonal
Age
sarcoma
Angiosarcoma
Multiple
<30
y, hypervascular,
capsule,
further
necrosis,
large necrotic
Usually very hypervascular, rhage
hypovascular (early/small + hepatic veins, Lipiodol
nodular
like HCC
large calcifications and hemorrhagic
lacunae
possible,
parts
necroses,
hemor-
le,,~}?PJwi
Benign: See above, often subcapsular
Hemangiomas Adenomas
Seeabove
Regenerative
nodules
Cirrhosis; tenuatmq
post trauma,
partial
resection,
when
siderotic
can be hyperat-
Malignant: Metastases Multifocal
See above, HCC
Lymphoma Epithelial
hemangioendothelioma
•
CT
most frequent
in macronodular
Seldom
clear margins,
sometimes
Angiosarcoma
multiplicity
Usually
cirrhosis,
hypervascular
hypoattenuating,
in interstitial
phase
hyperattenuating
Hypovascular, often capsular unenhanced CT best See above,
most frequently
after contrast
disseminated
Morphology
Focal fatty infiltration may affect individual hepatic segments (usually with a vascular pathogenesis), but most show a geographic or rounded pattern of involvement. Perihilar occurrence is as common as peripheral involvement adjacent to the capsule, Regions with an
retraction,
hypervascular
large granular
calcifications,
foci
atypical portal venous supply (see Flg.Il.Sc. d) are more commonly affected, and may undergo greater or less fatty infiltration than the surrounding liver parenchyma, Typical locations are in segment IV near the falciform ligament, or the posterior edge of segment IV, or in the gallbladder bed (Fig, 11.14),
429
11 The Liver
small as 5 mm, but in selected cases, adding delayed images is helpful. Small tumors do not require treatment, but surgical resection is indicated if the tumor is symptomatic (15%) or is causing biliary tract compression. •
b
a
Fig. 11.14 Pseudolesion caused by a region with near normal fat content in the gallbladder bed (arrow) in a patient with diffuse fatty infiltration of the liver (a). The site of the lesions coincides with artifact-prone regions in CTAP. The hypoattenuating region at the anterior border of segment IV is a frequent finding (b) and has been attributed to a varying amount of fat or glycogen.
The enhancement characteristics are pathognomonic, as the affected region shows an enhancement pattern parallel to the surrounding parenchyma. Fat distribution disorders can mimic hepatic lesions. However, there is no mass effect. In case of doubt, in-phase and opposed-phase Tl-weighted MRI will prove or exclude the presence of fat.
Hemangioma
Hemangioma is the most common benign liver tumor with an incidence at imaging of 5-7%. They are 2-5 times more common in females, are usually asymptomatic, and are common incidental findings during routine imaging. Histologically, a cavernous hemangioma is a well-circumscribed lesion composed of vascular lacunae that have a purely arterial blood supply from the periphery of the lesion. Blood flows very slowly in the lacunar spaces, and there is no extravascular compartment. Areas of fibrosis are common and thrombosis of the lesion can lead to hyaline degeneration and calcifications. Biphasic CT is the technique of choice for detection and characterization of lesions as
CT Morphology
Hemangiomas can be multiple in 10-50%. Lesions > 1Ocm are by convention referred to as "giant" hemangiomas. On noncontrast CT, hemangioma is usually a well-defined hypodense lesion with lobulated borders. A subcapsular location is typical. Calcifications occur in 10% of lesions with thrombosed areas. In the arterial phase, there is typically globular peripheral enhancement that parallels the enhancement of the aorta. Large peripheral feeding vessels may be demonstrated in large lesions. The lacunar spaces will gradually fill in towards the center, and will grow on portal venous and delayed imaging phases [lacunar enhancement, centripetal filling). By the end of the portal phase (vascular equilibrium), all enhancing lacunae are isoattenuating to the hepatic vessels (blood-pool effect, Fig. 11.15).Complete enhancement of the hemangioma can take seconds to more than 30 minutes. In the interstitial phase, when blood vessels and liver parenchyma are isoattenuating, the hemangioma can no longer be delineated although central regressive changes remain hypoattenuating in large lesions. Very large tumors may pose problems of differential diagnosis because of their mass effect and the presence of thrombosed or necrotic areas. Enhancement is often heterogeneous, but the detection of lacunae in the arterial and portal phases and the centripetal filling pattern seen on delayed scans at 10 to 30 minutes suggest the correct diagnosis. Small hemangiomas often show atypical features such as homogeneous hyperattenuation or hypoattenuation during the arterial phase. Hypoattenuating hemangiomas can show a characteristic central bright dot (best seen in portal venous phase, Fig. 11.16). Rapid enhancing hemangiomas can be associated with arterioportal shunting resulting in a wedgeshaped hyperattenuation in the surrounding parenchyma early after contrast administration (Fig. 11.16).Other rare features of hemangiomas include associated focal or regenerative nodular hyperplasia, hyalinization, pedunculation
Hemangioma
a
Fig. 11.15 Biphasic spiral CT of a small hepatic hemangioma with a typical subcapsular location. Without contrast medium (a) the mass is isoattenuating to blood. During the arterial phase (b) a small hyperattenuating lacuna is seen (differential diagnosis: hypervascular
a
c
b
b
Fig. 11.16 Bolus CT scanning of a large cavernous hemangioma. (a) On noncontrast CT the tumor has approximately the same CT attenuation as the aorta. The thrombosed areas with central hyaline degeneration are hypoattenuating. (b) During the arterial phase the tumor shows brightly enhancing peripheral lacunae.
(Fig. 11.17) and Kasabach-Merritt syndrome (coagulopathy with systemic fibrinolysis and thrombopenia). Differentiation of these small lesions from hypervascular malignant lesions is difficult, and a sensitivity of only 50% has recently been reported.
tumor). In the portal venous phase (c), the whole lesion becomes isoattenuating to blood. This blood-pool effect distinguishes hemangioma from a hypervascular metastasis.
c
d
(c) In the portal phase these areas appear larger and less distinct, as the lesion "fills in" from the periphery toward the center (the iris diaphragm sign), (d) In the late interstitial phase the enhanced portions of the hemangioma are isoattenuating to blood (blood-pool effect) .
In CT, only triphasic spiral scanning can characterize very small hemangiomas. Some show the same enhancement pattern as larger lesions, while others show homogeneous enhancement during the arterial phase and become isoattenuating to vessels during the portal phase and the delayed phase (bloodpool). Some lesions may be indistinguishable from small hypervascular metastases. A pro-
431
77
a
The Liver
b
Fig. 11.17 Atypical hemangiomas may be pedunculated (a), or may appear hypoattenuating on portal phase scans (b). Note the acute angle between pedunculated lesion and liver (a). Capillary hemangiomas
longed filling-in is more often present in large lesions but can also occur in small hemangiomas of 1-2 cm size. Such small lesions can then be misinterpreted for metastases if no delayed phase scan is performed.
Focal Nodular Hyperplasia (FNH)
FNH is the second commonest benign liver tumor and has an estimated prevalence of3%. It is seen in patients 20-50 years of age and is 4-10 times more frequent in women. It is often an incidental finding on routine imaging studies. FNH induction is unrelated to estrogen and, while in the past it has been reported that estrogen stimulation (contraceptives, pregnancy) might increase the risk of growth or hemorrhage, a recent large study could not find any relation between growth or complications and the use of oral contraceptives. It is thought that FNH is a benign hyperplastic response to a congenital arteriovenous malformation resulting in a hypervascular, lobulated solid mass that has a predominant arterial supply. The tumor consists ofhepatocytes, Kupffer cells, and bile ductules in an abnormally organized manner. Fibrous (myxomatous) tissue forms a "central scar" with radiating septa that surrounds the central feeding artery. Multiphasic helical or multislice CT is considered to be the optimal CT technique, inc1ud-
c (arrow) may present as a homogeneously hypervascularized mass (c). There are also a cyst (wide arrow) and typical small cavernous hemangiomas (arrowheads) demonstrated in this late arterial phase scan.
ing (late) arterial, portal venous, and (late) delayed phases. Detection sensitivities of 70-80% have been reported. •
CT Morphology
FNH is usually a solitary lesion, but can be multiple in 20%. Most lesions are located in the periphery of the liver and measure < 5 cm. Intralesional hemorrhage is rare. On the noncontrast scans, FNH nodules are well-defined, slightly hypoattenuating or isoattenuating masses. The central scar is often barely perceptible on non contrast scans. There is no capsule and calcifications are rare « 1%). Pedunculated forms can occur. Arterial phase CTtypically demonstrates intense enhancement of the entire tumor (Fig. 11.18). Enhancement is usually homogeneous but polynodular patterns exist. A hypoattenuating central scar is pathognomonic and can be visualized after contrast administration in up to 50% of cases. The central tumorfeeding artery can be demonstrated during this phase in up to 60%. Both are usually not visible in smaller tumors. A rapid wash-out of contrast material is characteristic, and most tumors become isoattenuating to the liver parenchyma during the portal venous phase. The central scar is hypoattenuating in this phase, but slowly increases in attenuation through diffusion of the contrast agent and may become relatively hyperattenuating on delayed images.
Focol Nodular Hyperplasia
a
b
(FNH)
c
Fig. 11.18 Focal nodular hyperplasia (FNH). The lesion has a central scar and is hyper perfused during the arterial phase (a). Thereafter the tumor quickly becomes isoattenuating, and by the portal phase it is detected
only by its subtle mass effect on hepatic vessels (b). In the interstitial phase there is subtle enhancement of the scar(c) .
A large percentage of lesions may show one or more atypical features (Fig. 11.19),which may be more frequent when multiphasic techniques are used. Early contrast enhancement can be heterogeneous. Larger lesions may be ill-defined on arterial phase scans. In the portal venous phase, some lesions may be hypo attenuating due to wash-out effects, and compressed parenchyma surrounding the lesion may appear as a hyperattenuating rim. Large draining vessels or sinusoids may be seen in the periphery on portal phase images and rarely also on arterial phase images. Differential diagnosis in-
eludes hepatic adenoma, (fibrolamellar) HCC, and hypervascular metastases. With atypical features, confirmation by other imaging techniques or biopsy is indicated.
a
Prolonged tracer uptake in Tc-sulfur colloid scintigraphy is pathognomonic. Color duplex ultrasound relies on similar features as CT, but MRI after the administration of liver-specific contrast material (SPIO or Mn- DPDP) shows characteristically prolonged contrast uptake in many cases.
c
b
Fig. 11.19 Atypical FNH (4 X 1/6, coronal MPR) may remain hyperattenuating in the portal phase (a) and even the interstitial phase (b). They may show a pseudocap-
sule due to Pedunculated FNH (c).
compression of surrounding tissue (b). nodule in another patient with multiple
433
11 The Liver
Hepatocellular Adenoma (HCA)
HCC and the odd malignant transformation, surgical resection is indicated in nearly all cases.
Hepatocellular adenoma is a rare benign tumor that occurs predominantly in young women. A causative relation with oral contraceptives and androgen steroids has been firmly established. The risk of the tumor increases with the duration of oral contraceptive use (factor 2.5 after 5 years, 25 after 9 years), and in these women an incidence of 3-4 per 100000 has been reported. Other risk factors are pregnancy, diabetes mellitus, glycogen storage disease (type la), and iron overload in thalassemia. HCA in men can arise spontaneously, but most often is associated with the use of anabolic steroids. Histologically, the lesions are well-described, encapsulated tumors that are composed of disorganized cords of hepatocytes and a variable amount of Kupffer cells. The lesions contain no bile ductules, portal, or hepatic veins. Large subcapsular vessels supplied by hepatic arterial branches account for its hypervascular nature. As lesions outgrow their vascular supply, foci of necrosis, infarction, and spontaneous hemorrhage are common and can lead to c1inical symptoms. CT is best performed with multiphasic helical or multislice technique in the noncontrast, arterial and portal phases. Given the risk of hemorrhage (up to 40% with large tumors), the uncertainties of histologic differentiation from
HCA is solitary in 80% with more tumors in the right liver lobe. Most lesions measure 5-10 em in size and are in a subcapsular location. In the noncontrast phase HCA is a well-defined, smooth hypo attenuating mass, but may be heterogeneous due to intratumoral hemorrhage (in 25-50%) or necrosis (up to 40% in larger lesions). If the hepatocytes within the adenoma contain large deposits of fat the tumors may show focal or global negative CT attenuation (Fig. 11.21).Tumor calcifications are uncommon. Viable tumor tissue shows moderate hyperattenuation during the arterial phase (Fig. 11.20). The mass may appear inhomogeneous due to necrosis or hemorrhage. In contrast to focal nodular hyperplasia, hepatic adenomas enhance less and do not have a central scar, although necrotic foci can mimic scar tissue. In the portal venous phase the lesion may be anything from hypo- to hyperattenuating to the surrounding liver. Because of a large fat content most adenomas are already isoattenuating during the portal venous phase. Some tumors with marked fatty infiltration may even be isoattenuating during the arterial phase and be hypoattenuating on noncontrast scans as well as during the portal and delayed phases
a
c
b
Fig. 11.20 Hepatic adenomas. During the hepatic arterial phase (a), adenomas display a copious arterial blood supply. Thus their enhancement pattern differs little from that of FNH (Fig. 11.18) and is indistinguishable from HCC (Figs. 11.32 and 11.35). In the portal venous phase (b) adenomas may appear isodense or slightly
•
CT Morphology
d
hypo- or hyperdense to the liver parenchyma. A large fat content may be present (c, d) and can obscure that the enhancement in the arterial phase (c) remains higher than in the protal venous phase (d ). Quantitative attenuation measurements can help prove arterial hypervascularization.
Inflammatory Pseudotutnor
a
c
b
Fig. 11.21 Intratumoral necrosis and hemorrhage is typical for hepatic adenomas. Unenhanced CT (a) shows an isodense adenoma with fresh intratumoral hemorrhage, marked by hyperdense clots (arrows) bordering hypodense serum collections. During the portal phase
(b), the hemorrhagic area is more clearly delineated but the tumor boundaries are seen only indirectly by the displacement of portal venous branches, (c) Large hypervascular adenoma without necrotic or hemorrhagic areas despite its size.
(Fig. H.21). The enhancement characteristics of adenomas resemble those of well-differentiated HCC, from which they are often morphologically indistinguishable. Differentiation from hypervascular metastases is also difficul t.
Angiomyolipoma
Radionuclide scans are negative in 80%. and MRI and ultrasound are relatively nonspecific. Even with the use of liver-specific contrast agents, a secure differentiation between an adenoma and a well-differentiated HCC may be impossible with MRI. Percutaneous biopsy is often unrewarding because of the similarity of hepatic adenomas to regenerative nodules and HCC. Resection is therefore indicated.
Lipoma
Hepatic lipomas are very rare benign mesenchymal tumors. They usually occur de novo and can range from a few mm to more than 10cm in diameter. •
CT Morphology
Alipoma has a pathognomonic appearance with fat attenuation values of approximately -100 HLJ, similar to subcutaneous fat. The lesions are circumscribed, have smooth margins, and do not enhance.
Angiomyolipomas are also very uncommon tumors in the liver. They are composed of smooth muscle cells, proliferating vessels, and variable amounts of fat. They can be multiple when associated with tuberous sclerosis. •
CT Morphology
Imaging characteristics depend on the amount of fat present. On noncontrast CT, the lesions are often heterogeneous with both fatty and softtissue densities. Postcontrast, the lesion enhances heterogeneously. Differential diagnosis includes metastasis from liposarcoma. Adenomas and HCC may contain regions with negative CT attenuation but these regions have higher attenuation than body fat (substantially greater than -100 HU) because they correspond to fat -containing hepatocytes rather than fat cells.
Inflammatory Pseudotumor
Inflammatory pseudotumor of the liver is a very rare entity defined by a proliferation of plasma cells, other inflammatory cells, and fibrovascular tissue. It has a high incidence in infants and
435
436
7 7 The Liver
a
c
b
Fig. 11.22 hancement
Infantile hemangioendothelioma with multiple capillary (a) and are isodense in the portal venous phase (b).
young adult males who usually have symptoms suggesting active inflammation. Small lesions may be asymptomatic. Most reported lesions are solitary and vary between 4-8 cm in size. •
CT Morphology
On non-contrast CT a solitary, well-defined hypo attenuating lesion is seen. After contrast administration the lesion can remain hypo attenuating or show irregular enhancement in the portal venous phase. Frequently, there is an (incomplete) hyperattenuation on delayed-phase images. (lHE)
IHE is the commonest benign mesenchymal neoplasm of childhood. It usually presents in young infants as an abdominal mass associated with congestive heart failure or coagulopathy. Most patients (85%) are younger than 6 months of age and in 50% there are cutaneous hemangiomas. Histologically, IHE is composed of a network of vascular channels lined by endothehal cells. •
that display
marked arterial
en-
Hepatic Alveolar Echinococcosis (HAE)
Patients with alveolar type of echinococcal disease often give a history of spending time in forests or fields (mushroom gatherers, forest workers, hunters). Endemic areas are Central Europe, Russia, Japan, and North America. The larvae of the fox tapeworm E. multilocularis proliferate by exogenous expansion and infiltrate the host tissue, producing lesions that can mimic malignant processes. Chronic granulomatous reactions lead to central necrosis, cavitation, and punctate calcifications of varying density. •
Infantile Hemangioendothelioma
hemangiomas
CT Morphology
The hepatic lesions of alveolar hydatid disease appear on noncontrast CT as heterogeneous masses of moderately low attenuation with illdefined margins and a geographic distribution
CT Morphology
Most masses are solitary and can measure up to 20cm in size. Noncontrast CT demonstrates a large, hypo attenuating mass that shows calcifications in "15% of cases. Large lesions may show focal areas of necrosis or hemorrhage. Following contrast administration there is early enhancement in the periphery with an enhancement pattern that mimics hemangioma with centripetal fill-in (Fig. 11.22). There may be a considerably larger supraceliac aortic diameter compared to the infraceliac diameter.
Fig. 11.23 Alveolar echinococcus with a tumor-like growth presenting as an infiltrating hypoattenuating mass. Note the peripheral calcification and the involvement of the retroperitoneum and inferior vena cava.
Metastases
(Fig. 11.23). Punctate calcifications of varying density may be scattered through the necrotic and granulomatous tissue. Pseudocystic areas of necrosis (10-20 HU) with a hyperattenuating perifocal reaction are typical. Inhomogeneous.
tumor-like enhancement is seen after contrast administration. Often the masses are morphologically indistinguishable from primary hepatic tumors. and a definitive diagnosis requires serologic testing.
Malignant Tumors The list of questions that determine the resectability of an hepatic tumor is exhaustive (Table 11.8) and need not be fully covered for every indication. The choice ofCT technique depends on the diagnostic problem (see Table 11.5). with more elaborate techniques being reserved for presurgical work-up.
Metastases Diagnosis of metastatic disease is one of the most common indications for liver imaging since the presence of liver metastasis is one of the main determinants of patient survival. The liver is fertile ground for the development of Table 11.8
Important
questions
before
partial
liver
section 1. Rule out/confirm
tumor
diagnosis
2.
Differential diagnosis • Rule out benign lesions • Metastasis or primary liver tumor? • Existence of bile duct carcinoma?
re-
metastasis because it receives 25% of the cardiac output. the sinusoids have a fenestrating basal membrane. and local humoral mechanisms may stimulate growth. Although the true prevalence of metastatic disease is unknown. metastatic disease is 20 times more frequent than primary liver malignancies. The commonest tumors with hepatic metastases are carcinomas of the colon (approximately 40%). stomach (>20%). pancreas (20%). breast (approximately 10%). and lung (approximately 10%). Metastasis can occur anywhere in the liver. but for unknown reasons the right lobe is more frequently affected. The great majority of hepatic metastases are multifocal (>90%). Although the neovascularity in malignancies is derived primarily from hepatic arterial branches. from an imaging standpoint liver metastases are commonly described as either hypovascular or hypervascular, relative to the surrounding parenchyma (Table 11.9). This will guide the most optimal CT technique to be used. Table
3. Tumor Show all lesions
• Segmental •
Size
Ascites factors
Central tumor location ; •
Compression,
thrombosis
or infiltration
hepatic veins, portal vein, hepatic • Bile duct obstruction •
Growth
6.
Prognostic criteria Distant metastases
•
Regional lymph node metastases Number of intrahepatic lesions complicating factors (see Point 5)
•••
beyond
organ
boundaries
tumors
of liver metastases
(carcinoid)
Islet cell tumors (insulinoma/gastrinoma) Malignant pheochromocytoma Thyroid carcinoma Choriocarcinoma/ovarian carcinoma Renal cell carcinoma (RCC) Adenocarcinoma (usually peripherally vascular) Breast carcinoma (usually peripherally vascular) Melanoma
• Cholesrasis 5. Complicating
Vascularization
Neuroendocrine
location
measurement
4. Functional status of the rest parenchyma Volume of rest parenchyma Signs of cirrhosis •
11.9
of IVC
hyperhyper-
Sarcoma
arteries
j
Hypovascular
lesions
Adenocarcinoma (Gltract. lungs) Breast carcinoma Squamous cell carcinoma (head and neck, anus) Lymphoma
lungs,
437
11 The Liver Most liver metastases are hypovasciilar and for these metastases it is generally accepted that scanning during the portal venous phase is optimal for tumor detection. Routine use ofbiphasic CT is not indicated, but addition of the arterial phase may be beneficial in selected cases such as the depiction of the hypervascular rim for characterization in colorectal metastases. It must. also be noted that independent of the hypo- or hypervascular nature of the primary tumor, in small lesions ( 1cm) the primary blood flow is supplied by the hepatic artery and arterial phase CT may have benefit in improving detection. Thus, it may be beneficial to use biphasic CT in the initial examination or before surgery. Also, liver perfusion studies showed higher risk for patients with increased arterial flow patterns, so that early perfusion imaging may be integrated with portal venous morphologic imaging. Delayed imaging is beneficial for differentiation against benign lesions. Metastases from hypervascular tumors, such as renal cell carcinoma, neuroendocrine tumors (islet cell, carcinoid), thyroid carcinoma, melanoma, sarcoma and breast carcinomas are hypervcisciilar (Table 11.10). However, hypervascular primaries also produce metastases that are hypovascular in comparison with the liver. Multiple studies have shown that biphasic CT is the recommended imaging technique for optimal detection and to assess tumor recurrence in hypervascular lesions treated with chemother-
apy. Although breast carcinoma metastases can be hypervascular, studies did not show added benefit of routine use of biphasic CT over portal phase CT. Routine use of the noncontrast phase is controversial. Some studies in melanoma or hypervascular metastases showed its use to be equivalent or better than arterial phase imaging when added to portal phase imaging and it has also been advocated as a third phase in carcinoid tumors. Other groups have advocated the addition of delayed imaging for improved characterization of lesions with rapid wash-out. For most optimal characterization, unenhanced plus biphasic (or even triphasic) CT is probably the best option, especially for the initial CT examination. Dynamic MRI is beneficial in (inhomogeneous) fatty livers and for the discrimination of small hemangiomas versus hypervascular metastasis. In presurgical evaluation, optimized detection can be achieved with angiographically assisted techniques at the expense of specificity. For hypovascular metastases, CTAP is used (Fig. 11.24), while in hypervascular metastases a combination of CTAP and CTHA gives the best results. One has to keep in mind, however, that pathologic studies have shown that for every metastasis > 1 cm there are on average 1.4 metastases •.; 1 cm in patients with colorectal cancer and up to 4 metastases - 1 cm in patients with other tumor entities. Thus, imaging modalities will only detect part of the total.
Table 11.10 Detection rates of multiphasic he Study
Primary
Hypovascular metastases: Scott. 2001 colon Vails. 2001 colon lnaba, 2000 colon Bluemke, 2000 mxed Ward. 1999 colon Vails, 1998 colon
Gold Standarc
NP
CR, S, H. IOUS CR. S, H, IOUS
HAP
PVP
HAP+PVP NP+PVP CTAP
67%
70% 85%
75%
CR, S, H CR. S, H, IOUS CR. S. H, IOUS CR, S. H, IOUS
Hypervascular metastases: Blake, 2000 melanoma CR,O CR, H/CFU Sheafor. 1999 breast Frederick, 1997 breast CR Oliver. 1997 CR, H (1 lesion) mixed
MRI+SPIO
87% 67% 74%
80% 81%
76%
72% 96% 61% 87%
84% 97% 59% 78%
86% 97% 85% 77%
74%
96%
96%
87%
96%
Gold Standard: S = surgery/palpation. H = histology. IOUS = intraoperativeUS, CR = consensus reading. CFU = clinical follow-up, Q= questionnaire
Metastases Fig. 11.24 Compared with portal phase conventional CT (a), CTAP (b) demonstrates significantly more small metastases from a colorectal carcinoma. Note the lack of enhancement of the inferior vena cava in CTAP (arrow).
a
•
CT Morphology
On iinenhcinced CT, a substantial numbers of metastases are indistinguishable from normal liver. Most. however, are hypoattenuating to the liver parenchyma. Lesions may become isoattenuating and may be missed if there is slight fatty infiltration of the liver. Greater fatty infiltration can reverse the usual density pattern, causing metastases to show higher attenuation than the surrounding liver (see Fig. 11.6a). Small lesions are nodular and homogeneous, while larger lesions are more irregular, heterogenous and with ill-defined margins. Calcification is common in mucinous CI tract carcinomas and in chemotherapeutically treated hypervascular lesions (e.g., carcinoid, islet cell tumors). After contrast administration, metastases from hypovasciilar tumors may have an enhancing rim that can be seen during the arterial phase and occasionally during the portal phase.
b
This 'target' sign is considered pathognomonic (Fig. 11.25). Lesions that do not have an enhancing rim show little if any enhancement during the arterial phase and become hypo attenuating during the portal phase. Hypervascular metastases show moderate to intense enhancement during the arterial phase (Figs. 11.26. 11.27) that may persist in the portal phase. However, most show a rapid wash-out and become isodense or hypodense to surrounding liver. Lesions with central necrosis also show the typical "target" appearance. The quicker the examination is performed during the desired phase of enhancement, the better the resul t in terms of lesion detection and discrimination. If scanning is performed too late, the detection rate will be lower because of the diffusion of contrast medium into the lesion ("vanishing" lesion).
Fig. 11.25 Hypovascular metastasis from gastric carcinoma, (a) A faintly enhancing rim is visible during the arterial phase, (b) The lesion is most conspicuous during the portal phase, and a second lesion is also seen (arrow).
a
b
439
11 The Liver Fig. 11.26 Recurrent hypervascular metastases from medullary thyroid carcinoma in a patient after a right hemihepatectomy. Most of the lesions are visible only during the arterial phase (a), but some lesions are not visualized until the portal phase (b).
a
b
a
b
c
Fig. 11.27 Metastases from a carcinoid tumor appear iso-, hypo- and hyperattenuating on the noncontrast scan (a). During the arterial phase (b) and portal phase
(c) a typical variegated pattern is seen, consisting of cystic areas, hypervascular lesions (arrows), and hypovascular lesions.
Detection rates (per lesion basis) of recent studies that specify numbers are summarized in Table 11.10. It must be noted that many studies include focal liver lesions in general, the accuracy in small lesions 1 cm is often considerably lower, and results on a per patient basis are generally better. MRI with liver-specific agents (SP10) often shows better results than helical CT and has a sensitivity equivalent to CTAP but an improved specificity. Thin-section multislice CT or the combination of SPIO with Gel-enhanced MRI may improve techniques even further.
retain their hyperenhancement during the portal venous and delayed phase. Differentiation from FNH may be difficult when the FNH does not show a central scar. Hypovascular metastases can be distinguished from focal fatty infiltration by their different enhancement pattern: metastases tend to become isoattenuating to surrounding liver over time. Necrotic lesions with rim enhancement require differentiation from abscesses, which have a center with lower attenuation, and from cholangiocarcinoma. In all problematic cases, delayed imaging often helps for differentiation.
•
Differential Diagnosis
Hypervascular metastases are to be differentiated from small hemangiomas which usually
Hepatocellular Carcinoma (HOC) 441 Metastases from very hypervascular tumors such as carcinoid, renal cell carcinoma, thyroid carcinoma, and insulinoma may be most conspicuous on noncontrast CT scans, The tumor mass may be slightly underestimated on contrast-enhanced scans, Postcontrast scans of carcinoid metastases often show a heterogeneous mix of cystic components with hypovascular and hypervascular regions (Fig, 11.27),
Hepatocellular Carcinoma (H CC) Hepatocellular carcinoma (HCC) is one of the world's commonest malignancies and the most common primary hepatic tumor, accounting for 80-90% of cases, In children, HCC is the second commonest tumor after hepatoblastoma. The incidence shows considerable geographic variation, mirroring that of chronic hepatitis Band C The annual incidence is low in most of Western Europe, the USA and Australia (1-3/100000), intermediate in Spain, Italy and Argentina (S-IOI 100000), high in Japan and Greece (10-301 100000) and very high in Southeast Asia, China, and sub-Saharan Africa (30-lS0/100000), Male to female ratio is 4-8: 1 and peak occurrence is in the fifth to seventh decades, In patients under 30, the tumor is SOO-fold more common in areas like Mozambique as compared to USA and Europe, Hepatic cirrhosis is the greatest risk factor and underlies HCC in 60-90% of cases, The annual incidence in cirrhotic patients is about 3-S%, While in the USA and Europe the most common under lying cause is alcohol, in Asia and Africa it is hepatitis B or aflatoxlns. Hepatitis B, endemic in Southeast Asia, China, and subSaharan Africa, is a well established risk factor, both in patients with a history of HBV and in chronic carrier states, Hepatitis C, endemic in Southern Europe and Japan, increases risk by promotion of cirrhosis but carriers are not at increased risk, Synergistic effects with hepatitis B have been described, The risk is also increased in persons exposed to chemical carcinogens (aflatoxin, cycasin, Thorotrast), and in patients with metabolic disorders (hemochromatosis, Wilson's disease, tyrosinosis, alpha-l-antitrypsin deficiency, etc.).
HCC may arise de novo in a noncirrhotic liver, but the majority of HCC in cirrhotic livers are believed to arise in a stepwise fashion: regenerating nodule-low-grade dysplastic nodule-high-grade dysplastic nodule-dysplastic nodule with focal HCC-small HCC and (large) HCC (Table ILl-I), Histologically, HCC are graded from well-differentiated to poorly differentiated according to Edmondson, Along with progression of the tumor, nodular lesions show a change in blood supply, Regenerative nodules and dysplastic nodules with or without foci of HCC have a predominant portal blood supply, while (small) HCC has a preferential arterial supply, There is, however, a substantial percentage of small HCCs in which portal supply predominates, Small HCC are usually nodular and can be divided in four types: Single nodular, nodular with extranodal growth, contiguous multinodular, and poorly demarcated nodular types, Large HCC are classified as expansive nodular, infiltrative, and diffuse types, Multinodular forms may consist of multiple small nodules or present as a dominant mass with small, daughter nodules, Well-differentiated tumors are frequently expansive, nodular and enclosed in a fibrous
Table
11.11 Terminology (lWP, 1995)
of hepatocellular
nodules
cirrhosis
New tenninology
Old terminologies
Regenerative nodule
Regenerative
Dysplastic
nodule,
low-grade
nodule
Macroregenerative Adenomatous
Dysplastic
nodule,
type I nodule,
high-grade
hyperplasia
Macroregenerative
nodule,
type II Adenomatous hyperplasia with atypia Borderline hepatocellular
lesion
Dysplastic nodule with focus ofHCC
Macroregenerative
nodule
with microscopicHCC Adenomatous hyperplasia with microscopicHCC
Early HCC Small
HCC «2
HCC(>2cm)
em)
Adenomatous hyperplasia with macroscopic HCC Early advanced HCC Advanced Hepatoma
HCC
in
77 The Liver
capsule. These encapsulated forms are found much more freq uently in Asian than in nonAsian populations and have a better prognosis. Large tumors tend to necrose or bleed centrally but no central fibrosis is present. HCC has a tendency for vascular invasion. more frequently in the portal vein than in the hepatic veins. Biliary invasion is seen less frequently. Lymphogenous spread occurs in the following order of frequency: porta hepatis > hepatoduodenal ligament> pancreatic bed > mesenteric root> mammary lymph nodes. A large tumor may directly invade the diaphragm. abdominal wall. or pancreas. Hematogenous metastasis can occur to the lung « 10%). peritoneum. adrenal glands. and bone. HCC is often diagnosed in advanced stages. with patients complaining of right upper quadrant pain. mass. fever. and weight loss. Asymptomatic cases can be diagnosed by screening risk populations with ultrasound and alphafetoprotein (AFP) measurements. AFP is elevated in 60-70% of HCC in USA and Europe and in 80-90% of HCC in Asia. False positives may be due to chronic hepatitis. cirrhosis. and pregnancy. Clinical staging is done by multiphasic CT or MRI using the TNM-classification (Table 11.12). Prognosis depends on the clinical stage. tumor histology. grade of cirrhosis (Child-Pugh classification) and patient's general condition (Karnofsky scale). Curative therapy is reserved for selected patients with small tumors and good hepatic function. The five-year survival rate is in the order of 30% after curative surgical therapy. Multiphasic CT has a role in the detection and staging of tumors, the exclusion of multifocal disease, and to select candidates for curative surgery. For planning surgical or embolotherapy, precise description of arterial variants helps facilitate these interventions. For confirmation of the diagnosis, MRI or angiographically assisted techniques like CTHA, CTAP or Lipiodol CT may be used. The Asian variant of HCC appears to differ from the Caucasian, tending to show typical patterns of Lipiodol storage more often than the variant seen in Caucasians. Nowadays, with improvements in CT and MRI technology, these invasive techniques are used less frequently and may be reserved for cases amenable to curative therapy. In surgical candidates, needle biopsy of HCC is controversial due
Table 11.12 TNMdassificationofprimarylivertumorsHCCandCCC(UICC, 1997)
TIr\M.I1
Description
T1
Solitary sion,
T2
tumor
without
Solitary
/\ 2 em in greatest vascular
tumor
2 em in greatest
<
dimen-
invasion dimen-
sion, with vascular invasion, or solitary tumor> 2 em in greatest dimension, without vascular invasion, or multiple tumors limited to one lobe, none >2 em in greatest vascular invasion T3
dimension,
without
Solitary tumor >2 em in greatest dimension, with vascular invasion, or multiple tumors limited to one lobe, none >2 em in greatest dimension with vascular mvasron. or multiple tumors limited to one lobe, any more than 2 cm in greatest dimension with or without vascular invasion
T4
Multiple tumors in more than one lobe, or tumor(s) involve a major branch of the portal or hepatic vein(s), or tumor(s) with direct invasion of adjacent organs otherthan gallbladder, or tumor(s) with perforation ofthe visceral peritoneum
N1
Regional
lymph
duodenal M1
Distant
node metastasis
(hepato-
nodes)
metastasis
at locations ligament)
outside
(including
lymph
nodes
the hepatoduodenal
to the risk of bleeding, tumor rupture, or peritoneal metastasis. Candidates for palliative therapy may undergo fine needle aspiration cytology. CT is also the primary diagnostic modality in the follow-up of patients after surgery or ablative therapies. •
CT
Morphology
The CT features of hepatocellular carcinoma are highly variable, depending on the size, vascularity, histologic composition, and growth pattern of the tumor. It is generally accepted that CT with images in the arterial and portal venous phases are the most useful for detection. Few tumors may only be seen on unenhanced scans. Delayed imaging may be helpful for characterization in hypervascular HCC. On unenhanced CT scans, most HCC present as solitary or multiple hypo- to isoattenuating masses. Calcification is seen in 5-10% of cases. Areas of necrosis or fatty metamorphosis appear as hypoattenuating foci, but fatty metamorpho-
Hepatocellular Carcinoma (HOC)
a
b
Fig. 11.28 Typical CT features of HCC. A heterogeneous hypervascular lesion (a) that may contain fat (arrow) and may show delayed enhancement in a tumor
sis is rare in non -Asian populations. In the presence of diffuse steatosis, the tumor may be undetectable or even slightly hyperattenuating. In small HCC, the nod ular type is often sharply demarcated from the liver and has a tumor capsule in 50% of lesions. This is often thin and may be difficult to see at CT. The other types show nodularity with ill-defined and irregular margins. In large HCC, the expansive nodular type is sharply demarcated. It is encapsulated in 70-80% and shows typical internal mosaic architecture. Infiltrative types are often segmental, heterogeneous, and with ill-defined margins. Strands of tumor frequently invade portal vein branches. The diffuse type is the least common and shows numerous small nodules scattered throughout the liver. All large HCC have a tendency for developing satellite nodules that represent intrahepatic metastases. In hypervasciilar tumors, arterial phase CT demonstrates feeding vessels and intense, enhancement throughout the tumor (Fig. 11.28). Larger tumors often are heterogeneous due to necrosis and hemorrhage. Especially with multislice CT, tiny hypervascular nodules may be detected (Fig. 11.29). If present, the capsule remains hypoattenuating. During the portal venous phase, HCCs wash out rapidly and become isodense to hypodense. relative to normal liver. Larger tumors are generally heterogeneous (mosaic pattern). On delayed phase images, the capsule and fibrous septa are relatively hyperattenuating and show prolonged enhancement
c capsule (b) ortumors associated with portal vein thrombosis (c) are suggestive of HCC. Note the contrast uptake in the tumor thrombus (arrows).
Fig. 11.29 Small hypervascular lesions in a cirrhotic liver (arrows) are characteristic of HCC. Such lesions are frequently multifocal and can be seen only in the arterial phase (here: 4 x 1/6. VRT).
while the tumors are iso- to hypoattenuating to the liver (Fig. 11.28). The diffuse type may not be detectable by imaging studies (including CTAP). and many HCCs are very poorly visualized wi thin a cirrhotic liver. Hypovascular tumors are not uncommon and usually represent early stages like dysplastic nodules with focal HCC orwell-differentiated small HCCs (Fig. 11.30). Often they are best seen on noncontrast scans. They show l ittle if any enhancement during the arterial phase. They may also be poorly visualized on later scans, with isodense to hypodense appearance depending on their portal blood supply.
443
11 The Liver
a
b
Fig. 11.301 HCC nodules are usually hypervascular (arrows) and ofteru only visible during I thearteriall phase (a) .. Hypovascular lesions" which are seen more' clearly during the portal phase, are less common.
Thrombosis of portal venous branches occurs in up to 40% of HCCs and is frequently caused by direct tumor invasion (Fig. 11.31). These tumor thrombi are only moderately hypoattenuating and enhance irregularly after contrast administration, especially in the arterial phase. Tumorous arterioportal shunts, which are characteristic of HCC, show early enhancement of intrahepatic portal branches on arterial phase images (Fig. 11.32). They may be associated with transient perilesional segmental or wedge-shaped areas of hyperenhancement (transient hepatic attenuation differences, THAD, Fig. 11.32). Large, proximal shunts can
a
b
a
b
Fig. 11.31 Tumor thrombus in the portal vein may demonstrate marked arterial enhancement (a) and become hypoattenuating in the portal venous phase (b).
Fig. 11.32 Hypervascular HCC withiai centrall hypodensitymimicking FNH. HCCis.highly likely becauseofarterio-porta] shunting, .which can Ibediagnosedl from uransient hyperattenuation of the perilesional liver parenchyma (arrowheads) (a) or from locally increased enhancement of peripheral portal branches despite substantially less enhancement of the central portal branches (arrow) best demonstrated on a axial thin-slab MIP(4x1/6)(b).
significantly alter perfusion and may lead to imaging features of portal hypertension with esophageal varices. The invasion of hepatic veins (approximately 15% of cases) is less common and may cause occlusion leading to a Budd-Chiari syndrome (see Fig.ll.55c). This occlusion leads to increased perfusion of liver parenchyma and lower attenuation of HCC in arterial dominant phase. Spontaneous rupture ofHCC «10%) is
c Note the hypotrophy of the right and hypertrophy of the left liver lobe. Liver vein infiltration (arrow) can progress to tumor thrombus in the vena cava (c).
Hepatocellular Carcinoma (HOC) 445 creased arterial flow and occurrence of pseudolesions. Detection rates (per lesion basis) of recent studies summarized in Table 11.13.The Peterson study is the only one with explantation results for both positive and negative imaging examinations and thus reflects the true sensitivity in screening situations. It must be noted that many studies deal with focal lesions in general; the accuracy in small lesions < 1 cm is considerably lower and results on a per patient basis are generally better. •
Differential
Diagnosis
In cirrhotic livers, typical hypervascular tumors with capsule and mosaic architecture represent Fig. 11.33 Spontaneous rupture of an HCC (arrow) HCC. If atypical findings are found, differential causes intraperitoenal hemorrhage (a). Fresh extravasadiagnosis includes hemangioma and metastasis, tion can be appreciated if bleeding is active (b). Adenoma, large FNH, or hemangiomas may cause similar but hypervascular metastases in this clinical setfindings. ting are rare. Tumors in non cirrhotic livers are more advanced and differential diagnosis is more difficult. HCC should be distinguished life-threatening and requires emergent treat- from metastases and other rare primary liver ment. It is manifested by hematoma (Fig. "11.33), tumors but also from benign lesions like FNH, lesion surface discontinuity, or separation of HCA or hemangioma. Even needle biopsy is tumor content from a peripheral enhancing rim often unrewarding, since the cytologic and histologic features may vary within the lesion and that may rupture intraperitoneally ("enucleaare very similar for all four diagnoses. Lipiodol tion sign"). Imaging accuracy is often more limited in CT may be helpful in larger tumors (Fig. 11.34). the cirrhotic liver because of architectural distortion with numerous regenerating nodules as well as altered portal hemodynamics with ina
b
Table 11.13
Detection rates of multiphasic
H E3i~/""'j"""j"""j"f
!
helical CT and CTAP/CTHA in the diagnosis
1'1'1'111IU11~~,~~,~:ij~'jIlH'JI!~M1.J11:Jpliasi<MUltiphasicCT AngiographicallyassistedCT'" ANI
MMiP AWP DPP
~-"";--'''----'';lffii!iiOf)lsS6tndaHtd
3 Phases
%%~~A,,_
;ASags®safB' WXf &
A. A.
30
S,IOUS
Peterson, 2000
430
E
69%1
Jang,2000
73
SorB
94%2
Makita, 2000
53
S, B, IOUS
Kirn, 1999
85
SorB,RFU
Hori,1998
176
RFUJOUS
1997
66
77%
57%
ClHA+
Lipiodol
96%
94%
99%
88%
85%
95%
14% 93%
E
85%
SorB
82%
74%
Baron, 1996
SorB
95%
82%
1 NP + HAP + PVP: prospective 2 HAP + PVP + early DP 3 NP + HAP + PVP
ClAP
91%
Kanematsu, 1997 100 326
GlHA
_:'";;.,,,,;;;:,).,;;;,.;,:.:,.s;,""
Choi, 2001
Spreafico,
of HCC (per lesion basis)
detection
31-67 days prior to transplantation,
Gold Standard: E == Explanted liver (transplantation), RFL) = Radiological follow-up
58%
87%
retrospective
rate: 85%
B = biopsy, S == surgery, IOUS = intraoperative
US,
11 The Liver
The role of CT is tumor staging and the presurgical evaluation of resect ability. •
Fig. 11.34 Lipioclol CT shows homogeneous retention of the multifocal recurrences of HCC the metal clips following the nodule.
pathognomonic, largely oily contrast material in in an Asian patient. Note resection of a solitary
Fibrolamellar Carcinoma (FLC)
Fibrolamellar carcinoma is a rare malignant neoplasm that is now seen as a separate entity. It accounts for 2% of all malignant liver tumors and there are no known risk factors. It is seen in young patients (5-35 years) without cirrhosis and there is no male preponderance. AFP levels are normal, but other markers like vitamin 812 or copper-binding proteins may be elevated. Histologically, FLC is a large mass made up of lamellar fibrous strands that separate markedly eosinophilic hepatocytes in cords. Centrally, a stellate fibrous scar is present. Hemorrhage and necrosis are infrequent features. FLC has a more favorable prognosis and is resectable in 50-75% of patients. Five-year survival is in the order of 60%.
CT Morphology
On noncontrast CT, FLC usually presents as a large, solitary, well-defined mass measuring 4-17 ern, but satellite nodules may be present. The tumor has lobulated contours and is usually hypo attenuating relative to liver. Centrally, a more hypo attenuating stellate central scar can be visualized that shows nodular or stellate calcifications in 55% of cases. After contrast administration there is moderate to marked enhancement of the peripheral part of the tumor, becoming iso to hyperdense compared to liver (Fig. 11.35). There is no enhancement of the scar, but contrast may slowly diffuse into the scar giving hyperattenuation on delayed phase images. Intrahepatic ductal dilatation and portal vein invasion may be seen, but the latter is much less frequent than in HCC. Cholangiocellular Carcinoma (CCC)
The rare intrahepatic (peripheral) cholangiocarcinoma accounts for 10% of primary hepatic tumors and is the second commonest tumor after HCC. Cholangiocarcinomas should not be confused with bile duct carcinomas (extrahepatic Cholangiocarcinomas), and comprise only 10% of all malignant bile duct cancers. Risk factors include a history of Thorotrast administration, primary sclerosing cholangitis, congenital fibrosis, and infection with Clonorchis sinensis (Chinese liver fluke). Histologically, the tumor has a large central core of fibrous tissue with adenocarcinoma cells
Fig. 11.35 Fibrolamellar carcinoma (FLC) is hypervasrular, similar to HCC (a). It may contain fat and is characterized by a central scar that not always contains calcifications (b).
a
b
Cholangiocellular Carcinoma (CCC)
a
c
b
Fig. 11.36 Varying CT appearances of cholangiocarcinoma: (a) diffuse type with disseminated nodules (may have the same appearance as HCC or metastases); (b) nodular tumor with peripheral hypervascularity (differ-
ential diagnosis: metastases); and (c) nodular hypovascular tumor (differential diagnosis: metastases). Fibrous retraction of the hepatic capsule (arrowhead) and small calcifications (arrows) may be observed.
at the periphery. The tumor frequently produces mucin. A nodular type (95%) is distinguished from a diffuse type (5%). As these tumors are usually clinically silent, most are diagnosed at an advanced stage. CT is used for preoperative staging of the nodular type; the diffuse type is difficult or impossible to detect with imaging procedures. •
CT Morphology
Cholangiocarcinoma appears on noncontrast CT scans as a large round or oval mass of low attenuation with irregular margins. Satellite lesions are common (65%) but often are poorly visualized. Centrally, stippled areas of hyperattenuation represent small mucinous substances. Irregular or punctate calcifications may be seen (in 20% of cases). Capsular retraction is typical but is not often visible. Mild, segmental bile duct dilatation is a common finding. Most tumors are hypovascular and show mild rim-like enhancement at the periphery (Fig. H.36) in both arterial and portal venous phases. Progressive tumoral contrast enhancement in the late delayed phase (W-15 minutes) is typical and can aid in detection and characterization (Fig. 11.37). Portal venous invasion is frequent. Some 50% of tumors show hyperenhancing THADs during the arterial phase because of an increased arterial flow to the liver segment.
Fig. 11.37 On interstitial phase images, some cholangiocellular carcinomas (CCC) demonstrate delayed enhancement.
•
Differential
Diagnosis
CCC have to be differentiated from metastases, most commonly of adenocarcinomas. This is not possible by imaging techniques but rather requires histologic proof. Capsular retraction and coarse calcifications may also be found in epithelioid hemangioendothelioma or in metastases from colorectal cancer.
11 The Liver
Hepatoblastoma
Hepatoblastoma is the commonest malignant hepatic tumor in children (>50%). It is most prevalent in children under 3 years of age but may affect children up to age 15. It is more common in patients with Beckwith-Wiedemann syndrome, Wilms' tumor, glycogen storage diseases, and Meckel's diverticulum. The tumor is composed of tissue that resembles fetal hepatocytes and mesenchymal cells. Pure epithelial types are homogeneous, while the more common mixed type may have areas of hemorrhage, osteoid, fibrosis, or cartilage. Alpha-fetoprotein levels are markedly elevated in two-thirds of patients. •
CT Morphology
Unenhanced scans show a solid, hypoattenuating mass that may occupy large portions of the liver. Around 20% are multifocal. The mixed type frequently contains necrotic areas, fibrotic bands, and coarse calcifications (30-50%). Arterial and portal phase images show a hypervascular mass that is often heterogeneous (Fig. 11.38). After chemotherapy, nodules may demonstrate hemorrhagic necrosis (Fig. 11.39). •
Fig. 11.38 Hepatoblastoma, appearing as a hypervascular mass containing areas of necrosis, in a 5-yearold boy.
Differential
Diagnosis
The differential diagnosis includes infantile hemangioendothelioma (fine granular calcifications, multiple hemangioma-like vascularized foci), metastatic neuroblastoma, mesenchymal
a Fig. 11.39 Hepatoblastoma in a 15-year-old girl after chemotherapy. There are multiple necrotic and hemorrhagic tumor nodules (arrows) in segments II to IV (a). A coronal reformation of the portal phase scan demon-
hamartoma (multiple cystic spaces of varying size and density, moderate contrast enhancement), and fibrolamellar HCC (age> 5 years). Undifferentiated Embryonal Sarcoma
Undifferentiated embryonal cell sarcoma is a rare malignant mesenchymal tumor found in older children and young adults (90% of patients are under age 15, with rare cases occurring up to age 30). It is a large solitary mass composed of spindle-shaped sarcomatous cells packed in whirls around a myxoid ground substance, and may contain hematopoietic foci. It is believed to
b strates an additional nodule (arrowhead) in segment VII that was treated by wedge resection (b). The enlarged hypervascular lymph node (wide arrow) proved to be hyperplastic at histology (c).
Epithelioid Heinangioendothefjoma
be the malignant variant of mesenchymal martoma. •
ha-
CT Morphology
CT scans usually show a very large mass (1025cm) located predominantly in the right lobe of the liver (75%). A near-water attenuation is typical (Fig. 11.40) and can be found not only in regions of necrosis and intratumoral hemorrhage but also in areas of solid tumor. The mass is sharply demarcated from surrounding liver by a fibrous pseudocapsule. The tumor parenchyma in the periphery of the lesion may show intense enhancement on arterial phase images, and there typically are hyperattenuating areas within the hypo attenuating central tumor portions (on CT very similar to cystic renal cell carcinoma) . •
(EHE)
Differential Diagnosis
Mesenchymal hamartoma may be distinguished from undifferentiated embryonal sarcoma by its enhancing soft -tissue component and lack of a pseudocapsule. Rhabdomyosarcoma may show extensive tumor necrosis. Epithelioid Hemangioendothelioma (EHE)
EHE is a rare solid vascular tumor which nowadays is considered to be a separate entity with low-grade malignant potential. Most affected patients are women between 20-40 years of age in whom the tumor is usually discovered incidentally. The tumor begins as multiplenodules ofepi thelioid -appearing endothelial cells that coalesce into large masses that are located in the periphery of the liver. Due to slow
Fig. 11.40 Non-d ifferentiated embryona I sarcoma is hard to distinguish from a cystic mass. However, even the cystic appearing portions are usually solid.
growth, compensatory enlargement can be seen in the non-affected liver parenchyma. •
CT Morphology
Noncontrast CT shows multiple well-defined, hypo attenuating lesions. Calcifications are frequently present. The lesions may have coalesced into larger tumors. In a subcapsular location retraction phenomena can be seen. After contrast administration, the hypervascular peripheral portion may enhance to the same amount as surrounding liver tissue. If an outer avascular zone is present, a characteristic target appearance can be seen (Fig. 11.41). When absent, contrast -enhanced CT may underestimate the tumor size. For this reason, noncontrast scans provide better tumor conspicuity and evaluation of tumor extent and should always be included in the work-up.
Fig. 11.41 Epithelioid hemangioendotheliomas present as a rnultifocal. mostly subcapsular hypovascular lesions.
a
b
77 The Liver
Angiosarcoma
Angiosarcoma is an extremely rare primary vascular hepatic tumor derived from endothelial cells. It is more common in adults that received the contrast agent Thorotrast in the years 19281950 (latent period > 15 years), and among patients with an exposure to inorganic arsenic or polyvinylchloride (latent period >4 years). An association with hemochromatosis and Recklingshausen's disease is also known. The tumor is highly aggressive and metastasizes early to the lung, spleen, and regional and peritoneal lymph nodes. Hemorrhagic ascites may occur. Median survival is only 6 months. •
CT Morphology
Unenhanced CT shows multiple or occasionally solitary hypoattenuating lesions with cystic and hyperattenuating components caused by necrosis and hemorrhage. After Thorotrast, strongly hyperattenuating nodular deposits may still be seen in the liver or spleen. Angiosarcomas tend to be very hypervascular and may be scattered throughout the liver (Fig. 11.42). Marked and prolonged peripheral enhancement is usually seen, without complete fill- in towards the center. It is very common to find invasion and tumor thrombosis of the portal vein and hepatic veins.
Angiosarcoma is the only tumor in adults besides cavernous hemangioma that may show peripheral lacunae after intravenous contrast administration. This possibility should be considered in lesions that show a marked increase in size over time. Necrotic and hemorrhagic areas and vascular involvement are helpful in suggesting the correct diagnosis.
Other Mesenchymal Tumors
Other sarcomas are even rarer, and include leiomyosarcoma or malignant fibrous histiocytoma. CT is nonspecific and shows large hypodense masses without calcifications. There is heterogeneous peripheral contrast enhancement. Lymphoma
Primary hepatic lymphoma is very rare and hepatic lymphoma is usually seen in association with systemic disease, both Hodgkin's disease and NHL. In advanced stages these are commonly associated with hepatic involvement (up to 60% in stage IV). There is however a rise in incidence due to growing numbers of AIDS patients and liver transplant recipients who are also at increased risk. Diffuse infiltration of the liver is more common in secondary forms, while large focal nodular disease is the most common presentation of the primary form. •
Fig. 11.42 Angiosarcoma with typical hypervascular portions that may simulate giant hemangioma. Aggressive behavior with infiltration of vascular structures is typical (arrows).
CT Morphology
Diffuse involvement is not detectable on CT scans. Focal lesions are isoattenuating and occasionally hypo attenuating on unenhanced scans, and most are poorly visualized. Portal phase images occasionally demonstrate multiple small hypo attenuating lesions, which may have ill-defined margins (Fig. 11.43). Large lesions frequently display an irregular contour. Primary lymphomas often appear hyperattenuating on delayed scans after contrast administration (due to the diffusion of contrast into the tumor interstitium). Concomitant abdominal lymph node enlargement, if present, will help to suggest the correct diagnosis.
Fatty Infiltration Table
11.14
Criteria
for differentiation
of predominantly
solid
liver lesions
(in order
of frequency)
Non-contrastCT Fat
Regional fatty infiltration> myolipoma
Cystic areas
Metastases (mucoid colon carcinoma, ovarian carcinoma, carcinoid> carcinoma, melanoma, sarcomas) > HCC, adenoma» mesenchymal toma, biliary cystadenoma/cystadenocarcinoma, angiosarcoma, embryonal sarcoma
Central
"scar"
Hyperdense
areas in fatty liver
Hemorrhage
FNH > cavernous
hemangioma
Regional
(no mass
sparing
» angiosarcoma, areas
Calcification
(central
effect)
lipoma,
thrombosis),
> solid
tumors
liposarcoma,
angiobronchial hamar-
FLC > adenoma » hemorrhage
Lipiodol,
undifferentiated
Thorotrast.
metallic
embryonal
foreign
bodies
sarcoma
(surgical
clips)
Granuloma (TBC. histoplasmosis). old hematoma, following cosis, HAE, brucellosis, coccidioidomycosis, toxoplasmosis,
calcification
Disseminated Contrast
adenoma»
Trauma, adenoma» FNH, hemangioma, HCC > metastases (colon carcinoma, RCC, mamma carcinoma, chorionic carcinoma, melanoma, thyroid carcinoma)
Very hyperdense
Timor
HCC,
nodules
abscess, echinococCMV, PC:P, schisto-
Metastases (mucoid carcinomas: colon, mamma, stomach> osteosarcoma> melanoma, mesothelioma, carcinoid, leiomyosarcoma) > CCC, FLC, hepatoblastoma > hemangioma, regenerative/dysplastic nodules (capsule), sclerosing HCC RegenerativeJdysplasticnodules,
metastases,
HCC, CCC, lymphomas
enhancement
Peripheral Arterial
nodular
Cavemous
hemangioma
» angiosarcoma
hypervascularization
homogenous/nodular
FNH. HCC, hypervascularmetastases
inhomogeneous
HCC > hypervascular metastases: RCC, colon carcinoma, mammacarcinoma > endocrine active tumors (carcinoid » insulinoma, gastrinoma, pheochromocytoma) > melanoma, ovarian carcinoma, chorioncarcinoma > hepatoblastoma » angiosarcoma, undifferentiated embryonal
Peripheral
> adenoma
enhancement
Abscess, metastases>
irregular
CCC
Portal hypovascularization CTAP
All lesions,
i.v. contrast media
Furtherexceptions:
Interstit. Vessel
enhancement infiltration
Preferential
except:
Lymphomas, HCC
many small adenomas,
HCCs
lesions
CCC, large fibrous tumor parts or large extracellular
» angiosarcoma,
large
space
metastases
locations
Liver hilum
Biliary cystadenoma/cystadenocarcinoma,
Subcapsular
Hemangioma endothelioma
Gallbladder Segm.IV,
well differentiated
small lesions. hypervascularized
bed
prepartal
Segm. IVb, lig. teres
Gallbladder (THAD)
> metastases, carcinoma,
focal
fatty
hematoma
fatty infiltration/sparing,
Focal fatty infiltration/sparing, Focal
bile duct carcinoma
subcapsular,
CT AP: perfusion
CTAP;
defect
(Klatskin)
» hemangioperfusion
defect
(THAD)
infiltration
Age Up to 3 years
Hepatoblastoma,
Up to 15 years
Hepatotxastoma.
FLC, hamartoma,
years
Adenoma,
FLC, FNH
Up to 40 years
Adenoma,
FNH, FLC, metastases
Up
to
infanWe hemangioendothelioma,
30
Up to 50 years 60 60
years years
FNH, metastases,
adenoma,
Metastases, CCC, HCC Metastases,
HCC, angiosarcoma
undifferentiated
epitheloid
mesenchymal embryonal
hemangioendothelioma
hamartoma sarcoma
451
11 The Liver
and the imaging features resemble lymphoproliferative disorders in other immunocompromised symptoms like AIDS-lymphoma. In PTLD however, the liver is the most common extranodal site. Apart from nodular or infiltrative disease, the liver may also show periportal lymphomatous infiltration which is unique to transplant patients. Other abdominal sites are the spleen, kidneys and small bowel. •
CT Morphology
In the liver three patterns can be seen: Fig. 11.43 Nodular hepatic involvement by NHL. Portal phase scan shows ill-defined hypoattenuating lesions that are indistinguishable from metastases. Concomitant extensive nodular involvement of the spleen suggests the correct diagnosis.
Post-transplant Lymphoproliferative Disease (PTLD)
PTLD occurs at a frequency of 2-5% and is the most common malignancy in patients following solid organ transplantation. It encompasses a spectrum of disorders ranging from lymphoid hyperplasia to (non- Hodgkin's) lymphoma. It represents an unregulated B-cell proliferation which is associated with Epstein- Barr virus infection. Although lymph node involvement is common, extranodal disease is much more frequent
• single or multiple low-attenuation nodules 1-4 em in diameter • diffuse infiltration with geographic or illdefined low-attenuation, and • a porta hepatis mass with extension into the biliary tree and associated hilar adenopathy. The diagnosis can be supported if lesions at other sites are found. The spleen can be enlarged, but nodular involvement of the spleen is less frequent. In the GI tract, the distal j ej un um or proximal ileum is most commonly affected, showing circumferential wall thickening, dilatation, ulceration, or a polypoid mass. In the kidney, unilateral solitary nodular lesions are seen most commonly. Enlarged lymph nodes, which do not show central necrosis, can be found in an intraperitoneal or retroperitoneal location.
Diffuse liver Disease
Except for hemochromatosis, diffuse liver diseases are not an indication for CT scanning. However, CT performed for other indications will commonly show findings that are referable to diffuse hepatic disease. Many of these diseases are associated with an increased incidence of HCC, and (multiphasic) CT should be performed in patients with elevated alpha-fetoprotein levels or equivocal sonographic findings. Fatty Infiltration
Fatty infiltration (steatosis) of the liver can be seen in 7% of abdominal CT examinations. It may result from metabolic causes (diabetes
mellitus, obesity, congested liver) or hepatic exposure to noxious agents (alcohol consumption, steroids, chemotherapy, etc.). Rapid changes in the degree of fatty infiltration (days to months) may be observed. Nonalcoholic steatohepatitis (NASH) and acute fatty liver of pregnancy (AFLP) are special types of hepatic steatosis. NASH is frequently seen in obese women with associated diabetes or hyperlipidemia, but has also been described in other patients. The cause is unclear but accumulation of fat may lead to focal inflammation and necrosis. It has a benign course. AFLP is a rare complication of first pregnancies and may progress to liver failure with encephalopathy and coagulopathy.
Cirrhosis Fig. 11.44 Focal fatty infiltration of the liver may lead to focal tumorlike regions of hypoattenuation (a). Non-steatotic areas in a fatty liver may mimic tumor (arrows) (b). Their site coincides with artifact-prone regions in CTAP.
a
b
•
CT Morphology
Hepatomegaly is common (> 7 5%) and is associated with rounding of the liver contours that is most apparent in the left liver lobe. In normal livers, the hepatic vessels are hypo attenuating relative to the liver parenchyma. As fatty infiltration of the liver commences (10% to 15% fat, see Fig. 11.6b), the hepatic vessels become isoattenuating to the surrounding liver. As the proportion of fat increases further, the normal pattern is reversed and the hepatic vessels become hyperattenuating to the liver parenchyma (see Fig.l1.6a) . If only post -contrast scans are available, fatty infiltration can be diagnosed if the attenuation of the (contrast -enhanced) liver parenchyma remains lower than that of the paravertebral muscles. Another indicator is the enhancement difference between liver and spleen on portal phase scans. If CT numbers of the liver are more than 20 HU lower than those of the spleen, fatty infiltration of the liver is very likely (Fig. 11.44). Incipient fatty infiltration can obscure hepatic lesions on unenhanced CT scans.
Cirrhosis
Hepatic cirrhosis is a generalized response of the liver to chronic hepatocellular injury. It is characterized by an irreversible destruction of the liver parenchyma with fibrosis, steatosis, and regenerative nodules, leading to an abnor-
mal reconstitution of the lobar architecture. Among many causes, chronic viral hepatitis and alcohol are most important. Less frequent causes are metabolic (e.g., hemochromatosis, Wilson's disease, glycogen storage disease IV, aI-antitrypsin deficiency, tyrosinemia), prolonged cholestasis (primary biliary cirrhosis, primary sclerosing cholangitis), toxins, malnutrition, and hepatic venous outflow blockage. Three important morphologic types of cirrhosis are distinguished according to the type of regeneration that occurs: • micronodular cirrhosis (nodules < 3mm): usually alcohol-related; • macronodular cirrhosis (nodules > 3 mm to several centimeters): usually related to chronic viral hepatitis; • mixed type: usually the result of chronic biliary obstruction. Sometimes, a fourth type, septal or posthepatitic cirrhosis is distinguished, but most of the time this is grouped under the macronodular type. Macronodular cirrhosis has a particularly high association with HCC (incidence rate> 40%). Multiphasic CT is used in evaluating patients for possible HCC and can help describe the amount and location of collaterals in patients with portal hypertension. For the follow-up of cirrhosis per se MRI is probably the preferred modality as it can perform these same tasks, and evaluate morphology and iron deposition. •
CT Morphology
Cirrhosis slowly progresses from compensated states to decompensated cirrhosis with compli-
453
11 The Liver
cations. Early cirrhosis shows a normal liver or enlargement of the hilar periportal space anterior to the right portal vein. This is associated with atrophy of the medial segment of the left hepatic lobe (segment IV). There may also be hepatomegaly, which is frequently combined with heterogeneous fatty infiltration. With progression, morphologic changes are common and include atrophy of the right hepatic lobe and left medial segment and enlargement of the caudate lobe and left lateral segment (see Fig. 11.31).The lateral segment is preferentially enlarged in viral causes, while the caudate lobe enlargement is more often seen in alcohol cirrhosis. Regenerating nodules produce a nodular liver contour, but usually the nodules themselves cannot be seen until later stages (Fig. H.45), when they may produce mass effects and show an attenuation difference relative to the surrounding parenchyma. With atrophy, the gallbladder fossa enlarges and fills with fatty tissue. Lymphadenopathy is also commonly seen in advanced-stage cirrhosis. In advanced disease, masses of confluent fibrosis can be seen in up to 15% of patients (Fig. 11.45), especially in cirrhosis from sclerosing cholangitis. These are usually wedge-shaped hypoattenuating areas radiating from the porta hepatis, although peripheral bands or total lobar involvement may also be seen. The preferential location is the medial segment of the left lobe and the anterior right lobe. There is often capsular retraction at the liver surface. Secondary extrahepatic changes of cirrhosis include splenomegaly, portal hypertension, portosystemic collateral vessels, varices, and ascites. Postcontrast scans in later stages of the
disease show inhomogeneous arterial enhancement and decreased enhancement of the portal vein and the liver parenchyma. This can hamper the search for HCC nodules and will lead to substantial perfusi on artifacts on CTAP scans. Areas of massive hepatic fibrosis are hypoattenuating on noncontrast scans as well as in the arterial and portal venous phases. The scars only enhance in the interstitial phase due to delayed diffusion of contrast medium back into the bloodstream (Fig. 11.45).
Regenerative Nodule (RN)
Regenerative nodules are normal structural features in hepatic cirrhosis and may also be seen after partial hepatic resections, regional chemotherapy, or liver trauma. They lack a true capsule, but they do contain venous vessels and bile ducts. RNs can be small (micronodular) enlarge (macronodular), but combinations occur as well. The blood supply is invariably portal, but due to increased arterial flow they can be hyperattenuating during CTHA. 25% contain iron (siderotic nodules) and are at an increased risk for developing HCC. •
CT Morphology
Most regenerative nodules are poorly defined on unenhanced CT scans, appearing isoattenuating or slightly hyperattenuating depending on the fat content of the parenchyma. Siderotic nodules may be relatively hyperattenuating. A mass effect may be seen. The nodules may be
Fig. 11.45 Hepatic cirrhosis leads to nodular changes in the liver parenchyma (arrows) that demonstrate similar contrast enhancement to liver parenchyma. Areas of massive fibrosis are hypoattenuating on pre-contrast (a), arterial or portal phase images (b) but become hyperattenuating on delayed phase images (c).
a
b
Hemochromatosis isodense or slightly hyperdense on contrast-enhanced scans, depending on their perfusion,
Dysplastic
Nodule
(ON)
Dysplastic nodules are found in 15-25% of cirrhotic livers and are considered premalignant. They are usually somewhat larger than RNs (1020mm), can have a bulging surface and are histologically divided in low-grade or high-grade types, depending on the cellular atypia present. Blood supply is usually from portal venous branches, but a minority have an arterial supply (better seen on CTHA) , Some may harbor small foci ofHCC and these lesions may be better visible on MRI showing a typical nodule-withinnoclule phenomenon, Siderotic DN that accumulate copper and iron can be seen with MRL •
CT Morphology
On nonenhanced scans, dysplastic nodules are often poorly defined, Following contrast administration, most DN are isodense to hypodense during all phases, Portal and hepatic veins may be visible in large nodules, The minority with arterial blood supply usually do not become hyperattenuating when intravenous contrast is used, but may become hyperattenuating when angiographically assisted CT techniques are used, Additional imaging is often required to distinguish these from well-differentia ted HCC, Dysplastic gutshable HCC.
Primary
nodules are frequently from well-differentiated
Biliary
indistin(small)
Cirrhosis
Primary biliary cirrhosis (PBe) is an infrequently occurring chronic cholestatic liver disease that leads to destruction of small intrahepatic bile ducts, portal inflammation, progressive scarring and eventually to liver failure, lt occurs most commonly in middle-aged women and is associated with disorders of the humoral and cellular immune system. In 95% of patients antimitochondrial antibodies are present. Associated disorders include connective tissue diseases (scleroderma, Sjogren syndrome, CREST syndrome) and hypothyroidism, Muliple stages
can be recognized and as in other types of cirrhosis, portal hypertension and formation of esophageal varices and ascites may be found, Development of HCC is uncommon, but the risk may be increased when coexistent viral hepatitis B or C is present. Advanced disease is usually treated by transplantation within 10 years from diagnosis, and PBC represents the third most common indication for orthotopic liver transplantation in adults, CT Morphology
In early stages of the disease, the liver is normal or increased in size and has a smooth contour. Segmental lace-like fibrosis and regenerative nodules can be seen in a third of patients, Varices and ascites are frequently seen, There is associated lymphadenopathy in 80-90%, With advanced disease, this condition can not be differentiated from other causes of cirrhosis, Typically a small irregular liver, splenomegaly and features of portal hypertension are present. The role of CT is to evaluate complications of cirrhosis and to monitor progression of disease, CT is usually done with noncontrast (fibrosis better seen) and portal venous phases, For improved results in the evaluation of vascular changes, disease progression or HCC detection, an additional arterial phase (triphasic C1) can be beneficiaL The arterial phase is mandatory during a pre-transplantation work-up for evaluation of vascular anatomy,
Hemochromatosis Primary hemochromatosis is an iron-storage disorder that is inherited as an autosomal recessive trait and is characterized by abnormally increased iron absorption from the gut and abnormal reticuloendothelial cells, lt leads to the deposition of iron in hepatocytes and other parenchymal cells (liver, pancreas, heart), leading to cellular toxicity and injury, Hepatomegaly is present in 90% of patients, diabetes mellitus in 50%, The peak incidence is in the fourth and fifth decades, with a 10:1 preponderance of males, In the early stage, the iron deposition is limited to the liver and the pancreas is spared, With advanced disease, also the pancreas shows iron accumultatlon. Characteristically, the spleen is spared because it is comprised of dys-
455
11 The Liver
functional reticuloendothelial cells. The primary form may be complicated by periportal fibrosis. which may progress to hepatic cirrhosis and carries an increased risk ofHCC (up to a 30% incidence rate). Hemosiderosis or secondary (transfusional) hemochromatosis is based on an increased iron intake due to multiple transfusions (chronic anemias), excessive iron ingestion, or increased iron absorption (alcoholic liver disease). This form involves deposition in Kupffer cells and also involves the spleen and bone marrow. In later stages this cannot be differentiated from the primary form. Hemosiderosis is reserved for increased iron deposition without organ damage. CT can be used for the noninvasive (semiquantitative) monitoring of hepatic iron content and is also useful for the detection ofHCC. More and more CT is being replaced by MRI as the primary imaging modality for this indication. CT Morphology
CT shows a generalized increase. or less commonly a focal increase. in liver attenuation (CT numbers> 70HU. sometimes 100-140 HU). Hepatic vessels. especially veins. are plainly visible on unenhanced scans (see Fig.ll.6d). The iron concentration in the liver is proportional to the difference between the measured attenuation of the liver and its normal attenuation (60HLJ). This is true. however. only if the liver is free of fatty infiltration; otherwise dual-energy CT measurements (e.g .. at 80 and 140 kVp) are required. Scans in more advanced cases demonstrate cirrhotic changes. Contrast administration is necessary only in patients evaluated for HCC.
Radiation
Injury
Radiation injury (hepatitis) to the liver may occur after fractionated doses of 35 Cy and more. Onset is typically 2-6 weeks after completion of therapy and patients present with RUQ. discomfort and abnormal liver function tests. It is believed that this injury is a form offibrotic veno-occlusive disease. Most patients show a complete recovery. but some will progress to a chronic stage with marked atrophy in the radiation portal.
CT Morphology
Noncontrast CT shows a sharply delineated band of hypoattenuation due to edema or fatty infiltration. After contrast administration. there is often hypo attenuation in the arterial or portal venous phases with hyperenhancement at delayed imaging. When conformal therapy is used this region may be oval and adjacent to the treated lesion. In chronic cases the irradiated part may become atrophic. with compensatory hypertrophy of the unaffected parenchyma. When patients develop diffuse steatosis. regions of radiation injury are typically spared.
Miscellaneous
Diffuse
Liver Diseases
Sarcoidosis
In 25-75% of patients with sarcoidosis. hepatic and splenic involvement can be seen at pathologic examination. but this rarely leads to hepatic dysfunction. Most commonly the disease produces hepatosplenomegaly. Nodules. ranging from 2-20 mm in size. can be identified in 5-30% of patients. Nodules are hypodense at noncontrast CT and show no substantial enhancement. Lymphadenopathy is often present in the porta hepatis or around the celiac axis. Differentiation of nodular sarcoidosis from lymphoma can be difficult.
Amyloidosis
Amyloid can be deposited in the spaces ofDisse between the endothelial gaps of the hepatic sinusoids. Primary forms are associated with immunoglobulin disorders like multiple myeloma and secondary forms are seen in chronic inflammatory processes such as rheumatoid arthritis. Liver involvement is usually uniform. but can be regional. Hepatomegaly is commonly seen. CT findings are nonspecific. with decreased attenuation of the involved areas on plain CT that show no mass effect. Decreased enhancement with persistent hypodense areas of involved areas is rule. The diagnosis may be suggested when there is an associated lack of splenic enhancement.
Abscesses
Metabolic Storage Diseases Rare
metabolic diseases may involve liver. Wilson's disease leads to an deposition of excess copper with hepatomegaly, chronic hepatitis, and slow progression to cirrhosis, In glycogen
Infectious
storage diseases (GSD) , enzyme-defiCiencies lead to glycogen accumulation with hepatomegaly, steatosis, and eventually cirrhosis, Type I is associated with an increased frequency of hepatocellular adenomas and HCC
Disorders
Hepatitis
Pyogenic Abscess
Hepatitis can be divided in acute or chronic forms, In acute hepatitis changes are present in the intralobular portions with hepatocyte swelling, Chronic hepatitis shows a portal inflammation with additional fibrous changes around the lobules, Causes include alcohol, medications, and viral infections, Neither acute nor chronic hepatitis is an indication for CT examination, The CT features are nonspecific and often unremarkable, In active viral hepatitis CT may show hepatomegaly and periportal low attenuation, while in chronic active forms lymph node enlargement may be noted at the porta hepatis also' In some regions of the world, CT is used in screening high risk populations for HCC
A pyogenic abscess is most often seen in older patients with debilitating diseases like cancer, bacteremia, or after prior surgery, It can result from infection by several routes: biliary (cholangitis) , portal venous (intra-a bdominal sepsis), hepatic arterial (sepsis), local extension or traumatic Patients usually present with fever, pain and malaise, Escherichia coli, Staphylococcus aureus, Enterococcus sp, and anaerobes like Klebsiella or Bacteroides are frequent causative microorganisms, and 50% of abscesses are polymicrobial.
Abscesses Hepatic abscesses may have a pyogenic cause (USA, 85%) or may result from amebiasis (6%) or fungal disease (9%), Regardless of etiology, abscesses progress through three stages of evolution, Within the first 10 days there is necrosis with small areas of liquefaction, At 10-15 days necrotic debris is resorbed and liquefaction will predominate, After 15 days, only little debris is left and is surrounded by a thick fibrous wall, For this reason, the radiological appearance may vary with the stage of the abscess, A mass that is indeterminate by ultrasound can be identified as an abscess based on its CT features, Suspicious lesions can be aspirationbiopsied and, if necessary, drained under CT guidance with good results, If biliary communication is present, drainage should be extended to longer periods, Even with appropriate therapy, large abscesses may take months to resolve,
CT Morphology CT is often nonspecific, but usually shows a round or irregularly shaped hypodense mass, Pus measures some 30 HU, but CT attenuation of superinfected cystic lesions may be much less, Abscesses do not show central enhancement, and peripheral rim enhancement occurs only in lesions walled off by granulation tissue « 50%), With strong enhancement of the liver parenchyma in the portal venous phase, the granulation tissue may appear hypodense relative to the liver. Ill-defined margins before and after contrast administration (caution: partial volume effects) are considered characteristic (Fig, 11,46), Sometimes a double-target appearance is seen due to hypodense edema around the enhancing abscess wall, Abscesses can be unilocular or multilocular with incomplete septa, When abscesses are multiple, they may be clustered C' cluster sign"), Gas formation is common in Klebsiella infection and although gas is a specific sign, it is only present in a minority of cases, Nonspecific accompanying reactions are hepatomegaly, elevation of the right hemidiaphragm, pleural effusion. and basal atelectasis or basal lung infiltrates,
457
11 The Liver
c
b
a
Fig. 11.46 Hepatic abscesses, (a) Large abscess (superinfected biloma) with a moderately enhancing rim. (b) Typical abscesses with indistinct margins, (c) Multiple
abscesses with CT features patic metastases.
Amebic Abscess
sions may have densities comparable with softtissue masses. The margins can be smooth or nodular (60%) and an incomplete ring of edema is often present. Septations are visible in 30%. Associated findings in amebiasis include pleural fluid, perihepatic fluid, and retroperitoneal extension of the disease.
Worldwide, amebiasis is the most frequent cause of a hepatic abscess. It is caused by the parasite Entamoeba bistolytica, which is endemic in tropic and subtropic climates, including the southwestern USA. Abscess formation is the commonest extra-intestinal complication that occurs in less than 10% of cases, 75% of which are found in the right lobe. It is primarily treated by medical therapy (metronidazole). •
CT Morphology
The CT appearance is nonspecific. It shows a solitary, well-defined, round to oval hypo attenuating mass with an enhancing wall (Fig. 11.47).Early le-
that closely
resemble he-
Fungal Abscess
These micro abscesses occur usually in immunocompromised patients, most commonly with hematologic malignancies. Causative organisms, in order of descending frequency, include Candida albicans, Aspergillus and Cryptococcus. Biopsy is usually required for definitive diagnosis as the CT appearance is aspecific and can mimic metastasis, lymphoma, sarcoidosis, or mycobacterial infection. •
CT Morphology
CT typically shows multiple, small hypoattenuating lesions scattered throughout the liver (Fig.11.48a). Some lesions may show rim enhancement. A hyperdense center is characteristic in larger lesions and represents clustered hyphae (Fig.11.48b). Concomitant involvement of the spleen and kidneys is frequent. Sterilized lesions may undergo calcification. Fig. 11.47 Large septated amebic abscess with a small rim of contrast enhancement.
Liver Tubercu/osis
Fig. 11.48 Fungal abscesses. (a) Multiple tiny hypoattenuating lesions are typical for candid sepsis. Aspergillus abscess may present as a lesion with a target enhancement in the portal venous phase (b).
a
b
Hydatid Disease
The dog tapeworm Echinococciis granulosus is endemic in the Mediterranean countries, Russia, the Middle East, Japan, Australia, parts of Africa, and South America, where sheep-farming is common. Its eggs hatch in the duodenum, releasing larvae that enter venous vessels in the wall of the small intestine and become lodged in various capillary filters (liver> lung> peritoneum > kidneys> spleen, etc.). The liver is affected in 55-75% of cases. The larvae develop into slow-growing hyclatid cysts, which are made up of three layers: the outer pericyst of infiltrated host tissue, which forms a fibrous protective zone, the middle laminated membrane, and the inner germinal layer. These latter two form the true wall or endocyst. As the cyst matures and degenerates, small daughter cysts form in the periphery. With further degeneration the pericyst eventually ruptures and the lesion may calcify completely. Cysts are commonly classified in five types, which is useful for treatment selection. Hydatid disease may be complicated by secondary infection and cyst rupture. This rupture can involve either the biliary system, peritoneal or retroperitoneal cavities or the gastrointestinal tract, and the spill of allergenic fluid can incite an anaphylactic response. Hydatid disease does not always produce a positive serologic test (upto 85 % sensitivity). CTis valuable in the detection ofthe cysts and can accurately define the extent of disease prior to surgery orintervention. It is also used in complications or to monitor therapy. Treatment is initially medical
(mebendazole). More recently, it was shown that percutaneous treatment is safe in selected cyst types and CT may be used for guidance. •
CT Morphology
Imaging findings depend on the stage of cyst growth and may show an unilocular cyst or a complex with daughter cysts. On noncontrast CT, a hydatid cyst is a well-defined round to oval cystic mass. Daughter cysts are seen as septate structures within the mother cyst or give it a multilocular appearance. The cyst usually has a fluid-density content, but the attenuation of the cyst contents may be as high as 45 HU (hydatid sand). Daughter cysts can have a lower attenuation than the mother cyst. Frequently the cyst is permeated by thin septa, and it has a high-attenuation wall. Cyst calcification occurs in the outer wall, and dense calcifications indicate non-viability of the lesion. Enhancement of the cyst wall and septa on post -contrast scans is a feature that distinguishes hydatid cysts from hepatic cysts (Fig. n.4g). Liver Tuberculosis
Hepatic TE is commonly associated with the miliary form of the disease. It can have a more common micronodular form or a rarer macronodular form. The latter spreads to the liver via para-aortic or hepatoduodenal ligament nodes. •
CT Morphology
The micronodular form shows moderate hepatomegaly. The liver parenchyma is homo-
459
11 The Liver
Oriental Cholang iohepatitis
a
b
Fig. 11.49 Ecchinococcus cysticus. Hydatid cyst with arterial enhancement of the pericyst (a) and excellent distinction between pericyst (arrow), cyst membrane (arrowhead) and germinative layer (wide arrow) of the endocyst in the portal phase (b). Ruptured hydatid cyst (c. d) shows a crumpled subcapsular membrane and an enhancing cyst wall and septa, accompanied by free intra-abdominal fluid.
geneous or heterogeneous. In the macronodular form, single or multiple hypo attenuating (3545 HU) nodules are scattered through the liver. The nodular lesions may calcify. Associated findings like adenopathy, ascites, and peritoneal disease suggest disseminated TE. Rare Infectious Diseases
Also termed recurrent pyogenic cholangitis, this condition is endemic in SE Asia and due to immigration is now more frequently seen in the West. It is a recurrent cholangitis that may result in progressive cholangiopathy and eventually liver failure. Patients have coliform bacteria in bile or previous Clonorchis infection. Major features include intra- and extrahepatic bile duct dilatation and stones. CT shows intrahepatic bile duct strictures and fusiform or varicose dilatation that is most common in the lateral segment of the left lobe. Other findings include high-attenuatingcalculi, pneumobilia, abscess formation, and segmental atrophy. Calculi are often multiple and in acute forms contrast enhancement of the bile duct wall may be seen. Clonorchiasis
The parasite Clonorchis sinensis (Chinese liver fluke) can infect humans after the ingestion of raw freshwater fish. It is endemic in South East Asia. Cercaria migrate to the small to medium bile ducts in the liver where the mature worms are formed. The number of infected persons increases with age and men are more commonly affected. CT shows diffuse, mild intrahepatic dilatation, usually uniform from porta to the periphery. Extrahepatic duct involvement is rare. In endemic areas, this infection is a common risk factor for HCC.
Schistosomiasis
Infection with Schistosoma japonicum is common in South East Asia. Liver involvement shows a characteristic pattern of capsular and parenchymal calcification perpendicular to the liver capsule. Linear and parenchymal band enhancement is seen. The liver surface can be nodular and fibrosis may show increased periportal fat. There is an increased risk for HCC.
Glanders Disease
This is a very rare infection with Burkholderia mallei causing oropharyngeal mucositis and abscess formation in lung, liver, and spleen. It is endemic in the Middle and Far East, infecting mainly horses and donkeys. Sporadic cases in the West include veterinarians or horse lovers. In the liver it causes multiple, small nodules that may coalesce into larger necrotic abscesses. The infection in humans is lethal when untreated.
Perfusion Abnormalities Vascular
Disorders
and Perfusion
Abnormalities
Vascular disorders used not to be a primary indication for CT and were originally evaluated by (color) duplex sonography. With CT angiography. and especially with the advent of multislice scanning. CT can be used as a primary imaging modality that not only evaluates the liver parenchyma but also vascular changes. A biphasic spiral or multislice CT examination can replace angiography in most instances and allows for preoperative investigations of the hepatic blood supply. Perfusion abnormalities have been known for a long time from experiences with angiographically assisted CTtechniques such as CTHA and CTAP. With the increasing use of multiphasic helical CT using higher injection flow rates. these perfusion abnormalities are frequently encountered in daily practice.
Third hepatic inflow tracts by systemic veins that drain non opacified venous blood into the liver are more frequent and are associated with THADs during multiphasic studies or as flow-related defects during CT during arterial portography. An aberrant right gastric vein may directly drain into the liver giving a perfusion defect in the posterior edge of segment IV. just anterior to the right PV (8% of patients. see Fig. 11.8). An aberrant left gastric vein can give perfusion defects in the posterior edge of segment IV (although less frequently) or in the posterior part of segment II. The cystic vein of the gallbladder can drain into intrahepatic portal vein branches around the gallbladder fossa and create a local portal perfusion defect (see Fig. 11.8). Paraumbilical veins that run into the falciform ligament drain into the left lobe portal branches and give portal perfusion defects anteriorly on both sides of the falciform ligament (Fig. 11.50).
Perfusion Abnormalities Any process (tumor. cirrhosis. etc.) that causes a local or generalized rise in intrahepatic pressure will cause a reduction of portal venous flow in the affected region. This is associated with a compensatory increase in arterial perfusion. Locally altered hemodynamics with increased arterial flow are known as transient hepatic attenuation differences (THAD). also called transient hepatic parenchymal enhancement (THPE). lt results in a regional areas of hyper attenuation in the hepatic arterial phase (HAP) that return to normal in the portal venous phase (PVP). These can be caused by a number of vascular lesions (described in the next section). but can also occur independently. Knowledge of the pattern of these abnormalities and their causes is important for correct interpretation (Table 11.14).
Arterioportal Shunts Arterioportal shunts can occur via several routes: through an iatrogenic fistula (post biopsy). trans-sinusoidal (blockage of venules), transvasal (via tumor thrombus), transtumoral
Table
11.15
Common
causes
and
locations
of perfu-
sion defects
'location IMan-occlusive Aberrant
R gastric
Aberrant
L gastric
Gallbladder
vein
Paraumbilical
Anatomic Variants
APshunt
Systemic arteries may aid in the arterial supply of the liver. especially in tumors or in situations with diminished hepatic arterial flow. The most common are the inferior phrenic. internal mammary. left gastric and intercostal arteries. These variants may give perfusion defects at CTHA.
Occlusive
edge
segment
IV
Posterior
edge
segment
IV
Posterior
edge
segment
II
along gallbladder Anterior Liver
Right hepatic Middle
veins
Posterior
hepatic
vein vein
ligament
periphery
Dorsal in anterior segment Ventral in posterior segment Ventral Medial
Left hepaticvein
falciform
fossa
in anterior
segment
segment
Throughoutleft lobe
11 The Liver
a
b
c
d
Fig. 11.50 Transient hepatic attenuation differences (THAD). (a) Hyperperfusion in the arterial phase due to collaterals from superior vena cava occlusion directly entering the left liver lobe. (b) Transient hyperperfusion in hepatic cirrhosis (negative biopsy), (c) Wedge-shaped arterial hyperperfusion due to a capillary hemangioma
with arterio-portal shunting in the arterial phase, (d) Segmental arterial hyperperfusion due to obstruction of a portal venous branch by a hypovascular metastasis (arrow). Note the concomitant biliary obstruction (wide arrow).
(via draining vein of a hypervascular tumor) or transplexal (peri biliary) . Tumoral shunts (HCC, cholangiocarcinoma or hemarigioma) are usually caused by vascular infiltration of portal vein branches. Nontumoral shunts due to liver cirrhosis, trauma, biopsies, and ethanol injection are less common. These shunts also reflect local blockage of portal venous venules and retrograde filling via trans-sinusoidal AP shunting. Regardless of their etiology, these shunts typically give wedge-shaped THADs in the liver periphery and there may be a dot -like or branching vessel visible resulting from early venous filling. These are best seen in the arterial dominant phase or during CTHA.
Portal Hypertension
Other Causes
Siphoning by hypervascular tumors that steal arterial blood from the surrounding parenchyma will result in hypo attenuation during the HAP and isoattenuation in the PVP. Infiammatoi-y changes caused by acute cholecystitis or hepatic abscesses lead to arterial hyperemia and venous flow stoppage that show typical THAD.
Normal portal flow rate is about llOOml/min and normal pressure is 5- WmmHg, which is 3-6mmHg above the IVC pressure. Portal hypertension is defined as a venous pressure gradient greater than 5 mm Hg and commonly results from increased portal venous resistance, the cause of which may be prehepatic, intrahepatic, or posthepatic. Rarely, portal hypertension may be due to a hyperkinetic condition caused by an arterial-portal fistula (traumatic, congenital, neoplastic). Clinically significant symptomatology is usually present with pressure gradients above 12 mm Hg, when complications will be seen. Prehepatic causes reduce portal blood flow (PV thrombosis, AP fistula), intrahepatic causes are associated with chronic liver disease (diffuse liver disease, cirrhosis) and posthepatic causes reduce outflow into hepatic veins or IVC (venoocclusive disease). Intrahepatic causes may be further divided in presinusoidal, sinusoidal and postsinusoidal. The pressure increase induces the formation of portosystemic collaterals, which are usually dilated pre-existing vessels but active angiogenesis is also seen. Common collateral pathways include the coronary (left gastric), posterior gastric, and short gastric veins to the esophageal and paraesophageal veins;
Portal Venous Obstruction gastrosplenic, gastrorenal and splenorenal shunts to the left renal vein; and several pathways including the omental, paraumbilical, hemorrhoidal, and retroperitoneal (Retzius) veins. Other complications include splenomegaly, encephalopathy and ascites. Banti's syndrome is an idiopathic, noncirrhotic form of portal hypertension that is secondary to obliterative venopathy of the intrahepatic portal venous branches. CT is not the firstline technique for imaging portal hypertension but imaging in the early portal venous phase is excellent for demonstration of complications and collateral pathways. •
Segmental portal hypertension results from the occlusion of splenic or mesenteric veins.
CT Morphology
An increased diameter of the portal vein (> 13 mm) and the superior mesenteric and splenic veins (> 1Omrn) with development of paraumbilical collaterals (>3mm) are highly specific but only moderately sensitive signs of portal hypertension. CT is able to demonstrate the various collaterals (Fig.'11.51). Large (spontaneous) splenorenal shunts may almost normalize the portal venous pressure and reduce the size of other collateral pathways. Paraesophageal varices have to be distinguished from esophageal varices because they cannot be detected with endoscopy. Collateral veins may be difficult to distinguish from collapsed bowel or lymph nodes on
a
unenhanced CT scans but are easily distinguishable after contrast administration. Collateral pathways are well demonstrated on axial sections but targeted (thin-slab) MIP or VRT yield a better overview of the anatomic situation (Fig. 11.52). Splenomegaly is frequently but not always present. Siderotic nodules (Gamna-Gandi bodies) in the spleen may appear as tiny disseminated calcifications.
b
Fig. 11.51 Portal hypertension in a patient with severe liver cirrhosis: enlargement of the portal vein diameter (» 13 mm) and development of prominent venous collaterals. This CTAP with selective contrast enhancement of the portal vessels demonstrates a spontaneous splenorenal shunt with contrast run-off into the renal
Portal Venous Obstruction
Causes of obstruction of the portal vein can be intrahepatic or extrahepatic. Extrahepatic obstruction can be the result of neoplastic involvement, PV thrombosis, extrinsic compression, iatrogenic or due to pressure on the parenchyma. Intrahepatic (presinusoidal) obstruction due to narrowing of portal venules by inflammation or fibrosis. Portal vein thrombosis is the major cause of prehepatic portal hypertension (and variceal bleeding) in the western countries. In young patients it is usually idiopathic or may result
c vein (arrow). After intravenous contrast administration such shunts are characterized by a dilatation of the left renal vein. Multiple hypoattenuating HCC nodules (arrowheads) are visible against a background of micronodular cirrhosis.
463
11 The Liver
to 20% of cases show cavernous transformation of the portal vein (local collateral veins). In main portal vein obstruction, reduced portal blood flow leads to dilatation of the hepatic artery and increased arterial inflow. This can take two patterns: • compensatory increase of regional flow in the area where portal flow is reduced (because of tumor thrombus, or compression or stricture of the portal vein or a branch); and • heterogeneous peripheral increase of arterial flow (in cavernous transformation of thrombosed portal vein), while collateral circulation supplies the central periportal parenchyma. Associated AP shunts may also be present. A chronic pel-fusion deficit (caused by arterial occlusion, portal thrombosis, arterioportal shunting, etc.) can lead to regional fatty infiltration. "
Fig. 11.52 The multitude of abdominal varices in a patient with portal hypertension is best demonstrated on volume-rendered images (4x 1/6, thin-slab VRT anterior to the aorta and SMA). Note the large perigastric varices.
from venous webs (mainly Asian countries). Secondary forms can result from inflammatory or infectious diseases such as pancreatitis, sepsis, myeloproliferative diseases, tumor compression or invasion, hepatic cirrhosis with portal hypertension, and hypercoagulable states (protein CIS or AT3 deficiency, antiphospholipid syndrome). The involvement may be partial or complete and can involve any portion of the portal system. The portal thrombosis induces the development of portosystemic, splenohepatic or mesentericohepatic collaterals, and up
a
b
Fig. 11.53 Non-tumorous portal venous thrombus must not enhance in the portal venous phase (a). Chronic thrombi in portal hypertension may not lead to occlusion but to thrombotic material adherent to the walls, often associated with calcifications (b). Pseudo-
•
CT Morphology
A portal vein thrombosis can only be identified on unenhanced scans if it is fresh with a hyperattenuating thrombus present (Fig. 11.53). The affected venous segment is enlarged (> 15mm in up to 40% of cases) and in long-standing thrombosis shows decreased attenuation. Calcifications may develop in a chronic thrombus or in the wall of the portal vein. They are especially frequent in patients with Banti's syndrome.
c thrombi are a frequent finding on arterial phase images and are due to lack of mixing of enhanced blood from the spleen and non-enhanced blood from the mesenteric circulation (c).
Hepatic Venous Outflow Obstruction Signs of portal hypertension (esophageal, fundal, and paraspinal varices) vary in their conspicuity, depending on the duration of the thrombosis. Ascites is a common secondary phenomenon of decompensation in the acute phase or late stage. Scans after contrast administration show nonopacification of the thrombosed segment or an intraluminal filling defect. Tumor thrombus shows moderate to marked and often irregular enhancement, best appreciated in the arterial phase. If cavernous transformation is present, a network of bypass veins can be seen in the area of the occluded segment (usually the porta hepatis, Fig. n.54). Parenchymal enhancement is often inhomogeneous with relative hypo density of the affected liver areas. The described arterial hyperperfusion patterns both produce THAD. The first pattern leads to a lobar or (sub)segmental hyperattenuation in the HAP (no collaterals), and the second to multiple inhomogeneous peripheral hyperattenuating areas (collaterals or cavernous transform). The peribiliary plexus takes over for periportal liver, and in chronic cases periportal hypertrophy can be seen with atrophy in the affected liver periphery. During the late arterial or early portal phase, contrast -enhanced blood from the splenic vein enters the portal vein while the mesenteric blood is still relatively nonopacified. In-
complete mIxmg of opacified and nonopacified blood can mimic a filling defect, creating a "pseudothrombus" in the portal vein or proximal superior mesenteric vein.
Hepatic Venous Outflow Obstruction
Blockage of the hepatic venous outflow can be due to Budd-Chiari syndrome, congestive rightsided cardiac failure, pericardia! disease, mediastinal fibrosis, or extrinsic compression on the portal vein or IVe. Obstruction of venous outflow will increase postsinusoidal pressure (hypertension), resulting in dilatation of the central veins with delayed portal venous inflow, with or without hepatomegaly. Eventually the flow in the portal vein may be reversed, turning it into the main hepatic draining system. The arterial inflow will be relatively increased. Suprahepatic thoracic inlet obstruction causes blood flow through collateral vessels like the azygos-hemiazygos and lateral thoracic veins to the IVe. The latter communicates via the paraumbilical veins with the portal circulation. Generally this diagnosis of hepatic outflow obstruction is established by ultrasound (color duplex sonography). CT may be used to confirm the diagnosis in selected cases. Budd-Chiari Syndrome
Fig. 11.54 With cavernous transformation of the portal vein, a network of collateral channels occupies the area of the former portal vein.
Budd-Chiari syndrome is a rare disorder that refers to a segmental or global obstruction of the hepatic veins. Acute cases (1/3) present with sudden abdominal pain due to hepatic congestion and ascites that is refractory to treatment (>95%). In the chronic stage, ascites is accompanied byjaundice and signs of portal hypertension with bleeding varices. The syndrome may be caused by hepatic venous thrombosis or nonthrombotic obstruction. It is more frequent in the Middle East, India, and Japan, where it is often caused by membranous webs or septa. In the USA and Europe, most cases are acquired and can be caused by hypercoagulable states (polycythemia vera, PNH, sickle cell disease, oral contraceptives, pregnancy and postpartum states), injuries to the vessel walls (trauma, phlebitis), or tumorous obstruction (HCC, renal
465
77 The Liver
cell or adrenal carcinoma). Concomitant portal vein thrombosis is present in 20% of patients. Based on the location of the obstruction, several types of the syndrome exist. Type I is defined as an occlusion of the IVC and type II patients have occluded major hepatic veins. In hepatic vena-occlusive disease, which has also been termed type III Budd-Chiari, resistance is at the level of the postsinusoidal venules following a progressive inflammation. This is often associated with chemotherapy and radiation ther-
apy. •
CT Morphology
In acute Budd-Chiari syndrome, noncontrast scans show hepatomegaly with decreased liver attenuation due to edema and rapid fatty infiltration. Ascites is present and hyperattenuating thrombi may be visible in the IVC or hepatic veins. The caudate lobe is rarely affected initially. After contrast administration, heterogeneous patchy enhancement (mosaic pattern) is seen in the arterial phase and there may be draining portal vein branches visible. The occluded area may be hyperenhanced during this phase due to arterial hyperperfusion (THAD). In the portal venous phase this patchy enhancement often persists, especially when there is hepatofugal portal vein flow (Fig.TI.55a). On 5-10 minute delayed scans, the patchy poly-
a
b
Fig. 11.55 Budd-Chiari syndrome. Occlusion of the hepatic veins (arrows) in chronic Budd-Chiari syndrome. Patchy areas of enhancement are seen around the larger portal branches, with intervening reticular areas of low attenuation. Parenchymal swelling has compressed the vena cava to a slit (arrowhead, a). Acute Budd-Chiari syndrome in another patient with intrahepatic venous
gonal enhancement of the affected areas disappears. The hepatic veins continue to show low or no luminal attenuation after contrast administration. Due to the edematous swelling of the liver, the IVC may be highly compressed in its hepatic portion. In the subacute stage, multiple hypervascular nodules may be found that correspond to regenerative nodules (Fig.TI.55b). Such nodules are amost impossible to differentiate form hypervascular tumors (Fig. T1.55c). In the chronic stage, the caudate lobe and unaffected liver areas are enlarged in 80-90% of cases, as it has its own veins that drain into the vena cava. The affected liver parenchyma shows the same patchy enhancement pattern with normal opacification of the portal vein and patchy enhancement around the terminal portal branches (Fig.TI.55b). Some patients may actually have a relative homogenous hepatic enhancement. Infarcted areas, especially at the periphery of the liver, become shrunken and fibrotic. They show low attenuation and very little enhancement. The hepatic veins are not visualized or are greatly reduced in caliber.
c collaterals (arrowhead) due to a patent left liver vein (wide arrow). Note the multiple hypervascular nodules (b). Budd-Chiari syndrome in a patient with a large central HCC and hypodense thrombosed hepatic veins (arrowheads) (c). In the periphery of the tumor it is no longer possible to distinguish between hypervascular benign nodules and HCC.
Rendu-Osler-Weber Disease Passive Hepatic Congestion
In passive hepatic congestion from heart failure or pericarditis, elevated venous pressure leads to decreased hepatic flow and arterial hypoxemia. This hypertension results in sinusoidal congestion and perisinusoidal edema. Patients can have hepatomegaly and elevated liver function tests. If uncorrected, this may progress to cardiac cirrhosis. •
CT Morphology
CT shows hepatomegaly with normal parenchymal density in the noncontrast phase. The caudate lobe is usually normal. Ancillary findings are ascites, pleural effusions, and cardiomegaly. After contrast administration, a diffusely mottled enhancement of the liver is seen and contrast -enhanced blood may reflux from the heart into the IVC or hepatic veins. There may be intrahepatic perivascular lucency and the IVC and hepatic veins are usually dilated (Fig. 11.56). This distinguishes this disorder from Budd-Chiarisyndrome. Hepatic Arterial Thrombosis and Hepatic Infarction
With obstruction of hepatic arterial flow, there is no increase in portal venous flow. Blockage of large arteries will directly induce new routes of flow from extrahepatic arteries (see Pel-fusion Abnormalities), e.g., through the bare area of the liver and via the suspensory ligaments. A delay in arterial flow may lead to a hypo dense region during arterial phase imaging.
Because of the dual blood supply of the liver, hepatic infarction is uncommon. It can result from coexisting hepatic arterial and portal venous obstruction. It is most commonly seen after liver transplantation. Other less frequent causes of hepatic artery occlusion include hepatic artery atherosclerosis, shock, sepsis, eclampsia, and oral contraceptive use. •
CT Morphology
Hepatic infarcts can take on three configurations. Peripheral wedge-shaped lesions, central or peripheral rounded lesions, or irregular shaped lesions paralleling bile ducts. With time, margins become less distinct. Bile duct necrosis may lead to gas collections or cystic bile collections. In long-standing cases the involved liver segment atrophies. Arterial obstruction will show increased caliber of extrahepatic systemic arteries with peripheral THAD during the arterial phase. Rendu-Osler-Weber Disease
Rendu-Osler- Weber disease (hereditary hemorrhagic telangiectasia) is a hereditary disorder in which multiple pathologic vascular formations occur throughout the body. Liver involvement is characterized by hepatic telangiectasia and usually arteriovenous, less commonly arterioportal, shunts. These shunts can cause a significant increase in cardiac output resulting in a hyperdynamic circulatory status.
Fig. 11.56 Severe hepatic congestion due to right heart failure, with patchy polygonal areas of periporta I enhancement in the liver (a). Note the poor opacification (a) or dilatation (b) of the hepatic veins (arrows).
a
b
468
11 The Liver
a
c
a
Fig. 11.57 Osler's disease is characterized by enlarged hepatic arteries and early enhancement of the hepatic veins already during the arterial phase (a). The liver parenchyma contains very small, hyperdense foci representing shunts (b). Individual portal veins are throm-
bosed (arrows). A volume-rendered display of an arterial phase scan in another patient (c) demonstrates marked enlargement of the hepatic artery and early enhancement of draining veins (arrowheads).
•
diminished on portal phase scans due to the high cardiac output, and little parenchymal enhancement is seen. The best examination techniques are spiral or multislice scanning performed during the arterial phase of enhancement.
CT Morphology
Noncontrast scans demonstrate enlarged vessels (especially the hepatic veins) and frequently also hepatic enlargement. Signs of (hyperkinetic) portal hypertension may be present. Arterial phase images demonstrate enlarged arteries • Differential Diagnosis with early opacification of enlarged hepatic veins due to arteriovenous shunting (Fig. 11.57). Hepatic hemangiomatosis appears as multiple The shunts usually involve the peripheral vessels circumscribed lesions that show a characteristic diffusely, and therefore they are difficult to pattern of peripheral to central enhancement (Fig. 11.58). define. Vascular opacification is markedly
a
b
Fig. 11.58 Liver hemangiomatosis in Gorham's disease with multiple lesions with patchy hypervascular areas in the periphery on arterial phase scans (a) that show progressive enhancement on portal phase images
c (b). There is marked enlargement of the hepatic arteries (arrowheads) (c) and early enhancement of liver veins (arrow) similar to Osler's disease (4 x 1/6, VRT).
Blunt Liver Trauma Fig. 11.59 Peliosis hepatis with multiple focal lesions that demonstrate modest enhancement on arterial (a) and more pronounced enhancement on portal venous phase scans (b).
b
a
Peliosis Hepatis
Peliosis hepatis is a rare disorder in which multiple blood -filled cystic spaces form within the liver parenchyma. These lacunar spaces do not have an endothelial lining and are prone to rupture. They may occur as a complication of aplastic anemia or androgen therapy, chemotherapy, diabetes mellitus and chronic renal failure.
•
CT Morphology
The lacunar spaces seen in peliosis hepatis appear almost isodense to the aorta, and their enhancement follows that of the aorta (Fig. 11.59). They are strongly hyperattenuating in the arterial phase and remain moderately hyperattenuating in the portal venous phase (bloodpool effect). Rupture with formation of intrahepatic and perihepatic hematomas is a dangerous complication.
Trauma Blunt Liver Trauma
The liver is the second most frequently injured solid organ and injuries occur in 15-20% of patients with blunt abdominal trauma. The right lobe is more often injured than the left lobe, and the posterior segments especially are prone to injury by compression or shearing forces. There may be associated rib fractures. In 80% there is associated hematoperitoneum and there may be retroperitoneal bleeding from rupture of the liver surface in the bare area. Injuries can be classified as vascular lesions, lacerations and fractures, intraparenchymal or subcapsular hematomas, contusions and periportal low -attenuation. Grading is often done using the organ injury scale of the AAST (Table 11.16),whereas imaging based classifications (e.g., Mirvis et ai, 1989) may also be used. CT can accurately define the extent of hepatic injury, quantitate hemoperitoneum, reveal associated injuries, and follow healing. This enables the surgeon to select the most appropriate
therapy. Standardization of reporting and grading enhances accuracy and can help in this triage. Based on clinical and imaging findings, 20-40% of patients are currently managed conservatively. Postcontrast scans usually suffice because they allow for the detection of hemorrhage and are more sensitive for parenchymal contusions and the extent of lacerations. If acute bleeding is suspected, a biphasic examination should be performed. CT is also indicated in the diagnosis and management of complications, such as biloma or abscess formation. •
CT Morphology
Vascular injuries are the most severe injuries but are fortunately rare. They most frequently include lesions of the (retrohepatic) inferior vena cava and hepatic veins, and management often requires a combination of surgery and catheter embolization. Lacerations are the commonest type ofinjury and frequently travel along the vessels and fissures. Perihilar lacerations are more frequently
469
11 The Liver Table 11.16 Grade
AAST organ injury scale for liver injuries (1994 resivion) Injury
Hematoma Laceration
Subcapsular, <10% surface area Capsulartear, <1 cm parenchymal depth
Hematoma
Subcapsular, 10-50% surface area intraparenchymaL <1Ocm diameter 1-3 cm parenchymal depth, <1Ocm length
Laceration III
Description
Hematoma Laceration
Subcapsular, >50% surface area or expanding, ruptured Subcapsular or parenchymal hematoma intra parenchymal hematoma> 1Ocm or expanding >3 cm parenchymal depth
IV
Laceration
Parenchyma! disruption involving 25-75% of hepatic lobe or 1-3 Cuinaud's segments in a single lobe
V
Laceration Vascular
Parenchymal disruption involving> 75% of hepatic lobe or >3 Cuinaud's segments within a single lobe Juxtahepatic venous injuries i.e. retrohepatic vena cava/central major hepatic veins
VI
Vascular
Hepatic avulsion
Advance
one grade for multiple injuries to same organ up to grade III
associated with biliary tract injuries, while lacerations through the proximal hepatic veins are important because repair is technically difficult. Lacerations appear as linear or branching low-attenuation areas with sharp margins. Multiple parallel lacerations have been termed "bear-claw" lacerations. Their location to the hepatic vessels and biliary radicles are important. A "liver fracture" is a laceration extending through the entire liver which may avulse a part of the parenchyma. Intraparenchymal hematoma is collection of blood within a laceration. Hemtomas appear as round or oval hyperattenuating areas in the liver (Fig. 11.60) that may persist for several weeks,
making it extremely difficult to estimate the age of the hematoma. Active extravasation or a traumatic pseudoaneurysm may be seen in the center. Post -traumatic cysts can result from liquefaction and cavitation of the hematomas. Subcapsular hematomas are blood collections beneath an intact capsule and usually lentiform in shape. They are most often located anterolateral to the right lobe and deform the liver contour. They may be difficult to distinguish from intraperitoneal hemorrhage at the same location. Contusions are areas of minimal hemorrhage or edema without a laceration. They may
a
b
Fig. 11.60 Large intrahepatic hematoma 14 days after blunt abdominal trauma, appearing as an area of increased attenuation on unenhanced CT (a). Hepatic
laceration with a contused area and a linear tear seen best on a contrast-enhanced scan (b).
Liver Resection present as hypodense areas in the liver, or regions of hepatic edema marked by reduced portal pel- fusion and arterial hyperperfuston. Periportal low attenuation may be the only sign of liver injury, In trauma patients it can be the result of hemorrhage along the portal vein branches, but it may be seen after vigorous fluid administration or other causes of elevated central venous pressure in the absence of liver injury, All traumatic lesions (hemorrhages, contusions, lacerations) are more conspicuous on portal phase scans, Hemoperitoneum is often seen in the hepatorenal recess (Mortsson's pouch) and extends via the right para colic gutter into the pelvis, It typically has an attenuation value of 40-50 HU, but more frequently blood collections may be in the range of 20-40 HU A "sentinel clot" of higher attenuation (45-70 HU) may be indicative of the site of bleeding, The presence of large hyperdense areas suggests active bleeding that requires prompt intervention, In the arterial perfusion phase such areas of active bleeding may be directly visualized as strongly hyperdense spots orthread-like extravasation (see Fig, 11.33),
Penetrating Abdominal Trauma
or gun shot wounds, Most stab wounds are superficial and when involving the liver commonly cause superficial lacerations, Gun shot wounds are more complex, depending on the type of weapon, projectile, and firing distance, CT should only be used for selected wounds in hemodynamically stable patients, It may be used to reveal the superficial nature of back or flank stab wounds, or it can direct which patients with gunshot wounds can be managed non-operatively, If associated bowel injury is suspected, explorative laparotomy is indicated and CT should not be performed, CT Morphology
CT morphology is similar to that of blunt trauma, In gunshot wounds, the path of projectile can be visualized as it disrupts normal tissue planes and leaves hematoma, gas bubbles, and metal fragments along its path, A retained bullet is easily seen as a large metal fragment with associated artifacts, CT interpretation may be complicated by altered hemodynamic status, respiratory motion artifacts, or incorrect timing of the contrast administration, Respiratory motion may mimic subcapsular hematoma around the right liver lobe, while unopacified blood vessels or beamhardening artifacts can simulate lacerations,
Penetrating abdominal trauma is less frequent than blunt trauma and is usually caused by stab
Postoperative
and Post-therapeutic
Changes
Liver Resection In patients without extrahepatic metastatic disease, several types of liver resection can be performed with low morbidity, The amount and functional status of the remaining liver largely determine the amount of liver that can be resected, Risk analysis, like the Memorial Sloan Kettering Risk Score (Fong et el.; 1999), can be used for improved patient selection and to guide adjuvant therapy, Resections can be combined with other therapies like chemoembolization or chemotherapy, which may improve results in selected patients, In the preoperative work-up of liver metastases, multiphasic helical or multislice CT as well as multislice CTHAICTAP playa major
role in detection and mapping, Additional highresolution CT angiograms for surgical planning can be reconstructed from the same datasers. Multiphasic CT is also an important tool in the diagnosis and management of complications, CT Morphology
The postoperative appearance varies with the type of resection that has been performed, There are often fluid collections in the resection bed and clips are commonly found at the surgical margin, Omental patches appear at the surgical margin as areas oflow attenuation approaching that of fat Hemihepatectomy is swtftly (within 1 week) followed by hypertrophy of the residual liver parenchyma, whereby the rest liver may nearly reach its preoperative size, If a wedge resection
471
11 The Liver
has been performed, hypo attenuating areas may persist in the area of the defect for several weeks or months and can be difficult to distinguish from recurrent tumor. Adjacent organs (kidney, transverse or right colon) may shift position into the newly created space. The commonest complications are hematomas, bile leaks with biloma formation and postoperative abscesses. Intraparenchymal hematomas can often be found as hypo attenuating areas near the resection margins. Most bilomas occur at the resection margin or in the porta hepatis. They are of water density « 15 H U) but are indistinguishable from seromas, liquefied old hematomas, and infected fluid collections. Collections that do not conform to the resected liver are of higher risk and represent serious complications like abscess formation. Postcontrast scans will usually demonstrate faint rim enhancement around infected areas. Doubtful cases are resolved by percutaneous aspiration. Tumor recurrence will result in up to 75% of patients, with 50% of recurrences in the regenerated liver. Dependingon location, CTis helpful in selecting candidates for repeat resection. Liver Transplantation
Transplantation of the liver is usually performed with cadaveric organs (orthotopic liver transplantation [OLT] or split-liver transplantation) or more recently using hemilivers from selected living liver donors. OLT implies four vascular anastomoses (two IVC anastomoses, hepatic artery, portal vein) and one biliary anastomosis, while partial liver transplantation has one vascular anastomosis less. Both can be fraught
by vascular complications, biliary complications, and rejection. Due to immunosuppressive therapy, the recipient is at increased risk for malignant tumors like post-transplant lymphoproliferative disease (PTLD) or lymphomas. In living donor transplantation complications from liver resection may also occur in the liver donor, such as abscess, bile leak, and vascular injury. CT plays an important role in the preoperative work-up of donor and recipient before transplantation. Multiphasic spinal CT is used for the detection of malignancy, combined with 3D CT angiographyformappingvascularanomalies (see p. 410 and p. 898). After transplantation, the role ofCTis to assess parenchymal pel- fusion and to diagnose and manage vascular complications (arterial and venous stenoses or thrombosis, hepatic infarction, AV fistulae oraneurysms ofthehepatic arteriesorportalvein), biliarycomplications (bile leak, anastomotic stricture), hematoma, and abscess formation. •
CT Morphology
CT demonstrates hyperdense suture lines at the superior and inferior caval anastomosis. In the early phase after transplantation periportal lowattenuation due to lymphedema is normal. If such a finding persists for more than 6 months, chronic rejection has to be considered. Portal phase scanning is frequently unsuccessful in revealing perfusion defects following partial hepatectomy or liver transplantation. Areas of arterial hyperperfusion in biphasic CT signify a relative deficiency of portal blood flow. Irregular hypo attenuating non enhancing areas at the liver margins often represent subcapsular necrosis, which has no prognostic significance. Direct visualization of the hepatic artery during
Fig. 11.61 Stenosis of the hepatic artery after liver transplantation (a). Post-transplant lymphoma at the porta hepatis (b).
a
b
Minimal Invasive Therapies
the arterial phase and the portal vein during the portal phase permits the detection of vascular complications such as thromboses, stenoses, and occlusions (Fig. H.61). Biliary leaks or obstructions occur in about "15%. Bile duct necrosis presents as marked biliary dilatation, which is either focal or diffuse. In such cases, a thin-section arterial phase CT (CTA) may reveal stenosis, or occlusion of the hepatic artery as the cause. Transjugular Intrahepatic Portosystemic Shunt (TIPS)
The TIPS procedure is a technique to decompress the portal venous system percutaneously by establishing a shunt between a branch of the portal vein and a liver vein (usually the right). An intra parenchymal tract is formed in the liver parenchyma and kept open by means of an expandable metallic stent (usually Wallstent). CT may playa role prior to a TIPS procedure in patients with liver anatomy distorted due to severe cirrhosis. In these patient the optimum vessel can be determined beforehand, thus reducing procedural duration and complications. In the follow-up after TIPS, CT is excellent for demonstrating acute complications. CTA also can demonstrate the patency of the shunt but color duplex ultrasound is the preferred firstline technique for follow-up because it enables assessment of stent morphology and intrastent velocity measurements.
a
b
Fig. 11.62 Patency of stents after TIPS procedures is best appreciated on curved planar reformations from thin-section scans: patent TIPS (a), occluded stent (b).
•
CT Morphology
In a patient with suspected acute complications, a noncontrast CT followed by a biphasic examination should be performed to detect all potential complications. These complications include intrahepatic, subcapsular or intraperitoneal hemorrhage (hypo- to hyperattenuating, depending on age and blood clot formation), hepatic artery pseudoaneurysms (best seen on arterial phase images), arterioportal fistulae (contrasted portal branches in arterial phase images), bile duct dilatations due to hemobilia, bile collections (fluid attenuation), and stent dislodgment to heart or lungs. For evaluation of TIPS patency, CTA with thin collimation of 3 mm or less should be used. While the detection of stent occlusion is usually not a problem, associated obstructions of the stent or the hepatic veins may be missed if too large a collimation is used. For this indication, thin-section (4xl mm) multislice CT is optimally suited (Fig.T1.62). To enable good quality CTA reconstructions, water should be used as oral contrast medium. Minimal Invasive Therapies
Because few patients are surgical candidates for resection and the poor results of radiation and chemotherapy for liver tumors, a number of minimal invasive therapies have recently been introduced. The reported I-year survival rates vary with technique and can reach 90% and more.
c Thick sections make the evaluation of the dense metallic struts (c).
impossible
because
473
11 The Liver
Percutaneous Tumor Ablation
Radiofrequency (RF) ablation is based on electric currents in the RF range (460kHz) that are applied via shielded needle electrodes to produce coagulative necrosis of focal lesions. Microwave ablation relies on microwaves for thermal coagulation The procedure is usually done percutaneously using 18 G needle electrodes. For both methods, patients with four or less lesions that are preferably smaller than 3-5 cm can be treated. Based on lesion size and ablation of a 5-10 mm cuff of normal tissue, multiple (1-12) sessions are needed for complete ablation. Interstitial laser plwtocoagulation (I L P) is based on percutaneous thermal ablation with neodymium :yttrium-aluminum -garnet (Nd:YAG) lasers using US or CT-guidance, although this method is also MR1 compatible. Patient selection is similar to other techniques, and with multiple needles necrotic lesions of 60-70 mm can be created. Cryosurgery uses freezing for coagulative necrosis, either with liquid nitrogen or argon gas. A number of cryoprobe designs are offered, with sizes ranging from 2 to 10mm in diameter. It is primarily an intraoperative technique using US for guidance but has been used with MR as well. Tumors are frozen using 1-2 probes in single procedures. Ablation by percutaneous etbanol injection (PEl) is one of the most frequently used techniques for therapy of primary liver malignancies like HCC. Ethanol (95% concentration) results in dehydration and coagulation necrosis followed by fibrosis. It is used in cirrhotic patients with an HCC load less than 30% ofliver volume, butis most effective in lesions smaller than 3cm. It can be performed either percutaneously or under general anesthesia. With percutaneous approaches in outpatients, multiple (4-12) sessions are needed for complete ablation; the inpatients' procedure use a one-shot approach. More detailed information can be found in chapter 6: CT interventions. •
CT Morphology
The necrotic tissue is invariably hypo dense on contrast-enhanced CT, and has a higher conspicuity on portal phase or delayed images. Persisting neoplastic tissue is enhancing on arterial phase images, but may be iso- to hypoattenuating on portal phase or delayed imaging
a
b
Fig. 11.63 Necrotic HCC nodule (a) after percutaneous alcohol injection (PEl). Formerly hyperattenuating HCC nodule after transarterial chemoembolization (TACE) (b). Note the gas inclusions in the lesion due to necrosis (b) or percutaneous puncture (a).
(Fig. 11.63). For this reason, two- or three-phase imaging is advocated in the follow-up after ablative therapies. Needle tracts can easily be visualized as band-shaped hypo attenuating areas on early post -therapeutic CT. After treatment, nodule size can be unchanged or larger, and often nodule shape is modified. An enhancing rim of inflammatory granulation tissue can persist for several months. A number of associated findings like perihepatic or pleural effusion, segmental biliary duct dilatation, portal vein thrombosis and hepatic infarction can be found. Transcatheter Arterial Chemoembolization (TACE)
TACE is a transcatheter technique for unresectable HCC and metastases from neuroendocrine and possibly colorectal primaries. It takes advantage of the preferential blood supply of malignant lesions via the hepatic artery. It allows for a 100-fold more concentrated chemotherapy and markedly increased contact time between the drugs and tumor in comparison with systemic infusion. TACE is a regional perfusion technique that will treat the entire liver (or liver segments) irrespective of size and location of the tumors, and is most often performed for lesions larger than 3 cm. It is performed by selective catheterization of the segmental or subsegmental hepatic arteries by
Minima/Invasive
2-3 F microcatheters and injection of a combination of chemotherapeutic drugs (cis pla tin, doxorubicin, mitomycin C) and embolization particles (PVA or Gelfoam), Iodized poppy seed oil (Lipiodol) is often used to act as a carrier and to increase drug contact time, Response rates are 60-80%, with l-year survival rates of the order of70%, Early randomized trials showed no benefit in long-term survival, but recent studies in selected subgroups show better results, In resectable cases, TACE prior to liver resection may improve the results significantly and it may also be used for recurrent tumors,
Therapies
CT Morphology
When iodized oil is used, embolized lesions are recognizable by the accumulation of iodized oll, In these cases, CT has limitations in detecting viable tumor tissue during follow-up, It has been shown that the necrotic part selectively takes up the oily contrast medium, To avoid mistakes in evaluating efficacy, a waiting period of 1 month is advised, Tumor size reduction does not correlate well with therapeutic effect, which is explained by the delayed resorption of necrosis after embolization, In hypervascular lesions, TACE efficacy is best evaluated using arterial phase CT as response is closely correlated with a decrease in arterial vascularity (Fig, n,63),
475
12 The Biliary Tract M. Prokop
Anatomy Examination Technique EnhancementCharacteristics Dilatation of the Biliary Tract
_
Caroli's Disease (p. 482) Choledochal Cyst and Choledochocele Obstructive Cholestasis (p. 483)
(p. 483)
Papillary Stenosis (p. 484) Mirizzi Syndrome (p. 485) Cholang itis(p. 485) Biliary-Enteric Fistula (p. 486)
Tumors Biliary Cystadenoma, Cystadenocarcinoma Bile Duct Carcinoma (p. 487) Gallbladder Carcinoma (p. 488) Cholelithiasis
(p. 487)
and Inflammatory Gallbladder Diseases
Cholelithiasis (p. 491) Cholecystitis (p. 491) Pericholecystic Abscess (p. 493) Laparoscopic Cholecystectomy Preoperative Evaluation (p. 493) Postoperative Complications (p. 493) Trauma and Postoperative Changes Traumatic injuries (p. 495) Chemoembolization, Chemotherapy Postoperative Changes (p. 495)
(p. 495)
72 The Biliary Tract
The present imaging procedures of choice for diagnosing biliary tract diseases are ultrasound and MRI combined with magnetic resonance cholangiopancreatography (MRCP). Invasive procedures such as percutaneous transhepatic cholangiography (PTC), and endoscopic retrograde cholangiography (ERC) are only necessary if biliary interventions have to be performed. CT
Table 12.1 Tumor
Indications for CT examination
Equivocal
ultrasound
of the biliary tract
Gallbladder
diagnosis:
findings:
is mainly important for tumor staging (Table 12.1) but is also used in patients with equivocal ultrasound findings. The majority of biliary tract diseases are incidental or associated findings at CT. With the advent of multislice CT. however, CT may become a potent competitor to MR techniques for the diagnosis of biliary diseases.
carcinoma;
Staging,
Bile duct carcinoma: Papillary stenosis:
Tumor
Cholestasis:
Tumor
resectability detection,
Detection
resectability
of macroscopic
detection,
extrinsic
tumor,
resectability
compression
of com-
mon bile duct
Acute
abdomen:
Cholangiolithiasis:
Detection
Acute
cholangitis:
Detection of tumor with extrinsic cholangiogenic abscess
Acute
cholecystitis:
Part
Complicated
Laparascopic
cholecystectomy:
Preoperative angiography)
Postoperative
complications:
Hemorrhage,
cholecystitis, detection
abscess,
with CT cholangiography
of diagnostic
cholangitis of variants,
urinoma,
workup
abscess, common
compression,
for acute
gall stone
adomen
ileus
bile duct stones
(CT chol-
biloma
Anatomy Bile
Ducts
The intrahepatic bile ducts run parallel to the portal venous branches and hepatic arteries (Fig. 12.1). Only ducts with diameters of2-3 mm or more can be seen on noncontrast CT scans and 1-2 mm on contrast-enhanced scans. As a result, intrahepatic bile ducts are rarely visible on conventional or spiral CT either with or without intravenous contrast medium. The only exceptions are the main trunks of the right and left hepatic ducts, which are occasionally visible on contrast-enhanced scans as hypoattenuating structures up to 3 mm wide. With thin-section multislice CT. however, even more peripheral intrahepatic bile ducts may be visualized as thin hypoattenuating lines parallel to the portal veins in the parenchymal phase of hepatic enhancement. The common bile duct is consistently visualized as a rounded or oblong structure of low
attenuation (see Fig. 14.1). The normal common bile duct has a maximum diameter of 5 mm or less; 6 to 7mm is considered borderline, and 8 mm or more signifies dilatation. The diameter of the duct within the pancreatic head and porta hepatis should not exceed 5 mm, although diameters up to 10 mm are still considered normal following cholecystectomy. Gallbladder
The gallbladder is located at the inferior hepatic border between segments IVandV (se e Fig. 11.1). I ts size is highlyvariable, ranging from a collapsed state after meals to marked dilatation after prolonged fasting (e.g., in patients on parenteral nutrition). A gallbladder size of IOx3.4 cm (50 ml) is considered normal. The wall ofthe gallbladder is up to 3 mm thick and is difficult to see on unenhanced scans. The bile within the gallbladder has a CT density ofo to 15 HU.
Anatomy
Right and left ~ hepatic ducts
\
~~
vein cystic duct
~
.,.,~~-
Hepatic artery Com m 0 n -\-----\:;
YJ:rr:rr CJ](]JcJJdJ Normal
cif
Type 2
Type 4
Type 5
Type 1
bile duct Pancreatic duct
------/
Type 3
or Type 6
Fig. 12.2 Survey of the most important variants of the biliary system.
Fig. 12.1 Biliary tract anatomy. The common bile duct runs anterior to the portal vein in the porta hepatis and enters the pancreatic head above the duodenum. The papillary region itself is rarely visible with CT, appearing as a very small projection into the duodenal lumen. As a variant (inset), the common bile duct and pancreatic duct may open separately into the duodenum.
The gallbladder neck is the roughly triangular, tapered part of the gallbladder, often angulated, that is directed toward the porta hepatis. The cystic duct is less than 2 mm in diameter and usually cannot be seen unless thin-section multislice CT is employed. In 95% of cases it opens posteriorly into the common bile duct. •
Variants
Variants of gallbladder and bile duct anatomy are noted in up to 10% of cholangiograms and usually involve multiplication and septations of
the gallbladder, atypical arborization or terminations of the bile ducts, and duplications of the cystic duct or common bile duct. These variants (Fig. 12.2) are important because of an increased frequency of postoperative complications (bile leak), but they can seldom be directly visualized with CT unless thin-section spiral or multislice CT with biliary contrast agents (CT cholangiography) is employed. Congenital hypoplasia or agenesis of the gallbladder is rare «0.1 %). Septations (longitudinal, transverse, or caused by angulation) and diverticula of the gallbladder are rarely detectable with CT and are of minor importance. An ectopic gallbladder has been described at various sites in the upper abdomen and may lie anterior, posterior, cranial, or caudal to the liver. Hepatic cirrhosis may be associated with an atypical gallbladder location in which the gallbladder fundus lies on the anterior surface of the liver, where it can mimic a hypo attenuating mass.
479
480
12 The Biliary Tract
Examination Technique
cr
scanning of the biliary tract is commonly performed as part of an upper abdominal examination (see protocols in Chapter 11). Special protocols are only required if there is a particular interest in the biliary system (Table 12.2).
Table 12.2
Acquisition
Parameters
Data acquisition is similar to liver scanning (Chapter 11). For optimum evaluation of the biliary system, however, a thin-section technique should be employed. With spiral CT, a collimation of 3 mm with a table feed of 5 mrn can be recommended but this technique will require long scan durations
Protocol for CT scanning of the biliary tract
General 500 rnl, administered 30 min prior to examination 1-1.51, administrated 60 min prior to examination
Oral contrast
Upper abdomen: Whole abdomen:
Patient position
Supine with elevated arms
Scan range
Liver
Respiratory
phase
Inspiration
Windowing
Noncontrast CT: Stones: Contrast-enhanced CT: CT cholangiography:
W/L~250/40 W/L~ 150/10 W/L~ 350/70 W/L ~ 700/150
Scan parameters
1-slice SqTF/RI
4-slice SC
a
16-slice" SI::
Liver + biliary system
3/5/2
t
1-1.25
t
0.5-1.5
(Contrast injection
1-slice
axial SW/RI
MPRb SW/RI
4/3
3/3 cor
~------~-----~~~~~~~~--~~-=------- V/F/D
-----
4-sliEe V+N/F/D
t
----
16-slice V+N/F/D'
Portal phase CT Biphasic CT
150/4/70 150/4/25 + 70 d
Delayed phase CT
10-15 mine
CT cholangiography-i.V.
100 ml Biliscopin drip infusion over 1 hr
CT cholangiography-oral
2 x 3 g Telepaque the night before the exam; fatty meal 20 min bevor exam
150 + 50/5/60 150 + 50/5/25 + 70 10-15mine
d
150 + 50/5/205 150 + SO/ 5/10A+80d 10-15 mine
SC~ slice collimation (mm), TF ~ table feed (mmjrotation], RI ~ recon. increment (mm), i t ~scanning direction, SW ~ effective section width (mm), MPR ~ multiplanar reformats, axial ~ axial sections, cor ~ coronal, sag ~ sagittal CM ~ contrast material, Contrast concentration ~ 300 mg/ml iodine V~ volume of CM (ml), N ~ volume of saline flush (ml), F~ flow rate (rnljs), D ~ start delay, X ~ trigger region , Pitch P~ TF/(NxSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI ~ 1-1.5/0.7 or 0.5-0.8/0.5; oblique coronal MPR through the common bile duct and pancreatic duct; CPRor MinlP parallel to the common bile duct c read: 205 (1OA)~ start delay 20 s (10 s) after reaching 50 HU enhancement in the spleen (desc. aorta) dread: biphasic scan (arterial + portal phase) after 25 s (lOA) for the 1st scan, 70-80 s for the 20d scan e read: delayed phase scanning after 10-15 min. following either a portal phase or biphasic scan
Image Processing 30-40 s) or only allow for coverage of the hilar region in less cooperative patients. Multislice CT yields best results with thin collimation (1-1.25 mm with 4- and 8-slice scanners. 0.5-1 mm with 16-slice scanners) and pitch factors P around 1.5 (P" = 6 for 4-slice scanners). For routine evaluation. 5-7.5 mm thick axial sections can be reconstructed but more detailed evaluation of the bile ducts requires thinner (2-4 mm) sections. Multiplanar reformations of the bile ducts rely on the reconstruction of a secondary raw data set. from which coronal and curved planar reformation. as well as 3D displays can be obtained.
In oral CTC, 6g of cholangiographic contrast (iopanoic acid) is administered in two doses the night before the CT examination. The CT examination should be scheduled 8-12 hours after the administration of the contrast medium. If a gallbladder is ;n situ, patients receive either a fatty meal 20-30 minutes before the examination or an intravenous injection of cholecystokinin during the examination to improve bile duct opacification. The technique is contraindicated during pregnancy and lactation as well as in patients with renal insufficiency or hepatic dysfunction. In order to obtain diagnostic levels of enhancement, the serum bilirubinate level should not exceed 2-3 mg/dl (35-50 mmolll).
Contrast Administration •
Image Processing
Noncontrast CT
A non contrast scan should be obtained in cases of unexplained cholestasis and in patients evaluated for tumors. as it will facilitate the differentiation of tumors and calculi. Thin-section CT is recommended. Contrast-enhanced
liver CT
Thin-section spiral or multislice CT imaging during the portal phase is best for demonstrating the intrahepatic bile ducts. as it provides optimum contrast between the hypoattenuating bile and the liver parenchyma. Biphasic CT of the liver is useful for tumor staging. since bile duct tumors may be either hyper- or hypovascular. Delayed scans after 10-15 minutes can aid in the detection of hypovascular cholangiocarcinoma (interstitial enhancement). •
CT Cholangiography
CT cholangiography (CTC) uses oral or intravenous administration of cholangiographic contrast medium to opacity the biliary tract prior to scanning. Intravenous CTC uses infusion of 100ml cholangiographic contrast (iotroxate) over 30-60 minutes directly prior to the CT study. To lower the risk of adverse reactions, a slow infusion rate should be used. In two large series only minor reactions were observed with a frequency of 0.5-0.7% of cases.
Thin-section multislice with intravenous contrast material application provides the basis for imaging of the central biliary system using thin curved planar reformations (CPR) of 2-3 mm thickness (see Figs. 12.4, 12.10, 12.13) or thin minimum intensity projections (mIP, MinIP) of 3-5 mm thickness. CT Cholangiography
After the administration of cholangiographic contrast material. MIP or VRT provide an overview of the biliary system but axial sections and multi planar reformations using optimized windows (W 500-800, L 70-150) are required to rule out intraluminal stones. •
Virtual CTCholangioscopy
This technique uses virtual endoscopy to evaluate the inner surface of the biliary system. Best results are obtained from CT cholangiography data sets with good positive enhancement of the biliary system. To optimize detail definition and keep image noise low, image reconstruction should involve a small targeted field of view (FOV) of some 15-20 em and a moderately smoothing filter kernel.
481
72 The Biliary Tract
Enhancement Characteristics
Portal phase scans and delayed scans most clearly demonstrate the intrahepatic bile ducts as a system of channels that are hypoattenuating relative to the liver parenchyma. Normally the walls of the bile ducts and gallbladder do not enhance with intravenous contrast. The presence of enhancement signifies an inflammatory process or, if focal or accompanied by irregular thickening, tumor invasion. Arterial phase enhancement has been reported in malignant tumors but enhancement may also be present in inflammatory disease. Normally, 1-2% of the administered dose of intravenous iodine is eliminated by hepatic excretion. If renal function is impaired or if a large
volume of intravenous contrast has been administered (> 1OOml), there will be significant hepatobiliary contrast excretion producing a variable degree ofgallbladder opacification. This can usually be demonstrated for 24 to 48 HLJours after the contrast injection and may persist as long as fasting is continued and the gallbladder is not stimulated to contract. Opacification of the intrahepatic bile ducts may be seen in patients with severe renal function impairment (rare). As a consequence of hepato biliary contrast excretion, contrast material may be detected in the bowel (especially the colon) up to a week after intravenous contrast administration.
Dilatation of the Biliary Tract
Congenital cysts of the intrahepatic and extrahepatic biliary tree are rare. Much more common are acquired dilatations leading to cholestasis. Caroli's Disease
Probably inherited as an autosomal recessive trait, Caroli's disease is a rare disease characterized by segmental saccular and cystic dilatations of the large intrahepatic bile ducts. Patients become symptomatic in childhood or early adulthood. The syndrome may be associated with medullary sponge kidneys, cystic kidneys (infantile form), congenital cirrhosis of the liver, or choledochal cysts (rare). Patients do not display signs of cirrhosis or portal hypertension. Complications include biliary calculi, recurrent episodes of cholangitis, hepatic abscesses, sepsis, and an increased risk of developing cholangiocellular carcinoma (CCC). The primary diagnostic modality is ultrasound. CT can be used for noninvasive confirmation of the diagnosis but is used mainly in the detection of complications. •
directed toward the porta hepatis (Fig. 12.3). The bile ducts may show a beaded pattern of saccular dilatation. The dilatations may communicate directly with the bile ducts, or individual "cysts" may be linked together by bridge-like connections. Some portal branches may be completely surrounded by dilated bile ducts (the "central dot" sign). Intraluminal projections, sludge, or calculi may be seen. The extrahepatic bile ducts are frequently ectatic. Differential diagnosis: dysontogenic Iiver cysts.
CT Morphology
Caroli syndrome presents with multiple intrahepatic cystic lesions that follow the course of the portal vessels and bile ducts and are
Fig. 12.3 Multiple cysts, some superinfected, are distributed along the portal system in Caroli syndrome.
Obstructive Cho/estasis
Choledochal Cyst and Choledochocele
A choledochal cyst is a segmental dilatation of the common bile duct caused by the reflux of pancreatic enzymes at an unusually high union of the pancreatic duct with the common bile duct. Other anomalies of the hepatobiliary system may also be present. There is an approximately 40% risk of developing gallbladder carcinoma and a 15% risk of CCc. Common bile duct stones are frequent (up to 50% of cases). A choledochocele (synonyms: gallbladder diverticulum, duodenal duplication cyst) is a cystic dilatation of the distal intramural part of the common bile duct that herniates into the duodenum. Calculi and sludge are commonly present. A choledochal cyst is usually diagnosed with ultrasound, while a choledochocele is confirmed by MRCP or ERCP. These lesions are rare incidental findings with CT, which is not indicated as a primary study. •
CT Morphology
A choledochal cyst appears on CT as a well-circumscribed cystic mass (2-" 15cm) that is located in the porta hepatis and often contains calculi (a pathognomonic sign). Spiral or multislice CT can occasionally demonstrate the connection of the cyst with the common bile duct. Approximately 15% of cases show dilatation of the intrahepatic bile ducts due to stenosis
a
b
Fig. 12.4 Cholestasis with dilatation of intrahepatic bile ducts, (a) Predominantly left-sided stasis caused by a bile duct carcinoma that is not demonstrated by CT. (b) Bilateral ductal dilatation caused by a large central
caused by compression or lithiasis. The bile ducts distal to the cysts are of normal size. A choledochocele appears as a cystic mass at the point where the common bile duct enters the duodenum. Good duodenal opacification is essential. Spiral or multislice CT is the best technique for localizing and identifying the lesion. Differential mesenteric hepatic or pancreatic
diagnosis: duodenal diverticulum; or omental cyst; renal, adrenal, pancreatic cyst; pseudocyst of the head.
Obstructive Cholestasis
An obstruction at the level of the intrahepatic or extrahepatic bile ducts leads to a local or generalized dilatation of the intrahepatic ducts. The obstruction may be caused by inflammation, scarring, calculi, intraluminal tumors, or extraluminal compression (by a tumor, lymph node, or gallbladder). Ultrasonography and MRCP are the primary imaging studies, but multislice CT may gain an increasing role. CT is used in the diagnosis of tumors or complications. •
CT Morphology
Noncontrast scans show the dilated intrahepatic bile ducts as hypo attenuating structures of water attenuation (Fig. 12.4). The ducts distal to the ob-
c mass (CCC, cholangiocellular carcinoma), (c) Moderate dilatation of the bile ducts due to a stenosis secondary to chronic pancreatitis (4 x 1/6, coronal oblique MPR).
483
72 The Biliary Tract
struction are not enlarged. The site ofthe obstruction is best appreciated on spiral or multislice scans acquired after IV contrast administration. Depending on the CT technique employed, the cause of the obstruction may not be directly visualized (Fig. 12.4a). Multislice CT, however, is frequently able to determine the etiology of an obstruction (Fig.12.4c). Inflammatory changes lead to irregular variations in ductal cali bers that are often difficult to detect. Intraluminal stones can usually be detected only if they are calcified or if the scans have been enhanced with cholangiographic contrast medium (see Fig. 12.14). The abrupt termination of a duct with no apparent cause suggests an intraluminal tumor, usually a cholangiocarcinoma (see Fig. 12.10). Papillary Stenosis
Papillary stenosis has many potential causes but is usually caused by a cicatricial stricture that follows the passing of a stone (usually a com-
mon bile duct stone). Periampullary tumor or sphincter dyskinesia are infrequent causes. Dilatation of the common bile duct is usually an incidental finding at CT, which is of limited value in detecting a periampullary tumor (moderate sensitivity of conventional and spiral CT). The imaging procedures of choice are MRCP and endosonography. Again, multislice CT may play an increasingly important role. •
CT Morphology
CT shows prestenotic dilatation of the common bile duct to 8 mm or more, which is often (80%) associated with enlargement of the pancreatic duct ("double duct sign") indicating an obstruction at the level of the papilla (Figs. 12.5, 12.6). Thin-section scanning improves the demonstration ofa papillary tumor, and multiplanar reformations along the duct may be required for optimum delineation of abnormalities. A papillary tumor may be hard to differentiate from an enlarged intraduodenal papilla unless it shows different contrast enhancement characteristics.
Fig. 12.5 Papillary stenosis caused by a small pancreatic carcinoma located near the papilla. The common bile duct and pancreatic duct (arrowheads) are obstructed (a). The tumor has begun to infiltrate surrounding fatty tissues (arrow) (b).
a
b
Fig. 12.6 Small carcinoid tumor of the papilla (a) with hypervascularization in the arterial phase (arrow) and a hypervascular lymph node metastasis (arrowhead). The double duct sign with dilatation of the common bile duct (arrowhead) and the pancreatic duct (arrow) is best appreciated on a semicoronal view (4 x 2.5/15) (b).
a
b
Mirizzi Syndrome Resectability issues in patients with papillary tumors are discussed in Chapter 14.
(oriental cholangiohepatitis) chapter 11. •
Mirizzi Syndrome
Mirizzi syndrome results from extrinsic compression of the common bile duct by a stone impacted in the gallbladder or cystic duct, often accompanied by a chronic inflammatory reaction. CT is indicated only in cases where MRCP and ultrasound cannot distinguish the syndrome from compression by an enlarged lymph node or gallbladder tumor. •
is described
in
CT Morphology
The intrahepatic and extrahepatic bile ducts are slightly to moderately dilated, and some ducts may contain gas. The amount of dilatation does not correlate with the severity of the cholangitis. Contrast enhancement of the bile duct wall is often present. Chologenic abscesses appear as multiple hypo attenuating intrahepatic lesions, often with ill-defined margins, that also may contain gas. These lesions are best appreciated on contrast-enhanced scans (Fig. 12.7).
CT Morphology
CT shows dilatation of intrahepatic and extrahepatic bile ducts proximal to the compression site and a gallstone impacted in the gallbladder neck or cystic duct. The relationship of the stone to the common bile duct is best appreciated with thin-section helical or multislice CT. Cholangitis
Acute (ascending) cholangitis may be secondary to a benign or malignant bile duct obstruction (stricture, stone, parasitic infestation, papillary carcinoma) and is usually caused by Gramnegative bacteria. It leads to symptoms of right upper quadrant pain,jaundice and sepsis (Charcot's triad). The suppurative form of ascending cholangitis has a high mortality rate and is often due to an 0 bstructing stone or tumor. CT is indicated to exclude tumor compression and to detect intrahepatic chologenic abscesses (see chapter 11). Recurrent pyogenic cholangitis
Primary Sclerosing Cholangitis (PSC)
Primary sclerosing cholangitis (PSC) is a chronic progressive, fibrosing inflammatory reaction of the bile ducts (pericholangitis) of unknown etiology. It is most prevalent during middle age (third to sixth decades) and occurs twice as frequent in men as in women. The disease leads to bile duct obstruction, cholestasis and eventually biliary cirrhosis and portal hypertension. The intrahepatic ducts are always involved and in 25% of patients the disease is limited to these ducts. Involvement of the extrahepatic and pancreatic ducts is more variable. PSC is often associated with inflammatory bowel disease, mainly ulcerative colitis, and up to 70% of PSC patients can have this disease. Less commonly associated diseases include Crohn's disease, retroperitoneal fibrosis, fibrosing mediastinitis, Riedel's thyroiditis, and Sjogren's syndrome. There is an increased risk of developing cholangiocellular carcinoma
Fig. 12.7 Cholangitis. (a) Acute cholangitis with bile duct dilatation and multiple ill-defined chologenetic abscesses with a hypoattenuating rim of granulation tissue, (b) Chronic cholangitis secondary to papillary stenosis. The walls of the bile ducts show contrast enhancement.
a
b
485
72 The Biliary Tract
(CCC), especially in long-standing and advanced cases. Initial diagnosis usually requires ERCP and biopsy, but follow-up MRCP is currently the diagnosic procedure of choice. CT can have a complimentary role to cholangiography in staging and is useful for detecting complications like bile duct stones, cirrhosis or CCC. Secondary sclerosing cholangitis can occur as a response to a variety of causes of bile duct damage like surgery, cholangitis, AIDS and hepatic artery infusion chemotherapy.
scans. The presence of scattered, dilated intrahepatic bile ducts with bile duct stones further suggests the primary disease.
•
•
CT Morphology
CT shows segmental dilatation, constriction, and beadingofthe intrahepatic bile ducts. Skip dilatations are highly suggestive for the disease. Strictures often involve the confluence of the left and right hepatic duct. The extrahepatic bile ducts also show alternating dilatation and stenosis. Nodular thickening may be seen in some of the duct walls. The walls of the extrahepatic bile ducts show contrast enhancement (Fig. 12.8), although this is a relatively aspecific finding. The cirrhosis ind end-stage PSC shows a different morphology than cirrhosis from other causes. The liver typically shows marked organ deformity with severe contour lobulations, atrophy of the lateral and posterior segments and hypertrophy of the caudate lobe (segment I) which may take the form of a pseudotumor. The right lobe may be hypodense (fibrosis) relative to the caudate lobe, best seen on non contrast
a
b
Fig. 12.8 Irregular dilatation of intrahepatic bile ducts in sclerosing cholangitis (a). The degree of dilatation is best appreciated on a mlP view after contrast administration (b). Primary sclerosing cholangitis (PSC) is
Biliary-Enteric Fistula
Biliary-enteric fistulae are most often caused by a gallstone perforating the duodenum or colon, but they may also result from ulcer disease or gallbladder carcinoma. The fistula itself cannot be directly visualized on CT scans. CT Morphology
Biliary-enteric fistulae are consistently associated with air in the biliary tree (differential diagnosis see Table 12.3). Oral contrast medium may occasionally enter the bile ducts through the fistula. While CT cannot define the fistula itself, it will show direct contact between a portion of the bowel (duodenum, colon) and the gallbladder or common bile duct. Perifocal inflammatory reactions with unsharpness and Table 12.3
Causes of air in the bile ducts
Sphincter ofOddi incompetence (in elderly patients) Prior ERCor ERCP(up to 2 days after study) Prior papillotomy Prior biliary-enteric anastomosis: choledochoenterostomy; cholecysto-, hepatico- or hepatojejunostomy Cholangitis
c characterized by wall enhancement of the common hepatic duct and secondary biliary cirrhosis with nodular transformation of the liver (c).
Biliary Cystadenoma, Cystadenocarcinoma
b
a
Fig. 12.9 Perforation of a gallbladder stone into the duodenum. A typical finding is air in the common bile duct and the gallbladder (a). The large stone is visible in a loop of jejunum (b). Note the small size of the distal
stranding in the surrounding fat may occur (differential diagnosis: gallbladder carcinoma). A gallstone ileus most often presents as a small bowel obstruction with marked dilation of bowel loops proximal to an occluding stone.
c bowel loops (arrowheads, b) and the accompanying dilatation of the proximal bowel segments (arrowheads, c).
These stones commonly are large and calcified. As an accompanying sign there is fluid around the gallbladder and almost invariably gas within the gallbladder and the common bile duct (Fig. 12.9).
Tumors
Ultrasound and MRIIMRCP are the primary modalities for diagnosing tumors of the bile ducts. CT may be used in tumor staging as an alternative to MRI to detect nodular tumor components and enlarged lymph nodes. Multislice CT may become an alternative for MRCP as well. Biliary Cystadenoma, Cystadenocarcinoma
Biliary Cystadenoma is a rare benign tumor that is similar to mucinous cystic neoplasms of the pancreas « 5% of all intrahepatic biliary cysts). The great majority occur in the intrahepatic bile ducts (85%). There is risk of malignant transformation to Cystadenocarcinoma. CT is performed in patients with equivocal ultrasound findings. •
CT Morphology
CT demonstrates a multilocular, septated, cystic-appearing intrahepatic mass (1-30 cm in diameter) of near water attenuation. The septa
and wall of the mass show faint enhancement. The diagnosis is established by the aspiration of mucinous or serous material. Lesions that degenerate to Cystadenocarcinoma display focal wall thickening or a soft-tissue component. Differential diagnosis: echinococcal cyst, necrotic hepatic metastasis, liver abscess, mesenchymal hamartoma.
Bile Duct Carcinoma
Bile duct carcinoma (extrahepatic cholangiocarcinoma) is far more common than intrahepatic cholangiocarcinoma (CCC). The peak incidence is in the sixth and seventh decades. Multiple risk factors are known. The cardinal symptom is painlessjaundice. The bifurcation of the hepatic duct is affected in 10% to 25% of cases (Klatskin's tumor), and more than 50% of the tumors arise in the
72 The Biliary Tract
common bile duct. The tumor spreads via the lymphatics, metastasizing chiefly to the hepatoduoclenal ligament (>30%) and the celiac lymph nodes (> 15%). Hepatic infil tration occurs in more than 20% of cases and peritoneal carcinomatosis in less than 10%. Hematogenous spread to the liver, lung, or peritoneum is extremely rare. While MRCP, PTC, or ERC are the procedures of choice for diagnosing bile duct carcinoma, CT is useful for excluding a visible tumor component, which is considered a poor prognostic sign and often makes the lesion irresectable. •
CT Morphology
Isolated dilatation of the intrahepatic bile ducts is the most important sign. Only about 40% of tumors can be directly visualized at the site of the obstruction with conventional CT but multislice scanning substantially improves tumor detection (Fig. 12.10). Exophytic tumor growth is clearly apparent, while intraductal polypoid growth is detectable in only 25% of cases on conventional CT, but this number will increase with the use of multislice CT. It has been demonstrated that biphasic scans improve the detection and staging of these tumors. Some intrahepatic bile duct tumors that contain abundant connective tissue are seen best on delayed scans acquired during the phase of interstitial enhancement. Spiral or multislice CT is excellent for detecting complicating factors such as encasement of the hepatic artery or portal vein.
a
b
Fig. 12.10 Bile duct carcinomas, (a) Central bile duct carcinoma (Klatskin's tumor) with obstruction of the intrahepatic bile ducts and peripheral metastases. (b, c) Intraductal carcinoma of the common bile duct (arrow-
Gallbladder Carcinoma
Carcinoma of the gallbladder is the commonest malignant tumor of the biliary system. The peak incidence is in the sixth and seventh decades, and the tumor is three to four times more common in women than in men. The risk of gallbladder carcinoma is increased in patients with a porcelain gallbladder (10-20% incidence), chronic cholecystitis, inflammatory bowel disease, or familiac polyposis. Most patients have a history of gallstones, but fewer than 1% of patients with gallstones go on to develop gallbladder carcinoma. More than 75% of gallbladder carcinomas spread beyond the gallbladder or metastasize to regional lymph nodes (TNM staging, Table 12.4). The tumors most commonly invade the liver (>50%), duodenum (> 10%), and colon (approximately 10%). Regional .lymph node metastasis (hepatoduodenal ligament, peripancreatic nodes, periportal nodes) occurs in more than 50% of cases. Hematogenous or intraductal spread is less frequent, but perineural spread is commonly seen with aggressive tumors. CT is used for tumor staging and is particularly useful for detecting spread beyond the gallbladder and lymphatic metastasis. •
CT Morphology
CT usually shows a hypo attenuating mass in the gallbladder bed infiltrating the adjacent fatty tissue and liver (Fig. 12.11). The tumor is rarely
c heads). There is abrupt termination of the dilated common bile duct (arrow) and contrast-enhancing neoplastic wall thickening best appreciated on semicoronal reformations (4 x 1/6).
Cholelithiasis Table 12.4 Tis
TNM Staging of gallbladder carcinoma
carcinoma in situ
T1 a
tumor infiltrates lamina propria
T1 b
tumor infiltrates gallbladder muscle layer
* T2
Table 12.5 Differential diagnosis noma
tumor infiltrates peri muscular connective tissue
T3
tumor perforates serosa or invasion of liver <2 cm or invasion of 1 other neighbouring organ
T4
invasion of liver >2 cm or into 2 or more neighbouring organs (stomach, duodenum, colon, pancreas, omentum, extrahepatic bile ducts)
N1
metastasis in nodes around cystic duct around CBD or in liver hilum (= in hepatoduodenal ligament)
N2
metastasis in peripancreatic head, in periduodenal, celiac and/or superior mesenterial nodes
Cholelithiasis
and Inflammatory
Usually generalized wall thickening < 10 mm
Xanthogranulomatous cholecystitis
Stones and a lobulated mass in the gallbladder
Polyps
Cholesterol polyp, hyperplastic polyp, granulation polyp: circumscribed nodular wall thickening
Adenomyomatosis the gallbladder
of
involve-
Tumor invasion from adjacent organs
a
Hyperplasia of the gallbladder wall, in some cases with a polypoid projection into the lumen Melanoma, lymphoma, or in leukemia Liver, pancreas, duodenum
b
Fig. 12.11 Gallbladder carcinoma with irregular wall thickening and hepatic invasion.
Gallbladder
Thin-section spiral or multislice CT may be used in a preoperative stetting before laparoscopic cholecystectomy (especially in combination with biliary contrast agents, as a CT cholangiography), in patients with acute abdomen and to help differentiate benign form malignant disease. MRI with MRCP is the most important competitor with the advantage that no biliary contrast material is necessary.
carci-
Acute or chronic cholecystitis
Metastatic ment
diagnosed in its early stages, which may be marked by asymmetric generalized or focal thickening of the gallbladder wall or a polypoid intraluminal mass. Tumor demarcation from the lumen is seen most clearly during the late phase after contrast administration. An increase in the CT density of an intraluminal feature by more than WHU is highly suspicious for a tumor. With a mucinous adenocarcinoma, fine granular or punctate calcifications may be seen. Enlarged regional lymph nodes are common. The tumors usually appear hypo attenuating after contrast administration, but some may show intense peripheral arterial enhancement. Table 12.5 provides the differential diagnoses for diffuse or focal thickening or tumors and tumor-like lesions of the gall bladder.
of gallbladder
Diseases
Cholelithiasis
Cholelithiasis is a common disease that predominantly affects middle-aged women. Gallbladder "sludge" is composed of gritty cholesterol crystals and calcium bilirubinate granules. Ultrasound is the imaging modality of choice, and cholelithiasis is usually an incidental finding at CT. For diagnosing choledocholithiasis,
489
72 The Biliary Tract
a risk stratification based on clinical, laboratory and ultrasound features can be used. For patients with low to intermediate risk MRCP is the method of choice. High risk patients and those with a positive US/MRCP should proceed to ERC with stone extraction. It has been suggested that CT cholangiography be used for the preoperative exclusion of cholangiolithiasis prior to laparoscopic cholecystectomy. This type of study requires use of a cholangiographic contrast medium and yields best results with thinsection mul tislice CT. •
CT Morphology
Stones in the biliary tract are visible on unenhanced scans only if their CT attenuation significantly differs from than that of the surrounding bile (0-20 HU). Thus, approximately 15-25% of biliary calculi are not detectable with CT. Cholesterol stones range from slightly hypo-
abc
attenuating (pure cholesterol, Fig.12.12a) to hyperattenuating (containing CaC03 and calcium bilirubinate, Figs. 12.12b, c). Pigment stones (calcium bilirubinate) are hyperattenuating (Fig.12.12d). Gas-containing stones result from dehydration and cavitation (Mercedes Benzsign, Fig.12.12b). In patients with cho]angiolithiasis, pre-contrast CT can demonstrate an obstructing stone in 50-90% of cases. Thin-section spiral and multislice CT yield the best results. If the radiation dose chosen is high enough to keep image noise low and if a narrow window setting (e.g., W/L= 150/10) is used, hypoattenuating noncalcified stones in the extrahepatic bile ducts can be detected to better advantage. However, isoattenuating and intrahepatic stones may be missed. The typical finding is a target or bull'seye pattern (Fig. 12.13). Nonenhanced scans are required to avoid confusing the calcified rim
d
Fig, 12.12 Gallstones: (a) hypoattenuatingr cholesteroll stone fl-oating in the gallbladder, (b) Mercedes Benz sign associatedl with gas-containing stones, (c) ring calcification around stones, arid (d) multiple pigment stones.
Fig. 12.13 Cholangiolithiasis in the common bile duct with a typical target pattern caused by an impacted, calcified stone. Note the decreased conspicuity after IV contrast enhancement.
a
b
Cholecystitis
Fig. 12.14 Hemobilia in the common bile duct causes a target pattern similar to the one seen in Fig. 12.13. (a) Semicoronal reformations demonstrate that the filling defect is not rounded and circumscribed but extends along wider portions of the duct (b) indicating sludge or hemobilia as the cause.
b
a
Table 12.6 Causes of increased CT density of the gallbladder contents Sludge Prolonged fasting (e.g., intensive care patients) Hepatobiliary excretion of intravenous contrast material Clotted blood in hemobilia Parasites (ascariasis) Gallbladder tumors
with a contrast-enhancing bile duct wall. However, sludge and hemobilia may also present as slightly hyperattenuating material in the bile ducts but may be differentiated from stones on multiplanar reformations using multislice CT (Fig. 12.14). Sludge has a moderately higher attenuation than the bile and shows layering in the dependent portion of the gallbladder. Prolonged fasting (intensive care patients) can cause the gallbladder contents to become isoattenuating to the liver tissue (differential diagnosis see Table 12.6). Cholecystitis
Acute cholecystitis is most prevalent in the fifth and sixth decades and affects women with a g3: 1 predominance. It usually results from transient obstruction of the cystic duct by an impacted gallstone. The diagnostic methods of choice are ultrasound and clinical examination (Murphy's sign). CT may be indicated in patients with an acute abdomen and atypical clinical findings. Acute cholecystitis may be found incidentally in CT follow-ups after intra-arterial
chemoembolization of the liver. As consequence of gallbladder hyperemia, there is an increased incidence of transient arterial hyperperfusion of adjacent liver parenchyma on biphasic liver scans. Chronic cholecystitis is the most common inflammatory gallbladder disease. It is an incidental finding in CT and requires differentiation from gallbladder carcinoma. Xanthogranolomatous cholecystitis is a rare chronic inflammatory disease characterized by foreign body giant cells and foamy histiocytes. Emphysematous cholecystitis affects men with a 5 :1 preponderance and shows a high mortality rate (15%). Gangrene and gallbladder perforation are the most frequent complications. The diagnosis is established by ultrasound and/or plain abdominal radiography. CT can detect emphysematous cholecystitis with high sensitivity during the evaluation of an acute abdomen. •
CT Morphology
Acute cholecystitis is characterized by thickening of the gallbladder wall > 3 mm (usually by 9mm). The wall enhances after contrast administration (Fig.l2.15a), occasionally showing a three-layered structure. Gallbladder hydrops (gallbladder width > 4ern) and gallstones are common. Pericholecystic fluid may represent an inflammatory exudate (Fig.l2.15a) or an abscess. Intraluminal air may signify a gallbladder empyema or emphysematous cholecystitis (Fig.l2.15c). With acute cholecystitis following chemoembolization of the liver, some of the highly hyperattenuating Lipiodol-containing embolic agent may occasionally be demonstrated in the gallbladder wall.
492
12 The Biliary Tract
a
b
Fig. 12.15 (a) Acute cholecystitis with wall enhancement and pericystitic fluid, (b) Edematous cholecystitis during chemotherapy for ovarian cancer, (c) Emphy-
c sematous cholecystitis with intraluminal air and airoutside the gall bladder lumen indicating perforation.
Xanthogranulomatous cholecystitis: There is Chronic cholecystitis: Gallstones are invariairregular gallbladder wall thickening, which bly present. Usually the gallbladder is increased in volume and shows smooth or irregular wall even may appear as a lobulated mass, rarely with calcifications. Stones are commonly presthickening (by an average of 5 mm). This thickening often affects the entire circumference of the ent. Pericholecystitic extension can simulate gallbladder, but may be asymmetric (Fig. 12.16). gallbladder carcinoma. Emphysematous cholecystitis: Starting about Postcontrast CT shows moderate wall enhancement, which may be seen only on delayed scans 24 hours after the acute event, air can be dem(Fig. 12.16a). In patients with calcifications ofthe onstrated in the gallbladder lumen (Fig.12.15c) gall bladder wall (Fig.12.16b, c) there is an in- and wall and rarely in the intrahepatic bile creased risk of developing gall bladder carci- ducts. Signs of perforation with a pericholecystic abscess are common (20%). noma. A distinction from malignant wall thickening may not always be possible based on CT criteria alone. Doubtful cases require sonographic follow-up or cholecystectomy.
a
b
Fig. 12.16 (a) Chronic cholecystitis with asymmetric thickening of the gallbladder wall (delayed phase), (b) Chronic cholecystitis with calcified gall bladder wall and
c multiple gall stones. CT cannot rule out the presence of aT1 gall bladder cancer (4 x 1/6, coronal MPR). (c) Porcelain gallbladder.
Postoperative Complications
Pericholecystic Abscess Gangrene that develops in the setting of acute cholecystitis can lead to a subacute perforation of the gallbladder wall. Ultrasound is the modality of choice. but CT may be indicated in cases where technical factors hamper ultrasound imaging (obesity),
Laparoscopic
Cholecystectomy
Preoperative Evaluation Laparoscopy has become the method of choice for the majority of patients scheduled for cholecystectomy, There are, however, a number of conditions that increase the risk of the procedure and may require prior intervention by ERe. The risk of bile duct injuries increases with anatomic variants, most importantly aberrant right hepatic ducts, and short cystic ducts, Chronic inflammatory changes with marked adhesions may complicate the procedure and may require open surgery, Bile duct stones are hard to extract during laparoscopic cholecystectomy and therefore should be removed prior to surgery, Asymptomatic patients with normal liver function tests and normal common bile ducts at ultrasound have a very low rate of bile duct stones, A positive detection of a stone at ultrasound mandates ERC with stone extraction, and a normal ultrasound rarely requires further work-up, Unless clinical, laboratory and US findings mandate direct ERC (e.g. cholangitis), MRCP is the technique of choice for diagnosis of anatomic variants and the detection of bile duct stones before laparoscopic cholecystectomy, CT cholangiography can be expected to yield similarly good results but requires the use of biliary contrast agents, •
CT Morphology
A zone oflow attenuation is usually found in the gallbladder bed and less commonly in the gallbladder wall (intramural abscess) or peritoneal cavity surrounding the gallbladder. Peripheral contrast enhancement may be visible but may be absent in early stages, Most intramural abscesses are detectable only after contrast administration.
CT Morphology
CT cholangiography using thin-section spiral or multislice CT can demonstrate filling defects in the extra- and intrahepatic bile ducts, As opposed to MRCP, CT can easily distinguish between intraductal air and stones due to their characteris tic CT numbers,
Lack of filling of the gallbladder suggests marked inflammatory reactions associated with substantial adhesions around the gallbladder. This finding has been suggested as a contraindication to laparoscopic approach and as an indication for open surgery, The detection of anatomic variants requires close scrutiny of axial images, best using an interactive cine display on a monitor. Volume-rendered images are superior to MIP to demonstrate the spatial relationship between the various portions of the biliary system. Typical variants of the cystic duct include a spiral cystic duct, a low junction of the cystic duct, a short cystic duct, and a cystic duct entering a right hepatic duct Variants of the right hepatic duct include aberrant ducts that drain directly into the common hepatic duct, common bile duct, the cystic duct, or the gallbladder (see Fig, 12,2),
Postoperative Complications Complications after laparoscopic cholecystectomy are strongly correlated with the experience of the surgeon, Laparoscopy itself may lead to injury of almost any abdominal organ, but liver, bowel. ureter, and vessels of the abdominal wall are most frequently involved, During laparoscopic cholecystectomy the common bile duct or a hepatic duct may be mistaken for the cystic duct and may be ligated, strictures may result from laser or cautery injuries, but also bile duct leaks may occur, Retention of stones in the common bile duct is possible, as is incomplete resection of the gallbladder. Clinically, leaks most frequently become symptomatic 1-2 weeks after surgery
493
12 The Biliary Tract
while structures take 2 weeks to several months to cause clinical symptoms. Patients may present with serologic signs of blood loss due to hemorrhage, or abdominal distension, mild pain and low-grade fever due to infected bilomas or bile peritonitis, or mildly abnormal liver function tests or minimal jaundice due to biliary obstruction. In the postoperative setting, CT is often superior over MRI because of less interference by perihepaticfl uid. •
CT Morphology
Normal postoperative findings that should resolve within the first week include pneumoperitoneum, subcutaneous emphysema, small amounts of fluid in gallbladder bed, adynamic ileus, pleural effusion, and lower lobe atelectasis. Postoperative ascites should resolve within two weeks. As a long-term effect after cholecystectomy there is compensatory dilatation of the common bile duct up to 1Omm (Fig. 12.17). Laparoscopy-related complications may present as peritoneal hemorrhage, liver lacerations, hematoma in the abdominal wall or in the rectus sheath, persistent pneumoperitoneum, or ascites. Suspected ureter leaks have to be
proven by extravasation of contrast -enhanced urine on delayed excretory phase scans after > 10 minutes. Bile leaks present as focal fluid collections, especially in the gallbladder fossa and the liver hilum, or as free fluid in the abdomen. Bile peritonitis will after some days cause peritoneal thickening and enhancement. The detection of fluid, however, is nonspecific because it can also be caused by seroma, lymphocele, hematoma, or abscess. A suspected bile leak requires confirmation by fine needle aspiration, CT cholangiography, or ERe. Ligation or stricture of a normal bile duct (most commonly an aberrant right duct or the common bile duct) leads to marked cholestasis in the affected segment. Prolonged cholestasis due to complete obstruction may lead to atrophy and cirrhosis of the drained liver segment. Bile duct ligation can be associated with injuries to the right hepatic artery (hemorrhage, stenosis, occlusion). Stones that have dropped into peritoneal space or that remained in the common bile duct can rarely be detected. (Optimum visualization would require CT cholangiography.)
Fig. 12.17 Unusually pronounced compensatory dilation of the common bile duct in a patient following cholecystectomy.
a
b
Trauma
and Postoperative
Changes
Traumatic Injuries
Postoperative Changes
Blunt abdominal trauma or penetrating injuries may cause hemobilia or biliary leaks. Liver lacerations usually involve the bile ducts. CT is an excellent tool to detect hematoma or biloma, but it is less well suited to prove a bile leak or hemobilia.
General postoperative complications after open cholecystectomy, bile duct surgery, or liver transplantation are similar to those after laparoscopic cholecystectomy, but they occur with different frequency. The most typical complication after bilodigestive anastomosis (usually choledochojejunostomy) is a stricture at the site of the anastomosis sometimes complicated by secondary cholangitis. After liver transplantation, structures at the site of the bile duct anastomosis may develop and are most often caused by ischemia due to a stenosis or thrombotic occlusion of the hepatic artery. In an occluded remnant of the cystic duct, mucoceles may form.
•
CT Morphology
Hemobilia may be hard to detect. It may present as an increase in bile attenuation or as hyperattenuating clots in the biliary system (see Fig. 12.14) or the gallbladder. Blood contents pool in the dependent portion of the gallbladder or of a dilated duct. Such hyperattenuating material. however. may also be caused by sludge, milk of calcium bile, small gallstones, or contrastmaterial. Traumatic avulsion of the gallbladder can best be diagnosed on reformations from thinsection spiral or helical CT that are positioned parallel or perpendicular to the gallbladder axis. There is an increased distance between liver surface and gallbladder and free fluid in the gallbladder fossa. Perihepatic fluid collections may be due to hematoma, biloma, abscess or loculated ascites.
Chemoembolization, Chemotherapy Chemoembolization or intra-arterial chemotherapy may lead to benign bile duct strictures due to focal ischemia. As a consequence offocal or generalized bile duct obstruction, secondary cholangitis and even chologenetic abscesses may develop. Chemoembolization may lead to cholecystitis if the cystic artery received some of the embolic agent. These agents are commonly mixed with radio-opaque contrast material and can be identified as small high-attenuation foci in the hepatic arteries, the liver or the gall bladderwall. CT is not the primary imaging tool but can help detect inf e c ti ous c om plica tions.
CT Morphology
Normal postoperative findings are the same as those described for laparoscopic cholecystectomy. After open surgery, they may persist for a longer period of time, depending on the complexity of the surgical procedure. Air in the biliary system is a normal finding, even years after bilodigestive anastomoses. Periportal hypoattenuation after liver transplantation is a normal finding that may persist for up to four months and is due to lymphatic congestion in the transplanted liver. This finding should not be confused with dilated bile ducts which are more tubular and usually extend more into the periphery. Mucoceles of the cystic duct remnant appear as rounded fluid collections in the liver hilum and often cannot be differentiated from loculated ascites, lymphocele, biloma, abscess, or a fluid-filled Roux-en-Yjejunalloop. Strictures of the bile duct anastomosis lead to a dilation of the intrahepatic bile ducts. Presence of gas in the bile ducts indicates patency but does not rule out strictures. Dilatation without gas implies a tight stenosis. Other findings (biloma, bile leak, free fluid, peritonitis) correspond to the ones described for laparoscopic cholecystectomy.
The
spleen
C. Schaefer-Prokop,
M. Prokop
Anatomy Examination Technique Enhancement Characteristics Accessory Spleen, Polysplenia.
Splenosis
Cystic Splenic Lesions
.
Congenital Splenic Cysts (p. 502) Post-traumatic Pseudocysts (p. 503) Postpancreatitic Pseudocysts (p. 503) Echinococcal Disease (p. 503) Splenic Abscess (p. 503) Lymphangiomas (p. 504) Solid Splenic Lesions
..........................••....
Benign Splenic Tumors (p. 504) Malignant Primary Tumors of the Spleen (p. 505) Metastases (p. 506) Lymphoma (p. 506) Diffuse Splenic Diseases, Splenomegaly Inflammatory
Diseases, Parainfectious
Lymphoma (p. 508) Portal Hypertension Sarcoidosis (p. 508)
(p. 508)
Trauma Vascular Disorders Splenic Infarct (p. 511) Splenic Vein Thrombosis (p. 511) Splenic Artery Aneurysm (p. 511)
Splenomegaly (p. 507)
13 The Spleen
The spleen is rarely a primary target organ for a CT examination but it is regularly included in an upper abdominal examination (Table 13.1). CT has a second-line role in locating an infectious focus (splenic abscess) or detecting splenic rupture after bl unt abdominal trauma and should be used whenever ultrasound findings are equivocal. CT is not sensitive enough to detect lymphomatous involvement of the spleen, but it is excellent for detecting splenic metastases from other tumors. CT is not helpful for the etiologic assessment of splenomegaly. The major competing modality is ultrasound. CT is superior to ultrasound in cases where sonographic imaging is difficult. MRI can also be used for splenic imaging but has not yet achieved much practical importance. MRI may gain increasing importance for detecting splenic
Table 13.1 spleen
Indications
for CT examination
of the
CT indicated Tumor staging
As part of upper abdominal examination
Search for splenic abscess
If ultrasound findings are equivocal
Blunt abdominal trauma (splenic rupture)
If ultrasound findings are equivocal
CT not indicated Splenic involvement Investigation
by malignant lymphoma
of splenomegaly
involvement in lymphoma using contrast agents specific for reticuloendothelial tissue.
Anatomy
The size and shape of the spleen are highly variable, and the diagnosis of splenomegaly is subjective, at least in the borderline range. The following biometric determinations can be made. The craniocaudal length of the spleen (L) has a normal range of 10-15 cm and is determined by the craniocaudal extent on CT. The maximum diameter (D) on any transverse image is 4-8 cm. The maximum thickness (T) is 3-5 cm (Fig. 13.1) measured at the level of the splenic hilum. If the splenic shape is markedly irregular, a number of measurements may be averaged to yield the thickness T. The splenic index LxDxT should be less than 480. A recent estimate of the splenic volume from spiral CT data has been provided by Prassopoulos et al. (1997): V-30+0.58xLxDxT
The authors found the splenic volume in adults to be largely independent of age and gender, height, weight, and body mass index. The mean volume was 215ml (range, 110340ml). The splenic weight can be estimated by multiplying the volume V by 1.05, the density of soft tissue. The splenic weight ;n vfvo is probably greater than the published weight from post mortem studies due to the large amount of blood contained within the spleen in vi'vo.
Splenic index= Lx W x T «480 is normal) D=maxJmum diameter T =thickness at splenic hilum L=craniocaudal length
Fig. 13.1 Cross-sectional diagram showing the relation of the spleen to the tail of the pancreas, the adrenal, stomach, and left colic flexure.
Examination
A list of causes of a "small" spleen is given in Table 13.2. while causes of splenomegaly are listed in Table 13.6. Normal splenic parenchyma has a CT attenuation of 40-50 HU (lOHU less than the liver). Changes in the relationship of splenic to hepatic attenuation are generally due to hepatic disease rather than a change in the spleen. Increased attenuation values are seen in primary hemochromatosis. sickle-cell anemia. and thalassemia. The spleen is an intraperitoneal organ that relates closely to the gastric fundus, the tail of the pancreas, and the left kidney and adrenal gland (Fig. 13.1). The spleen is sufficiently soft and pliable in texture that left upper quadrant abdominal masses or organ enlargement can cause considerable displacement and deformity in shape. Thus, tumors of these organs commonly displace the spleen but rarely invade it. Perisplenic abscesses or hematomas lead to considerable deformity that does not necessarily indicate splenic infiltration or rupture. The shape of the spleen may vary as a result of increased lobulation or impressions on the
Examination
13.3
13.2
Small
spleen
Congenital
Polysplenia syndrome Atrophy Prior irradiation Infarct Sickle cell anemia
(homozygous
form)
spleen from neighboring organs. Fissures in a lobulated spleen should not be mistaken for splenic rupture. Splenic indentation can mimic a pararenal or pancreatic mass at ultrasound but is easily recognized on CT scans. The splenic artery is often tortuous, and a long segment of the vessel may show calcifications, which should not be mistaken for pancreatic calcification. The splenic vein runs just posterior to the pancreas. Its confluence with the superior mesenteric vein can be identified on axial scans by a slight dilatation of the vessel. The inferior mesenteric vein enters the splenic vein several centimeters to the left of the venous confluence.
Technique
In most cases the spleen is scanned as part of an upper abdominal CTexamination. The examination technique and contrast protocol are geared toward the primary organ of interest (Table 13.3). A non contrast examination of the spleen is useful only for the detection offresh hemorrhage in trauma patients. Most other splenic lesions are difficult to detect on unenhanced scans. Table
Table
Technique
Protocol for CT scanning
of the
The parenchymal phase of enhancement is the most rewarding for diagnostic evaluation of the spleen. During the arterial phase, the splenic enhancement is markedly heterogeneous and may hamper an accurate evaluation ofthe splenic parenchyma (see below). Since scan delays of 40 to 60 seconds are used in examinations of the pancreas and liver, such artifacts should always be considered in evaluations of the spleen.
spleen
General Oral contrast
not mandatory
Patient
Supine
position
Upper
Scan range
with elevated abdomen
arms
from diaphragm
to lower pole of liver, in splenomegaly
includ-
ing pelvis Respiratory Windowing
phase
Inspiration
Noncontrast CT: Contrast-enhanced
CT:
W/L - 350/40 W/L =400/60 Continue
A_
13 The Spleen Table 13.3
(Continue)
I Scan parameters Standard, trauma Focal lesion
1-1.2St
Contrast injection Standard
IS-slice3 SC
axial SWIRl
1-1.5 I
5/4
0.75-1.251
4/3
3/3 cor 16-slice V+IM/F/D'
4-slice V+N/F/D 12012/60
MPR" SWIRl
120 + 5013160
120+50/3/205
SC = slice collimation (rnrn). TF = table feed (mm/rotation), RI = recon. increment (mrn), f [ = scanning direction, SW= effective section width (rnrn). MPR= multiplanar reformats, axial = axial sections, cor = coronal, sag = sagittal CM = contrast material. Contrast concentration = 300 mglml iodine V '"=volume of CM (rnl). N = volume of saline flush (ml), F = flow rate (mils), D = start delay, X '"=trigger region a
Pitch P=T FI (NxSC) should be chosen close to 1.5with 4-slice scanners, 1.3-1.5with 16-slice scanners
b
MPR are reconstructed from a "secondary raw data set" with SWIRl =1-1.510.7
c
read: 20S = start delay 1Os after reaching 50 HU enhancement in the spleen
Enhancement Characteristics
During the arterial phase of enhancement (e.g., in a biphasic liver study), the splenic parenchyma shows a characteristic, heterogeneous enhancement pattern (Fig. 13.2). This effect has not been fully explained yet but has been attributed to a differential perfusion of the
red and white pulp that produces an arciform and wave-like pattern of sharply defined enhancing and non enhancing areas. Because these heterogeneities can mimic focal lesions, the arterial phase should not be used for lesion detection in the spleen.
a b Fig. 13.2 Atypical striped enhancement pattern ofthe spleen is seen on arterial phase images and is caused by differential enhancement of the red and white pulp (a).
In rare cases (b) this pattern can mimic a lesion (arrow). Adjacentscans (c) suggest the correct interpretation by resolving the lesion into band-like structures.
c
Accessory Spleen, Polysplenia, Splenosis
During the parenchymal phase, which begins 60 to 90 seconds after the start of the injection, the normal splenic parenchyma has a
Accessory Spleen, Polysplenia,
CT Morphology
Accessory spleens are generally located in close proximity to the spleen (hilum, gastrosplenic ligament). Rarely they may occur in the pancreas or lesser pelvis. Most are less than 1 cm in
Fig. 13.3 Polysplenia in a patient with Karthagener syndrome and situs inversus. The multiple accessory spleens show an enhancement pattern that is characteristic of splenic tissue. L = liver, P = pancreas.
diameter, but accessory spleens as large as lOcm have been described. They are identified by their rounded shape, smooth margins, and enhancement characteristics that parallel that of the spleen (Fig. 13.3). Splenosis presents with multiple round or oval masses in a patient with a history of splenic trauma (Fig. 13.4). In most cases, no spleen is
Fig. 13.4 Splenosis after splenectomy for splenic trauma (4 x 1/6). There are multiple nodular peritoneal masses (arrows) that enhance similarly to splenic tissue (a) displayed on an oblique coronal section, as well as aVRTimage(b).
a
enti-
Splenosis
The developing spleen fails to fuse normally in 10-30% of the population, and multiple spleens develop in approximately 10-20% of these individuals. More than two accessory spleens occur in less than 5%. Polysplenia denotes a condition in which the spleen is subdivided into multiple parenchymal segments that usually have a rounded shape. It is mostly associated with various other complex abnormalities of the abdominal vasculature and gastrointestinal tract. Splenosis develops when splenic tissue is seeding within the abdomen or chest following trauma. This autotranplanted splenic tissue will undergo hypertrophy, leading to the formation of masses in the peritoneum, along the diaphragmatic surface, the mesentery and omentum, the liver, or even the pleura. Accessory spleens may undergo hypertrophy after splenectomy and may cause recurrence of symptoms in patients with hematologic disorders associated with hypersplenism. •
homogeneous appearance. Consequently, ties during this phase signify pathology.
b
501
13 The Spleen
present (resection after injury). All masses have a homogenous enhancement pattern but the degree of enhancement may vary depending on their blood supply.
Accessory spleens can mimic abnormal lymph nodes or pancreatic, adrenal, or intraperitoneal lesions on unenhanced CT scans.
Cystic Splenic Lesions (see Table 13.4) Table 13.4
Cystic splenic lesions
Congenital Splenic Cysts
Congenital cyst
Hypoattenuating «20 HU), smooth margins, homogeneous, possible wall calcifications
Post-traumatic cyst
Hypo- to hyperattenuating, smooth margins, homogeneous, possible wall calcifications
Pancreatic pseudocyst
Hypoattenuating, often very large, signs of previous pancreatitis
Echinococcal cyst
Calcifications (usually in the wall, occasionally in septa), daughter cysts, contrast enhancement of wall and septa
Abscess
Hypoattenuating, occasionally with indistinct margins; usually does not show rim enhancement
Candida sepsis
Numerous small focal hypodensities, possible wedge-shaped perfusion defects
Lymphangioma
Multiple cysts, relatively small and confluent with well-defined margins; a rare lesion
a Fig. 13.5
(b).
Splenic cysts frequently represent incidental findings at CT. Congenital splenic cysts (also called epidermoid, mesothelial, or primary cysts) are rare. They are epithelialized and are often associated with cystic kidney disease. •
CT Morphology
Cysts appear as well-circumscribed, homogeneous lesions that usually have attenuation values less than 20 HU. They do not enhance and do not show increased peripheral vascularity. Wall calcifications may occur in less than 15% and septations in less than 20% (Fig. 13.5). In 80% of cases, congenital splenic cysts are unilocular and solitary. Infrequently, the cysts show wall trabeculation or peripheral septation. Very rarely, congenital epidermoid cysts may contain cholesterol crystals (negative attenuation values) or blood.
b Splenic cyst (4 X 1/6) with tiny septation
and calcification
on axial image (a) and volume-rendered
image
Splenic Abscess Post-traumatic
Pseudocysts
Pseudocysts after trauma or infarction are not epithelialized and result from the cystic degeneration of hematomas. They are substantially more common than congenital cysts.
• cr
Morphology
Higher attenuation values may be seen after trauma or infarction. Post-traumatic cysts have been reported to become as large as 13 em. Wall calcifications (50%) and peripheral septations (85%) are more frequent in post-traumatic cysts than in congenital ones.
Postpancreatitic
Pseudocysts
Intrasplenic pseudocysts are very rare.
• cr
after acute pancreatitis
Morphology
Pseudocysts secondary to pancreatitis that arise in the tail of the pancreas adjacent to the splenic hilum occasionally extend beneath the splenic capsule or even into the splenic parenchyma. Such intrasplenic pseudocysts can reach considerable size. They are usually surrounded by a well demarcated wall that enhances after intravenous administration of contrast material. Signs of previous pancreatitis will suggest the correct diagnosis. Further splenic complications of pancreatitis include subcapsular hematoma. rupture. infarction. splenic artery pseudoaneurysm, and splenic vein thrombosis.
Echinococcal
Disease
involvement by Ecbinococcus granulosus (cysticus) is present in less than 5% of patients. It is rare, except in endemic areas (e.g., Argentina, Greece, Spain).
Splenic
•
echinococcosis. The cyst wall enhances after contrast adminis tra tion. Percutaneous aspiration should be avoided in an untreated patient because of the risk of allergic reaction to cyst content and spread of infection.
CT Morphology
Echinococcal (hydatid) cysts are well circumscribed low density lesions that enlarge the spleen. Wall calcifications are extremely common (» 50%), while intracystic calcifications are rare. The mother cyst (10-45 HLJ) may contain daughter cysts of slightly lower attenuation at its periphery, confirming the diagnosis of
Splenic Abscess
Splenic abscesses are more often multiple than solitary. Most develop in the setting of septicemia. Microabscesses are commonest in imrnunocompromised patients and usually result from a fungal infection (Candida. Aspergillus, Cryptococcus). Splenic abscesses in AIDS patients are additionally caused by mycobacteria and Pneumocystis carinii. Pyogenic abscesses may develop secondary to superinfection of an infarcted area or post-traumatic hematoma. Predisposing primary infections are endocarditis, urinary tract infection, surgical wound infection, pneumonia, and appendicitis. CT can detect abscesses only a few millimeters in size. The clinical presentation must be considered in order to make an accurate differential diagnosis. CT Morphology
Abscesses appear on CT scans as focal areas of low attenuation (Fig. 13.6). Usually there is no inflammatory rim. Many abscesses remain uniformly hypoattenuating after contrast administration and do not show peripheral enhancement (Fig.13.6a). The margins of the abscess may be ill-defined. Gas collections are very rarely seen (Fig.13.6b). Splenic abscesses are commonly associated with similar lesions in the liver (15-20%). Multilocular abscesses have a fungal etiology in more than 60%, while unilocular abscesses have a bacterial etiology in more than 90%. In Candida sepsis the spleen is permeated by numerous small lesions, and the occlusion of small arteries can lead to wedge-shaped infarcts (Fig. 13.6c). Calcifications are seen in treated Candidamicroabscesses and in lesions caused by other fungi, by mycobacteria, and Pneumocystis carinii. Focal hypoattenuating areas composed of granulomatous tissue may persist after therapy.
503
13 The Spleen
c
b
a
Fig. 13.6 (a) Staphylococcal sepsis with large subcapsular splenic abscesses, (b) Superinfected splenic infarctwith small gas inclusions, (c) Multiple small splenic
A CT scan without focal abnormality does not exclude the possibility of early infection, particularly in hematogenously disseminated fungal disease.
Lymphangiomas
Lymphangiomas are benign tumors that are categorized as capillary, cavernous, or cystic, depending on the size of the abnormal lymphatic channels. In the spleen the cystic type is com-
and hepatic abscesses are typical of Candida immunosuppressed patients (here: AM L).
sepsis in
monest. Lymphangiomas are more frequent in the neck and axilla than in the spleen. •
CT Morphology
Cystic lymphangiomas are composed of multiple, relatively small, sharply circumscribed confluent cysts, often in a subcapsular location, which are filled with lymphatic fluid and do not enhance after contrast administration. CT attenuation varies between 15 and 35 HU. Curvilinear calcifications have been reported. There is generalized enlargement of the spleen.
Solid Splenic Lesions Benign Splenic Tumors
The spleen is rarely affected by benign primary tumors (hemangioma, lymphangioma, hamartoma, fibroma, myxoma, chondroma, osteoma, desmoid tumors) (Fig. 13.7). Hemangioma is the commonest of these tumors (0.01-0.14 % in autopsy series). Splenic hemangiomas are freq uently associated with hemangiomas in other organs. In the spleen, cavernous hemangiomas are more common than capillary hemangiomas. A new entity specific for the spleen is the littoral cell angioma. It is characterized by multiple nodules composed of vascular channels of red pul p, and is more often diffuse than focal. A malignant angiosarcomatous variant has been described.
•
CT Morphology
Hemangiomas may be solitary or multiple and range in size from a few millimeters to 15cm. Capillary splenic hemangiomas present as homogeneously hyper- to isoattenuating nodules. Their dynamic enhancement pattern almost parallels that of the aorta but with a markedly lower peak enhancement. Cavernous hemangiomas may present with similar imaging characteristics as hepatic hemangiomas. On unenhanced scans they may be isoattenuating to the aortic blood and slightly hypo attenuating to the splenic parenchyma. Cystic components are frequent, and may be the dominant feature in some tumors. Multiple calcifications are com-
Malignant Primary Tumors of the Spleen Fig. 13.7 Hamartomaofthe spleen (4 x 1/6). Multiple hyperenhancing masses (arrows) seen only on arterial phase images (a) appear isoattenuating in the portal phase (b). There is a small cystic component.
a
mon, and are either punctate and curvilinear, or radially oriented from the center. Rupture and hemorrhage are reported in up to 25% of cases. Littoral cell angioma shows multiple small hypo attenuating nodules in the portal phase, ranging in size from 5-Wmm. The nodules become isodense to splenic tissue on delayed postcontrast images. Table 13.5
b
The CT features of these and other benign lesions are summarized in Table 13.5. Malignant Primary Tumors of the Spleen
Malignant tumors that arise in the spleen include (littoral cell) angiosarcoma, fibrosarcoma,
Solid splenic lesions
Tumor-like lesions Perfusion artifact
Early perfusion phase, ring-like or band-like; compare with adjacent slice
Splenic infarct
Band-like or wedge-shaped defect broadly based on the splenic capsule; coronal or sagittal sections may be required
Sarcoidosis
Rounded hypovascular lesions
Amyloidosis
Rounded hypovascular lesions
Hematopoiesis
Rounded hypovascular lesions
Gaucher's disease
Rounded hypovascular lesions
Peliosis
Isoattenuating to blood, enhancement characteristics similar to splenic artery
Benign tumors Hemangioma
Solitary or multiple, isoattenuating to blood on noncontrast and delayed scans, multiple calcifications, some may show same enhancement pattern as liver hemangioma. Capillary hemangiomas are homogenous, cavernous types often contain cystic portions
Hemang iopericytoma
Multiple small confluent nodules; solid and cystic; intense enhancement (solid portion)
Littoral cell angioma
Multiple small hypodense nodules; isoattenuating on delayed phase; diffuse> focal
Hamartoma
May contain fatty components and amorphous hancement
Fibroma, myxoma
Rounded hypovascular lesions after contrast administration; rare
Desmoid tumors
Hypovascu lar lesions; rare
Chondroma
Rounded hypovascular lesions with amorphous calcifications;
Osteoma
Rounded, homogeneous plaquelike calcifications;
calcifications;
heterogenous en-
rare
rare Continue
>
505
13 The Spleen Table 13.5
Continue
Malignant tumors Lymphomas
Rounded hypovascular lesions, rarely showing rim enhancement or heterogeneities (detected injust 20% all lymphomas involving the spleen)
Metastases
Heterogeneoushypo- or hypervascular lesions, usually multiple. Central necrosis possible. Most common primaries: melanoma > breast carcinoma > bronchial carcinoma > colon carcinoma > renal cell carcinoma >others
Angiosarcoma
Heterogeneous,very hypervascular, history ofThorotrast administration; rare
Kaposi'ssarcoma
Immunocompromised patients, heterogeneous hypervascular
Hemangioendothelioma
Heterogeneous hypervascular lesions after contrast administration; rare
and leiomyosarcoma, malignant teratoma, and malignant fibrous histiocytomas. The development of angiosarcomas has been linked to the prior administration of Thorotrast, a thoriumcontaining contrast agent used in the 1930s to 1950s. Some authors have used the term (littorial cell) hemangioendotheliomas to refer to vascular tumors of borderline malignant potential. •
CT Morphology
The CT appearance of these tumors is nonspecific. Primary splenic malignancies appear as relatively large, heterogeneous masses that may contain cystic and solid components. Hypervascular tumors (especially angiosarcomas) show marked, heterogeneous enhancement following contrast administration and often contain areas of necrosis. Attention should be paid to obvious criteria of malignancy such as distant metastases and signs of infiltration (Fig. 13.8). The tumors may be multifocal, involving both the liver and the spleen. Splenic rupture has been reported to occur in one- third of patients. Metastases
Splenic metastases are most commonly seen in patients with advanced primary tumors. The most frequent primaries are malignant melanoma (approximately one-third of melanoma patients), breast carcinoma, and bronchial carcinoma (see Table 13.5). •
CT Morphology
Metastases may appear as ill-defined areas of low attenuation on unenhanced scans. They are more clearly delineated after contrast adminis-
Fig. 13.8 Malignant hemangioendothelioma of the spleen with nodular hypervascular components and necrotic areas. Note the different enhancement characteristics of the hepatic metastases.
tration. Cystic areas of necrosis may occur. Multifocal lesions are the rule, but solitary splenic metastases have been described even in the absence of other metastatic disease. Diffuse infiltration is reported in up to 10% of affected patients. Cystic lesions may occur with metastasis from melanoma, and cancer of the ovary, breast, and endometrium. Calcifications may occur in patients with mucinous adenocarcinoma as a primary tumor. Lymphoma
Lymphoma is the commonest primary malignancy of the spleen. Primary splenic lymphoma is rare (1-2% of all lymphomas) and is usually a non-Hodgkin Lymphoma (NHL) of the small cell type.
Inflammatory Diseases, Parainfectious Splenomegaly Secondary splenic involvement is frequent in both Hodgkin's disease (HD) and NHL. Approximately 25-33% of all patients with HD and NHL are affected. Reported numbers for sensitivity/specificity of CT to diagnose splenic involvement by lymphoma vary substantially, ranging from 30% and 7"!% to 90% and 90%, respectively. Diffuse lymphomatous involvement need not be associated with splenomegaly. Nodular lesions are seen in fewer than 20% of patients with splenic involvement. However, this is the only type of involvement that can be reliably detected by CT.Therefore CTis used primarily for lymph node staging. •
CT Morphology
Splenic involvement in lymphoma can have four forms: homogeneous enlargement, miliary nodules, multifocal lesions of A to 10cm, or a solitary mass. CT can detect focal or multifocal disease more reliably than diffuse infiltration. Commonly, the spleen appears homogeneous on unenhanced scans. Larger focal lesions may cause bulging of the splenic contour. During the parenchymal phase of enhancement, focal lesions are clearly demarcated as hypodense areas (Fig. 13.9). Occasionally their margins are ill-defined. Rim enhancement is rarely observed.
Fig. 13.9 Nodular involvement of the spleen by NHL. The CT morphology is indistinguishable from hypovascular metastases or nodular involvement by sarcoidosis. The slight enhancement of the lesions distinguishes them from abscesses.
Necrosis of large lesions has been reported and can cause an irregular cystic appearance that may mimic an abscess in patients with lymphoma and associated fever. Calcifications are rare before, and more frequent after therapy. In NHL, splenic involvement is associated with para-aortic lymph node involvement in approximately 70% of patients.
Diffuse Splenic Diseases, Splenomegaly (see Table 13.6) Table 13.6
Ca uses of splenomegaly
Portal congestion: Portal hypertension, cirrhosis Portal or splenic vein thrombosis Right heart failure Neoplasms: Leukemia (AML > CML > ALUCLL) Lymphoma Metastases, primary neoplasms Infection: Echinococcus, malaria, leishmaniasis, hepatitis, mononucleosis, brucellosis, tuberculosis, typhus, syphilis Others: Sarcoidosis Storage diseases Hemodialysis Hemolytic anemias Extramedullary hematopoiesis Collagen diseases
Inflammatory Diseases, Parainfectious Splenomegaly
Reactive splenomegaly can develop as a parainfectious process in response to mononucleosis, typhus, bacterial endocarditis, or other infectious diseases. •
CT Morphology
Parainfectious splenomegaly generally has a homogeneous appearance. Abscesses are not present. Intrasplenic hemorrhage may occur as a rare complication. Infiltration of the spleen by Pneumocystis cetinii (mostly in AIDS patients) may show a fine speckled pattern of hypo attenuating lesions after contrast administration (differential diagnosis: fungal abscesses) or may present as splenomegaly with or without multiple punctate calcifications.
13 The Spleen
Calcifications also occur after other granulomatous inflammations such as tuberculosis or histoplasmosis (Table 13.7). Other organs, such as the liver, kidneys, or adrenals, may also be affected.
phomatous involvement, even in a patient with confirmed lymphoma. Diffuse infiltration may not alter the enhancement characteristics of the spleen, but a few cases show a pattern of fine, nodular hypo attenuating lesions.
Lymphoma
Portal Hypertension
Diffuse lymphomatous infiltration of the spleen is more common in NHL than Hodgkin's disease. It is mostly due to low grade lymphomas with associated blood involvement.
Portal hypertension leads to splenomegaly and stimulates the development of portosystemic collaterals.
•
In most cases the enlarged spleen presents a homogeneous pattern on unenhanced scans and during the parenchymal phase after contrast administration. The splenic vein is often dilated (> 1Omm) , and collateral vessels are found in the hilum of the spleen. Portosystemic collaterals involving fundic, gastroepiploic. and gastroesophageal veins are recognized by their increased tortuosity and their sometimes grotesquely enlarged diameters (differential diagnosis: poorly opacified bowel loops) . Enlargement of the left renal vein should prompt a search for a spontaneous splenorenal shunt (see Fig. 11.51).Ascites maybe present. Frequent causes are hepatic cirrhosis (e.g., small liver with a nodular surface, hypertrophy of the left and caudate lobes) or a thrombosis in the splenic or portal vein.
CT Morphology
CT is able to detect splenomegaly (splenic index >480) but is often not useful for the direct detection of diffuse splenic involvement. Diffuse infiltration may or may not lead to splenomegaly. Conversely, the presence of splenomegaly is not necessarily caused by lymTable 13.7
Increased CT density, splenic calcifications
Diffuse increase in CT density Hemochromatosis Sickel cell anemia Thalassemia Disseminated foci of increased density Granulomas (tuberculosis, histoplasmosis, brucellosis) Pneumocystitis carinii Phleboliths Gamma-Candy bodies Thorotrast Prior lymphography Isolated calcificatiohs Granulomas Solid tumors (mostly benign) Capsular or parenchyma! calcifications Abscess (pyogenic, tuberculous) Infarction Hematoma Splenic artery calcification, splenic artery aneurysm Calcified cyst wall Congenital cyst Post-traumatic cyst Echinococcalcyst Epidermoid cyst
•
CT Morphology
Sarcoidosis
Between 20% and 60% of patients with sarcoidosis develop splenomegaly. CT cannot detect diffuse involvement of the spleen and is useful only for detecting the less common focal involvement. •
CT Morphology
Noncontrast CT shows a uniformly enlarged spleen. On postcontrast scans heterogeneous enhancement may be noted. The spleen may contain sharply circumscribed hypo attenuating lesions corresponding to aggregated granulomas. Coexistent abdominal lymphadenopathy is frequently seen in patients with splenomegaly. The chest radiograph may be normal in up to 25%. The presentation may be morphologically indistinguishable from focal lymphomatous involvement (Fig. 13.9).
Trauma
The spleen is the organ that is most frequently injured by blunt abdominal trauma. The extent of the injury can range from subcapsular hemorrhage and parenchymal contusion, to a parenchymal laceration (with or without capsular injury), or a complete splenic rupture. The detection of a subcapsular hemorrhage is important because there is the potential of a lifethreatening hemorrhage in the event of a capsular rupture. Injury to the vascular pedicle leads to splenic infarcts of variable extent. CT has greater than 95% sensitivity in the detection of splenic injuries and is indicated whenever ultrasound findings are inconclusive. •
CT Morphology
A hematoma may appear hyperattenuating (after fresh hemorrhage, often with clot formation), isoattenuating (older hemorrhage or hyperacute extravasation without separation of red cells and serum), or even hypo attenuating to the splenic parenchyma, depending on the age ofthe collection. Contrast -enhanced scanning is required in addition to precontrast scans for adequate evaluation of splenic trauma, as areas of hematoma or laceration may be isoattenuating to the splenic parenchyma on noncontrastenhanced CT and the splenic injury is more clearly demarcated after contrast administration. On the other hand, fresh hematomas may
a
b
Fig. 13.10 Splenic trauma, (a) Traumatic splenic laceration with a concomitant rupture of the diaphragm. The splenic capsule is disrupted, but the intrasplenic hemorrhage is poorly visualized, (b) Splenic rup-
appear isoattenuating to the splenic parenchyma on enhanced scans. A subcapsular hemorrhages appears as a peripheral crescent-shaped collection that indents and displaces the splenic parenchyma. It may reach a considerable size. An intra parenchymal hematoma following splenic contusion appears as a patchy, ill-defined area of low attenuation with hyperattenuating components (Fig. IS.10a). A laceration or rupture produces a linear, wedgeshaped, or rounded parenchymal defect of low density that shows absent or reduced enhancement (Fig. 13.10b) and covers part or all of the organ cross section. Hemoperitoneum usually accompanies this type of injury. Contrast extravasation is a sign of active hemorrhage that nearly always needs immediate surgical treatment. Traumatic splenic false aneurysms show a CT attenuation similar to blood vessels after contrast administration (Fig. 13.10c) and predispose for a delayed splenic rupture. In rare cases, initial CT scans are normal despite the presence of a splenic injury. Splenic hemorrhage may be absent initially due to hypotension from severe blood loss, but it may commence later when volume replacement is instituted. A delayed splenic rupture is marked by intraperitoneal blood collections that are often denser (fresher) than perisplenic or intrasplenic blood collections.
c ture with a band-like laceration of the parenchyma. (c) Post-traumatic pseudoaneurysm of the spleen with a fine parenchymal Jaceration. Delayed splenic rupture occurred one week later.
13 The Spleen count the integrity of the capsule, the size of the hematoma, the number and length of lacerations, vascular involvement, and the extent of parenchymal devascularization, The later schemes have improved upon older ones, and significant correlation with patient outcome is noted, However, these correlations are not strict for grade III lesions and conflicting results from nonsurgical management have kept the discussion alive, The results of nonsurgical management may be improved by more liberal use of endovascular therapies like embolization, CT can play an important role in patient selection, Lack of splenic parenchymal enhancement without infarction may occur in the setting ofprofound hypotension, It has to be differentiated from arterial disruption after abdominal trauma, Splenic injuries may take several months to fully resolve on follow-up CT examinations, Peritoneal blood and perisplenic hematoma typically resolve in 1 to 3 weeks, Infection may complicate hematoma resolution and produce a splenic abscess, Pseudocysts may develop as a late sequel to parenchymal hematomas,
Even if a parenchyma! defect cannot be directly visualized, the detection of perisplenic blood clots (so called sentinel clots, patchy hyperattenuating structures with CT numbers >60HU as compared to 35-45 HU of a hemoperitoneum) provides an indirect sign of parenchyma! injury, Faint focal hyperenhancing regions in an otherwise homogenous splenic parenchyma may be an indicator of laceration and are associated with an increased risk of delayed rupture, Strongly hyperenhancing regions suggest a traumatic false aneurysm, Splenic laceration may be mimicked by beam-hardening or streak artifacts due to ribs, an air-contrast interface in the stomach, nasogastric tubes, or monitor material. The streaks usually extent beyond the splenic parenchyma, Fissures in a lobulated spleen should not be mistaken for a splenic laceration or rupture, An over-distended stomach compressing the medial border of the spleen may obscure a small laceration, Several CT grading systems (e,g" by Mirvis, 1989 or Federle, 1998) have been proposed to predict which patients can be successfully managed conservatively and which need immediate surgical intervention (Table l3.8), They take into ac-
Table 13.8
AAST
organ
injury scale for splenic
injuries-modified
Hematoma
Subcapsular,
nonexpanding,
Laceration
Capsulartear,
nonbleeding,
Hematoma
Subcapsular,
non expanding,
Intra parenchymal, Laceration
1111
Hematoma
I Laceration
IV
V
Hematoma
«zcm
active
<10%
surface
(Federle et
al..
1998)
area
<1 em parenchymal
depth
10-50% surface area diameter, nonexpanding
bleeding,
1-3 em parenchymal
depth
Subcapsular, >50% surface area or expanding Ruptured Subcapsular hematoma with active bleeding Intra parenchymal, >2 em diameter or expanding >3 em parenchymal Ruptured Involving
segmental
Laceration
Shattered
spleen
Hilarvascular
one grade for multiple
depth
intraparenchymal
Laceration
Vascular Advance
Capsulartear,
CT version
injuries
hematoma
or hilar vessels
with
injury that devascularizes to same organ
active
producing
up to grade
spleen III.
bleeding
devascularization
>25%
Splenic Artery Aneurysm
Vascular Disorders Splenic Infarct
Splenic infarcts may be caused by thromboemboli ejected from the heart or an aortic aneurysm. Thrombotic occlusions of splenic arteries can result from atherosclerosis, subendothelial
infiltration
in myeloid
leukemia,
in-
flammatory or neoplastic pancreatic diseases, vascular encasement by gastric carcinoma, hemagglutination in sickle cell anemia, or traumatic intimal •
injury.
CT Morphology
Unenhanced CT demonstrates an acute splenic infarct as a subcapsular area of low attenuation, that classically is well demarcated and wedgeshaped with the base at the splenic capsule. Not infrequently, however, infarcts may appear as more rounded or linear poorly marginated lesions that are hard to distinguish from other splenic lesions. Extension to the capsule suggests the correct diagnosis (Fig. 13.11a). Splenic infarcts show little or no enhancement after contrast administration. Occasionally an enhancing strip of parenchyma is seen just beneath the capsule. This rim sign is caused by opacification via capsular vessels (Fig.l3.11 b). Secondary hemorrhage may occur, causing the lesion to appear hyperattenuating. Over time, the lesion becomes isoattenuating to the surrounding parenchyma on noncontrast scans. A pseudocyst may develop from the infarcted area, or the infarct may undergo secondary calcification.
CT in patients with homozygous sickle cell anemia and chronic recurrent splenic infarcts shows a shrunken, heavily calcified spleen. Patients with the heterozygous form have an enlarged spleen showing multiple subcapsular infarcts with calcifications. Splenic Vein Thrombosis
Thrombosis of the splenic vein may occur in the setting of an inflammatory or neoplastic pancreatic disease, peritonitis, or after trauma. •
CT Morphology
With an acute thrombosis, the splenic vein usually appears dilated and hyperattenuating. Later the vein shrinks and acquires soft-tissue density. It does not enhance after contrast administration. Collaterals develop via short gastric, gastroepiploic, gastroesophageal, and splenorenal venous channels. Splenomegaly is invariably present. Splenic Artery Aneurysm
Aneurysms of the splenic artery are the commonest abdominal artery aneurysms. An incidence of 0.01-2% is reported in autopsy series, rising up to 10% if aneurysms between 0.5 and 1 cm are included. Predisposing conditions include multiparity, portal hypertension, and arteriosclerotic disease. Congenital splenic artery
Fig. 13.11 Splenic infarct. (a) MUltiple linear and wedge-shaped infarctions in a patient with a hematopoietic disorder and splenomegaly (4 x 2.5/15, VRT). (b) Splenic infarct appears as a hypodense wedge-shaped region bounded by an enhancing subcapsular strip of parenchyma. This rim sign is only seen occasionally and can be difficult to distinguish from an abscess membrane.
a
b
511
512
73
The Spleen
Fig 13.12 Patient with chronic renal failure, kidney transplantation, and a stenosis of both native renal arteries (arrowheads) who also has a large partially calcified aneurysm of the splenic artery (arrows). Volume-rendering of a multislice eTA data set (4 x 1/6) in an AP direction (a) and caudocranial direction (b).
a
b
aneurysms are frequently accompanied by aneurysms of the hepatic or celiac artery. A false aneurysm of the splenic artery can develop as the result of a shearing injury (seat-belt trauma) . Splenic artery aneurysms are a common incidental finding at CT. Embolization or resection is recommended if the aneurysm is symptomatic, larger than 2 em, increasing in size, seen in pregnant women or in women of childbearing age. Preoperative visualization can be accomplished with CTA.
•
CT Morphology
CT demonstrates a circumscribed mass in the course of the splenic artery, mostly in the distal third of the artery (Fig. 13.12). The mass enhances strongly during the arterial phase after contrast injection, and its diameter exceeds that of the adjacent vascular segments. It can range from a few millimeters to several centimeters in size. Larger aneurysms are associated with wall calcifications and peripheral thrombosis.
14 The Pancreas c. Schaefer-Prokop
Anatomy Examination
Technique
Patient Preparation (p. 517) Acquisition Technique (p. 517) Image Processing (p. 519) Enhancement Characteristics Cystic Pancreatic
Lesions
.
Dysontogenic Pancreatic Cysts (p. 520) Pancreatic Pseudocysts (p. 520) Cystadenoma, Cystadenocarcinoma (p. 522) Solid Pancreatic Lesions
.
Pancreatic Carcinoma (p. 524) Islet Cell Tumors (p. 527) Solid Papillary Epithelial Tumor (p. 529) Pleomorphic Carcinoma (p. 529) Lymphoma (p. 529) Metastases (p. 529) Pancreatitis Acute Pancreatitis (p. 530) Complications of Acute Pancreatitis Chronic Pancreatitis (p. 536) Diffuse Pancreatic
(p. 534)
Changes
Cystic Fibrosis (Mucoviscidosis) Hemochromatosis (p. 539) Lipomatosis (p. 539) Trauma and Postoperative
. (p. 538)
Changes
Pancreatic Trauma (p. 539) Pancreatic Resection (p. 540) Pancreatic Transplantation (p. 540)
14 The Pancreas The current mainstays in the radiological evaluation of the pancreas are ultrasonography, computed tomography and magnetic resonance imaging, Diagnostic endoscopic retrograde pancreatography (ERCP) is presently being substituted by MR cholangiopancreatography (MRCP), Endoscopic ultrasound is important for small
Table
14.1
Tumor
Indications
diagnosis
Oncologic
follow-up
for CT examination
tumors or ambiguous findings at CT or MR At present, CT offers the broadest spectrum of information in the diagnosis of pancreatic diseases (Table 14.l), Thin-section spiral and multislice CT have increased the sensitivity of CT for the detection of small tumors and have improved the accuracy of tumor staging,
of the pancreas
Pancreatic
carcinoma:
Endocrine
pancreatic
Therapeutic
response,
Trauma
Pancreatic
Pancreatitis
Differentiation
rupture,
tumor
detection,
tumors:
potential
for curative
resection
localization
suspected
recurrence
pancreatitis
of exudative
and necrotizing
forms;
Pretherapeutic CT to determine extent of inflammatory Detection and identification of pancreatic pseudocysts; Evaluate for suspected abscess
spread;
Anatomy The pancreas is located in the anterior retroperitoneal space (Fig, 14.1), which is bounded in front by the parietal peritoneum and behind by Cerota's fascia, The tail of the pancreas usually sweeps upward, and ends intra peritoneally within the splenorenalligament The normal pancreatic parenchyma has CT attenuation values in the range of 30-60 HU, Pancreatic attenuation decreases due to fatty infiltration (which occurs normally with aging) and is increased in hemochromatosis, The size and shape of the pancreas are highly variable and depend on age and fat content (normal upper values for the transverse diameter: 3cm in the head, 2,5cm in the body, 2cm in the tail), A more important criterion of normal pancreatic morphology is an even tapering of the pancreas from the head to the tall. The tail of the pancreas usually sweeps upward, but in 25% of cases it turns downward and can mimic a renal or adrenal mass, or a recurrent tumor following nephrectomy, Atrophy or fatty infiltration may give the pancreas a lobulated appearance, The uncinate process is embedded in the duodenal loop (Fig, 14.1, see also Fig, 12.1), While the mesenteric artery is usually surrounded by a thin layer of fat, the mesenteric
vein is in direct contact with the parenchyma of the uncinate process, The venous confluence lies just posterior to the pancreatic head, and the splenic vein runs along the posterior border ofthe body of the pancreas, The splenic artery is very tortuous, so on axial sections only segments of the vessel can be imaged posterosuperior to the pancreas, The main pancreatic duct (ofWirsung) extends along the central axis of the pancreas, its distal portion having a diameter of 3-5 mm near the papilla, The duct can be visualized in most patients up to the tail only with thin-section spiral or multislice CT (3 mm thickness or less), The main pancreatic duct should not be confused with the fine fatty layer that runs parallel to the splenic vein, A diameter greater than 4mm is considered pathologic. The accessory duct (of Santor in i) is detectable only with thinsection CT The distal portion of the common bile duct runs through the pancreatic head, where its diameter is less than 5 mm, and opens into the duodenum at the papilla of Vater. The common bile duct and the main pancreatic duct have either one orifice in common or two separate orifices lying next to each other (see Fig, 12.1),
Anatomy Fig. 14.1 Diagrams illustrating the sectional anatomy of the pancreas. Pancreatic duct -______. Common hepatic artery Common bi Ie -duct
.__ . /
Splenic a Splenicvein
Celiac (arterial) trunk
Gastroduoden.al ~ artery Superior --mesentericvein Superior mesenteric artery Pancreatic head
Inferior mesenteric
•
Anatomic Variants
Pancreas divisum (5-10%) is the commonest anatomic variant. It results from a failure of fusion of the anterior and posterior halves of the developing pancreas and is defined as a completely separate pancreatic ductal system in a grossly undivided gland. The main portion of the pancreas, including the anterosuperior part of the head, the body and the tail, is drained by the dorsal pancreatic duct through an accessory papilla. The posteroinferior part of the head and uncinate process is drained by a short, narrow ventral pancreatic duct that joins the common bile duct in the ampulla. Pancreas divisum is associated with an increased rate of pancreatitis. Although the overall size of the pancreas may be normal, CT demonstrates broadening of the pancreatic head in the craniocaudal and/or an-
-------" _
vein
teroposterior extent. In rare cases the ventral and dorsal moieties may be distinctly visible, separated by a thin fat plane. Thin-section spiral or multislice CT can detect the separate ductal systems. Annular pancreas is a developmental anomaly in which the main pancreatic duct encircles the duodenum and produces a constriction. An annular pancreas is visible with CT only if there is sufficient parenchyma! involvement in the anomaly. It should not be mistaken for duodenal wall thickening. An annulus composed entirely of connective tissue cannot be demonstrated with CT. An accessory pancreas is subject to many variations. Ectopic pancreatic tissue in the stomach or small bowel usually cannot be detected with CT.
515
516
74 The Pancreas
Examination Technique The examination technique varies strongly between evaluation for pancreatic tumors or pancreatitis (Tables 14.2, 14.3).
Table 14.2
Suggested protocols for CT scanning of the pancreas
General Preparation
Tumor diagnosis:
patient should fast for 6 hours
Oral contrast
Pancreatitis:
depent on clinical situation either none or 500 - 1000 ml contrast medium or water, 30 min prior to examination 1000-1500 ml water, 30 min prior to examination additional 250 ml of or water in RLD posltlon on CT table N-butyscopolamine iv (to relax the duodenum)
Tumor diagnosis:
Patient position
Supine with elevated arms Tumor diagnosis: Right lateral decubitus (optional)
Scan range
Tumor:
pancreas only (detection, T-staging), pancreas + liver (M-staging) Upper abdomen to small pelvis
Pancreatitis: Respiratory
phase
Windowing
Inspiration Noncontrast CT: Contrast-enhanced
Scan parameters
1-slice SCfTF/RI
Pancreatic tumors
2/4/2
L
Pancreas + liver
5/8/4
L
Upper/lower
7/12/6
abdomen
Contrast injection
CT: 4-slice' SC
16-slice' SC
axial SW/RI
MPRb SW/RI 3/3 cor 4/4 sag
0.5-1.25
L
3/2
2-31
0.5-1.25
L
5/4
2-31
1-1.5
1-1.25
L
W/L~350/40 W/L~400/60
L
L
8/6
1-slice V/F/D
4-slice V+N/F/D'
16-slice V+N/F/DX'
Tumors/chronic pancreatitis: Arterial phase Parenchymal phase Portal phase
150/4/20 150/4/40 150/4/70
150 + 50/5/5A 150 + 50/5/20A 150 + 50/5/60
150 + 50/5/5A 150 + 50/5/25A 150 + 50/5/205
Acute pancreatitis
100/2/60
100 + 50/3/50
80 + 50/3/30A
SC~ slice collimation (mm), TF ~ table feed (rnmjrotatlon), RI ~ recon. increment (mm), i L ~ scanning direction, SW ~ effective section width (mm), MPR ~ multiplanar reformats, axial ~ axial sections, cor= coronal, sag ~ sagittal CM ~ contrast material, Contrast concentration ~ 300 mg/ml iodine V ~ volume of CM (ml), N ~ volume of saline flush (ml), F~ flow rate (ml/s), 0 ~ start delay, X ~ trigger region " Pitch P~ TF/(Nx5C) should be chosen close to l.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI ~ 1-1.5/0.7 or 0.5-0.8/0.5; oblique coronal MPR through the pancreas, sagittal MPR between the common bile duct and the inferior meso vein; CPR parallel to the pancreatic duct c read: 5-30A (205) ~ start delay 5-30 s (20 s) after reaching 50 HU enhancement in the desc. aorta (spleen)
Acquisition Table
14.3
Suggested
examination
"Indication Pancreatic
technique
Scan range
Phase carcinoma
Islet cell tumors
Islet cell tumors
Parenchyma! Portal phase
(benign)
(malignant)
Pancreatitis
(acute)
Pancreatitis
(chronic)
a not mandatory, b 60 s delay suffices
Arterial Portal
Pancreas Liver + pancreas
phase
Pancreas Liver + pancreas"
phase
phase"
Liver + pancreas Liver + pancreas
Arterial phase Portal phase Parenchymal Noncontrast
Trauma
Technique
Upper abdomen extend to lesser
phase scan"
Parenchyma!
phase
Pancreas Pancreas
Parenchymal
phase"
Upper abdomen
for blunt abdominal
to iliac crest, pelvis if necessary
+ Pelvis
trauma
Patient Preparation The duodenum is difficult to opacity because of its brisk peristalsis, but adequate duodenal distension is essential for tumor diagnosis, A right lateral decubitus position improves filling of the duodenum, so the patient should ingest the last cup of oral contrast while lying in this position on the CT table, A spasmolytic drug (e,g" 20-40 mg N-butylscopolamine IV) should be administered to dilate the duodenum, Afterward the patient is moved to the supine position, and scanning is initiated, If N-butylscopolamine is contraindicated, the examination may be performed while the patient remains in the right lateral decubitus position, There is a sound rationale for using negative contrast media (water, milk, or gas) in the duodenum instead of positive contrast media, as this leaves open the option for CT angiography of the pancreatic vessels, does not mask bile duct stones and may be advantageous for evaluation of gastric or duodenal wall infiltration, lfthe clinical situation allows, oral contrast medium should be used for bowel opacification even in patients with acute pancreatitis, In doubtful cases (patients on n.p.o.) the referring physician has to be consulted, Positive oral contrast material aids in differentiating bowel from abscesses, peripancreatic fluid collections, and pseudocysts, Some institutions, however, prefer water as oral contrast agent as it facilitates the evaluation of the pancreatic head, the periampullary region, the distal common bile duct, and
the pancreatic duct Such a procedure is mainly recommended when CT is performed to rule out biliary pancreatitis but it is inferior for identification of complications (e,g" differentiation between water-filled bowel loops and pseudocysts),
Acquisition Precontrast
Technique Scan
Generally, there is no diagnostic need for precontrast scans of the upper abdomen in patients with suspected pancreatic tumors unless there is coexistent chronic pancreatitis, U nenhanced scans may be used, however, to tailor the subsequent contrast-enhanced scan optimally to the tumor size, U nenhanced CT may be used to verify bile duct or pancreatic duct stones as etiologic agents for acute pancreatitis and to detect calcifications in the chronic or chronic relapsing forms of the disease, However, non-calcified stones may not be visualized with this technique, and calcified stones as well as parenchymal calcification usually can also be detected on contrast-enhanced CT Tumor
Diagnosis
Staging of malignant tumors and detection of islet cell tumors requires thin sections to ensure optimum pancreatic imaging, Because an optimum attenuation difference between tumor and parenchyma can be achieved for only a brief
14 The Pancreas
time after contrast injection, trade-offs must be made between scan length and section thickness in single-slice CT For malignant tumors, the liver has to be examined as well because of the high frequency of hepatic metastases, Patients with potentially resectable pancreatic carcinomas are challenging, Most authors suggest biphasic scanning, The first scan is focused only on the pancreas and uses a thin collimation of 2-3 mm with single-slice spiral CT 1-1.25 mm with 4-slice scanners and 0,5-1.25 mm with 16-slice scanners depending on patient size, It is performed during the parenchyma! phase of pancreatic enhancement with a scan delay of 30-40 seconds, depending on the contrast injection rate (A-4mlls is advised for multiphasic studies), In this phase the contrast between parenchyma and lesions is highest, and there is still sufficient enhancement of the arteries to evaluate vascular tumor involvement Often portal enhancement is also already beginning, To increase spatial resolution, images may be reconstructed using a small fieldof-view that is focused on the pancreas, or with a 10242 matrix and a full field-of-view (ifavailable) , The second scan covers the liver and pancreas during the portal venous phase (60-100 second scan delay) and may use a greater section thickness (5mm with single-slice CT 2-2,5 mm with 4-slice CT 1-1.5 mm with 16-slice C1), In patients with large pancreatic carcinomas that are known to be unresectable, a portal phase scan of the upper abdomen suffices, CT angiograpby of the peripancreatic vasculature has become an integral part of the preoperative staging procedure and can be derived from data acquired during biphasic scanning of the pancreas, A scan in the pancreatic parenchyma! phase is used to create CTA images of the peripancreatic arteries but arterial contrast is improved if the scan delay is reduced, For practical purposes a scan delay of about 30s (or better, 10-15 s after bolus triggering in the aorta) usually provides the best compromise between parenchyma! and arterial enhancement CTA in the portal venous phase is best if the scan delay is around 60s, Thin-section scanning with multislice CT has intrinsic advantages for CTA Imaging of the pancreatic duct has been termed CT pancreatography It relies on a thinsection spiral or multislice CT data set acquired in the parenchyma! phase, Curved reformations (CPR), volume rendering (VRT) or minimum in-
tensity projections (MinlP) may be used to visualize the duct (see Figs, 14,6 and 14,7), Virtualendoscopic views (virtual pancreatoscopy) have been employed to demonstrate ductal irregularities or intradu ctal masses bu trarelyyield su perior information over other display techniques, Endocrine pancreatic tumors are generally hypervascular and often can be visualized only in the arterial phase of enhancement If a benign lesion is to be localized, scanning can focus on the pancreas and should employ thin sections (2-3 mm with single-slice spiral CT 1-1.25 mm with 4-slice scanners and 0.5-0,75 mm with 16slice scanners) in order not to miss a small lesion, Scanning may start shortly after the contrast arrives in the aorta (3-4 s after bolus triggering in the aorta) and should be completed shortly after peak aortic enhancement This requires short data acquisition times of20s or less for optimum contrast between tumor and parenchyma, If a malignancy is suspected, the liver has to be covered in the arterial phase as well, and the section thickness may have to be increased to 3-5 mm (Single-slice spiral C1) or 2-2.5 mm (4-slice scanners), depending on the performance of the scanner. Most authors suggest biphasic scanning with a portal phase scan similar to that for pancreatic carcinoma, In most cases, however, this scan only contributes to the staging of larger tumors with venous involvement If arterial phase scanning of the liver is not possible, a noncontrast examination of the liver is recommended, Pancreatitis
For pancreatitis, most institutions perform only a contrast-enhanced scan in the parenchyma! or portal phase of enhancement A contrast dose of 70-100 ml is sufficient to differentiate between exudative and necrotizing pancreatitis (scan delay 40-60 s). Thicker sections (7-8 mm) can be used for evaluation of the entire abdomen to detect spread of exudates or necrosis, Because it is a benign disease and pathology is often gross, dose reduction may be considered, Patients with chronic pancreatitis in whom a tumor shall be ruled out have to be scanned according to the tumor protocol described above, B
Pancreatic Trauma
In patients who have sustained blunt abdominal trauma, noncontrast CT scanning should be performed on the entire upper abdomen for the de-
Enhancement Characteristics
tection cluded to that nation
of hemorrhage. Parenchymal injury is exwitha contrast-enhanced protocol similar used in pancreatitis. and a survey examiof the pelvis should also be performed.
Image Processing Image processing is rarely required for pancreatitis or pancreatic trauma. For the evaluation of tumor resectability. however. various types of image processing can be recommended. As a standard. axial images should be complemented by oblique coronal reformations parallel to the pancreatic head or tail. depending on the tumor location. An MPR thickness of 3-4 mm reduces image noise. It is usually sufficient to include the para-aortic nodes posteriorly and the superior mesenteric vessels anteriorly in the reconstructed range of MPR. Such coronal sections display the cranial and caudal extent ofa tumor and are optimum for assessing fatty tissue invasion. paravascular spread. and the relation of the abnormality to the common bileduct(cf. Fig. 14.19). Sagittal MPR covering a range that includes the common bile duct and the inferior mesenteric vein are well suited to evaluate proximal vascular spread around the superior mesenteric artery (SMA) and celiac artery as well as the commonbileduct(CBD) .
Enhancement
Characteristics
The blood supply to the pancreas is entirely arterial. Consequently. the attenuation of the pancreatic parenchyma rises more rapidly after contrast administration than in the liver. and the enhancement persists for a shorter time due to the absence of a portal phase. Pancreatic enhancement begins shortly after the start of the aortic enhancement. reaches a peak approximately 5 s after the end of the aortic plateau. and rapidly declines to a level that depends on the remaining intravascular contrast concentration. •
Coronal curved planar reformations (CPR) through the pancreatic duct are excellent for displaying the level of ductal involvement (see Fig. 14.7c). CPR through the superior mesenteric artery. hepatic artery or the superior mesenteric vein and the portal or splenic vein are well suited to demonstrate the relation of the tumor to these critical structures. Thin-slab maximum intensity projections (MIP) are excellent for displaying vessel involvement (irregularities. stenoses). They should be as narrow as possible but still include the whole vessel of interest. The plane of the slab should be tilted appropriately. Thin-slab minimum intensity projections (mIP) may be used to display the pancreatic duct. Slabs should be chosen parallel to the pancreatic body and tail. and to course of the duct in the head and uncinate process. Best results are gained with a relatively narrow slab thickness (some 5 mm). Volume rendering with appropriate cut planes allows for excellent 3D display and evaluation of venous as well as arterial involvement. Opacity settings should be chosen to make the fat transparent but keep the enhancing pancreas opaque. By cutting into the volume. the relation of tumor relative to the arteries and veins can be directly displayed (see Fig. 14.8).
Anerial Phase
The arterial phase starts as contrast material arrives in the abdominal arteries (about 15-25 seconds after the start of the injection). During
the course of the arterial phase there is a steady increase in enhancement of the pancreatic parenchyma with a gradual transition to the parenchyma! phase. In general. the arterial phase lasts from about 20 to 40 seconds after the start of the injection of 150 ml of contrast at 4mlls. It is the only phase in which many hypervascular neuroendocrine tumors can be detected. because at later phases the tumors very quickly become isoattenuating to the pancreatic parenchyma and are no longer detectable. The intense opacification of the peripancreatic arteries during this phase is useful for detecting arterial invasion by tumor. but there is still insufficient opacification to detect the invasion of portal venous vessels.
519
14 The Pancreas
•
Parenchymal Phase
artifacts in the mesenteric and portal venous trunks that can mimic thrombosis or tumor invasion (pseudothrombus artifact).
The parenchyma! phase is characterized by strong enhancement of the pancreas • Portal Phase parenchyma. It overlaps with the arterial phase and lasts from about 30-70 seconds after the The portal phase of hepatic enhancement (peak start of the injection (peak at about 40s), de- at about 60-90 s) overlaps with the parenchypending on contrast volume and flow rates. The mal phase of pancreatic enhancement but permore contrast material is injected, the longer sists for a longer time. this phase persists. At constant volume, higher At 50-70 seconds after the start of the inflow rates increase the level but reduce the du- jection opacification of the portal and superior ration of pancreatic enhancement. mesenteric veins is optimal for detection of Pancreatic carcinomas, necrotic areas, pseu- tumor infiltration and thrombosis. Hypovascudocysts, and abscesses are most clearly demarlar hepatic metastases are well delineated. For cated during this phase, which also gives op- this reason, a scan is performed in tumor tim um delineation of the pancreatic duct and patients during the portal phase after arterial or common bile duct. Arterial opacification is still parenchymal phase scanning has been sufficient to define the vessels, but irregularities completed. in vascular contours may not be depicted as When the contrast material bolus is inclearly as during the arterial phase. The splenic creased to 200ml at 4mlls and scanning is carvein opacities approximately 40 seconds after ried out with a 60s delay, a hybrid phase often the start of the injection, and the mesenteric provides high arterial and portal venous opacifivein opacities about 5-10 seconds later. This cation and a good pancreatic and sufficient hedisparity in opacification times produces flow patic parenchymal enhancement.
Cystic Pancreatic Lesions (see Table 14.4) Dysontogenic Pancreatic Cysts
"True" epithelium-lined pancreatic cysts are found in 10% ofpostmortem examinations. They are clinically silent and are commonly (in 25-50%) associated with cysts in other organs (liver, kidneys, spleen, thyroid, ovaries and testis). Berry aneurysms of the cerebral arteries are present in approximately 10% of patients. Polycystic disease of the kidneys and HippelLindau disease is frequently associated with multiple pancreatic cysts. •
CT Morphology
CT demonstrates sharply circumscribed round or oval masses of low attenuation that do not enhance after contrast administration. In most cases the cyst wall is not detectable or appears only as a thin, faint line (Fig. 14.2). Pancreas pseudocysts generally have thicker walls.
Fig. 14.2 True pancreatic cyst. confirmed at operation. CT demonstrates a faintly visible cyst wall.
Pancreatic Pseudocysts
Pancreatic pseudocysts are loculated collections of necrotic material, old blood, and enzymatic fluid that develop as a sequel to acute or chronic pancreatitis. Pseudocysts form when areas of exudation and necrosis become walled off by
Pancreatic Pseudocysts Table 14.4
Differential diagnosis of cystic lesions of the pancreas
Dysontogenic cyst
Sharply circumscribed with a faint or invisible wall; septations does not enhance after contrast administration
Pancreatic
Thick granulating wall; rarely contains septations; history or direct evidence of pancreatitis
pseudocyst
frequently extrahepatic;
Abscess
Gas collection
Cystic fibrosis
Small cysts may occur; no pancreatitis; fatty involution of pancreas
Lymphangioma
Malformation with cluster of small cysts; calcifications tain lymphatic fluid
Microcystic adenoma
Multiple (>6) cysts 1 -20 mm in diameter with hypervascular septa and central calcifications (40%); central scar; unilocular variant very rare
Macrocystic adenoma
Unilocular or multilocular large cystic structure with irregularly hypervascular septa; calcifications «15%); no central scar
Cystadenocarcinoma
Same as macrocystic adenoma (often indistinguishable); foci of infiltration; hypoattenuating soft-tissue components; lymph node enlargement; metastases
IPMT -main
Diffuse/segmental dilation of pancreatic duct; intraductal papillary soft tissue excrescences; bulging papilla; vague intraductal calcifications
IPMT -branch
duct type duct type
IPMT = intraductal
papillary
(rare), granulating
are very rare;
rim history consistent with cystic fibrosis may occur; cysts con-
thickened,
Multilocular cystic lesion, usually in uncinate process, segmental pancreatic duct dilatation; communication of cystic lesion with main duct; bulging papilla mucinous tumor
non-epithelialized granulation tissue. There is a continuous transition from a peripancreatic fluid collection to a pseudocyst. A pseudocyst that persists longer than six weeks is very unlikely to resolve spontaneously. Interventional treatment (percutaneous or transgastric drainage) or surgery (cystogastrostomy, cystoduodenostomy, or cystojejunostomy) is then required depending on cyst size and complications (hemorrhage, superinfection). CT is used for treatment planning and for follow-up (see below p. 534).
or infection. CT morphology alone, however, is unreliable for the diagnosis of superinfection. The presence of gas bub bles within a pseudocyst may be caused by gas-forming bacteria but in rare cases can also be caused by a fistula between the pseudocyst and the gastrointestinal tract without concomitant infection.
IS CT Morphology
Pancreatic pseudocysts are highly variable in their size and location. They may be completely intrapancreatic, but it is more common to find extrapancreatic lesions located in the omental bursa or along Cerota's fascia. Rarely, pseudocysts may occur in the mediastinum or retroperitoneum or may erode into the liver or spleen. The cyst wall is often several millimeters thick and is composed of surrounding tissue and reactive granulation tissue (Fig. 14.3). Calcifications may be visible in the cyst wall. Septations are rare. CT numbers of the fluid of more than 25 HU are indicative of coexisting fat necrosis, hemorrhage,
Fig. 14.3 Pancreatic pseudocyst occurring at a typical site in the omental bursa and displacing the stomach. The thick granulating rim is a common finding. Note the substantially reduced attenuation of the liver due to fatty infiltration in this alcoholic patient.
521
74 The Pancreas
Cystadenoma,
Cystadenocarcinoma
•
CT Morphology
Benign micro cystic (serous) adenoma in more than 80% of cases occurs in patients > 60 years of age, with a female predilection of3 :2-9:2. It is distinguished from potentially malignant macrocystic (mucinous) adenoma and from cystadenocarcinoma, which mainly affects women 40 to 60 years of age. Cystadenomas account for 10-15% of all cystic pancreatic lesions, while cystadenocarcinomas comprise only a few percent of all pancreatic malignancies. Approximately 60-80% of subtypes of cystic neoplasms can be correctly categorized by CT. Recently, the intraductal papillary mucinous tumor (IPMT) was recognized as a separate new subtype of cystic pancreatic neoplasms. In the past, this rare tumor has also been named mucinous ductal ectasia or ductectatic mucinous cystic tumor. It originates from proliferating mucin-secreting cells in the ductal wall that form papillary folds. There is an abundant secretion of viscous mucin leading to obstructive dilatation of the ducts. Unlike other mucinous tumors, the tumor is more frequent in men and presents in the seventh decade. The tumor can be either benign or (low grade) malignant. Its location affects the macroscopic features and several subtypes are recognized: the main duct type, branch duct type, and combined type. Correct identification is important as this tumor has a more favorable prognosis than adenocarcinoma or mucinous cystadenocarcinoma and can be treated by curative segmental or total surgical excision.
The benign microcystic serous adenoma is characterized by mul ti pIe (> 6) cysts that are 1-20 mm in diameter with hypervascular septa (Fig. 14.4). The cysts form a lesion that may exceed 5 cm in size, has a multilobulated contour and a honeycomb-like configuration. The tumor may appear solid on noncontrast scans (especially when consisting of only small cysts < 2 mm), but enhancing septations are usually visible after contrast administration. Scans typically show stellate or "sunburst" calcifications (15-40%) within a central fibrotic scar. This scar may show prolonged enhancement on delayed images. Central calcifications are seen more frequently in serous cystadenoma than in any other pancreatic tumor. The tumor shows no signs of local invasiveness. Very rarely, serous cystadenoma may present as a macrocystic variant. The macrocystic mucinous adenoma consists ofa uni- or multilocular cystic mass (Fig. 14.5) that may reach considerable size (always >2cm, many over 10cm). Sometimes multiple small daughter cysts are found on the inner surface of a large cyst. The contents of the cyst may show water or soft-tissue density. The septa are hyperattenuating on postcontrast scans, but generally they are thicker and more irregular than in the microcystic form (important for differential diagnosis) Macrocystic adenomas most commonly involve the body or tail of the pancreas and peripheral calcifications are specific for this entity.
Fig. 14.4 Macrocystic mucinous adenoma (4x 1/6) presenting as a cystic mass in the pancreatic head with a polypoid intracystic lesion (a, arrow) and a dilatation of
the pancreatic duct (b). No intraductal found at histology.
a lesions
were
Cystadenoma, Cystadenocarcinoma
a
b
c
Fig. 14.5 (a) Macrocystic adenoma appears as a sharply circumscribed, unilocular cystic lesion with an enhancing rim. (b) Small cystadenocarcinoma in the pancreatic tail with peripheral enhancement, (c) Cyst-
adenocarcinoma of the pancreatic tail (arrowheads) with infiltration of the peri pancreatic tissue posteriorto the SMA (arrow). Note the massive infiltration of the spleen.
Macrocystic adenomas may undergo malignant change [cystadenocarcinoma') and are therefore classified as premalignant lesions. The malignant process may involve only a small area of the cystic mass. The thickness of the septa does not correlate with the propensity for malignant transformation. The septa exhibit cacifications in less than IS70 of cases, but there is no central fibrotic scar. A solid tumor component, if present, typically shows less contrast enhancement than healthy parenchyma (Fig.14.4c). While the tumor has a smooth external surface similarly to the microcystic form, the presence of solid papillary excrescences and thick septations enhancing after contrast material application practically exclude the diagnosis of a benign microcystic adenoma. Mucinous cystadenoma cannot be confidently distinguished from cystadenocarcinoma on CT scans (Fig.14.4b). They are also not readily distinguished by cytological evaluation of fine needle aspiration biopsy sampling, since undersampling of portions of the tumor containing carcinoma is likely. Thus surgical resection is indicated. For intraductal papillary mucinous tumors (lPMT) thin section helical or multislice CT is essential. The main duct type will show segmental or diffuse dilatation of the pancreatic duct with or without distal parenchymal atrophy. Intraductal papillae show solid attenuation in the duct wall and sometimes intraductal amorphous calcifications may be seen. As a result of
obstructing mucus, a bulging papilla may be identified (25% of cases). The branch duct type is usually located in the uncinate process, and much less freq uentl y in the body or tail. At CT it shows a lobulated mass of clustered small cysts with central septations (Fig. 14.6). Less frequently a unilocular cyst is demonstrated. Associated dilatation of the main pancreatic duct is segmental. There also may be a bulging papilla. Solid papillary excrescences of the tumor are often difficult to visualize. In the combined type features of both main and branch duct types are present. Differentiating malignant versus benign tumor is difficult (sensitivity 65-70%), but the risk of malignancy is increased when the following features are present: solid mass, dilatation of the pancreatic duct > 1Omm, bulging papilla, diffuse or multifocal involvement, or attenuating intral uminal content. Differential diagnosis includes macrocystic mucinous adenoma, microcystic serous adenoma, obstructive pancreatitis with pseudocyst, and necrotic adenocarcinoma. For correct diagnosis of this tumor, the communication of the cysts with the main pancreatic duct should be demonstrated. If thin-slice CT fails, MRIIMRCP or ERCP are the modalities of choice.
14 The Pancreas
Fig. 14.6 Intraductal papillary mudnous tumor (IPMT) (4 x 1/6). The tiny septa are hard to appreciate (a). A curved reformation (CPR) best demonstrates the location within the pancreatic head (b).
a
b
Solid Pancreatic lesions Pancreatic Carcinoma
Up to 90% of pancreatic carcinomas are adenocarcinomas, and most (80%) occur in the pancreatic head. Tumors in the body (15 %) or tail of the pancreas (5 %) do not cause typical symptoms and usually are not diagnosed until they have reached an inoperable stage. The only effective treatment of pancreatic carcinoma is surgical resection. Because of early lymphogenous and hematogenous spread and a propensity for invasive growth, only a small percentage of pancreatic cancers are resectable at the time of diagnosis (10-30%). Vascular invasion is a relative criterion for the lack of resectability, depending on the extent of vascular involvement and reconstructive options. Metastatic spread makes a tumor irresectable and is most often caused by liver metastases and peritoneal metastatic implants. Because peritoneal and omental metastases may measure only "1-2 mm in size and then can only be detected by direct visualization, some institutions perform a preoperative laparoscopy before the patient is subjected to laparotomy. The sensitivity of CT for the detection of small tumors and accurate tumor staging (Table 14.5) can be significantly improved by the use ofthin-section spiral or multislice CT techniques. •
CT Morphology
Pancreatic carcinomas are almost always isoattenuating to the parenchyma on unenhanced CT scans. The presence of a tumor is suggested by
Table 14.5
TNM Staging
of pancreatic
tumors
Tis
carcinoma
in situ
T1
tumor limited to pancreas,
< 2 cm
T2
tumor limited to pancreas,
>2 cm
T3
invasion of duodenum, or peri pancreatic fat
T4
infiltration of stomach, large vessels
N1a
singular
N1 b
multiple regional
regional
common spleen,
bile duct, colon, or
lymph node metastasis lymph node metastases*
* superior/i
nferior/ anterior (pa ncreaticod uadena I, proximal mesenteric/posterior (pancreaticoduadenal, common bile duct, proximal mesenteric)/splenic/cel iac
focal enlargement of the pancreas (usually the head or uncinate process) or by a circumscribed bulge in the gland contour. The uncinate process may loose its wedge-shaped configuration and become round. Neither of these changes is an early sign, however. In patients with pancreatic lipomatosis, tumors appear as homogeneous areas of increased parenchyma! density or lead to a focal obliteration of the lobular pattern. Fibrofatty septa within the normal parenchyma help to differentiate hypertrophic (benign) parenchyma! lobulation from a malignant tumor. On contrast-enhanced scans during the parenchymal phase adenocarcinoma demarcates as a hypoattenuating mass from the normal parenchyma. Small tumors may become
Pancreatic Carcinoma
isoattenuating just 30 seconds later, so only a brief time window is available for optimum tumor detection. Dilatation of the pancreatic duct is suggestive of a tumor, but beaded dilatation of the duct may be also seen in chronic pancreatitis in the absence of a tumor. Signs strongly suggestive of a tumor are an abrupt cutoff of the duct or dilatation of the duct in the pancreatic body or tail with a normal duct caliber in the pancreatic head. Long-standing ductal dilatation is associated with atrophy of the surrounding pancreatic parenchyma. A concomitant obstruction of the intrapancreatic portion of the common bile duct with no evidence of a stone may be caused by a carcinoma in the pancreatic head (double duct sign, Fig.14.7a). Even in the absence of a demonstrable tumor, the possibility of a periampullary or (intraductal) papillary carcinoma should be considered. Because small tumors that lead to early obstruction of the common bile duct make up the majority of the surgically resectable pancreatic tumors, identification of these lesions is particularly important. Foci of intratumoral necrosis can simulate a pseudocyst [pseudo-pseudocyslj, but their neoplastic origin is suggested by the higher density of the tumor necrosis, a relatively thick and shaggy cyst wall, and the absence of detectable calcifications. Vascular involvement is not an absolute contraindication forresection, but makes the surgical
intervention much more difficult, and careful preoperative planning is necessary. Assessment of the celiac and hepatic arteries, the superior mesenteric artery and vein (SMA, SMV) and the portal vein (PV) are crucial to successful resection. Identification of tumor spread to the retroperitoneum behind the SMA is important (d. Fig.14.5c), since direct intraoperative visualization ofthis area is diffi cuI tun til to the final stage of the operation. Arterial invasion (celiac trunk, superior mesenteric artery or hepatic artery) is less common and for most surgeons an absolute contraindication for tumor resection (Fig. 14.8a, b). Invasion of the superior mesenteric vein (Fig. 14.8c), venous confluence, or portal vein, if it is locally limited, does not preclude surgery. Various grading systems for vascular involvement have been described that assess the resectability of tumors adjacent to a vessel (Table 14.6), and positive predictive values of> 95% can be achieved. Direct contact between the tumor and a vessel with loss of the intervening fat plane does not automatically mean vascular invasion. Vascular involvement is confirmed, however, by a tumor thrombus, by vascular occlusion (nonvisualization or vessel cutoff) or by abrupt changes in vessel caliber. Indistinct vessel contours and a visible vessel "wall" does not necessarily indicate tumor infiltration but may be caused by peritumorous inflammatory reaction. The pancreaticoduodenal vein (PDV) produces an arcade of collateral connections be-
a
c
b
Fig. 14.7 Carcinoma of the pancreatic head. (a) Periampullary tumor (arrow) with associated dilatation of the common bile duct and pancreatic duct (double duct sign, arrowheads), (b) Tumor of the pancreatic head (arrow) with local lymphangitis in the mesenteric fat (broad arrow) and a small lymph node (arrowhead) in a
patient who previously underwent endoscopic stent insertion in the common bile duct. (c) Large tumor of the pancreatic head without contact to the SMA and SMV but causing a double duct sign (CPR, 4 x 1/6). Note the large metastatic node (arrowhead).
525
74 The Pancreas
a
c
b
Fig. 14.8 Volume-rendering (VRT) and MIP displays of vascular involvement in pancreatic adenocarcinoma (4 x 1/6). (a) Carcinoma of the pancreatic body infiltrating the splenic artery (arrow) and causing occlusion of the splenic vein with extensive collaterals (VRT). (b) Carcinoma of the pancreatic head (arrowheads) with en-
casement and narrowing of the SMA (arrow) indicating infiltration and irresectability (VRT). (c) Cystadenocarcinoma of the head and uncinate process causing occlusion (arrow) and stenosis (arrowheads) of the SMV. MIP demonstrates the course of the mesentericosplenic collaterals.
Table 14.6
carcinoma
Criteria for vascular
involvement of pancreatic
Grade
Vascular involvement (Raptopoulos et aI., 1997)
Circumferential vessel Involvement (Lu et al., 1997)
Surgery
Grade 0
normal vessels
no contiguity
Grade 1
loss of fat plane with or without smooth displacement of the vessel
:£25% circumference
resectable resectable
Grade 2
flattening or slight irregularity on one side of any vessel
> 25-50%
circumference
questionably
Grade 3
tumor around two sides of artery rowing of venous lumen
> 50-75%
circumference
unresectable
Grade 4
vessel occlusion
nar-
tween the portal vein and the SMV. Asymmetric enlargement of one vein may be interpreted as a secondary sign of encasement of another major vein and may be the only sign ofunresectability. Dilatation of the posterior superior pancreaticoduodenal vein (> 5m m) is an indirect sign of extrapancreatic extension with compression or occlusion of the pancreaticoduodenal venous arcade. However, the sensitivity of this sign is rather low and detection of abnormalities in these vessels is uncommon even in patients with unresectable tumors. Further secondary signs of encasement of major veins are an increased enhancement of the wall of small bowel loops and a decreased enhancement of the portal vein. A tethered, teardrop-shaped SMV was found to be a reliable sign of the unresectability of tumors.
> 75%
circumference
resectable
unresectable
Indistinct margins and stranding in the peripancreatic fat may be caused by pancreatic tumor spread due to focal carcinomatous lympitangiitis or by a perifocal inflammatory reaction (in 10% of cases) (Fig.14.7b). Tumor invasion into adjacent organs signifies unresectability, with the exception of the duodenum which is resected anyway and therefore poses no contraindication for surgery. Ascites is usually seen in advanced stages and signifies peritoneal carcinomatosis or invasion of the lesser sac. Irregular increase in density of the omental fat is highly suggestive of peritoneal carcinomatosis. It is common to find (hypovascular) hepatic metastases, but splenic metastases are very rare. However, even with spiral CT, up to 20% of patients that were considered operable by CT were found to have un-
Islet Cell Tumors detected liver lesions and peritoneal implants. Preoperative laparoscopy with ultrasound is therefore recommended in some institutions. In 40-65% of cases. lymph node metastases are already present in the area of the mesenteric root. celiac trunk. porta hepaus, or in para-aortic and paracaval or retrocrural sites at the time of tumor detection. CT has important limitations in the detection of nodal metastases because metastatic lymph nodes in this disease are frequently not enlarged. Regional lymph node metastases need to be considered when peripancreatic lymph nodes are at the upper limit orjust beyond the normal size (1-1.5 cm) or when there is a cluster of still normal sized lymph nodes (Fig.l4.7b). Rarely lymph node metastases may be quite extensive and present as a peripancreatic mass. Regional lymph node metastases (N 1) do not necessarily render the tumor unresectable. However. the 5-year survival of patients with positive nodes found at surgery is significantly worse than the prognosis of other patients. Microscopic involvement oflymph nodes is fairly common even in small pancreatic tumors and is not considered as contraindication to resection.
endocrine neoplasia) is associated with hyperplasia or adenoma of the parathyroid in 90%. adrenal cortical tumors in 40%. thyroid tumors in 20%. pituitary gland adenomas in 40% and endocrine pancreatic tumors in 80%. Islet cell tumors may also be seen in association with von Hippel-Lindau disease (17%). Functional tumors produce early symptoms resulting from their hormonal activity. and usually they are only about 1-2 cm in diameter when diagnosed. Most nonfunctional tumors are considerably larger at diagnosis. because they do not produce early symptoms. Less than 10% ofinsulinomas undergo malignant transformation. compared with 60% of gastrinomas. Most malignant tumors are nonfunctional. In functional tumors malignancy may be difficult to determine histologically'. it is indicated by local invasion of adjacent structures as well as distant metastases. CT has a role in the preoperative localization of functional tumors and in the staging of nonfunctional tumors. The results of the examination depend critically on the use of an optimal scanning technique.
Islet Cell T umars
Functional islet cell tumors are usually isoattenuating and rarely hypoattenuating to the surrounding pancreas on unenhanced CT scans. Most islet cell tumors are hypervascular (80% of all insulinomas) and show a marked attenuation increase on arterial phase scans (for differential diagnosis. see Table 14.9). Small tumors « 1 em) can be identified by their enhancement dynamics rather than by a contour irregularity in the gland (Fig. 14.9). Some tumors remain visible in the parenchyma! or portal venous phase of en-
CT Morphology
Islet cell tumor is a collective term applied to tumors arising from endocrine cells of the islet cell apparatus of the pancreas. Approximately 90% of these tumors are benign. and 75% are functional (hormonally active). The principal types are reviewed in Table 14.7. Functional pancreatic tumors are frequently associated with tumors involving cells originating from the APUD system. MEN syndrome type 1 (multiple Table
14.7
Islet cell tumors
Insulinoma
((3 cells)
Gastrinoma
(a cells)
Vipoma
(6 cells)
Glucagonoma Somatostatinoma
(a cells)
Frequency
Malignancy
Site of occurrence
Remarks
60-75%
5-10%
Entire pancreas
10% multiple
20%
60%
Pancreatic head (50%), penpancreatrc or duodenal (35%)
20% multiple; calcifications may occur; thickened bowel folds in CT; hepatic metastases
4%
60%
Body or tail of pancreas
vasoactive intestinal peptide secretion; calcifications may occur; thickened bowel folds CT; hepatic metastases
1%
80%
Body or tail of pancreas
<1%
50-90%
Pancreatic
head
in
14 The Pancreas
a
b
Fig. 14.9 InsLilinoma. (a) The tumor is not demonstrated on precontrast scans, (b) Postcontrast scan during the late arterial phase shows a hypervascular le-
Table 14.8
Criteria for differentiating
adenocarcinoma
Adenocardnoma Hypoattenuating Calcifications
c sion in the pancreatic tail. (c) The lesion is poorly visualized during the parenchymal phase.
from a nonfunctioning
islet cell tumor
Islet cell tumor after contrast
administration
present in 2%
Usually hyperattenuating Calcifications
after contrast
present in 25%
Central necrosis
Central necrosis and cysts are rare
Vascular encasement
No vascular
Locally invasive
Noninvasive
hancement but most are isoattenuating and can no longer be detected in these phases. Nonfunctional tumors, which can range from 3-25 cm in size, show less enhancement after contrast administration (Fig. 14.10). Inhomogeneities represent areas of necrosis. Calcifications are found in approximately 25% of functional and nonfunctional tumors, whereas adenocarcinomas very rarely exhibit calcifications (Table 14.8). The secondary effects oflarger tumors consist of ductal dilatation and atrophy of the pancreatic parenchyma. Signs of local invasion and the detection of metastases are the only means available for distinguishing malignant from benign lesions. Arterial phase imaging improves the detection rate of liver metastases.
administration
encasement
Fig. 14.10 Extensive, cell tumor (gastrinoma) metastases.
nonfunctioning malignant islet with hepatic and lymph node
Metastases
Solid Papillary
Epithelial Tumor
The low-grade solid papillary epithelial tumor (papillary epithelial neoplasm, papillary cystic neoplasm) is a very rare neoplasm that almost exclusively affects women between 20 and 30 years of age. It is a separate histopathologic entity that shows little propensity for metastasis. •
CT Morphology
The tumor most commonly involves the pancreatic tail and can reach considerable size (> 1Oem). It contains areas of cystic degeneration, necrosis, and solid material in varying proportions, and its CT morphology resembles that of macrocystic adenoma. It shows little or no contrast enhancement, and cystic lesions do not contain septations. The absence of metastasis, even with large tumors, is typical. Pleomorphic
Carcinoma
Pleomorphic carcinoma is a rare tumor (2-7% of all pancreatic tumors) that is histologically similar to sarcoma. It is characterized clinically by a downhill course, and death usually occurs within three months after diagnosis. •
CT Morphology
Pleomorphic carcinoma is commoner in the body or tail of the pancreas than in the head, and most tumors have reached considerable size when diagnosed. The tumor may contain cystic necrotic areas and shows a corresponding inhomogeneoUs pattern of enhancement. Typically there is widespread metastasis by the time of diagnosis, including spread to unusual sites such as the mediastinal lymph nodes, kidneys, and adrenals. Other distant metastases may be found in the liver, lung, and bone. Abdominal lymph node metastases are so extensive that the tumor may be misinterpreted as a lymphoma. Lymphoma
Primary lymphomas of the pancreas are rare « 1 %ofall pancreaticneoplasms). Themajority are non- Hodgkin lymphomas. Secondary lymphomatous involvement of the pancreas is also rare and is generally associated with extensive abdominal lymphomas.
Fig. 14.11 Pancreatic lymphoma with diffuse organ enlargement and extensive stranding of the surrounding mesenteric fat.
•
CT Morphology
Pancreatic lymphomas may appear as large homogeneous masses, but it is more common to find central tumor necrosis. Approximately 60% are larger than 4cm when diagnosed. They can be difficult to distinguish from pancreatic carcinoma. Extensive, diffuse infiltration of the peripancreatic fat and disproportionately mild pain symptoms are suggestive of malignant lymphoma (Fig. 14.11). Lymphomas displace the peripancreatic vessels without invading them, and the pancreas may be displaced anteriorly by enlarged peri pancreatic nodes. With secondary pancreatic involvement by malignant lymphoma, images will typically show extensive manifestations of extra pancreatic disease. Metastases
Carcinomas of the lung, breast, thyroid, kidneys, ovaries, testes, liver (HCC), and melanomas can produce intrapancreatic metastases. With its proximity to other organs, the pancreas may also be invaded by direct extension of tumors from the stomach, colon, gallbladder, liver, or kidney. •
CT Morphology
Pancreatic metastases may be hypovascular or hypervascular, depending on the primary tumor (Table 14.9). There is an even distribution throughout the pancreas. They may be morphologically indistinguishable from primary pan-
529
74 The Pancreas Table 14.9 Differential pancreatic tumors Primary
diagnosis
of
cr
creatic neoplasms on scans, but there is less propensity for infiltration of the peripancreatic fat and vascular involvement.
nypervascular
tumors
Islet cell tumors Micro- and macrocystic
adenomas
(septa)
Metastases Carcinoidtumor Melanoma Renal cell carcinoma Adrenal carcinoma Thyroid carcinoma Angiosarcoma Leiomyosarcoma
Pancreatitis Pancreatitis is classified into various forms according to the course and severity of the inflammatory processes. This classification has significant prognostic and therapeutic implications. has two major roles. staging the severity of the inflammatory process and detecting complications. especially the identification and quantification of parenchyma! and peripancreatic necrosis. should be used for diagnosis only in patients with equivocal clinical findings. (See recommendations for use of CT. below.)
cr
cr
Table 10.10 Acute Pancreatitis (clinical classification according to the Atlanta conference of 1992) Mild acute pancreatitis
minimal
Severe acute pancreatitis
associated with organ failure or local complications such as necrosis, abscess, or pseudocyst
Acute fluid collections
lack of a defined wall
Pancreatic
nonenhanced pancreatic parenchyma >3 cm or <30% of pancreatic area (CT numbers after eM <SOHU) infected vs. uninfected necrosis by fine needle aspiration biopsy (FNAB)
Acute pseudocyst
defined wall. usually >4 weeks after onset of pancreatitis
Pancreatic abscess
defined wall. fungi or pus
Acute Pancreatitis On the basis of the severity of the pathologic changes. clinical manifestations and laboratory values. acute pancreatitis is classified as mild or severe (Table 14.10). The predominant macroscopic feature of mild acute pancreatitis is interstitial edema. While parenchymal necrosis is macroscopically not visible. small foci of acinar cell necrosis occasionally may be found. Necrosis of peripancreatic adipose tissue is common. It represents the more common form (70%) and generally runs an uncom plicated course. The diagnosis is based on clinical and laboratory findings and does not require imaging studies. Severe acute pancreatitis is characterized by extensive changes of the peripancreatic fat and hemorrhagic fluid within the retroperitoneal space. There are patchy and confluent zones of parenchymal necrosis with associated necrosis of
organ
dysfunction
little or no necrosis,
vessel walls, foci of hemorrhage and necrosis, and disruption of the pancreatic duct. Clinically, severe pancreatitis is indicated with three or more Ranson criteria or eight or more APACHE 11 criteria any time during the course of disease. Because the pancreas lacks a true capsule, enzymerich pancreatic secretions that escape from the inflamed pancreas can easily reach the peripancreatic fat, producing areas of fat necrosis and leading to more extensive inflammatory spread, manifested by tracks of necrotic material. While the pancreatic parenchyma is still intact in serous exudative pancreatitis, areas of (often hemor-
Acute Pancreatitis rhagic) pancreatic necrosis develop in the hemorrhagic necrotizing form. The superinfection of these areas leads to a suppurative form of pancreatitis with abscess formation. The latter has a poor prognosis with a high mortality rate. While the mild form of pancreatitis shows no or only mild organ dysfunction and usually responds rapidly to conservative treatment, the severe form is associated with much more serious physical and laboratory symptoms (shock, pulmonaryinsufficiency, renal failure, CI bleeding, metabolic disturbances, etc.). The frequency of complications increases with the severity of the pancreatitis. Progression from mild to severe can occur, but is a rare phenomenon. A failure of clinical response within 48 to 72 hours should result in additional diagnostic tests to re-evaluate the initial diagnosis and depict any complication. •
CT Morphology
The mild acute pancreatitis (edematous or interstitial pancreatitis) is characterized by swelling of the affected portions or the entire gland (Fig. 14.12). The edematous areas appear slightly enlarged and show decreased attenuation. The margins of the gland are blurred, and there may be thickening of Cerota's fascia. Postcontrast scans usually show slightly inhomogeneous enhancement of the pancreatic parenchyma with no perfusion defects (necrotic areas). In 30% of patients with definite clinical and laboratory signs of pancreatitis, the pancreas appears completely normal on CT scans. The detection of pancreatic enlargement in itself is inconclusive, given the normal range ofvariation in the size of the gland. Hypoattenuating, non-enhancing acute fluid collections (exudates, < 15 HU) that lack a defined wall are found in serous exudative pancreatitis. Fat necrosis may be hard to differentiate from such exudates but they are less well defined, and have a higher than fluid attenuation (> 25 HU). Inhomogeneities are highly suggestive of fat necrosis. The pathways of such extrapancreatic inflammatory spread initially conform to pre-existing fascial spaces (Fig. "14.13). Inflammatory exudate can spread anteriorly and laterally in the anterior pararenal space along Cerota's fascia, the mesenteric root, and the
Fig. 14.12 Mild acute pancreatitis with interstitial edema, characterized by swelling of the pancreas and blurring of its margins (Stage B, CTSI = 1).
gastrohepatic, gastrosplenic, and gastrocolic ligaments (Fig. 14.14). Infiltration of the pararenal space is usually greater on the left side than on the right. Rarely, inflammation may track around the renal fasciae to involve the posterior pararenal space. Downward extension into the lesser pelvis may also occur. Increased density (infiltration) of the pararenal space with sparing of the fat-filled perirenal space is called the renal halo sign. Enzymatically active exudates must penetrate the fascial planes in order to enter the omental bursa, the perirenal spaces, and the mediastinum. Even early stages of inflammatory spread may incite a small accompanying ascites and a reactive pleural or pericardial effusion. In severe acute pancreatitis (extensive exudative pancreatitis or liemorrliagic necrotizing pancreatitis] the gland is markedly enlarged, nonhomogeneous, and poorly demarcated from surrounding tissues. Necrotic areas in the pancreas appear hypoattenuating, parenchyma! sequesters are isoattenuating, and hemorrhagic areas are hyperattenuating. Intravenous contrast should be used to determine the extent of pancreatic necrosis: viable parenchyma shows contrast enhancement while necrotic parenchyma does not (Fig. 14.15). Extrapancreatic exudates and fat necrosis spread like in the mild form, but the inflammatory and necrotic changes are much more extensive and severe. Absorption of the exudates occurs over a period of weeks to months, with serous elements undergoing more rapid absorption than
531
14 The Pancreas Omental
bursa
Fig. 14.13 Pathways of inflammatory spread in acute pancreatitis. CT demonstrates focal or generalized enlargement of the gland. Inflammatory collections may spread along the anterior pararenal spaces, into the mesenteric root, or into the omental bursa.
Lateroconal
1111111111111
Anterior
pararenal space
Perirenal space
I07m/;7A
a
Posterior Gerota's-fascia
Posterior pararenal space
c
b
Fig. 14.14 Severe acute pancreatitis with moderate serous exudates (4x1 /6). Note the small focus of parenchymal necrosis (arrow) seen on axial (a) as well as coronal sections (b). Parenchyma! perfusion is otherwise unchanged. The caudal spread of the exudates is
hemorrhagiccomponents.Asabsorptionoccurs, the periphery of the inflammatory mass becomes walled off by granulation tissue, first producing an encapsulated fluid collection that subsequently develops into a pseudocyst (Fig. 14.3). These acute pseudocysts evolve at least four weeks or more after onset of extrapancreatic fluid.
best seen on a sagittal volume-rendered image (c). Exudates are located anterior and posterior to the pancreas, in the omental bursa/lesser sac (8). and along Gerota's fascia (arrowheads) extending into the pelvis (Stage E, CTSI=6) .
CT Staging of the Severity of Acute Pancreatitis
Several studies have shown that initial clinical evaluation may miss approximately 60% of patients with severe acute pancreatitis who are likely to develop complications with a high morbidity and mortality. This is because clinical evaluation alone measures only the physiologic or systemic response of the patient. CT can depict
Acute Pancreatitis
533
b Fig. 14.15 (a) Necrotizing pancreatitis after ERCP. Note the stent in the common bile duct, a postinterventional hematoma in the duodenal wall (") and a small infected fluid collection in the porta hepatis (Stage E, CTSI= 6). (b) Hemorrhagic necrotizing pancreatitis with
a perfusion defect in the body of the pancreas. Large inflammatory exudates are present in the left anterior pararenal space (= paracolic gutter). There is free fluid in the peritoneum (Stage E, CTSI = 8).
the extent of glandular damage and identify complications prior to clinical manifestation. Such complications include extensive peripancreatic fluid collections, abscess formation, and vascular, biliary, or gastrointestinal involvement. CT can differentiate the mild edematous from the more severe exudative and necrotizing forms with 95% accuracy, but it is only 60% accurate in distinguishing hemorrhagic necrotizing pancreatitis from the suppurative form. The false negative rate of CT for detecting parenchymal necrosis was reported to be some 20% in patients with minor necrosis and less than 10% in patients with major necrosis. The specificity approaches 100%.
Balthazar and colleagues (1990) defined a composite CT Severity Index (CTSI, Table 14.11) to assess the severity of parenchymal gland necrosis (scores 0 to 6) and the extraglandular inflammatory process (stages A to E, scores 0 to 4). The CTSI ranges from 0 to 10 and shows excellent correlation with the clinical severity, morbidity and mortality rates. Patients with CTS1 of 0 or 1 had no mortality and no morbidity, while patients with an index of 2 had no mortality and only 4% morbidity. In contrast, a CTSI of 7-10 resulted in a 17% mortality and a 92% complication rate.
Table 14.11 CT severity index (Balthazar et al., 1990) Assessment
of acute inflammatory process
StageA:
score 0
Stage B:
score 1
intrinsic pancreatic changes: focal or diffuse gland enlargement, mild heterogeneity of parenchyma, small intrapancreatic fluid collections «3 ern)
StageC:
score 2
like stage B but with mild inflammatory changes of the peri pancreatic tissue
Stage D:
score 3
like C but more prominent peripancreatic inflammatory changes, but not more than one fluid collection
StageE:
score 4
multiple or extensive extrapancreatic fluid collections or abscess
normal pancreas
Assessment of pancreatic necrosis
Normal:
score 0
no necrosis, regular enhancement
Mildly necrotic:
score 2
<30% of pancreatic parenchyma
Moderately necrotic:
score 4
30-50% of pancreatic parenchyma
Extensively necrotic:
score 6
>50% of pancreatic parenchyma
Recommendations for Use of CT in Patients with Acute Pancreatitis
An initial CTscan is indicated: • in patients with clinical diagnosis in doubt (best after at least 48 hours after onset of symptoms); • in patients with clinical signs of severe pancreatitis (Ranson score >3 or APACHE 11 score >8); in patients who do not show adequate improvement within 72 hours of initiation of conservative medical treatment; in patients with suspected necrotizing pancreatitis: after onset of clinical symptoms glandular necrosis develops often within the first 24 hours and virtually always within the first 72 hours; therefore the initial CT scan should be obtained after these 72 hours if the clinical conditions allows; • in patients who develop a sudden change in clinical status (fever, pain, decreasing hematocrit, hypotension, etc) after initial improvement suggesting complications,
Follow-up eTscans
should be obtained:
• after 7 to "10 days andlor before hospital discharge in all patients with a grade D or E pancreatitis (CTSI score of 3-10) to exclude clinically silent complications', • in patients with no or inadequate clinical improvement or deterioration under treatment; • in patients with grade A-C pancreatitis (CTSI 0-2) only when complications are suspected,
Complications of Acute Pancreatitis The complications of acute pancreatitis include pseudocyst formation, infection, hemorrhage, formation of pseudoaneurysms, and venous thrombosis,
CT Morphology
CT occasionally shows a circumscribed peri pancreatic fluid collection of high attenuation (>60HLJ), but it is difficult or impossible to differentiate hemorrhage into the peripancreatic tissue from hemorrhage into a pseudocyst based on CT findings,
Pseudocyst Pseudocysts (see also p, 520) do not develop until 4 to 6 weeks after initialization of the pancreatitis, Mostly they are related to acute fluid collections seen at the onset of pancreatitis, In about 50% of patients these collections resolve; in the remainder pseudocyst formation occurs, For preoperative differentiation of a pseudocyst from other cystic lesions (cystadenoma), cytological studies and tests evaluating enzymatic activity of the peripancreatic fluid can be advocated, CT Morphology
Pseudocysts present as rounded, well encapsulated fluid collections of varying attenuation (0-25 HU), There is contrast enhancement of the granulation tissue in the cyst wall, A reliable correlation between the degree of wall thickness or wall enhancement and the maturity of a pseudocyst (consistency of its capsule) cannot be established, Septations are uncommon, Most frequently pseudocysts are located close to the pancreas but may also be found in the liver, spleen, or mediastinum, A suspected infection (pancreatic abscess) requires needle aspiration because it cannot be ruled out on the basis of CT morphology, Gas bubbles are suggestive of abscess formation, but they are quite infrequent and may also be caused by enteric Itstulatton, or following percutaneous puncture.
Infectious Complications Hemorrhagic Pancreatitis The diagnosis of hemorrhagic pancreatitis is based on histopathologic criteria and not on the demonstration of acute hemorrhage by CT, which is seen in only 2-5% of patients with acute pancreatitis,
Even today, infectious complications of pancreatitis are associated with a high mortality rate, A pancreatic abscess must be differentiated from an infected necrosis because the prognosis is substantially different and the treatment is different: an abscess can be treated effectively
Complications of Acute Pancreatitis with percutaneous catheter drainage while infected necrosis requires surgical debridement. Infected necrosis is an early complication of acute pancreatitis involving a disseminated bacterial inflammation of necrotic pancreatic and peri pancreatic tissue. Affected patients are septic, and the mortality rate ranges between 40% and 70% depending on the extent of infected necrotic tissue (involvement of <50% or >50% of glandular tissue). The frequency of bacterial contamination in patients with necrotizing pancreatitis increases with time (24% within 1st week, 36% within 2 weeks and 71 % within 3 weeks). It is very important to confirm bacterial contamination (usually by fine needle aspiration) because well over 50% of patients with severe pancreatitis have clinical symptoms of sepsis but only 50% of them actually suffer from superinfection. Pancreatic abscesses are circumscribed intraabdominal collections of pus located in proximity to the pancreas gland. They contain little or no necrosis and usually develop from infected pseudocysts. Pancreatic abscess is a late complication (> 5 weeks), occurring at a stage when the acute manifestations of pancreatitis are already subsiding. The mortality rates described in the literature vary between 20% and 50%. •
CT Morphology
Infected necrosis shows the same CT features as necrotic pancreatitis, including non-enhancing portions of the gland and peripancreatic fluid collections. Infection may be present in solid or liquefied regions of the gland, or peripancreatic fluid collections, or fat necrosis. Regions of fat necrosis present as heterogeneous hypoattenuating areas that are often ill-defined and have near soft-tissue attenuation (> 20 HU). Gas bubbles in areas of pancreatic gland necrosis or dispersed in necrotic peripancreatic tissue (emphysematous pancreatitis) are highly suggestive of infection. Pancreatic abscesses show CT features similar to pancreatic pseudocysts. They are wellcircumscribed regions of fluid attenuation (1030 HU) with a contrast -enhancing "wall". Pancreatic abscesses are multiple in 30% of cases, and many are located some distance from the pancreatic bed. The detection of gas inclusions in the pancreas and peripancreatic tissue is the only morphologic sign suggestive of bacterial superinfec-
Fig. 14.16 Pancreatic abscess with small gas inclusions (arrow). There is inflammatory spread along Gerota's fascia (Stage E, CTSI=6). Rim enhancement indicates the formation of granulation tissue (arrowheads), but whether this region is also superinfected cannot be determined based on CT criteria alone.
tion, but gas is detected in only 30-50% of cases (Fig. 14.16) and can be also caused by a gastrointestinal fistula in the absence of infection, or by previous percutaneous intervention. If gas collection is not present, it is not possible to distinguish infected from noninfected necrosis, or a pseudocyst from an abscess, based on CT findings alone. Both the abscess membrane and the inflammatory rim surrounding a pancreatic fluid collection show the same pattern of peripheral enhancement. Clinical parameters such as persistent leukocytosis and fever suggest the diagnosis. Equivocal cases should be resolved by fine-needle aspiration biopsy of the suspicious area for microbiologic analysis. A positive Gram stain or culture indicates the need for intervention. Vascular Complications
Hemorrhage results from the enzymatic erosion of small blood vessels. Larger vessels such as the hepatic, splenic and gastroduodenal arteries may develop pseudoaneurysms in up to 10% of cases. Pseudocysts and chronic pancreatitis predispose to thrombotic occlusions of the portal venous system. The splenic vein is most commonly affected, which may lead to splenic infarction.
535
14 The Pancreas •
CT Morphology
Pseudoaneurysms are most clearly depicted during the arterial phase of enhancement (Fig. 14.17) but can also be detected during the parenchymal phase. Venous occlusions lead to non opacification of the affected vessel and incite the development of peripancreatic collaterals. Intraluminal filling defects due to thrombotic material can occasionally be detected in the affected vessels. Chronic Pancreatitis
Chronic pancreatitis is most commonly the result of alcohol abuse or biliary disease. There may be recurring episodes of acute exacerbation (chronic relapsing pancreatitis), or the disease may take a slowly progressive course. Chronic pancreatitis is characterized by irreversible damage to the organ with progressive exocrine andend ocrin eins uffi ci en cy. Patients with pancreatic calcifications [chronic calcifying pancreatitis) show an increased incidence of pancreatic carcinoma. Chronic pancreatitis and a pancreatic tumor coexist in up to 5% of cases. The goals of CT scanning are to detect intraductal lithiasis as the cause of the disease, exclude a pancreatic tumor, stage the severity of disease, detect complications such as pseudocysts, ductal stenosis, and fistula formation,
a Fig. 14.17 Vascular complications of necrotizing pancreatitis. (a) Pseudoaneurysm of the gastroduodenal artery with partial thrombosis that has led to cholestasis with hydrops of the gallbladder, (b, c) Large infected
and assist in the selection of treatment alternatives. •
CT Morphology
CT findings are normal in 10% of patients with clinically diagnosed chronic pancreatitis. The inflammation may lead to focal or diffuse enlargement of the pancreas (hypertrophic form), but more frequently the gland is atrophic. This atrophy is characterized by reduced gland size, decreased structural density of the parenchyma, and ductal dilatation. The detection of ductal dilatation or calcifications signifies the pancreatitis- induced form of atrophy as opposed to the "senile atrophy" that occurs after age 60. Calcifications are more common in the alcoholrelated form (50%) than in the biliary form (approximately 20%, see Table 14.13). Most calcifications are located adjacent to or within the ductal system. Pseudocysts are found in approximately 30% of cases and may have an intra- or extrapancreatic location (Fig. 14.18). Pancreatic ductal calcifications occur in up to 90% of patients with severe, and in up to 55% of patients with only mild to moderate, pancreatitis; thus calcifications per se are a poor predictor of disease severity. Widening of the pancreatic duct to more than 4mm is found in more than 50% of patients. The duct is more often irregular (beaded) than smooth due to distortion by scarring (Fig. 14.19). This contrasts with tumor -related obstruction, which is more likely to cause
b
c
pseudoaneurysm of the splenic artery with gas inclusions and a layered thrombus, which has caused erosion of the duodenum (arrow, c).
Chronic Pancreatitis
Fig. 14.18 (a) Moderate chronic calcifying small cysts, and dilatation of the pancreatic
pancreatitis with enlargement of the gland, multiple calcifications duct. (b) Severe chronic calcifying pancreatitis (VRT, 4 x 1/6).
c
b
a
Fig. 14.19 Pseudotumor of the pancreatic head due to severe chronic pancreatitis (4 x 1 /6). Note the intraductal calcifications on the axial sections (a). The pancreatic duct is markedly dilated and shows irregularities towards the pancreatic tail (b, coronal oblique section).
Table 14.12 1996)
Cambridge
classification
of pancreatic
The pancreatic head is markedly dilated and inhomogeneous but no focal tumor can be discerned (c. coronal section). Note that it may be impossible to differentiate pseudotumor from carcinoma in patients with chronic pancreatitis.
morphology
in chronic pancreatitis
(adapted
from
Freeny,
Changes
ERCP
None
no abnormal
Equivocal
MPDnormal
One of the following: <3 abnormal LSB, MPD 2-4 mm, gland enlarged over 2 times normal size, heterogeneous parenchyma
Mild
MPDnormal
Two or more signs for diagnosis: >3 abnormal LSB, MPD 2-4 mm, slight gland enlargement. heterogeneous parenchyma
Moderate
Severe
CT and US LSB
and
normal gland size and shape, homogenous
parenchyma
MPD changes
small cysts <1 Omm, MPD irregularity
LSB changes
focal acute pancreatitis « 1/3 of the gland), increased echogenicity of MPD walls, gland contour irregularity
any of the above changes plus one of the following: cyst> 1Omm, intraductal filling defects, calculi, MPD obstruction irregularity, or contiguous organ invasion
LSB = lateral side branch ducts, MPD = main pancreatic duct
or stricture,
enhancement!
severe
MPD
537
538
14 The Pancreas
smooth dilatation and abrupt termination of the duct. However, the shape and extent of the dilatation are not reliable indicators for differentiating benign from malignant disease. An acute exacerbation of chronic pancreatitis can cause focal or diffuse enlargement of the gland with decreased contrast enhancement and occasional infiltration of the peripancreatic tissue. These changes may be morphologically indistinguishable from pancreatic carcinoma at this stage, necessitating further diagnostic studies such as ERCP and biopsy. The International Cambridge Classification is the most widely used system for the morphologic staging of the severity of chronic pancreatitis (Table 14.12). Correlation between morphologic findings and function tests, and between morphologic changes and clinical stages, is high in advanced disease but only moderate in mild or early chronic pancreatitis. Complications of chronic pancreatitis are similar to those of acute pancreatitis and include formation of fluid collections and inflammatory involvement of the vascular, gastrointestinal and bile duct system. Imaging studies playa crucial role in deciding whether conservative medical, surgical, or interventional treatment is more appropriate. Indications for surgical intervention are suspected tumors, persistent pain, and complications such as biliary or duodenal obstruction, or pseudocysts.
Table 14.13 cations
Differential diagnosis of pancreatic calcifi-
Chronic pancreatitis
Alcohol-related pancreatitis (20%-50%) Biliary pancreatitis (20%) Hereditary pancreatitis (35%-60%) Idiopathic pancreatitis Pseudocyst(12%-20%) Kwashiorkor: tropical pancreatitis, indistinguishable from alcohol-related pancreatitis
Tumor Microcystic adenoma (40%, stellate) Macrocystic adenoma «15%) Adenocarcinoma (2%) Cavernous hemangioma, lymphangioma (phleboliths) Metastases (e.g., from colon carcinoma) Hyperparathyroid
ism
Chronic pancreatitis in 10%, indistinguishable from alcohol-related pancreatitis Cystic fibrosis
Fine granular calcifications in advanced stage
Diffuse Pancreatic Changes Cystic Fibrosis (Mucoviscidosis)
Cystic fibrosis may induce degeneration of the pancreas. •
fibrosis
or fatty
CT Morphology
Usually the pancreatic tissue is nonhomogeneous clue to fibrosis and fatty degeneration. Cystic structures (ranging from millimeters to centimeters in size) and calcifications have been described. Atrophy occurs in the late stage. Complete fatty involution is possible (Fig. 14.20). Fig. 14.20 Complete fatty involution of the pancreas is typical for patients with cystic fibrosis.
Pancreatic Trauma Hemochromatosis
Primary terized organs docrine sis (e.g. mia) is •
CT
idiopathic hemochromatosis is characby iron storage in the pancreas and other (liver, spleen, heart, lymph nodes, enorgans, skin), Secondary hemochromatocaused by repeated transfusions for anenot associated with pancreatic changes, Morphology
The CT attenuation of the pancreas is homogeneously increased in hemochromatosis, reaching values of70-130 HU
Lipomatosis
senile atrophy, the end stage of chronic pancreatitis, cystic fibrosis, or following occlusion of the pancreatic duct Lipomatous pancreatic atrophy in children is a rare disorder caused by a vtraltnfectton. CT
Morphology
The fatty tissue deposits cause decreased structural density of the pancreatic parenchyma, accompanied by the displacement of individual gland lobules, The mesenteric vessels and pancreatic duct appear very prominent, but the duct is of normal caliber «5mm), The gland may be small with normal contours, In extensive cases, the parenchyma in total may show fat attenuation (Fig, 14,20),
Increased fat deposition in the pancreatic parenchyma may occur in the setting of obesity,
Trauma and Postoperative Changes Pancreatic
Trauma
Pancreatic injury is infrequent in blunt trauma and is usually associated with injuries to other organs such as duodenum and liver. It is seen more frequently in children or young adults, Usually, the pancreas is compressed from the anterior against the spinal column, e,g" by a seat-belt or steering wheel in car accidents, This may produce a contusion, hematoma, laceration, or fracture that can be graded with the MST organ injury scale (Table 14.14), Lacerations and fractures that rupture the pancreatic duct require emergency surgery in which part of the organ may have to be resected, Overlooking these injuries may lead to recurrent pancreatitis with pseudocyst or abscess formation, Other possible late sequelae include ductal strictures, and fistulization, •
CT
Morphology
In laceration or fracture, CT shows interruption of the pancreatic body with associated hemorrhage in the anterior pararenal space, Less severe injuries may present as edematous swelling, a focal mass, or contour deformities, which may be associated with peripancreatic fluid collections (Fig, 14,21), Indirect signs to look for are fluid collections around the superior mesenteric
Table 14.14 juries Grade
AAST organ
injury scale
for pancreatic
Injury
Description
Hematoma
Minor contusion duct injury
Laceration
Superficial laceration without
Hematoma Laceration
Laceration
duct
in-
without
injury
Major contusion without duct injury or tissue loss Major laceration without duct injury ortissue loss Distal transection or parenchymal injury with
duct
injury
IV
Laceration
Proximal transection or parenchymal injury involving papilla
Laceration
Massive disruption creatic head
Advance one grade for multiple organ up to grade III. Proximal pancreas mesenteric vein.
injuries
is to the right ofthe
of pan-
to same
superior
artery, transverse mesocolon, lesser sac and between the pancreas and splenic vein, In severe cases with ductal laceration, necrotizing pancreatitis may develop,
539
74 The Pancreas
•
Fig. 14.21 Poly traumatized patient with pancreatic contusion (arrow), signs of exudative pancreatitis, free air (arrowhead) in the gall bladder bed due to small bowel perforation, and acute perisplenic contrast extravasation (wide arrow) due to splenic rupture.
Traumatic injuries to the surrounding organs should be carefully sought for, which may require additional oral contrast. Evaluation of duct integrity is the terrain ofMRCP or ERCP, depending on the stability of the patient. Occasionally, the pancreas may be deceptively normal at CT due to isodense lacerations or lack of contour irregularities. Thickening of Cerota's fascia is a very nonspecific sign of an acute or old pancreatic injury. Pancreatic Resection
Various operations such as the Whipple procedure or pancreatectomy involve partial or complete resection of the pancreas. It may also be necessary to remove the spleen and portions of the stomach, omentum, or bowel. In partial pancreatic resections, anastomoses may be constructed to provide for drainage of pancreatic secretions to the stomach or bowel.
CT Morphology
Following resection ofthe pancreatic tail, the rest of the gland retains an essentially normal shape and position. With a resection of the pancreatic head, the tail is anastomosed to the stomach or to a transposed loop of small bowel. Generally the anastomotic site is not clearly visible on CTscans. When the pancreatic head has been resected, the rest of the gland may atrophy. Peri operative fluid collections may signal complications such as seroma, hematoma, abscess, biloma, or pseudocyst. A leaky pancreatic anastomosis can lead to postoperative pancreatitis. In patients who have undergone a pancreaticoduodenotomy for carcinoma, lesions of soft-tissue density in the pancreatic bed that are not identified as prolapsed loops of small bowel should raise suspicion of tumor recurrence. Pancreatic Transplantation
Aside from general postoperative complications, there are no indications for CT scanning after pancreatic transplantation. •
CT Morphology
The pancreatic allograft lies within the lesser pelvis in contact with the iliac vessels and bowel loops. Its enhancement characteristics are typical of pancreatic tissue. Adjacent bowel loops should be adequately filled with oral contrast medium so that they can be positively identified. The morphologic features of acute rejection range from a normal-appearing CT scan to edematous enlargement with regional parenchymal hypoattenuation. Chronic rejection is manifested by atrophy of the pancreatic allograft.
15 The Gastrointestinal Tract C. Schaefer-Prokop,
M.Jorgensen
Anatomy Esophagus(p .543) Stomach and Duodenum Small Intestine (p. 546) Colon (p. 543)
. (p. 543)
Examination Technique Esopha gus(p.548) Stomach (p. 548) Small Intestine (p. 549) Colon (p. 550) Acute Abdomen (p. 551) Esophagus
Duplication Cysts and Diverticula (p. 552) Benign Tumors (p. 553) Esophageal Carcinoma (p. 553) Esophageal Fistula (p. 555) Perforation of the Esophagus (p. 555) Esophagitis (p. 555) Achalasia (p. 556) Strictures (p. 556) Hiatal Hernia (p. 556) Esophageal Varices (p. 557) Dysphagia lusoria (p. 557) Postoperative Changes (p. 558) Stomach Cysts and Diverticula (p. 558) Benign Gastric Tumors (p. 559) Gastric Carcinoma (p. 559) Gastric Lymphoma (p. 561) Other Malignant Gastric Tumors (p. 562) Gastric Sarcoma (p. 562) Kaposi Sarcoma (p. 562) Gastric Metastasis (p. 562) Gastritis, Ulcer Disease (p. 563) Varices(p.563) Postoperative Changes (p. 563)
.
542
15 The Gastrointestinal Tract
Intestine Cysts
564
and
Benign
(p.
Diverticula
Tumors
Malignant
564)
(p. 564)
Small
Bowel
Gastrointestinal
Tumors
Stromal
(p. 565)
Tumor
(p.
566)
Carcinoid (p. 567) Lymphoma (p. 568) Colorectal Colon
(p.
Carcinoma Cancer
Tumor
(p.
Recurrence
Inflammatory
and
Appendicitis Diverticulitis Tuberculosis
(p.
(Regional
(p.
Colitis
versus
Whipple's
Diseases
(p.
573)
Enteritis)
(p.
Enterocolitis
Host
(p. 577)
579)
Pseudomembranous Graft
Bowel
577)
Disease
Ulcerative
(p. 571) 573)
Infectious
(p. 573) (p. 576)
Grohn's
569)
Screening
(p.
Disease
580)
581)
Disease (p. 581)
Postirradiation
Changes
(p. 581)
Small Bowel Obstruction (SBO) (p. 582) Mechanical Bowel Obstruction (p. 583) Strangulation/Ischemic Volvulus,
Closed
(p.
Intussusception Abdominal
Internal Bowel
Pneumatosis Postoperative
Disease
(p. 584)
(p. 585)
586)
(p. 586) (p. 586) (p. 586) (p. 588) (p. 588)
Hernias
Hernias Hernias
Perforation
Mesenteric
Loop
Herniation
Diaphragmatic External
Bowel Bowel
Ischemia cystoides Findings
and
Infarction
intestinalis
(p.
(p. (p.
589)
591)
591)
Trauma
Endoscopy is the commonest initial diagnostic procedure for evaluating diseases of the upper and lower gastrointestinal tract. Laparotomy is used for the staging of tumors of the stomach and colon. Only small bowel studies still rely primarily on imaging procedures. CT in this setting can image the bowel wall and lumen in addition to surrounding structures and therefore is often superior to conventional barium studies, As a rule, conventional CT is not used for the staging of primary gastrointestinal tumors (except esophageal carcinoma). CT is used to investigate suspected hepatic metastases, however, and it is applied in advanced tumor cases to direct the planning of adjuvant pre-
593
operative therapy and to monitor therapeutic response. Presently, new techniques in computed tomography such as hydro-CT, CT enteroclysis, and CT colonography, especially when cornbined with multislice scanning, have increased the sensitivity of CT in detecting and staging gastrointestinal tumors or in evaluating inflammatory diseases, although the relative merits of these techniques have not yet been clearly established. Virtual colonoscopy, in particular, holds great potential as a screening tool for colorectal cancer but has not yet been shown to be cost-effective. CT has gained new indications, such as the evaluation of suspected appendici-
Anatomy tis, acute abdomen, suspected bowel ischemia, diverticulitis, and acute bleeding (Table 15.1). Ultrasound is a potent competitor in pediatric patients, suspected appendicitis, or in patients with an acute abdomen. Endosonography has been widely accepted for evaluating the submucosal extent of tumors or cancers in the esophagus, stomach, and colon. MRI holds great Table 15.1 Tumor
Indications
for CT examination
diagnosis
potential for the evaluation of benign disease such as inflammatory bowel disease, localization of acute gastrointestinal bleeding using blood pool contrast agents, and for the evaluation of rectal cancer with endorectal coils, but has so far failed to be widely accepted for indications concerning the alimentary tract.
of the gastrointestinal
Esophageal,
tract
gastric, and colorectal cancer: therapy planning in advanced M staging (liver)
tumor stages, Detection of small
bowel
tumors
Lymphomas Re-staging after therapy of primarily unresectable tumors Onocologic follow-up in gastric and rectal carcinoma Colon cancer screening Inflammatory
bowel
disease
Mural
involvement,
Acute abdomen
Appendicitis,
General
Postinflammatory,
Other
follow-up indications
ileus,
skip lesion,
fistulae,
conglomerate
obstruction,
search
for hemorrhage
postoperative
masses,
obstruction
post-traumatic
Indeterminate abdominal complaints Equivocal radiographic, sonographic, Retrocardiac mass
or endoscopic
findings
Anatomy Esophagus
The cervical portion of the esophagus lies directly posterior to the trachea. Often it is not clearly demarcated from neighboring structures due to a paucity of intervening fat. The intrathoracic portion of the esophagus is intimately related to the trachea, the left main bronchus, and the left atrium of the heart. Portions of the intrathoracic esophagus may be poorly delineated by CT due to transmitted pulsations. On passing through the diaphragm, a short segment of the esophagus acquires a peritoneal covering as it turns left and enters the gastrtc fundus. For correlation with endoscopic measurements it is helpful to note that the upper esophageal sphincter is located 15cm from the incisor teeth (15cm aboral). From a surgical standpoint, the esophagus is divided into three segments:
• proximal third (extending from the upper , esophageal sphincter to the tracheal bifurcation; 15-25 cm aboral); • middle third (extending from the fourth to seventh thoracic vertebra; 25-32 cm aboral); • distal third (extending from the seventh thoracic vertebra to the gastric cardia; 32-42 em aboral). The wall of the air-filled esophagus should not exceed 3 mm in thickness, although areas of wall thickening that can been seen on CT may be nonspecific. The esophageal mucosa is rarely visible even on contrast-enhanced scans.
Stomach and Duodenum The gastric cardia starts approximately 3 cm past the site where the esophagus passes through the diaphragm. Even when the stomach
544
75
The Gastrointestinal
Tract
is well-distended with contrast medium, the cardia has a thicker wall than the other gastric segments and may produce a focal thickening that mimics carcinoma. Scanning in the left lateral decubitus position can aid in evaluating the region of the cardia. The gastric fundus lies directly infradiaphragmatically and is in contact with the spleen, while the body of the stomach extends anteriorly and establishes contact with the anterior abdominal wall (Fig.15.1 a). The junction of the body and antrum of the stomach
usually has a constriction that should not be mistaken for the pylorus. The pylorus and duodenal bulb turn posteriorly and are intimately related to the left hepatic lobe and the head of the pancreas (Fig. 15.Ib). CT visualization ofthese structures requires the use of oral contrast medium and may require scanning in the right lateral decubitus position. The duodenum may extend far downward and may protrude between the inferior vena cava and the aorta, sometimes simuFig. 15.1 8, b Crosssectional anatomy of the gastrointestinal tract. Left colic flexure
Pylorus DuodenalBulb
Anatomy Fig. 15.1 c-e Crosssectional anatomy of the gastrointestinal tract.
+onzontai oart of :luodenum
. Descending colon mesentericvein ~ight colic
Superior mesenteric
Iexure Inferior epigastric
artery
Small intestine Cecum-~ Appendix~
"----- Ureter
-
Perirectal fascial sheath
-
artery
545
75 The Gastrointestinal Tract lating the appearance of lymphatic tissue. The mesenteric vessels cross the horizontal portion of the duodenum (Fig.l5.1c). Further distally. thejunction of the duodenum with the jejunum is marked by the duodenojejunal flexure. The normal gastric wall ranges from 4-10 mm in thickness, depending on the degree of gastric distention. There may be various degrees of fatty changes in the submucosa of the gastric wall that appear as a hypoattenuating layer on CT sections. Pathologic wall thickening (>4mm) should be diagnosed only when the stomach is optimally distended. Contrast administration improves the delineation of the gastric mucosa.
Small Intestine The small intestine (small bowel) is approximately 3 m long. Loops of small bowel in the left mid-abdomen generally represent jejunum, while bowel loops in the lesser pelvis and right lower abdomen are part of the ileum. The ileocecal region is identified by an abrupt change in the caliber of the bowel loops in the right lower quadrant and by the distinctive gas patterns: intraluminal gas in the small bowel occupies the anterior part of the lumen in the supine patient while the intraluminal gas is present in small bubbles in the non-distended colon. The ileocecal valve appears as a structure of soft-tissue density within the cecal lumen and should not be mistaken for a mass. It may contain fat in the submucosal layer. Small bowel that is not filled with contrast medium is generally in a collapsed state. When the bowel is well distended, the Kerckring folds (valvulae conniventes) can be seen. Adequately distended small bowel segments have a normal wall thickness of 1-3 mm. When the bowel has been filled with water-equivalent contrast material, CT following intravenous contrast administration demonstrates the intestinal mucosa as a thin, moderately hyperattenuating line.
Colon Loops of large intestine (colon) are distingutshed by their larger caliber and the presence of haustral markings and gas-containing feces.
The gas inclusions appear as multiple small bubbles retained within the feces in the bowel lumen. The ascending and descending colon have a retroperitoneal location (Fig.l5.1d). When not distended by contrast material. the descending colon is often in a collapsed state and often can only be distinguished from small bowel loops by its position (far posterior, just lateral to Cerota's fascia). Frequently the sigmoid colon can be identified on CT scans only by carefully tracing the descending colon or the rectum to the sigmoid on sequential scans. Portions of an elongated sigmoid colon may even be found on the right side. The rectum is approximately 12 em long, occupies a presacral extraperitoneal position, and is surrounded by perirectal fat. The perirectal fascial sheath appears as a thin line on CT images (Fig.l5.1e). The ischiorectal fossa is the largest of the anorectal spaces: it is bounded medially by the levator ani muscle and the external sphincter muscles, laterally by the internal obturator muscle and inferiorly by the glutens maximus muscle and the sacrotUberous ligament. The levator ani muscle separates the ischiorectal fossa from the extra peritoneal space of the pelvis (supralevator space) and constitutes the most important anatomic and surgical landmark of the region. Colon that is not distended with contrast medium generally contains feces and has a wall less than 2mm thick. Optimum evaluation of the lumen requires prior colonic cleansing followed by the instillation of positive or negative contrast medium. When the colon has been filled with water-equivalent contrast material, CT scans after intravenous contrast injection will display the colonic mucosa as a thin line of moderately high attenuation.
Examination Technique
547
Examination Technique Examination technique and scan interpretation are particularly challenging in cr examinations of the stomach and bowel. Optimum distension (for hydro Cl, cr enteroclysis, cr colonography, or virtual colonoscopy) combined with the seamless volume coverage of spiral or multislice CT markedly improve diagnostic accuracy compared with conventional cr. Interactive tracing of the stomach and bowel loops on the worksta-
Table lS.2
Protocol
for CT scanning
tion monitor is an essential part of scan interpretation. Injection of intravenous contrast material requires the use of negative intestinal contrast material, and the timing of contrast injection should be either optimized for the liver (if metastatic disease has to be ruled out) or for the enhancement of the intestinal mucosa (3060 s p.i., cr. Table 15.2).
of the gastrointestinal
tract
General Preparation Oral/rectal
see Table lS.3 contrast
see Table 1S.3
Patient position
Supine with elevated
Scan range
Esophagus: Stomach
arms From T1 to below posterior lung recesses; if necessary coverage may be extended to CS or below liver From diaphragm to below kidneys; coverage may include lesser pelvis Form left hemidiaphragm to ischial tuberosities
and duodenum:
Small bowel and colon: Respiratory
phase
Windowing
Inspiration Noncontrast CT: Contrast-enhanced
W/L=300/40 W/L = 400/60
CT:
16-slice"
Scan parameters
l-slice SC/TF/RI
4-slice'
sc
sc
Standard
S/10/S1
2-31
1-1.S1
Volumetric imaging
3/6/31
1-1.25
t
t
MPRb SW/RI
axial SW/RI 5/4
t
0.5-1.25
1-1.251
0.5-01.251
1-1.251
0.5-01.25
3/2
3/3 cor
3/2
3/3 cor
3/2
3/3 cor
CT enteroclysis
3/6/3
CT colonography
5/10/5
Contrast injection
l-slice V/F/D
4-slice V+N/F/D
16-slice V+N/F/DX'
Esophageal tumors
90/3/30
90 + 50/3/30
90 + 50/3/25A
Standard (portal phase)
120/2/60
120 + 50/3/60
120 + 50/3/205
Tumor detecting/ staging Biphasic CT
150/4/50
120 + 50/4/40
120 + 50/5/20A
t
150/4/25
+ 70 d
120 + 50/4/25
t
+ 70
d
120 + 50/5/20A
+ 80 d
SC= slice collimation (mm), TF =table feed (rnmjrotation], RI= recon. increment (mm), it = scanning direction, SW = effective section width (mm), MPR = multiplanar reformats, axial = axial sections, cor=coronal, sag = sagittal CM= contrast material, Contrast concentration = 300 mg/ml iodine V=volume of CM (ml), N = volume of saline flush (ml), F =flow rate (rnlrs), 0 = start delay, X= trigger region a Pitch P = TF/(Nx5C) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI = 1-1.5/0.7 or 0.5-0.8/0.5 c read: 20-25A (205) = start delay 20-25 s (205) after reaching 50 HU enhancement in the desc. aorta (spleen) dread: biphasic scan (arterial + portal phase) after 25 s (20A) for the 1" scan, 70-80 s for the 2nd scan
II
15 The Gastrointestinal
Tract
Esophagus Patient
direction and the use of a saline flush injected immediately after the contrast material. Preparation
The use of oral contrast medium is controversial. Positive oral contrast material may improve delineation of the esophageal lumen. Barium pastes (e.g .. Esophocat®) have a high viscosity and therefore may improve the coating of the esophageal mucosa but they are contraindicated if there is a risk of aspiration. However. evaluation of uptake of IV contrast material in the esophageal mucosa may be hampered. and often the oral contrast material passes the esophagus as a bolus. which leaves large portions of the esophagus without intraluminal contrast. For these reasons, we prefer not to use any positive oral contrast material. If CT angiography is planned (e.g., dysphagia lusoria), oral contrast material should always be avoided. Acquisition
Stomach Patient
Technique
A CT examination of the esophagus is part' of a thoracic CT. The scan length may have to be extended in the cranial or caudal direction, depending on the site of a suspected esophageal tumor. Spiral and multislice CT provide the best results when based on a thin-section protocol (Table IS.2). Curved planar reformations along the esophagus best delineate the longitudinal extent of an abnormality. Contrast
Injection
Intravenous contrast material is usually only injected to differentiate vascular structures from other soft tissues. The same protocols as for general mediastinal examinations are employed. Injecting larger amount of contrast material at higher flow rates may improve the detection of hypervascularization of the esophageal mucosa or tumorous structures. Because of streak artifacts in the brachiocephalic veins, we recommend a caudocranial scanning Table
15.3
Patient
preparation
and enteric
Esophagus:
No oral contrast
Stomach
500-1000
Small
and duodenum:
bowel/colon:
Sellinklsmall
bowel
contrast
enema:
medium:
administration
medium
necessary
ml oral contrast
medium
1000-2000
ml oral contrast
up to 2000
ml of dilute barium
methylcellulose Rectal
contrast
Rectum: Colon:
Preparation
An empty and distended stomach is an essential prerequisite for a good examination. Thus, the patient should not eat for 12hours nor drink for 6 hours prior to the CT examination. Hydro-CT is based on optimum distention of the stomach with negative oral contrast medium (water, juice, or a methylcellulose preparation) in conjunction with intravenous contrast material. The patient drinks SOO-IOOOml of contrast medium shortly before the examination. While on the CT table, the patient should be asked to drink another 2 SOml of contrast medium (Table IS.3). A prone position has been recommended for lesions of the gastric body and antrum of the stomach. Scanning in the right lateral decubitus position may be advantageous for the pylorus and duodenal bulb, and scanning in the left lateral decubitus position for the gastric cardia. An antispasmodic agent (N-butylscopolamine, Buscopan®) may be administered for optimum distension of the stomach and relaxation of the gastric muscles. For virtual gastroscopy, gas is required as a negative contrast agent. The patient has to be examined early in the morning to reduce the amount of gastric secretions in the stomach. The same principles of patient preparation hold true as for conventional radiographic barium studies of the stomach. The stomach is distended by giving the patient a C02- producing powder and a little water. The patient then has to wait for some 2-3 minutes to make sure the powder is completely dissolved. CT scanning in both the prone and supine positions improves evaluation ofthe anterior and posterior portion of the gastric wall.
administered
or water;
antispasmodic
medium or by duodenal
intubation
500 III contrast medium or water 1000-2000 ml contrast medium or water
(Buscopan)
Examination Technique •
Acquisition Technique
CT examinations of the stomach and duodenum usually are performed as part of an upper abdominal survey that includes the liver, spleen, and kidneys, For staging examinations, the scan length is extended into the lesser pelvis to detect drop metastases, As a rule, scanning protocols are used that also permit optimum evaluation of the liver (see Chapter 11), Thin-section scanning with single-slice or multislice spiral CT improves the evaluation of horizontally oriented portions of the stomach (TableIfi.Z). With 4-slice scanners, a 4xlL25mm collimation is to be preferred if the patient is not too obese, In obese patients, the increased image noise requires use of a wider collimation (e,g" 4x2, 5 mm): With 16-slice scanners, even thinner sections may be acquired (SC= 0,5-0,75 mm). For image review, thicker sections (SW=4-6mm) should be reconstructed from the raw data or an overlapping secondary raw data set Multiplanar reformations (3-4 mm thick) that are aligned to the position of the stomach (semicoronal and sagittal) are highly recommended for optimum display of abnormalities, •
Contrast Injection
Intravenous contrast injection usually is performed to optimize evaluation of the liver because detection of liver metastases is a major issue for gastric malignancies, Since most gastric carcinomas cause hypovascularized metastases, the upper abdomen is scanned during the portal venous phase, During this phase, there is enhancement of the gastric wall as well as the perigastric vessels, which makes it easier to differentiate them from lymph nodes, During an earlier phase that corresponds to the late arterial phase in liver CT, however, there is superior differentiation of the various layers of thegastricwalL Small •
Intestine
Patient Preparation
Standard examinations require that a sufficient volume of oral contrast medium (10002000ml) is given to the patient over a period of at least 60 to 90 minutes before the examination, The patient should be handed portions of
500ml contrast medium every 15-20 minutes to more evenly distribute the contrast agent within the bowel. CT enteroclysis is a superior technique for indications that focus on the small bowel. The patient should take in ample fluid but no solid meal for 12 hours prior to the examination to cleanse the small bowel. Prior to CT a duodenal tube is placed distal to the duodenojejunal flexure, under fluoroscopic guidance, Then the patient is brought to the CT suite and a total volume of up to 2000 ml contrast medium is instilled at a rapid rate (80-120 mllmin) by a roller pump, The contrast material may consist ofa highly diluted barium suspension, a methylcellulose preparation, or a stearin-containing preparation (Table 15,3), We prefer negative contrast for most indications because it allows for excellent visualization of the bowel wall following intravenous contrast administration, Positive contrast material is better suited to demonstrate stenoses or fistulous tracts, The instillation of contrast material should be timed so that it just ends during or just after the CT scan. Acquisition Technique
Scanning of the small bowel is performed during the course of upper and lower abdominal examinations that include the lesser pelvis, Spiral and multislice CT with a thin-section protocol (Table 15,2) provide the best results and the same technique as for evaluation of the stomach can be used, With 8- and 16-slice scanners, however, 1-1.5 mm collimation generally suffices, Image review is based on 4-6 mm thick axial sections and 3-4 mm thick coronal reformations, Problem-adapted sections may further optimize the display of abnormalities, Contrast Injection
Intravenous contrast material injection relies on similar principles as for gastric examinations, If a hypovascular tumor is suspected, scanning in the portal venous phase is usually sufficient Hypervascular tumors, such as carcinoid tumors, require biphasic examinations, The best delineation of small bowel mucosa also occurs during the late arterial phase and early portal phase, which should preferably be used for the evaluation of inflammatory bowel disease,
75 The Castrointestinal
Tract
Colon Patient Preparation
For a standard abdominal examination that includes the colon, the administration of a large volume of positive oral contrast material (10002000ml) should be given to the patient over a period of at least 60 to 90 minutes prior to the examination, In patients with distal colon lesions, the rectal instillation of contrast medium (SOO-1000 ml) improves bowel distension and facilitates evaluation of the bowel wall, As in all enemas, care has to be taken to avoid perforation of the rectum, CT colonography is a focused examination of the colon, which requires the same patient preparation (bowel cleansing) as a barium enema (Table IS.3), The patient should eat no solid foods for two days, but drink copious amounts of liquids, Laxative medication using a "dry" preparation (e.g. Fleet) yields superior results over a "wet" preparation (e,g" Colite or Golytely), which is commonly used for endoscopy, Fecal tagging is possible if the patient drinks SO-80 ml of a diluted barium suspension with each meal (total of 200ml) on the day prior to the examination, This technique will help opacity stool and distinguish it from real polyps, Colon distension is achieved by rectally insufflated air or COs as a negative contrast material. Colonic distension is better with C02, and there is less colonic spasm and discomfort for the patient during and after the examination because C02 is absorbed the blood and eliminated via respiration, pumps that provide constant disention are now available, Due to the propensity of the colon to collapse, scanning should be performed in a prone and supine position, The filling state of the colon is checked on a low-dose scanogram before starting each spiral or multislice CT scan, If distention is not complete, antispasmodic agents and turning the patient (usually right lateral decubitus position) may improve colonic filling, Commonly used antispasmodic agents are O,S-LO ing glucagon or 20-40 mg N-butylscopolamine (e.g. Buscopan), which are injected subcUtaneously or intravenously, It is not yet clear, however, if they provide an advantage for every patient or whether they are better reserved for patients that experience colonic spasm during the exam.
Water or positive contrast material can also be used as an alternative for colonic distention but handling is more complex and time-consuming, Acquisition Technique
The colon is usually examined as part of an upper and lower abdominal examination, Scan coverage should include the region from the left diaphragm to the ischial tuberosities, It should not be limited to the pelvic region in rectal tumor examinations, since distant metastases to the liver could be missed, CT colonography yields best results using a thin-section protocol (Table IS,2), A 4x 1-1.2S mm collimation should be used for 4slice scanners whenever possible, Obese patients require use of a wider collimation (4 x 2-3.7S mm): With 8- and 16-slice scanners, 1-1.S mm collimation is sufficient even for CT colonography. Images of 4-6 mm thickness can be reconstructed for general colonic examinations but CT colonography requires thinner sections of 2-2.5 mm section width for optimum detection of small lesions and differentiation between polyps and stool, Coronal reformations (2-3 mm thick) should always be performed for dedicated examinations of the colon, Sagittal reformations are useful for the rectum, and further improve the detection rate for virtual colonoscopy, A low-dose technique (CTDlvoi=3-6mGy) with two scans in a prone and supine position without injection of intravenous contrast material is used for screening purposes, A wide window setting decreases the influence of image noise, For clinically indicated colon examinations (e,g" incomplete colonoscopy, cancer staging, or detection of secondary lesions in stenoses that cannot be passes with an endoscope), one scan (e.g. in prone position) can use a low-dose technique, while the second scan (e,g" in supine position) is performed with regular dose (6-12 mGy) and intravenous contrast administration. Contrast Injection
Intravenous contrast administration should be performed in a way that both the bowel and the liver can be adequately evaluated, For this reason, some 120-lS0 ml of contrast material are injected at 4mlls and scanning is initiated some IS-20 s after contrast arrival in the aorta just above the bifurcation, The scan is then per-
Examination Technique formed in a caudocranial direction in order to yield good bowel opacification and good portal phase images in the liver. Image Processing
Evaluation relies on axial sections and coronal multiplanar reformations, preferably using interactive display on a 3D workstation, In case of doubt, additional virtual endoscopic views may help differentiate polyps from thickened folds, There is data that suggests that virtual colonoscopy of the whole colon improves detection of small polyps but also holds the danger of an increased false positive rate, Other display modes (see Fig, 2,50) are also available but there is no consensus yet about which technique is best suited in clinical practice, Automated detection algorithms are being developed that in the future may direct the radiologist's attention to potential polyps,
Acute Abdomen Spiral CT is increasingly used for the diagnostic work-up of patients with acute abdominal pain, Tailoring the examination parameters to the Table 15.4
CT protocols
Suspected
in patients
diagnosis
with
acute
abdomen
suspected markedly
clinical diagnosis improves the diagnostic
Patient
Preparation
Oral contrast material is not usually necessary, especially with multislice scanning, Oral contrast material may improve the differentiation between inflammatory lesions (suspected appendicitis, diverticulitis, abscess), tumors, and collapsed small bowel loops, If possible, it should be administered over a period of at least 60 minutes prior to the scan to ensure adequate bowel opacification, thus making this procedure difficult to use in an emergency situation, In patients with suspected gastric or duodenal disease, pancreatitis, or intraductal gallstones, negative oral contrast (water) results in better distention and demarcation of the stomach and upper small boweL Oral contrast material is contraindicated in patients with suspected high-grade small bowel obstruction or ureter obstruction; it may obscure findings in patients with suspected gastric disease or gastrointestinal bleeding, In patients undergoing 3D CT angiography, oral contrast interferes with the volume-rendering procedure,
tailored
to the
Rectal contrast
Oral contrast
clinical
working
iv contrast'
Urolithiasis
obstruction
PVP
-/water -/water -/water 1000ml -/water
Aortic disease
Abdominal ischemia Venous
thrombosis
Abscess Pancreatitis Bowel
-/500ml
Volumetric
AP+PVP AP+PVP PVP PVP
-/water
Pyelonephritis/abscess Pelvic inflammatory
Standard Volumetric
-/500ml
-/water
Trauma
3
AP
phase,
b
phases in brackets see Table 15.2
Volumetric Volumetric
PVP
=
portal venous
may add information
NP(+DP)
-/500ml
phase,
bu.
NP
=
nephrographic
are not mandatory
Standard Volumetric
PVP phase,
DP
Standard Standard
PVP
lOO0ml
Enteric tumor
Standard
(PVP)
(PVP) NP (+ DP)
-/1000ml
disease
Volumetric
PVP (AP+)NP
-/500ml -/500ml
Diverticulitis
Volumetric
Standard Standard
PVP
Renal infarct Appendicitis
arterial
Standard Volumetric
AP
perforation
Biliary obstruction
=
Scan parameters"
Standard
Hematoma bowel
diagnosis
Volumetric
AP+PVP
Acute abdominal hemorrhage Small
(Table 15.4) accuracy,
=
delayed
phase
75 The Gastrointestinal
Tract
Positive rectal contrast (diluted iodinated contrast agents) can improve the evaluation of suspected diverticulitis and appendicitis. Acquisition Technique
Scanning parameters depend on the size of the suspected abnormalities. For the detection of hematoma. abscess. pancreatitis, renal infarction, or pyelonephritis, thicker sections (SW= S-8 mm) suffice. Thus, spiral scanning can rely on a SI10/S or 7/12/6 protocol. Multislice scanning with a 4x2.S-4x3.7Smm collimation has the advantage that thinner sections can be reconstructed if necessary. For the evaluation of small bowel disease, appendicitis, diverticulitis, urolithiasis, acute bleeding, or suspected perforation, thinner sections should be preferred if possible. With 1 s spiral CT scanners, however, the whole abdomen with a length of some 40cm can only be covered with a S mm collimation. For this reason, the scan range should be reduced wherever possible, and the reconstruction increment should be narrowed to improve z-axis resolution. With subsecond spiral CT or dual-slice scanners 2.S-3 mm collimation can be used for this purpose. For the evaluation of biliary and vascular disease, an even smaller collimation should be chosen, usually in the 3 mm range for
spiral CT. The 4-slice scanners can either employ 4 x 2-3 mm collimation for fast volume coverage or 4x 1-1.2S mm collimation for optimum spatial resolution. With 8- and 16-slice scanners, 1.2S-2 mm collimation suffice. Evaluation relies on interactive viewing of axial sections in a cine display.lfa perforation is suspected or intraperitoneal free air is to be excluded, the images should be viewed using a lung window setting as well, to identify air collections that are not associated with bowel loops. Interactive multiplanar may be used as a problem-solving tool. Volume-rendered displays are favored for vascular disease, and curved planar reformations help to better demonstrate findings in the bile ducts and the ureters. Contrast Injection
No contrast injection should be used for the evaluation of suspected urolithiasis and for suspected abdominal hemorrhage. Even for other indications, intravenous contrast injection is not always necessary although it can substantially improve evaluation if no oral or rectal contrast could be given. Early portal phase scanning provides a good compromise for most indications, even for patients with suspected active hemorrhage.
Esophagus Duplication Cysts and Diverticula
Duplication cysts of the esophagus are rare congenital anomalies (0.S-2.S% of all tumor-like lesions of the esophagus) that may be noted incidentally on a conventional chest radiograph as an indeterminate mediastinal mass and require further investigation by CTor MRI. Esopbageal diverticula may be of the pulsion or traction type. Whereas pulsion diverticula occur predominantly in the pre sphincteric region (cervical, epiphrenic), most traction diverticula occur at parabronchial sites. They are detected incidentally on CTscans. CT Morphology
Duplication cysts are smoothly
marginated, homogeneous masses with water-equivalent attenuation that most commonly occur in the
lower esophagus (60%). They are intimately related to the esophagus on multiple axial scans but rarely communicate with it. The cysts may have a paraesophageal (Fig. IS.2a) or intramural location. Infected cysts may incite a perifocal reaction in the mediastinal fat. The differential diagnosis of esophageal duplication cysts includes bronchogenic and pericardial cysts. Depending on the density of the cyst content they may simulate hematoma, neurofibroma, leiomyoma, or lipoma. Diverticula appear as air-, water- or contrast-filled masses that most commonly occur posteroinferior to the cricoid cartilage as Zenker diverticula (70%, Fig.lS. 2b). Parabronchial diverticula (20%) and epiphrenic diverticula (10%) are not usually visualized on axial CT scans but may be noted on coronal or sagittal reformations of spiral or multislice CT data sets.
Esophageal Carcinoma
Fig. 15.2 (a) Zenker's diverticulum at a typical location in the lower necklposterior superior mediastinum. There is an air-fluid level and the trachea is displaced anteriorly and to the opposite side. (b) Esophageal duplication cyst.
b
a
Benign Tumors
Benign tumors of the esophagus are rare « 1 %), and most are asymptomatic. In rare cases the tumors are manifested clinically by upper intestinal hemorrhage. More than 50% of benign esophageal masses are leiomyomas. Benign esophageal tumors in asymptomatic patients are an incidental finding at CT examination. •
CT Morphology
Scans demonstrate a smooth, well-defined thickening of the esophageal wall. Benign tumors usually appear as a nodular or eccentric mass that differs from the typical circumferential growth pattern of esophageal carcinoma (Fig. 15.3). Even so, benign tumors cannot be confidently distinguished from carcinoma by their CT features alone.
Fig. 15.3 Esophageal leiomyoma presenting as an eccentric mass in the esophageal wall, which can also be appreciated on virtual endoscopy.
Leiomyomas are most commonly found in the distal esophagus and may reach 2-8 cm in size. The eccentric masses, which are sometimes misdiagnosed as extrinsic lesions, tend to distort the esophageal lumen and obstruct the passage of food. Leiomyomas show diffuse contrast enhancement and are the only tumors that may contain calcifications. The less common adenomas are smoothly marginated intraluminal masses that show moderate contrast enhancement. Hemangiomas show intense enhancement. Lipomas are readily identified by their fat attenuation (HU < 0). Esophageal
Carcinoma
By far the most common malignant tumor of the esophagus is squamous cell carcinoma (95 %), followed by adenocarcinoma (4 %). The peak incidence is between 50 and 60 years of age, and men are predominantly affected. Sites of predilection are the three physiologic constrictions of the esophagus, with "15% of lesions occurring in the upper third, 50% in the middle third, and 35% in the lower third. Adenocarcinoma predominates in the distal esophagus and shows a high association with endobrachyesophagus (Barrett's esophagus). A continuum exists between this tumor and carcinoma of the gastric cardia. Most tumors spread by longitudinal extension along the submucosal plane and metastasize early to paraesophageal, mediastinal, cervical, perigastric, and celiac lymph nodes, aided by the lack of a serosa on the cervical and thoracic segments of the esophagus. The re-
553
75 The Castrointestinal
Tract
gional lymph nodes for the cervical esophagus are the cervical nodes, and those for the in- • CT Morphology trathoracic esophagus are the mediastinal and Esophageal carcinomas appear as eccentric or areas of wall thickening perigastric nodes. Affected celiac nodes are circumferential (> 5 mm) that usually involve a shorter segment classified as distant metastases. Hematogenous of the esophagus than inflammatory processes spread occurs to the liver (rare with squamous (Fig. 15.4). Postcontrast CT shows enhancement cell carcinoma, common with adenocarcinoma), of viable tumor tissue with improved delinealungs, adrenals, and bone. The role of CT in staging esophageal neo- tion of tumor boundaries. The gastroesophageal plasms has been controversial. Currently it is junction is a problem area in terms of diagnostic used more in treatment selection (curative ver- evaluation because focal wall thickening at the gastroesophageal junction can be normal, desus palliative) rather than to place the patient into a distinct staging level. In a neoadjuvant sit- pending on the degree of gastric distension, but may mimic neoplasm. uation with chemotherapy or radiation therapy An important criterion for transmural prior to surgery, the role of CT is the initial tumor spread is the obliteration of the paraassessment of the tumor bulk for (radiation) therapy planning and monitoring tumor re- esophageal fat plane, which may be difficult to appreciate in emaciated patients. Even in norsponse to tumor-reducing therapy. Distant mal individuals the fat plane between the metastases and tumor spread in contiguous structures (stage T3, Table 15.5) preclude cura- middle third of the esophagus and the aorta is frequently absent. A small triangle of fat, tive surgery. Thus, the primary task of preoperative CT (chest and upper abdomen) is to detect however, is usually preserved between the esophagus, aorta, and spine in nearly all intransmural tumor growth and distant dividuals. Encasement of 90° or more of the aormetastases (liver) because of the substantial surgical mortality that is associated with tic contour makes invasion highly probable, while less than 45° of contact implies that invaesophageal resection. Invasion of the tracheal bifurcation, bronchi, pericardium, and aorta are sion has probably not yet occurred. The range of critical importance from a surgical stand- between 45° and 90° is considered indetermipoint. nate. Tumor extension into surrounding tissues is definitely indicated by displacement and
a
b
Fig. 15.4 Esophageal carcinoma, (a) Esophageal dilatation with retained secretions and eccentric wall thickening caused by carcinoma. Note the impression on the left main bronchus, (b) Asymmetric, circumferential carcinoma in the distal third of the esophagus. The tumor is still separated from the pericardium by a thin fat plane but is in broad contact with a vertebral
c
d
body. (c) Coronal reformation curved to follow the course of the esophagus demonstrates the longitudinal extent of a middle-third esophageal carcinoma, (d) Recurrence of esophageal carcinoma following resection and gastric transposition. The aorta is partially encased by tumor.
Esophagitis Table 15.5 TNM
Staging
criteria for esophageal
carcinoma
Bt\~~"I3BHmB""8 Ci RBffi BEN mBBBlhEW
staging
•
w
T1
Tumor <5 cm in longitudinal extent with no circumferential wall involvement
T2
Tumor >5 cm in longitudinal extent with circumferential wall involvement or luminal obstruction
T3
Extension tissue
of tumor
T4
Invasion
of adjacent
into surrounding
N1
Regional lymph node involvement mediastinal, perigastric)
M1
Distant metastases; lymph nodes
fatty
structures
involvement
(cervical,
of celiac
compression of the airways with intraluminal convexity, fistulation to the tracheobronchial tree, and vertebral body erosion, The presence of a fistulous tract is indicated by the passage of orally administered air or contrast medium (do not use barium-containing agents) into the airways, mediastinum, or pleura, Periesophageal lymph nodes larger than 10mm, and infradiaphragmatic nodes of 8mm or larger are considered to indicate metastatic involvement (Table 15,5),
Esophageal Fistula Most esophageal fistulae in children are congenital, while fistulae in adults may be secondary to esophageal carcinoma, esophageal trauma, infections, or radtochemotherapy. Generally CT scanning is not indicated in children, but CT may be necessary in adults for localization and surgical planning, CT can also be used to detect pleuropulmonary or mediastinal inflammatory reactions to esophageal fistulae, • CT Morphology CT can demonstrate fistulous connections between the esophagus and the tracheobronchial system, pleura, pericardium, or mediastinal fat if the fistulous tract is of sufficient size and contains air or oral contrast medium, CT can also detect perifocal reactions in the form of pleuropneumonia, mediastinitis, or mediastinal abscess formation (see also Chapter 10),
Perforation of the Esophagus Perforation of the esophagus is an acute, lifethreatening event that requires prompt diagnosis, More than 50% of esophageal perforations result from iatrogenic manipulations (e,g" endoscopy, dilatation esophageal stenoses, or attempted intubation), Other causes are swallowed foreign bodies, blunt or penetrating chest trauma, caustic ingestion, and spontaneous perforation due to a sudden rise in intraluminal pressure, Esophageal diseases such as strictures, achalasia, and tumors predispose to perforation, Boerhaave syndrome is a special type of esophageal perforation involving a transmural injury of the distal esophageal wall approximately 2-3 cm above the esophagogastric junction, It is precipitated by extreme overeating and severe vomiting, CT Morphology
Esophageal injuries are manifested by wall thickening associated with periesophageal fluid, extra luminal air (pneumomediastinum, subcutaneous emphysema), pleural effusions, and possible contrast extravasation, CT also allows the visualization of very small collections of mediastinal air or contrast material in cases with small tears, It may also demonstrate penetrating foreign bodies that are only faintly opaque and therefore invisible on CI studies, Often the location of the esophageal perforation cannot be accurately defined by CT While distal perforations are usually associated with the development of a left-sided pleural effusion or a hydropneumothorax, a mid-esophageal perforation is more often characterized by a right-sided pleural effusion or pneumothorax, Thus, the side of the effusion or seropneumothorax can be critical in determining the route of the surgical approach,
Esophagitis Inflammatory thickening of the esophageal wall may be noted incidentally during the course of a CT staging examination or postoperative CT follow-up, It is not a primary indication for CT evaluation,
555
15 The Castro intestinal
Tract
CT Morphology
The inflamed esophageal mucosa shows uniform, circumferential wall thickening that usually involves a relatively long esophageal segment Eclematous wall segments appear hypoattenuating after intravenous contrast administration, while inflammatory infiltrates show substantial enhancement The transmural spread of an inflammatory process is uncommon, Intramural nodular densities, deep ulcerations, and fistulae, accompanied by mediastinal lymphaclenopathy, are seen in association with tuberculosis, Inflammatory and neoplastic wall changes cannot be reliably distinguished based on CT findings, Short segments of ulcerative wall thickening are more suggestive of a malignant lesion while longer segments are more consistent with an inflammatory process, Biopsy is usually necessary to establish the diagnosis,
Achalasia Achalasia is a motility disorder in which deglutition evokes incomplete relaxation of the lower esophageal sphincter, leading to dilatation of the esophagus, It is commonest from the third to sixth decades of life, The cardinal symptom is dysphagia, The primary diagnosis of achalasia is based on contrast esophagography or endoscopy, In rare cases (2- 7%), long-standing achalasia is associated with esophageal carcinoma, Cases that present unusual findings at esophagography or endoscopy should be further evaluated by CT CT Morphology
CT shows uniform dilatation affecting a long segment of the esophagus (4,Scm average diameter at the level of the carina) with no wall thickening and with normal-appearing boundary surfaces and mediastinal fat The esophagus narrows abruptly at the esophagogastric junction with no evidence of an intramural or extrinsic obstructive lesion, In contrast to a stricture, the esophageal wall is not thinned at the site of narrowing, and the wall is not thickened as it is with an esophageal tumor or esophagitis,
Strictures Stricture formation with obstructive narrowing of the esophagus may result from caustic ingestion, irradiation, trauma, surgery, or inflammation, The only role of CT is to exclude esophageal compression by an extrinsic mass. CT Morphology
Fibrosis and stricture formation are marked by luminal narrowing with thinning of the esophageal wall, In most cases the outer diameter of the esophagus is reduced over the site of the stricture, and the wall does not show abnormal enhancement It is common to find postinflammatory changes in the adjacent fat or at intrapulmonarysues.
Hiatal Hernia Hiatal hernia is the commonest form of diaphragmatic hernia, accounting for 90% of cases, The hernial opening is the esophageal hiatus, A distinction is drawn between an axial hernia (99%), in which the cardia is displaced into the thoracic cavity, and a paraesophageal hernia (1 %), in which portions of the stomach herniate alongside the esophagus, Mixed forms also occur. Hernias are an incidental finding at CT CT Morphology
With an axial hernia, portions of the stomach that emerge above the diaphragm may be misinterpreted as a retrocardiac mass (Ffg.lfi.Sa) but the demonstration of gastric folds and air in this mass is frequent and pathognomonic. Adequate gastrointestinal opacification is required in indeterminate cases, The diaphragmatic crura show more than IS mm of separation, The fat layer around the distal esophagus may be augmented by herniated portions of omentum, A paraesophageal hernia is associated with fixation of the gastric cardia, As a result, portions of the stomach herniate anterior to the esophagus and present as a contrast-filled intrathoracic mass, Upside-down stomach is an extreme form in which all of the stomach has herniated into the thoracic cavity (Fig, IS,Sb, c), and no portions of the stomach can be detected below the diaphragm, As the hernia enlarges the risk of life threatening complications such as incarceration, volvulus, strangulation, and infarction increases.
Dysp hagia/usoria
a
b
c
Fig. 15.5 Hiatal hernia, (a) Axial herniation of the stomach has resulted in a large retrocardiac mass. (b) Massive paraesophageal hernia (upside down stomach) with focal thickening of the wall due to esophageal
cancer. (4x1 /6). (c) Paraxial hernia with a gastric cancer in the fundus (4 x 1/6, coronal MPR). Note the retained contrast material in the proximal esophagus.
Esophageal Varices
structures that show marked pooling of intravenous contrast medium. Paraesophageal varices are often larger, and have a more serpiginous structure (see Fig. 10.13). Venous aneurysms are relatively frequent. While esophageal varices can be appreciated easily by endoscopy, paraesophageal varices are only seen on CT. In patients with portal hypertension, convoluted venous vessels can be seen in other areas of the upper abdomen that appear on the scan, but hepatic changes may be absent (depending on the etiology of the portal hypertension) . CT following uncomplicated sclerotherapy may show areas of esophageal wall thickening, scant mediastinal fluid, thickening of the diaphragmatic crura, pleural effusions, and small areas of pulmonary atelectasis.
Esophageal and paraesophageal varices are most commonly the result of portal hypertension. These" uphill" varices establish a collateral circulation from the portal vein through the azygos venous system to the superior vena cava. They occur predominantly in the lower half of the esophagus. "Downhill" varices are rare and may develop in the upper half of the esophagus when obstruction or occlusion of the superior vena cava causes blood to flow through the azygos system into the inferior vena cava and portal venous system. Esophageal varices may present as a mediastinal mass on chest radiographs and require CT for further evaluation. Often, however, they are an incidental finding at a thoracic or upper abdominal CT examination. •
CT Morphology
Esophageal varices have a variable CT appearance. Noncontrast scans show only sites of nonspecific esophageal wall thickening and nodular paraesophageal or retrocardiac masses. The enhancement of the varices after contrast injection requires a sufficiently long scan delay (at least portal venous phase), otherwise they may remain isoattenuating to the esophagus. Esophageal varices present as intraluminal (intramural, submucous) tubular, often dot-like
Dysphagia lusoria
Aberrations in the course of the aortic arch or supra-aortic branches can displace or compress the proximal esophagus, leading to dysphagia. The most frequent cause is an aberrant right subclavian artery that arises from the descending aorta and passes behind the esophagus. Other causes are a duplicated aortic arch or an aortic aneurysm.
557
15 The Gastrointestinal Tract
CT angiography (CTA) has already replaced intravenous and intra-arterial DSA in the evaluation of thoracic vascular anomalies and has become the diagnostic procedure of first choice (see Chapter 24, p, 871), Contrast-enhanced MR angiography has recently emerged as a competitive modality especially in young patients in whom, radiation exposure is an issue.
Postoperative Changes Gastric transposition and colon interposition are two reconstructive techniques that are commonly used to restore alimentary continuity after a partial or complete resection of the esophagus, The esophageal substitute may be placed within the esophageal bed or in a presternal or retrosternal position, CT has an essential role for the detection of recurrent esophageal carcinoma in the post esophagectomy patient The strength of CT is to identify an extraluminal mass in the mediastinum and to detect distant metastases, In patients with dysphagia, endoscopy and upper
Gl series may be more helpful in identifying a mucosal lesion or an anastomotic obstruction, but CT is superior in patients with general symptoms such as weight loss and abdominal pain, CT Morphology
Depending on its choice and placement, the esophageal substitute will generally show slight dilatation and a greater outer diameter than the original esophagus, When gastric transposition has been performed, part of the infradiaphragmatic stomach will be absent An interposed colon segment is recognized by its haustrations, Recurrent tumor presents as wall thickening, or as an intra- or extraluminal mass, usually at the site of anastomosis, Extraluminal mediastinal masses or encasement of the esophagus or esophageal substitute are most often present at the site of maximum tumor size before therapy (see Fig.l5.4d), Lymph node enlargement may be found in a paraesophageal or upper abdominal location but also in other site in the mediastinum, Pulmonary or hepatic metastases may be present
Stomach Cysts and Diverticula
Duplication cysts are congenital,
intramural masses of the stomach, Gastric lesions are lined by secretory epithelium and may enlarge over time, Most of these masses are detected at an early age, Diverticula of the stomach are very rare and arise in two locations, True gastric diverticula are more common and occur just below the gastroesophageal junction at the posterior portion of the lesser curvature, They contain all gastric layers and may become quite large (I-IOcm), An antral diverticulum is rarer and represents a focal invagination of the mucosa into the muscularis propria of the gastric wall, Antral diverticula are intramural and rarely larger than a few millimeters, They are almost never detected on CT
CT is indicated in patients with indeterminate obstructive thickening of the gastric or duodenal wall on endoscopy, CT Morphology
Duplication cysts appear
as intramural cystic areas of gastric or duodenal wall thickening with an associated mass effect Cysts on the greater curvature of the stomach may reach more than !Oem in size, The masses are oriented parallel to the gastrointestinal lumen, Generally the cysts are not opacified by oral contrast medium, because they rarely communicate with the lumen of the organ, Diverticula appear as air-, water- or contrast-filled masses located on the posterior wall of the gastric fundus,
Gastric Carcinoma
Benign Gastric Tumors
Benign tumors of the stomach may originate from any layer in the gastric wall. The commonest are adenomas (polypous adenomas), leiomyomas, lipomas, leiomyoblastoma, neurofibromas, and angiomas. Polypoid adenomas have a high risk of malignant transformation. Hyperplastic gastric polyps have no malignant potential; however patients are at increased risk for harboring separate, coexisting gastric carcinomas (between 8% and 28%). Leiomyoblastomas are uncommon smooth muscle tumors usually arising from the antrum, that can become quite large. Dependent on their size they have a tendency for malignant transformation with metastases to the liver or other structures occurring in 10% of patients. Metastases seldom occur with lesions smaller than 6cm. Approximately two-thirds of all leiomyomas occur in the stomach. A large subset of the leiomyomas and leiomyoblastomas that are positive for KIT (CD 117) are currently believed to belong to the group of gastrointestinal stromal tumors (GIST, pS66) There are only few cases (lipomas) in which CT can differentiate benign from malignant wall lesions, so a definitive diagnosis must rely on endoscopic biopsy. •
CT Morphology
Adenomas of sufficient size appear as intral uminal masses on CT scans. Leiomyomas are sharply circumscribed, uniformly enhancing masses with an average size of some Scm. They may contain irregular streaks or clumps of mottled calcifications and
are the most commonly calcified benign gastric tumors. They usually project into the gastric lumen, but occasionally may have an extragastric appearance. Contrast-enhancing normal mucosa usually covers the tumors (Fig.1S.6a) although ulcerations may occur. Neurofibromas are sharply circumscribed oval lesions within the gastric wall that are covered by normal contrast -enhancing mucosa. The tumors usually do not calcify and enhance less strongly than most leiomyomas (Fig. IS.6b). Lipomas range from 1-3 cm in size and are easily distinguished from other masses by their negative attenuation values «-80HU). Up to 7S% of lipomas are located in the antrum. They are endogastric lesions that arise from the submucosal fat in more than 90% of cases and grow towards the lumen. Large lesions develop superficial ulcerations that may bleed. Angiomas show intense enhancement after intravenous contrast administration. Gastric
Carcinoma
Adenocarcinoma (9S%) is one of the commonest tumors in humans with a high incidence in China, Japan, and East Asia, and the lowest incidence in North America and North Africa. There is a declining incidence of distal tumors but increasing number of tumors in the cardia and esophagogastric junction. It is most prevalent after SO years of age and predominantly affects men (2: 1). Most carcinomas infiltrate the gastric wall in the regions of the antrum or pylorus, along the lesser curvature, or in the gastric fundus
Fig. 15.6 Benign gastric tumors. (a) Leiomyoma of the anterior gastric wall, demonstrated by hydroCT in the prone position. The tumor appears as a well-circumscribed, intensely enhancing mass. Absence of infiltration of the gastric mucosa confirms the submucous location of the tumor. (b) Neurofibroma of the gastric wall with a similar morphology but less contrast enhancement.
a
b
15 The Gastrointestinal Tract near the carclia. Early lymphatic metastases occur to the lymph nodes of the lesser and greater curvatures and along the celiac trunk, the gastrocolic and gastrosplenic ligaments, and the greater omentum. Retroperitoneal (paraaortic) spread may occur along the thoracic duct to the left sLipraclavicular region (sentinel node). Affected retropancreatic, mesenteric, and para-aortic lymph nodes and affected nodes along the hepatoduodenal ligament are classified as distant metastases. Advanced disease is characterized by transm ural tumor growth with invasion of the colon, pancreas, or liver, and by peritoneal carcinomatosis with ascites (Table 15.6). Mucinous gastric carcinoma in women tends to seed drop metastases to the ovaries (Krukenberg tumors) or into the cul-de-sac. Hematogenous metastasis occurs to the liver, adrenals, lung, ovaries, bone, and brain. Tumor staging is performed histologically in a surgical specimen. CT is used to assess the presence and extent of transmural or extragastric spread of tumor, thereby assisting in the selection of surgical or conservative therapy. In patients with transm ural tumor extension with peritoneal spread and distant metastases (liver, lung), chemotherapy or palliative surgery are performed. If CT indicates a localized tumor, staging laparotomy and curative surgery will be attempted. CT is also used in "downstaging" to assess therapeutic response and evaluate the Table 15.6
Staging criteria for gastric carcinoma
T1
Tumor confined to the mucosa or submucosal
T2
Infiltration of the muscularis or serosa
T3
Penetration of the serosa
T4 a
Invasion of surrounding tissue
T4b
Invasion of other organs, diaphragm, or abdominal wall
N1
Lymph node involvement on greater or lesser curvature (up to 3 cm from primary tumor)
N2
Involvement of perigastric or celiac lymph nodes (>3 cm from primary tumor)
M1
Distant metastases (liver, lung, ovaries, adrenals, bone, brain), involvement of paraaortic, hepatoduodenal, mesenteric, or retropancreatic lymph nodes
secondary resectability of locally advanced tumors. With thin-section spiral and multislice CT, the accuracy of tumor staging has substantially improved, but its diagnostic role has not yet been established. •
CT Morphology
CT of the adequately fluid-distended stomach reveals focal areas of wall thickening (>4mm) that may be irregular or nodular, or intral uminal masses with asymmetric thickening of the gastric folds. Scirrhous carcinomas (linitis plastica) tend to cause concentric wall thickening with lack of distension of the affected segment (Fig.15.7a). The loss of a normal fold pattern is best detected when the stomach is distended with negative contrast medium (water). Differentiation of nodular tumor from hyperplastic gastric folds is required and improves with interactive cine display or multiplanar reformations. Larger tumors may obstruct the passage of contrast medium and cause prestenotic dilatation. Mucin- producing adenocarcinomas contain hypoattenuating central areas and may appear completely hypodense. They may contain stippled calcifications. Variable enhancementof gastric carcinomas is seen after intravenous contrast injection, with scirrhous carcinomas showing pronounced enhancement. Polypoid carcinomas present as soft tissue masses protruding into the lumen. Mural thickening due to gastric cancer ranges from 6mm to 4cm, the likelihood of transmural tumor-extension is directly correlated to the wall thickness and is quite common when the wall thickness exceeds 2cm. It can be difficult to assess transm ural tumor extension in emaciated patients due to the lack of conspicuous fat planes. Irregularities of the outer gastric margin, perigastric stranding, areas of increased structural density, and ill-defined boundary surfaces, especially in the perigastric fat, are suspicious for transmural tumor extension (stage T3, serosal involvement), although most of the stranding represents desmoplastic reaction. Irregular areas of increased attenuation in the mesenteric fat and omentum, sites of peritoneal thickening, and ascites with solid components are suggestive of peritoneal carcinomatosis in patients who have not had previous abdominal surgery. Proximal gastric cancer may cause direct invasion into the
Gastric Lymphoma
a
c
b
Fig.15.7 Gastric carcinomas (4x1/6, semicoronal MPR). (a) Scirrhous carcinoma with concentric gastric wall thickening and effacement of folds despite the slight degree of gastric distention. The inner layers of the gastric wall show appreciable enhancement, (b) T2
tumor of the gastric fundus with small, round and contrast-enhancing lymph node metastasis (arrow), (c) T4 tumorwith local infiltration of the transverse colon (wall thickening) and the pancreatic tail (arrows).
adjacent esophagus via the gastrophrenic ligament, the left lobe of the liver via the gastrohepatic ligament or the spleen via the gastrosplenic ligament. Distal gastric tumors can invade the neighboring duodenum, the pancreas, or the periaortic nodes, via the inferior aspect of the lesser omentum and the hepatoduodenal ligament, and into the transverse colon via the gastrocolic ligament. Metastases occur most frequently in lymph nodes along the gastrohepatic ligament and along the attachments of the omental folds to the great curvature. For the diagnosis of lymph node metastases, CT shows a sensitivity and a specificity of60-66%. Newer criteria for multislice CT (rounded shape with> S mm; contrast enhancement, no fatty center LIT ratio < 2 and short -axis diameter, see also Chapter 22) promise to improve the staging accuracy (Fig. IS.7b, c).
present. Low-grade MALT lymphoma has a favorable clinical course. Thoracic and abdominal CT is used for tumor staging.
Gastric Lymphoma
The stomach is the commonest site (SO%) of involvement by primary lymphoma or generalized lymphomatous disease in the gastrointestinal tract, with lymphomas accounting for just 3-S% of all gastric malignancies. Most of the tumors are histiocytic or lymphocytic non-Hodgkin lymphomas (90-9S%). MALT lymphomas arise from acquired mucosa-associated lymphatic tissue, with Helicobacter pylori invariably being
•
CT Morphology
Gastric lymphoma may infiltrate the entire stomach diffusely, or it may produce sharply circumscribed areas of pronounced polypoid or nodularwall thickening (Fig.18a, b). Ulcerations may be present (Fig. IS.8c). These areas do not contain necrotic foci and rarely show fistulation. Lymphomas tend to infiltrate the stomach homogeneously and show little contrast enhancement (Fig. IS.8). The mean thickness of the affected gastric wall is 4 to S cm despite extensive lymphomatous infiltration the stomach usually remains pliable and distensible without significant luminal narrowing (Fig.1S.8a). Advanced lesions may become very large. Only very rarely gastric lymphoma presents as cavitated lesion caused by tumor necrosis and excavation. Unlike gastric carcinoma, gastric lymphoma most commonly affects the greater curvature of the antrum and the body. Direct invasion of the pancreas, spleen, colon, and liver may occur. Transpyloric spread of lymphoma into the duodenum is quite common and represents a differentiating feature to adenocarcinoma that invades the duodenum in less than 10%. While splenic involvement is suggestive of lymphoma, hepatic metastases are characteristic of adenocarcinoma (Table IS.7).
561
562
75
The Gastrointestinal
Tract
Fig. 15.8 Gastric lymphoma. (a) Large MALT lymphoma with thickening of the gastric wall along the major curvature and in the antrum. Note the extra gastric tumor extent with encasement of the splenic artery, (b) Lymphoma of the lesser curvature with nodular submucosal tumor growth.
a
b
Table 15.7 Differentiating features between gastric adenocarcinoma CTfeatures
Gastric Lymphoma
Gastric carcinomc
Mural thickness
4cm (1-7 cm)
1.8cm (1-3cm)
Gastric wall
Thickened folds with smooth outer contours
Nodular outer contours + perigastric stranding
Inhomogeneous wall enhancement
uncommon
common
Extent
often diffuse
often focal
Spread into adjacent organs
40%
>70%
Lymphadenopathy below renal hilum without perigastric adenopathy
>40%
Other Malignant Gastric Tumors Gastric Sarcoma
Up to 3% of all gastric malignancies are leiomyosarcomas, which may occur at a young age. Patients present clinically with gastrointestinal obstruction and hemorrhage. Sarcomas tend to metastasize hematogenously to the liver, lung, peritoneum, omentum, and retroperitoneum. More than 90% involve the fundus and the body. Leiomyosarcomas may be difficult to distinguish from leiomyomas; malignancy correlates with size, and tumors larger than 6cm are almost always malignant. A specific subset of these sarcomas that express the c-kit oncogene (CD 117) are currently categorized as gastrointestinal stromal tumors (CIST, see p. 566). •
and lymphoma
CT Morphology
In patients with leiomyosarcoma, CT demonstrates lobulated, irregular areas of wall thickening or focal masses that show intense, nonho-
mogeneous enhancement and may contain areas of hypo attenuating necrosis and deep ulceration. Cavitation with air-fluid levels may be seen. Calcifications are occasionally present. Preferred sites are the anterior and posterior walls of the body of the stomach. The tumors may reach more than 12cm in size, resulting in a large extragastric component. Kaposi Sarcoma
Kaposi sarcoma develops in about 35% of patients with AIDS. Gastrointestinal lesions are seen in about 50% of them and are commoner in homosexuals than in drug abusers. Cutaneous involvement is common. •
CT Morphology
Kaposi lesions represent as submucosal masses of 0.5-3 cm diameter. They are markedly hypervascular. As these masses enlarge, they frequently undergo central necrosis or ulceration (target appearance).
Postoperative
Also polypoid masses or diffuse infiltrations are seen and may be indistinguishable from adenocarcinomas. Since the gastric findings alone are nonspecific. CT is helpful in assessing retroperitoneal lymphoma. splenomegaly or other evidence of Kaposi sarcoma suggesting the diagnosis.
Gastric
•
CT Morphology
Gastric metastases may appear as solitary submucosal masses or as diffuse infiltration. the latter being indistinguishable from a primary scirrhoLis carcinoma. Unenhanced CT attenuation values and enhancement characteristics depend on the primary tumor.
Gastritis,
Ulcer Disease
Endoscopy is used for the diagnosis and followup of gastritis and gastroduodenal ulcer disease. Generally CT is not indicated. but it does have selected indications such as the assessment of complicated cases prior to surgical intervention. •
Varices
Portal hypertension can lead to the portocaval shunting of blood through enlarged venous channels in the gastric wall. or varicose short gastric veins and the coronary vein of the stomach. Like esophageal varices. they are incidental findings at CT. CT Morphology
Metastasis
The stomach may be involved by metastases from malignant melanoma. carcinomas of the breast. lung. colon. or prostate. leukemia. or secondary lymphoma.
CT Morphology
Scans in severe gastritis demonstrate areas of gastric wall thickening that enhance after intravenous contrast administration. Nodular intramural densities suggest a tuberculous process. but CT cannot differentiate benign from malignant wall changes. Intramural air collections are found in complicated forms of bacterial gastritis {phlegmonous gastritis}. Scans in gastroduodenal ulcer disease may reveal local wall thickening that enhances with intravenous contrast. Due to the retroperitoneal position of the duodenal loop. its perforations are usually well contained and do not always present as a pneumoperitoneum (see Fig. 15.33a). Pockets of extraluminal gas may be seen in one or both anterior pararenal spaces or in the soft tissues of the upper abdomen.
Changes
Gastric varices appear as areas of gastric wall thickening (fundus. lesser curvature) with enhancing varicose veins. The varices can be traced as tubular wall segments on contiguous sections. They are best appreciated on MIP or volume rendered images if a negative contrast agent has been used.
Postoperative
Changes
Surgical procedures on the stomach and duodenum (gastrojejunostomy or esophagoje]unostomy) can radically alter the anatomy and physiology of the upper intestinal tract and can significantly affect the transit of food and contrast medium. Some reconstructive procedures (e.g .. a Billroth II reconstruction) may preclude the possibility of endoscopy. A complicated postoperative course or unexplained luminal compression requires evaluation by CT. •
,CT Morphology
A leaky anastomosis is manifested byextraluminal fluid collections with possible contrast extravasation or abscess formation. Contrast extravasation may be hard to distinguish from blind portions of small intestine that are frequently found after these procedures. In these cases. conventional radiological examinations or multislice CTwith multi planar reformations have to be performed. Postoperative luminal narrowing at the anastomotic site may have inflammatory or neoplastic causes (anastomotic recurrence). which are indistinguishable by CTif there is only a moderate mass effect. Given the frequent absence of oral contrast opacification in the afferent loops (e.g .. at the porta hepatis). these loops are easily mistaken for a mass lesion. lymph nodes. or an abscess with air inclusions. Such loops can be opacified by oral administration of biliary contrast agents (see Chapter 12).
563
75 The Gastrointestinal Tract Intestine Cysts and Diverticula
Duplication cysts are rarely seen in the duodenum. Diverticula in the duodenum are more common than in the jejunum and ileum. Diverticula of the small bowel are often located on the mesenteric side of the proximaljejunum. The majority of duodenal diverticula are acquired, asymptomatic diverticula but symptomaticinverted (in tral uminal) duodenaldiverticula in young adults (congenital lesions) may be responsible for unexplained pancreatic or biliary stasis. Asymptomatic colonic diverticula are much more frequent and become increasingly common with aging. Approximately two-thirds of all colonic diverticula involve the sigmoid colon. CT is indicated in cases of unexplained pancreatic or biliary stasis to exclude a tumor of the pancreatic head, or in patients with suspected diverticulitis (see p. 576). Duodenal or colonic diverticula often present as an incidental finding. •
CT Morphology
Duplication cysts appear as cystic areas within the duodenal wall and have an associated mass effect. Small bowel diverticula appear as air-, water- or contrast-filled masses located on the inner aspect of the duodenal C loop (Fig. 15.9) or less commonly on the mesenteric side of the proximaljejunum. Withjuxtapapillary diverticula, dilatation of the pancreatic duct and bile ducts may occur as an accompanying reaction. A
diverticulum containing retained food may simulate an abscess, although the wall of the diverticulum is thin and enhances like other bowel structures. Inverted diverticula present as filling defects in the duodenum and require good luminal opacification by oral contrast material for diagnosis. The differential diagnosis of indeterminate round or oval masses in the gastrointestinal tract should always include small bowel diverticula in addition to tumors. Retained stool in small bowel diverticula can be a cause for the "small bowel feces sign" that is due of abnormal bacterial overgrowth. Colon diverticula present as air-; water- or contrast -filled outpouchings of the colon wall and usually measure between 3 and 10mm in diameter. Chronic diverticular disease causes wall thickening (3-5 mm) in the affected colon segment. Previous inflammatory processes are characterized by residual densities in the pericolic or mesenteric fat. Benign Tumors
Tumors of the small bowel are very rare (4 % of all intestinal tumors), and approximately threefourths of all small bowel tumors are benign. Lipoma of the duodenum is quite common. The commonest benign tumor of the more distal small bowel is leiomyoma, which involves the
Fig. 15.9 Duodenal diverticulum with food remnants and air in a juxtapapillary location in a patient with NHLand retroperitoneal lymph node involvement (a). The normal duodenal lumen is marked by an asterisk. A volume rendered coronal image (4 x 1/6) in another patient demonstrates the typical location of such diverticula (b).
a
b
Malignant Small Bowel Tumors jejunum in 50% of cases. Adenomas most commonly occur in the ileocecal region, and hemangiomas are commoner in the jejunum. Neurofibromatous tumors occur in the setting of Recklingshausen disease. The commonest benign colonic tumors are adenoma's (polyps) and lipomas. The risk of malignant transformation in adenomas depends on the size and histology of the tumor (see Table 15.9). Polyps of more than lcm diameter and those with villous histological features are associated with a higher risk for malignant transformation. Lipomas are found predominantly in the right -sided colonic segments (cecum and ascending colon). Another common benign tumor is endometriosis, with 70% of intestinal deposits showing a perirectal or perisigmoidal location. Leiomyomas, fibromas, neurofibromas, hemangiomas, and lymphangiomas are rare benign colonic tumors. •
CT Morphology
Lipomas usually appear as well-defined round or oval intramural masses of fat density measuring 1 to 3 em in diameter. Lipomas of the horizontal portion of the duodenum (Fig. 15.10a, b) and of the ileocecal valve are relatively common. Lipomas may simulate extramural masses on endoscopy but can be readily identified by CT (see Fig.15.17b). Leiomyomas are moderately hypervascular masses or areas of bowel wall thickening that are prone to necrosis if their size is large. Hemangiomas appear as hypervascular, nodular areas in the bowel wall and may be
a
b
Fig. 15.10 Submucosal lipoma in the duodenum (a). A close relation to the papilla of Vater is seen on a coronal image (4 x 1/6) in another patient (b). Multiple in-
pedunculated. Endometriosis appears as an extramucosal tumor-like infiltrate. Adenomas of sufficient size may be demonstrated as pedunculated or broad-based intral uminal masses of soft-tissue density (Fig.15.1 Oc). In the small bowel, they are best detected by CT enteroclysis, and in the colon by CT colonography or virtual colonoscopy techniques (see p. 571). Malignant Small Bowel Tumors
Carcinomas of the small bowel are rare lesions that occur predominantly in the jejunum, while sarcomas are commonest in the ileum. Malignant tumors of the duodenum [adenocarcinomas, leiomyosarcomas, carcinoid tumors, lymphomas) are very rare. •
CT Morphology
Carcinomas of the small bowel most often involve the jejunum but may also be seen in the duodenum, or least frequently, the ileum. In the duodenum, they predominantly affect the posterior wall. Carcinomas present as areas of wall thickening or intraluminal masses (Fig. 15.11) that enhance after intravenous contrast administration. As opposed to small bowel lymphomas, carcinomas do not tend to cavitate with a communication to the bowel lumen but instead lead to focal small bowel stenoses or obstruction. As a result, signs of bowel dilatation or ileus may be present. The tumors grow
c traluminal pedunculated polyps in the small bowel in a patient with Cowden disease (multiple hamartoma syndrome) (c).
565
75 The Gastrointestinal Tract
a
b
Fig. 15.11 (a) Carcinoma of the small bowel. CT reveals wall thickening and subtotal stenosis of thejejunum by a nonhomogeneous mass, accompanied by borderline enlargement of mesenteric lymph nodes (arrow heads). (b) Hypervascular duodenal metastasis of a renal cell carcinoma.
locally invasive with stranding of the peritumoral fat as a sign of infiltration. Advanced duodenal tumors may invade the pancreatic head, obstructing the pancreatic duct and causing biliary stasis. These lesions may be indistinguishable from primary tumors of the pancreatic head. Leiomyosarcomas reach more than 6cm in size and appear as nodular masses with necrotic components that show heterogeneous enhancement and may contain calcifications. In most cases (>60%), the nodular mass is mainly located extrinsically. The mucosa may be stretched and ulcerated (50%), which may result in a fistula communicating between the bowel lumen and the necrotic tumor center. Metastases to the bowel are rare and are almost impossible to differentiate from primary tumors (Fig.15.11 b). Gastrointestinal Stromal Tumor
Castrointestinal stromal tumors (GISTs) are a heterogeneous group of tumors that represent the most common mesenchymal tumors in the gastointestinal tract. They are defined as KIT (CD n7)-positive spindle cell or epitheloid neoplasms and include most tumors previously designated as leiomyoma, leiomyoblastoma, and leiomyosarcoma. However, CISTs differ clinically and pathogenetically from true leiomyosar-
comas (very rare in the GI tract), from leiomyomas (common in esophagus and colon) and from neurogenic tumors like schwannoma. Although not fully understood, they are believed to originate in the plexus myentericus or in a pI uri potential stem-cell that can differentiate into Cajal and smooth muscle cells. Based on the immunohistochemical findings, four categories can be differentiated: tumorwith differentiation towards smooth muscle cells, towards neural elements (gastrointestinal autonomic nerve tumors, CANT), without a specific differentiation (uncommitted type), and with differentiation towards smooth muscle andneural elements. The histologic differentiation can be very difficul tor even impossible even with use of modern immunohistochemical methods. GISTs typically present in older (>50 years) individ uals and are most common in the stomach (60-70 %), followed by the small intestine (2025%), colon and rectum (5%), esophagus «5%) and mesentery and omentum « 5%). Theycanbe benign or malignant (in up to 30%) or have a borderline to low malignant potential. Malignant tumors frequently show metastases in the liver, peritoneum, lung, bone or lymph nodes. Therapy is primarily surgical and for nonresectable tumors very promising results with imatinib (ST1571, Glivec) chemotherapy has recently been reported. The most important prognostic factors are tumor size (or preferably tumor volume), tumor grade, tumor necrosis, and infiltration and metastasis to other sides. •
CT Morphology
GISTs may be submucosal, subserosal or intral uminal. They occur as solitary or m ul ti pIe lesions. Submucosal GISTs appear as round, smoothly defined filling defects while subserosal GISTs appear as extrinsic masses (Fig.15.12a), displacing adjacent bowel loops. Hypervascular intraluminal GIST may cause ulceration and secondary hemorrhage. Calcifications are rare. Benign GISTs are homogeneous with soft tissue density similar to muscle. Malignant GISTs appear as irregular lobulated masses with central liquefactive or cystic necrosis (Fig.15.12b), with or without a fluid-air level. They typically show a tumor diameter >5cm, intralesional necrosis, and infiltration ofthe surrounding structures. Vascular encasement may occur. A smooth contour, however, is also seen
Malignant Small Bowel Tumors Fig. 15.12 Gastrointestinal stroma cell tumor (GIST), (a) Subserosa I GIST of the stomach presenting as an extrinsic mass in the left upper quadrant. (b) Malignant GIST of the ileum with central necrosis.
b
a
in malignant lesions, and in some cases it may be impossible to distinguish benign from malignant types. The enhancement pattern in benign GIST is usually homogeneous, while in malignant types inhomogeneous peripheral enhancement is seen. Although liver metastases are most frequently hypodense, hypervascular types are not infrequent. Therefore, as with other sarcomas, metastatic disease detection in the liver is optimal when Diphasic CT is used. In HIV-positive patients, multiple hypervascular GISTs may simulate multiple Kaposi's sarcoma. Lymphoma, metastases, carcinoid, carcinoma or other mesenchymal tumors may show indistinguishable morphologic findings. The morphology following Clivec therapy has not yet been elucidated, but personal experience suggests that dimishing enhancement and increase in necrosis may precede actual tumor size reduction. Liver metastases may become purely cystic. Carcinoid
Carcinoid tumors are the commonest primary neoplasms of the small bowel (ileocecal region) and appendix. Only about 5% of intestinal carcinoids occur in the colon, and 17% involve the rectum. Carcinoids are neuroendocrine in origin (APUDoma=amino precursor uptake and decarboxylation, representing a cell line arising from islet of Langerhans). Men are affected twice as often as women. Patients are asymptomatic (60%) or suffer from pain (20%), obstruction, or weight loss (20%). Patients with liver metastases
may develop a carcinoid syndrome caused by excess serotonin levels bypassing the liver (diarrhea in 70 %, right heart failure clue to rightsided endocardial fibroelastosis and tricuspid regurgitation (35 %), asthma (15 %), skin lesions (5 %), cutaneous flushing « 5%). While small bowel and colonic carcinoids metastasize relatively early, regardless of their size, the occurrence of metastasis of rectal carcinoids is size-dependent. CT is used for nodal staging and for detecting or excluding distant metastases (lung, liver, bone). •
CT Morphology
The primary tumor can be very small and escape imaging, but the associated desmoplastic reaction and mostly bulky lymphadenopathy is readily seen. Calcifications occur in the majority of cases. In addition to hypervascular areas of bowel wall thickening or mass lesions, CT reveals retraction of the mesentery due to a desmoplastic reaction (Fig. 15.13). Adjacent bowel loops may be displaced, unusually separated, or sharply bent. Encasement of mesenteric vessels leads to chronic ischemia with segmental bowel thickening. Fibrotic retraction of mesenteric vessels may simulate hypervascularity. Venous occlusion may cause mesenteric varices. Affected mesenteric and retroperitoneal lymph nodes contain necrotic components of low attenuation. Advanced tumor stages may be associated with hepatic metastases (that may be near cystic, hypo-, iso-, or hyperattenuating after contrast injection, see Fig. 11.27), pulmonary nodules, and mostly osteoblastic skeletal metastases.
15 The Gastrointestinal
Tract Fig. 15.13 Carci noid. (a) Carcinoid tumor of the ileum with hypervascular nodular wall thickening. (b) Desmoplastic reaction leads to retraction of the mesentery.
a
b
Lymphoma
Lymphomas are the commonest primary small bowel tumors but are rare in the colon (approximately 1.5% of all abdominal lymphomas). The incidence of small bowel involvement by lymphoma is substantially increased in immunocompromised patients. Anatomically, the ileum is the most usual site of small bowel involvement. Multifocal involvement is occasionally seen. The most frequent site of occurrence in the large bowel is the cecum. N on- Hodgkin lymphoma can involve the gastrointestinal tract due either to secondary spread from nodal disease elsewhere or as a primary disease. Primary gastrointestinal lymphoma occur more frequently in the setting of inflammatory bowel disease, immunosuppression, sprue, and Helicobacter gastritis. CT is used in tumor staging to evaluate local tumor manifestations and lymphadenopathy. •
CT Morphology
Intestinal involvement by lymphoma may consist of a discrete mass or diffuse infiltration, resulting in pronounced eccentric or circumferential bowel wall thickening or a focal soft-tissue mass (average size Scm) that shows only slight enhancement (Fig. 15.14). The intestinal lumen may be narrowed as consequence of wall thickening. However, it can also show dilated caliber making it difficul t to differentiate from a small bowel obstruction. Large ulcerations and necrotic areas are also found that may be in continuity with the bowel lumen and thus contain contrast material. Polypoid lesions may act as the leading edge of intussusception, especially
Fig. 15.14 Intestinal lymphoma. CT shows segmental involvement of the small bowel by NHL with pronounced, asymmetric wall thickening and bizarre cavitation caused by complete destruction of the intestinal mucosa.
in children. Fistulae between adjacent loops may simulate Crohn's disease. Scans may also show pronounced regional lymph node enlargement along with mesenteric and retroperitoneal lymphadenopathy. The "sandwich" sign seen with involvement of mesenteric lymph nodes is caused by masses that encroach upon the mesenteric vessels from both sides. Malignant lymphomatous polyposis is a special condition with multiple (> 100) small polypoid lesions in the colon; frequently a large polyp or cecal mass is seen. Typically it is a low grade lymphoma most frequently occurring in men beyond 50 years. As a rule, lymphoma should be considered in the differential diagnosis ofvirtually all gastrointestinal tumors or inflammatory processes.
Colorectal Carcinoma Colorectal Carcinoma Colorectal carcinoma is the commonest malignant tumor of the gastrointestinal tract and the second commonest malignant tumor in men and women. The peak incidence is between 50 and 70 years of age. The predominant histological type is adenocarcinoma (70%). followed by mucfnous carcinoma (20%. known also as signetring or colloid carcinoma). and anaplastic carcinoma. Squamous cell carcinoma and adenoacanthoma are rare. Approximately 90% of all colorectal carcinomas arise from benign adenomatous polyps. Most carcinomas show an exophytic. polypous type of growth with frequent central degeneration. and they tend to infiltrate the bowel wall circumferentially. More than 50% of colorectal carcinomas occur in the rectum and sigmoid colon. Regional lymph node metastases of the colorectal carcinoma involve the pericolic and perirectal nodes and the nodal chains along the mesenteric vessels and internal iliac arteries on both sides. More advanced stages may also metastasize to the para-aortocaval nodes and mesenteric fat nodes. The most common sites of distant metastasis are the liver. lung. andadrenals. Potential complications of colorectal carcinoma include bowel obstruction. perforation. hemorrhage. fistula formation. and unilateral or bilateral hydronephrosis due to ureteral encasement. CT colonography is only indicated to detect additional lesions proximal to a stenotic primary that cannot be passed by an endoscope. From a surgical point of view. CT has no role yet for T-staging of patients with known colon cancer because most primary tumors are resected. even if metastases are present. Crosssectional imaging is indicated for the detection or exclusion of distant metastases (lung, liver) and to direct planning of radiotherapy in patients with locally advanced disease. With thin-section multislice CT, tumor staging becomes more accurate, and CT may regain a role in the pretherapeutic work-up, especially in the setting of neoadjuvant radio-chemotherapy. •
CT Morphology
CT demonstrates sites of focal or circular wall thickening or intramural masses that show enhancement with intravenous contrast. Markedly
hypoattenuating masses with hypoattenuating lymph nodes are characteristic of mucinous adenocarcinoma. Mucinous carcinoma tend to develop calcifications, both in the primary tumor as well as in distant metastases. Wall thickening distal to a large, fungating type of colonic carcinoma in the ascending colon may be due to edema or colitis. Tumor staging (Table 15.8) is problematic with CT but improves substantially with thinTable 15.8 Staging criteria forcolorectal
carcinoma
Dukes classification Dukes A: Dukes
B:
Dukes
C:
Dukes
Tumor
confined
Invasion Lymph
to the bowel wall
of the serosa
or mesenteric
fat
node metastases
Plus tumor bowel wall
growth
C1: Dukes
Plus tumor
extension
confined
to the
into fatty tissue
C2: Dukes D:
Distant metastases
TlIIIM staging
(UICC. 1991)
Tis
Carcinoma
T1
Tumor confined to mucosa
in situ confined to mucosa
T2
Tumor
T3
Transmural penetration into the subserosa or into non-peritonealized pericolic or perirectal structures
T4
Extension
confined
and submucosa
to muscularis
into retroperitoneum,
or adjacent organs (including distant bowel segments) N1
propria
-S3 positive
perirectal
mesocolon, infiltration
or pericolic
of
lymph
nodes N2
> positive nodes
M1
Distant
Tumor
perirectal
or pericolic
lymph
metastases
Stages
(UICC.
1997)
Stage 0
TIS,
NO,
MO
Stage I
rtz.
NO,
MO
Dukes A
Stage II
U4,
NO,
MO
Dukes B
MO
III
anyT,
Nl-2,
Stage IV
anyT,
any N,
Stage
Residual
tumor
classification
Dukes C
Ml after surgery
(R-classification) RO
no evidence
for residual
R1
microscopic
evidence
R2
macroscopic
evidence
tumor
for residual for residual
tumor tumor
569
75 The Castrointestinal
Tract
section multislice scanning. Even multislice CT is not able to differentiate between TI and T2 tumors (Fig.15 .15a). Tumor extension beyond the colon wall (T3) produces stranding (due to tumor and desmoplastic reaction), an indistinct boundary, and nodular protrusions in the pericolic fat (Figs.15.15b, 15.16). Absence of intervening fat planes, broad-based contact, and indistinct boundaries are suggestive of tumor extension to adjacent organs (T4). CT yields best results for staging of tumors that occur on the mesenteric or retroperitoneal wall of the colon. At the antimesenteric side of the colon, only a
a
b
Fig. 15.15 Colon carcinoma, (a) Small T2 rectal tumor (arrow) that is almost impossible to detect in standard CT despite rectal contrast administration. The small, round, contrast-enhancing lymph node (arrow) contained tumor at histology, (b) T3 rectal carcinoma
a
b
Fig. 15.16 -Recurrent colon carcinoma after right hemicolectomy. (a~ T3 tumor (arrow) withl hiqlu grade stenosis that could not.be.passed by an endoscope. (b) Virtual endoscopic view demonstrating the exophytic
very narrow rim of fat (1-2 mm) separates the colon from the peritoneal surface. Therefore, there is very little difference in infiltration depth from a T2 to a T4 tumor located at this side, and an ill-defined outer margin of such a tumor is already suggestive ofT 4 disease. There is a predictable pattern of lymph node drainage from the various sites of the colon. The mesocolic and the inferior mesenteric arterial nodes drain the left-sided colon, rectum and anus, while the right-sided colon and the cecum metastasize to the peripancreatic lymph nodes, simulating primary pancreatic
c with initial infiltration of the perirectal tissue (arrow heads) and round, contrast-enhancing pathologic lymph nodes (arrows), (c) Perforated colonic carcinoma with local abscess formation (arrow).
c nature or tne. lesion, (C)I Tfssue transition projection I resembling eft double-contrast barium enema. The tumor shows: a typicall apple' core configuration, ,andJthere 'is 31 second stenosing tumor at the ileo-colic anastomosis.
Colorectal Carcinoma 571 Table 15.9
Risk of malignancy in colonic polyps (Ferrucci 200_1)
_
Poly size
Likelihood
likelihood adenoma
If adenoma, likelihood cancer i1 IOyrs.
Overall likelihood
<Smm
<0.01%
30% 50% 80%
<5%
1% 2-5% 5-10%
5-9mm 10-15 mm
<1%
1-5%
cancer and may even result in biliary obstruction. Lymph nodes are recognized as nodular densities in the pericolic fat and are commonly classified as metastatic when larger than 1 cm in diameter (Fig.lS.lSb). With multislice CT, however. the three-dimensional shape oflymph nodes (rounded versus oval) may provide another clue to metastatic involvement (Fig. 15.15 a, b), but there are no hard data available yet that define new size and form criteria. Local complications of color ectal cancer include perforation, obstruction, intussusception or fistulation (Fig.1S.1Sc). Frequently, the complication causes more symptoms than the primary tumor itself.
Colon Cancer Screening
Colorectal cancer most frequently arises from a benignadenomatous polyp (in90%). Endoscopic polypectomy is an effective and minimally invasive technique for prevention of progression into a colonic cancer. Even if there is malignant transformation but no invasion of the submucosal. the cancer can be definitively cured. If a colonic cancer is detected at a later stage there is a marked drop in S-year survival rates (some 90% for stage I colon cancer, versus 60% for stage II cancer, and 40% for stage III cancer). For this reason, screening of a high-risk patient group for colonic polyps and colon cancer appears desirable. Tests for occult blood in stool samples are simple but lack reliability and yield a high rate of false positives if patient compliance is not optimum. Colonoscopy is highly reliable but it may fail to detect polyps hidden behind a haustral fold. Rates of incomplete colonoscopy in which the ileocecal valve has not been reached differ strongly between institutions but may be as high as 20%. Colonoscopy is especially difficult and painful for the patient if the colon is elongated and tortuous. In patients with (sub)total bowel obstruction, the proximal portion of the colon may not be evaluated at all.
5-10% 10-15%
cancer
in IOyrs.
In this setting of cancer screening, CT colonography and virtual colonoscopy offer a practical alternative to conventional screening techniques. The goal is to detect polyps of > 1Omm diameter that have an increased risk (still only between 1% and 6%) for being a colonic cancer (Table IS.9). Adequate patient preparation is an essential prerequisite, and low-dose scanning with rectally insufflated air or C02 yields a good display of the interior surface of the colon. While CT colonography, especially with C02, is well tolerated by the patient, the bowel preparation on the days prior to the examination is regarded with dislike similar to that for endoscopy. In patients at high risk for colon cancer (e.g., positive test for occult blood in stool), in whom endoscopy or barium enema could not be successfully completed, CT colonography can be used to detect a suspected colon tumor. In this setting, intravenous contrast administration and a higher exposure dose should be employed because a malignant tumor is a probable finding. There is a lesion type, however that can rarely be detected by CT. Such depressed or flat neoplasms account for less than 3% of the total number of colorectal neoplasms detected. Histologically they represent adenomas or carcinomas of less than 10 mm diameter with a high grade of dysplasia. Submucosal invasion is present in about 2S% even if they are smaller than S mm in diameter. The majority of cases is detected with specialized colonoscopy. CT Morphology
Colonic polyps present as rounded sessile or pedunculated lesions that protrude into the colonic lumen (Fig. IS.17). Larger polyps may be hard to delineate because they may almost completely fill the colonic lumen (Fig. IS.17c) and then have to be distinguished from a collapsed bowel loop. Scans in the prone and supine position have to be compared to detect lesions that might be hidden in collapsed segments. Pseudolesions may be caused by retained stool (air inclu-
572
15 The Castro intestinal Tract
a
b
c
Fig. 15.17 CT colonoqraphy. (a) Intraluminal polyp seen on virtual endoscopy (4 x 1/6) in a patient with colonic polyposis. (b) Large colonic lipoma with tiny polyp in the transverse colon not seen on initial colonoscopy. The lipoma is classified on the cross-sectional im-
ages, the polyp (arrowhead) is best detected by virtual endoscopy. (c) Large adenomateous polyp in the rectum that cannot be differentiated for a carcinoma in CT colonoqraphy.
sions are pathognomonic), thickened haustral folds, stool-filled diverticuli, or respiration artifacts (Fig. "15.18).An atypically appearing ileocecal valve may mimic a cecal tumor. Sessile polyps located on or close to a haustral fold may be difficult to distinguish from a thickened fold but differentiation is easily
possible on virtual colonoscopic images. This technique presently serves as a problem-solving tool, but optimum accuracy for detection of colonic polyps has been observed when multiplanar reformations were used in conjunction with virtual colonoscopy for the evaluation of the whole colon. Semiautomatic detection
a
b
Fig. 15.18 Pitfalls of CT colonoqraphy. (a) Thickened haustral fold mimicking a polyp, (b) Stool in the colon can mimic polyps but often has a characteristic appearance with hypodense gas inclusions and an acute angle
c between stool and colon wall. (c) Cross-sections images allow for correct identification of diverticula that may mimic polyps in CT colonoscopy and CT colonoqraphy (arrow). Note the stool-filled diverticulum (arrowhead).
Inflammatory and Infectious Bowel Diseases schemes are being developed that simplify polyp detection and cut down the time required for reading such a study. Even polyps smaller than 5 mm can be detected with thin-section multislice CT scanning but the number of false positive findings due to residues of fecal matter increases. From a clinical point of view, polyps smaller than 5 mm are not very important because the risk for a malignancy in polyps of this size is < 0.1 %. Flat adenomas impose another problem; they are raised less than 2mm from the colonic mucosal surface and are virtually impossible to detect by CT colonography. Tumor Recurrence
Tumor recurrence is commonest during the first two years after resection (up to 80% of cases) and usually occurs in the anastomotic region. T3 tumors with infiltration of the subserosa have a significantly higher risk for local recurrence. If previous proctectomy has been performed, CT or MRI are usually the only means for detecting a recurrence, in which case MRI is superior to CT in differentiating recurrent tumor from scar tissue. •
CT Morphology
CT during the early postoperative period (up to three months) usually cannot differentiate among inflammation, reparative processes, and tumor recurrence. Consequently, an initial follow-up CT scan should not be obtained until 4-6 months after surgery. Local recurrences are usually extraluminal, so any asymmetric soft-tissue density in the anastomotic region should raise immediate suspicion of a recurrence. Differentiation is required from abscesses, fistulae, and scar tissue (Fig.15 .19a), although this cannot always be accomplished. A recurrent tumor takes up contrast earlier and more strongly than scar tissue, which enhances most during the interstitial phase after 5 minutes (Fig. 15.19b) . Inflammatory
and Infectious Bowel
Diseases
Inflammatory reactions in the bowel initially lead to mucosal swelling and increased local perfusion. These findings are transient and re-
a
b
Fig. 15.19 Follow-up of rectal carcinoma, (a) Scar tissue following abdominoperineal proctectomy. The presacral soft-tissue mass has streaky margins and shows very little enhancement. The scar tissue has tethered the bladder posteriorly, retracting it into a sharply tapered shape, (b) Local recurrence after abdominoperineal proctectomy appears as a moderately enhancing nodular mass with ill-defined margins in the pararectal fat. Note the posterior distortion of the seminal vesicles (arrows) by the mass.
versible in acute catarrhal processes, and do not require CT evaluation. If clinical complaints persist, a CT examination may be useful for evaluating extraluminal extension, abscess formation, fistulation, and formation of stenoses or other complications. Inflammatory bowel diseases vary in their primary site of occurrence, pattern of involvement, clinical course, lymph node involvement, accompanying reactions, and complications (Tables 15.10, 15.11). Appendicitis
Appendicitis is attributed to lymphoid hyperplasia of the appendix (60%) and luminal obstruction by a fecalith (33%). A diagnosis based on clinical and laboratory parameters alone is relatively unreliable. About 20% of patients taken to surgery for appendicitis without preoperative imaging have had a normal appendix removed. Ultrasound and CT have become first -line imaging modalities for evaluation of suspected appendicitis. In pediatric patients and young females, however, ultrasound should be preferred. CT is then reserved for patients with
75 The Gastrointestinal Table 15.10
Inflammatory
Tract bowel wall changes
Disease
Site of occurrence
tJ
Diverticulitis
Sigmoid colon rest of colon
Diverticula,
Crohn's
disease
Any part ofthe gastrointestinal terminal
>
features circumferential
wall thickening
(often
> 10 cm); intramural air or contrast material; in pericolic fat; thickening offascial planes Segmental tract;
wall thickening
(upto
15 rnrn. mean
11 mm)
Acute: irregular,
ileum
stranding
enhancing
podense
outer contour;
homogeneous,
hy-
inner layers (mural stratification)
Chromc: irregular, nonenhancing outer wall contours Isolated right-sided colon involvement fibrofatty mesenteric proliferation abscess Ulcerative
colitis
Colon,
rectum
formation
Continuous, circumferential wall thickening mean 8mm) Abscess formation very rare
«10
rnrn.
Acute: smooth outer contour; irregular mucosa; stratification of bowel wall by submucosal target
pattern
with hyperdense
fat deposits;
inner and outer rings
Chromc: 'smooth
Yersinia
enteritis
Terminal
ileum
inner and outer contours
Segmental
with thickening;
normal-appearing
mesenteric
Tuberculosis
lIeocecalregion >ascending colon
Bowel wall thickening (20-4SHU)
Pseudomembranous enterocolitis
Whole
Diffuse circumferential enhancement
Ischemic colitis
Distal colon
Postirradiation
Ileum>
colon
colon,
changes
Extraintestinal
Related to disease to therapy directed Independent
of underlying
Due to inadequate or related
complications
activity and responsive at bowel disease bowel disease
intestinal
function
to long term therapy
wall thickening
ascites
with homogeneous
Symmetric, lobulated segmental wall thickening with irregular luminal narrowing; intramural and intravascular air inclusions rectum
Homogeneous
wall thickening
gins; target pattern; fascial planes
Table 15.11
fat
with lymphadenopathy,
of inflammatory
stranding
with irregular in pericolic
bowel disease Iritis Athritis Primary sclerosing cholangitis Ankylosing spondylitis Cholelithiasis Urolithiasis Hydronephrosis Fatty liver infiltration Liver abscess Pancreatitis Psoas abscess Osteonecrosis Osteomyelitis
bowel
mar-
fat; thickening
Inflammatory and Infectious Bowel Diseases
a
b
c
Fig. 15.20 Appendicitis, (a) Normal appendix, (b) Inflammatory changes in the appendix with a fecalith (arrowhead) and a perityphilitic abscess (arrow). (c)
Stranding and increased density of the periappendiceal fat in phlegmonous appendicitis (arrowhead). Note the terminal ileum (arrow) and cecum (C).
equivocal findings or in cases with negative ultrasound but strong clinical suspicion. CT was found to have an overall accuracy between 93% and 98%. In patients in whom CT is negative for appendicitis, CT is able to detect alternative diagnoses in 35-80%.
In appendicitis, the periappendiceal fat is typically clouded with linear stranding (Fig. 15.20). Focal fluid collections in the right lower quadrant are seen in up to 18%. A petityphlitic abscess or a phlegmon (ill-defined soft tissue mass with or without central hypodensity) occurs in up to 30% in adults and in up to 50% of pediatric patients. Small bowel obstruction is a complication more typical for the pediatric patient. Small extraluminal air bubbles (intraperitoneal or retroperitoneal) are seen in up to 30% of patients and are suggestive for a perforation. The cecum, terminal ileum, and sigmoid colon may also show wall thickening. Focal thickening of the cecum at the origin of the appendix is described as the arrow bead sign if the imaging plane shows the funnel-shaped edema of the cecal wall en face. In up to 60% of patients there are enlarged inflammatory pericecal lymph nodes. Scans may show intraluminal calcifications (in 25-50%) representing fecaliths. Such an appendicolith may be an incidental finding and does not in itself signify an inflammatory process. An appendicolith may cause obstructing appendicitis but is then associated with other signs of inflammation. Intraperitoneal appendicitis leads to circumscribed adhesion of the peritoneal leaves with associated peritoneal thickening and enhancement. With retroperitoneal appendicitis, the inflammatory process can spread along the psoas muscle.
•
CT Morphology
The normal appendix is a thin tubular structure that is closely related to the base of the cecum and is variable in its location. It may be collapsed, air-or fluid-filled. The presence of intraluminal air is strongly predictive ofa normal appendix and virtually excludes an inflammatory process (Fig.15.20a). The wall is normally very thin, and the diameter of the appendix measures less than 6 mm. The surrounding fat is clear without suggestive stranding or linear opacifications. It is important for correct localization of the appendix to trace it to its blind-ending tip, for which the cine mode of overlapping thin sections (3 mm width) is most helpful. The abnormal appendix is characterized by wall thickening (2-6 mm) and distention >6mm (mostly 1-1.5 em, rarely up to 4cm). On precontrast scans the wall may be denser than the intraluminal fluid. The strong enhancement of the wall after contrast injection is called target sign or ring sign. A double-ring sign (in 7%) is suggestive for a necrotizing appendicitis. Other findings in necrotizing appendicitis include focal wall thinning, focal loss of wall continuity and focal loss of wall enhancement.
15 The Gastrointestinal Tract False positive or false negative diagnoses are caused by mistaking small bowel loops for the appendix, by an unusual location of the appendix (a transverse or unusually high or low position of the cecum), by the presence of an appendolith without an appendicitis, or by the presence of cecal diverticulitis or ileitis, Problems also may occur with recurrent stump appendicitis or so-called tip appendicitis (8-20%), Frequent alternative diagnoses in patients with suspected appendicitis are mesenteric adenitis, diverticulitis, colitis, or ileitis, In young women, a right ovarian process or pelvic inflammatory disease must be considered,
Diverticulitis Asymptomatic colonic diverticula are a common incidental finding in older patients and do not have pathologic significance, Inflammation of the diverticula, approximately two-thirds of which occur in the sigmoid colon, leads to symptomatic diverticulitis, The inflammatory process usually spreads to surrounding tissues and may involve extensive portions of the bowel along with the peri- and retroperttoneum. The course of the disease may be complicated by confined perforations with abscess formation, hemorrhage, pneumoretroperitoneum, fistula formation, and postinflammatory stenosis, CT has become the primary imaging modality for evaluation of suspected diverticulitis, especially if complications (perforation, abscess formation) are suspected, Up to 33% of patients taken to surgery without preceding CT show no inflammatory process, The overall accuracy of CT in establishing or excluding the diagnosis ranges between 84% and 99%, In 50-60% of patients that are negative for diverticulitis, CT establishes an alternative diagnosis, In some cases it may be impossible to differentiate between a carcinoma and complicated diverticulitis based on CT morphology alone, CT demonstrates the extent of disease and helps with the decision between a conservative and surgical treatment
CT Morphology
The most frequent findings in patients with diverticulitis include the presence of diverticula (85-97%), wall thickening (70-94%), and fat stranding (98-100%), Intramural sinus tracts or fistulae are indicated by the presence of oral or rectal contrast material within the colon wall (up to 10%), Small intramural air inclusions are a frequent finding, Colonic wall thickening (3-5 mm in a distended colon) may be due to muscular wall hypertrophy and inflammation, Muscular hypertrophy is characterized by circumferential wall thickening with preserved haustral markings, Inflammation first leads to asymmetric thickening in the area of pericolic inflammation, and later causes circumferential thickening in advanced stages (Fig, 15,21 a), The inflammatory changes may cause narrowing of bowel segments with functional obstruction, Perforation or fusion of inflamed diverticula can lead to abscess formation, whichis demarcated by rim enhancement on postcontrast scans (Fig, 15,21 b) or a diffuse phlegmonous infiltration (20-50%), Depending on its location, the process may incite local peritoneal or retroperitoneal reactions with fluid collections and extraluminal air collections in the peritoneal space or the retroperitoneum. Intravesical air collections indicate the formation of a colovesical fistula, Spread of the inflammatory reactions to the ureters may lead to urinary obstruction, There is an overlap in the imaging appearance of colon cancer and diverticulitis, Pericolonic inflammation and segmental involvement greater than lOem were the most significant findings for diverticulitis; pericolonic lymph nodes and a luminal mass were the most significant findings for colon cancer. Fluid at the root of the mesentery, extra luminal fluid, and extraluminal air are additional findings that are highly suggestive, yet slightly less sensitive for diverticulitis, The presence of a shoulder formation (= overhanging edges), segmental involvement ofless than 5 em length, or the presence of an intraluminal mass are suggestive for a neoplasm, especially in conjunction with the presence of pericolonic lymph nodes, Using these criteria, an unequivocal and accurate diagnosis could be made, based on CT morphology, in about 50% of cases, Cases with less confident diagnosis need further work-up including endoscopy and biopsy,
Inflammatory and Infectious Bowel Diseases
b
a
c
Fig. 15.21 Diverticulitis. (a) Moderate inflammation with multiple sigmoid diverticula and increased density of the pericolic fat. (b) Wall thickening (>4mm). fat stranding, and diverticula in an Scm-long segment of the sigmoid colon, (c) Diverticulitis of the transverse
colon with abscess (arrowheads) in the greater omenturn (4 x 1/6, coronal MPR). Note the sharp delineation of the inflamed omentum to the normal mesentery and the contrast material remnants in a small diverticulum (arrowhead).
Tuberculosis
right lower quadrant adjacent to an abnormally thickened terminal ileum and cecum. Concomitant micro abscesses in the liver or spleen are caused by hematogenous dissemination.
Intestinal tuberculosis is a granulomatous inflammation that mainly affects young adults (age 20 to 40) and is usually acquired by the enterogenous route or, less commonly, by the hematogenous-lymphogenous route. Up to 40% of patients have concomitant pulmonary manifestations of the disease. The most frequent site of involvement in the gastrointestinal tract is the ileocecal region (80-90%), followed by the colon (usually the right side). Involvement of the stomach or other bowel segments is far less common. Ulcerations with enterocutaneous fistulae and perforations may complicate the course. •
CT Morphology
CT demonstrates bowel wall thickening in the ileocecal region or right colonic segments (segmental involvement) and enlarged lymph nodes (peripancreatic, mesenteric, retroperitoneal), which may show partial calcification or central hypoattenuation representing foci of caseating necrosis. Irregular masses of soft-tissue density may be seen in the omentum or mesentery. It is common to detect ascites with attenuation values between 20 and 45 HU. Intestinal adhesions can lead to small bowel obstruction. The typical pattern of Mycobacterium tuberculosis colitis in AIDS patients is a cluster of low density lymph nodes in the mesenteric root or
Crohn's
Disease (Regional
Enteritis)
Crohn's disease is a chronic, granulomatous bowel disease that can affect any portion of the gastrointestinal tract. The commonest site of primary involvement is the terminal ileum, followed by the proximal colon. The acute form of the disease is distinguished from a chronic stage. Frequent complications are fistulae, abscesses, adhesions, and stenosis leading to bowel obstruction. Perforations are uncommon, and if present, are usually contained. CT can define the extent of inflammatory bowel wall changes and detect extramural complications. CT enteroclysis is the preferred technique. •
CT Morphology
CT usually shows thickening of the wall of affected segments (up to 2cm), often associated with luminal narrowing (Fig. 15.22). Obstruction should only be diagnosed if the bowel segments proximal to the stenotic segments are dilated. Multiple noncontiguous bowel segments may be involved (skip lesions). The disease most frequently occurs in the terminal ileum but other portions of the small bowel, colon,
15 The Gastrointestinal
a
Tract
b
c
Fig. 15.22 Crohn's disease, (a) Long segments of the terminal loops of ileum show thickened walls with marked enhancement of the affected mucosa. There is an intervening thin-walled normal segment. (b) Fistulous tracts (arrows) connecting small (posterior) and large bowel loops (anterior). Note the marked mu-
cosal enhancement and bowel wall thickening in acute inflammation, (c) A long segment of the sigmoid colon shows wall thickening and fibro-fatty proliferation of the adjacent fat (arrowheads). Large abscesses with air inclusions mimic large bowel loops (arrows).
duodenum, stomach or esophagus may be also affected. The irregular outer contours of the bowel wall show enhancement during the acute stage but no longer enhance after the disease has progressed to a chronic stage. During the acute (noncicatrizing) phase, the bowel maintains its mural stratification with the target or double halo sign: the inner soft tissue ring corresponds to the mucosa, surrounded by a low density ring with attenuation near water (submucosal edema) or fat (fatty infil tra tion), which in turn is surrounded by a higher density ring of the muscularis propria (Fig. 15.23). The intensity of enhancement after application of contrast media was shown to correlate with disease activity. Mural thickening and liiminal obstruction may be reversible as long as it is due to inflammation and edema. The demonstration of mural stratification indicates that transmural fibrosis has not yet taken place and that medical therapy may be successful in normalizing the 1uminal reduction. With disease progression, mural stratification is lost and the affected bowel wall demonstrates homogeneous contrast enhancement. Thistransm uralinflamm ati onisaccompaniedby irreversible fibrosis, which responds less to antiinflamma tory therapy .Infl amm atoryandfi brotic alterations more frequently affect the mesenteric side of the bowel (Fig. 15.23 b).
Bowel loop separation seen on enteroclysis or upper GI series can be attributed by CT to an abscess, phlegmon, creeping fat, or fibrofatty proliferation. The latter is the most usual cause for bowel loop separation. The boundary between bowel wall and mesenteric fat becomes ill-defined. The attenuation value of the fat is substantially increased due to the influx of inflammatory cells and edema. Reactive hypervascularity of the mesenterium is called vascular jejunization of the ileum or the "comb" sign. Hypervascularity suggests active Crohn's disease and may be helpful for differentiating Crohn from lymphoma or metastases. Mesenteric lymph nodes range typically between 3 and 8mm in size; larger lymph nodes should suggest the diagnosis of lymphoma or carcinoma. Some 15 to 20% of patients with Crohn's disease develop abscess formations. Abscesses may be located between small bowel loops, in the ischi orectal fossa, the psoas m uscl e, or in solid abdominal organs. Abscesses usually result from fistulas, perforations, or surgery. A phlegmon is an ill-defined inflammatory process of the mesentery or omentum that will either resolve or progress to form an abscess. In CT, a phlegmon produces loss of definition of surrounding organs and a streaky infiltration of the mesenteric or omental fat. Fistulae can be detected as soft-tissue bands extending into the
Inflammatory and Infectious Bowel Diseases Fig. 15.23 Crohn's disease, (a) Coronal section with inflammatory wall thickening (arrowheads) and fistula (arrow), (b) Antimesenteric dilatation of a large bowel loop (arrow) and pseudopolyp formation seen on a tissue transition projection in longstanding Crohn's disease.
b
a
paraintestinal fat. They may be connected to an abscess, the skin surface, or another bowel loop (Figs.15.22b, 15.23a). They are better delineated if positive contrast material is given. Conglomerate masses will form if there is involvement of multiple bowel segments or a long segment together with fistulation, and abscess formation due to contained perforation. The differential diagnosis includes lymphoma, carcinoma with contained perforation, or other tumorous lesions. Infectious enterocolitis and inflammatory bowel disease may have identical morphologic features, and the ultimate diagnosis has to be based on histological and bacteriologic evaluation. The differentiation between Ctohn's disease and ulcerative colitis is important because of different therapeutic and prognostiC implications. • Mural stratification is seen in more than 60% of patients with ulcerative colitis, but in less than 10% of patients with Crohn's disease. • Mean colon wall thickness is approximately 8 mm with smooth outer contours in ulcerative colitis, while in Crohn's disease wall thickness is more pronounced and the outer contour is irregular. Yersinia enteritis also causes segmental bowel wall thickening with involvement of the terminal ileum and can be distinguished from Crohn's disease by an absence of concomitant reactive changes in the mesenteric fat.
Ulcerative Colitis
Ulcerative colitis is an inflammatory bowel disease that predominantly affects the colonic mucosa but may later extend to other layers of the bowel wall. Its etiology is not yet fully understood. The disease usually starts as a continuous, ascending acute inflammation of the rectal mucosa that may ultima tel y involve the entire colon (pancolitis). The most severe complication is toxic megacolon (in up to 5% of patients) with risk of perforation and peritonitis. Unlike Crohn's disease, ulcerative colitis is associated with little if any abscess or fistula formation, and lymphadenopathy is less pronounced. Chronic forms lead to scarring and rigidity of the colon wall with stenosis and polypoid islands of regenerating mucosa (pseudopolyps). The chronic irritation of the bowel wall promotes regenerative errors that may progress through severe dysplasia to malignant transformation. CT is indicated in complicated forms of ulcerative colitis with extramural involvement. It provides valuable information that supplements the findings of endoscopy and barium enema studies. •
CT Morphology
The subtle mucosal abnormalities that characterize the early stages of ulcerative colitis are at present beneath the spatial resolution CT. In these cases barium enema studies remain the most effective radiographic technique for making the diagnosis. With progressive disease, CT shows diffusely inflamed mucosa with continuous, circumferential wall thickening «Wmm) that predominantly affects the rec-
15 The Gastrointestinal
a
Tract
b
c
Fig. 15.24 Ulcerative colitis, (a) Extensive wall thickening in the rectum and sigmoid colon shows no skip areas and displays a smooth outer contour with slight mucosal irregularities, (b) Backwash ileitis with wide open
ileocecal valve (arrowheads), (c) Volume rendering of the colon in advanced ulcerative colitis with lack of haustration in the transverse colon.
turn but may involve the entire colon. Scans in the acute stage show mucosal irregularities 'accompanied by a smooth outer contour of the bowel wall. Mucosal ulceration can denude portions of the colonic wall, leading to inflammatory pseudopolyps. Noncontrast scans show a typical pattern of a three-layered bowel wall that is further accentuated after intravenous contrast administration (Fig. 15.24). When sectioned perpendicular to the bowel these mur al changes produce a target or halo appearance: the inner, contrast-enhancing ring of mucosa, lamina propria and hypertrophied muscularis mucosae is followed by a low density ring due to edema or fatty infiltration of the submucosa. The outer contrast -enhancing layer represents the muscularis mucosae. This m ural stratification is not specific and can also be seen in the acute phase Crohn's disease, ischemic disease and various types of enterocolitis (Table 15.10). Mural thickening in ulcerative colitis may cause narrowing of the rectal lumen. Mural stratification and thickening is accompanied by increased paracolic vascularity with only slight lymph node enlargement. As a consequence of inflammation there may be focal proliferation of the mesenteric or retroperitoneal fat. In particular, the typical widening of the presacral space is caused by proliferation of the perirectal fat (see Fig. 16.24). The perirectal fat then has an increased attenuation (by 10-20 HU) and contains an increased number of nodular and streaky soft
tissue densities as compared to normal extraperitoneal or mesenteric fat. Toxic megacolon is characterized by marked distention of the colon (>5.5cm in the transverse colon), mural thinning, pneumatosis and unsuspected perforation. The chronic stage is characterized by a loss of haustration (Fig.15.24c) and may present with stenotic areas with prestenotic dilatation. Destruction and incompetence of the ileocecal valve will cause backwash ileitis (Fig. 15.24b). Colon wall thickening usually is a sign of active inflammation. Marked focal wall thickening or larger polypoid lesions should raise the suspicion of malignant transformation with development of colon cancer. Pseudomembranous Enterocolitis
Pseudomembranous enterocolitis is a complication of antibiotic therapy caused by a cytotoxin produced by the bacterium Clostridium diffjcile. The clinical hallmarks are profuse watery diarrhea and abdominal cramps. Histologically the pseudomembranes are composed of necrotic debris, mucus, and inflammatory cells. CT may reveal an unsuspected pseudomembranous enterocolitis in patients on broad spectrum antibiotic therapy who present with sepsis and nonspecific abdominal complaints.
Inflammatory and Infectious Bowel Diseases •
CT Morphology
The imaging findings depend on the severity of the inflammation. CT demonstrates small-bowel or colonic wall segments with diffuse, circumferential wall thickening. Mostly the distal colon and rectum are involved, however, disease can also be limited to the right colon and transverse colon. A sign described as quite specific for pseudomembranous enterocolitis is the "accordion' sign, which represents the combination of broad transverse bands of closely spaced nodular haustra and thin columns of trapped contrast material. Hyperemia of the affected segments leads to homogeneous mucosal contrast enhancement (Fig. 15.25). Graft-versus-Host Disease
Graft-versus-host disease (GVHD) occurs when the immunocompetent graft reacts to the immune- incompetent host. GVHD of the bowel is frequently seen after bone marrow transplantation. The acute form predominately affects the small bowel, less frequently the colon also; the chronic form more frequently involves the skin. •
CT Morphology
The CT appearance is nonspecific with diffuse mucosal thickening and increased enhancement after intravenous contrast administration (Fig. 15.26). Thickening of the colon wall may mimic the accordion sign of pseudomembranous enterocolitis (Fig.15.26a) but the entities can be differentiated if small bowel is also
b
a
Fig. 15.26 Graft-versus-host disease with marked inflammatory thickening of large (a) and small bowel loops(b).
affected in GVHD (Fig.15.26b). Ascites is a frequent associated finding. Strictures may develop as long term sequelae. Whipple's Disease
Whipple's disease is an intestinal lipodystrophy of unclear etiology that is presumed to result from a streptococcal infection. Histologically Whipple's disease is characterized by an abnormal proliferation of macrophages in the intestinal submucosa of the proximal small bowel. •
CT Morphology
CT demonstrates thickening of the bowel walls of the duodenum and jejunum with preserved architecture of the folds and associated with marked, hypo dense lymph node enlargement (3-4 ern) in the region of the mesenteric root and retroperitoneum (Fig. 15.27). The affected loops of small bowel show little or no dilatation. Concomitant splenomegaly and ascites are frequent findings. The differential diagnosis of low density lymph nodes includes tuberculosis, AIDS, squamous cell carcinoma, ovarian and testicular carcinoma, and lymphoma. Postirradiation Changes
Fig. 15.25 Pseudomembranous enterocolitis with marked colonic edema and a hyperenhancing mucosa.
Radiation therapy exceeding 45 Gy can cause obliterative vasculitis in the intestinal wall and mesentery with consecutive radiation enteritis.
582
15 The Gastrointestinal
Tract
may simulate a neoplasm. Proliferation perirectal fat (> 10 mm) may occur.
of the
Small Bowel Obstruction (SBO)
Fig. 15.27 Whipple's disease, characterized by enlarged, hypoattenuating lymph nodes in the region of the mesenteric root and stranding in the mesenteric and retroperitoneal fat. The bowel loops show no dilatation or wall thickening.
Early complications include necrosis, ulceration, bleeding, and perforation with abscess'and fistula formation. Late complications may manifest clinically only after up to 15 years following radiotherapy, but permanent radiologic changes become apparent within one or two years. Women with gynecologic tumors and patients with bladder tumors are predisposed. The ileum is predominantly affected, and there are accompanying changes in the colon and rectum. Complications include bowel obstruction and fistula formation. Clinical hallmarks are cramping abdominal pain with persistent diarrhea and intestinal bleeding. CT is used to exclude extramural causes and in tumor patients for exclusion of recurrent disease. •
CT Morphology
In the early phase, the affected bowel segments show homogeneous wall thickening with consecutive luminal narrowing and irregular outer contours. The circumferentially thickened hypoattenuating submucosal layer may produce a target pattern in the bowel wall. Extramural changes consist of fibrotic strands in the perirectal or mesenteric fat along with thickened fascial planes and peritoneal adhesions. In the late phase, fibrosis lead to single or multiple minimally enhancing narrowed bowel segments with varying degrees of proximal dilatation. Fistulae, perforation, and abscess formation are rather rare complications. Onlyoccasionally, a shoulder formation at the strictures
Intestinal obstruction is a frequent problem affecting both, in-patients as well as out -patients and may account for up to 20% of hospital admissions. About 80% of obstructions are of small bowel origin. Amechanicalsmall bowel obstruction (SEO) or ileus is indicated by hyperactive high pitched bowel sounds. Prolonged obstruction, however, reduces peristalsis and may lead to atonic bowel and absence of bowel sounds. A history of previous surgery (5 days to many years earlier) favors the diagnosis of adhesions as the cause of obstruction. Other etiologies for a mechanical small bowel obstruction include inflammatory disease (Crohn's disease, post irradiation), tumors (metastases, carcinoid, primary bowel tumors), intussusception, or internal hernias (Table 15.12). The presence of continuous abdominal pain, signs of peritonitis, and a heart rate over 1l0beats/min are suggestive for strangulation of bowel segments. Strangulation implies interference with the blood supply associated with an obstruction that has not necessarily to be complete. There is a growing importance of CT for the diagnosis of SEO for a variety of reasons: 1) it does not rely on peristaltic propulsion and can
Table 15.12
Classification
of bowel obstruction
Mechanical bowel obstruction Strangulation:
Incarceration Volvulus Intussusception Closed loop
Obstruction:
Adhesionsladhesive Tumors Atresia Gallstone ileus Stenosis Duplication Coprostasis
Paralytic ileus Metabolic Reflex Toxic
bands
Smoll Bowel Obstruction (SBO)
be used in cases of complete obstruction or paralytic ileus, 2) it provides global assessment of the abdomen, 3) it does not require oral administration of contrast media, and 4) it provides ready assessment of strangulation. CT has become the diagnostic method of choice to demonstrate or exclude strangulation, which would require immediate surgical intervention. When CT fails to provide a definite diagnosis, enteroclysis may be performed for a more detailed analysis. In patients with low grade or intermittent obstruction, (CT) enteroclysis remains the primary diagnostic modality. Mechanical Bowel Obstruction
A mechanical bowel obstruction is classified as acute, subacute, chronic, or chronic recurring, depending on the time course of the obstructive process. It is further classified by severity, as partial or complete obstruction, and by the anatomic level, as high or low small bowel obstruction or as colonic obstruction. CT is only indicated in patients with mechanical obstruction. •
CT Morphology
In mechanical bowel obstruction, dilated bowel loops with gas- fluid levels are seen proximal to the obstruction. The site of obstruction is characterized by an abrupt caliber change. CT usually provides more detailed information on the cause and location ofthe obstruction. Ifperforation occurs, extraluminal air is seen in the abdominal cavity and ascites is sometimes observed. As opposed to strangulated obstruction with interference of the blood supply (see below), a simple obstruction usually shows no mesenteric changes such as vascular engorgement, there is none, or only minimal haziness and none, or only minimal ascites. Localized peritoneal fluid accumulations should not be misinterpreted as mesenteric haziness. A non-obstructive paralytic ileus is characterized by uniformly dilated loops of small and large bowel with gas-fluid levels and no demonstrable lesion causing the obstruction (Fig. 15.28). The key finding of a mechanical obstruction is the so-called transition zone of obstruction. A transitions zone has three components. There is a proximal dilated segment, that abruptly terminates with a straight or slightly curved demarca-
Fig. 15.28 Paralytic ileus in a patient with retroperitoneal rupture of an abdominal aortic aneurysm after stent graft placement. Note the dilated fluid-filled small bowel loops with an air-fluid interface (arrowheads) but normal mesenteric vessels and lack of bowel wall thickening.
tion. The distal bowel segment has a normal caliber, is collapsed, or of reduced diameter. Between these two components, there is a short, sharply marginated segment that has a compressed lumen (Fig.15.29a). Depending on the underlying disease (adhesions, postinflammatory stricture, tumors), there is a normal wall, a thickened wall, or a mass in or close to the narrowed segment. The "small bowel feces sign" is characterized by gas-inclusions in small bowel loops that resemble colonic contents. The sign has. a sensitivity of around 80% but should be used with caution if it is the only sign present. Obstruction by an extraluminal tumor (Figs. 15.29b), or by a broad adhesive band, produces a less abrupt beak-shaped transition zone. Metastases may cause a more high grade obstruction because of associated desmoplasia, while primary tumors are characterized by lower grade of obstruction without desmoplastic change. The diagnosis of metastases as opposed to a primary tumor is further supported by evidence of additional metastases elsewhere in the abdomen. Carcinoid tumors cause intraluminal filling defects and typically rather obvious desmoplastic changes in the mesenteric fat that retract and displace bowel loops. Adhesions can be produced by extrinsic inflammatory or postoperative changes or are due to intrinsic inflammatory processes (e.g., Crohn's disease). The latter can be differentiated by the wall thickness itself.
584
75 The Gastrointestinal Tract
Fig. 15.29 Mechanical obstruction, (a) Simple obstruction with dilatation of bowel loops but regular wall enhancement. Note the transition point from enlarged
to normal-sized bowel loops (arrow), (b) Obstructing cecal carcinoma with proximal bowel dilatation.
Strangulation/Ischemic Bowel Disease
Interference with the blood supply may occur either from twisting of the bowel on the mesentery or from distention of a closed loop. Obstruction and strangulation are not necessarily combined, but the more severe the obstruction, the higher is the incidence of strangulation. The longer the involved bowel segment, the more severe are the clinical complications, ranging from perforation and peritonitis to shock and death. Pathophysiologically there are different stages of strangulation that can also be differentiated in CT. Initially, the venous return is compromised but arterial infl ux into the closed loop
Strangulation implies an interference of the blood supply associated with an obstruction of the bowel lumen that is not necessarily complete. CT was found to have a high sensitivity (above 90%), but rather low specificity (about 45%) for demonstrating intestinal ischemia. Strangulation occurs as a complication of intussusception, torsion, volvulus or any type of closed loop obstruction. Mesenteric vascular occlusion is also included in the strangulation category.
Table 15.13
Differentiation
of simple and strangulated
CT features
bowel obstruction
Strangulation
(modified from Ha et al., 2000) Simple Obstruction
Bowel wall changes Beak sign Target sign Wall thickness Enhancement
serrated » smooth frequent mean 5 mm, frequently thicker poorer none
normal
moderately frequent moderately frequent frequent moderately
frequent
frequent very rare moderately frequent very rare very rare very rare
frequent frequent
frequent very rare moderately frequent
smooth
rare
mean 3.5 mm, rarely thicker
Mesenteric changes Focal vascular engorgement Diffuse vascular engorgement Focal mesenteric haziness Diffuse mesenteric haziness Unusual vascular course Mesenteric thrombosis
frequent frequent
Ascites Small Large High attenuation
frequent moderately moderately
Small Bowel Obstruction (SBO)
continues, leading to increasing intravascular pressure and vessel engorgement. Hemorrhage into the bowel lumen and the bowel wall occurs. Eventually perfusion ceases in the bowel wall as result of arterial spasm due to increasing pressure from interstitial hemorrhage, anoxia, edema and capillary stasis. •
CT Morphology
CT scans demonstrate a moderately dilated, fluid filled bowel segment that has a radial or Y-shaped configuration (Table 15.13). Converging and engorged edematous vascular bundles represent congested mesenteric vessels (Fig. 15.30a). Venous strangulation of the bowel wall is suggested when CT shows circumferential thickening of the bowel wall with increased attenuation, hazy mesenteric vessels, and increased enhancement after intravenous contrast administration (target sign). Arterial strangulation may be assumed when there is delayed, reduced, or no contrast opacification of the bowel wall within the closed loop. The latter is frequently associated with the presence of intramural or extraluminal air. The poorly perfused strangulated segments show a stronger delayed enhancement in the interstitial phase 10-15 minutes after completion of spiral CT. The strangulated segments may thus appear hyperor hypo attenuating relative to the adjacent bowel loops, depending on the scan phase. Volume rendering of the superior mesenteric
a
b
Fig. 15.30 Volvulus, (a) A whorled pattern is created by the bowel loops twisted around the fixed mesentery. The torsion has led to strangulation (absence of bowel enhancement), mesenteric edema (arrow), and small amounts of free fluids around the affected bowel segment. (b, c) In another patient, the cecum had rotated
vascular course can be helpful to diagnose the spiral course of twisted branches of the SMA or SMV (Figs.l5.30b, C). Coexisting vascular engorgement can be appreciated as thickened vessels distal to the twisting point. A serrated beak sign due to thickened mucosal folds is suggestive of strangulated obstruction as opposed to a smooth beak seen in patients with simple obstruction. Volvulus, Closed Bowel Loop
A volvulus is a malrotation or axial twisting of portions of the small or large bowel associated with a long mesentery. It usually occurs in the elderly patient. The commonest form is a colonic volvulus, which can lead to bowel obstruction. A closed loop obstruction refers to entrapment of a bowel loop by a contracting adhesive band. A similar situation is seen in patients with an incarcerated internal hernia. All three situations may lead to symptoms of strangulation, dependent on the interference of the vascular supply. •
CT Morphology
The colonic volvulus can affect the cecum or the sigmoid. CT demonstrates a whirled pattern caused by the torsion of the bowel segments (afferent and efferent loops) around the fixed and twisted mesentery. In patients with a rotated cecum, the distended cecum is usually located
c around the terminal ileum and is displaced upwardly, resulting in a typical whirl signs (b). The volume rendered display of the vessels (c ) demonstrates venous stenosis (wide arrow) and engorgement of the right colic veins (ReV).
15 The Castrointestinal Tract in the left upper quadrant. A twisted sigmoid leads to greatly distended paralyzed loops affecting both the small and large bowel. The whirl sign, convergence of mesenteric vessels toward the twisted site or a reversed position of the mesenteric artery and vein, are CT signs indicative of a volvulus or a closed bowel loop. However, the whirl sign is also seen in asymptomatic subjects (approximately 7% of asymptomatic patients), especially at the anatomic level just below the inferior pole of the right kidney. This sign can also be seen in patients after previous abdominal surgical (e.g., gastrectomy). Therefore it is necessary to combine the presence of the whirl sign with a suggestive abdominal symptoms or other morphological findings such as dilated bowel loops. In closed loop obstruction demonstrates a markedly dilated segment of (usually small) bowel with two transition zones. Vascular engorgement and other signs of strangulation may be present (Fig.15 .30a). Intussusception
Intussusception is defined as invagination or prolapse of an intestinal segment (intussusceptum) into the lumen of the adjacent intestine (intussuscipiens). It is much more common in pediatric patients (94%) than in adults (6%). Intussusception is the commonest abdominal emergency of early childhood (75% in patients <2 years) and the leading cause of acquired bowel obstruction in childhood. The etiology is mostly idiopathic with the predominant location at the ileocecal valve. In less than 5% of patients there is a lead point (e.g., Meckel's diverticulum, lymphosarcoma, polyp, granuloma, appendicitis) mostly occurring in patients> 6 years. In adults, more than 80% of patients have a specific cause (tumor, lipoma, Meckel's diverticulum, adhesions, prolapsed mucosa, chronic ulcer) serving as an anatomic lead point. Possible complications are hemorrhage, bowel ischemia, and the development of bowel obstruction. CT is only indicated in adults and in pediatric patients with complicated forms or suspected extrinsic compression. •
CT Morphology
A section (axial or multiplanar) perpendicular to the intussusception demonstrates three concen-
Fig. 15.31 Intussusception. CT shows typical stratification of the bowel wall in longitudinal section and concentric rings in cross-section (from outside to inside: intussuscipienswith air, mesenteric fat, and intussusceptum).
trie rings formed 1) by the canal and wall of the intussusceptum (inner layer), 2) the mesenteric fat (middle layer), and 3) the intussuscipiens as the outer layer (Fig. 15.31). Proximal to the intussusception there may be an obstruction with dilated bowel loops. Rarely is it possible to detect a tumor, polyp or metastasis as the cause of an intussusception because it is hidden within the collapsed and thickened inner layer. Abdominal Herniation
Abdominal hernias are a common clinical problem and asymptomatic hernias are a frequent incidental CT finding. External or abdominal wall hernias are differentiated from diaphragmatic and internal hernias. Approximately 95% of intestinal hernias are external, and only about 5% are internal. Diaphragmatic Hernias
Hiatal hernias are described on p. 556. Bochdaiek hernias are located posterolaterally and are more frequent (9: 1) than the anteromedially located Morgagni hernias. Bochdaiek hernias are usually congenital and caused by an incomplete closure of the pleuroperitoneal fold by the ninth week of gestation; they are rarely the result of surgery or trauma. They are more frequently located on the left than on the right side (5 :1). They are found in small children and
Abdominal Herniation can become rather large. Morgagni hernias are due to an anteromedial diaphragmatic defect caused by maldevelopment of the transverse septum. Morgagni hernias are more frequent on the right than on the left (5: 1). They are seen in older children or adults and are usually small. •
CT Morphology
Diaphragmatic hernias are quite frequent. A sliding hiatal hernia is seen in 10% of adults. When the hernial defect enlarges, other structures such as duodenum, colon, pancreas, or the whole stomach may protrude in to the thoracic space (see Fig. 15.4). In paraesophageal hernias, the stomach protrudes partially or completely through the esophageal hiatus with the cardia remainingin place. Posterolateral Bochdaiek hernias are common: a defect in the posterolateral portion of the diaphragm can be seen quite frequently. Dependent on the size of the posterolateral diaphragmatic defect, bowel, spleen, left hepatic lobe, kidney and pancreas may partially herniate. On the right usually parts of the liver herniate. Morgagni hernias are located anterior to the liver close to the midline. Herniated structures consist of bowel, omentum, stomach or spleen. External Hernias
Sites of predilection for external hernias are the diaphragm, the abdominal wall, scars, the umbili-
a
b
Fig. 15.32 Abdominal hernias, (a) Incarcerated abdominal wall hernia with decreased wall enhancement and dilated small bowel loops. Note the unrelated psoas hematoma. (b) Scrotal hernia containing portions
cal and femoral region, the inguinal canal, the pelvic floor, and the back muscles. CTimagingmaybe appropriate in complicated hernias or in patients who have been operated on previously. •
CT Morphology
External abdominal wall hernias are classified according to their location. CT allows for visualization of the wall defect, the contents of the hernial sac, and potential complications. Hernias that involve only a part of the bowel wall are called Richter hernias; obstruction and incarceration are rare in these cases. Even when no bowel loops are herniated, CT will demonstrate the abdominal wall defect. Ventral hernias protrude through the linea alba and are divided into epigastric (superior to the umbilicus) and hypogastric (inferior to the umbilicus). Spigelian hernias represent a protrusion of short segments of the large or small bowel or the omentum between layers of the abdominal wall muscles (Fig.15.32a). Sncisional hernias occur in up to 5% of patients after surgery, typically in the first postoperative year. Areas for surgical scarring should be scanned while the patient performs a Valsalva maneuver. Lumbar hernias may contain bowel loops, retroperitoneal fat, or the kidneys. They protrude dorsolaterally into the superior and inferior lumbar spaces through weak soft tissue triangles between the wall muscles, the 12th rib, and the iliac crest. Indirect inguinal hernias are the commonest type of abdominal wall hernias.
c of the sigmoid colon (tissue transition projection from a CT colonoqraphy data set), (c) Large inginual hernia in a female patient that contains large and small bowel loops.
581
15 The Castrointestinal
Tract
Small bowel loops protrude through the inguinal canal and emerge at the external inguinal ring. In men the hernia can extent into the scrotum (Fig.15.32c). in women into the labia major (Fig.15.32b). The hernia is typically located lateral to the inferior epigastric vessels. Preoperative determination of the contents of the hernial sac is essential. Direct inguinal hernias protrude directly through the lower abdominal wall and are located medially to the inferior epigastric vessels. In femoral hernias, pre peritoneal fat, omentum, or small bowel loops protrude below the inguinal ligament into the femoral canal. Obturator, sciatic, or perineal hernias occur only very rarely.
through the foramen of Winslow into the lesser sac now lying posteriorly to the stomach that is displaced anteriorly. Herniated structures are absent from their usual location. Pericecal hernias usually involve an ileal loop that herniates through a defect in the cecal mesentery and is lies lateral to the cecum. Defects in the small bowel mesentery cause transmesenteric hernias that are associated with a high frequency of incarceration. The differential diagnosis from a volvulus or entrapment by adhesions is sometimes impossible. Heal orjejunal loops may protrude through a defect of the sigmoid mesentery and lie then between two sigmoid loops.
Bowel Perforation Internal Hernias Acute bowel obstruction secondary to incarceration is the typical manifestation of an internal hernia. Internal hernias are difficult to diagnose and CT is most helpful for delineation of clinically occult abdominal wall hernias or internal hernias. It provides information about the anatomic location, the herniated structures and potential complications (incarceration, obstruction, volvulus and strangulation). CT Morphology
The key findings ofinternal hernias are fixation of bowel loops, abnormal locations of bowel loops in suspicious regions such as the lesser sac, crowding of small bowel loops, and signs of bowel loop obstruction. The commonest types are paraduodenal. the foramen Winslow, pericecal. transmesenteric, supravesical. and intersigmoid. Paraduodenal hernias are the most frequent internal hernias {> 50%). A right-sided duodenal hernia (25% of duodenal hernias) is associated with an absent ligament of Treitz. Bowel loops lie lateral and inferior the descending portion of the duodenum. The superior mesenteric artery and branches run along the anterior wall of the right-sided paracluodenal hernia. In a more frequent left-sided paracluodenal hernia the jejunum passes within a sac-like structure through the foramen of Landzert behind the inferior mesenteric vein, and extends into the nonfused mesentery of the descending colon. Small bowel loops and occasionally terminal ileum, cecum, ascending colon, transverse colon, gall bladder, and omentum may protrude
CT is by far more sensitive than plain film radiography for the detection of a small pneumoperitoneum and unsuspected gastroduodenal perforation. Perforations can be the result of an injury (seat belt, steering wheel, or iatrogenic) or an intrinsic lesion (postbulbar ulcer, Crohn'sdisease. diverticulitis). CT Morphology
Perforation of the stomach or duodenal bulb cause extra luminal gas in the peritoneal cavity while perforation of the duodenal loop may lead to air bubbles in the porta hepatis (Fig.l5.33a) or the retroperitoneum e.g., anterior pararenal space. Small pockets of extra luminal air may be trapped beneath the mesenteric leaves or the greater omentum and may be better detected with lung window settings. Larger collections of air are usually localized anterior to the liver in the upper abdomen if the patient is examined in supine position. Perforation of the jejunum, ileum, or colon may cause substantial amounts ofinfra-abdominal free air. Perforations of the ascending, descending, and sigmoid colon as well as the rectum may cause retroperitoneal air that may even extend into the mediastinum. Contained perforations are especially frequent in Crohn's disease or tumors. They present only small air bubbles close to the site of perforation. Usually there are inflammatory changes that seal off part of the peritoneal cavity (Fig.l5.33b, see also Figs. 15.22 c, 16.11 b). Perforations may incite inflammatory reactions in adjacent organs (pancreas, colon, liver).
Mesenteric Ischemia and Infarction
jtig. 15.33 Free abdominal air. (a) D~odenal perforation with free air in the porta hepatis (arrow) and a small fluid collection ("). (b) Bowel ischemia due to strangulation with pneumatosis and a perforation (arrow) leading to a local abscess formation (arrowheads) with free air.
(c) Free air (arrows) in the right infrahepatic space in this elderly patient was not associated with perforation at surgery. Note the streak artifacts caused by placing the arms next to the body (4 x 1/6, coronal MPR).
Fluid extravasation (containing hydrochloric acid and digestive enzymes) from gastroduodenal perforations into the peritoneal cavity gives rise to peritonitis and abscess formation. Postinflammatory strictures may be responsible for bowel obstruction. Free abdominal air is most commonly caused by perforation (Table "15.14) but in rare cases, no perforation can be found at surgery (Fig.15.33c) .
caused by arterial occlusion, non-occlusive ischemia, or venous thrombosis. Embolization of the SMA with emboli originating from the left heart account for more than 50% of cases. Thrombosis of pre-existing atherosclerotic lesions accounts for another 25% of cases with acute mesenteric ischemia. Nonocc1usive mesenteric ischemia accounts for 20-30% of cases with acute mesenteric ischemia, and is related to low mesenteric flow states such as systemic hypotension, cardiac failure, sepsis, or systemic administration of alpha-adrenergic drugs. Chronic intestinal ischemia or intestinal angina is usually caused by atherosclerotic stenosis and less commonly by arterial thromboembolism or compression. Venous thrombosis of the mesenteric vessels may result from extrinsic compression (by a tumor) or be related to risk factors such as thrombophilia, portal hypertension, visceral infections, previous abdominal surgery, or abdominal trauma. The clinical consequence of mesenteric ischemia varies with the sites of involvement. Infarction ofthe small bowel has a mortality of more than 80% and successful treatment requires immediate surgical intervention. By contrast, ischemic colitis is usually self-limiting and responds to supportive care and administration of fluids. One of the major protectors against ischemia of the gut is the presence of extensive collaterals. The small bowel has alternative pathways of blood supply provided by the primary,
Mesenteric Ischemia and Infarction
Mesenteric ischemia can be classified into acute and chronic forms. Acute ischemia can be Table 15.14
Reasons for free abdominal air
Free air in the retroperitoneum Perforation:
rectum cecum
Free air in the peritoneum Perforation:
stomach
duodenum small bowel colon
Free air in the peritoneum Perforation:
sigmoid
and retroperitoneum
colon
589
15 The Gastrointestinal
Tract
secondary, and tertiary arcades with multiple connections between the superior and inferior mesenteric artery and the celiac trunk. In the colon, the marginal artery runs parallel to the mesenteric border. The end arteries-called vasa recta-arise from this vessel. Thesplenicflexureis one of the commonest sites for ischemic colitis because no primary or secondary arcades serve as back-up in this segment. In addition, the vasa recta in the right ascending colon are fewer in number and more prone to vasospasm. Acute mesenteric ischemia is associated with high morbidity and mortality. It is a diagnostic problem because of its nonspecific clinical symptoms and its high mortality if not diagnosed and treated properly and quickly. Arterial angiography remains the gold standard for patients in whom acute mesenteric infarction is suspected clinically. CT angiography, especially with multislice technique, is excellently suited to demonstrate emboli in the SMA and its first segmental branches, and to detect manifest signs of bowel ischemia. •
CT Morphology
The majority of emboli lodge approximately 3-10 cm from the ostium of the SMA,just distal to the origin of the middle colic artery. Approximately 15% of emboli lodge at the origin of the SMA. The arterial embolus may be identified as contrast filling defect or abrupt cutoff of vascular opacification (Fig. 15.34a). Frequently the arterial occlusion is followed by secondary venous thrombosis clue to stasis. In these cases, there is a
a
b
Fig. 15.34 Acute mesenteric ischemia, (a) Direct demonstration of an embolus in the superior mesenteric artery (4 x 1/6). (b) Mesenteric venous thrombosis, (c)
lack ofopacification of the mesenteric veins after injection of contrast media (at> 70 seconds p.i.). CT reveals moderate « 15mm) bowel wall thickening. Submucosal hemorrhage may lead to asymmetric high attenuation areas. Involved bowel segments may be dilated and are often completely atonic with air- fluid levels (paralytic ileus). Scans may also show intramural air inclusions and air collections in the portal and mesenteric veins indicating bowel gangrene (Figs.15.35a, b: see also Fig.15.33b). While in the early evolution of ischemic injury there is intense mucosal enhancement after administration of intravenous contrast (due to slower transit of blood through capillaries), in later phases hypoperfusion of affected bowel segments may be seen (Fig. 15.35c). Gas in intrahepatic portal venous branches accumulates ventrally and extends much closer to the liver capsule than gas in the biliary tree (Fig. 15.35d). In cases of mesenteric venous thrombosis, a prominent intraluminal filling defect in the mesenteric vein can be identified (Fig.15.34b). Care has to be taken not to misinterpret inhomogeneous contrast filling of the mesenteric vein as a thrombotic filling defect if the scan is performed too early. Spiral CT scans acquired during the late arterial or early portal phase after intravenous contrast administration may cause pseudothrombus artifacts that mimic mesenteric vein thrombosis because of differences in transit time for various mesenteric branches and the splenic artery (Fig. 15.34c).
c Pseudothrombus due to laminar flow artifacts contrast enhanced blood from a non-affected segment in Crohn's disease.
of less bowel
Postoperative
Changes
Fig. 15.35 Acute mesenteric infarction. Mural necrosis is evidenced by distended bowel loops and by gas in the bowel wall (a) and mesenteric veins (b), and by lack of
enhancement of the affected segments (c). Air in intrahepatic portal branches extents more close to the liver capsule than aerobilia (d).
There are alternative interventional nonsurgical therapeutic options for treatment of acute arterial or venous occlusion. Preconditions for non-surgical treatment, however, are the lack of bowel wall necrosis (complete lack of enhancement, pneumatosis) and absence of signs of peritonitis (thickening of peritoneal leaves, ascites). Some institutions, therefore, request an exploratory laparotomy for visual inspection of the bowel to exclude manifest intestinal infarction.
tected incidentally at CT and lack specific clinical manifestations. Therefore, the finding of pneumatosis should always be interpreted within the clinical context.
Pneumatosis cystoides intestinalis
This condition involves the appearance of gas bubbles in the subserosa and mucosa of the small and large bowel (mainly the distal colonic segments) in young adults. Primary pneumatosis (15 %) is distinguished from the commoner secondary form (85%) (Table 15.15). Most cases of pneumatosis cystoides intestinalis are de-
•
CT Morphology
CT demonstrates multiple, thin-walled, noncommunicating, air-filled cysts of varying size distributed along the subserosal or submucosal layer of the bowel wall that may mimic pneumatosis due to bowel infarction. The perforation of a cyst can lead to an asymptomatic pneumoperitoneum that may persist for a prolonged period of time. Postoperative Changes
There is a large variety of surgical procedures involving the small bowel and colon. Apart from classical tumor resections, segmental resections, ileocolonostomy, palliative bypass operations in
591
15 The Gastrointestinal Table 15.15 Differential testinal pneumatosis
Tract diagnosis
of secondary
Intestinal trauma
Ingestion Gastrointestinal endoscopy Jejuno-ileal bypass surgery Postoperative bowel anastomosis Abdominal trauma Parenteral nutrition Barium enema
Intestinal ischemial infarction
Necrotizing enterocolitis Mesenteric vascular disease
Intestinal obstruction
Pyloric stenosis Hirschsprung's disease Meconium plug syndrome Neoplasm
Infection
Parasites, tuberculosis Perforated diverticulum Peritonitis Steroid therapy
Inflammation
Pyloric/duodenal ulcer Inflammatory bowel disease Collagen vascular diseases Whipple disease
COPD
in-
Empyhsema Bullous lung disease Chronic bronchitis Asthma Artificial ventillation
COPD = Chronic obstructive disease
bronchopulmonary
tumor patients, enteroplasties to increase the lumen of stenotic segments can be performed, and various techniques are available for permanent or temporary percutaneous stomata. CT is indicated only to rule out complications. •
CT Morphology
Anastomotic leaks manifest as extraluminal fluid collections, postoperatively persisting or increasing intraperitoneal air, oral contrast extravasation (bari urn preparations are contraindicated in such patients), or abscess formation. Postoperative fluid collections are quite common but require further work-up (aspiration biopsy, drainage) only if there are concomitant signs of infection. Hematoma or hemorrhagic ascites may be present (see Fig. 16.9). At the site of an enterostomy, a hyperattenuating suture line may be found. Postoperative 1uminal narrowing at the anastomotic site may have an inflammatory or neoplastic cause
Fig. 15.36 Postoperative dilatation of multiple small bowel loops in a patient with Crohn's disease and multiple corrective surgical interventions for intestinal stenoses (4 x 1/6, coronal MPR).
(anastomotic recurrence), which are indistinguishable by CT if there is only a moderate mass effect. Stenosis of an anastomosis is only considered clinically significantifthere is dilatation of the pre-stenotic bowel loop. Multislice CT provides superior delineation of complex bowel changes after surgical interventions (Fig. 15.36).
Trauma Blunt or penetrating abdominal trauma can involve the small or large bowel and can lead to hemorrhage or perforation. Injury from blunt trauma to the gastrointestinal tract is uncommon, and occurs most frequently in rear-seat passengers in motor vehicle accidents. It can lead to contusion. intramural hematoma. or perforation of the bowel. Contusion or hematoma formation are usually treated conservatively. although the latter may cause obstruction when large. Perforation from laceration or transection requires urgent surgery. Associated vascular lesions can lead to ischemia or infarction with risk of perforation or stricture formation. Peritoneal lavage has been substituted for by sequential abdominal ultrasound but CT is becoming a first-line technique in poly traumatized patients (see also Chapter 25. p. 943) for demonstrating direct and indirect signs of bowel injury. Injuries to the stomach usually occur in children and are most frequently located on the ventral wall. In the duodenum the second and third segments are most frequently injured and there may be associated injuries to surrounding organs (liver. pancreas and spleen). Small bowel injuries involve both mobile parts and parts where the bowel is fixed. Colon injuries are very rare. The transverse part is most often involved. but diagnosis is difficult in a collapsed bowel. When suspicion is high. the patient should be rescanned following the administration of 7501000 ml rectal contrast medium. Mesenteric injuries usually occur in combination with injuries to the bowel. Fresh hematomas should prompt a thorough search for associated bowel injuries. Lacerations of major vessels may lead to life-threatening bleeding. CT requires adequate contrast in the bowel for optimal results. Usually only 500-750 ml is administered via a nasogastric tube directly before the scan and consequently only the stomach. duodenum and part of the jejunum are opacified when there is adequate peristalsis (often trauma will lead to reflex paralysis and gastric dilatation). Nevertheless. CT sensitivity has been reported to be 88-93%. with better results obtained with spiral and multislice CT.
CT Morphology
Penetrating bowel injury with perforation
almost always causes free intraperitoneal. sometimes retroperitoneal air. A wide window setting (lung window) should be used in order not to miss small air bubbles. (Hemorrhagic) ascites may bepresent. Active hemorrhage can be demonstrated by scanning preferable in the arterial phase of enhancement. Hemorrhage may involve the peritoneal cavity but may also be present in the mesentery. If hemorrhage is suspected no positive oral contrast should be administered in order not to mask the findings. Blunt trauma is more difficult to diagnose. CT signs of bowel injury are usually subtle. and diagnosis requires a careful search of all bowel segments in cine mode. Contusion or hematomas will show focal bowel wall thickening with mesenteric infiltration. Free fluid is nonspecific but highly sensitive. Interloop fluid presenting as triangular fluid accumulation between bowel loops can be a sign of bowel perforation. A near-water attenuation suggests s perforation. while attenuation> 30 HU suggests hemorrhage. A "misty mesentery" with increased density of the fat. engorgement of vessels. bowel edema. and may be seen. Hemorrhagic areas within the bowel wall are rare. Sbocl-: bowel is characterized by marked wall thickening an increased contrast enhancement that is thought to be due to reperfusion injury after initial vasoconstriction. In shock. the spleen may become small and hypoattenuating due to internal blood redistribution. and vasoconstriction may be detected in mesenteric arteries. A "caved in" (flattened) cava is caused by hypovolemia. Peritoneal or retroperitoneal air. pneumatosis. loculated fluid between bowel loops. and active contrast extravasation are signs that often prompt direct laparotomy.
16 Peritoneal Cavity and Retroperitoneum M. Jorgensen,M.
Prokop
Anatomy
.
Peritoneal Cavity (p. 596) Supramesocolic Compartment (p. 597) Inframesocolic Compartment (p. 599) Pelvic Cavity (p. 600) Retroperitoneum (p. 600) Examination
Technique
Patient Reparation (p. 601) Acquistion Parameters (p. 601) Contrast Injection (p. 602) Peritoneal
Fluid Collections without Solid Components
Mesenteric, Enteric, and Omental Cysts (p. 603) Lymphangioma (p. 604) Mesenteric Pseudocyst (p. 604) Ascites (p. 604) Hemorrhage (p. 605) Bile Leak (p. 605) Peritonitis (p. 606) Abscess (p. 606) Peritoneal Fluid Collections with Solid Components Pseudomyxoma peritonei (p. 608) Mesothelioma (p. 608) Peritoneal Carcinomatosis and Metastases (p. 609) Solid Processes of Peritoneum and Mesenterium The "Misty" Mesentery (p. 610) Retractile Mesenteritis (p. 611) Lipoma, Liposarcoma (p. 612) Carcinoid Tumor (p. 612) Desmoid Tumor (p. 612) Retroperitoneal
Fluid Collections
Inflammations and Abscesses Hematoma (p. 614) Urinoma (p. 616) Cysts (p. 617) Lymphocele (p. 617)
and Infection
(p. 613)
..
Solid Retroperitoneal
Processes
.
Retroperitoneal Fibrosis (p. 617) Pelvic Fibrolipomatosis (p. 619) Primary Retroperitoneal Tumors (p. 619) Neuroblastoma (p. 620) Rhabdomyosarcoma (p. 620) Teratoma (p. 620) Lipoma (p. 621) Liposarcoma (p. 621) Malignant Fibrous Histiocytoma (p. 622) Leiomyoma, Leiomyosarcoma (p. 623) Secondary Retroperitoneal Masses and Lymphadenopathy
CT has become the diagnostic method of choice for the evaluation of the peritoneal cavity and the retroperitoneum, except for pediatric patients and young, slim patients, in whom ultrasonography is preferable. CT is superior for patients who are hard to examine by ultrasound due to obesity or superimposing bowel gas. Ultrasound is superior in patients in whom there Table 16.1
617
(p. 623)
is little abdominal fat that serves as a natural negative contrast agent in CT The reproducibility of sectional planes makes CT particularly useful for follow-up examinations. CT is also used to guide diagnostic needle aspirations and biopsies and to direct interventional procedures such as percutaneous drainage of abscesses (Table 16.1).
Indications for CT examination of the peritoneal cavity and retroperitoneum
Tumor diagnosis
Primary evaluation of peritoneal or retroperitoneal masses Nand M staging of other tumors Oncologic follow-up
Suspected abscess
Search for focus
Hematomalhemorrhage
Detection, localization, active bleeding?
Preoperative,
Priorto open abscess drainage or drain insertion
interventional
Diagnostic needle aspiration or biopsy
Indeterminate masses cr fluid collections
Other
Lesions indeterminate by ultrasound Trauma
Anatomy The abdominal cavity is traditionally divided into the peritoneal cavity and the retroperitoneum. The peritoneum, mesenteries, and various fascia and ligaments form the boundaries of the peritoneal and retroperitoneal spaces. A thorough knowledge of the various spaces is helpful for understanding the pathways of spread of infectious and neoplastic disease. It is particularity helpful for successful percutaneous drainage of abdominal fluid collections.
Peritoneal Cavity The peritoneal cavity refers to the portion of the abdominal cavity that is lined by the peritoneum. It consists of the space between the two layers of serosa and the enclosed organs. A ligament is formed by two folds of peritoneum that support a structure within the peritoneal cavity. It may contain lymph nodes, vascular structures, and ducts. It is usually named according to the two structures it connects.
Peritoneal Cavity Table 16.2
Compartments
of the peritoneal
Upper abdomen = supramesocolic Subphrenic
cavity
Separated
space
--H .1
compartment into right and left halves by the falciform
Right subhepatic space
Between
liver and right kidney
Left suphepatic space
Between
liver and stomach
Omental bursa (lesser sac)
Between pancreas hilum
Lower abdomen = inframesocolic
compartment
Right paracol ic gutter
Right of cecum and ascending
Left paracolic
Left of descending
gutter
and stomach, extends to caudate
lobe and splenic
colon
colon
Supra mesenteric space
Above
Inframesenteric
Below root of mesentery
space
ligament
root of mesentery
Pelvic cavity Vesic-outerine
Between
bladder and uterus
Cul-de-sac (in females)
pouch (in females)
Between
uterus and rectum
Rectovesical pouch (in males)
Between
bladder and rectum
An amentum is a large reflection of the peritoneum containing also fat, lymph nodes, and vascular structures. Both the greater and the lesser omentum are connected to the stomach. Ligaments form parts of the omentum. A mesentery connects a portion of bowel to the retroperitoneum. It contains vascular and lymphatic structures and is covered on each side by a peritoneal lining. The space between the parietal and visceral peritoneum is subdivided into various compartments by the organs enclosed by peritoneum and by the corresponding ligaments (Table 16.2). These compartments are not separate but communicate broadly with one another anterior to the transverse colon. This area of communication is continued into the pelvis (Figs. 16.1,16.2).
space, the perisplenic space, and the left subhepatic space. The left triangular hepatic ligament that runs between the superior aspect of the left hepatic lobe and the diaphragm does not compartmentalize the left subphrenic space. The immediate subphrenic space is located between the diaphragm and the gastric fundus.
Right ~ subphrenic Subhepatic
Supramesocolic Compartment
colic
The transverse mesocolon separates the supramesocolic from the inframesocolic compartment. The supramesocolic space consists of the right and left subphrenic spaces, right and left subhepatic spaces, the perisplenic space, and the lesser sac (Fig. 16.1). The falciform ligament separates the right and left subphrenic spaces.
Supr-amesenteric
•
Left Subphrenic
Left subphr-enic
\
\
Hepatoduodenal ligament
mesenteric
Mesenteric root
Space
The left -sided subphrenic space is large and can be divided into the immediate subphrenic
Fig. 16.1
Compartments
of the peritoneal
cavity.
597
16 Peritoneal Cavity and Retroperitoneum
Epiploic foramen
omentum Transverse mesocolon
Transverse colon
Omental ---bursa (lesser sac) "" Greater ----~/"
~H-"'?\--\7"'---'~-\----
Lo 0 P of small bowe - Sigmoid colon
omentum Vesicouterine pouch
Rectum Cul-de-sac
__
Fig. 16.2 Sagittal ity and pelvis.
---f
section through the abdominal
cav-
It is crossed by the gastrophrenic ligament and represents a common location for fluid, abscesses, hemoperitoneum, and gastric processes. The perisplenic space is separated from the left paracolic gutter by the phrenicocolic ligament that runs between the proximal descending colon and the diaphragm. This ligament inhi bits a free flow of ascites between these spaces. The perisplenic space is a common location for splenic processes and processes involving the tail of the pancreas. The left subhepatic space (= gastrohepatic recess) is located between the left hepatic lobe and the stomach and may be involved with processes of the gallbladder, duodenal bulb, left hepatic lobe and lesser curvature of the stomach. •
Right Subphrenic Space and Right Subhepatic Space
The right triangular ligament runs posterolaterally from the dome of the right liver lobe to the diaphragm. It divides the right subphrenic and the right subhepatic space.
The latter is further divided into anterior and posterior portions. The anterior right subhepatic space communicates with the lesser sac via the epiploic foramen (foramen of Winslow) and is involved with processes of the gallbladder, pancreas, stomach, and liver. The posterior right subhepatic space is also described as Morisons pouch or the hepatorenal fossa; it is the most dependent part of the intraperitoneal space of a patient in a supine position and therefore a frequent location of fluid or peritoneal metastases. •
Lesser Sac
The lesser sac [omental bursa} is the largest recess of the peritoneal cavity (Figs. 16.1-16.3). It communicates with the greater peritoneal cavity through the epiploic foramen at the inferior border of the hepatoduodenal ligament, which contains the common bile duct anteriorly, the proper hepatic artery centrally, and the portal vein posteriorly. The lesser sac has three recesses: the superior, splenic, and inferior recess. The superior recess surrounds the medial aspect of the caudate lobe, the splenic recess extends across the midline to the splenic hilum, and the inferior recess extends inferiorly between the stomach, pancreas, and transverse mesocolon. Encapsulated fluid produces a cyst-like bulging ofthe lesser sac. An isolated fluid collection in the lesser sac is suggestive for a pancreatitis or a perforated gastric or duodenal ulcer. A disproportionate amount of fluid in the lesser sac may be evidence of malignant disease. The anterior border of the lesser sac is the lesser omentum, which consists of the hepatogastric ligament and-to a lesser degree-the hepatoduodenal ligament and the gastrocolic ligament. The hepatogastric ligament connects the liver with the lesser curvature of the stomach. It contains the left gastric artery, the coronary vein, lymph nodes, and may contain coronary varices in patients with portal hypertension. It is the primary pathway for the spread of a pancreatic phlegmon or for metastatic disease from the esophagus, stomach, or biliary tumors. The hepatoduodenal ligament forms the inferior part of the hepatogastric ligament and runs from the proximal duodenum to the porta hepatis. It contains the common hepatic duct, common bile duct, hepatic artery, and portal
Peritoneal Cavity Fig. 16.3 T opogra phy of the upper abdominal cavity at the level of the lesser sac (omental bursa).
Left subhepatic Perihepatic space
- Lesser sac space ~
vein, and is an important primary pathway for the spread of biliary neoplasms. The gastrocolic ligament connects the greater curvature of the stomach with the superior aspect of the transverse colon. Gastric neoplasms may spread via this structure and infiltrate the transverse colon or the splenic flexure. It represents the superior aspect of the greater omentum and contains the gastroepiploic vessels. Posteriorly and posterolaterally the lesser sac is bordered by the gastrosplenic and splenorenal ligaments. The gastrosplenic ligament runs between the greater curvature of the stomach and the splenic hilum and contains the short gastric vessels. It may be involved by gastric or pancreatic tail processes. The splenorenal ligament connects the posterior aspect of the spleen and the anterior pararenal space and may be involved with pancreatic processes. The mesocolon borders the lesser sac posteroinferiorly. Inframesocolic
Compartment
Mesocolon and mesentery are identified in CTby their fatty consistency, their anatomic location, and by the vascular structures they contain. The transverse mesocolon suspends the transverse colon from the retroperitoneum lying
anteriorly and inferiorly of the pancreas in the upper abdomen. It contains the middle colic vessels. Processes in the pancreas spread via the mesocolon to the inferior border of the transverse colon and vice versa. Most of the omentum is formed by a peritoneal duplication. The greater omentum is suspended from the transverse colon and hangs over the lower abdominal viscera like an apron (Fig. 16.2). The gastrocolic ligament forms the upper portion of the greater omentum that can be located anterior to the small bowel on axial CT sections. It is formed by four layers of peritoneum and therefore predisposed for becoming involved in neoplastic as well as inflammatory processes. Confluent omental masses are referred to as omental cake. The lesser omentum represents a peritoneal duplication between the lesser curvature of the stomach, the superior border of the duodenum and the porta hepatis. The root of the small bowel mesentery suspends the small bowel from the retroperitoneum and extends from the left upper quadrant (ligament ofTreitz) to the right lower quadrant passing in front of the horizontal part of the duodenum. It contains the superior mesenteric vessels and more than 100 lymph nodes which are normally smaller than 1 cm in size. The lymph nodes can be divided into three groups: adjacent to the bowel wall, adjacent to the
599
600
16 Peritoneal Cavity and Retroperitoneum upper trunk of the superior and inferior mesenteric vessels, and parallel to the major branches of the mesenteric vessels, The mesentery subdivides the inframesocolic compartment into a supramesenteric space and an inframesenteric space, Both spaces communicate with each other above the duodenojejunal flexure (Fig, 16.1), Laterally they extend to the ascending and descending colon, Superiorly they are bounded by the right and left halves of the transverse colon, Inferiorly they communicate with the bilateral paracolic gutters and the pelvic cavity, The superior and inferior ileocecal recesses are located above and below the terminal ileum, The cecum normally lies completely retroperitoneal; however, a retrocecal space may be created when there is an additional peritoneal reflection dorsal to the cecum. The right and left paracolic gutters are located lateral to the ascending and descending colon, The sigmoid mesocolon suspends the sigmoid from the posterior wall of the pelvis and contains the sigmoid vessels, It is the target area of diverticulitis, perforated colon carcinoma, or Crohn's disease, The intersigmoid recess is located in the left lower quadrant along the undersurface of the sigmoid mesocolon,
Pelvic Cavity Lateral to the urinary bladder, the peritoneum is deepened on each side to form the perivesical fossae, The deepest part of the peritoneal cavity lies posterior to the bladder. In females, the uterus subdivides this deep recess into the vesico-uterine pouch anteriorly and the recto-uterine pouch (culde-sac orpouc/i o[Douglas) posteriorly (Fig, 16,2), This cul-de-sac is bounded anteriorly by the broad ligaments and the uterus, and is the most dependent portion of the pelvis in women, In males both peritoneal recesses are combined to form a single space between the rectum and bladder called the recto-vesical pouch. Non-encapsulated fluid collections in the supine patient tend to collect in these deep spaces, where they are detectable by imaging procedures, Larger pelvic fluid collections can extend up the paracolic gutters into the subhepatic and subphrenic spaces (Fig, 16.1),
The broad ligaments envelop the parametria, fallopian tubes, ovaries, uterine vessels and ureters, and run from the uterus to the pelvic sidewall. They are involved in neoplastic as well as inflammatory disease, The round ligaments serve as the anterior suspensory ligaments of the uterus and contain lymphatic channels, They pass lateral to the deep epigastric vessels and continue into the ingumal canal, finally inserting on the major labia, The median umbilical ligament lies in the midline and forms a peritoneal reflection over the obliterated urachus, The lateral umbilical ligaments consists of two parts: the medial part runs along the obliterated umbilical arteries to the falciform ligament The lateral part forms a peritoneal reflection over the deep inferior epigastric vessels,
Retroperitoneum The retroperitoneum is bounded superiorly by the diaphragm, inferiorly by the linea terminalis of the lesser pelvis, anteriorly by the posterior layer of the peritoneum, and posteriorly by the muscles of the back and by the bony pelvis, It is continuous inferiorly with the subperitoneal space of the pelvis, and anteriorly with the preperitoneal space, It has connections with the intra-abdominal organs through the mesenteric root and the sites of attachment of the pancreas, liver, duodenum, ascending colon, and descending colon (Fig, 16.4), The layers of Cerota's fascia (anterior and posterior renal fasciae) subdivide the retroperitoneum into three compartments at the level of the kidneys (Figs, 16.4, 16,5): the perirenal space, formed by the anterior and posterior layers of Cerota's fascia; the anterior para renal space between the posterior layer of the peritoneum and the anterior layer of Cerota's fascia; and the posterior para renal space between the posterior layer of Cerota's fascia and the transverse fascia, Viewed in sagittal section, the perirenal space is sharply tapered inferiorly, The anterior and posterior para renal spaces communicate with the anterior pre peritoneal space approximately at the level of the iliac crest A connection also exists between the right and left perirenal spaces, The retroperitoneal fasciae have a normal thickness of 1 -2 mm, CT can define the fascial
Retroperitoneum Peritoneum
~
Lateroconal fascia
Perirenal space Posterior pararenal space Fig. 16.5 Retroperitoneum (fascial spaces) at the level of the kidneys.
space Perirenal space
Fig. 16.4 Parasagittal section through the peritoneal cavity and retroperitoneum.
Examination
Technique
Patient Preparation
Sufficient oral contrast medium should be administered to improve the delineation of pathologic processes. This is especially important when looking for abscesses. In postoperative patients some bowel loops may not fill after oral contrast administration (e.g., roux-en- Y loops). Here oral administration of a biliary contrast agent 12 hours prior to the CT examination leads to opacification of bile secretions and consequently of the excluded bowel loop if there is a biliodigestive anastomosis (see also CT cholangiography, Chapter 12). Acquisition
planes in approximately 50% of patients. The fasciae can be seen if they are perpendicular to the CT scan plane and if the retroperitoneum contains a sufficient amount of fat. Thin-section imaging improves fascial delineation. Fluid collections and sites of fascial thickening improve the delineation of the fascial planes.
Parameters
CT scanning of the peritoneal cavity and retroperitoneum is performed in the supine patient as part of an upper and lower abdominal survey that includes all the abdominal compartments
and the pelvis (Table 16.3). Longitudinal coverage should extend from the diaphragm to the ischial tuberosities. Standard protocols use 5-7 mm section thickness but the evaluation of more complex intra- and retroperitoneal disease is improved with thinner sections. Multislice CT allows for creating excellent coronal or sagittal reformations that are particularly helpful for demonstrating the involvement of the various retroperitoneal spaces and for display of the mesenteric and retroperitoneal vasculature. However, most indications do not require such elaborated scanning techniques. Best results are obtained with 0.75-1.5 mm high-resolution protocols on 4- to 16-slice scanners. On 4- or 8-slice scanners, thicker collimation (e.g., 4-8 x 2.5 mm) is useful in more obese patients to decrease image noise. Similar results, however, can be obtained with thin section collimation if 5-7 mm wide sections are reconstructed in the axial plane and 3-5 mm wide sections are reformatted in the coronal plane.
601
602
16 Peritoneal Cavity ond Retroperitoneum Table 16.3
Suggested
protocol
for CT scanning
of the peritoneal
cavity and retroperitoneum
General Preparation
Tumor diagnosis:
Oral contrast
1000-1500 ml contrast medium, given over 60-90 min prior to examination if necessary administered by gastric intubation. Negative contrast agents (water) may be advantagenous for multislice CT
Patient
position
Scan range
patient
Supine with elevated From diaphragm
Respiratory
phase
Windowing
should fast for 6 hours
arms
to ischial tuberosities
Inspiration Noncontrast CT: Contrast-enhanced
CT:
W/L=300/40 W/L=400/60
Scan parameters
1-slice SC/TF/RI
4-slice' SC
16-slice' SC
Standard
7/12/6
2-4 !
1-2!
1-1.25!
0.75-1.25
Volumetric Contrast
imaging injection
!
5/8/4 ! 1-slice
Standard
CT
5-7.5/5 !
16-slice V+N/F/DXc
100 + 50/2/70
120/2/70 + 70
120 + 50/4/25
3/4 cor 4/4 sag
5/4
4-slice V+N/F/D
V/F/D
MPRb SWIRl
axial SWIRl
100 + 50/2/70 + 70
120 + 50/4j20A
Biphasic CT
150/4/25
Interstitial (delayed) phase
5min
5min
5min
CT urography
2:15 min
2:15 min
2:15 min
d
d
+ 80
d
SC = slice collimation (mm), TF = table feed (mmjrotation), RI = recon. increment (mm), T! = scanning direction, SW = effective section width (mm), MPR = multiplanar reformats, axial = axial sections, cor = coronal, sag = sagittal CM = contrast material, Contrast concentration = 300 mg/ml iodine V = volume of CM (ml), N = volume of saline flush (ml), F = flow rate (mljs), 0 = start delay, X = trigger region a Pitch P=TFj(NxSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI = 1-1.5/0.7 c read: lOA = start delay 10 s after reaching 50 HU enhancement in the desc. aorta dread: biphasic scan (arterial + portal phase) after 25 s (lOA) for the 1" scan, 70-80 s for the 20d scan
Contrast Injection Many peritoneal and retroperitoneal abnormalities originate from abdominal organs, and intravenous contrast administration should be geared toward the primary lesion. In patients with neoplastic disorders, the IV contrast administration should be tailored to allow optimum evaluation of the liver (hepatic metastases). I!' hemorrhage is suspected, the initial scans should be obtained without intravenous contrast. Scans in the interstitial
phase of contrast enhancement are helpful in fibrous processes such as retroperitoneal fibrosis, while late scans some 15 min after contrast application are required to detect leaks of the urinary tract. With multislice CT, late scans (thin-sections, reduced dose) can provide excellent visualization of the ureter (CT urography). As an alternative, 20 ml of contrast material may be given up to 20 minutes before the examination to provide opacification of the collecting system during arterial or portal phase scanning. An overview is given in Table 16.4.
Mesenteric, Enteric, and Omental Cysts Table 16.4
603
Suggested examination technique Precontrast
Arterial phase
Portal phase
Interstitial phase
Active hemorrhage
+
+
(+)
Trauma
+
(+)
+
(+)
Urinoma
(+)
+
+
+
(+) (+)
+
(+)
Mesothelioma
+ +
(+)
Other tumors
+
+
Abscess Peritonitis Retroperitoneal fibrosis Sarcomas. carcinoid
MPR= multiplanar
(+)
reformations;
CT urography'
(+) (+)
(+) + (+)
(+)
+ suggested, (+) may yield additional information,
(+)
+
- not recommended
• Preopacifying the pyelocaliceal system with 20 ml of contrast medium 5-20 min before scan replaces excretory phase scanning (not recommended in suspected urolithiasis)
Peritoneal Fluid Collections without
Solid Components
The detection of peritoneal fluid is not a primary indication for CT scanning. CT is the modality of choice, however, for the differential diagnosis of abdominal fluid collections and especially for the exclusion of abscesses and hemorrhage. Table 16.5 provides the most important differential diagnoses for intra-abdominal fluid collections.
with epithelium and a layer of smooth muscles. They occur anywhere in the gastrointestinal tract, but are most common in the esophagus (10-15%) and the small bowel (see also Chapter 15, p. 541). •
IT Morphology
Mesenteric cysts are located between
Mesenteric,
Enteric,
and Omental
Cysts
Mesenteric cysts are asymptomatic lymphatic hamartomas that may contain variable fluid components and are usually detected incidentally. Approximately 80% of the cysts occur in the mesentery of the small bowel. Alimentary tract (enteric) duplication cysts result from developmental errors and are lined
Table 16.5 Differential fluid collections
diagnosis
of intra-abdominal
Ascites Bowel (dilated bowel loops, bowel obstruction) Biloma Cyst Seroma Hematoma Inflammation (empyema, abscess) lymphocele Urinoma Necrotic tumor
the leaves of the mesentery, mostly of the small bowel (Fig. 16.6). CT demonstrates a single cyst or multilocular cysts that may reach several centimeters in size. The cyst may exhibit water or softtissue CT attenuation, depending on its contents. A fluid-fluid level may be present if the cyst contains fat. Peripheral calcifications may be found. Omental cysts are located in the greater omentum or lesser sac. They are usually pedunculated. Twisting of the pedicle can only be detected in rare cases. Bowel obstruction is marked by dilatation of the proximal bowel segments. Enteric duplication cysts are usually unilocular and have a well-defined enhancing wall, which contains a muscular and mucosal layer. They contain serous or occasionally hemorrhagic fluid. Gas inclusions indicate a communication with the bowel. Most often they are located at the mesenteric surface of the ileum, more rarely at the ileocecal junction, the duodenum or at the greater curvature of the stomach. CT shows elongated tubular or spheri-
II
604
16 Peritoneal Cavity and Retroperitoneum Mesenteric Pseudocyst
Mesenteric pseudo cysts often contain blood and evolve from prior hematomas or abscesses. •
CT Morphology
Pseudocysts can be uni- or multilocular. Differentiation from pancreatic pseudocysts is based on the lack of other signs of inflammation in the retroperitoneum, and thus may be difficult. The wall is usually thicker than in simple cysts and has a stronger contrast enhancement. Fig. 16.6 Mesenteric cyst with wall calcification small bowel mesentery.
in the
cal cystic masses that are intram ural in location or run parallel to the normal bowel lumen. The differential diagnosis includes abscess and old hematoma (mesenteric pseudocyst, see below). Lymphangioma
Lymphangiomas are congenital disorders. They represent the most common cystic tumors of the mesentery. •
CT Morphology
Lymphangiomas may be multilocular with very thin cystic walls that merge with the bowel wall. Imperceptible cystic walls lead to an illdefined opacification of the mesenteric fat that obscures the cystic nature of the tumor (Fig. 16.7a). The fluid within a lymphangioma is variable and may be serous, chylous, or hemorrhagic. The presence of streaky low densities within a multi-cystic lesion are very suggestive for the diagnosis (Fig.l6.7b). A fat-fluid level within cystic spaces is seen only rarely.
Ascites Numerous diseases, including cardiovascular, inflammatory, and neoplastic entities, may be associated with the accumulation of ascitic fluid in the peritoneal cavity. The collection may be transudative or exudative in nature. Special forms are hemorrhagic ascites (secondary to tuberculosis or peritoneal carcinomatosis) and chylous ascites. •
CT Morphology
Ascites often presents as a hypoattenuating rim that surrounds intraperitoneal organs. Fluid volumes of approximately 50 ml or more can be detected on smooth organ surfaces such as the liver. Adequate bowel opacification helps to detect small amounts of fluid in the infracolic spaces. The distribution of ascites is determined by the anatomy of the intraperitoneal compartments, by gravity, and by the position of the patient. In supine patients, nonencapsulated fluid quickly gravitates to the more dependent spaces (the cul-de-sac and posterior subhepatic space). Larger fluid collections usually do not enter the lesser sac. Large amounts of fluid can
Fig. 16.7 Lymphangioma. Tiny cysts are imperceptible on CT but present as an ill-defined opacification of the mesenteric fat (a). Large cystic components are very suggestive of the diagnosis (b).
a
b
Bile Leak
a
c
b
Fig. 16.8 Ascites. (a) A large perihepatic fluid collection has caused medial displacement of the liver. The subphrenic space is subdivided by the falciform ligament. There is a small perisplenic fluid collection and pleural effusions. Atelectasis of the lung bases (arrows) should not be mistaken for the diaphragm (arrow heads), (b) Ascitic fluid in the subhepatic space, the
cause separation of small bowel loops and the fatty layers of the mesentery (Fig. 16.8). Adhesions (postoperative, inflammatory, neoplastic) give rise to cystic fluid collections that may be difficult to distinguish from abscesses. Fibrin septa usually cannot be directly detected by CT (as opposed to ultrasound with which they are excellently displayed). Indirect signs of septations incl ude scalloping of the liver contour. The CT attenuation of ascites ranges from 0-30 H, depending on the protein content of the fluid. Higher attenuation values are characteristic of ascites due to tuberculosis, ovarian or appendiceal tumors, and intraperitoneal hemorrhage. In c/ironfc ascites there may be slight thickening of the parietal peritoneum. Enlarged peritoneal veins should not be mistaken for peritoneal metastatic deposits. Intraperitoneal fluid with solid components, thickening of the peritoneum, and a large amount of fluid in the lesser sac should raise suspicion of peri ron eo I carcinomatosis. Hemorrhage
Infra-abdominal hemorrhage most commonly results from sharp or blunt abdominal trauma (organ or vascular injury), bowel perforation, tumor rupture, ectopic pregnancy, or anticoagulant medication. In hemodynamically stable patients, CT is the modality of choice for detecting or excluding an intra-abdominal hematoma.
supra- and inframesenteric space, and in the right and left paracolic gutters, causing marked displacement of the bowel loops and mesentery. Note the retroperitoneal location of the duodenum (arrows) as well as the ascending and descending colon (arrow heads). (c) Ascites in the supine patient guickly gravitates into the cul-de-sac between the rectum and uterus.
•
CT Morphology
Fresh hemorrhagic collections show soft-tissue attenuation values because there is not yet a separation between high-attenuation erythrocytes and serum. Fresh hemorrhage may be better detected on contrast -enhanced images as a hypo attenuating rim around the liver or spleen. Clots have high CT attenuation, but scans often show only streak-like or rounded, ill-defined areas of increased attenuation within the collection. The CT attenuation of the intraperitoneal hematoma fades within days, reaching a value of 0-20 HU by 1 to 2 weeks. Layering effects (ventrodorsal attenuation gradient, fluidfluid level) are most commonly seen in hemorrhagic ascites, especially in the lesser pelvis (Fig. 16.9). Bile Leak
Intraperitoneal bile leakage may be caused by iatrogenic manipulations, trauma, or spontaneous rupture of the bile ducts. The leak incites a local inflammatory response that usually leads to rapid encapsulation of the intraperitoneal fluid. CT is indicated prior to surgical or interventional procedures and during follow-up. •
CT Morphology
CT generally shows a localized fluid collection (biloma) in the region of the porta hepatis or at
605
606
76 Peritoneal Cavity and Retroperitoneum Fig. 16.9 Hematoma. (a) Fresh hematoma anterior to the left liver lobe, which is hyperattenuating to muscle but isoattenuating to the liver and thus almost imperceptible on the non-contrast scan. (b) On the post-contrast scan, the hematoma is clearly demarcated from the liver parenchyma.
a
b
the hepatic margin. It usually appears encapsulated and has a CT attenuation of about 10-20 HU (Fig. 16.10). Loculation due to fibrin strands is rarely seen on CT. Fibrin strands are much better detected by ultrasound. Peritonitis
Peritonitis usually results from an infection of the abdominal cavity due to peritoneal injury (e.g., surgery) or bacterial seepage through the peritoneum. It may involve the entire peritoneum (diffuse peritonitis) or may be a localized process (local peritonitis). Fibrinous,
purulent exudates cause adhesions to form between the peritoneal layers, leading to localized abscess formation. Adynamic (paralytic) ileus may develop through reflex mechanisms. Clinically overt peritonitis is a serious disorder and an absolute indication for surgical treatment. It does not require further radiological assessment. But if clinical presentation and ultrasound findings are equivocal, CT examination may be required. S
CT Morphology
Acute peritonitis is characterized by ascites with dilatation of mesenteric vascular structures and a hazy increase in attenuation of the mesenteric fat. The peritoneal layers are smoothly thickened (Fig.l6.11a) and enhance after intravenous contrast administration. Chronic forms may be associated with focal peritoneal thickening and calcifications (Fig.l6.11b). Abscess
Fig. 16.10 Biloma. CTfollowing a left-sided partial hepatic resection. This demonstrates a fluid collection at the resection margin and a second, large parahepatic fluid collection on the right side. The latter has lead to substantial liver compression with transient hepatic attenuation difference of the compressed parenchyma (arrow heads).
An isolated inflammatory process secondary to a hollow viscus perforation or surgery may incite a local, circumscribed peritonitis that becomes encapsulated and forms an abscess. The adjacent abdominal compartments may be involved. An encapsulated abscess may also persist after diffuse peritonitis. CT is used in patients with unexplained fever to exclude an abscess or to localize an abscess prior to intervention.
-Abscess
607
b Fig. 16.11 Peritonitis, (a) A postoperative colon perforation led to fecal peritonitis with large amounts of free fluid. The pneumoperitoneum (broad arrows) indicates perforation. The extraluminal gas within the fluid collection (arrows) indicates a high viscosity of the fluid, in this case due to fecal material. Note that there is only
a thickening and enhancement of the peritoneum on the right side (arrow heads) where the perforation was found at surgery, (b) A confined perforation in this patient incited a local peritoneal inflammatory reaction with small amounts offree fluid and extraluminal gas.
•
subphrenic space. Despite the direct communication between Morison's pouch and the lesser sac, a generalized peritonitis usually does not affect the lesser sac because the slit -like connection is often sealed off by adhesions. Inflammatory processes within the lesser sac are highly suggestive for being primarily caused by adjacent organs such as a perforated ulcer from the stomach or duodenum. Left-sided subphrenic abscesses mostly result from complications after gastric, colonic, or splenic surgery (Fig. 16.12). Rarely they may result from a perforated ulcer in the stomach or duodenum.
CT Morphology
An abscess typically appears as a localized fluid collection of "1O-40HU attenuation. Thickened, enhancing peritoneum may be seen as a manifestation of the peritoneal inflammatory response. Inflammatory spread to adjacent structures is manifested by ill-defined boundary surfaces and the obliteration of fat planes. Extral uminal gas is pathognomonic, but postoperative air collections may be detectable for up to 14 days. Secondary signs include scoliosis, ileus, or changes at the lung bases. Developing abscesses may resemble encapsulated ascitic fluid. Thus, any isolated fluid collection is suspicious for an abscess when corresponding clinical signs are present. It requires fine needle aspiration for differentiation. Sites of predilection for abscess formation are the right subphrenic space, the Morison pouch (subhepatic space), the cul-de-sac (see Fig. 16.23), and the rectovesical pouch. The frequency of subphrenic and subhepatic abscesses is 2-3 times greater on the right than on the left. The commonest site is the Morison pouch. Abscesses of Morison's pouch and the right subphrenic space often coexist. This is explained by the fact that negative subdiaphragmatic pressure causes flow from the pelvis up the right paracolic gutter. The right subphrenic space is affected only after Morison's pouch has been contaminated. The falciform ligament avoids direct passage from the right to the left
Fig. 16.12 Large peritoneal patic space.
abscess
in the left subhe-
16 Peritoneal Cavity and Retroperitoneum
As effect of gravity, fluid in the inframesocolic compartment invariably seeks the pelvic cavity (pouch of Douglas, lateral paravesical
fossa) and makes the pelvis the most frequent location of residual abscess formation following generalized peritonitis.
Peritoneal Fluid Collections with Solid Components Pseudo myxoma peritonei
Pseudomyxoma peritonei is a rare condition caused by the rupture of a cystic lesion with the release of mucinous, gelatinous material and mucus- forming epithelium into the free abdominal cavity. The source of the material may be a mucinous cystadenocarcinoma of the ovaries or the gastrointestinal tract. A similar pattern is produced by an appendiceal mucinous adenoma (degenerative mucocele). Once the material has been seeded into the abdominal cavity, it continues to form mucus, leading to adhesions, bowel obstruction, and occasionally to abscess formation.
masses do not enhance with intravenous contrast, but slight enhancement is observed in the septations and solid components of the lesions (Fig. 16.13). The peritoneum itself may also be thickened. The differential diagnosis includes pancreatitis with pseudocysts, pyogenic peritonitis, and widespread echinococcal disease. Mesothelioma
CT demonstrates extensive, hypoattenuating intraperitoneal masses of varying size that progressively displace the bowel loops and mesentery posteriorly. The masses may cause a characteristic scalloping of the liver contour. The thin-walled cystic masses (approximately 15-30 HU) appear as bulky, septated areas of pseudo-ascites. Ring-like or curvilinear calcifications rarely develop in chronic forms and are highly suggestive of pseudomyxoma. The
Primary neoplasms of the peritoneum are very rare, the commonest being peritoneal mesothelioma. The peak incidence is between 55 and 65 years of age. Men are predominantly affected. The disease is associated with asbestos exposure. The peritoneum is the second commonest site of involvement (30%) after the pleura (60%). Benign forms are localized; malignant forms show early spread along the peritoneal surfaces. A cystic type of mesothelioma occurs in young to middle aged women with a prior history of abdominal surgery, pelvic inflammatory disease or endometriosis. This type has no association with asbestos exposure.
Fig. 16.13 Pseudomyxoma peritonei with extensive, hypoattenuating intraperitoneal masses impressing on the liver surface and displacing bowel loops and
mesentery. The solid components and septa show slight contrast enhancement, and peritoneal thickening is observed.
•
CT Morphology
Peritoneal Carcinomatosis and Metastases
deposition and growth of seeded neoplasms in the abdomen depends on the flow of ascites within the peritoneal spaces and is similar to the spread of peritoneal abscesses. Peritoneal Carcinomatosis may be detected as an incidental finding or during CT tumor staging. Because implants may be as small as 2 mm they pose significant problems for CT diagnosis. While CT clearly depicts diffuse nodular peritoneal thickening, it is rather inaccurate for the detection of small lesions especially in the pelvis. • Fig.16.14 Mesothelioma with nodular peritoneal thickening (arrows) on the bowel surface and parallel to the mesenteric vessels (arrow heads).
•
CT Morphology
In malignant" mesot-heh'oma, CTreveals foci ofirregular, nodular peritoneal thickening as well as circumscribed masses that occur predominantly in the anterior parietal peritoneum. There is frequently thickening of the mesentery, omentum, and bowel walls (Fig. 16.14). Focal calcifications may be seen. The masses and sites of nodular peritoneal thickening are more clearly demarcated after IV contrast administration. The disproportionately small amount of clear ascites is suggestive for the diagnosis. A singular nodular mass of the peritoneum is highly suggestive of benign mesothelioma or mesorhe !iaffibroma. In patients with cystic mesothelioma CT shows multiple tiny cysts with thin fibrous septa that are thought to represent inclusion cysts secondary to entrapment of mesothelial cells after laparotomy. Differentiation from a lymphangioma may not be possible based on morphological criteria. Peritoneal Carcinomatosis and Metastases
Secondary peritoneal tumors are far commoner than primary peritoneal neoplasms and most commonly originate from carcinomas of the stomach, colon, ovaries, gallbladder, pancreas, and uterus. Initial tumor extension is a local subperitoneal growth, followed by spread to the peritoneum and intraperitoneal seeding. The
CT Morphology
Peritoneal neoplasms are generally associated with pronounced sero-fibrinous or hemorrhagic ascites with solid components. Peritoneal thickening may be generalized without major nodular components (Fig.16.15a). Small nodular densities are best appreciated on smooth organ surfaces such as the liver or at posteriorly located peritoneal surfaces. They are best detected with thin-section techniques such as multislice CT. Even if solid components are not found, the presence of ascites in tumor patients should always be taken as a potential sign of peritoneal Carcinomatosis. A large fluid collection in the lesser sac reinforces the suspicion of carcinomatosis in a patient with no apparent gastric or pancreatic abnormalities. Sites of predilection for peritoneal Carcinomatosis are the pouch of Douglas (50%), the ileocecal junction (40%), the superior aspect of the mesosigmoid (20%), and the right paracolic gutter (20%). Along the ruffled small bowel mesentery there are a number of peritoneal recesses that run obliquely towards the right lower quadrant. Malignant cells in the ascites will follow these recesses and will finally arrive in the lower recess of the small bowel mesentery in the right infracolic space. Metastases typically displace the distal ileal loops and infiltrate the medial contour of the cecum. In the left lower quadrant the superior border of the sigma is affected in a similarly way. Coalescence of multiple metastatic deposits is typically associated with a dense fibrotic (desmoplastic) reaction. Metastatic implants in the right subphrenic space along the liver surface and along the falciform ligament are most characteristic for ovarian cancer (see Fig. 16.13). They may be more the result of lymphatic drainage to the right-
7 6 Peritoneal Cavity and Retroperitoneum
a Fig. 16.15 Peritoneal carcinomatosis. (a) Broad, plaque-like thickening of the greater omentum (omental cake), which shows marked contrast enhancement.
(b) Peritoneal seeding of cholangiocellular carcinoma with nodules at the liver surface, on the bowel surface and in the mesentery (4 x 1 /6, coronal MPR).
sided diaphragm rather than the result of direct intraperitoneal spread. More rarely metastatic implants can occur in the right paracolic gutter and the Morison pouch. Occasionally peritoneal metastases from mucinous ovarian carcinoma may calcify. Gastric or more rarely colon mucinous adenocarcinomas tend to seed bilateral metastases [Kmkenberg tumors) to the ovaries, which appear as cystic lesions or masses of soft-tissue density, which are usually associated with massive ascites. Metastatic spread into the umbilical region is described as Sister Mary Joseph nodule. This may be caused by carcinomas of the stomach, ovary, colon, or pancreas.
Invasion of the mesenteric fat is shown by a reticular or nodular pattern of increased density. Fluid deposition causes an apparent thickening of the mesenteric vessels. A stellate pattern is produced by stretching and increased density of the mesenteric root. Affected portions of the omentum have a nodular or streak-like increase in attenuation of the fatty tissue, resulting in sheet-like or plaque-like masses called omental cake (Fig.l6.15b). CT following IV contrast administration shows enhancement of the peritoneum and of solid components. Paired vascular structures in the mesenteric fat (arteries and veins) are more easily distinguished from lymph nodes on contrast -enhanced scans.
Solid Processes of Peritoneum and Mesenterium The "Misty" Mesentery
Infiltration of the mesenteric fat by cells, fluid (edema, lymph, blood), tumor or fibrosis has been termed the "misty" mesentery. CT Morphology
The CT attenuation, which is normally around -100 HU, increases to values of -10 HLJ to -60 HU. As a consequence, vascular structures within the opacified mesenteric fat may be obscured and can only be delineated after application of contrast medium. The visceral peri-
toneum-normally not visible-may become recognizable due to fluid deposition. A diffuse mesenteric edema has multiple etiologies (Table 16.6). It extends from the surface of the bowel to the root of the mesenteric vessels. Due to edema, the bowel wall becomes illdefined. If associated with a systemic process the mesenteric edema is mostly combined with a generalized subcutaneous edema or with ascites (Fig. 16.16). Mesenteric edema due to arterial or venous mesenteric thrombosis tends to remain localized (see Fig. 15.34).
Retractile Mesenteritis
Table 16.6 mesentery" Diffuse
Differential
diagnosis
of
the
"misty
meseirlfferic edtenria
As cites Portal hypertension Severe cardiac congestion
Inflammatory processes Pancreatitis Appendicitis Diverticulitis Inflammatory bowel disease Tuberculosis Retractile mesenteritis
Hemorrhage Trauma
Tumors AV malformation, Anticoagulation
mesenteric aneurysm therapy
Lymphedema Congenital anomalies Inflammation Leukemia Non-Hodgkin's lymphoma Mesothelioma Metastases (pancreas, colon, breast, carcinoid, ovaries) Surgery Radiation therapy
Inflammatory processes such as pancreatitis, appendicitis, diverticulitis, or inflammatory bowel disease result in mesenteric inflammation that is more or less locally defined. In abdominal tuberculosis, infiltration of the mesentery is usually patchy. There is an increased number of enlarged lymph nodes that show a strong peripheral enhancement with low-density centers. Concomitant findings in tuberculous infection include high-density ascites, bowel wall thickening, and peritoneal thickening and enhancement. Fresh blood in the mesentery shows typical high CT attenuation (50-70 HU). It may originate from the mesenteric vessels, the bowel wall, the retroperitoneum, or the pelvis. In trauma-related bowel lacerations the alterations in the adjacent mesenteric fat may be more conspicuous than the injury itself. Lymphedema of the mesenteric fat occurs after obstruction of the lymphatics due to congenital anomalies, inflammation, neoplasms,
Fig. 16.16 Mesenteric edema and anasarca with increased (fluid-like) density of the mesenteric, retroperitoneal, and subcutaneous fatas well as ascites and pleural effusion (notshown) in a patientwith hypoalbuminemia. Also note the marked edema of the transverse colon with mucosal enhancement (accordion sign).
surgery, or radiation therapy. In patients with non- Hodgkin's lymphoma it was found that the infiltration of the mesenteric fat becomes more obvious and may persist over years, while the lymphadenopathy may decrease under radiation or chemotherapy. Metastatic spread of tumors of the pancreas, colon, breast, and of carcinoid, leukemia, ovarian cancer, or mesothelioma may develop a similar appearance. Retractile Mesenteritis
Retractile mesenteritis (also called mesenteric lipodystrophy, chronic fibrosing mesenteritis, sclerosing mesenteritis, mesenteric panniculitis) is a very rare disease of unknown etiology characterized by fibrolipomatous thickening of the small bowel mesentery, chronic inflammation with fat necrosis and calcifications, and finally mesenteric fibrosis. The peak incidence is in the sixth decade, and men are predominantly affected. The disease has a nonspecific clinical presentation and is usually discovered incidentally. •
CT Morphology
CT shows a mesenteric mass {mesenteric panniculitis; with a CT attenuation slightly above that of fat that is permeated by soft tissue (connective tissue) and may contain calcifications. The lesion may be sharply demarcated and then has a rim of higher density due to connective
612
16 Peritoneal Cavity and Retroperitoneum Fig. 16.17 Mesenteric panniculitis with increased attenuation of the mesenteric fat (arrows), which is clearly demarcated against normal retroperitoneal fat. The coronal MPR (4 x 1/6) demonstrates multiple small nodules (arrowheads) within the mesentery (b).
a
b
tissue (Fig. 16.17). Mesenteric vessels traverse the lesion in an undisturbed fashion and may be surrounded by a hypo attenuating halo of normal fat. Lesions are well delineated even on unenhanced scans. In more advanced cases, a softtissue dense fibrous mass may develop in the thickened mesentery (fibrosing mesenteritis) is recognized by its stellate configuration. Small bowel loops may be retracted or narrowed, but without signs of obstruction (proximal bowel dilatation). A single nodule or multiple nodular masses (fibromas) may be present in the mesentery. The findings are characteristic but have to differentiated from liposarcomas and carcinoid tumors. Lipoma, Liposarcoma
See Solid Retroperitoneal Processes (p. 617). Carcinoid Tumor
Up to 95% of carcinoid tumors are located in the gastrointestinal tract, mostly in the ileum. The tumors may be multiple, and either metastases or the primary tumor may be located in the mesentery. •
CT Morphology
Carcinoid tumors within the mesentery present as an ill-defined mass with a stellate radiating pattern (see Fig. 15.13). More than 70% show coarse and dense calcifications. Extension of the tumor to the small bowel causes focal wall thickening. There may be multiple nodular tumor metastases in the adjacent mesentery.
The tumor-whether lying within the small bowel or within the mesentery-is typically associated with an excessive peritumoral desmoplastic and fibrotic reaction leading to retraction of the mesentery, beading of the mesenteric vascular bundle, and displacement of adjacent bowel loops (see Fig. 15.13). Encasement of mesenteric vessels may lead to chronic ischemia. Liver metastases occur in 90% of patients with carcinoid syndrome and in >85% of patients with tumors larger than 2 cm. Desmoid Tumor
Desmoid (aggressilivefibrom a tosis) is a rare infiltrative tumor that arises from the aponeurosis or the fascia of muscle groups, or directly form the mesentery. Abdominal desmoid predominantly affects women of childbearing age, and its biologic behavior (non-metastasizing, high recurrence rate) places it in the category oflowgrade sarcomas despite its benign histology. CT is performed as a staging examination. Desmoids can be divided into those that involve the abdominal wall or mesentery, and those that affect other regions (extra-abdominal desmoids). They are associated with Gardner syndrome, multiple pregnancies, and prior trauma. •
CT Morphology
Scans demonstrate large masses measuring 5-20 cm in size, located in the mesentery or in the aponeurosis of the rectus abdominis muscle or internal oblique muscle (Fig.16.18a). The masses may be well-circumscribed at initial
Inflammations and Abscesses
a
c
b
Fig. 16.18 Desmoid tumor, (a) Large mass arising from the fascia of the rectus abdominis muscle. The tumor has invaded the subcutaneous fat anteriorly, and posteriorly it has expanded the muscle and penetrated the peritoneum. Desmoid tumor of the mesentery that is
hypoattenuating during the portal phase (b) after contrast injection but displays marked late enhancement in the interstitial phase (c) due to a large fibrous component.
presentation and even have a pseudocapsule. Recurrent tumors, however, often are less welldefined with respect to the fatty tissue and adjacent organs. Twisting and distortion of bowel loops indicates infiltration of the mesentery. Due to their fibrous origin lesions are slightly hyperattenuating to muscle tissue. They show little or no enhancement in the portal phase after IV contrast administration but en-
hance substantially during the interstitial phase due to their large fibrous component (Fig. -16.18b). Differentiation from lymphoma or other more malignant lesions is based on the presence of a single mass in the mesentery without retroperitoneal adenopathy. Late enhancement is characteristic.
Retroperitoneal Inflammations
Fluid Collections and Infection
and Abscesses
Local reactions or abscesses may develop in the retroperitoneum as a complication of retroperitoneal inflammatory disease (Crohn's disease, retrocecal appendicitis, pancreatitis, pyelonephritis, osteomyelitis, spondylodiscitis), trauma (pancreas, kidneys, vessels), extraperitoneal hollow viscus perforation (gastrointestinal tract), or the superinfection of hematomas and urinomas. Apart from tuberculosis (developing countries), the majority of psoas abscesses are of pyogenic origin (Staphylococcus aureus and mixed gram-negative organisms). Acute abscesses or inflammations lead to serious illness. Chronic forms present with nonspecific clinical signs and symptoms.
CT is indicated in patients with equivocal ultrasound findings, and it is used in the planning of open or percutaneous abscess drainage. H
CT Morphology
Inflammatory reactions in the retroperitoneum may be localized by encapsulation, may be confined to fascia-defined spaces, or may involve the entire retroperitoneum. Extensive processes may invade the fascial space of the psoas muscle and spread within the muscle compartment to the groin. Concomitant involvement of the iliopsoas muscle leads to generalized or focal expansion of the muscle that may be associated with intramuscular abscess formation (Fig. 16.19a). The density increase of an inflammatory collection in the retroperitoneum depends on the protein content and age of the process, resulting
7 6 Peritoneal Cavity and Retroperitoneum
a
b
c
Fig. 16.19 Retroperitoneal abscess, (a) Expansion of the psoas muscle with intramuscular abscess formation. Air inclusions and rim enhancement are pathognomonic. (b, c) Tuberculous abscesses arising from a spondylodiscitis LS/S 1 (c). Fluid collections are adjacent
to the psoas muscles. The collections (arrowheads) between the iliac vessels were located anterior to the sacrum (b). Small calcification are frequently seen in tuberculous abscess formations (arrow).
in attenuation values that usually range from 10-30 HLJ. The fascial planes are most often thickened and may show enhancement with intravenous contrast. Early stages of abscess formation may not yet show the typical rim enhancement seen in more mature abscesses. Exudative processes (e.g., phlegmons, pancreatitis) permeate and obscure the retroperitoneal fat, causing a general increase in CT density. Fat necrosis presents as regions of low attenuation within the fatty tissue and may not be distinguished from abscess formation. The fluid collections that may form in pancreatitis should not be misinterpreted as abscesses, even though these collections may be thick-walled due to granulation tissue. CT does not provide the means of making a definitive diagnosis of abscess in a patient with pancreatitis unless there are gas inclusions, and the patient had not received any type of prior aspiration biopsy. Aspiration biopsy is often the only way to confirm the suspected diagnosis. Tuberculosis used to be the commonest cause of abscess formation in the spine, paraspinal region, and the iliopsoas compartment.
Today there has been resurgence in tuberculous paraspinal infections in HIV -positive patients. Bone destruction, involvement of the disk space with spread along the anterior longitudinalligament, enlargement of the psoas muscle by fluid collections, and associated calcifications within the abscess are typical findings (Fig.16.19b). As opposed to pyogenic infections, tuberculous infections are characterized by a relatively thick and potentially calcified abscess rim, multiple abscess cavitations and only minimal new bone formation. CT-guided aspiration and drainage can be used for confirmation of diagnosis and for treatment. Hematoma
Retroperitoneal hematomas can result from vertebral or pelvic fractures, injuries of the pancreas or urogenital tract, or vascular injuries. Among these causes, trauma of the kidneys is most frequent (see Fig. 18.46). Spontaneous retroperitoneal or pelvic hemorrhage in adults requires exclusion of an aortic rupture (see Chapter
Hematoma Table 16.7 hemorrhage
Reasons
for
spontaneous
retroperitoneal
Tumor Angiomyolipoma Renal cell carcinoma Adrenal tumors Vascular lesions Aortic disease Renal artery aneuryms Congenital AV malformation Infarction Renal venous thrombosis Ruptured gonadal artery Systemic causes
•
Polyarteritis nodosa Anticoagulation therapy Fibrinolytic agents
24, p. 878). It may also be due to neoplastic diseases, hemorrhagic diathesis, or anticoagulant medication, which is the most frequent cause of spontaneous psoas hematoma. It is important to look for causative lesions such as tumors (Table 16.7) in patients with spontaneous retroperitoneal bleeding under anticoagulationtherapy. Retroperitoneal hematomas may be localized (e.g., to the perirenal space) or may involve
a
the entire retroperitoneum, depending on the primary lesion and the extent of the collection. Clinical suspicion of retroperitoneal hematoma (e.g., trauma, vascular disease, or a falling hematocrit in anticoagulated patients) or indeterminate retroperitoneal fluid seen at ultrasound requires further evaluation by CT. CT will also influence treatment planning (percutaneous drainage vs. open surgery) and is useful for evaluating response to conservative therapy. Contrast -enhanced CT in the arterial or portal phase can be used to demonstrate active bleeding, which in many cases will mandate immediate therapy, such as surgery or interventional radiological treatment by embolization therapy.
b
Fig. 16.20 (a) Retroperitoneal hematoma with internallayering due to separation between blood serum and cellular blood components. A clotted portion is seen laterally (wide arrow). Note that there is contrast extravasation signifying an active hemorrhage. It arises posteriorly (arrowheads), seeps through and layers on
CT Morphology
Hematomas appear as masses of varying size whose CT attenuation depends largely on the age and size of the collection. Hemorrhage in acutely bleeding patients may be is 0 attenuating to soft tissue (see Fig. 7.12d) or to be more precise to the blood in the aorta. This is the case if fresh blood has not yet separated into its solid and fluid components. Very rapidly, however, sedimentation effects or clot formation occur. Sedimentation causes layering of blood components with a gradient towards the dependent portions of the hemorrhagic material (Fig.16.20a; see also Fig.7.12c). Fluid-fluid levels may occur. In general, sedimentation indi-
c top of the denser portions of the hemorrhage (arrow). There is marked displacement of bowel loops, (b. c) Old retroperitoneal hematoma with typical hyperattenuating center on the noncontrast scan (b) and rim enhancement due to granulation tissue on the postcontrast scan (c). There were no signs of infection.
16 Peritoneal Cavity and Retroperitoneum cates lack of coagulation within the blood collection. Clot formation leads to typical hyperattenuating regions within the hemorrhage (up to 70 HU), which often are ill-defined and rounded in appearance (see Fig. 7.12a). Clot formation indicates that simple drainage will not suffice for removing the hemorrhagic collection. Serous components in acute hemorrhage are hypoattenuating and coexist with hyperattenuating regions. With increasing duration of the hemorrhage serous components are absorbed, and the hemorrhagic area may appear homogeneously or inhomogeneously hyperattenuating. Contrast extravasation in to a hemorrhagic area indicates active bleeding (Fig.l6.20a). It can best be seen if the CT examination has been performed similar to a CTA. A second portal phase scan may be used to confirm increase in size of the actively bleeding focus. The site of extravasation may be used to guide surgical or interventional radiological treatment. Recurrent hemorrhage can be suspected if there is marked heterogeneity in the hemorrhagic area. Large hemorrhages tend to form separate collections due to fibrin strands. As the collection matures, its CT attenuation decreases to near-water density. Hematomas do not enhance after intravenous contrast administration. With increasing resorption of the hemorrhage there will be a contrast -enhancing rim (Fig.l6.20b) that should not be mistaken for superinfection. The only secure sign of superinfection is the presence of gas bubbles in patients who have not had any type of percutaneous puncture. A primary retroperitoneal hematoma may become an intraperitoneal hemorrhage if it ruptures the posterior parietal peritoneum. Hematomas of sufficient size can displace and compress adjacent organs. Unilateral hematomas of the iliopsoas muscle are initially manifested by expansion of the muscle, often without a significant increase in CT density. A chronic hematoma may be confused with an abscess or necrotic mass, and percutaneous needle aspiration may be necessary for definite diagnosis. Urinoma
Injuries of the renal collecting system and lower urinary tract (traumatic, iatrogenic) may allow urine to extravasate into the perirenal fat along
the ureter or collect at perivascular sites. Weeks or years may pass between the injury and the appearance of clinical symptoms. Infection of the urinoma leads to retroperitoneal abscess formation. CT is a very sensitive tool for detection of urinoma but requires late scans after the injection of contrast material for detection of persistent urinary leaks. •
CT Morphology
Any indeterminate postoperative fluid collection detected in a perirenal, periureteral, or perivascular location should be imaged on delayed scans (> 15 min) after contrast administration. Contrast extravasation into the fluid confirms the diagnosis of persistent urinary leak (Fig. 16.2-1). Chronic urinoma, however, may no longer be accompanied by a contrast extravasation and may require needle aspiration for diagnostic confirmation. Old, liquefied hematomas may be indistinguishable from chronic urinomas by CT morphology alone. CT attenuation around 30 HU on pre-contrast scans, and rim enhancement following IV contrast administration should raise suspicion of a superinfection. The absence of an enhancing rim does not exclude infection in the acute stage.
Fig. 16.21 Urinoma resulting from injury to the renal pelvis during the removal of a postoperative drain. Opacified fluid is visible around the kidney 30 minutes after contrast administration.
Lymphoceie
Fig. 16.22 Lymphoceles. (a) Typical bilateral Iymphoceles of water density following lymphadenectomy (metal clips), (b) Atypicallymphocele that appears asawell-encapsulated, nonhomogeneous, nonenhancing mass containing components between -20 HU and 35 HU. The mass is compressing the ureter.
a
b
Cysts
Lymphocele
Cysts in the retroperitoneum are rare. Congenital cysts may arise from the bowel or urogenital tract but are less common than acquired posttraumatic or postinflammatory cysts, which are usually in direct contact with the site of the primary lesion (e.g., the pancreas). Differentiation is required from old hematomas and urinomas and must rely on the clinical presentation in addition to imaging findings.
Chylous fluid collections most commonly occur after lymphadenectomy for prostatic or testicular cancer and occasionally for upper abdominal tumors. They are also seen in patients after kidney transplantation, where they may cause secondary renal obstruction due to compression effects on the ureter. Lymphoceles are usually an incidental CT finding during follow-up of tumor patients.
•
•
CT Morphology
Cysts appear as masses of varying size that have smooth margins, are typically well encapsulated, and are usually of water attenuation, Cysts do not enhance on postcontrast scans and do not show rim enhancement. Teratogenic cysts and dermoid cysts include various embryonic elements and may contain calcifications oreven teeth.
Solid Retroperitoneal Retroperitoneal
CT Morphology
Lymphoceles appear as round or oval masses of water density along abdominal lymphatic pathways (Fig. 16.22a). Heterogeneous areas of nearfat attenuation may also be present (Fig.l6.22b). The mass usually has a very thin wall that cannot be distinguished on CT images and does not enhance with contrast medium. Surgical clips are often found adjacent to the mass.
Processes
Fibrosis
Retroperitoneal fibrosis may occur in a primary form (Ormond's disease), which probably has an autoimmune pathogenesis and is responsive to corticosteroids, or in a secondary symptomatic form (Ormond's syndrome), which can result from inflammation, neoplasms, aortic aneurysms, radiotherapy, drug ingestion, or trauma. Both forms lead to fibrotic induration of
the retroperitoneum. The fibrotic process may involve blood vessels, the ureters, the kidneys, and retroperitoneal portions of the bowel and biliary tract. An acute inflammatory stage is distinguished from a chronic fibrotic stage. Retroperitoneal fibrosis is most prevalent in middle age and predominantly affects males. The malignant form of retroperitoneal fibrosis « 10%) is caused by infiltration of the retroperitoneum by malignant cells. This kind of
76
Peritoneal Cavity and Retroperitoneum
desmoplastic malignant reaction of the retroperitoneal fat is associated with breast, stomach, colon, and lung cancer, and Hodgkin's disease, lymphoma, carcinoid, and sarcoma. CT is used to investigate unexplained hydronephrosis, exclude retroperitoneal masses of other origin, and evaluate response to steroid therapy. However, the differentiation of inflammatory tissue and mature fibrosis is superior with MRI. •
CT Morphology
Retroperitoneal fibrosis presents as masses of soft-tissue density enveloping the aorta and vena cava that may be nodular or plaque-like. It is frequently accompanied by proximal ureteral dilatation. The margins of the mass may be sharply defined or indistinct (Fig. 16.23). Aside from medial deviation of the ureters, no displacement or invasion of retroperitoneal structures will be found in most patients. Retroperitoneal fibrosis usually covers the anterior part of the great vessels and extends to the lateral wall of the vena cava and aorta. In the primary form, the posterior wall of the aorta is typically preserved. The soft-tissue process tends to progress in a caudocranial direction along the midline. The process commonly begins in the area of the bifurcations of the aorta and inferior vena cava. In the commoner limited form of disease the fibrous mass is centered over the fourth or fifth lumbar vertebra. In the rarer extensive form, the fibrotic process may extend downward, spreading along the ureters and ileum into the lesser pelvis (Fig.16.23b), laterally beyond the outer
edge of the psoas muscles and upward through the diaphragm and into the mediastinum. Atypical manifestations of retroperitoneal fibrosis occur mainly in extensive disease. Lesions may appear as a huge mass and may mimic retroperitoneal lymphoma or other malignancies. The renal hilum may be involved. Encasement of retroperitoneal vascular structures can lead to stenosis and thrombosis. Differentiation of the malignant form of retroperitoneal fibrosis is often impossible based on morphological CT signs. An anterior displacement of the aorta due to extension around the posterior wall is only seen in malignant disease. The absence of multifocal, usually asymmetric lymph node enlargement distinguishes retroperitoneal fibrosis from lymphoma. Primary retroperitoneal tumors usually occur at eccentric sites and show a propensity for invasion and metastasis. The detection of cystic areas is more consistent with a malignant tumor. On non contrast CT images, the attenuation values are similar to those of muscle tissue and the mass is very difficult to distinguish from vessels. Therefore IV contrast administration is required. Scans in the late arterial or portal phase best demonstrate contrast uptake in acute inflammation while scans in the interstitial phase of contrast enhancement show contrast uptake in areas of fibrosis. In the secondary form of retroperitoneal fibrosis, it may be possible to demonstrate the causative lesion (tumor, aneurysm) after contrast injection. Caution is required for application of contrast media in patients with renal failure.
Fig. 16.23 Retroperitoneal fibrosis. (a) The soft tissue mass surrounding the aorta is relatively sparing of its posterior aspect. a sign that is more often seen in primary retroperitoneal fibrosis. The mass has streaky, iII-defined margins and has caused medial deviation of the ureters. Subtle contrast enhancement can be seen in the interstitial phase, indicating a chronic fibrotic stage, (b) The symmetrical, plaque-like presacral mass that has encased the veins, arteries, and ureters extends from the small pelvis to the thoracoabdomina Ijunction, indicating an extensive disease.
a
b
Primary Retroperitoneal Tumors Pelvic Fibrolipomatosis
Pelvic fibrolipomatosis is a disease of unknown etiology in which increased fatty tissue deposition predominates over fibrous and inflammatory components. Men from 25 to 60 years of age are affected much more often than women. The clinical features are nonspecific, and the disease is usually discovered incidentally at CT. •
CT Morphology
The small fibrous component causes the pelvic mass to display predominantly fat density on CT scans. The mass causes displacement, elevation, and narrowing of pelvic soft-tissue structures (rectum, sigmoid colon, bladder, ureters) but no effective compression. There might be mild dilatation of the upper urinary tract, however. The rectum appears elongated, straightened and fusiformly narrowed. The ureters are displaced medially. The lipomatous mass may completely fill the pelvis and can expand the presacral space to more than 10 mm width (Fig. 16.24). Pelvic lipomatosis can be associated with cystitis glandularis or cystitis cystica. Primary Retroperitoneal Tumors
Retroperitoneal neoplasms may arise from any of the three germ layers (Table 16.8). The origins Table 16.8
Classification
of primary retroperitoneal
Mesenchymal (40-80%)
Fig. 16.24 Fibrolipomatosis in a patients with Crohn's disease causes anterior displacement. elongation, and straitening of the rectum (arrowhead). and anterior displacement of the bladder. The presacral space is markedly expanded and filled with a mass offat density with a small fibrous component. The ureters are displaced medially (arrows).
of such tumors are muscles, fasciae, loose connective and fatty tissue, blood and lymphatic vessels, nerve trunks and sheaths, heterotopic tissue and embryonic remnants. Mesenchymal tumors are the commonest, followed by neurogenic, dysontogenic, and epithelial tumors. At least 80% of all primary retroperitoneal neoplasms are malignant, and benign lesions have a strong propensity for malignant transformation.
tumors
Benign tumors
Malignant
Lipoma Leiomyoma
Liposarcoma Leiomyosarcoma Malignant fibrous histiocytoma Fibrosarcoma Rhabdomyosarcoma Lymphangiosarcoma Hemangiosarcoma
Fibroma Rhabdomyoma Lymphangioma Hemangioma Hemangiopericytoma Xanthogranuloma Neurogenic
Neuroma Neurofibroma Ganglioneuroma Sympathicoblastoma Benign neuroblastoma Paraganglioma Pheochromocytoma (extra-adrenal)
(10-50%)
Dysontogenic
(5-25%)
and epithelial
Teratoma Chordoma Urogenital vestigial tumors
tumors
Neurofibrosarcoma Malignant ganglioneuroma Malignant Sympathicoblastoma Neuroblastoma Malignant paraganglioma Malignant pheochormocytoma Malignant teratoma Malignant chordoma
620
16 Peritoneal Cavity and Retroperitoneum
Neuroblastomas and embryonic rhabdomyosarcomas are most prevalent in children, while neurogenic and teratogenic tumors are commonest in young adults, An increased incidence of neurogenic tumors has been documented in patients with von Hippel-Lindau disease, tuberous sclerosis, multiple endocrine neoplasia (MEN syndrome), and familial neurofibromatosis. Sarcomas have a peak incidence between 50 and 60 years of age. The CT features of these neoplasms depend on the size and origin of the tumor, and the clinical manifestations are largely nonspecific. N onfunctioning tumors are usually diagnosed at a late stage. Good tumor delineation in the absence of metastases does not exclude a malignant process, and therefore the diagnosis must be confirmed by biopsy and histological evaluation. Invasive growth, bone destruction, and distant metastases (lung, liver) are unmistakable signs of malignancy. CT is used to establish the identity of indeterminate retroperitoneal masses and to stage known tumors. The differentiation of retroperitoneal masses is based on the age of the patient, the location of the mass, laboratory findings, and CT morphology. Precontrast CT images are helpful for demonstration of fatty components and calcifications. Contrast media application is necessary for evaluation of the vascular architecture. MRI can be used in place ofC T owing to its good tissue discrimination. MRI is particularly useful in children and young adults. M ultislice CT offers improved tumor delineation by its multiplanar imaging capabilities. The commonest benign and malignant retroperitoneal tumors are described below.
Neuroblastoma Neuroblastomas have approximately the same incidence as Wilms' tumor (nephroblastoma), accounting for 7-10% of all pediatric malignancies. Fifty percent of children are affected before 1 year of age and 90% before age 7. As derivatives of primitive sympathetic ganglion cells, neuroblastomas develop in the region of the sympathetic trunk and adrenal medulla. The commonest site of occurrence is the retroperitoneum (75%), followed by the mediastinum ("15%) and the neck and pelvis (5% each). Neuro-
blastomas metastasize early to the lymph nodes, bone marrow, skeleton, liver, lung, brain, and skin. The tumors almost always secrete catecholamines, which can be detected in the urine.
cr
Morphology
CT reveals a large, nonhomogeneous mass in the adrenal bed or paravertebral area in the region of the sympathetic trunk. Up to 50% of the lesions contain areas of hemorrhage and necrosis, and 30-70% contain calcifications. Mass effect is apparent on adjacent organs. Unlike nephroblastoma, the tumor deforms the kidneys without causing renal destruction. Advanced disease is manifested by retroperitoneal lymphadenopathy and distant metastases to the liver and bones. Invasion of the spinal canal with erosion of the pedicles may be present. Involvement of major retroperitoneal vascular structures may occur (see Fig. 17.8).
Rhabdomyosarcoma Rhabdomyosarcoma is the commonest softtissue sarcoma in children, accounting for approximately 5% of all pediatric malignancies and showing a peak incidence between 2 and 6 years of age. The principal sites of occurrence are the head, neck, urogenital tract, trunk, and extremities. CT Morphology
Masses of muscle attenuation involving the urogenital tract or the muscles of the trunk should raise suspicion of rhabdomyosarcoma. The masses may appear nonhomogeneous due to intralesional hemorrhage and necrosis. The diffuse tumor vascularity causes the masses to enhance intensely after IV contrast administration. Retroperitoneal lymphadenopathy is usually present as an associated finding.
Teratoma Teratomas occur mainly in the gonads, but a few lesions (approximately 4%) may have an intracranial, mediastinal, or retroperitoneallocation. A distinction is drawn between mature and immature teratomas.
Primary Retroperitoneal Tumors •
CT Morphology
The diverse tissue components of teratomas may give them a nonhomogeneous CT appearance. The coexistence of fat, soft-tissue elements, bone, or teeth within a mass is considered pathognomonic. Malignant teratomas may invade the inferior vena cava and may remain relatively homogeneous, and thus have a nonspecific appearance on CT (Fig. 16.25). Lipoma
Lipoma is the commonest benign mesenchymal tumor, showing a peak incidence in the fifth and sixth decades. The tumor may have a superficial (subcutaneous) or deep site of occurrence (retroperitoneum, chest wall, hands, feet). Fewer than 10% oflipomas are multifocal. •
CT Morphology
Lipoma appears as a well-circumscribed, homogeneous mass of fat density (approximately -100 HU). Unlike liposarcoma, the mass does not contain soft-tissue components and does not enhance with intravenous contrast. Intramuscular lipoma very rarely shows locally invasive growth with separation of the surrounding muscle planes [invasive lipoma).
621
Liposarcoma
Liposarcoma is the second commonest malignant retroperitoneal soft -tissue tumor in adults. Like lipoma, it has a peak incidence in the fifth and sixth decades of life. Four histological types are recognized, the myxoid type (40- 50%) being the most common. The tumor presents as a painful mass in only about 10-15% of cases. Ten percent of the myxoid lesions show multicentric involvement. Low-grade liposarcomas have a high recurrence rate but rarely metastasize, while high-grade liposarcomas frequently metastasize to the liver, lung, and bone. •
CT Morphology
The mass may consist predominantly of fat or soft-tissue components, depending on its grade and histology. A lipomatous mass is very likely to be a liposarcoma if it contains streaky densities or solid areas of soft -tissue density in addition to areas of pure fat density (Fig. 16.26). Areas with negative attenuation values that exceed fat density (as compared to other region of abdominal fat) are also suspicious for liposarcoma. Well-differentiated tumors contain large fatty components and tend to have expansile rather than infiltrative margins, whereas poorly differentiated myxomatous, round-cell or pleomorphic liposarcomas have large soft-tissue components and are frequently invasive. Only a few liposarcomas contain no grossly detectable fatty components, and all but a very few show
b Fig. 16.25 (a) Benign teratoma in a 8-year-old girl. There is fat, soft tissue, and ossified portions (arrows) interspersed in this large mass. The small bowel is displaced to the left. (b) Malignant teratoma that presents
as a slightly heterogeneous left-sided pre- and paravertebral mass causing erosion of the anterior vertebral margin.
16 Peritoneal Cavity and Retroperitoneum
a Fig. 16.26 Liposarcoma. (a) CT shows a nonhomogeneous, lipomatous mass that contains streaky components of soft-tissue density in addition to areas of fat density. The mass is expansile and infiltrative. (b) This
liposarcoma appears as an encapsulated mass in the left lower quadrant containing a combination of fat and streaky soft-tissue components.
negative attenuation values. The nonfatty portions of liposarcomas may show contrast enhancement. Calcifications are found in up to "12% of the tumors. It may be difficult to differentiate between an extrarenal angiomyolipoma and a liposarcoma of the retroperitoneum, especially when the tumor is large. In the early stage liposarcomas are most often confined outside the perirenal fascia, while angiomyolipomas grow in the perirenal fat. A fat line between the mass and the kidney, compression of the renal surface and preservation of the renal cortex are suggestive for a liposarcoma, while discontinuity of the renal cortex and a spur sign are suggestive for angiomyolipoma.
•
CT Morphology
Soft -tissue involvement is characterized by masses of muscle density that contain hypodense necrotic areas and frequently measure at least 5 cm in diameter. The masses generally show intense contrast enhancement. The lesions may infiltrate the muscles of the back or abdominal wall, especially the psoas muscle (Fig. 16.27). Scans occasionally demonstrate osseous involvement with periosteal reaction and cortical erosion. Invasion of the inferior vena cava or renal vein usually is not found. This distinguishes malignant fibrous histiocytoma from advanced renal cell carcinoma.
Malignant Fibrous Histiocytoma
Malignant fibrous histiocytoma is the commonest primary malignant soft-tissue tumor of the extremities and retroperitoneum in adults. The average age at diagnosis is 50 years, but all age groups are affected. Primary soft -tissue involvement by the tumor is distinguished from skeletal involvement and from pulmonary involvement, which is very rare. Many patients clinically present with a painless mass that has been evident for several months. The tumors have a high recurrence rate after resection, and 50% have already metastasized (lung, liver, bone, lymph nodes) by the time of diagnosis.
Fig. 16.27 Malignant fibrous histiocytoma that has infiltrated the left psoas muscle (arrow heads) and has led to an anterior displacement of the kidney. The tumor is heterogeneous with hypervascular areas and areas of necrosis.
Secondary
Leiomyoma,
Leiomyosarcoma
These tumors arise from blood vessels, the spermatic cord, or embryonic rests and are seldom benign. They most frequently metastasize to the liver, lung, mediastinum, and soft tissues. •
CT Morphology
The tumors appear on CT as a nonhomogeneous, peripherally enhancing mass. Because of their rapid growth, 75% of the tumors contain areas of necrosis, cystic degeneration, and hemorrhage (Fig. 16.28). Calcifications are rare. Secondary Retroperitoneal Masses and Lymphadenopathy
Secondary retroperitoneal masses result from tumor invasion or tumor seeding into the retroperitoneum. The retroperitoneal lymph nodes may be involved by systemic lymphatic diseases (Hodgkin's disease, non-Hodgkin's lymphoma) or by tumors of the pancreas, kidneys, stomach, or lesser pelvis (testes, ovaries, uterus, bladder, and prostate). Inflammatory processes can also cause changes in the retroperitoneal lymph nodes. CT is the method of choice for the detection of retroperitoneal lymph nodes but its accuracy to detect abnormality in normal-sized lymph nodes is limited.
•
Retroperitoneal
Masses and Lymphadenopathy
CT Morphology
Tumor invasion into the retroperitoneum is seen in large tumors, especially of the kidneys, adrenals, pancreas, and retroperitoneal bowel structures. Usually the primary tumor site is well-delineated and differentiation from primary retroperitoneal tumors poses no problems. Normal para-aortic or paracaval lymph nodes are seen as small, round or oval structures in the fat adjacent to the great vessels. The diagnosis of lymphadenopathy at CT is mainly based on the size criterion, i.e., the minim um diameter on axial sections. Thresholds between benign and suspected malignant involvement vary depending on the anatomic location (Table 16.9; see also Tables 22.6, 22.7). Additional criteria are the presence of a fatty center (fibrolipomatous degeneration) that indicates a benign node. Inhomogeneities or strong enhancement after contrast administration suggests necrotic or hypervascularized nodes either due to tumor or infection. With multislice CT, the longitudinal extent and threedimensional shape of lymph nodes can be evaluated as well. Round nodes (maximum diameter < 1.5 x minim um diameter) suggest malignancy even though they may be smaller than the threshold size.
Table 16.9 Retroperitoneal lymphadenopathy: suggested thresholds for minimum (short axis) diameter Threshold diameter
Location Retrocrural
nodes
Gastrohepatic
Fig. 16.28 Leiomyosarcoma presenting as a septated, peripherally enhancing mass that shows foci of cystic degeneration. The mass cannot be differentiated from the inferior vena cava and has encased the aorta.
nodes
6mm 8mm
Pancreaticoduodenal
10mm
Mesenteric
10mm
High preaortic and celiac
10mm
Para-aortocaval
11 mm
nodes
77 Adrenal Glands M. Galanski
Anatomy Shape and Size (p. 626) Examination
Technique
Contrast Injection (p. 627) Patient Preparation (p. 627) Acquisition Technique (p. 627) Endocrine Diseases Cushing's Syndrome (p. 628) Conn's Syndrome (p. 629) Adrenogenital Syndrome (p. 630) Adrenocortical Insufficiency (Addison's Disease) (p. 630) Pheochromocytoma (p. 630) Nonfunctioning
Adrenal
lesions
Adrenal Cysts (p. 632) Adrenal Hematoma (p. 632) Adrenal Calcifications (p. 632) Myelolipoma (p. 633) Nonhyperfunctioning Adenoma (p. 633) Ganglioneuroma, Ganglioneuroblastoma (p. 634) Neuroblastoma (p. 634) Adrenal Carcinoma (p. 634) Adrenal Metastases (p. 635) Lymphoma (p. 635) Incidentally Detected Adrenal Masses (p. 636)
77 Adrenal Glands
The only primary indication for a selective CT examination of the adrenal glands is an abnormality of adrenal function (Table 17.1).In those cases, diagnostic evaluation usually starts with the clinical examination and laboratory tests since their findings will determine the need for imaging procedures and direct their selection. Because of its superiority for estimating the fat content of the lesion (in phase and opposed phase GRE images), MRI offers great potential, especially for the differentiation of benign and malignant enlargement of the adrenals. MRI used to have the further advantage of providing coronal imaging, a feature thatis also available with spiral CT. Multislice CT now provides even better quality. CT is the primary method for excluding metastatic disease in the adrenals. Frequently,
Table 17.1 Indications adrenal glands Impairment
for
CT
examination
of the
of adrenal function
Localizing diagnosis
Cushing's syndrome Conn's syndrome Pheochromocytoma
Adrenal mass Incidentaloma
DO benign/malignant
Tumor staging
Adrenal carcinoma Adrenal metastasis
however, the radiologist is confronted with incidentally detected adrenal lesions and is challenged to develop diagnostic strategies for differentiating benign from malignant lesions.
Anatomy
Because the adrenal glands are embedded in the retroperitoneal fat, they can be clearly visualized with CT in almost every patient. Occasionally fusion with the liver or with the renal capsule is encountered. Both adrenal glands are located in the upper part of the pararenal space, enclosed in the pararenal fascia. The right adrenal gland lies anteromedial and superior to the upper pole of the kidney, in the angle between the right lobe of the liver and the crus of the right diaphragm, just posterior to the inferior vena cava. The left adrenal gland lies posterolateral to the aorta and is more anteromedial than superior to the upper renal pole (Fig.17.1 a). Occasionally the lower pole of the gland extends very close to the renal vascular pedicle. Shape and Size
An axial CT scan through the adrenal glands shows each gland to consist of a body and two limbs. The right adrenal gland usually displays a linear, comma, or V shape, while the left adrenal is more often triangular, or Y-shaped. The contours of the glands are slightly concave and sharply marginated (Fig.17.1 b). Corticomedullary differentiation is sometimes possible with contrast-enhanced multislice CT. The adrenal glands measure 4-6 cm in their vertical dimension and 2-3 cm in width. The
~
right adrenal
left adrenal
Fig. 17.1 Anatomy of the adrenal glands, (a) Relationship to adjacent organs, (b) Variations in the shape of the adrenal glands.
limbs are 6-8 mm thick. Measurement of the adrenal glands has no significance because of the large range of variation. The qualitative assessment of shape, size, and contours is of greater diagnostic importance.
Contrast Injection Examination
Technique
No patient preparation is necessary for evaluation of the adrenal glands unless pheochromocytoma is suspected. In the presence of a known secreting pheochromocytoma alpha- and beta-adrenergic blockade prior to contrast material injection can be administered to prevent an adrenergic hypertensive crisis. Recently. it has been shown that specific blockade may not be required before contrast medium enhanced scanning with nonionic contrast media.
Thin-section scanning (3 mm or less) is required for detecting small endocrine tumors and for correct measurements of CT attenuation. Section thickness may be increased to 5 mm for malignant diseases or if the scan length has to be extended. Multislice scanning should employ 0.5-1.25 mm collimation on 4-16 slice scanners. Axial section of 2-3 mm thickness usually suffice for image review but coronal (and sometimes sagittal) reformations of 2-3 mm thickness may help to better detect multifocal lesions and improve delineation oflarge adrenal tumors in their relationship to adjacent organs.
Acquisition Technique
Contrast Injection
The technique for CT examination of the adrenal glands is guided strictly by the indication and the anticipated findings. This particularly applies to examinations that focus primarily on the adrenal glands. such as the evaluation of abnormal adrenal function (Table 17.2). With single-slice spiral CT, a low-dose (localizer) scan of the upper abdomen (7-10 mm section thickness) can be performed to provide a clue to the pathology to be expected. and to optimally define the scan range. This range only needs to include the adrenals in case of benign Cushing's or Conn's disease but should be expanded in malignant tumors or suspicious findings on the localizer scan. For localization of a suspected pheochromocytoma the abdomen should be imaged, at least down to the aortic bifurcation.
Imaging in the arterial phase (25-35 sec) is helpful in distinguishing the adrenals from adjacent structures. However, it is generally accepted that early contrast enhancement (arterial phase) does not contribute to the differential diagnosis of benign versus malignant lesions. The mean attenuation difference between nonadenomas and adenomas was found to be highest on scans acqulred 60-90 sec after injection of contrast material due to the faster wash-out of the contrast material in nonadenomas. Delayed scanning 15 minutes after injection of 150ml of contrast material (300 mg iodine! ml) was reported to have a sensitivity and specificity each over 95% in differentiating adenomas from nonadenomas.
Patient Preparation
Table
17.2
Protocol for CT seaming
ofthe
adrenal
glands
General Oral contrast
Unnecessary (adrenal SOOml, administered
Patient
Supine
position
with elevated
Adrenal
Scan range
region) 30min priorto
region:
Retroperitoneum: Respiratory
phase
(retroperitoneum)
chosen
according
from diaphragmatic
to a thick-section crura to aortic
localizer
scan
bifurcation
Inspiration (spiral C'T) Expiration
Windowing
examination
arms
(conventional
NoncontrastCT: Contrast -enhanced CT:
C'T) W/L=300/40 WiL = 400/60 Continue
>
627
628
17 Adrenal Glands Table 17.2
Continue
Scan parameters
l-slice SC/TF/RI
Adrenals (endocrine
2/3/2 disease)
L
Adrenal region (tumor staging)
3/5/3 t
Retroperitoneum (pheochromocytoma)
5/10/5
Contrast
injection
t
4-slice" SC
16-slice" SC
1-1.25t
0.5-1.25
'/6 t
0.75-1.25
2-3 t
1-2
t
L
MPRb SWIRl
3/2
3/3 cor
5/4
3/3 cor 4/4 sag
5/4
3/4 cor
4-slice
16-slice
N/F/D
V+N/F/D
V+N/F/Dc
100/3/30
100 + 50/3/30 15min
80 + 50/3/15A 15min
l-slice
Parenchymal phase Delayed phase (wash-out)
t
axial SWIRl
15min
SC = slice collimation (mm), TF = table feed (rnmjrctation), Rl = recon. increment (mm), i t = scanning direction, SW = effective section width (mm), MPR = multi planar reformats, axial = axial sections, cor = coronal, sag = sagittal CM = contrast material, Contrast concentration = 300 mg/ml iodine V = volume of CM (ml), N =volume of saline flush (ml), F= flow rate (mils), D = start delay, X = trigger region a Pitch P = TF/{NxSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SWIRl = 1-1.5/0.7 or 0.5-0.8/0.5 C read: 15A = start delay 15 s after reaching 50 HU enhancement in the desc. aorta
Endocrine Diseases Fig. 17.2 reviews (he morphologic changes in the adrenal glands that may be observed in endocrine diseases.
Cushing's Syndrome Endogenous Cushing's syndrome perproduction of corticosteroids)
(adrenal hyis due to a
pituitary overproduction of ACTH with a resulting adrenal hyperplasia in 70% of cases, to an autonomous production of cortisol by a neoplasm of the adrenal cortex in 15%, and to an ectopic ACTH production in a further 15%. In adrenal Cushing's syndrome, it is important to differentiate between adenoma and carcinoma. •
CT Morphology
are easy to detect with CT because they usually measure 2-5 em in diameter, and the obesity in Cushing's syndrome creates very favorable imaging conditions. The lesions appear as well-circumscribed round or oval masses of homogeneous density that show little enhancement after IV contrast administration (Fig. 17.3a). Portions of the gland unaffected by tumor and the contralateral gland appear atrophic. Areas of hemorrhage, necrosis, or calcification are rare and are seen only in large tumors. In Cushing's syndrome due to central hypothalamic-pituitary dysfunction (note that this condition does not represent a primary indication for cross-sectional imaging), the ACTH stirnulaCushing adenomas
Diffuse hyperplasia
Nodular hyperplasia
OO~A
Macroadenoma
Microadenoma
Pheochromocytoma
Fig. 17.2 Changes in adrenal morphology that may be found in endocrine diseases. Pheochromocytomas enhance inhomogeniously after contrast injection.
Conn's Syndrome
Fig. 17.3 (a) Adenoma in Cushing's syndrome appears as a homogeneous, sharply circumscribed, slightly hypoattenuating nodule. The "atrophy" affecting the remaining portions of both adrenal glands suggests an ACTH-independent autonomous adenoma rather than macronodular hyperplasia in central Cushing's syndrome, which also may produce adenoma-like nodules.
(b) Adrenal cortical hyperplasia in central Cushing's syndrome. The adrenal glands are enlarged and show no contour irregularities. Given the highly variable configuration of the adrenals, the diagnosis of a central Gushing syndrome requires corroborative evidence of an endocrine disorder.
tion of the adrenal glands induces a sometimes subtle bilateral adrenal hyperplasia (Fig.l7.3b). The shape of the adrenals remains in proportion. This secondary adrenal hyperplasia can be difficult to differentiate from endogenous tnecronodular adrenal hyperplasia, a rare condition that is also characterized by the development of adenoma-like nodules. Both with ACTH-dependent adrenal hyperplasia and with macronodular adrenal hyperplasia the whole gland is hyperplastic or irregularly thickened, but in patients with an autonomous adrenal adenoma the non tumor-bearing portions of the gland are atrophic. An adrenal Cushing syndrome may be caused by an adrenal cortical carcinoma, but in most cases the CT features of these tumors can be clearly differentiated from benign conditions.
scanning is presently the technique of choice and should rely on thin-section spiral or multislice techniques.
Conn's Syndrome
Conn's syndrome (hyperaldosteronism with the clinical hallmark of hypokalemic hypertension) is caused by a unilateral aldosterone-producing adenoma in 70% of cases and by bilateral idiopathic adrenal cortical hyperplasia in 30% of cases. This differentiation has key therapeutic implications, since adenomas are treated surgically while hyperplasia is managed with drug therapy. Because Conn adenomas are small, CT
•
CT Morphology
Adenoma detection is the key to diagnosis, because idiopathic hyperplasia (aside from rare macronodular forms) cannot be identified by its gross morphologic features on CT scans. Aldosterone-producing adenomas are generally less than 2cm in diameter, and some measure only a few millimeters. CT scanning with a thin slice collimation can consistently detect lesions as small as 10mm, especially if they are favorably located (i.e., in the peripheral part of the gland). They appear as small, round or oval, homogeneous masses with slightly hypodense CT attenuation values (Fig. 17.4). The detection of idiopathic hyperplasia is a more difficult task than the localization of adenomas. Only the rare macronodular forms allow a positive diagnosis of hyperplasia, but these lesions are apt to be misinterpreted as aldosterone-producing adenomas. Determination of the affected gland in Conn's syndrome relies on adenoma detection. Therefore the disease cannot be classified when scans show normal-appearing adrenals, even if there is strong clinical suspicion of idiopathic hyperplasia. The detection of bilateral adenomas also precludes lateralization. The evaluation of
77 Adrenal Glands
Fig. 17.4 Aldosterone-producing adenoma in Conn's syndrome. CT typically shows a microadenoma less than 2cm in diameter that appears as a well-defined, homogeneous, hypoattenuating nodule « 10 HU).
these cases requires bilateral adrenal venous sampling for differential hormone assay, or NP59 scintigraphy. Adrenogenital Syndrome
Primary adrenogenital syndrome (ACS) is based on a congenital enzyme defect in steroid synthesis resulting in absent or deficient cortisol and/or aldosterone production and excessive androgen secretion. The cortisol deficiency leads to constant ACTH stimulation of the adrenal cortex, resulting in adrenal hyperplasia. ACS perse is not an indication for CT examination. •
CT Morphology
CT demonstrates marked adrenal cortical hyperplasia in which the adrenal glands show coarse tumorous enlargement with preservation of normal shape. This type of change is not seen in any other adrenocortical endocrine disease. Adrenal cortical hyperplasia may be absent or mild in ACS patients who receive early replacement therapy. Adrenocortical Insufficiency (Addison's Disease)
Primary adrenocortical insufficiency (hypoadrenalism, Addison's disease) results from destruction of the adrenal cortex, while secondary
adrenocortical insufficiency reflects a deficiency of pituitary stimulation. The most frequent cause of pituitary insufficiency is a neoplastic disease. Primary adrenocortical insufficiency is most commonly due to autoimmune processes (idiopathic adrenocortical atrophy). Less frequent causes are specific or nonspeclftc granulomatous processes (tuberculosis, mycosis), hemorrhage, metastases, amyloidosis, hemochromatosis, lymphoma, etc. The condition is differentiated into chronic (more than 2 years), subacute (less than 2 years) and acute forms of insufficiency. At least 90% of the adrenal cortex must be destroyed before clinical manifestations of adrenal insufficiency appear. Imaging studies are used only in the etiologic investigation of primary adrenocortical insufficiency. •
CT Morphology
CT has no practical role in the diagnosis of idiopathic adrenocortical insufficiency, as the only CT change consists of bilateral atrophy in adrenals that otherwise appear normal. Autoimmune-based atrophy is the commonest cause of chronic insufficiency and is characterized by atrophy of the adrenals without calcifications. Active infectious-inflammatory and granulomatous processes (e.g., in AIDS patients with Mycobacterium evium, Mycobacterium intracellulare or CMV) cause subacute adrenal insufficiency with mild to moderate adrenal enlargement, which is usually bilateral and often symmetrical. The basic shape of the glands is unchanged, but frequently their contours are ill-defined. Heterogeneities may result from necrosis or abscess formation. Calcifications suggest a granulomatous infection (tuberculosis, histoplasmosis). Acute adrenal insufficiency is most frequently caused by adrenal hemorrhage seen after severe hypotension, shock, or septicemia. On CT the adrenals are enlarged and hyperattenuating owing to the presence of blood. Pheochromocytoma
Ninety percent of pheochromocytomas arise from the adrenal medulla and 10% from extraadrenal sympathetic tissues (paraganghomas). Ten percent of pheochromocytomas are bilateral, and 10% are malignant.
Pheochromocytoma Extra-adrenal pheochromocytomas occur predominantly in the retroperitoneal space or pelvis (Organ of Zuckerkandl around the aortic bifurcation). Only 3% are supradiaphragmatic, where they are more often mediastinal than cervical. Intracardiac and paracardiac lesions have been described (see Fig. 10.10). The diagnosis is established by laboratory tests (urinary catecholamine excretion, serum levels of epinephrine and norepinephrine). Imaging procedures are used only for preoperative localization. Contrast media administration is therefore often unnecessary, especially since it may trigger an adrenergic hypertensive crisis in some patients. •
CT Morphology
Because almost all pheochromocytomas are larger than 2 cm when diagnosed, CT can easily detect tumors located at a typical adrenal or juxta-adrenal site in the retroperitoneum. CT demonstrates a well-circumscribed round or irregularly shaped mass. Small tumors are relatively homogeneous, but larger tumors are more likely to contain areas of hemorrhage, necrosis, cystic changes, and calcifications (3-10%). The intense enhancement of pheochromocytomas serves to distinguish them from adrenal adenomas (Fig.17.5). MRI and specific radionuclide studies (iodine-131 meta- iodo benzy lguanldtne (MI BG) scintigraphy) are superior to CT in screening for extra-adrenal tumors. • Generally differentiation between benign and malignant tumors cannot be made, although metastasis is a reliable criterion for malignancy. • Pheochromocytomas can mimic some of the symptoms of a neurocutaneous syn-
Fig. 17.5 Pheochromocytoma containing areas of regressive change (necrosis and calcification). Only the peripheral parts of the tumor show typical intense enhancement. Note: intravenous contrast is not essential in patients with biochemically confirmed pheochromocytoma. If required, it should be used with premedication (receptor blockers).
drome or multiple endocrine neoplasia (MEN) syndrome. Bilateral and ectopic pheochromocytomas are frequently associated with MEN. Localization is particularly difficult in MEN syndrome type II (medullary thyroid carcinoma, parathyroid adenoma, pheochromocytoma) and type III (medullary thyroid carcinoma, pheochromocytoma, ganglioneuroma, marfanoid body habitus), because the tumors are often small and bilateral and develop from medullary hyperplasia, which cannot be detected with imaging studies. The incidence of pheochromocytoma is increased in patients with von Hippel- lindau syndrome (5-15%) and neurofibromatosis(5%).
Nonfunctioning Adrenal Lesions
This group consists of lesions that do not cause hormonal abnormalities or other symptoms and therefore are often discovered incidentally during examinations performed for other indications. They include nonhyperfunctioning adenomas, cysts, hematomas, granulomatous changes, myelolipoma, and rare instances of
metastases and primary adrenal carcinoma (Table 17.3). Only some of these lesions have CT features specific enough to suggest a spot diagnosis or presumptive diagnosis. These include myelolipomas, cysts, hemorrhage, and coarse calcifications without a mass effect.
631
77 Adrenal elands Table 17.3 findings Benign
Differential diagnosis of incidental adrenal
lesions
Hypodense(O-10HU).no enhancement
Cysts
Hematomas
Hyperdense (>60HU), no enhancement
Calcifications
Very hyperdense
Silent
tumor (irregular wall thickening), The presence of irregular wall thickening or contents that show a density greater than that of water do not support the diagnosis of a simple cyst Adenomas, adrenal carcinomas, and pheochromocytomas (determination of urinary catecholamine levels) may also represent as cystic tumors,
No definite criteria «10 HU, <3cm)
Adrenal Hematoma
Myelolipoma
Fatty components «-10 HU)
Granuloma
No
Adrenal hemorrhage is the commonest cause of an adrenal mass in neonates, If bacteremia occurs, an abscess may develop (4-6 weeks after birth), In adults hemorrhage in the adrenals is rare, It is seen after severe hypoperfusion, hypotension, shock, burns, under anticoagulation therapy in coagulopathies, and after trauma, It is frequently associated with centrilobular liver necrosis and renal tubular necrosis, Extensive hemorrhage associated with septicemia is called Waterhouse-Friderichsen syndrome, Traumatic hemorrhage is mostly unilateral and right sided,
adenoma
definite criteria (calcifications may be present)
Malignant lesions Neuroblastoma
Age <5 years, (80%), large soft-tissue heterogeneous
Adrenal
carcinoma
component, enhancement
definite criteria (s-eem, ill-defined margins, No
heterogeneous ment)
Metastases
calcifications
enhance-
No definite criteria (>5 em, >10H)
Adrenal Cysts Adrenal cysts are very rare, They may be classified by their pathogenesis as endothelial (45%), epithelial (9%), parasitic (Echinococcus, in 7%), or pseudocystic (40%), The last represents the end stage of a remote hemorrhage, CT Morphology
The typical CT appearance is that of a cystic mass with smooth, uniformly thin walls and contents of fluid attenuation, Many cysts display a calcified rim, Wall thickness seldom exceeds 3 mm, Rim enhancement but no intralesional enhancement can be seen. On unenhanced CT scans it may be distinguish between a benign cyst attenuation adenoma, Ultrasound adjunct for confirming that the liquid,
difficult to and a low is a useful content is
The differential diagnosis of cystic lesions should include the possibility of regressive changes in a
CT Morphology
On CT the adrenals appear round or oval having a similar shape to those with adenomas or metastases, Adrenal hemorrhage can be diagnosed with high confidence when it is fresh and displays high attenuation values, but with increasing delay between the bleeding and the CT examination attenuation values decrease. As in cysts but unlike solid tumors, contrast enhancement is not seen, Older hematomas often appear cystic, in which case they are indistinguishable from pseudocysts of other etiology,
Adrenal Calcifications Pronounced calcifications in the adrenal gland most commonly result from adrenal hemorrhage in early childhood, They may also be residual to adrenal tuberculosis or a different granulomatosis, Tumor-associated calcifications are easily recognized by demonstrating the accompanying mass, Calcifications are present in up to 80% of neuroblastomas, up to 30% ofcarcinomas, and up to 10% of pheochromocytomas,
Nonhyperfunctioning Adenoma Myelolipoma
Adrenal myelolipoma is a benign developmental tumor (hamartoma) that contains varying proportions of fat and hematopoietic tissue elements. In 10% of patients this is associated with endocrine disorders like Cushing's or Conn'ssyndrome. •
CT Morphology
Characteristic feature of these lesions is the presence of fat, which is seen in almost all myelolipomas (Fig. 17.6). The proportion of fat to bone marrow cells varies, and tumors may be completely fatty or retain only a few foci of fatty elements. The myeloid elements show higher precontrast attenuation and significant enhancement, while the enhancement of the fatty structures is minimal. The size of the well-circumscribed tumors varies between some millimeters and 30cm, with Scm being the average. Retroperitoneal and intralesional hemorrhage has been described. In "10% of patients bilateral lesions are seen, and up to 30% of lesions have calcifications that are often punctate. Myelolipomas with a high fat content are easily diagnosed, and only lipoma (fat, no other soft tissue) and adenoma (CT attenuation usually >-10HU) need be considered in the differential diagnosis. With more inhomogeneous appearance, the further differential diagnosis includes a liposarcoma or an upper pole renal angiomyolipoma. The larger the myeloid component, the more difficult it is to identify the tumor. Tumors
Fig. 17.6 Typical near-fat adrenal myelolipoma.
CT attenuation
values
of
that contain very little fat cannot be identified by their CT features, and biopsy is required. Nonhyperfunctioning
Adenoma
At least 30% of all adrenal masses are incidental findings. Nonhyperfunctioning adrenal cortical adenomas have no pathologic significance. If such a lesion is discovered incidentally, e.g., on a CT scan of the abdomen, it is called "incidentalorna". For differentiation between benign and malignant disease see incidentally detected adrenal masses (p. 636). •
CT Morphology
These adenomas can vary greatly in size (O.S-Wcm), but most are smaller than 3cm. They appear as round or oval masses with welldefined margins. The shape of the adrenals is preserved. Adenomas are usually homogeneous and show CT attenuation lower than or equal to that of muscle. CT attenuation below 10 HU on non-contrast scans is considered highly predictable of adenoma, while attenuation above this cutoff level is considered indeterminate (10-40% of lesions, lipid -poor adenomas). Inhomogeneities and calcifications are rare. Contrast enhancement may vary but is usually high initially (more than nonadenomas) and rapidly washes out. Most adenomas have a density below 30 HU (> 40% de-enhancement within 15 min) on delayed scans 1 h after contrast administration (Table 17.3;Fig. 17.7).
Fig. 17.7 Silent adenoma (incidentaloma) has exactly the same CT features as functioning adrenal cortical adenomas. Incidentalomas are elliptical, sharply circumscribed, usually measure below 10 HU.
633
77 Adrenal Glands Ganglioneuroma,
Ganglioneuroblastoma
Canglioneuroblastoinas are generally located in the retroperitoneum (65%), otherwise in the neck, mediastinum, or in the adrenals. Ganglioneuroblastoma is a complex tumor containing both mature ganglion cells and primitive neuroblastic cells. They are seen in children under 10 years of age. Ganglioneuromas are benign tumors that arise from the sympathetic chain or rarely from the uterus, ovary, skin, or GI tract. Adrenal ganglioneuromas occur predominantly between the third and fifth decade. •
CT Morphology
Canglioneuroblastoinas show similar imaging features to neuroblastomas, and require histological proof. Adrenal ganglioneuromas are indistinguishable on CT from adrenal carcinomas, but no signs of local or vascular extension are seen. Specific diagnosis requires biopsy or surgical removal.
infiltrate the bone marrow in such a Way as to be confused with leukemia. The goal of cross-sectional imaging is to stage local extent and metastasis, for which MRI is the current imaging modality of choice. Multislice CT probably will yield similarly good results. •
CT Morphology
Most neuroblastomas display a heterogeneous tumor matrix with areas of necrosis and hemorrhage. CT shows calcifications in up to 80% of tumors (Fig. 17.8), which are finely stippled and linear or ring shaped. Solid tumor portions show moderate contrast enhancement. Local staging is based on tumor size, location, extent, spread across the midline, and supradiaphragmatic involvement. Invasion of the kidneys by neuroblastoma may make it difficult to distinguish this from nephro blastoma (Wilms" tumor). The tumors frequently metastasize to the liver, bone, and lymph nodes. Adrenal Carcinoma
Neuroblastoma
Neuro blastoma is the commonest pediatric abdominal malignancy. The peak incidence is in the second year of life, with 75% of cases occurring before age 2; the remainder are found at all ages. They originate from the adrenal medulla and from the sympathetic ganglia. The origin of large tumors cannot always be determined (see Chapter 12). The tumor may be quite highly differentiated and contain ganglion cells (ganglioneuroblastoma) or be undifferentiated and
Primary adrenal carcinomas are very rare. They remain clinically silent for some time before they are detected by their mass effect or endocrine symptoms (Cushing'ssyndrome, virilization). By that time, many of the tumors have already metastasized. There are two age peaks: before the age of 5 years and between 40 and 50 years. •
CT Morphology
Adrenal carcinomas are large (90% > 6 cm; up to 25cm) at the time of clinical presentation. Thus, while the tumor is easily detected by CT, it can
Fig. 17.8 Neuroblastoma of the right adrenal gland. Calcifications are visible on the precontrast scan (a). On the coronal MPR (3/5/2) after contrast administration, the tumor displaces the kidney downward but is clearly demarcated from it (b).
a
b
Lymphoma be difficult to assign the lesion to a specific organ and delineate it from adjacent organs. Differential diagnosis is aided by the fact that adrenal carcinomas displace and deform adjacent organs for a long time before infiltrating them. Large carcinomas have irregular margins and are heterogeneous due to areas of regressive change (necrosis, hemorrhage). Solid tumor components show marked contrast enhancement (Fig. 17.9). Calcifications are observed in 25-30% of adrenal carcinomas. The distinction between benign and malignantcortical neoplasms is virtually impossible if no signs of infiltration or metastases are present. Large size of tumors, necrosis, and intratumoral hemorrhage are suggestive for malignancy. It can also be difficult to differentiate an adrenal cortical neoplasm from a renal cell carcinoma, since both show similar enhancement characteristics, and patterns of metastatic spread. Here, multiplanar reformations, especially from thin-section helical or multislice CT are particularly helpful.
635
geneous, while larger metastases are heterogeneous due to regressive changes caused by hemorrhage or necrosis (Fig. 17.10). Generally the lesions enhance after IV contrast administration (Table 17.4). Calcifications are rare but do occur in mucin-producing tumors. Approximately 50% of adrenal metastases are bilateral. Adrenal glands that appear grossly normal or hyperplastic (with their basic shape unchanged) do not rule out metastases (see Fig.9.10b). A differentiation between benign and malignant lesions cannot always be made (see discussions below). Lymphoma
Primary adrenal lymphoma is extremely rare. Secondary involvement of the adrenals is more frequent (in approximately 25%). The most common cause is NHL.
Table 17.4 Incidentally detected adrenal masses
Adrenal
Metastases
The tumors that most frequently metastasize to the adrenal gland are breast carcinoma (55 %) , bronchial carcinoma (35%), renal cell carcinoma (25-40%), and melanoma (60%). •
CT Morphology
Adrenal metastases have a highly variable CT appearance. Small metastases are usually homo-
Consistent with adenoma
Size <3cm Round or oval shape Well-defined margins Homogeneous Hypoattenuating «10HU) Rapid eM wash-out
Sus~icious for mafignancy Size >5cm Irregular shape III-defined margins Heterogeneous Isoattenuating Marked enhancement
a
b Fig. 17.9 Adrenal carcinoma (3/5/2) presenting as a large mass with lymph node metastasis (arrowhead). On the contralateral side there is a small silent adenoma (arrow).
17 Adrenal Clands
a Fig. 17.10 Adrenal metastases. (a) CT reveals large, focal adrenal metastases in a patient with bronchial carcinoma. The size and bilateral occurence of the lesions are suspicious for metastases. (b) Metastases that
mimic adrenal hyperplasia are rare (here: bronchial carcinoma). Although the gland is enlarged, it retains an approximately normal configuration.
are a challenge to the radiologist, who must determine their significance. • CT Morphology Choosing an appropriate diagnostic Involvement is bilateral in about 50% of strategy for differentiating between a benign lepatients. CT attenuation values range between sion and a metastasis becomes an important 40 HU and 60 HU; there is little contrast en- issue when an adrenal mass is detected in a hancement (Fig.17. Tl). Necrosis can be seen in patient with a known primary malignancy. The rapidly growing lesions. detection of an adrenal lesion is suspicious for metastasis in such a patient, but it is not conclusive. Published reports indicate that the majorIncidentally Detected Adrenal Masses ity of adrenal masses found in oncologic patients are nonmalignant. Therefore it is cliniIncidentaloma is the term applied to an adrenal cally important to be able to differentiate bemass larger than 1 em, discovered incidentally tween benign and malignant disease in the main a patient with no evidence of endocrine dys- jority of patients. function. These masses are detected in approxi• CT Morphology mately "1-4% of abdominal CT examinations. Thus, incidentalomas are by no means rare and Morphologic differentiating criteria have been established for the etiologic classification and differentiation between benign and malignant adrenal masses discovered by CT (Table 17.4).The most reliable criteria are, in order of importance, attenuation, size, and enhancement characteristics, although these criteria overlap. A useful rule of thumb is that small « 3cm) lesions of uniformly low attenuation « 10 HU) or that show rapid wash-out of the contrast enhancement on delayed scans taken 15-60 minutes after contrast injection are probably benign. Larger heterogeneous masses (> 5 ern) that show marked contrast enhancement are strongly suspicious for malignancy or metastasis, especially in a patient with a known primary tumor. Fig. 17.11 Primary adrenal large cell lymphoma, diffuse enlargement of both adrenal glands.
with
Incidentally Detected Adrenal Masses Diagnostic
indicated, Asymptomatic masses dcm that have a CT attenuation < 10 HU may be classified as benign with high confidence, If the lesion has an attenuation> 10 HU, then thin-section contrast-enhanced CT with early and 15 minute delayed images is done, When the lesion has a relative contrast wash-out >40% or when the density at the delayed scan measures dOHU, benignity is assumed, Further investigations are unnecessary in non-oncologic patients, Oncologic patients can be managed with a wait-and-see approach that includes a single follow-up, provided the mass is not a new lesion. A new lesion is presumed to be metastatic 3. For asymptomatic lesions that are detected incidentally during routine contrast-enhanced abdominal CT. 15 minutes delayed imaging is performed, The lesion is considered benign if it shows a relative contrast wash-out> 60% at 15 minutes, 4, Large (>5cm), heterogeneous, enhancing lesions are suspicious for malignancy and require a definitive diagnosis (fine needle aspiration biopsy), 5, If the lesions do not fulfill the above criteria they are considered indeterminate, Spectflc gutdelines cannot be offered for such lesions, The procedure in a given case depends on the totality of clinical findings, If therapeutic implications compel a decision, fine needle aspiration biopsy should be performed,
Strategy
A recent analysis showed that up to 40% of adenomas are relatively lipid-poor and do not fulfill the densitometry criteria on noncontrast CT. Applying densitometry criteria on delayed contrast CT may be difficult due to differences in CTscanners or contrast injection protocols, leading to significant numbers of indeterminate lesions, Although the density of lipid-poor adenomas resembles that of nonadenomas, their contrast kinetics are similar to lipid-rich adenomas, Using a cutoff value of 40% relative wash-out (("I HUeariy/HUdeiay)x 100%) at 15 minutes, these adenomas can be identified with a high specificity and acceptable sensitivity, For lipid-rich and lipidpoor combined a more appropriate value is 60 %, A practical implication of this technique is the fact that delayed scanning can easily be combined with routine abdominal CT without having to reschedule the patient for a second examination, The following considerations should direct the diagnostic strategy with CT (in those patients where there is an apriori knowledge of a lesion, chemical shift MRI is a good alternative)-see also Table 17,5: L lf an adrenal mass shows features typical of a specific benign lesion (myelolipoma, cyst, hemorrhage) that is consistent with the clinical presentation, further diagnostic studies are not required, 2, When an adrenal lesion has been diagnosed previously, thin-section noncontrast CT is Table 17.5
Differential diagnosis of adrenal masses
Noncontrast
CT
Fa' Cystic areas Hemorrhage Calcification
Myelolipoma
Cyst, old hematoma, Trauma,
neuroblastoma,
Adrenal hemorrhage
echinococcosis, neuroblastoma adrenal carcinoma, metastases in early childhood, adrenal tuberculosis,
matosis
Calcification and CT attennation
Postcontrast
a mass
Neuroblastoma,
adrenal
carcinoma,
<-10 HU Myelolipoma <10 HU Adenoma >10 HU Adenoma, metastasis,
pheochromocytoma
carcinoma,
CT
Arterial hyperperfusion Relative washout (1Smin delayed): Densitometry (1Smin delayed): Vascular invasion
Pheochromocytoma >40% Adenoma <40% Lymphoma. carcinoma, metastases <30 HU Adenoma >30 HU Lymphoma, carcinoma, metastases Adrenal carcinoma, neuroblastoma
neuroblastoma
granulo-
637
KIDNEYS CSchaefer-Prokop,
M. Prokop
Anatomy
641
Examination Technique
643
Enhancement Characteristics
647
Congenital Anomalies
.
649
Hypoplasia, Aplasia, Agenesis (p. 649) Lobar Dysmorphism, Persistent Fetal Lobulation (p. 649) Duplication, Duplex Kidney, Ectopia (p. 649) Fusion Anomalies (Horseshoe Kidney, Cake Kidney, Crossed Fused Ectopy) (p. 650) Obstructive
Uropathy
651
Hydronephrosis (p. 651) Urolith iasis(p. 653) Cystic Renal Masses
655
Renal Cysts (p. 655) Echinococcal Cysts (p. 657) Polycystic Kidney Disease (p. 658) Multicystic Dysplastic Kidney (p. 659) Acquired Cystic Kidney Disease (p. 659) Renal Cystadenoma (Perlmann's Tumor) (p. 659) Cystic Renal Cell Carcinoma (Cystic Hypernephroma) (p. 660) Benign Solid Tumors
660
Pseudotumors (p. 660) Renal Adenoma (p. 661) Oncocytoma (p. 662) Angiomyolipoma (p. 662) Other Mesenchymal Tumors (p. 664) Malignant Solid Tumors
.
664
Renal Cell Carcinoma (Hypernephroma. RCC) (p. 664) Metastases (p. 668) Lymphoma (p. 668) Nephroblastoma (Wilms' Tumor) (p. 669) Tumors of the Renal Pelvis and Ureter (p. 670)
Continue>
Diffuse and Inflammatory
Renal Diseases
.
Acute Pyelonephritis (p. 672) Intrarenal or Perirenal Abscess (p. 673) Pyonephrosis (p. 674) Renal Tuberculosis (p. 674) Xanthogranulomatous Pyelonephritis (p. 674) Chronic Pyelonephritis (p. 675) Chronic Interstitial Nephritis (Analgesic Nephropathy) Glomerulonephritis (p. 676) Vascular
Disorders
672
(p. 676)
.
676
Renal Infarction (p. 676) Renal Cortical Necrosis (p. 677) Renal Vein Thrombosis (p. 677) Trauma and Postoperative
Changes
678
Renal Trauma (p. 678) Postoperative Changes (p. 680) Renal Transplantation (p. 680)
The kidney can be excellently analyzed with computed tomography. This is due to its typical enhancement characteristics, which allow for an optimal visualization of the renal parenchyma, renal lesions, and the renal vessels. Helical CT provides all necessary information for surgical management, especially for nephron-sparing resection techniques. Helical CT has improved the detection and characterization of small renal neoplasms although the differentiation between small benign and malignant tumors (
modalities, CT is in a competitive position with ultrasound and MRI. As a rule, ultrasound should be used as the primary imaging procedure. CT and MRI are largely equivalent in their diagnostic capabilities. CT offers advantages in the detection of stones and calcifications. MRI should be preferred in patients with impaired renal function or intolerance towards iodinated contrast agents. MRI is advantageous
Table 18.1 neys
Indications for CT examination of the kid-
Suspected tumor:
Exclusion or confirmation of a renal tumor
Tumor staging:
Renal cell carcinoma Transitional cell carcinoma Nephroblastoma Lymphoma
Complications:
Hemorrhage (postoperative, polycystic disease) Abscess (pyelonephritis, etc.)
Trauma:
Hemorrhage, contusion Renal vascular injury or occlusion
Hydronephrosis:
Differentiation between calcull, uretertumor, external compression
Renal colic/calculi: Substitute for intravenous urography
Anatomy
for the differential diagnosis of small complicated cysts and for imaging of the renal collecting system without prior injection of contrast
material (in patients with impaired function) .
excretory
Anatomy
The kidneys are located in the central compartment of the retroperitoneum (Cerota's space, perirenal space; Fig. 18.1) between the anterior renal fascia (Gerota fascia) and posterior renal fascia (Zuckerkandl's fascia or posterior Gerota fascia). They are enclosed in a fatty capsule and have an average volume of approximately 150ml. They measure 9-11 cm in their craniocaudal dimension, the left kidney measuring
about 1 cm more than the right. The coronal diameter of the kidneys at the renal hilar level is approximately 5-6 ern, and the sagittal diameter is approximately 4cm. On contrast-enhanced CT, the renal cortex and medulla can be distinguished. The renal sinus contains the main vascular branches, the pyelon and calyces, and the perirenal fat (Figs. 18.2, 18.3). Major blood vessels and the ureter leave the kidney at the
~ //-
Fig. 18.1 The kidneys are located in the central compartment of the retroperitoneum between the anterior renal fascia (Cerota's fascia) and posterior renal fascia (Zuckerkandl's fascia or posterior Gerota fascia).
Anterior ~ Gerota fascia
Posterior pararenal space
Gerota fascia
Renal vein -___.
Renal cortex
Renal medulla ~ Renal pelvis
Anterior pararenal space
Perirenal space
Renal artery
Fig. 18.2 Cross-sectional anatomy at the level of the renal hilum.
Peritoneum
-------I
641
642
78
The Kidneys
/
Renal cortex
r
Fig. 18.3 kidney.
Coronal
section through
the
Renal column
r: Renal medulla (pyramids)
Renal vein
j
Renal pelvis -
.../ "-
Ureter----
Renal calyx Fatty tissue in renal sinus
renal hilum. Normally the hilum is directed anteromedially toward the aorta, but increased ganterior rotation is not uncommon. The width of the renal parenchyma is approximately 1.5cm and tends to decrease with age. The cortex and medulla have a CT attenuation value of 30-50 HU on unenhanced scans, and IV contrast administration is necessary for corticomedullary differentiation. The renal cortex contains the vascular system, glomerula, and tubules. The medulla contains the collecting ducts and consists of approximately 18 pyramids. The medullary pyramids are separated from one another by extensions of the renal cortex (the renal columns, or columns of Bertin), which extend to the renal sinus. The apices of the pyramids converge to form 6-12 papillae, which convey urine into the pyelocaliceal (collecting) system. Intravenous contrast may or may not be needed to define the pyelocaliceal system on CT scans, depending on whether the system is of the ampullary or dendritic type. Normally the wall of the pyelocaliceal system is not seen on CT sections or appears only as an extremely fine line. Thickening of the wall signifies an inflammatory or neoplastic process. The ureter courses on the psoas muscle to enter the lesser pelvis. It is often visible already on un enhanced CT scans and is accompanied laterally by the gonadal ves-
sels. The ureter crosses over the iliac vessels at about the level of the promontory and descends on the retroperitoneal pelvic wall to the ureteral orifice on the posterolateral bladder wall. The ureter often shows a slight dilatation proximal to the site where it crosses the iliac vessels (compression effect). The relatively small-caliber renal arteries lie posterior to the renal veins. The longer left renal vein runs in front of the aorta to the inferior vena cava. The surgeon should be made aware of anatomic variants such as accessory renal arteries, early branching of a renal artery, duplicate veins, or a retroaortic renal vein. The presence and location of other small vessels entering the renal vein (e.g. gonadal veins, vertebral veins) is important for renal donor evaluation, especially if an endoscopic procedure is planned.
Potient Preparation
Examination
Technique
The technique of cr examination depends on whether the study is limited to the kidneys or whether the clinical problem warrants a general survey of the upper abdomen or the entire abdomen (Table 18.2).
contrast material. Oral hydration can be combined with bowel opacification. Bowel opacification with oral contrast medium is not mandatory but improves the differentiation between lymph nodes and bowel loops. The volume of oral contrast depends on the region to be examined: 500 ml is sufficient for examining the kidneys alone, but at least 1000 rnl should be administered for examinations of the pelvis. Whenever a cr angiography is required during the arterial or venous phase of contrast enhancement, water should be used as a negative oral contrast agent.
Patient Preparation Oral or intravenous hydration is important for optimum display of the renal collecting system and for protecting against nephrotoxic effects of
Table 18.2
643
Protocol
for CT scanning
of the kidneys
General Preparation in patients with impaired renal function
500-1000
ml of 0.45% saline iv over 4-12 h prior to the examination
Oral contrast
Upper abdomen
only:
500 ml administered tion 1-1.51, administered tion
Whole abdomen:
Patient
Supine with elevated
position
Respiratory
phase
both kidneys according kidneys to ischial bone
Expiration (conventional Inspiration (spiral CT)
Windowing
Noncontrast CT: Contrast-enhanced
60 min prior to examina-
arms
Kidneys: Urinary tract:
Scan range
30 min prior to examina-
to scanogram
CT)
CT:
W/L~ 300/40 W/L ~ 400-500/50-100
Scan parameters
l-slice SC/TF/RI
4-slice' SC
16-slice' SC
axial SW/RI
MPRb SW/RI
Kidneys
3/5/21 5/8/4
t
1-1.251 2-31
0.75-1.251 1-1.51
4/3 5/4
3/3 cor 3/3 cor
Urinary tract
5/10/51
1-1.251
0.5-1.251
4/3
Contrast
l-slice
4-slice'
16-slice'
V/F/D
V+N/F/D
V+N/F/DX
injection
Kindey/urinary
tract
120/2/X 120 + 50/3/X Scan delays X vary with the clinical problem (see Table 18.3)
3/3 cor
120 + 60/4/X
SC'" slice collimation (mm), TF= table feed (mmfrotation). RI"" recon. increment (mm). i ! = scanning direction, SW = effective section width (mm), MPR '" multiplanar reformats, axial = axial sections, cor= coronal. saq= sagittal eM = contrast material, Contrast concentration = 300 rnqjrnl iodine V'" volume of eM (ml). N -volume of saline flush (ml), F = flow rate (mils), 0 = start delay, X"" trigger region Contrast concentration" 300 mqjrnl iodine a Pitch P '" TF/(NxSC) should be chosen close to 1.5 with a-slice scanners. 1.3-1.5 with 16-slice scanners b
MPR are reconstructed
from a "secondary
raw data set" with SW/Rla1-1.5/0.7
or 0.5-0.8/0.5
II
644
18 The Kidneys Aquisition Technique Depending on the clinical imaging task, the required spatial resolution varies. For most dedicated indications for the kidneys as well as for obstructive uropathy, a section collimation of 3-5 rnm should be used. The smaller the collimation, the better the quality of multiplanar reformations of the renal parenchyma and excretory system. In addition, the differentiation of small cysts from tumors, as well as the detection of calculi, improves because of less partial volume effects (see Fig. 18.16). With multislice CT, moderately thin sections (4 x 2.5 mrn) suffice for most indications although the quality of coronal reformats is substantially improved with 0.5-1.25mm collimation on 4- to 16-slice scanners. Such thin-section protocols become important, however, for a subtle evaluation of the renal calyces. For obstructive uropathy, best image quality is obtained with a thin-section, 0.5-1.25 rnrn protocol and curved planar reformats (CPR) along the course of the ureter (see Fig. 18.11 b). A 4 x 2.5 mm protocol will allow for a somewhat lower patient dose (only relevant in 4-slice scanners). Coronal reformats should become standard for optimum demonstration of the extent and
Table 18.3
location of lesions. Unlike MRI, the coronal plane can be optimally adjusted after the data set has been acquired. Image quality often can be improved if thick MPR (section width 3 mrn) are employed. Maximum intensity projections may be helpful to demonstrate renal calculi and to demonstrate the urinary tract during the excretory phase. •
Exposure
Settings
The mAs settings should be chosen depending on the clinical indication and patient size. For noncontrast scans in obstructive uropathy, a low-dose protocol can be employed (CTDlval= 2-4mGy). For CT urography, low-dose scanning may also be feasible. Staging of known malignant tumors, on the other hand, requires optimum image quality for planning of surgery, and a multiphasic examination with a CTDlval = 8-20 mGy per phase (depending on the size of the patient) can be recommended. For a patient with suspicious renal masses on ultrasound, the number of phases should be reduced to two (precontrast and nephrographic, or nephrographic and late excretory phase), with a choice of moderate dose (CTDlvol= 5-10 mGy per scan). Note, however, that slim patients may require less, and very obese patients may require substantially more dose for adequate image quality.
Tailoring the phase of CT imaging to the clinical problem
----------------------------~
Phases after contrast administration 25 sec
100 sec
>15 min
Noncontrast
Arterial
Parenchymal
>5 min Excretory'
Indeterminate lesion (suspicious for RCC)
+
+
+
(+)
Suspected RPC
+
+
+
(+)
Ureteral lesion
+
+
+
(+)
Scan delay Clinical problem
(+)'
+
Pyelonephritis
Delayed Excretory
+
Abscess
(+)
Urinary tract stone
+
Trauma
+
(+)
+
Vascular injury
+
+
+
Urinoma
+
(+)
+
+
RCC= renal cell carcioma, RPC= renal pelvic carcinoma • Preopacifying the pyelocaliceal system with 20 ml of contrast medium 5-15 min before scan replaces excretory phase scranning
Aquisition
Noncontrast
CT
A non contrast examination of the kidneys is appropriate for almost all investigations with primarily renal indications (Table 18.3). Noncontrast scans are mandatory for detecting small calculi and are helpful for identifying hemorrhage and for measuring the density of fatty tumor components, e.g., in angiomyolipomas. Precontrast scans also provide a baseline for evaluating the enhancement characteristics of cystic lesions (
1OHU). This technique is especially helpful if no precontrast scan was preformed because the primary organ of the scan was not the kidney. Except for specific investigations of urinary tract calculi, noncontrast scans should always be followed by intravenous contrast administration. •
Contrast-Enhanced
CT
The contrast-enhanced examination may be conducted as a single- or multiphase study depending on the clinical question (Table 18.3). The time to peak level of vascular and renal tissue enhancement are highly dependent on volume and injection rate of contrast media as well as individual patient characteristics (cardiac output, renal blood flow, and renal function). This explains the relatively broad range of times listed in the literature for the various perfusion phases. Nephrographic Phase CT
With few exceptions, scanning should always include the nephrographic phase of enhancement (also called the parenchymal phase; scan delay approximately 100-180 seconds). During this phase the cortex and medulla show equal levels of enhancement while lesions appear markedly hypoattenuating to the renal parenchyma (see Fig. 18.4). For this reason, the rate of tumor detection is best in this phase. During the nephrographic phase of enhancement, special care should be taken in the evaluation of small cysts: partial volume effects may mimic an enhancing tumor ("pseudoenhancement". see Fig. 18.16). Conversely, a small tumor seen during the nephrographic phase should not be mistaken for a cyst. In cases of un-
Technique
certainty, a short-term follow-up is required using thinner collimation. If the renal examination is done as part ofa complete abdominal survey, scan delays are often too short for optimum imaging of the kidneys. Contrast enhancement then corresponds more to the corticomedullary phase than to the nephrographic phase. In these cases, any uncertainties that arise can be resolved by performing additional renal scans up to 3-5 minutes after the start of the contrast injection. Even though the nephrographic phase is preferred for evaluation of the kidney, it does not justify the routine use of additional parenchymal phase images in abdominal CT. Corticomedullary
Phase CT
Scanning during the corticomedullary phase (arterial phase, vascular phase, vascular nephrogram phase, delay 20-35 seconds) is a useful addition for demonstrating the vascular anatomy of the kidneys, especially in patients scheduled for nephron-sparing surgery. A start delay of some 30 s ensures opacification of the renal vein and enhancement of hypervascular tumors, while a shorter delay of some 20s (or better, bolus triggering 5 s after arrival of contrast in the aorta) will improve delineation of the arterial vascular anatomy (CT angiography). Scanning in the arterial phase may detect acute hemorrhage in patients with suspected renal injuries and may help distinguish small hypervascular tumors from pseudoenhancing cysts. In the vast majority of cases, hypervascularity suggests renal cell carcinoma. Arterial phase scanning alone may miss hypovascular lesions that are confined to the renal medulla. Arterial phase scanning is therefore not mandatory, and imaging in the nephrographic phase in general will suffice for detection and staging ofrenal malignancies. Excretory Phase CT, CT Urography
For studies focusing on the excretory portion of the urinary tract, scans should be obtained during the excretcny phase {pyelogram phase] more than 5 minutes after the start of the contrast injection. For dose reduction purposes, the contrast protocol can be modified by injecting 20-40 ml of contrast medium immediately after the non contrast examination to opacity the renal collecting system. The definitive contrast injection for evaluating the renal parenchyma is then performed some 5 minutes later.
18 The Kidneys
With this technique, the collecting system will be opacified on the nephrographic and corticomedullary phase images as well, making an extra excretory phase scan unnecessary, Best results are obtained with a compression device placed over the region of L5 to ensure optimum evaluation of the renal calyceal system. The first excretory scan is performed with the compression device inflated and includes the kidneys down to the iliac crest After release of the compression, the pelvis is scanned immediately afterwards, In patients with obstructive uropathy or impaired renal function, the scans may have to be delayed substantially: in these cases a delay of >30 minutes can be recommended, with a test scan at the level of the renal hilum to assess the contrast excretion into the collecting system before starting the scan, If multislice scanning is available, the maximum spatial resolution should be used if the pyelocaliceal system is to be evaluated (CT urography), In the excretory phase, scanning can be performed with substantially reduced dose because of the high contrast between the opacified pyelocaliceal system and the surrounding parenchyma, Visualization can be enhanced by administering fluids and a small dose offurosemlde. Delayed excretory scans acquired more than 15 minutes after the injection are useful for evaluating urinomas and for detecting the contrast retention in the renal tubules that is characteristic of acute inflammatory processes, They may also be used to evaluate hyperattenuating lesions incidentally detected during post-contrast scans of the abdomen, If there is contrast wash-out of > 10 HU, a solid tumor must be suspected, otherwise the lesion can be considered a hyperattenuating cyst, and ultrasound follow-up is sufficient
Contrast Use in Patients with Renal Failure In patients with pre-existing impairment of renal function (serum creatinine >l.5mg/dlor > 130 [unolll), the expected information gain provided by intravenous contrast administration must be weighted against the risk of further renal damage from the contrast medium itself Alternative diagnostic modalities like MRI or US must considered, Based on risk factors and
creatinine clearance, a risk profile can be established, which determines further pretreatment (see Tables 18,2, 18.3), A basic requirement is that the patient must be adequately hydrated to avoid intratubular precipitation of the contrast material, which could provoke acute renal failure, The European Society of U roradiology suggests preparing inpatients by infusing 100-150 mllh of 0.45% saline solution starting 4-12 h before the examination, In outpatients, 1000mi can be infused, starting approximately 1 hour prior to contrast administration, Intravenous or oral hydration should be continued for 12-24 h after the examination, Low osmolar contrast medium should be used and the volume should be limited to 100-150 ml depending on the size and renal function of the patient Repeat applications of contrast medium within 72 hours should be avoided, Isoosmolar contrast medium has not shown advantages over low osmolar non-ionic substances, Nephrotoxic medication (nonsteroid anti-inflammatory drugs, dipyramidamole, metformin) should be stopped 48 h prior to CT whenever clinically possible: acetylsalicylic acid (aspirin) should be reduced to 150mg/d, Proteinuria (diabetic nephrosclerosis, amyloidosis, light-chain plasmacytoma) may increase the risk of tubular contrast precipitation, Consequently, stringent selection criteria should be applied in these patients, and MRI can be performed as an alternative to CT in many cases, See chap, 3 for more information,
Image Processing With thin-section imaging using single or multislice spiral CT, coronal MPR of 3-4 mm thickness provide a better overview of the localization offocal Iestons. The plane should be slightly tilted to be coronal to the kidneys, Thin-slab MIP (5mm) are useful for analyzing the renal calyces on excretory phase scans but require very thin sections (0,5-1.25 mm section collimation) for optimum results, Thick MIP (3 em) give an overview of the collecting system and should be seperately performed for the kidneys, the midportion of the uereters and small pelvis, Thin and thick curved planar reformations (3 mm and l Omm) along the course of the ureter can display the localization of calculi or ob-
Nephrographic Phase (Parenchymal Phase)
structing ureteral lesions. Volume rendered displays are helpful for the planning of tumor re-
section and for the display of vascular anatomy, e.g. in living renal donors.
Enhancement Characteristics
The characteristic enhancement pattern of the kidneys on CT scans results from the high blood flow through the renal cortex (20% of the cardiac output) and the subsequent excretion of contrast medium through the collecting ducts of the renal medulla into the pyelocaliceal system. Corticomedullary Phase (Arterial Phase, Vascular Phase)
fading of attenuation in the renal cortex. Nevertheless, the corticomedullary phase usually persists during the nephrographic portal venous phase of abdominal CT (up to 80s p.i.). Enhancement of the renal arteries appears immediately after the contrast medium arrives in the aorta. Strong arterial enhancement persists throughout the aortic enhancement plateau. Because of the high renal blood flow the renal vein opacities rapidly, some 5-15 seconds after the start of the aortic plateau. This early part of the corticomedullary phase is called the arterial phase or vascular phase by some authors (20-40 s p.i.).
The corticomedullary phase of renal enhancement starts rapidly after the contrast medium arrives in the aorta (usually some 20-25 s after the start of contrast injection). It is characterized by intense enhancement of the renal cortex Nephrographic Phase and renal columns (Fig. 18.4). Maximum cortical enhancement is reached shortly after the end of (Parenchymal Phase) the plateau phase of aortic enhancement. DeThe attenuation of the renal medulla slowly pending on the contrast injection parameters, this happens some 30-50 s after the start of rises until it becomes isoattenuating or even hycontrast injection. The medullary pyramids are perattenuating to the renal cortex (Fig. 18.4). demarcated as hypoattenuating areas that show Such a hyperattenuation of the papillae may be increasing enhancement over time. The corti- observed as a transitory effect during the early comedullary phase persists as long as there is nephrographic phase. The nephrographic phase stronger enhancement of the renal cortex as begins some 60-80 s after the start of the aortic compared to the medulla. But as aortic en- plateau, and some 80-100 s after the start of inhancement declines, there is a relatively rapid jection. It lasts for a few minutes but there is
Corticomedullary phase
Pre-contrast a
b
Fig. 18.4 CT appearance of hypervascular (a) and hypovascu tartumor stb). aswell as a hyperdense renal cyst (c) during various phases of enhancement. While the cor-
Nephrographic phase
c ticomedullary phase is useful fortumor characterization (hypervascular vs. hypovascular), the nephrographic phase is better for delineating the tumor margins.
18 The Kidneys
relatively rapid decrease of enhancement during this time. The nephrographic phase overlaps with the excretory phase. A delayed onset of the nephrographic phase indicates impairment of renal function. Excretory Phase (Pyelogram Phase)
Opacification of the pyelocaliceal system normally begins 3-5 minutes after the start of the contrast injection. Some 3 minutes after contrast injection there often is still sufficient enhancement of the renal parenchyma to detect tumors. Thus, some authors suggest scanning at this point because of the concomitant opacification of the pyelocaliceal system. In patients with acute or chronic obstruction, enhancement of the urine may be substantially delayed. CT attenuation of the opacified urine varies with the patient's state of hydration and renal function. Peristaltic waves cause variations in the luminal width of the ureter following contrast administration, and portions of the ureter may not be opacified. Tumor Enhancement
Renal masses can be characterized by their pattern of contrast enhancement (Fig. 18.4). Cysts do not enhance with intravenous contrast medium «' 10 HU attenuation difference before and after contrast administration), and abscesses may show little or no enhancement.
Nephrographic phase
Corticomedullary phase
a
Hypervascular lesions (arterial phase) should always raise strong suspicion of renal cell carcinoma (RCC). In rare cases, benign tumors such as renal adenomas or angiomyolipomas may present as hypervascular lesions as well. Their enhancement pattern is initially similar to that of the renal cortex, but their level of enhancement declines more rapidly. Shortly after the end of the aortic plateau, hypervascular tumors may become isoattenuating to the parenchyma and can no longer be detected. During the nephrographic phase (> 100 seconds after the start of the injection), the tumors become hypo attenuating to the renal parenchyma. Hypovascular lesions may represent benign tumors, lymphoma, metastases, or, less commonly, RCC. During the arterial phase of enhancement they are well demarcated from the renal cortex but are poorly demarcated from the medulla. Accurate assessment of the extent of these lesions requires scanning in the nephrographic phase. Most transitional cell carcinomas are hypovascular. Because of their location in the pyelocaliceal system, they are may be more clearly depicted on excretory phase images. A tumor thrombus in the renal vein is distinguished from a "true" fibrin thrombus by its (frequently nonhomogeneous) enhancement during the arterial or parenchymal phase. If tumor enhancement equals the enhancement of the renal vein (such as in strongly hypervascular tumors), special care has to be taken to detect inhomogenetics as a sign of a tumor thrombus.
b
Fig. 18.5 CT appearance of a renal pseudotumor, e.g., hypertrophic column of Berlin (a), acute pyelonephritis (b), and focal bacterial nephritis (c). Pseudotumors
Excretory phase
c demonstrate identical enhancement relative to the renal cortex, while inflammatory lesions may lead to striate enhancement on excretory phase images.
Duplication, Duplex Kidney, Ectopia
Atypical Enhancement Characteristics Renal infarction is marked by an absence of parenchymal enhancement in the affected area (focal or wedge-shaped "defects"), although capsular arteries can maintain a rim of peripheral enhancement (see. Fig. 18.43). Chronic parenchymal disease (e.g.. glomerulonephritis) leads to parenchymal narrowing, delayed and diminished corticomedullary enhancement, and delayed excretion of contrast into the collecting system. Acute dysfunction of the renal tubules causes a radial, striate pattern of pel-fusion defects (see. Fig. 18.37). These areas display a reversed enhancement pattern during the ex-
Congenital
Anomalies
Most congenital anomalies of the kidneys can be evaluated by ultrasound imaging. They are not primary indications for CT scanning, but may be discovered incidentally. CT or MRI are the preferred modalites for differentiating lobar dysmorphism and persistent fetal lobulation from renal neoplasms.
Hypoplasia,Aplasia,Agenesis Renal agenesis denotes the complete absence of a kidney including its ureter and blood vessels. In aplasia, the kidney is present but poorly developed. A hypoplastic kidney has a reduced number of pyramids and a hypoplastic, somewhat bulbous pyelocaliceal system. •
cretory phase due to impaired outflow from the collecting ducts: the affected areas are hyperattenuating to the normal parenchyma and remain so for a period of hours to days. In cases of acute renal failure, varying degrees of damage to the renal tubules result in patchy areas of sustained parenchymal enhancement (see. Fig.l8.9b) . Normally the lymphatic system is not visualized, but acute obstruction of the renal vein may cause contrast medium to enter the renal lymphatics (see. Fig. 18.46). Fig. 18.5 gives an overview over various enhancement patterns of nontumoral diseases.
CT Morphology
An aplastic kidney appears as a structure of softtissue density in the renal bed that has an aortic vascular supply. There is compensatory hypertrophy of the contralateral kidney. A hypoplastic kidney has a well-proportioned shape. It can be differentiated from an acquired atrophic kidney by its clinical history, CT morphology, signs of intact renal function, and lack of evidence of parenchymal scars.
Lobar Dysmorphism, Persistent Fetal Lobulation Lobar dysmorphism refers to the hypertrophy of a renal column. Persistent fetal lobulation is characterized by shallow parenchymal indentations that are directed toward the renal columns. CT Morphology
Lobar dysmorphism and fetal lobulation can mimic tumors on non contrast CT scans. They can be clearly identified by their enhancement characteristics, however, which precisely match those of the rest of the renal cortex during all phases of enhancement.
Duplication, Duplex Kidney, Ectopia Renal duplication denotes separate kidneys, while a only separate pyelocaliceal (complete or partial ureteral
the presence of two duplex kidney has systems and ureters duplication).
CT Morphology
The presence of two renal pelves in a duplex kidney is easily missed on axial scans. The anomaly may be suggested by noting the greater length of the affected kidney in a side-to-side comparison. It is best appreciated on oblique coronal MPR or thin-slab MIP from spiral or multislice CT scans in the excretory phase (Fig. 18.6). The
650
78
The Kidneys
(intrathoracic kidney). The renal artery arises from the aorta or iliac artery at the level of the ectopic kidney. An abnormally low kidney (nephroptosis) differs from an (ectopic) pelvic kidney in that it has an artery arising from the aorta at the typical level (L lIL2) and a long, tortuous ureter. Fusion Anomalies (Horseshoe Kidney, Cake Kidney, Crossed Fused Ectopia)
a
b
Fig. 18.6 Duplex kidney with duplication of the caliceal system and the ureters (ureter fissus) demonstrated on a low-dose CT pyelogram (4 x 1/6, CTDlvoi = 1.8 mGy): curved 2 cm thick MPR (a) and posteroanteriorVRT after automated bone removal (b).
solid parenchymal isthmus connecting the two renal segments may mimic a tumor at the level of the presumed renal hilum when viewed in axial sections, but the true nature of the structure can be appreciated on multiplanar reformations. Hydronephrosis, usually involving the upper segment, is frequently accompanied by massive ureteral dilatation (Fig. 18.7) and is often caused by a ureterocele at the vesico-ureteral orifice. Renal ectopia denotes an abnormally positioned kidney that is most frequently located in the pelvis or, very rarely, above the diaphragm
a
b
Fig. 18.7 Duplex kidney with chronic obstruction of the upper caliceal system and marked dilatation and tortuosity of the ureter. The lower caliceal system and
The commonest type of fusion anomaly is the horseshoe kidney, in which the kidneys are joined at their lower poles by a pre aortic isthmus of renal parenchyma or fibrous tissue. Cake kidney is an anomaly characterized by the complete fusion of both kidneys. The fused kidney occupies a prevertebral or presacral position. In crossed fused ectopia, the kidneys both lie in one side of the abdomen and one ureter will cross the midline. Crossed ectopia without fusion is rare. •
CT Morphology
With a horseshoe kidney, the longitudinal axes of both kidneys converge inferiorly. The renal pelves are directed anteriorly and usually show slight dilatation. While a parenchymal isthmus is easy to detect (Fig. 18.8), a fibrous tissue band is hard to demonstrate on axial CT sections and often can only be presumed from the position of the kidneys. Multiple renal arteries are frequently present, that may lead to obstruction.
c second ureter are normal. Note the pronounced focal thinning of the parenchyma and the variations in ureteral opacification caused by diminished renal function.
Hydronephrosis In crossed fused ectopia CT reveals a single large kidney. At CTA aberrant renal arteries can be seen. One of the ureters will cross the midline and frequently inserts in the trigone of the bladder.
Fig. 18.8 Horseshoe kidney with typical rior rotation of the renal hilum.
Obstructive
Uropathy
Hydronephrosis
CT is performed in patients with known hydronephrosis to detect or exclude an extra- or periureteral cause of the obstruction (Table 18.4). Hydrocalices develop as a result of intrarenal segmental obstruction. In cases with clinical symptoms of ureterorenal obstruction, contrast -enhanced CT is performed, in addition to a precontrast scan, if causes other than lithiasis are suspected. Images are acquired in the nephrographic phase to provide information about the renal parenchymal status and in the excretory phase (10 minutes to hours after contrast injection, depending on the degree of obstruction and functional impairment). Coronal reformats or coronal thin slab MIPs of images obtained during the excretory phase are best suited to display the collecting system. With multislice CT, the quality of such images is often superior to that of an intravenous urogram. CT provides information complimentary to cystoscopy or retrograde ureterography. •
anterosupe-
CT Morphology
The renal collecting system in hydronephrosis is dilated. On noncontrast scans, the CT attenuation ranges from 0-10 HU, depending on the concentration of excreted substances. Fine, sharp contours of the collecting system are the most important criterion for distinguishing hydronephrosis from renal pelvic carcinoma or
Table 18.4 system
Causes
of dilatation
of the renal collecting
Supravesicalobstructive causes JHHR "'" ••••••,
.
"iIliIlBIlllil91i
A
Luminal obstruction: Urolithiasis Trauma (blood clots) Renal pelvic tumors, ureteral tumors Papilloma, endometriosis, metastases
Intramural causes: Atresia Strictures
(congenital,
radiogenic,
Ureteral kinking or compression:
inflammatory) *
Lymphoma, Iymphocele Pelvic tumors Enlarged uterus (caution: pregnancy) Retroperitoneal tumors Retroperitoneal fibrosis Aneurysm Hematoma, abscess Inflammatory bowel loops Ureteral kinking (renal transplantation)
Congenital anomalies: Aberrant vessels Ureteral anomaly Subpelvic stenosis Horseshoe kidney Nephroptosis
Functional Causes Vesico-ureteral reflux Neurogenic causes
* Indications for CT
;! - A5«1«»»)i£i>ilC!>!i»WilaSII\lll\t13:::W1
..
651
18 The Kidneys
pyonephrosis. Visible wall thickening is suggestive of an inflammatory process or a tumor. At the onset of urinary obstruction, the protruding portion of the renal pelvis is indistinguishable on noncontrast scans from an ampullary type of pyelocaliceal system. On postcontrast scans, however, the layering of contrast medi urn in the renal pelvis points to urinary obstruction rather than an ampullary pelvis. Further progression of the urinary obstruction causes dilatation of the major and minor calices (Fig. 18.9). Late stages are marked by papillary atrophy, and loss of the medullary pyramids due to the high intral uminal pressure. Finally the kidney becomes a hydronephrotic sac surrounded by a thin rim of parenchyma. Renal function at this stage is usually impaired, causing a delay of corticomedullary enhancement
a
b
Fig. 18.9 Acute right-sided urinary obstruction with differential contrast enhancement between the two sides (a), subacute bilateral urinary stasis (b) with pyelocaliceal dilatation and loss of corticomedullary differen-
and delayed contrast excretion into the renal pelvis. Hydronephrosis should be distinguished from ureteropelvic junction (L1Pl) obstruction (Fig. 18.10), which is a congenital anomaly that does not cause dilatation of the ureter. The ureter is dilated proximal to an obstruction (diameter proximal to the obstruction > 5 mm, further proximally> 3 mm) and shows delayed opacification. The ureter distal to the obstruction is collapsed and poorly defined. If dilatation of individual calices is noted (hydro calices) , several possible causes should be considered: crossing vessels, calculi, tumors, or inflammatory changes (common in tuberculosis). CT can demonstrate the cause only if thinslice helical or multislice scanning is performed.
c tiation (scan acquired at 100 seconds), and a hydronephrotic sac representing the end stage of chronic hydronephrosis (c).
Fig. 18.10 Ureteropelvic junction (UPJ) obstruction (4 x 1/6, paracoronal MPR) with marked dilatation of the renal pelvis and normal size of the ureter. Normal size kidney (a). marked thinning of the cortex in untreated massive obstruction (4 x 1/6, sagittal MPR) (b).
a
b
Urof'ithiasis
a
c
b
Fig. 18.11 Urolithiasis. Small calculus in the ureter with soft-tissue rim sign and stranding of the periureteric fat (a). A curved MPR (4 X 1/6) demonstrates the location relative to the ureter (b). A prevesical calculus
d
(arrow) has to be distinguished from phleboliths (arrowhead) (c). Note the comet tail sign due to a small vein (4x 1/6, coronal MPR) (d).
Fig. 18.12 Unilateral perinephric stranding in acute obstruction (a) as compared to bilateral perinephric stranding in elderly patients with chronic renal disease (b). The perinephric fluid in (a) is extensive, suggesting rupture of the pyelocalyceal system.
a
b
Urolithiasis
Noncontrast CT plays an increasingly important role in patients suspected of suffering from a renal colic. Spiral CT is more effective than intravenous urography in identifying ureteral stones and as effective as intravenous urography in determining whether there is renal obstruction. CT is considered the best method of determining the stone volume (with a mean error of < 5 %) and thus providing important information for treatment planning (lithotripsy for stones> 2.5 cm has considerably higher complications). Residual calculi after shock wave lithotripsy are readily assessed. Low-dose scanning usually suffices for making the diagnosis. Best results are obtained with multislice CT.
•
CT Morphology
CT is more sensitive than other imaging modalities for detecting urinary calculi. It can even detect small noncalcified stones that are radiolucent on abdominal radiographs, as their attenuation values are always greater than 100 H (Fig. 18.11).Calcium-containing stones (90% of all stones) and cystine stones have attenuation values of 450-1500 HLJ, while xanthine stones are in the range of 100-600 HLJ. Generally, however, urinary stones cannot be differentiated by their chemical composition on the basis of CT attenuation measurements. Acute obstruction may lead to swelling of the kidney, increased thickness of the renal cortex due to edema, and delayed or reduced contrast enhancement of the affected kidney. Ureteral stones in combination with hydronephrosis, hydroureter, or perinephric fluid
18 The Kidneys
establish the diagnosis of an acute renal obstruction (Fig. 18.12). Dilatation of the collecting system without detectable stone occurs if the stone has already passed into the bladder or was extracted. In acute ureteral 0 bstruction the ureteral wall will initially appear normal but will eventually thicken with the onset of edema or inflammatory reactions. Acute rupture of the renal pelvis presents as substantial amounts of fluid around the affected kidney in patients with complete ureteral obstruction. Perinephric fluid alone is a good predictor of acute obstruction if it occurs unilaterally, and the patient has no history of acute trauma, previous or chronic renal disease (Table 18.5). Papillary
calcifications
and arterial
calcifi-
cations are usually distinguished by their location from stones in the collecting system (Fig. 18.13). The following features on noncontrast CT that are helpful to distinguish ureteral stones from extra-ureteral calcifications such as phleboliths in the small pelvis: • A soft tissue rim around an impacted stone (tissue rim sign) is thought to represent the ureteral wall thickened by edema or inflammation. • Stranding around the calcification represents an inflammatory reaction in the periureteral fat. • Phleboliths may be characterized by a central region of low attenuation (magnified images with bone window settings), a bifid peak on the line histogram and the comet
a
b
Table 18.5
Urolithiasis
/\
Signs of acute obstruction due to urolithiasis
Hyperattenuating focus (>120 H) in the collecting system Dilation of the intrarenal collecting system Hydroureter (ureteral obstruction) Stranding of the perinephric fat (unilateral) Perinephric fluid (markely more pronounced on one side) Unilateral increase in renal cortical thickness Unilaterally reduced cortical enhancement DD phleobolith/ureter
stone
Plebolith:
Comet sign Central low attenuation
Ureter stone:
Tissue rim sign Periureteral stranding Continuity with ureter on curved reformations
sign. The latter is caused by an eccentric, tapering, soft tissue mass adjacent to the round calcification, which represents the noncalcified portion of the pelvic vein containing the phlebolith. With multislice CT longitudinal reformats can directly demonstrate the relation of the ureter to the suspected stone (Fig. 18.11), even in situations where conventional spiral CT remains inconclusive.
c
Fig. 18.13 (a) Papillary calcifications located at thejunction of the medulla and pyelocaliceal system, (b)Severe nephrocalcinosis in oxalosis. (c) Medullary sponge kidney.
Cystic Renal Masses
Differential diagnostic features are provided in Table 18.6. Renal Cysts
Renal cysts are a frequent incidental finding that has no pathologic significance; more than 50% of patients over age 50 have renal cysts. Multiple cysts are common. Cysts may be intraparenchyTable 18.6
mal or parapelvic, and they can reach considerable size. Complicated cysts result from superinfection or hemorrhage into the cyst. While the majority of uncomplicated renal cysts are easily diagnosed, complicated cysts can be very difficult to distinguish from tumors. •
CT Morphology
Uncomplicated cysts appear on noncontrast scans as masses of uniformly low attenuation
Differential diagnosis of cystic renal lesions
Solitary lesion
Uncomplicated cyst
No calcification, no septa, no contrast enhancement «10 HU attenuation difference)
Abscess
III-defined margins, air bubbles, rim enhancement, renal parenchyma
Hematoma, traumatic cyst
Hyperattenuating, nonenhancing
Echinococcalcyst
Thin mural calcification, enhancing capsule and septa
Renal cystadenoma
Multiple cysts enclosed in a capsule, enhancing capsule and septa
Cystic hypernephroma
Enhancement in nodular portion of wall
inflammatory reaction of
Multiple lesions
a
Uncomplicated cysts
No septa, no contrast enhancement
Adult polycystic disease
Numerous cysts of varying density scattered throughout the kidney, accompanied by cysts in the liver and pancreas
Echinococcalcysts
See above
Multicystic and dysplasia
Cluster-of-grapes parenchyma
cysts with wall calcifications, no demonstrable renal
b
Fig. 18.14 Renal cysts, (a) Large, uncomplicated renal cyst with a marginal beak sign caused by the stretched renal capsule, (b) This sign should not be mistaken for a solid tumor component, (c) Parapelvic cysts can be dis-
c
d
tinguished from a dilated renal pelvis by their lack of enhancement on late scans, (d) Smm-thick coronal MPR of parapelvic cysts (4 x 1/6).
18 The Kidneys
(0-20 HU) with well-defined margins (beware of partial volume effects). The cyst wall is not visible with CT, and the attenuation of the mass does not increase after IV contrast administration « 10 HU). Renal cysts can be classified by their location as subcapsular, cortical, medullary, or parapelvic (Fig. 18.14). Parapelvic cysts cause impressions on the pyelocaliceal system (Fig. 18.14c, d), but segmental obstruction ofthe collecting structures is rare. They can best be differentiated from an obstructed renal pelvis on late scans. A beak sign occurs with cysts that stretch the renal capsule (Fig. 18.14a, b) and can mimic wall thickening in cysts that are located at the renal pole.
a
c
b
Fig. 18.15 Complicated cysts (a). Septated benign cyst (4 x 1/6, coronal MPR). ( b) Multiple cysts in von-HippelLindau disease, associated with a renal angioblastoma (arrow). Note the distorted renal parenchyma, (c) Table 18.7
Cysts are called complicated if they have an increased CT attenuation, thickened septa, calcifications, wall thickening, or contain gas (Fig. 18.15). A high-density renal cyst (> 20 HU, up to 90 HU) may be caused by intracystic hemorrhage, infection, a high protein content, or milk of calcium. Calcifications most commonly result from previous infection or intracystic hemorrhage. Infection can lead to uniform wall thickening with rim enhancement or the presence of intracystic gas (see below). The Bosniak classification has been established to differentiate between benign and potentially malignant cystic lesion (Table 18.7).
d
Thickened wall and extensive mural calcification in renal tubercolosis. (d) A hyperdense cyst in the nephrographic phase may simulate tumor (4x 1/6, coronal MPR).
Management of cystic renal lesions (according to Bosniak, 1994)
Category I: Categories 11-IV: • Hyperattenuating • septa or • calcifications Category II:
Uncomplicated cyst
IMo follow-up
Complicated cysts cyst contents
(>20 HU) or
Probable benign cyst
Follow-up at 3, 6, and 12 months
Homogeneous attenuation before contrast administration (narrow window setting) and no enhancement with contrast medium (narrow window setting, <10 HU attenuation difference) <3 cm diameter and less than 1/3 of the cyst extends outside the kidney and has smooth margins possible history of infection (not mandatory) Category III:
Possible malignancy
Resection (except in high-risk patients)
• One of the first four criteria in Category II does not apply Category IV:
Probable malignancy
• Detection of a solid component • Irregular, ill-defined margins
and
Resection
(no matter how small) that enhances with contrast medium or
Echinococcal Cysts Partial volume effects can cause indistinct margins and may simulate increased CT attenuation, especially in small cysts (see Fig. 7.24, 18.16). This increased attenuation depends on the attenuation of the surrounding parenchyma and is higher after contrast enhancement. As a consequence, small cysts may seem to take up contrast medium (contrast enhancement > 10 HU) if too wide a section collimation (SC) was employed for scanning. Thin slice spiral or multislice CT scanning with overlapping reconstruction improves diagnostic accuracy (Fig. 18.16). The collimation SC for scans in the parenchymal phase should be chosen equal to or smaller than for the precontrast scans. Problems may occur if no precontrast scan was performed, e.g., because the scanning was primaril y targeted on the liver, and a cyst -like lesion with increased attenuation (> 15-20 HU) is seen in the renal parenchyma during a contrast-enhanced examination. In these cases, late scans more than 15 minutes following intravenous contrast administration demonstrate a reduction in CT attenuation if the lesion is a vascularized tumor, while CT attenuation remains unchanged « 10HU difference) if it is a cyst (Fig. 18.17). •
Differential
a
Fig. 18.17 Even without a precontrast scan, a suspicious lesion measuring 45 HU on the nephrographic phase (a) can be identified as a tumor by demonstrating de-enhancement to 30 HU in the excretory phase (b).
the renal sin us. In isolated cases it may be difficult to distinguish from parapelvic cysts or transitional cell carcinoma. The slight, delayed enhancement of fibrolipomatosis differentiates it from cysts, however. Metastatic melanoma may show very low attenuation on pre contrast scans, but it differs from cysts in its marked enhancement after contrast administration. Some difficulties may be encountered in distinguishing complicated cysts from renal cell carcinoma. Some helpful differentiating features are listed in Tables 18.6 and 18.7.
Diagnosis
Parapelvic cysts are best distinguished from an ampullary renal pelvis or hydronephrosis during the excretory phase. They show no enhancement, and they cause smooth impressions on the collecting system. Pelvic fibrolipomatosis has a negative to above water-equivalent CT attenuation, depending on the connective tissue content of the fat in
Fig. 18.16 Pseudoenhancementwith increase of enhancement from 20 HU before contrast injection (a) to 35 HU in the nephrographic phase may be seen with small cysts if wide sections (here;SW= 10 mm) are used (b). This partial volume effect is reduced with thin-section imaging: the cysts measures only 23 HU on 3-mm sections reconstructed form the same multislice CT data set (c).
a
b
Echinococcal Cysts
After the liver and lung, the kidney is the third commonest organ infected by Ec1iinococcus cysticus, with renal involvement occurring in 2-7% of
b
c
657
18 The Kidneys
patients with systemic hyclatid disease. Renal involvement by Echinococcus alveolaris is unknown. •
CT Morphology
Renal echinococcosis is marked by the development of sharply circumscribed cysts. Calcification of the cyst wall is seen in approximately one-third of cases. Septations and daughter cysts within the mother cyst are characteristic, and the attenuation difference between the mother and daughter cysts is pathognomonic. The cyst wall is visible with CT, and both the wall and septa enhance with intravenous contrast. Rarely the cysts may contain solid components, which also show intense enhancement. These lesions are generally not recommended for biopsy. Cysts that show septa or wall thickening may be difficult to distinguish from cystic hypernephroma. Hydatid cysts in other organs and serologic test results will suggest the correct diagnosis, however. POlycystic Kidney Disease
The infantile form of polycystic disease (Potter type 1) is inherited as an autosomal recessive trait and is often fatal during the first months of life. It is commonly associated with anomalies of the hepatobiliary system. CT scanning is not usually indicated. The adultform (Potter type I I I) is transmitted as an autosomal dominant trait and causes progressive renal impairment starting at about 30
a
b
Fig. 18.18 Polycystic kidney disease, (a) Multiple bilateral cysts in a 42-year-old-patient. (b) Massive renal enlargement with multiple hyperdense (chocolate)
years of age. It is associated with hepatic cysts (30-60%), pancreatic cysts (10%), and an increased incidence of arteriovenous malformations and aneurysms. There is, however, no increased incidence of renal cell carcinoma in these patients. CT is used to diagnose complications such as hemorrhage and infection and to detect or exclude a tumor in patients with hematuria. •
CT Morphology
In the infantile form, the kidneys are markedly enlarged but retain a normal shape. The numerous cysts are 1-2 mm in size and commonly cannot be defined by CT. Contrast-enhanced scans reveal faint radial striations in the parenchyma, which does not show corticomedullary differentiation. Urinary stasis in the dilated collecting ducts leads to stone formation with medullary nephrocalcinosis that is visible even on plain scans. In the adult form of polycystic disease, the kidneys are enlarged during adolescence and young adulthood and are diffusely permeated by cysts of varying size (Swiss cheese pattern). The number and size of the cysts increase with further progression of the disease, until finally the grotesquely deformed kidneys fill the entire abdomen. Recurrent intracystic hemorrhage is very common (70%) and produces areas ofvariable high attenuation with homogeneous contents (Fig. 18.-18).Calcifications (in> 50%) result from small calculi, hemorrhage, or infection. Intravenous contrast administration improves the
c cysts in a 55-year-old patient (4 x 1/6, coronal MPR). (b). Fresh intracystic hemorrhage (arrow) is less homogeneous and has a fuzzy border (c).
Renal Cystadenoma (Perlmann demarcation of the residual parenchyma from the collecting structures. Areas of fresh intracystic hemorrhage present as heterogeneous regional hyperattenuating structures within a renal cyst (Fig.18.18c) and may warrant surgical intervention in symptomatic patients. Acute infection can lead to thickening and enhancement of the affected cyst wall and may also be manifested by intracystic gas. Malignanttumors (renal cell carcinoma, lymphoma) are easily missed in polycystic kidneys. Multicystic Dysplastic Kidney
Multicystic dysplastic kidney (Potter type II) is a condition marked by severe structural disorganization of the kidney. It is a unilateral anomaly, since bilateral multicystic kidneys are incompatible with life. •
CT Morphology
The kidney is composed of approximately 10 to 20 cysts, each smaller than 4cm, that show wall calcifications and a c1uster-of-grapes appearance. The complete absence of functioning (enhancing) renal parenchyma is pathognomonic (silent kidney). The contralateral kidney is hypertrophic and in 40% of cases show associated abnormalities such as malrotation or ureteropelvicjunction stenosis. Acquired Cystic Kidney Disease
Acquired renal cysts are found in 40-60% of patients who have been on hemodialysis for 3-5 years, and in up to 80% of patients on dialysis
for 5-10 years. The number and size of cysts increase with duration of hemodialysis. Complications and suspected tumors are the only indications for CT. There is an increased incidence of renal carcinoma of about 5%. Ultrasound follow up is indicated once per year for patients who are on dialysis for longer than 3 years. •
CT Morphology
Scans reveal multiple small cortical cysts less than 2 cm in diameter (acquired cystic renal disease). The renal parenchyma appears shrunken (Fig. 18.19), and fatty atrophy is apparent in the renal sinus. Hemorrhage into cysts causing increased attenuation may occur. Diffuse cortical calcifications result from the deposition of oxalate crystals in the cortical tubules. Renal Cystadenoma (Perl mann's Tumor)
Renal Cystadenoma (multilocular cystic nephroma) is a benign encapsulated tumor composed of multiple cysts filled with myxomatous material. It is most prevalent in small boys « 4 years) and older women (40-70 years). •
CT Morphology
The tumor ranges from 3-30 cm in size and consists of numerous cysts, up to lOcm in diameter, which are sharply demarcated from surrounding structures (Fig. 18.20). There are multiple, well-defined septa permeating the cysts. Calcifications are common (10-50%), but areas of hemorrhage and necrosis are rare. Cysts with mucinous contents show soft-tissue attenuation and are indistinguishable from solid tumors on precontrast scans. The cysts do not communi-
Fig. 18.19 Renal changes associated with chronic hemodialysis (aquired polycystic kidney disease, APKD). (a) Multiple small, bilateral renal cysts are a typical finding, (b) Uncomplicated cysts on the left side and a small, hypervascular renal cell carcinoma on the right side in a patient with APKD.
a
s Tumor)
b
659
18 The Kidneys
phroma in adults and cystic Wilms' tumor in children, surgical resection is always indicated, especially since the tumor may undergo sarcomatous transformation in adults. These lesions are in the Bosniak type 3 category (see Table 18.7). If CT shows thickened or nodular enhancing septa the mass must be considered renal carcinoma until proven otherwise. Cystic Renal Cell Carcinoma
Fig. 18.20 Renal cystadenoma (Perl mann's tumor) appears as a focal polycystic lesion with fine enhancing septa.
cate with the collecting system but, not uncommonly, a portion of the tumor herniates into the distorted collecting system. Only the septa and capsule enhance after IV contrast administration. Solid nodular components are not found. Because the CT features of Perlmann's tumor are similar to those of cystic hyperne-
In approximately 20% of all renal cell carcinomas, cystic characteristics are observed in portions of the tumor. •
CT Morphology
The hypernephroma may display low attenuation values (approximately 10-30 HU) because of cystic components, fat, or necrotic areas. An enhancing cyst "wall" of irregular thickness is highly suspicious for a tumor (Fig. 18.21). If even a small portion of a cystic mass displays a feature that is not consistent with a cyst, a malignant tumor should be suspected (see criteria in Table 18.7). Fig. 18.21 (a) Cystic renal cell carcinoma with peripheral, septa-like internal structures that show contrast enhancement, (b) Large cystic hypernephroma with an irregular, enhancing soft-tissue component and irregular calcifications.
a
b
Benign Solid Tumors
CT is excellent for tumor detection but, except for angiomyolipomas, cannot positively identify a renal tumor as being benign. For this reason, tumors that appear benign (Table 18.8) are generally managed by surgical resection (enucleation). Biopsy is not indicated because of its low specificity.
Pseudotumors
Hypertrophic renal columns or persistent fetal lobulation can mimic renal tumors (see congenital anomalies, p.649). Secondary pseudotumors result from the nodular hypertrophy of healthy areas of parenchyma following the atro-
Renal Adenoma Table 18.8
Differential diagnosis of solid renal lesions
I Benign renal lesions:
Noninvasive, no extracapsular
extension
Pseudotumor
Same enhancement pattern as renal cortex (renal columns, fetal lobulation, parenchymal scars)
Angiomyolipoma
Fatty areas «-20 HU) accompanied by areas of soft-tissue attenuation
Lipoma
Fat attenuation only «-80 HU)
Oncocytoma
Stellate central scar, nonhomogeneous enhancement pattern; differential diagnosis: renal cell carcinoma
Adenoma
<3 cm, nonspecific enhancement pattern; differential diagnosis: renal cell carcinoma; lesions 3 cm or larger are potentially malignant
Fibroma, hemangioma, etc.
<3cm. nonspecific
Papilloma
Like renal pelvic carcinoma, but noninvasive; often multifocal (differential diagnosis: renal pelvic carcinoma)
Malignant renal lesions:
Invasive, possible extracapsular
Renal cell carcinoma
Usually hypervascular, rarely hypovascular; tends to infiltrate the renal vein
Nephroblastoma (Wilms tumor)
Children; nonhomogeneous hypervascularity; often very large; tends to infiltrate the renal vein
v.Sarcomas
extension (T3 or higher)
1:
tAmMB
May be hypervascular; liposarcoma contains heterogeneous fatty components
Metastases
Usually hypovascular, often bilateral, multifocal; primary tumor?
Renal pelvic carcinoma
Located in renal sinus; frequent pyelocoliceal obstruction; drop metastases
Lymphoma
Bilateral> unilateral; diffuse infiltration is common; abdominal lymphadenopathy; hypovascular
phy of damaged parenchymal regions due to inflammation, trauma, or infarction. •
CT Morphology
The lesions are in continuity with the renal cortex and display normal corticomedullary enhancement characteristics, so generally the differential diagnosis is straightforward (Fig.18.22). Renal Adenoma
The incidence of renal adenomas is increased following pyelonephritis, in benign nephrosclerosis, and in patients on hemodialysis (up to 20% of all autopsies). They are occasionally multifocal and can range from a few millimeters to several centimeters in size. There is much controversy whether adenoma should be classified as a distinct benign neoplasm or whether it is a premalignant lesion that eventually becomes a renal cell carcinoma.
Fig. 18.22 Renal pseudotumor due to postinflammatory scarring in a patient with recurrent pyelonephritis. In the corticomedullary phase, there is a hypervascular pseudolesion that appear isoattenuating to the renal parenchyma in the nephrographic phase.
In the past, a well encapsulated solid neoplasm of the renal cortex smaller than 3 em in size with no metastases, hemorrhage, or histological evidence
661
78 The Kidneys
for mitoses or atypia was considered an adenoma. On the other hand, most small renal cell carcinomas detected by CT have identical CT characteristics, and the incidence of adenoma is increased in patient groups that are also at high risk for developing renal cell carcinoma. Because renal adenoma cannot be confidently distinguished from carcinoma (also not by fine needle biopsy), close follow-up examinations or surgical resection are indicated. •
CT Morphology
Renal adenomas appear on noncontrast CT scans as nodular masses that are roughly isoattenuating to the renal parenchyma. Calcifications are rare. Tumors may be hyper- or hypovascular (Fig. 18.23; see also Fig.18.19b) but are most clearly delineated in the parenchyma! phase.
Fig. 18.23 Renal cell adenoma in an obese patient with acquired polycystic kidney disease. The hypervascular adenoma cannot be distinguished from a renal cell carcinoma by its CT morphology. Uncomplicated cyst on the left side.
They do not infiltrate but are otherwise identical to renal cell carcinoma with respect to their CT features (compare Fig. 18.26). Larger tumors may contain central low attenuation due to necrosis. Oncocytoma
Oncocytomas are benign tumors (3-5% of all renal tumors) that arise from the epithelial cells of the proximal renal tubule and may reach an enormous size. The peak incidence is in the sixth and seventh decades, and males are affected twice as often as women. •
CT Morphology
Oncocytoma is a usually single, well-demarcated, uniformly expansile mass. It appears slightly hypo- or isoattenuating on unenhanced scans. The tumor is hypovascular and in > 80% of cases shows a typical spoked wheel pattern of enhancement with a central stellate scar of low CT attenuation (Fig. 18.24). Rare findings are cystic changes or more than one scar. Rarely, oncocytomas occur as multiple lesions (oncocytomatosis) or become very large (giant oncocytoma) . A confident differentiation from renal cell carcinomas, however, cannot be made: lesions smaller than 5 cm may show no scar and on the other hand, a central scar in a renal cell carcinoma can have a very similar appearance Angiomyolipoma
Angiomyolipoma is a relatively common benign hamartosis composed of varying proportions of
Fig. 18.24 Oncocytoma with a faint central scar in the right kidney (a). Large tumor in the right kidney with a central scar (b) that proved to be a renal cell carcinoma at histoloqy.
a
b
Angiomyolipoma fat, atypical muscle fibers, and ectatic vessels. Angiomyolipomas predominantly affect young women, are usually unilateral, and present as lesions of 2-5 cm in size. They may reach 15cm, however, and these large tumors are often symptomatic (flank pain, hematuria). Lymphangioleiomyomatosis (LAM) of the lung and tuberous sclerosis are associated with an increased incidence of angiomyolipoma (4080%). The lesions in tuberous sclerosis are small, multiple, bilateral, asymptomatic, and can be detected in adolescence. In young patients, however, a macroscopic fatty tumor component may be absent. Malignant transformation is unknown. Local lymph node enlargement and even foci in the liver, spleen, or uterus are a reflection ofm ul ticentric disease rather than metastases. Complications include potentially severe retroperitoneal hemorrhage due to rupture, hypertension, and impaired renal function. The frequency ofcomplications increases with a tumor size of> 4 cm. Because angiomyolipomas can enlarge over time, they may warrant therapy at some point. Tumor growth appears to be greatest in patients with multiple angiomyolipomas, regardless of whether the stigmata of tuberous sclerosis are present. It is not clear yet to which extent followup examinations are required. •
CT Morphology
Fatty components are almost always present, and their CT detection suggests the correct diag-
a
b
Fig. 18.25 Angiomyolipoma. (a) CT demonstrates a large angiomyolipoma with characteristic fatty components and angiomatous soft-tissue components, (b) Multiple, often tiny angiomyolipomas offat density are seen in tuberous sclerosis (caution: these should not be
nosis. It is sufficient to identify even small areas with attenuation values less than -20 HU (Fig. 18.25). In doubtful cases, the accuracy of attenuation measurements can be improved by acquiring thin slices with spiral or multislice CT. The tumor grows by expansion and not by infiltration, but it may rupture its capsule or extend into the renal sinus, causing pyelocaliceal compression. Predominantly extrarenal extension occurs in approximately 25% of cases. The presence of an angiomatous component leads to heterogeneous enhancement. Hemorrhage into the tumor or perirenal space appears as a hyperattenuating area. Problems of differential diagnosis may arise if the diagnostically important fatty component of the tumor is very small or in rare cases absent (Fig.18.25c). This can make the angiomyolipoma indistinguishable from renal cell carcinoma. Very rarely, a renal cell carcinoma or nephroblastoma may contain a small amount of fat. Renal lipomas and liposarcomas are extremely rare. Care should be taken not to mistake adipose tissue of the renal sinus engulfed in a renal carcinoma as the fatty component of a tumor suggesting an angiomyolipoma. The combined presence of fat and calcifications is suspicious for carcinoma since angiomyolipomas very rarely contain calcifications. Angiomyolipomas never show an extension into the renal vein.
c confused with small cysts), (c) Large angiomatous component showing intense, nonhomogeneous enhancement in an adolescent patient with tuberous sclerosis (mimics renal cell carcinoma).
663
664
18 The Kidneys
Other Mesenchymal Tumors Fibromas, lipomas, hemangiomas, hemangioendotheliomas, leiomyomas, and histiocytomas are very rare « 1 %) and are generally small,
CT Morphology Only lipoma can be identified by its negative attenuation values and the absence of a softtissue component, although it may be mistaken for a much more frequent renal angiomyolipoma. All other tumors display nonspecific morphologic features on CT scans,
Malignant Solid Tumors Renal Cell Carcinoma (Hypernephroma, RCC) The clinical hallmarks of renal cell carcinoma (RCC) are hematuria (60%) and Hank pain (50%), Approximately one-third are discovered incidentally, Males are affected with a 3: 1 predominance, and the peak incidence is after 50 years of age, The risk of RCC is increased in Hippel-Lindau syndrome (35%), in tuberous sclerosis, and after years of dialysis (6%), The tumors in Hippel-Lindau syndrome are bilateral in 75% of cases and multicentric in more than 85%, Tumor histology is a more important prognostic indicator than tumor size, Hypovascular tumors (5-15%) have a better prognosis than the commoner hypervascular forms, The majority of renal cell carcinomas are of the clear cell subtype (70%), that are mostly (80%) hypervascular; 10-20% of them are cystic. Papillary cell type tumors (10-15%) or chromophilic cell carcinoma « 5 %) are hypovascular and resemble cortical cysts (papillary type 1), lymphoma or metastases (pa pillary type 11), CT is more than 90% accurate in the detection and staging ofRCC CT Morphology RCC typically causes distortion of the renal contours, It is approximately isoattenuating to the renal parenchyma on non contrast scans, but it may also contain hyper- or hypoattenuating areas (due to hemorrhage or necrosis), Fatty components are very rare and may make the tumor difficult to distinguish from angiomyoltpoma: Less than 20% of the tumors contain central or peripheral curvilinear calcifications, The likelihood of calcifications increases with size.
The more common hypervascular tumors show intense enhancement during the early arterial phase after contrast administration (Fig, 18,26), During the parenchymal (nephrographic) phase, RCC is always hypoattenuating to the renal parenchyma regardless of its vascularity (Figs, 18,26, 18,27), Larger tumors with necrotic components show heterogeneous enhancement An indistinct margin between the mass and the renal parenchyma is suggestive of malignancy, This, however, is only common in large tumors, Over 80% of small renal tumors are well demarcated, some of them even showing a pseudocapsule consisting of a thin low-attenuation border. Demonstration of the exact localization of the tumor within the renal parenchyma, its size and relation to the renal vessels is essential prior to nephron-sparing surgery (partial nephrectomy), and can be best accomplished using volume-rendering techniques (Fig, 18,28), The preoperative staging ofRCC (Table 18,9) is based on the invasion of adjacent structures, lymph node status, and the involvement of venous structures, Tumor thrombi are often in contact with appositional thrombi that may extend into the vena cava and right atrium, Table 18.9 T1 T2 T3
TN Staging
to the kidney
> 7 em, confined
to the kidney
Perirenal
invasion
inferior vena T4
Extension
N1
Solitary
node
N2
Solitary
node
<Scm N3
criteria for renal cell carcinoma
7 em, confined
1\
>5cm
and/or
involvement
of the
cava or renal vein
through <:
Gerota's
fascia
2 em
>2 em or multiple
nodes,
all
Renal Cell Carcinoma (Hypernephroma, RCC)
a
b
Fig. 18.26 Renal cell carcinoma, (a) Small T1 mass on the renal border that is hyperattenuating during the arterial phase. Nearby is an intensely enhancing paraaortic metastatic lymph node (N2). (b) Multiple masses
c in both kidneys (4 x 1 16) due to multifocal renal cell carcinoma. (c) Large hypovascular T1 mass (5 cm) due to a papillary renal cell carcinoma without lymph node metastases.
Fig. 18.27 Small renal cell carcinoma at the lower pole of the kidney. A volume-rendered display (4 x 1/6) demonstrates the spatial relationship of the tumor to the kidney, the renal vessels, and the ureter.
a
b
a
b
Fig. 18.28 Extensive, non homogeneously enhancing, partially necrotic renal cell carcinoma with extension into the renal pelvis (T3) (a). The low-dose CT pyelogram (4 x 1/6) of a similar patient demonstrates the distortion of the renal pelvis (b).
665
18 The Kidneys
Thrombus enhancement after contrast administration indicates that the thrombus contains tumor components (Fig. 18.29). Involvement of the vena cava is three times more frequent with right -sided tumors. Depiction of tumor extension into the renal vein and inferior cava is eq ually possible with spiral or mul tislice CT as it is with MR1. A thrombus in the renal vein is often seen best during the corticomedullary phase (peak enhancement of the vein), while detection of an inferior vena cava (lVC) thrombus is superior in the nephrographic phase. Caution has to be exercised that flow artifacts in the IVC, related to nonopacified blood originating from below the renal veins, are not misinterpreted as a tumor. In case of doubt, coronal reformats may help, and delayed scans show a homogeneous enhancement in the inferior vena cava. Lymph nodes are considered pathologic when their short axis diameter is \ cm or more, with initial nodal involvement occurring at the level of the renal hil um. Nodal metastases by hypervascular tumors may show intense enhancement and require differentiation from dilated veins (Fig. 18.29). The sensitivity of CT for detection of lymphadenopathy has been reported to be over 95%. Attention should be given to invasion of the adrenal glands which occurs in less than 10% of cases and rarely also to the contralateral side. Unless proven otherwise, any change in the adrenal gland, even if confined to the con-
a
• Small hypovascular tumors may not be detected during the corticomedullary phase. • Cystic forms of RCC are relatively frequent (up to 20%). • The combined presence of fat and calcifications suggests that the tumor is RCC, whereas isolated fat deposits are more consistent with angiomyolipoma. • It is difficult to distinguish between neoplastic and inflammatory infiltration of the perirenal fat. •
Management
of Small
«
1.5cm)
Solid Renal Tumors
With improvements in imaging techniques (most notably spiral CT), there has been an increase in the incidental detection of small renal lesions « 1. 5cm). With such small lesions, CT attenuation measurements and the assessment of lesion margins have become more difficult due to increasing partial volume effects. For lesions smaller than 1.5cm that cannot be identified as benign or malignant with CT, angiography, or MRI, due to a lack of specific cri-
c
b
Fig. 18.29 Large renal cell carcinoma (T3, panied by tumor thrombus in the dilated lymph node metastases, and invasion of the fat (a). Small tumor thrombus extending
tralateral gland, is suspicious for metastasis. Infiltration of the psoas muscle is manifested by expansion and a change in attenuation. Direct spread to contiguous organs indicates irresectability. However, this is very uncommon, and diagnosis should not be based on the obliteration of fat planes alone.
N2) accomrenal vein, perinephric into a side
branch of the renal vein in another patient (b). Coronal reformation (4 x 1/6) of a large primitive neuroectodermal tumor (PN ET) infiltrating the inferior vena cava in a 30-year old male (c).
Reno! Cell Carcinoma
terta, Bosniak proposes the diagnostic and therapeutic guideltnes summarized in Table 18.10, His "watchful waiting" approach in doubtful cases, especially for elderly patients and those who are at risk for surgery, is based on the experience that renal cell carcinomas have a relatively slow growth rate and, with rare exceptions (2 % of 254 observed tumors), tended to metastasize only when larger than 3cm in diameter. Tumor growth rates vary considerably (between Ocm and 1.6 ern/year in a study by Birnbaum), and can be determined by close follow-up examinations, Knowledge of growth rates allows for a more informed decision regarding the need for surgel)', Management
of Indeterminate
Renal
Tumors
With careful technique and use of the diagnostic criteria previously described as well as consideration of patient history, very few renal masses should be left undetermined, Technically indeterminate CT images are often the result ofa scan done for unrelated purposes, not directed at the evaluation of a renal mass, Repeat scans with optimized technique will provide definite information in most cases, The kidneys should be examined with a precontrast scan, and a biphasic post-contrast examination in the corticomedullary and nephrographic phase using spiral or multislice CT A slice thickness not larger than 3 mm should be employed during all scans, While the nephrographic phase is most sensitive for detection of lesions, the corticomedullary phase is useful for lesion characterization, Hypervascular lesions
Table 18.10 I.
Management
Attenuation
II. Attenuation Patient>
<20
of small «1.5 em), potentially solid renal lesions (after Bosniak, 1995)
HU
with
ReC)
are highly suggestive of renal cell carcinoma, However, the diagnosis should never be based on the corticomedullary phase alone: enhancement differences between the medulla and the cortex may be misinterpreted as medullary lesions, and hypovascular tumors may be misinterpreted as cysts because only some 70% of neoplasms enhance more than 10 HU during the arterial phase, Small hypervascular lesions should not be misinterpreted as a prominent column ofBertin in the arterial phase, Measured attenuation in an identical ROI should be compared for the pre contrast scan and the nephrographic phase, Cystic masses in the Bosniak type 1Il category may be considered indeterminate, because they have about a 50% chance of being malignant In such cases, clinical factors, including age, other illnesses, and patient risk for surgery, direct patient management either towards close follow up (every six months for two years) or surgery, Rarely do other diagnostic procedures aid in assessment of these lesions, Other indeterminate lesions include angiomyolipoma with little or no fat, oncocytoma, adenoma, and chronic inflammatory processes [xanthogranulomatous pyelonephritis} that all may show a CT appearance indistinguishable from a malignant neoplasm. Delayed CT (at least "15 minutes after the initial postcontrast C1) was found to be helpful in differentiating renal neoplasms from high density cysts by demonstrating de-enhancement as a proof of vascularity and, hence, neoplasm (see Fig, 18.18),
>20 HU
7 5 years
(Hypemephroma,
m.iltiple rrorbidey
and
high
surgical
Benign uncomplicated
cyst
Differential diagnosis:
cyst or tumor
Watchful waiting
risk Younger patient and/or acceptable surgical risk Lesion does not enhance «10 HU attenuation Lesion is hypervascular (>SOHU attenuation
during
the arterial
difference) enhancement
Lesions shows >10 HU cular -> No interval growth -c- Interval growth
difference)
phase
but is not hypervas-
Spiral CTwith intravenous contrast medium: Cyst Suspected hypernephroma; surgical removal Follow-up at 6 and 12 months Benign lesion (adenoma, Surgical removal
cyst)
18 The Kidneys
Metastases
Lymphoma
The most frequent source of renal metastases is bronchial carcinoma, followed by tumors of the breast, colon, stomach, cervix, ovaries, pancreas, prostate, skin, and the contralateral kidney. The kidney is the fifth commonest organ involved by metastases. In patients with known metastatic disease, each newly discovered renal tumor is most likely a metastasis. Renal metastases are bilateral in 50% of cases. Generally they are small and are accompanied by metastases in other organs.
Primary renal involvement by lymphoma is very rare, because the kidneys do not contain lymphatic tissue. But secondary renal involvement by non- Hodgkin lymphoma is a relatively common occurrence (30-50%). Outside the CNS, the kidney is the organ most frequently involved by post-transplantation lymphomas (PTLD).
•
CT Morphology
Metastases are isoattenuating or slightly hypo attenuating on noncontrast CT. Generally they are hypovascular and show relatively homogeneous enhancement (Fig. 18.30). Because they occur in the peripheral part of the renal parenchyma, they rarely cause hydronephrosis. Metastases from mucinous carcinoma (e.g., of the colon) may contain calcifications. Metastatic melanomas tend to infiltrate the perirenal space. If renal metastasis cannot be diagnosed with sufficient confidence from the clinical context, a fine needle aspiration or core biopsy may be performed.
•
CT Morphology
Four different patterns of involvement have been described (Table 18.11).Bilateral multifocal involvement (Fig.18.31a) is three times more common than a solitary focal mass. The other manifestations are retroperitoneal disease extending into the kidney and diffuse renal infiltration causing function impairment (Fig.18.31b). All forms appear on CT scans as hypovascular lesions that may be visible only after IV contrast administration. Invasion of the psoas muscle is a relatively common finding. Encasement of the renal artery is more common than with hypernephroma. Poorly marginated lesions, diffuse renal enlargement, and diffuse infiltration of the perirenal tissue are morphologic criteria that are more consistent with lymphoma than with a primary renal tumor. Because of the different therapeutic implications (surgical removal vs. chemotherapy), biopsy confirmation should be obtained in doubtful cases.
Table 18.11 Patterns of renal involvement by malignant lymphoma Bilateral multinodular involvement: Multiple poorly marginated lesions (1 -Scm) visible only after contrast administration. Kidneys are usually enlarged. Solitary nodular mass: Circumscribed hypovascular mass that may distort the renal contour and may infiltrate the perirenal fat. Fig. 18.30 Renal metastases usually present as hypovascular lesions (arrows). Note the left adrenal (arrowhead) and multiple liver metastases from bronchogenic cancer.
Retroperitoneal
mass:
Extensive mass that spreads to the renal sinus or entire kidney and encases the ureter. Diffuse renal enlargement: No circumscribed mass, decreased contrast enhancement (parenchyma! phase) and reduced excretion.
''''x
II
Nephroblastoma (Wilms' Tumor) Fig. 18.31 Renal involvement by malignant lymphoma. (a) Multiple hypoattenuating nodules, similar to the pattern seen with renal metastases. (b) Diffuse renal involvement with encasement of the renal pelvis and concomitant involvement of the retroperitoneum and vena cava (arrow).
a
Nephroblastoma (Wilms* Tumor)
Nephroblastoma arises from immature renal parenchyma and is the commonest malignant abdominal tumor in children 1 to 5 years of age. Up to 15% of cases are bilateral. There is early invasion of the renal vein and early hematogenous metastasis to the lungs (10%). CT and MRI are useful for detecting tumor spread and particularly for excluding involvement of the contralateral kidney. •
CT Morphology
Nephroblastomas are usually quite large when diagnosed, but they very rarely infiltrate adjacent organs. The tumor appears as a hypoattenuating mass that causes marked distortion and displacement of the renal parenchyma (Fig. 18.32). Areas of intratumoral hemorrhage and necrosis cause heterogeneous attenuation.
b
Only about 10% of nephroblastomas contain calcifications, and 7% contain fatty tissue. Following contrast administration, the viable portions of the tumor show inhomogeneous hypervascularity. Invasion of the renal veins can be demonstrated in 30-40% of cases and is commoner on the left side. Enlarged para-aortic and renal hilar lymph nodes are metastatic in more than 20% of cases, but most nodal enlargement is due to reactive hyperplasia. Very large tumors can be difficult to distinguish from adrenal neuroblastoma. However, neuroblastomas are more often calcified, and Wilms' tumors more strongly distort the shape and contour of the kidney. Multiplanar reformations from thin-section spiral or multislice CT can help determine the kidney or the adrenals as the organ of origin.
Fig. 18.32 Wilms'tumor in a 3year old patient appears as a large, heterogeneous mass causing renal deformation, (a) Axial scan. (b) Coronal reformation (3/5/3).
a
b
669
18 The Kidneys
Tumors of the Renal Pelvis and Ureter
Tumors of the renal pelvis account for only 10% of all renal neoplasms. The clinical symptoms of tumors of the renal collecting system are hematuria, hyclronephrosis, and flank pain. Papillomas are usually multiple and may involve any part of the urinary drainage system from the renal pelvis to the bladder. They are classified as premalignant lesions. Urothelial carcinomas (transitional cell carcinomas) are the commonest malignant tumors of the renal pelvis (80-90%) and are often associated with multiple papillomas (>50%). From 25-40% of urothelial carcinomas are multiple and additionally involve the ureter or bladder. Approximately 10% are bilateral. Males are predominantly affected (4: 1), and most patients are over 60 years of age. Various industrial materials (dyes, chemicals) can cause urothelial carcinoma with a latent period of 20 years or more. Squamous cell carcinoma (approximately 10%) and adenocarcinoma (approximately 1%) are associated with chronic epithelial irritation (renal calculi in 50% of patients) and leukoplakia. Squamous cell carcinoma has a poor prognosis and tends to invade extrarenal structures. •
CT Morphology
Squamous cell carcinoma and urothelial carcinoma cannot be reliably differentiated with CT. Small tumors may be easily missed due to the lack of obvious deformity or an obstructive effect. The attenuation values of urothelial carcinoma range from 8-40 HU and increase only moderately after contrast administration (ap-
proximately 30 HU). Dependent on size and location, the tumor can be seen as a soft tissue mass within the fluid-filled pelvis on a precontrast scan. During the excretory phase of enhancement, the tumor appears as a filling defect in the collecting system. CT urographic images in 2D or 3D technique (MPR, MIP or VRT) are a useful adjunct to the axial images and provide a comprehensive overview of the collecting system. The tumor may present a nodular or lobulated contour of the renal pelvis (Fig.18.33a, papillary tumor), it may infiltrate the renal parenchyma (Fig.18.33b, infiltrating tumor), or it may form a plaque-like growth on the wall of the renal pelvis (Fig. 18.34). Obstructive tumors lead to hydronephrosis or the obstruction of a caliceal group. Occasionally an "obstructive" nephrogram (delayed onset of the parenchymal phase) is the only evidence of a renal pelvic tumor. Blood clots may form in the urinary collecting system from an acutely bleeding tumor. They have smooth margins, show higher attenuation than tumors (depending on the age of the clot), and do not enhance with intravenous contrast medium. If fibrolipomatosis of the renal pelvis is particularly dense, it maybe hard to distinguish it from a transitional cell carcinoma of the renal pelvis (Fig. 18.35). Ureter carcinomas frequently present only with indirect signs such as hydronephrosis and prestenotic ureteral dilatation. The tumor itself appears as a soft tissue mass with a higher attenuation than the urine-filled, dilated ureter proximal to it, or as a filling defect or ill-defined thickening of the ureteral wall (Fig. 18.36). It has been stated that all tumors except those smaller than 5 mm in diameter can be detected by CT.
Fig. 18.33 Renal pelvic carcinomas. (a) Small transitional cell carcinoma appearing as a polypoid filling defect during the excretory phase after contrast administration. An uncomplicated cyst is present as an associated finding. (b) This obstructive transitional cell carcinoma forms a plaque-like growth on the walls of the renal pelvis, leading to secondary hydronephrosis.
a
b
Tumors of the Renal Pelvis and Ureter
With multislice CT urography, this size limit may be further reduced. In addition to the information provided by retrograde pyelography, CT reveals the extraureteral soft-tissue extension and nodal involvement. Two-thirds of all tumors involve the distal third of the ureter. The possibility of multifocallesions in the ureter or bladder should be considered whenever a urothelial carcinoma in the renal pelvis is detected.
Fig. 18.34 Transitional cell carcinoma of the renal pelvis. The extent of disease is excellently displayed on multiplanar reformations.
• When hydronephrosis is present, wall enhancement in the pyelocaliceal system or an indistinct wall boundary suggest a neoplastic or inflammatory cause of the obstruction. • CT cannot differentiate a chronic inflammatory, postoperative, or radiogenic ureteral stenosis from a tumor-related stenosis.
Fig.18.35 The renal sinus is completely filled with solid tumor material (transitional cell carcinoma) that displaces the calices (a). Fibrolipomatosis may present with similar findings but only minor displacement of the calices (b).
a
a
b
b
c
Fig. 18.36 Ureteral carcinoma may present with massive excretory tract dilatation and may spread from the ureters (arrowheads) to the renal pelvis (a). Multislice
CT may be able to demonstrate a small tumor (b), which can best be evaluated on curved reformations (4 x 1/6, CPR c).
671
100HU. CT is therefore particularly useful in distinguishing radiolucent stones from soft-tissue filling defects. Whenever a urothelial tumor (carcinoma or papilloma) is suspected, the entire urinary tract should be evaluated.
A primary ureter carcinoma can usually be distinguished from secondary invasion by cervical, bladder, prostatic, or rectal carcinoma on the basis of associated findings. Ureteral stones, even when nonca1cified, have an attenuation of greater than
Diffuse and Inflammatory
Renal Diseases
Acute Pyelonephritis
Acute pyelonephritis is an acute bacterial interstitial nephritis with involvement of the renal pelvis. Focal lesions may occur in addition to diffuse interstitial involvement. Acute focal bacterial nephritis is a severe circumscribed form of pyelonephritis. Empliysematous pyelonephritis is a particularly severe form of acute pyelonephritis caused by infection with gas-forming organisms. The infection is mostly due to Escheridna coli or other gram-negative organisms. It occurs mainly in diabetics (90% of all affected patients), and 40% of cases are associated with urinary tract obstruction. CT is the most sensitive modality for detecting this infection, which has a fatal outcome in 50% of patients. Acute pyelonephritis is not a primary indication for CT, but a CT examination is indicated in patients with suspected complications
a
b
Fig. 18.37 Acute pyelonephritis, (a) Parenchymal phase image of a focal bacterial nephritis presenting as ill-defined tumor-like lesions, (b) Parenchymal phase image in focal bacterial nephritis (4 x 1/6, coronal MPR) shows a wedge-shaped hypodense area with a subcorti-
(abscess formation, gas formation, or spread of infection into the perirenal (perinephric) space. •
CT Morphology
In acute pyelonephritis, precontrast scans may be normal or may show focal or diffuse renal enlargement with slightly decreased CT attenuation due to edema (Table 18.12). There may be round or wedge-shaped focal areas of decreased attenuation (necrosis, edema) or increased density (hemorrhage) on the pre contrast scan. The spread of inflammation to the perirenal space is manifested by streaky infiltration of the perirenal fat and thickening of Cerota's fascia. During the parenchymal phase of contrast enhancement these rounded, linear, or wedgeshaped areas show a diminished enhancement with a typical radially oriented pattern (Fig. 18.37a) across the corticomedullary boundary. Later the enhancement pattern is reversed,
c
d
cal abscess (arrow), (c) Delayed scans in acute pyelonephritis show a striated pattern of contrast retention in the affected collecting ducts, accompanied by multiple microabscesses. (d) Emphysematous pyelonephritis with multiple gas inclusions.
Intrarenal or Perirenal Abscess Table 18.12
Differential
diagnosis
of large kidneys
Focal or diffuse nephritis Acute pyelonephritis Xanthogranulomatous pyelonephritis Lymphoma Renal plasmacytoma Hypernephroma Functional hypertrophy Acute nephropathy
with delayed scans (> 15 min) showing a typical radial striated pattern of contrast retention caused by the obstruction of collecting ducts by inflammatory material (Fig. 18.37b). Focal bacterial nephritis (focal pyelonephritis) is characterized by a circumscribed mass-like renal enlargement. During the parenchymal phase of enhancement, the affected area appears hypo attenuating and poorly demarcated with respect to the surrounding parenchyma. The appearance is similar to that of focal lymphomatous involvement or a hypovascular tumor. As in acute pyelonephritis, delayed scans demonstrate radial striations in the inflammatory focus (Fig.l8.37b, c). The presence of liquefaction (nonenhancing area) indicates that the focal nephritis has developed into an abscess. In empHysematous pyelonephritis, the renal parenchymal changes are accompanied by gas collections at parapelvic, subcapsular, or retroperitoneal sites (Fig.l8.37 c). Gas is always seen also in the parenchyma and not only in the collecting system as may be the case with reflux of gas from the bladder or gas due to fistulae. Further differentiation is required from gas collections caused by an abscess, therapeutic procedures (embolization, aspiration biopsy, catheterization), orpost-traumaticinfarction. The CT manifestations of complicated pyelonephritis may persist for months even with appropriate anti biotic treatment.
rial nephritis, or septicemia. In most cases, microabscesses coalesce to form a larger abscessed area that establishes a communication with the renal collecting system or perirenal space. Extrarenal abscesses may develop in a setting of severe nephritis or a disseminated retroperitoneal infection (e.g., psoas abscess), or they may reach the perirenal space by hematogenous spread from a distant focus. Superinfected cysts and hematomas can lead directly to abscess formation, bypassing the stage of focal microabscesses. •
CT Morphology
Abscesses usually show higher attenuation (ID30 HU) than uncomplicated cysts or the pyelocahceal system and may produce a mass effect. The lesions are most clearly delineated during the parenchymal phase of enhancement (Fig. 18.38). An enhancing rim is not always visible, even when the abscess is extrarenal. Fungal infections may produce multiple microabscesses that are scattered throughout the parenchyma and may calcify. Fungal mycelia in the renal pelvis or ureter may have an obstructive effect. Chronic, asymptomatic abscesses are very difficult to distinguish from cystic renal cell carcinoma by their CT features alone. An abscess can be differentiated from an uncomplicated cyst by detecting a perifocal reaction and inflammatory changes in the surrounding renal
Intrarenal or Perirenal Abscess
Renal abscesses are rare, accounting for only about 2 % of all renal masses. They are either the result of ascending infection (gram-negative urinary pathogens) or result from hematogenous seeding [Staphylococcus aureus). As such, they may develop as a complication of hydronephrosis, acute pyelonephritis, focal bacte-
a
b
Fig. 18.38 Abscesses. A superinfected cyst presents a cystic mass with a thickened enhancing wall (a). renal abscess has a hypodense center surrounded by ill-defined rim of granulation tissue (arrowheads, Note the concomitant focal nephritis (arrow).
as A an b).
674
18 The Kidneys
parenchyma. In equivocal cases, a needle biopsy may be required. Placement of drains has been show to improve outcome. Pyonephrosis
Pyonephrosis results from infection of a hydronephrotic kidney. It is most frequently caused by an obstructing calculus. The term pyocalix is used if only one or a group of calyces is affected. Renal function is usually impaired. CT is used only to exclude a concomitant abscess in patients with equivocal ultrasound findings. •
CT Morphology
The collecting system shows local or generalized dilatation. Excretion of contrast material into the renal pelvis is poor and delayed. The inflamed wall of the collecting system is frequently thickened and shows contrast enhancement (Fig. "18.39).The attenuation values ofthe infected fluid range from 20-30 HU (Table 18.13), and dependent layering may occur between the urine and pus.
a
b
Fig. 18.39 Pyonephrosis with enlargement of the renal pelvis and wall enhancement (a). There is stranding of thefatinthevicinityoftheinflamed renal pelvis. Notethe caliceal diverticulum (arrow) in the excretory phase (b).
Table 18.13 Differential tenuation values (20-80 Renal pelvic tumors Blood clots Pyonephrosis Fungal mycelia Malacoplakia
diagnosis of increased atH) in the pyelocaliceal system
Renal Tuberculosis
Renal symptoms of tuberculosis develop in 4-8% of immunocompetent patients and in 20% of immunocompromised patients within W-20 years after the initial pulmonary infection. Urogenital tuberculosis is the commonest extrapulmonary form of the disease. Tubercle bacilli reach the kidneys by the hematogenous route, establish intrarenal granulomas, and may be reactivated following a dormant period of 5-20 years. Urographic studies are superior to standard CT in demonstrating papillary changes and will usually suggest the correct diagnosis. Multislice CT can be expected to become the technique of choice since it can show the collecting system as well as the parenchymal and extra parenchymal abnormalities. •
CT Morphology
The CT findings in renal tuberculosis are highly variable and usually nonspecific. The productive form presents with multiple hypo attenuating miliary tubercles in the renal cortex that heal without sequelae, leaving behind disseminated punctate calcifications. The ulcerative-cavernous form is associated with parenchymal destruction and deformation of the pyelocaliceal system. Abscess formation may occur (Fig. 18.40). Caliceal strictures develop, producing hydrocalices. Extensive patchy or eggshell-like calcifications develop in the necrotic material (see Fig.18.15c). Further progression leads to parenchymal atrophy and Pyonephrosis secondary to ureteral strictures. Unless adequately treated, the infection progresses to an end stage in which the kidney is shrunken and completely calcified (caseous pyonephrosis, mortar kidney). Xanthogranulomatous
Pyelonephritis
Xanthogranulomatous pyelonephritis is a rare disease that predominantly affects middle-aged women. It is a chronic, destructive inflammation of the renal medulla and cortex that is usually secondary to an intrarenal urinary obstruction. The affected parenchyma is permeated by characteristic xanthomatous cells (lipid-laden macrophages-xanthoma cells). The changes are almost always unilateral. CT is used to differen-
Chronic Pyelonephritis
Fig. 18.40 CT in tuberculous pyonephrosis may show ill-defined enhancement of the narrow parenchymal rim and an extensive perirenal abscess (a). Later in the disease, calcifications (arrowheads) and deformation and congestion of the calices (arrows) occur (b).
a
tiate this condition from tumors and to detect perinephric extension. •
CT Morphology
Diffuse xanthogranulomatous pyelonephritis is distinguished from the less common focal form. Typically the kidney is enlarged but retains a reniform shape (Fig. 18.41, see Table 18.12). Approximately 70% of cases show renal pelvic stones (often staghorn-shaped), caliceal stones, or renal parenchymal calcifications. Multiple focal low attenuation masses (-15 to 20 HU) are scattered throughout the involved portions of the kidney; they represent dilated debris-filled calyces and xanthoma collections. They may be arranged in a rosette-like pattern that resembles hydronephrosis. Usually the renal pelvis is poorly delineated due to encasement by inflam-
b
matory tissue, but obstruction of the renal pelvis is uncommon. CT cannot differentiate inflammatory distortion and enlargement of the calices from abscesses or necrotic cavities. Inflammatory involvement of the perirenal fat is seen in more than 90% of cases. If disease is extensive, the involvement of bowel loops can lead to the formation of a convoluted mass with fistulous tracts. Thrombosis of the renal veins or vena cava may occur. •
Differential
Diagnosis
Differentiation is required from renal cell carcinoma (tumor calcifications, rarely calculi), pyonephrosis (dilatation of renal pelvis), and lymphoma (usually bilateral, higher CT numbers, rarely calculi). Chronic Pyelonephritis
The diagnosis is based on clinical and urographic criteria. The suspicion of chronic pyelonephritis is not a primary indication for CT. •
Fig. 18.41 Xanthogranulomatous pyelonephritis manifested by hydronephrosis-like enlargement of the kidney and a calculus in the non-dilated renal pelvis.
CT Morphology
CT shows irregular narrowing of the renal parenchyma (Table 18.14). The calices are deformed by the parenchymal retraction, and their tips may extend to the fibrous capsule (Fig.l8.42a). Renal function is severely impaired. Focal areas of normal parenchyma appear as pseudotumors that are distinguished from neoplasms by their normal enhancement characteristics (see Fig. 18.23).
675
676
18 The Kidneys
Fig. 18.42 Small kidney. Chronic pyelonephritis with multiple scars (a), chronic glomerulonephritis with reduction of cortex and medulla (b).
b
a
Table 18.14
Differential
Inflammatory:
diagnosis
of small kidneys
Chronic pyelonephritis Glomerulonephritis Tuberculous mortar kidney
Vascular:
Renal arterial stenosis Renal infarction Prior renal cortical necrosis
Other:
Postobstructive atrophy Nephrosclerosis Chronic urate nephropathy Oxalosis Collagen disease
Chronic Interstitial Nephritis (Analgesic Nephropathy)
Abuse of analgesics (aspirin, acetaminophen, NSAIDS like indomethacin-doses of at least 1 g daily for 3 years, or a total of 3 kg) is the most frequent cause of chronic interstitial nephritis. The main clinical features are colicky pain (due to the passage of sloughed papillae) and chronic renal failure. There is no primary indication for CTevaluation. •
CT Morphology
CT in the late stage of disease shows a symmetrical decrease in renal size along with papil-
Vascular
lary calcifications Fig.18.13a) . •
Differential
at
typical
sites
(see
Diagnosis
Apical papillary calcification may also be seen in diabetic nephropathy, tuberculosis, sickle cell anemia, and in urinary tract obstruction with secondary infection. Glomerulonephritis
Glomerulonephritis is responsible for 30% of cases of end-stage renal failure. CT does not have a role in the diagnosis of inflammatory and immunologic changes. •
CT Morphology
In the acute stage the kidneys are normal-sized or slightly enlarged with a widened parenchymal rim that shows reduced enhancement on postcontrast scans. The chronic stage is marked by uniform narrowing of the parenchymal rim, with atrophic changes predominantly affecting the renal cortex. The calices and papillae are of normal size, but there is relative enlargement of the renal sinus and perirenal space (Fig. 18.42c).
Disorders
Renal arterial abnormalities Chapter 24, p. 900.
are described in
Renal Infarction
Acute renal infarction is most often caused by thrombembolic occlusion of the renal artery or its side branches. The principal sources of emboli are the heart (atrial fibrillation, mitral valve defects) or an aortic aneurysm. The clinical hall-
Reno/ Vein Thrombosis Fig. 18.43 Renal infarction with a wedge-shaped perfusion defect seen on axial sections (a) as well as a shaded surface display (b).
a
b
marks are hematuria and sudden flank pain. Thrombotic infarction may result from arteriosclerosis, vasculitis (polyarteritis nodosa), sickle cell anemia, or trauma. Renal infarction can be accurately diagnosed with CT. If an acute embolism is suspected, however, angiography is preferred for its combined diagnostic and interventional capabilities. •
CT Morphology
Infarcted areas of the renal parenchyma appear as wedge-shaped areas of decreased attenuation on contrast-enhanced scans (Fig. 18.43). The subcapsular parenchyma still has a collateral blood supply and may appear as a thin rim of enhancing tissue (the cortical rim sign). Septic embolism may lead to abscess formation. Late sequelae include scarring and contour distortion of the renal cortex and shrinkage of the infarcted tissue.
Fig. 18.44 CT in acute cortical necrosis shows an enhancing renal medulla adjacentto a nonenhancing renal cortex. Subcapsular perfusion is maintained by capsular arteries.
After about one month, calcifications develop in the cortex (in 50% of cases) and the kidney undergoes progressive atrophy.
Renal Cortical Necrosis
Renal cortical necrosis is a rare cause of acute renal failure that may occur in various clinical settings (sepsis, shock, complications of pregnancy). Renal tubules, glomeruli, and connective tissue become necrotic as a result of intravascular coagulation. There is no indication for CT scanning. •
CT Morphology
In the acute stage, contrast enhancement of the kidney is confined to the renal medulla (Fig. 18.44). The hypo attenuating cortex is surrounded by a thin peripheral enhancing rim. Contrast excretion is observed.
Renal Vein Thrombosis
Renal vein thrombosis may be caused by various neoplastic, infectious, and metabolic disorders. It may be the only sign of an occult renal malignancy. It is quite common (up to 33%) in patients with nephrotic syndrome. CT can clearly define renal vein thrombosis when optimum technique is used (rapid bolus injection, imaging during maximum vascular enhancement), but CT scanning is rarely indicated as the primary diagnostic mode.
677
18 The Kidneys
•
CT Morphology
With a complete occlusion, the renal vein is enlarged to more than 1.5cm in diameter. A filling defect is seen after IV contrast administration, and opacification of the vein is absent or diminished compared with the opposite side (Fig. 18.45). Contrast enhancement in the thrombus suggests a neoplastic pathogenesis. Collateral flow leads to dilatation of the gonadal veins, capsular veins, and periureteral veins. Perfusion defects may be found in the renal parenchyma. Fig. 18.45 Bilateral renal vein thrombosis in a patient with nephrotic syndrome.
Trauma and Postoperative Changes Renal Trauma
Table 18.15 Traumatic
Causes of renal hematomas causes
Blunt trauma is responsible for 80% of renal inBlunt abdominal trauma juries and is frequently caused by car accidents, Stab injury falls or fights. The principal types ofrenal injury Percutaneous renal biopsy are contusion (intrarenal edema), hematoma Lithotripsy (subcapsular or parenchymal), laceration (subPostoperative capsular parenchymal tear with extravasation of Montraumatic causes blood and urine), and rupture (a parenchymal tear with capsular injury). Subcapsular hemaTumors tomas are found in 15% of patients after shockCoagulation disorders Nephritis wave lithotripsy and in up to 30% of patients Tuberculosis after percutaneous renal biopsy (Table 18.15). Renal cysts (cystic kidneys) They are treated conservatively. Hydronephrosis Renal injuries can be graded according to Panarteritis nodosa the organ injury score of the American AssociaAtherosclerosis Renal artery aneurysm tion for the Surgery of Trauma (Table 18.16). Accurate grading has therapeutic consequences and is related to outcome. Category I-III lesions are contusions and lacerations (80%) that are managed conservatively, while category V le- individual patient. A suspected vascular injury, sions require surgery in all cases, usually ne- which is more common in children and typically follows deceleration trauma, requires evaluation phrectomy. There is controversy over category by arterial angiography or CTA. Suspicion of inIV lesions, which may be managed conservatively with or without interventional radiology jury to the renal colI ecting system or ureters retechniques. quires delayed imaging during the excretory Hemodynamically stable patients with mi- phase. croscopic hematuria are treated conservatively • CT Morphology and probably do not need additional imaging. Fresh hemorrhage appears on noncontrast scans CT is the method of choice for evaluating trauma patients with macroscopic hematuria and plays as a hyperattenuating area that may persist for an important role in selecting treatment for the longer than four weeks. Mild injuries may cause
Renol Trauma Table 18.16 Grade
AAST organ injury scale for renal injuries (Moore, 1989) Mechanism
Description
Contusion Hematoma
Microscopic or gross haematuria, urological studies normal Subcapsular, nonexpanding without parenchymal laceration
II
Hematoma Laceration
Nonexpanding perirenal haematoma confined to renal retroperitoneum <1 cm parenchymal depth of renal cortex without urinary extravasation
III
Laceration
>1 cm depth of renal cortex, without collecting system rupture or urinary extravasation
IV
Laceration
Parenchymal laceration extending through the renal cortex, medulla and collecting system Main renal artery or vein injury with contained haemorrhage
Vascular
v
Laceration Vascular
Parenchymal laceration extending through the renal cortex, medulla and collecting system Avulsion of renal hilum which devascularizes kidney
Advance one grade for multiple injuries to same organ
a simple perirenal hematoma that is not accompanied by renal changes. Subcapsular hematoma is the result of a renal injury that leaves the fibrous capsule intact. Large Subcapsular hematomas can cause renal compression with secondary hypertension (Page kidney). A heterogeneous reduction in parenchymal enhancement following contrast administration reflects a renal contusion with a local pel-fusion deficit caused by edema or diffuse intrarenal hemorrhage. Hematomas are iso- to hyperattenuating to the renal parenchyma on precon-
a
b
Fig. 18.46 Renal trauma, (a) Renal contusion with a crescent-shaped Subcapsular hematoma. (b) Intrarenal pseudoaneurysm (arrow) in a ruptured kidney. Note the concomitant hepatic hematoma (H). (c) Low-density
trast scans and better demarcated on postcontrast scans (Fig.18.46a). Lacerations and ruptures are accompanied by perirenal hematomas of variable size, not compressing the kidney but potentially displacing the organ. Injuries of the renal arteries lead to segmental or global pel-fusion defects (intimal lesion with consequent arterial thrombosis). Severe parenchymal injuries may be associated with the development of a false aneurysm (Fig.18.46b). Complete avulsion of the renal artery leads to spontaneous contraction of the
c areas of infarction (arrowheads) are accompanied by increased contrast retention in the parenchyma and contrast extravasation into lymphatic vessels (arrows).
679
18 The Kidneys
vessel and complete infarction (Fig.l8.46c), while avulsion of the renal vein leads to an extensive retroperitoneal hematoma. Injury to the renal collecting system is usually manifested by a localized extravasation of urine.
Postoperative Changes Early postoperative complications include hematoma and abscess. Collections of retroperitoneal gas may persist for more than a week after surgery. CT Morphology
After a right-sided nephrectomy. the right colic flexure and the duodenum shift posteriorly to occupy the nephrectomy site. After a left-sided nephrectomy. the renal bed is occupied by loops of small bowel and by the tail of the pancreas. which falls posteriorly. Adequate bowel opacification with oral contrast medium is important to avoid errors of interpretation. Ipsilateral thickening of the psoas muscle is suspicious for recurrent or residual tumor. CT after partial nephrectomy demonstrates a small kidney and scar tissue. The cortical defect left by a wedge resection or enucleation may be hard to detect or may be occupied by retroperitoneal fat.
Renal Transplantation Hematoma. lymphocele, urinoma, and abscess are the most important immediate complications of renal transplantation. They are usually adequately examined by ultrasound and radionuclide techniques. CT is used in ecquivocal cases but not useful for detecting complications such as acute rejection or cyclosporine- induced changes. Chronic complications of renal transplantation include chronic rejection, development of stenoses or thrombosis of graft vessels, ureteral stenosis, and the development of posttransplant lymphoproliferative disorder (PTLD) or other malignant diseases. While renal perfusion in chronic rejection is best evaluated by ultrasound, CT is superior to detect tumorous complications of transplantation. For vascular evaluation, MRA should be considered as an alternative in patient with impaired renal function.
The CT technique strongly depends on the diagnostic question. Noncontrast scans are performed to rule out hematoma, CTA is used for detecting stenosis of the graft artery in patients with equivocal ultrasound findings, scans in the nephrographic phase are indicated for suspected lymphoceles or abscesses, and scans in the excretory phase (c- 15 minutes p.t.) are necessary for urinoma and may be helpful to detect ureter stenosis with spiral or multislice CT. CT Morphology
Attenuation measurements are helpful in differentiating post-transplantation fluid collections. The commonest collections are lymphoceles, which may become quite large and are of water density. Urinomas can be identified by the presence of opacified urine on delayed scans. Urine extravasation signifies a leaky ureteral anastomosis or ureteral necrosis due to ischemia. Postoperative hydronephrosis may be a result of transient edema at the ureteral implantation site, extrinsic compression of the ureter, or ureteral obstruction by blood clots. CT urography is an excellent alternative to IVU to distingutsh between strictures and extrinsic ureteral compression. PTLD may present as pathologic lymph node enlargement anywhere in the body but may also present as a focal mass in the region of the transplanted kidney (Fig.18.47a). Chronic rejection may present as a hilar mass that is caused by local inflammation and thus may mimic PTLD (Fig.l8.4 7b). For CTA evaluation of the graft arteries, 3D reconstructions (e.g., volume rendering) provide the best initial orientation (Fig.l8.47c) due to the frequently complex anastomotic relationships (patch grafts, multiple renal arteries). The findings can then be scrutinized on axial images and multiplanar reformats.
Renal Transplantation
a
b
Fig. 18.47 Renal transplantation with post-transplant lymphoma presenting as ill-defined tissue surrounding the renal pelvis (a). A mass-like appearance can also be seen with inflammatory tissue due to chronic rejection
c (b ). Volume rendering ney (c).
of a transplanted
horseshoe
kid-
681
19 Urinary Bladder C. Schaefer-Prokop
Anatomy
684
Examination
Technique
Enhancement Congenital
685
Characteristics
Anomalies
688 688
Persistent Urachus (p. 688) Diverticula (p. 6S9) Benign Generalized
Bladder Wall Thickening
Bladder Tumors
689 691
Bladder Carcinoma (p. 691) Mesenchymal Tumors (p. 693) Inflammatory
Bladder Diseases
694
Cystitis (p. 694) Malacoplakia (p. 694) Enterovesical Fistulae (p. 694) Trauma Postoperative
695 Changes
......................................................
Cystectomy and Urinary Diversion (p. 696) Recurrent Tumor (p. 697)
696
19 Urinary Bladder
The primary modality for examining the urinary bladder is cystoscopy. CT can be used for tumor staging but is inferior to MRI for differentiating the layers of the bladder wall. Generally the indications for CT are limited chiefly to the detection of postoperative complications and followup (Table 19.1).
Table 19.1 der
Indications
for CT examination
of the blad-
Tumor diagnosis
Bladder carcinoma: advanced stages
Oncologic
After cystectomy: every 6 months for the first 2 years, then once yearly Suspected tumor recurrence
follow-up
Postoperative
Complications: hemorrhage, abscess, urinoma, fistulation
Anatomy
The wall of the fully distended bladder is approximately 2-3 mm thick. The nondistended bladder wall is thicker but should not exceed 8 mm. The ureters enter the bladder at the ureteral hiatus on the posterosuperior bladder wall (Fig. 19.1). The ridge between the ureteral orifices is called the interureteric crest. It forms the superior boundary of the trigone, a triangular area on the interior base of the bladder that is bounded by the two ureteral orifices and the internal urethral orifice.
The bladder is surrounded by perivesical fat. In the female it lies adjacent to the uterus and ovaries. In the male the prostate abuts the base of the bladder while the seminal vesicles are related to the posteroinferior bladder wall (Fig. 19.2). No fat planes are interposed between the prostate and bladder. The sigmoid colon is posterosuperior to the bladder, and the rectum is posterior. The superior half of the bladder is covered by peritoneum and is in contact with the terminal loops of the small intestine.
Common femoral artery and vein
Obturator and vein
artery
---------,.
(F.:::::::=:::::::::=:::z:i.",,~---+----J_----
Pelvic venous plexus Sciatic nerve
----------\----I------,i"I
---------~'L
'------
Fig. 19.1
Ureteral orifice Seminal vesicle
"-----I-----}'--------
Sectional anatomy of the male urinary
bladder
at the level of the ureteral
Internal obturator muscle
orifice.
Examination
Technique
Peritoneum Suspensory ligament of ovary »r--:
~
~
\~-14~-7"""'7,/----1--7'-=""~-
Bladder wall
/~ /
Ureteral orifice Semin aI
Ureter Fallopian tube Ovary
~ Uterine round ligament uterine pouch
vesicle
- Bladder wall - Urethra Ureteral orifice Vagina space Corpus cavernosum
Urethra ~
a
b
* Cul-de-sac
(pouch of Douglas)
Fig. 19.2 Sagittal sections through the male pelvis (a) and female pelvis (b). The thickness of the bladderwall with the degree of bladder distention. S = sigmoid, R = rectum, U = Uterus
varies
Examination Technique
The ability to define and evaluate the bladder with CT depends on its degree of distension. Distension makes the bladder wall and intravesical space easier to evaluate, and it raises the interface between the uterus and superior bladder surface from a parallel to a more vertical orientation relative to the scan plane.
der distention can be optimized as needed by the instillation of water or dilute (1 :20) contrast medium. In follow-up examinations, the same degree of bladder distention should be used. Acquisition Technique
Patient Preparation
For patients without indwelling bladder catheters, adequate fluid intake should be provided prior to the CT examination. Having the patient drink 1000-1500 ml of oral contrast agent or water over a time span of some 60-90 minutes prior to the examination yields optimum bladder distension and in addition leads to adequate luminal opacification of bowel loops in the pelvis. Bowel distention is especially important for primary tumor staging and for follow-up scans to detect or exclude recurrent tumor. Indwelling bladder catheters should be occluded at least 30 minutes prior to the examination to allow for adequate bladder filling. Blad-
Suggested protocols for CT scanning of the bladder are summarized in Table 19.2. Partial volume effects can hamper evaluation of the bladder roof and floor if the reconstructed sections are too wide. A section thickness 5mm improves the visualization of details. For optimum evaluation of these regions, however, thin-section spiral CT and multislice CT (/\ 2 mm) should be performed. From this acquisition, highly overlapping images (multislice CT: 1-2 mm section width at 0.7mm increments; spiral CT: 1.5mm increments) should be reconstructed to obtain a secondary raw data set from which high-quality multiplanar reformations in the coronal and sagittal planes can be generated. This allows for
686
79 Urinary Bladder Table 19.2
Suggested
protocols
for CT scanning
of the bladder
General Oral contrast
1000-1500
Bladder
300-400ml of 0.9% NaCI or 300-400 ml of contrast medium
Patient
distention
position
Scan range
Respiratory
ml, administered
Supine with elevated
Windowing
diluted
1 : 10-20
arms total bony pelvis small pelvis from sacroiliac joint to lower border of ischiadic bone
Pelvis: Bladder:
phase
60 min prior to examination
Inspiration Noncontrast CT: Contrast-enhanced CT cystography:
CT:
W/L=300/40 W/L=350/60 W/L=1500/300
Scan parameters
l-slice SC/TF/RI
4-slice' SC
16-slice' SC
axial SWIRl
MPRb SW/RI
Trauma
5/8/4 ~
2-3
1-1.5
5/4
3/4 cor
Tumor staging
3/5/3
CT cystography
2/4/1.5
Contrast
injection
1-slice V/F/D
4-slice V+N/F/D
16-slice V+N/F/DX'
Standard
CT
120/2/70
100 + 50/2/70
100 + 50/2/80
Tumor staging
120/4/40
120 + 50/5/40
120 + 50/5/20A
Excretory phase
>30min
>30min
>30min
t t
1-1.25
0.75-1.25
4/3
3/3 cor
1-1.25
0.75-1.25
4/3
3/3 cor
SC = slice collimation (mm), TF = table feed (rnrnjrotatlon), RI= recon. increment (mm), it = scanning direction, SW = effective section width (mm), MPR = multiplanar reformats, axial = axial sections, cor = coronal, sag = sagittal CM = contrast material. Contrast concentration = 300 mg/ml iodine V = volume of CM (ml), N = volume of saline flush (ml), F = flow rate (ml/s), D = start delay, X = trigger region a Pitch P=TF/(NxSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI = 1-1.5/0.7 radial reformations perpendicular to the bladder wall improve tumor staging c read: 20A = start delay 20 s after reaching 50 HU enhancement in the infra renal aorta
excellent evaluaLion of the roof and floor of the bladder as well as the distal ureters. To improve signal-to-noise ratio, the thickness of the reconstructed sections should be between 2 mm and 3 mm for coronal images and 4 mm and 5 mm for sagittal images. Best results for tumor staging require perpendicular sections through the expanded bladder wall. In this respect, radial multiplanar reformations can be recommended. The fixed center point should be placed so that the tumor-
bearing portions of the bladder wall are sectioned perpendicularly (Fig. 19.3). Additional CT scans in the lateral decubitus position can be helpful in differentiating between intraluminal and mural filling defects. They may also provide useful information for differentiating T3a lesions from T3b lesions that have spread through the bladder wall.
Examination
Technique
Fig. 19.3 Positioning of radial MPR of the bladder (4 x 1/6) in a patient with exophytic bladder tumor (a) and the resulting oblique section perpendicular to the base of the tumor and bladder wall at the level of the infiltrated distal ureter (b).
a
Contrast Administration
Noncontrast scans should be obtained if intravesical hemorrhage or bladder calcifications are suspected (see Table 19.3). Intravenous contrast is essential for tumor staging, especially the evaluation of the bladder wall and transmural extent, and the differentiation between lymph nodes and veins. Best results for the bladder wall are obtained during the parenchymal phase of enhancement (30-50 s p.i.) if the bladder is filled with negative contrast material (water or urine). High flow rates of4-5mlls lead to enhancement of the bladder mucosa and to an improved contrast uptake in most malignant lesions. Such injection protocols are optimum for tumor staging. Lymph node staging in this phase may be hampered by incomplete filling of the iliac veins. Iliac venous enhancement relies on inflow of contrast-enhanced blood from the lower extremities and requires between 2 and 3 minutes afterinjection. Late scans in the excretory phase usually are not necessary. They lead to contrast -enhancement of the bladder lumen and may improve delineation of wall irregularities and tumors. Iodinated contrast medium is usually denser than urine and therefore layers in the dependent portion of the bladder (Figs. 19.4, 19.5, 19.7c). Horrnogeneous enhancement takes quite some time. For this reason, the patient should be asked to sit and walk around before being positioned on the CT a second time for excretory scans after 30 minutes, or even better, after an 1 hour. Scanning too early will obscure lesions at the interface between contrast -enhanced and
b
non-enhanced urine because the bladder lumen may then appear isoattenuating to the bladder wall. This effect can occur during normal abdominal examinations, e.g., if a biphasic scan of the liver is performed first, and scanning of the bladder is somewhat delayed. If the bladder is an organ of interest, this effect should be avoided, and one should try to scan the whole abdomen in a caudocranial direction (40-50 s delay) starting with the parenchymal phase in the bladder and scanning the liver the portal venous phase. CT Cystography
CT cystography is the preferred technique for evaluation of trauma patients and for selected dedicated bladder CT examinations. Some 300400ml of a diluted solution (1 :10-20) ofaposi-
Fig. 19.4 "Jet effect" produced by opacified tering the bladder from the right ureter.
urine en-
19 Urinary Bladder
tive iodinated contrast medium is instilled in the bladder via a Foley catheter after which the catheter is clamped. Thin-section CT is then performed as described above. Comparative studies have shown that this technique is superior to conventional CT techniques for diagnosing urine leakage following bladder or posterior urethral rupture. Alternatively. air can be used as a negative contrast agent. This technique is best suited to detect small tumor nodules if combined with thin-section scanning and virtual cystoscopy.
Virtual Cystoscopy Virtual cystoscopic images require filling the bladder with a positive or negative contrast agent. either via a Foley catheter or by scanning
Enhancement
Characteristics
Bolus injection of intravenous contrast medium provides information on the perfusion ofpathologic processes. In the parenchymal phase, some 30-50 s p.i., most malignant tumors appear hyperattenuating in relation to the water-dense content of the bladder (urine or instilled water). In the interstitial phase, more than 100s p.i., bladder tumors will appear homogeneously hyperattenuating (apart from areas of hemorrhage) because of increased diffusion of contrast material into the interstitium due to leaky tumor capillaries. In this phase, there may already be excreted contrast material in the bladder. This may obscure smaller lesions in dependent positions. If a scan is to be performed at such a late phase, patients with tumors in the
Congenital
in a delayed phase after intravenous contrast material injection (at least 30 min, better 60 inin p.t.). Thin-section imaging is necessary and virtual endoscopy can be performed by perspective rendering using either shaded surface displays or volume-rendering techniques. An upper threshold or inverted opacity curve has to be chosen to accommodate for the contrast-enhanced bladder lumen (very similar to virtual angioscopy). Air can be used as a negative contrast agent. It has the advantage that there is excellent display of the bladder wall, and injection of intravenous contrast material remains possible. However, there is always some fluid in the bladder which makes scanning in both, the prone and supine position mandatory. With this technique excellent virtual cystoscopy displays can be created (see Fig.2.48a) that may be superior for the detection of small lesions in the bladderwall.
posterior portions of the bladder should be examined in a prone position. Delayed scans taken in the excretory phase after the contrast medium has drained into the bladder (beginning 5 minutes after the end of the contrast injection) show a reversed pattern in which bladder wall lesions and intravesical processes have a lower attenuation than the hyperattenuating bladder content. "Jet effects" may be produced by opacified urine entering the bladder from the ureters (Fig. 19.4). Layering in the bladder lumen is common. Inverted layering may be present with high density urine (insufficient patient hydration) in the most dependent position, followed by contrast material and less concentrated urine anteriorly.
Anomalies
Persistent Urachus The urachus (the canal connecting the bladder with the allantoic duct in fetal life) normally closes after birth and persists as the median umbilical ligament, which stretches between the bladder and umbilicus. A persistent urachus may present as a diverticulum on the bladder
roof, an intra ligamentous cyst, or a tubular structure. Malignant transformation is rare and predominantly affects the inferior intra- or perivesical portion (<1 % of bladder tumors, approximately 95% are adenocarcinomas). This tumor is more common in males (75-80%) and has a very poor prognosis because the lack of symptoms can greatly delay the diagnosis.
Benign Generalized Bladder Wall Thickening
•
CT Morphology
CT demonstrates a cystic or solid structure located on the midline between the bladder roof and umbilicus (Fig. 19.5). Calcifications may be present. A urachal cyst (66%) appears as a cystic mass that is separate from the bladder, while a vesico-urachal diverticulum (33%) forms an outpouching at the anterosuperior pole of the bladder. Diverticula
With a true diverticulum, all layers of the bladder wall protrude through a congenital, circumscribed area of weakness in the bladder wall. This is less common than a pseudodiverticulum, in which only the bladder mucosa prolapses through a congenital muscle defect. Sites of predilection are the ureteral hiatus, the bladder sidewall, and the area of the interureteric crest.
a
Fig. 19.5
•
Benign Generalized
remnant (arrows)
CT Morphology
Bladder diverticula are identified by their CT attenuation values, which are equal to those of the bladder lumen (Fig. 19.6). In some cases adhesions may occlude the neck of the diverticulum and prevent contrast filling. Bladder diverticula may contain calculi or, in rare cases, tumors (5 % of all bladder carcinomas).
c
b
Fig. 19.6 Bladder diverticula. (a) Right-sided periureteral diverticulum. (b) Left-sided bladder diverticulum. (c) 3D volume-rendering of multiple diverticula in
Urachal
patient with bladder carcinoma spiral CT data set (2/4/1.5).
(arrowheads)
from a
Bladder Wall Thickening
Benign, diffuse thickening of the bladder wall may result from a postvesical obstruction (trabeculated bladder), inflammation (see below), early postirradiation edema, or longerterm fibrotic changes (Table 19.3).
•
CT Morphology
CT reveals uniform thickening of the bladder wall «Smm), which shows homogeneous enhancement after intravenous contrast adminis-
689
19 Urinary Bladder Table 19.3
Differential
diagnosis
of bladder calcifications,
shape changes,
Bladder wall calcifications
Focal thickening
Bladder tumor (carcinoma, hemangioma) Schistosomiasis Tuberculosis Echinococciosis Malacoplakia Amyloidosis Cytostatic cystitis Incrusted cystitis (deposition of phosphate crystals due to pharmacologic alkalinization of the urine)
Benign:
Change in bladder shape: benign extravesical causes
of bladder wall
Aftertransurethral resection Focal inflammation Malacoplakia Trauma (hematoma) Papilloma Granuloma Endometriosis
(TUR)
Malignant: Bladder carcinoma, Generalized
Adjacent bowel loops Retroperitoneal fibrosis Pelvic lipomatosis Hematoma Lymphocele Venous collateral vessels Inflammation (diverticulitis,
and wall thickening
sarcoma
thickening
of bladder wall
Benign:
abscess)
Change in bladder shape: malignant extravesical causes
Inadequate distention Trabeculated bladder (postvesical Neurogenic bladder Cystitis • Acute, hemorrhagic • Tuberculosis, schistosomiasis • After radiotherapy • After chemotherapy
obstruction)
Invasion by: • rectal carcinoma • uterine or ovarian carcinoma • tumors of the prostate or seminal vesicles
Fig. 19.7 Benign thickening of the bladderwall. (a) Radiogenic cystitis following radiotherapy for prostatic carcinoma. CT demonstrates concentric bladderwall thickening and stranding of the perivesical fat. (b) Asymmetric thickening of the right lateral bladderwall following TUR. Such changes may persist for up to 6weeks and, as in this example, may be indistinguishable from a neoplasm.
a
b
(ration Postvesical obstruction is usually caused by an enlarged prostate. Due to muscular hypertrophy of the bladder wall, an increased trabeculation of the bladder wall is seen. Hypertrophied folds may mimic tumor appearance. Virtual endoscopy views help differentiate such folds from true masses. Following radiotherapy the bladder volume is generally reduced. Fi-
brotic stranding may be observed in the perivesical fat (Fig.19. 7a). Scans following transuretbral resection (TUR) may show focal areas of bladder wall thickening due to local inflammation and edema that are indistinguishable from tumor (Fig. 19.7b). Consequently, staging examinations should always be performed prior to
TUR.
Bladder Carcinoma Bladder Tumors
CT is unable to differentiate the various layers of the bladder wall, but it is useful for detecting or excluding perivesical tumor extension (t.e., distinguishing T3a from T3b lesions) and evaluating the lymph node status. Thin-section scanning, combined with high flow rates, has the potential to improve the evaluation of bladder wall infiltration for T-staging (Fig. 19.8). Contrast-enhanced MRI is so far superior to CT for staging local tumor extent. Bladder Carcinoma
Approximately 95% of bladder carcinomas are transitional cell (urothelial) carcinomas. The commonest sites of involvement are the floor and lateral wall of the bladder (60% or more), followed by the bladder neck, trigone (25%), and bladder roof (5-10%). Men are affected more often than women, and the peak incidence is between the fifth and eighth decades of life.
Papillary urotlielial carcinoma is an exophytic neoplasm that does not become invasive until a relatively late stage and therefore has a relatively favorable prognosis. Approximately 25% ofurothelial carcinomas are already multifocal when diagnosed. Bladder malignancies include benign papilloma, which is considered a grade I urothelial carcinoma and is multiple in 25% of cases. Solid, urothelial carcinomas are endophytic growths that become invasive at an early stage, resulting in a poor prognosis. Adenocarcinomas are rare tumors «! %) that may arise from the urachal epitheli um or in patients with glandular cystitis (cystic cystitis). Approximately 35% are located on the roof or anterior wall of the bladder, where their detection is frequently delayed. Squamous cell carcinomas « 5%) develop in chronic cystitis and schistosomiasis, and their propensity for early invasion implies a poor prognosis. At least 15% of patients with urothelial carcinomas of the upper urinary tract will eventually develop bladder tumors. •
1 - Epithelium 3 - Muscle
2 - Subepithelial connective 4 - Perivesical fat
tissue
Fig. 19.8 T-staging of bladder carcinoma. T4b tumors infiltrate the pelvic or abdominal wall (from: Sobu LH, et al. LJICC TNM classification of malignant tumors, 6th ed.-New York: J. Wiley & Sons, 2002).
CT Morphology
Even unenhanced scans demonstrate an irregular wall thickening that may form a plaque-like growth on the bladder wall or may project into the bladder lumen as a sessile or pedunculated polypoid mass projecting into the bladder lumen. Calcifications may be seen. Tumors in the trigone region frequently cause hydronephrosis due to obstruction of the ureteral orifice. Virtual cystoscopy may be helpful in detecting plaque-like lesions and in indeterminate cases in which it is difficult to differentiate between real tumor and hypertrophied folds. Sufficient distention and opacification of the bladder lumen is a prerequisite for good results. The CT attenuation of a bladder neoplasm is similar to that of normal bladder wall on noncontrast scans, but is higher than that of normal bladder wall in the parenchyma! phase (>40s) after bolus injection of intravenous contrast material. Intra- and extravesical tumor components are most clearly identified during this phase (Fig. 19.9). Tumors are always hyperattenuating to the non-contrast -filled bladder. Surface evaluation is best accomplished with thin section scanning (e.g., with multislice CT)
691
19 Urinary Bladder
b
a
c
Fig.19.9 Bladder carcinoma T1-T3a. (a) T1 tumor (arrows) spreading along the bladder surface with no signs of transmural growth (coronal multislice CT image), (b) Multifocal bladder tumor with infiltration into the outer layer of the bladder wall (T3a; arrows) without infiltration of the perivesicular tissue. Note the
dilated submucosal veins (arrowheads) that simulate further tumor manifestations (coronal volume-rendered multislice CT image), (c) The differentiation between the dilated submucosal veins (arrowheads) and tumor (arrows) is facilitated on semitransparent volume-rendered images (sagittal view).
through a well-distended bladder either in the parenchymal phase while there is still no excretion of contrast material into the bladder, or during the delayed phase when the bladder is completely opacified. If the tumor has infiltrated the bladder wall (T2 or higher), bladder distention will lead to marked retraction of the less compliant tumor-affected wall areas (Fig.19.9b). CT usually cannot distinguish among stages Tl, T2, and T3a of bladder carcinoma. This prognostically important classification is established by transurethral resection (TUR) (Table 19.4).
The type of bladder wall invasion dictates the therapeutic approach: superficial tumors (TaTl) are removed by TUR, while infiltrative tumors (T2 or higher) are managed by cystectomy. TUR is followed by inflammatoryedematous bladder wall thickening that may be detectable with CT for up to 6 weeks and should not be mistaken for residual tumor. CT can differentiate a stage T3a tumor from stage T3b with an accuracy of over 80%. Multislice CT with coronal and sagittal reformations will probably be even better. The main criteria for diagnosing stage T3b lesions (perivesical invasion) are visible tumor projection beyond the unaffected bladder wall contour and fine stranding of the perivesical fat (Fig.19.9c). However, it is difficult (both for CT and MRI) to determine whether perivesical fat stranding is due to neoplastic infiltration or a reactive inflammatory reaction. Obliteration of the fat planes separating the bladder from other organs is suspicious for T4 disease. T2 and T3 tumors (wall invasion) lead to retraction effects in the fully distended bladder (Fig.19.9b). Stage 4 lesions may invade the prostate, seminal vesicles, and internal obturator muscle (Fig.19.1 Ob). The uterus and cervix are rarely involved (Fig.19.1 Oa). With infiltration of both the bladder and prostate, it can be difficult to establish the origin of the tumor. The overall accuracy of CT in detecting perivesical
Table 19.4
Staging criteria for bladder carcinoma
Ta
Confined to the mucosa
T1
Confined to the lamina propria
T2
Invasion of the superficial muscle layer
T3a
Invasion of the deep muscle layer
T3b
Transmural invasion of the perivesical fat (focal Iymphangiosis)
T4a
Extension past the floor and roof of the bladder
T 4b
Invasion of the abdomen and pelvic wall
N1
Solitary lymph nodes "s 2 cm
N2
Solitary lymph nodes 2 - 5 ern. multiple lymph nodes A 2cm
N3
Lymph nodes >5 cm
M1
Distant metastases
Mesenchymal Tumors
a
c
b
Fig. 19.10 Bladder carcinoma T3b-T4. (a) Multifocal polypoid tumor with initial tumor extension through the bladder wall (stage T3b). (b) Marked retraction effects associated with tumor extension through the bladder
wall (stage T3b). (c) Stage T4 bladder carcinoma has spread beyond the bladder and invaded the uterus. A large metastasis is visible in the obturator lymph nodes on the leftside (arrows), and both ureters are obstructed.
and seminal vesicular involvement is in the range of 65-85%, the accuracy in detecting lymph node metastases ranges from 70 to 90% with a false negative rate of 25-40%. Lymph nodes larger than 1.0 cm in short axis diameter are considered suspicious for metastasis, but on CT scans this finding cannot be distinguished from a comparable degree of reactive inflammatory lymph node enlargement. Multislice CT has the potential of using lymph node size and enhancement to distinguish between tumor-bearing and inflammatory nodes. Round, moderately enhancing lymph nodes are considered suspicious even below the 1.0cm threshold (Fig.l9.1 Oc). However, no cutoff values between benign and malignant disease have been established yet. Initial lymphatic drainage is to the ipsilateral lymph nodes of the obturator group and the middle external iliac group, followed by the nodes of the internal iliac group, the common iliac group, and the presacral nodes. Distant hematogenous metastases occur in the liver, lung, and bone.
generally small. Hematuria malignancy.
Mesenchymal Tumors
Mesenchymal tumors of the bladder (fibroma, neurofibroma, lipoma, hemangioma, adenoma, leiomyoma, rhabdomyoma, pheochromocytoma, and their sarcomarous forms) are very rare tumors that may produce an obstructive effect, depending on their location. Benign tumors are
•
is suspicious
for
CT Morphology
Mesenchymal tumors have a nonspecific CT morphology with two exceptions. Neurofibromas display low precontrast attenuation values (20-30 HU) and enhance after contrast administration. Asymmetrically enlarged sacral foramina suggest the correct diagnosis. Lipomas can be identified by their negative attenuation values. Two-thirds of /el'omjlomas are extravesical and cause displacement of the bladder wall. Fifteen percent of all pheochromocytomas are extra-adrenal, and 1-2% of these tumors occur in the bladder wall, usually the roof or trigone. Approximately 80% are hypervascular. Nephrogenic adenomas are rare benign tumors that typically develop as a sequel to trauma or inflammatory disease (latent period 1-6 years) and appear on CT as a circumscribed wall thickening or mass. Hemangiomas frequently display calcifications within a well-circumscribed mass. Rhabdomyosarcoma is a pediatric tumor that accounts for 10% of all soft-tissue tumors in children. It is characterized by early and very aggressive infiltrative growth and causes diffuse thickening of the bladder wall. Benign and malignant forms of mesenchymal tumors cannot be confidently distinguished by their CT features. But it is helpful to note that benign tumors tend to form a smooth- filling de-
693
19 Urinary Bladder
feet on the bladder wall while malignant tumors typically appear as an irregularly lobulated and sometimes ulcerative mass. Signs of invasion and metastasis are definite criteria of malignancy.
Inflammatory
Bladder
Diseases
Cystitis Cystitis is diagnosed by clinical and laboratory findings and is not an indication for CT scanning. o
When extensive pelvic tumors are detected. it is often no longer possible to determine the site of tumor origin (bladder. prostate. spermatic cord. testicular sheaths. uterus. vagina. pelvic muscles).
CT Morphology
Simple acute cystitis may produce no CT abnormalities. Diffuse. smooth (edematous) wall thickening may be observed. Following contrast administration. the thickened. hypervascular inflamed mucosa may be demarcated as a brightly enhancing stripe. Scbistosomiasis causes irregular polypoid thickening of the bladder wall (>Smm) and asymmetric contraction of the anterior wall. Calcifications are typical and suggest the correct diagnosis. The incidence of neoplastic disease is increased following schistosomiasis. Tuberculosis also causes irregular bladder shrinkage. and calcifications are a common finding. Tuberculous bladder abnormalities are always accompanied by changes in the kidneys and ureters. Radiogenic cystilis is caused by irradiation to the bladder. and chemically induced cystitis may follow the local instillation of chemotherapeutic drugs or may follow systemic cyclophosphamide therapy due to breakdown products that damage the bladder mucosa. These forms of cystitis are characterized by irregular thickening of the bladder wall and uniform bladder shrinkage. Cystitis cystica (cystitis glandularis. or follicularis) is a rare proliferative form that develops in response to chronic irritation. It is character-
ized by thin-walled cysts of water density of 1-10 em diameter. Cystitis cystica is considered a premalignant lesion and is associated with an increased incidence of adenocarcinoma. A rare inflammatory disorder is necrotizing granulomatous cystitis, which occurs in the setting of Wegener's disease and causes nonspecific thickening of the bladder wall.
Malacoplakia Malacoplakia is a rare, postinfectious, matory granulomatous wall thickening responsive to antibiotic therapy.
inflamthat is
CT Morphology
There is a circumscribed nodular or plaque-like area of wall thickening that may contain central necrosis and calcifications. It is indistinguishable from a bladder malignancy on CT scans.
Enterovesical Fistulae Enterovesical fistulae can occur in the setting of Crohn's disease, sigmoid diverticulitis, or colonic malignancy. o
CT Morphology
Enterovesical fistulae are manifested by gas in the bladder lumen. !ffocal bladder wall thickening is present, the chronic inflammatory cause is morphologically indistinguishable from an invasive tumor.
Trauma
Extraperitoneal bladder ruptures are more Approximately 10% of patients with pelvic trauma sustain a bladder injury. The risk of in- frequent (60-75% of major bladder injury) and jury is greater in a distended bladder than an associated with penetrating trauma or fractures empty bladder. CT cystography (usually done as of the pelvis, usually the pubic rami or symphysis. part of a general abdominal CT examination) is It is caused by a penetrating bone fragment or by superior to conventional cystography for detect- other mechanical effects relating to trauma-like ing small bladder extravasations, including ret- shearing forces or stress on the puboprostatic ligaments. The site of rupture can be far removed rovesical collections. It also yields information from the site of the bony fracture. In 5% of major on trauma-associated changes in the extravesibladder injuries a combination of both intra- and cal soft tissues. Symptoms of gross hematuria, pelvic fluid and specific pelvic fractures are extraperitoneal ruptures can be found. highly correlated with bladder rupture and may • CT Morphology guide patient selection. The most important traumatic lesions are Optimum CT evaluation of the bladder requires bladder contusion, intraperitoneal bladder rup- contrast instillation through a Foley catheter. Inture and extraperitoneal bladder rupture. A con- juries to the upper urinary tract should be excluded by conventional urography or CT tusion of the bladder wall is an incomplete tear of the mucosa that produces a localized (delayed scans in the excretory phase acquired > 10 min after IV contrast administration, best intramural hematoma. It is the commonest results with m ul tislice CT). form of bladder injury and is treated conservatively. Bladder contusions show a focal area of hyperattenuation in the bladder wall (Fig.19.11 b) An intraperitoneal bladder rupture accounts for 20-35% of major bladder injuries. It may be and are best visualized on non-contrast or early scans with a full bladder. Intravesical hemorcaused by a sudden pressure rise within the full bladder caused by a forceful external blow rhage presents as a hyperattenuating mass on (often after a long automobile ride). The site of early scans and a filling defect on delayed scans (Fig. 19.11a). In intraperitoneal ruptures contrast predilection is the bladder roof. It is a typical medium is seen in the paracolic gutters and outseat-belt or steering wheel injury, that requires lining small bowel loops or other abdominal orprompt surgical intervention.
a
b
Fig. 19.11 (a) Traumatic bladder tamponade with sharply circumscribed filling defects (blood clots). Small air collections surround the inflated tip of the Foley catheter, (b) Sagittal reformation (4 x 1/6) demonstrating a traumatic bladder injury with intramural (arrow-
c heads) and supravesicular hematoma (arrow) that was hard to detect on axial sections, (c) Impression of the contrast-filled bladder ba large abscesses secondary to pelvic trauma.
19 Urinary Bladder
gans. Extraperitoneai ruptures can be simple or complex. In simple ruptures. extravasation is limited to the pelvic perivesical spaces. In complex ruptures the pelvic fascial boundaries have ruptured and contrast extends into the thigh. scrotum. penis. and perineum. Extravasated
Postoperative
Changes
Cystectomy and Urinary Diversion CT can be performed after radical cystectomy with lymphadenectomy (usually combined with prostatectomy/hysterectomy or with resection of the seminal vesicles/ovaries) to detect or exclude postoperative complications such as hematomas. abscesses, and urinomas (Table 19.5). In a classic urinary conduit procedure, the detached ureters are implanted into a suitable isolated bowel segment (ileum, transverse or descending colon) and the urine is diverted through a percutaneous stoma into an external receptacle. A more modern alternative is the continent urinary diversion, in which part of the bowel is used to construct a reservoir pouch for urine. The various procedures differ in the selecTable 19.5 Diagnostic procedures various surgical complications Complications
procedures
recommended
for
Diagnostic
Postoperative Anastomotic Infection,
leak (urinary
leak)
Cystography,
Hemorrhage
CT CT
formation
Ureteral
CT
CT
abscess
Ischemia Fistula
Cystography,CT
obstruction
Cystography,
complications
Reflux Diurnal
Cystography or nocturnal
Infection Ureteral
in conti-
Cystography Laboratory tests, CT
stricture
Calculi
IVP,CT Cystography,
IVP, CT Metabolic
acidosis
Laboratory
tion of the bowel segment used to construct the bladder substitute, the nature of the anti-reflux and continence-preserving mechanisms, and the mode of urine elimination. In all these procedures, small bowel and/or colon is used to construct a "neobladder" into which the ureters are reimplanted. The procedures are subject to numerous modifications that influence the CT appearance of the reconstruction and should be interpreted in consultation with the urologic surgeon. The presence of a functioning external urethra! sphincter permits the construction of an "orthotopic neobladder" that occupies a normal intrapelvic position. Otherwise the reservoir may be placed in the right lower abdomen and fitted with a cutaneous stoma (usually an umbilical stoma) to allow intermittent catheterization. Immediate postoperative complications are distinguished from late complications. (Voiding) Cystography after contrast instillation, and urodynamic studies are the methods of choice for diagnosing ureteral reflux, incontinence, or poor pouch filling due to excessive pressure within the reservoir. Calculi may develop on metal clips that are exposed to urine. CT is used to detect extravesical changes such as abscesses, fluid collections, and enterovesical orvesicocutaneous fistulae (Table 19.6). CT Morphology
IVP, CT Late
urine in the prevesical space may extend cephalad in the abdominal wall or surround the peritoneal cavity, thereby mimicking intraperitoneal ruptures. Extravesical hemorrhage appears as a hyperattenuating mass that may produce bladder impression or displacement.
tests
The reservoir pouch appears as a fluid-filled structure that may be located in the right lower abdomen or may occupy an orthotopic position in the pelvis, depending on the reconstructive technique used (Fig. 19.12). The bowel should be adequately opacified with oral contrast so that the pouch can be distinguished from functioning bowel. A precontrast examination should be performed to detect hyperattenuating fluid collections and calculi. Delayed scans (» IOmin after
Recurrent Tumor Table 19.6 Postcystectomy cedures (selection
urinary
diversion
697
pro-
Orthotopic bladder replacement (urinary diversion through the urethra) Carney bladder (historical technique) U-shaped ileal loop with each ureter implanted into one end of the loop; used only in males. Ileal neobladder (Hautmann procedure) Reservoir constructed ofdetubularized* terminal ileum Studer bladder Reservoir constructed ofdetubularized* ileum with the (shortened) ureters implanted into a cranially transposed ileal loop
a
Hemi-Kock pouch Reservoir constructed ofdetubularized* ileum, using intussusception (ileal nipple) as an anti-reflux mechanism Orthotopic Mainz I pouch Reservoir constructed of cecum and terminal ileum (ileocecal pouch) Heterotopic bladder replacement (urinary diversion through a stoma) Mainz I pouch Ileocecal pouch, usually with an appendiceal umbilical stoma Kock pouch Ileal pouch with a continence and anti-reflux nipple, often combined with an umbilical stoma Indiana pouch Ileocecal pouch
b Fig. 19.12 Kock pouch in the right lower abdomen with an umbilical stoma (a). Ileum conduit (4x1/6 VRT). The arrows mark the anastomoses ofthe ureters to the conduit. Note the contrast remnants in colonic diverticula (arrowheads) following biliary contrast excretion after an CT urogram (b).
Ureteral implantation into the bowel (urinary diversion through the rectum) (historical technique) Implantation of the ureters into the sigmoid colon
Coffey operation
Mainz II pouch Implantation of the ureters into a reservoir constructed of sigmoid colon and rectum * Detubularization = incising the bowel longitudinally and suturing it to create a reservoir
IV contrast administration) should also be obtained to evaluate the supravesical urinary tract, including the sites of ureteral reimplantation in the pouch. If a fistula or urinary leak is suspected, the pouch should be scanned after percutaneous instillation of contrast material. Intravesical gas is a normal finding after percutaneous catheterization and should not be
mistaken for an abscess. However, extravesical gas collections, like fluid collections, are always abnormal (unless scanning is performed in the immediate postoperative period) and may signify hemorrhage (> 2OHU), urinary leak «20HU, contrast extravasation), or infection with abscess formation (20-40 HU). Recurrent Tumor
MRI is superior to CT for differentiating recurrent tumor and postoperative fibrosis, but it is less effective in differentiating acute radiationinduced changes, tumor, inflammation, and infection. A CT examination to detect tumor recurrence should cover the abdomen and pelvis including the perineum.
19 Urinary Bladder
CT Morphology
Adequate bowel opacification must be obtained in the pelvis to ensure that coils of small intestine are not mistaken for recurrent tumor. A new (enhancing) soft-tissue mass that is detected in the bladder bed more than 3 months after surgery is suspicious for a recurrent tumor (Fig. 19.13). Some 70% of cases with local recurrences also have enlarged pelvic lymph nodes. On the other hand, more than 50% of postoperative patients are found to have abnormally enlarged pelvic lymph nodes that are not associated with a local recurrence. Consistent with the lymphatic pathway that drains the bladder, enlarged para-aortic retroperitoneal nodes are generally associated with enlarged pelvic nodes. There have been reports of local tumor recurrence in the pelvic floor involving only the perineum or vagina and appearing on CT as an isolated thickening of the anterior rec-
Fig. 19.13 carcinoma.
Urethral recurrence
(arrows)
of a urothelial
tal wall or anterior vaginal wall. As with a recurrence of rectal carcinoma, a recurrent bladder carcinoma may directly infiltrate the bony pelvic ring.
20 Female Pelvis K. J. Lehmann, A.
J. van der Molen
Anatomy Examination Technique Anomalies ofthe Uterus and Vagina Predominantly
Cystic Masses
NabothianCysts(p.706) Functional Ovarian Cysts (p. 706) Theca Lutein Cysts (p. 706) Polycystic Ovary Syndrome (Stein-Leventhal syndrome) (p. 706) Paraovarian Cyst (p. 706) Endometriosis (p. 706) Dermoid Cyst (p. 707) Cystadenomas (p. 708) Hydrosalpinx (p. 708) Ovarian Carcinoma (p. 708) Fallopian Tube Carcinoma (p. 711) Benign-Malignant Differentiation in Cystic Lesions (p. 711) Solid Tumors Uterine Leiomyoma (p. 712) Fibroma and Thecoma (p. 713) Brenner Tumor (p. 713) Ovarian Torsion (p. 713) Cervical Carcinoma (p. 714) Endometrial Carcinoma (p. 716) Recurrent Malignancies of the Uterus (p. 718) Gestational Trophoblastic Disease (p. 719) Other Primary Malignant Tumors (p. 720) Metastases (p. 720) Infectious and Vascular Diseases Pelvic Inflammatory Disease (p. 721) Abscess(p.721) Endomyometritis and Pyometra (p. 721) Fournier's Gangrene (p. 722) Ovarian Vein Thrombosis (p. 722) Pelvic Varices (Pelvic Venous Congestion Syndrome) (p. 723) CT in Pregnancy
20 Female Pelvis
organs and should be used for assessment of pathologic structures. The first imaging modality that is used after the physical examination of the pelvis is ultrasound, preferably using an transvaginal probe. In diseases of the female pelvic organs, such as the staging of uterine tumors, MRI is increasingly being used as the first-line modality, although multislice CT techniques may yield similarly good results in patients with more advanced diseases. CT scanning of the female pelvis is usually performed for reasons other than primary gynecologic indications (Table 20.1).
CT is not among the first-line procedures that are used in the diagnosis of female pelvic diseases. Unlike ultrasound and MRl, CTcannot define the structures of the uterine wall, particularly the endometrium and its changes during the menstrual cycle. On the other hand, the pelvis is often included as part of an abdominal CT study, and therefore knowledge of normal and pathologic appearances of the pelvic structures is essential. Similar to MRl, the multiplanar imaging capabilities of spiral or multislice CT (sagittal and coronal views) are most helpful for delineation of the interfaces between the pelvic
Table 20.1 Tumor
Indications
for CT examination
diagnosis
Planning
follow-up
For advanced Optional: carcinoma Suspicion
Suspected
In patients
abscess
Interventional
Abscess
Other a
CT can be used
pelvis FICO stage FIGO stage FIGO stage
carcinoma Cervical carcinoma"" Endometrial carcinoma'
of radiotherapy
Oncologic
of the female
Ovarian
indications,
endometrial,
vaginal,
or vulval
carcinoma'
postoperative baseline examination (ovarian carcinoma, 5T2, endometrical carcinoma =:*T3) of recurrent
or cervical
tumor
with equivocal
ultrasound
findings
drainage
Equivocal for these
stages of cervical.
II or higher II or higher or hgher
III
ultrasound
findinqs '
but MRI is preferred
Anatomy The uterus is covered by peritoneum, which anteriorly continues into the peritoneal covering of the bladder roof. Posteriorly the peritoneum descends to the dorsal vaginal fornix and then along the anterior wall of the rectum, forming the posterior cul-de-sac (pouch of Douglas) . The vagina appears on CT scans as a transverse oval structure of soft-tissue density located anterior to the rectum and medial to the levator sling (Fig. 20.1). Unless a vaginal tampon has been inserted, the vaginal lumen, and the anterior and posterior fornices, usually cannot be visualized with CT, nor can it be demarcated from the cervix. However, after rapid injection of contrast material the vaginal mucosa often enhances and can be better appreciated if no tampon is present.
The cervix displays uniform soft-tissue density, is sharply demarcated from the surrounding fat, and superiorly continues into the uterine body (Fig. 20.1). The shape, size, and position of the uterus are variable and are influenced by many factors including age, hormonal status, pregnancy, the presence ofleiomyomas, and the degree of bladder distension. Viewed in axial section, the cervix of a woman of childbearing age normally measures no more than 3 em in diameter, and the uterine body measures no more than 5 cm. The uterine cavity is delineated as an area of low attenuation only if it contains fluid (e.g., in the perimenstrual period). Although the cervix can be hard to distinguish from adjacent portions of the uterus and vagina (even if a tampon is inserted), they nevertheless can be sepa-
Anatomy Fig. 20.1 Cross-sectional anatomy of the uterus, ovaries, parametria, and vagina.
~-
Common femoral artery and vein
Vagina
Sartorius muscle
\
)
Sciatic Inerve Obturator
Urinary bladder
---.l
artery and vein
Interna pudendal artery and vein
Inferior epigastric artery and vein Broad lliqarnent External iiliac artery .and
Ureter
Levator ani muscle
Peritoneum Uterine round ligament Uterine tube
Ovary Peritoneum
rated from one another by their configuration: the uterine body usually is somewhat triangular in shape, whereas the cervix has a more rounded appearance. The broad ligament is formed by two layers of peritoneum, which drape the uterus and extend laterally from the uterus to the pelvic sidewall. The superior free edge is formed by the fallopian tube medially and the suspensory ligament of the ovary laterally. The lower edge of the broad ligament ends at the cardinal ligament (Fig. 20.2).
The parametrium is a broad sheet of loose extraperitoneal connective tissue, smooth muscle, and fat that extends laterally from the uterus to the pelvic wall between the layers of the broad ligament. The parametrium contains the fallopian tubes, and the ovarian and cardinal ligaments. In addition, it contains uterine and ovarian vessels, nerves, and lymphatic vessels and a portion of the ureter that runs 2 cm lateral to the cervix through the parametrium. Although the broad ligament is rarely seen unless
701
20 Female Pelvis
~ ~
Uterine round ligament Suspensory of ovary
ligament
Fig. 20.2 Schematic diagram of the parametria, viewed from the posterior aspect.
Uterine tube Ovary Broad ligament Ureter Cardinal
ligament
Uterosacral
ascites is present, its position can be determined by structures that it contains or abuts. The only structures that can be consistently visualized with CT in this region are the principal structures of the uterine ligaments. The uterine round ligament is a band of fibromuscular tissue that passes anterolaterally forward and downward from the lateral angle of the uterus to the internal inguinal ring. It appears on CT as a band of soft-tissue density that gradually tapers from its broad base at the uterus (Fig. 20.1). The cardinal ligament (lateral or transverse cervical ligament, Mackenrodt ligament) passes laterally from the cervix and upper vagina to the fascia overlying the obturator internus muscle (Fig. 20.2). It forms the base of the broad ligament but cannot always be defined over its entire length in CT. The uterine artery runs along its superior aspect. The cardinal ligaments represent the border between the parametrium and the paravaginal tissue [paracolpos). There is a broad variation in the anatomic configuration of the cardinal ligament with respect to thickness, shape and contour. The uterosacral ligament extends laterally and posteriorly from the cervix to the sacral bone and encloses the rectum. The vesicouteral ligament extends laterally and anteriorly and encloses the dorsal wall of the bladder. It is
ligament
common for the uterine ligaments to appear thickened following a course of radiotherapy. The ovaries cannot always be identified as discrete structures on conventional CT scans but they are almost always seen with multislice scanning. They are located anterior to the ureters and posterolateral to the uterus in the ovarian fossa. The origin of the Fallopian tube and the ovarian ligament (between the uterus and the ovary) is usually visible at the lateral angle of the uterus and provides an additional anatomic landmark for identifying the ovary. The ovaries may be of uniform soft tissue density. More commonly, however, the ovaries contain soft tissue stroma admixed with small cystic areas representing normal follicles. The size of the follicles varies with the menstrual cycle. Just before ovulation they may be as large as 2.5 cm. The uterine vessels pass from the internal iliac artery and vein to the inferior border of the broad ligament and cross in front of the ureter before forming the uterovaginal vascular plexus, which can be identified after IV contrast administration. The paired ovarian arteries (which arise directly from the aorta) can be visualized only on arterial phase multislice CT scans, but the ovarian vein can be identified with most techniques on the psoas muscle lateral to the ureter, especially during the postpartum period or if there is varicose dilatation.
Examination
Technique
Patient Preparation
Acquisition
Adequate oral contrast opacification is essential to avoid mistaking nonopacified bowel loops for pelvic masses or fluid collections. The fluid absorbed from oral administration of 10001500ml of an iodine- or barium-containing solution also distends the bladder. displacing the coiled loops of small intestine upward and shifting the uterus to a more upright position. It is important to have the patient drink the oral contrast agent continuously. spread over a period of at least 45 minutes. If scanned too early the distal small bowel loops will often not yet be opacified. while drinking the contrast agent all at once may result in non-opacification of proximal loops. Full bowel distension usually is not necessary because a small amount ofintraluminal contrast already identifies bowel structure. With multislice CT scanning. negative contrast agents such as water or methyl cellulose preparations may be advantageous for better display of the vasculature but this risks failing to distend distal loops. which makes it harder to distinguish them from pelvic masses. With thinsection multislice scanning. however. bowel loops can be followed and thus be distinguished from masses, even if distension is suboptimum. For optimum display of the pelvic structures, a moderate distension of the bladder is ideal. Patients should therefore be asked to urinate about 20-30 minutes before the scan. A moderate amount of rectal contrast (approx. 100ml of 5-10% diluted CM) is advantageous. However, an overdistended rectum leads to compression of the pararectal fat and can make the scans difficult to interpret. Peritoneal instillation of 1000-2000 ml of diluted (1: 20) sterile nonionic contrast agents is not a standard procedure but may be used to optimize staging and diagnosis of recurrence of ovarian cancer. Prior to intraperitoneal chemotherapy such an approach may help determine which portions of the peritoneal cavity are reached by the chemotherapeutic agent when applied via a intraperitoneal catheter.
Thin slices should be used for selective helical CT examinations of the uterus and adnexa (Table20.2). although a 4x2.5 mm collimation will usually suffice for 4-slice scanners. For optimum evaluation of local tumor invasion, however, multislice scanning with 0.75-1.25 mm sections (4- to 16-slice scanners) is advisable to obtain a secondary raw data set from which thicker axial sections (3-7 mm, to improve signal-to-noise ratio) and coronal sections of 2-4 mm thickness as well as sagittal sections of 3-6 mm can be reformatted. The variations in section thickness are required because of the special noise characteristics in the pelvis. Image noise in the pelvis is anisotropic because absorption varies with the rotational angle of the x-ray tube; it is higher for lateral projections than for sagittal projections. The appearance of noise on the scans, therefore, will not be punctuate but instead slightly needle-shaped, oriented in a horizontal direction. Software filters that average raw data sets from high-absorption views (either in-plane or along the z-axis also) can improve image quality substantially and may even allow reduction in dose. They are available for spiral CT and are currently being developed for multislice CT. Another technique is adaptive dose modulation in which the mA settings of the x-ray tube are continuously varied and adapted to the higher requirements in the lateral projections and the lower requirements in AP directions. This technique hold the potential for up to 30% dose reduction at comparable image quality. With primary staging of known malignancies, however, dose is only a minor issue. For this indication, radiation exposure should be chosen high enough for optimum image quality. Especially with obese patients this requires choosing 140 kVp for better penetration, thicker sections (e.g., 4 x 2.5-4 x 3.75 mm for 4-slice scanners), and the highest possible mAs setting. Low pitch factors for 4-slice scanning (P" = 3) make it possible to use higher mAs settings in very obese patients. For spiral CT or dual-detector CT. a 360' LI interpolation with a pitch of 1 (P*=2 for dual detectors) substantially improves signal-to-noise ratio (equivalent to increasing mAs settings by a factor of two) but at the expense of a 30% wide section profile.
Technique
704
20 Female Pelvis Table 20.2
Suggested protocols for CT scanning of the female pelvis
General Oral contrast Rectal contrast
1000-1500 ml positive contrast, administered
Vaginal contrast
100 ml (optional) Insertion of vaginal tampon (optional)
Patient position
Supine with elevated arms Pelvis from iliac crest to below ischiac bone
Scan range Respiratory
60-90 min before examination
phase
Windowing Scan parameters
Inspiration (shallow respirations is also possible) Contrast-enhanced
SC
t
2-3
5/10/5
Small pelvis
3/5/21
Contrast injection
l-slice
Small pelvis Infectous disease Tumor staging d
W/L~400/70
4-slicea
l-slice SC/TF/RI
Routine (whole abdomen)
Routine (whole abdomen)
CT:
i
1-1.251
16-slice" SC 1-1.5
t
0.75-1.251
axial SW/RI
MPRb SW/RI
5/4
4/3
3/3 cor 4/4 sag
V/F/D
4-slice V+N/F/D
16-slice V+N/F/DX'
120/2/60
120 + 50/3/25
120 + 50/3/155
120/2/80 120/4/40
120 + 50/2/80 120 + 50/5/40
120 + 50/2/90 120 + 50/5/20A
SC= slice collimation (mm), TF
For all indications that involve benign disease (e.g., abscesses), radiation dose should be kept as low as possible. In the small pelvis, the choice of 140 kVp tube voltage is also advisable, and automated dose control techniques should be used, if available, to spare slim patients from excessive radiation exposure.
Contrast Injection
a examination
of the pelvic organs is generally performed after intravenous contrast injection. Unenhanced scans are useful only for the detec-
tion of acute hemorrhage. For a spiral or multislice a staging examination that includes the upper abdomen, scanning should proceed in the caudocranial direction because the best parenchymal opacification of pelvic organs and parametrial arteries occurs at an earlier time (3040 s p.i.) than the portal venous phase ofthe liver (50-90 s p.i.). Pelvic veins, however, are opacified later, usually after more than 90 s. Delayed imaging after 3-5 minutes can be beneficial for evaluating stromal invasion, abscesses, and infiltration of the bladder wall or distal ureters.
Predominantly Cystic Masses
(Anomalies of the Uterus and Vagina
Uterine anomalies (incidence of 2-3% with an increased prevalence in women with infertility or high rate of abortions) such as an arcuate, septate, bicornuate, or didelphic uterus can sometimes be detected (Fig. 20.3) but generally cannot be sufficiently evaluated with CT. Highresolution MRI is effective in making the important preoperative differentiation between a septate and bicornuate uterus. Ultrasound is used for diagnosing reproductive tract anomalies in children and adolescents, and MRI is added only for evaluating complex forms of gonadal intersexuality. CT is not indicated. Fig. 20.3
Bicornuate uterus.
Predominantly Cystic Mass<
In the pelvis a variety of cystic masses can be present (Table 20.3). Most cystic tumors origiTable 20.3
nate from the ovaries. Differentiation of the various conditions can be difficult using CT.
Differential diagnosis of cystic ovarian masses
Benign lesions Nabothian cyst
Localized in the cervix
Retention cyst (follicle, corpus luteum)
Thin walls, smooth margins, serous contents «20HU),
Ovarian cyst
Near-water septations
size"
density «20 HU), homogeneous, sharp margins, no
Polycystic ovaries (Stein-Leventhal)
Microcystic enlargement of both ovaries, signs of viriliz.atior=H
Endometrial cyst (chocolate cyst)
Smooth margins, intracystic hemorrhage reach considerable size
Dermoid cyst (mature teratoma)
Often contains fat «-10
Benign cystadenoma serous mucinous
(hyperattenuating),
may
HU), may also contain bone or trtl:]
Thin walls, smooth margins, serous contents «20 HU), size oftsin >6cm U Indistinguishable from cystadenocarcinoma '''''1
Hydrosalpinx
Tortuous tubular structure of fluid attenuation, wall
Tubo-ovarian abscess
Fever, marked rim enhancement
slightly enhancing
Malignant lesior Ovarian carcinoma
Criteria for malignancy (Table 20.4)
Tumors of ovarian stroma
Hormone-producing, criteria for malignancy (Table 20.4) (granulosa cell tumor, theca cell tumor, etc.)
Dysgerminoma
Young women, early lymph node metastases
Malignant teratoma
Solid tumor containing elements of all three germ tissue, fat, bone), extension to adjacent organs
Krukenberg's tumor
Metastatic, usually from cancer of the gastrointestinal trao: (e.g., gastric carcinoma)
705
20 Female Pelvis
Nabothian Cysts
These cysts results from mucinous distension of endocervical glands. They are relatively common and may reach 2-4 cm in size. No therapy is indicated. •
CT Morphology
Well-defined cysts in the cervix, usually small. There is no enhancement of the cyst wall. Functional Ovarian Cysts
Functional cysts are very common in reproductive women and include follicular cysts, corpus luteum cysts, and corpus albicans cysts. Hemorrhage in a cyst is usually painful. •
CT Morphology
These are typically well circumscribed, thinwalled and have a low attenuation. Due to hemorrhage, the attenuation may be higher (up to 50 H) in corpus luteum or hemorrhagic cysts. Thicker, well-vascularizecl rims are also suggestive of corpora lutea. Differentiation of hemorrhagic cysts and endometriomas is generally not possible. Papillary projections or a solid nodule should raise suspicion of a malignant cystic neoplasm. Theca Lutein Cysts
These cysts typically develop in patients with high levels of p-hCC, such as in gestational troyyphoblastic disease or ovarian hyperstimulation syndrome (in vitro fertilization patients). Cysts can be present up to 4 months after evacuating a mole pregnancy. •
Fig. 20.4 Polycystic ovaries in Stein-Leventhal syndrome. Because the patient underwent previous partial removal of the right adnexa, the typical microcystic pattern of ovarian enlargement is seen only on the left side. The patient presented with marked clinical signs of vir ilization.
CT Morphology
The ovaries are enlarged and edematous. Large, septated cysts are present bilaterally. There are no mural nodules or septal thickening. POlycystic Ovary Syndrome (Stein-Leventhal syndrome)
Polycystic ovaries (Fig. 20.4) may be seen in association with Stein-Leventhal syndrome, a combination of hirsutism, amenorrhea, and in-
fertility. There is an imbalance of the normal cyclical changes of FSH and LH leading to chronically stimulated but unruptured follicles and hypertrophy of the ovarian stroma. Obesity is frequently present. •
CT Morphology
Multiple cysts of uniform size are peripherally located in enlarged ovaries. The central stroma and the cortex of the ovary is prominent and there is often an associated hypoplastic uterus. The disease is always bilateral. Occasionally, the cysts are too small to be discernible by CT and the ovaries appear as two enlarged soft tissue masses. Paraovarian Cyst
All types of ovarian cysts can also occur in the broad ligament (parovarium). These account for more than \0% of adnexal masses. CT morphology is not different from ovarian cysts. Endometriosis
In endometriosis, there is functional endometrial tissue present outside the uterine cavity. It is a common disease in women of childbearing age, occurring in 15% of the general female population. It is clinically characterized
Dermoid Cyst
by the triad of dysmenorrhea, dyspareunia, and infertility. It is most frequently found in the ovaries, the cul-de-sac, or the broad and uterosacral ligaments. Atypical implantation sites may be found rarely in the GI tract, urinary tract, chest, or in cutaneous scars. Abdominal adhesions are a common complication. Endometriotic cysts (endometriomas) are the commonest manifestation. These can become as large as 15cm, frequently show intracystic bleeding ("chocolate" cysts) and adhere to adjacent structures. Malignant transformation (endometrioid carcinoma) is rare. While endometrial cysts can be clearly visualized, imaging studies generally will not detect the small nodular foci only a few millimeters in diameter, which may occur throughout the lesser pelvis. If there is suspicion for endometriosis, radiologic diagnosis must rely on transvaginal US or preferably (fat -suppressed) MRI. The gold standard is laparoscopy, which includes the sampling of biopsy material. There is no indication for CT scanning. •
CT Morphology
Endometriomas may appear as a uni- or multilocular cystic mass. The cysts may be anything from hyperattenuating to near-water attenuation, and may even present as a mixture of areas with varying attenuation (Fig. 20.5). There may be a thick, enhancing fibrous capsule around the endometrioma. Multiplicity and distorted shape can help in differentiating an endometrioma from a simple hemorrhagic ovarian cyst, which is mostly solitary and round. Differentiation may be difficult, and both may show a density gradient within the cyst (due to blood products). Endometriosis can also present as solid fibrotic
Fig. 20.5 Endometrioma presenting cyst in the left ovarian fossa.
nodules implanted on the peritoneum, and may thus simulate peritoneal carcinomatosis. Dermoid Cyst
A dermoid cyst (mature cystic teratoma) is the commonest ovarian tumor in premenopausal women. Dermoids are bilateral in 8-15 %, and predominantly cystic with a varying amount of fat that is present in over 90% of cases, calcifications or bony structures in < 50%, or a fat -fluid level in > 10% of patients. Complications are torsion, rupture or malignant transformation (in 3%). •
CT Morphology
Dermoids are generally unilocular cysts with a mural nodule, the dermoid plug or Rokitansky body. The cyst is typically filled with greasy liquid, made up of keratin and sebum that has a low attenuation (Fig. 20.6). Other components
Fig. 20.6 Dermoid cyst ("mature teratoma") presenting as a rightsided ovarian mass that shows fat density in addition to solid components. Ascites is present in the cul-de-sac.
a
as a thick-walled
b
707
20 Female Pelvis
such as fat, hair, teeth, or even bone, are variably present. The cyst may be septated or contain debris with layering. There may be calcifications in the mural nodule. Most mature cystic teratomas contain only a small soft tissue component; the presence of a large (> 1Ocm), irregular soft tissue mass should raise suspicion of malignant transformation (immature teratoma). Cystadenomas
Cystadenomas account for 40% of benign ovarian tumors. Serous Cystadenomas are common in patients aged 20-50 years and 20% are bilateral. Serous tumors are more often unilocular than multilocular and filled with serous fluid (Fig. 20.7). Mucinous Cystadenomas are commoner after age 40 and bilateral in only 5%. These tumors are more frequently multi' ocular, filled with mucinous fluid and are often very large at presentation. •
CT Morphology
Depending on the serous or mucinous content of the cysts, the CT numbers vary between water and soft tissue density. The cystic walls and internal septa are of varying thickness and regularity; however, thickness should not exceed 3 mm. Malignant cystadenocarcinomas cannot be reliably distinguished from their benign counterparts by CT.
Fig. 20.7 Serous ovarian cystadenoma presenting as a cystic mass of homogeneous water density with welldefined margins and no internal septations.
Hydrosalpinx
Occlusion of the fimbriated end of the fallopian tube will lead to hydrosalpinx. The tube may be filled with serous fluid (hydrosalpinx), blood products (hematosalpinx), or pus (pyosalpinx). It is most commonly caused by pelvic inflammatory disease, but adhesions, endometriosis, or obstructing tumor may also cause this condition. •
CT Morphology
CT shows a coiled or serpentine cystic structure. Depending on the contents of the tube, the CT numbers vary, but are generally of lower attenuation than muscle. After contrast the thin tube wall will enhance. Ovarian Carcinoma
Ovarian carcinoma is the fifth highest cause of cancer death in women in the USA and has the highest lethality rate of all gynecological malignancies because in more than two-thirds of cases diagnosis is made when metastases are already present. Nearly all malignant ovarian tumors have combined solid and cystic components. Histologically, 85-90% are epithelial tumors and include serous, mucinous, endometroid, and clear cell adenocarcinomas. The remaining are malignant germ cell tumors (dysgerminoma, teratoma), sex cord stromal tumors (granulosa cell tumor), and sarcomas. A subset of borderline malignant tumors have a more favorable prognosis. Metastatic spread occurs early as peritoneal implants, hematogenously via the ovarian vein to the liver, and lymphogenously to the pelvic and para-aortic lymph nodes. Pelvic examination and transvaginal ultrasound are the mainstays in diagnosis of ovarian carcinoma. Surgical treatment is always indicated, since resection or debulking is beneficial even in patients with advanced tumors. The staging of ovarian carcinoma (Table 20.5) is done accordingly, and prognostic factors are based on the findings at surgery and the histological or cytological results that are obtained. CT or MRI are equally useful for improving patient selection and presurgical planning in advanced cases (stage III -IV disease) who are not amenable to optimal surgical staging or de-
Ovon'on Carcinoma Table 20.4 Criteria for the benign-malignant differentiation of a cystic ovarian mass Suggestive of
Suggestive
of malignancy
te tt<
benign disease
Homogeneous internal structure
Heterogeneous internal structure, solid components
Unilateral occurrence
Bilateral occurrence
Smooth margins
III-defined or lobulated margins
Unilocularity
Multilocularity
No contrast enhancement
hancement
Age <30 years
Heterogeneous contrast enAscites, peritoneal carcinomatosis, lymphadenopathy, metastases
bulking. CTand MRI are very useful for delineating peritoneal, bowel, and liver involvement. Accuracy varies substantially with the CT and MRI technique employed (between 60-90%), and with local expertise. Spiral or multislice CT is more widely available, is simpler to perform and cheaper, and is less susceptible to motion artifacts. MRI is better for characterization of malignant ovarian tumors and may be better for detecting malignant lymph node involvement and initial skeletal metastases. For the follow-up of ovarian tumors, a baseline examination may be obtained at six weeks after surgery. Subsequent examinations are scheduled to monitor therapy or if clinical, laboratory (CA-125) or ultrasound findings raise suspicion of recurrence. Second-look laparoto-
mies can be obviated in a large percentage of cases. •
CT Morphology
A benign-malignant differentiation is often difficult to make by assessing the CT morphology of a lesion (Fig. 20.8). CT features suggestive of malignancy include lesion diameter greater than 4cm, presence of papillary projections, walls and septa more than 3 mm thick, a mixed solid and cystic mass, a lobulated solid mass, and the presence of tumor vessels on contrastenhanced studies (Table 20.5; Figs. 20.9-20.11). CT signs of tumor extension in the pelvic organs include distortion of uterine contour, irregular interface between tumor and uterus, loss of tissue plane between tumor and sigmoid colon or bladder, encasement or direct extension into the sigmoid, distance from pelvic side wall < 3 mm, and tumor surrounding or displacing the iliac vessels. Preferred sites for peritoneal spread include the posterior cul-de-sac, infundibulopelvic ligaments, omentum, right paracolic gutter, and the undersurface of the right hemidiaphragm. Other common sites are the liver surface, porta hepatis, Morison's pouch, and bowel mesenteries. Implants are recognized by nodular thickening and enhancement of peritoneal surfaces (see Chapter 16) and by the invariable presence of ascites, both in the greater and lesser sac. Early stages may not be detectable with CT, while advanced cases are characterized by reticulonodular infiltration of the mesentery and greater omentum ("omental cake"). Calcification is frequently seen in metastatic serous adenocarci-
o
o
10
o Completely nonFig. 20.8 Correlation between the gross morphologic features of an ovarian mass and the likelihood of malignancy (after Schillinger, 1985).
margins Scalloped margins Loculated Well-defined margins
homogeneous Solid-cystic Cystic-solid Solid (cystic) Cystic (solid) Solid Watery
709
710
20 Female Pelvis Table 20.5 TIMM
TIMMand
FIGO
T1
ofovarian
carcinoma
(UICC, 1997)
Description Tumor limited to the ovaries
IA
T1a
Tumor limited to one ovary; capsule intact; no tumor on ovarian surface; nant cells in ascites/peritoneal washings
no malig-
T1b
Tumor limited to both ovaries: capsules intact: no tumor on ovarian surface; malignant cells in ascites/peritoneal washings
T1c
Tumor limited to one or both ovaries with any of the following: ruptured tumor on ovarian surface, malignant cells in ascites/peritoneal washings
T2
no
capsule,
Tumor involves one or both ovaries with pelvic extension T2a
II A
Extension
T2b
II B
Extension to other pelvic tissues;
T2c
IIC
T2a or T2b with malignant
T3
and/or
implants on uterus and/or tubes; no malignant no malignant
cells in ascites
cells in ascites/peritoneal
cells in ascites/peritoneal
washings
washings
Tumor in one or both ovaries with microscopically confirmed peritoneal outside the pelvis (and/or regional lymph node metastasis (= N1))
metastasis
T3a
III A
Microscopic
T3b
III B
Macroscopic
T3c
IIIC
Peritoneal metastasis beyond pelvis more than 2cm in greatest dimension regional lymph node metastasis (= N1)) K i : y ;
peritoneal peritoneal
metastasis
beyond pelvis
metastasis
beyond pelvis 2 cm or less in greatest dimension
Regional lymph node metastasis (internal sacral, para-aortic and inguinal nodes)
N1 MI
FIGO classification
IV
Distant metastasis
(excludes
Fig. 20.9 Cystadenocarcinoma appears as a multicystic, partially septated ovarian mass of nonhomogeneous density with internal calcifications. The patient underwent a previous proctectomy for colon carcinoma.
peritoneal
iliac, common
and liver capsule
iliac, external
(and/or
iliac, lateral
metastasis)
Fig. 20.10 Ovarian carcinoma appears as a multicystic mass with ill-defined margins and a nonhomogeneous internal structure. There is concomitant peritoneal carcinomatosis.
Benign-Malignant
Fig. 20.11 Functioning theca cell tumor presenting as a hypervascular, nonhomogeneous mass that is sharply demarcated from adjacent fat.
noma, while mucinous carcinoma may. develop pseudomyxoma peritonei. Lymph nodes larger than 1 cm in short axis diameter are considered suspicious. The external, internal, and common iliac as well as the para-aortic nodes are most frequently involved. With multislice CT, increased contrast enhancement of lymph nodes and a round shape are further criteria suggesting malignancy, although no hard data are yet available. Hematogenous distant metastases can be found in the liver, lung, adrenals, pancreas, spleen, bone, and kidney. Most parenchymal liver lesions are the result of growth of serosal implants (Fig. 20.12) and are not true hematogenous metastases (important for staging). For morphologic differentiation compare Chapter 7 (p.20l) .
Differentiation
in Cystic Ovarian Lesions
Fig. 20.12 Peritoneal metastasis of an ovarian noma that infiltrates the liver. Note the acute (arrow) at the interface between a portion of the and the liver indicating an extrahepatic origin mass.
•
carciangle lesion of the
CT Morphology
The most usual finding is a small solid mass close to the ovarian fossa that enhances after contrast administration. Often there is peritumoral ascites and hydrometra. Hydrosalpinx is not always present. Benign-Malignant Differentiation Cystic Ovarian Lesions
in
A benign-malignant differentiation usually cannot be made by assessing the CT morphology of a lesion, especially in complex ovarian masses (Fig. 20.8). A young age suggests benignancy, since only 3% of ovarian masses in women under age 30 are malignant. An ovarian mass can be classified as benign , with reasonable confidence if the lesion is unilateral and unilocular, has smooth margins, disFallopian Tube Carcinoma plays homogeneous CT density (regardless of Fallopian tube tumors are rare. Most of the the density level), and does not enhance with malignant tumors are adenocarcinomas. They intravenous contrast medium (Fig. 20.6). The more solid elements the lesion contains and the usually affect women over 65 years that present with abnormal bleeding or discharge and pain. less homogeneous its internal structure, the Tumors are diagnosed by pelvic examination or greater the likelihood that it is malignant radiological studies. They spread early to the (Figs. 20.8-20.11). "Classic" criteria of maligperitoneal cavity and to the pelvic lymph nodes nancy, such as a lobulated contours and ill-defined margins, are of lesser importance. Asand can be staged using TNM or FICO classifications. Patient prognosis is poor and is related to sociated findings that suggest malignancy are ascites, lymphadenopathy, peritoneal implants, tumor burden and stage. and parenchymal metastases.
711
20 Female Pelvis
If CT and ultrasound findings do not show evidence of malignancy (Table 20.5), it is sufficient to schedule the patient for ultrasound fol-
low-up. If doubt exists or if findings are suspicious for malignancy, a definitive diagnosis should be established by laparoscopy.
Solid Tumor*
Benign and malignant solid tumors in the female pelvis occur more often in the uterus than in the adnexa. When seen on routine abdominal studies, CT may provide morphologic clues to their diagnosis. Most often, however, diagnosis is made with US or MRI. Uterine Leiomyoma
Leiomyomas (fibroids) are benign tumors of smooth-muscle cell origin and are the commonest benign uterine neoplasms (in 20-30% of premenopausal women). Since the growth of leiomyomas depends on estrogen secretion, they are first observed in adulthood and tend to regress after menopause. Uterine leiomyomas may become symptomatic as they enlarge and/ or degenerate (dysmenorrhea, menorrhagia, displacement of adjacent organs, infertility). Torsion of a subserous mass and sarcomatous transformation are rare complications. Uterine leiomyomas are diagnosed by manual pelvic examination and transvaginal ultrasonography or MRI. CT is not indicated, but leiomyomas are a common incidental finding. Patients are treated by hysterectomy or by uterine artery embolization. 5
•
CT Morphology
Leiomyomas are usually found in submucous, intramural, or subserous locations. The uterus is enlarged with a lobulated outer contour, focal myometrial thickening, or deformity of the endometrial cavity (Fig. 20.13). Pedunculated tumors may lie in the adnexa, iliac fossa, or anterior to the aortic bifurcation. Multiplanar reformations or curved planar reformations are helpful for pedunculated lesions, especially when using multislice CT. Lesions often are round and sharply demarcated. Leiomyomas are can be anything from hypo- to iso- or hyperattenuating to the adjacent myometrium on contrast -enhanced studies (Figs. 20.14, 20.15). Leiomyomas tend to show hyperattenuation more frequently on early phases and may turn hypo attenuating on later phases (Fig. 20.15). Smaller lesions are homogeneous while larger lesions are more often heterogeneous. Low-density areas are the result of hyaline or cystic degeneration, necrosis, or infection, while hyperattenuation may result from bleeding (Fig. 20.14). Coarse, dystrophic calcifications are seen in 10% and are the most specific CTsign. Following uterine artery embolization, CT will show contrast retention in the leiomyomas
Fig. 20.13 CT appearances of uterine leiomyomas (fibroids). (a) Enlarged uterus with a homogeneous internal structure. (b) Calcifications coexisting with areas of central necrosis. a
b
Ovarian Torsion lipid-laden (thecoma). but usually a combination of both (fibrothecoma). Thecomas are common in peri- and postmenopausal women and there is often co-existing endometrial carcinoma. Fibromas are premenopausal lesions that are usually asymptomatic. •
CT Morphology
CT findings of fibromas and thecomas are similar and show a solid soft-tissue mass. Calcified foci may be present. Contrast enhancement is negligible. Differentiationfrom pedunculatedfibroids or a solid Brenner tumor is often difficult. Fig. 20.14 Markedly enlarged uterus with a nonhomogeneous internal structure caused by multiple leiomyornas. some displaying hypodense central necrosis. CT cannot exclude a malignant tumor, but the surgical specimen showed no evidence of malignancy.
shortly after the procedure. Central necrosis will manifest as cystic areas. Differentiating posttherapy effects from degenerated fibroids can be difficult, especially when multiple lesions are present. Subserous or pedunculated simulate an ovarian tumor.
leiomyomas may
Brenner Tumor
Brenner tumors are rare fibrous tumors of the ovary. Most tumors occur in patients over 40 years of age. The tumor can be solid, cystic, or mixed. They are usually benign, but malignant types have been described. •
CT Morphology
CT demonstrates an unilateral, sharply marginated tumor that is usually solid, or mixed solid and cystic. The cystic parts are often multilocular. Extensive amorphous calcifications of the solid parts are highly characteristic. The tumor enhances mildly after contrast.
Fibroma and Thecoma
Ovarian Torsion
These are the commonest solid benign tumors of the ovary and are derived from stromal cells. They are either purely fibroblastic (fibroma) or
Ovarian torsion is most frequent before puberty or during pregnancy. Presence of an ovarian mass (dermoid) predisposes to this condition.
Fig. 20.15 Uterine leiomyoma that demonstrates a hyperattenuation during the arterial phase (a) and hypoattenuation during the venous phase (b) of contrast enhancement. a
b
20 Female Pelvis Acute torsion presents with acute, severe lower quadrant pain. Subacute torsion is much more difficult to diagnose clinically, presenting with vague and unspecific symptoms. Venous flow is compromised first and leads to enlargement and edema. Arterial compromise results from progressive twisting and results in hemorrhagic infarction. CT Morphology
There is a solid mass with deviation of the uterus to the involved side. The adjacent blood vessels are engorged, there is free intra-abdominal fluid and the surrounding fat planes are usually obliterated. There is lack of enhancement of the mass after administration of contrast medium.
Cervical Carcinoma Cervical carcinoma is the third commonest malignancy of the female genital tract. It has an incidence of 16 per 100000 women and shows two peaks between 35 and 40 and between 60 and 65 years of age. Patients usually present with intermenstrual or postmenopausal bleeding. Risk factors include first intercourse at early age, sexual precocity, low socioeconomic background, and infection with human papilloma virus. Early detection with PAP-smears has lead to a significant decrease in mortality. In young women, carcinomas are usually very aggressive. Histologically, 80-90% are squamous cell carcinomas. The remaining types include adenocarcinomas, adenosquamous carcinomas or mucin-secreting adenoma malign urn. These latter tumors are more aggressive. Tumors spread through direction extension (vagina, uterine body) and lymphatic metastasis to the pelvic nodes. Para-aortic and inguinal lymph nodes are only involved when the pelvic wall or lower vagina is infiltrated. Hematogenous spread to the liver and lungs is infrequent and only seen with advanced stages. A poor prognosis of cervical carcinoma is related to young age, non-squamous cell tumor, tumor size larger than 4cm, stromal invasion greater than 5mm and malignant lymphadenopathy. Lymph node staging, however, is not included in the FICO system. Approximately 5% of stage lA-2 and 30-40% of stage 18 lesions are associated with pelvic lymph node metastasis. In doubtful cases, a staging laparotomy could be
performed (the role ofMRI with USPIO contrast media is also still evolving). Cervical tumors are diagnosed clinically by pelvic examination under general anesthesia and endocervical curettage. Clinical staging uses TNM or FICO classifications (Table 20.6). As the stage of the disease increases, bimanual examination becomes less accurate for estimating tumor extent, and staging errors increase from some 15-20% for stage I to over 60% with stage II disease. Transvaginal ultrasound, CT, and MRI are more accurate in defining the depth of infiltration. MRI is superior to ultrasound or CT, but multislice CT may improve the accuracy of CT staging in the future. The role of preoperative imaging in early carcinoma is to select surgical candidates and to provide more accurate staging. The exclusion of metastases and detection of parametrial invasion are the most important in this group. Tumors with stage I or IIA are treated by hysterectomy with or without lymphadenectomy, while tumors with stages lIB and higher are treated primarily by radiotherapy. In clinically more advanced disease, imaging can provide accurate staging and guide the selection of the most appropriate therapy. Finally, CT imaging plays a central role in radiotherapy planning. CT Morphology
Cervical tumor is relatively hypovascular and presents as a hypoattenuating mass or eccentric thickening of the cervix on contrast-enhanced CT. The uterus may be enlarged or fluid-filled if the tumor occludes the ostium (Fig. 20.16). Vaginal extension leads to focal thickening of the vaginal wall. CT criteria for parametrial involvement (Fig. 20.17 a) are: • prominent parametrial soft tissue strands or eccentric soft tissue mass • obliteration of the periureteral fat plane • irregular or unsharp demarcation of the lateral cervical border. A perifocal inflammatory or edematous reaction, e.g., after invasive diagnosis, can lead to a false-positive diagnosis of parametrial invasion. Also parauterine vessels, and the normal round, cardinal, and uterosacral ligaments should not be mistaken for lateral tumor extension. Extension to the pelvic wall is manifested by obliteration of the fat planes towards the
Cervical Carcinoma Table 20.6 TNM and FIGO stages of cervival carcinoma (UICC, 1997) TNM
FIGO
Definitiion Carcinoma in situ Carcinoma strictly confined to the uterus
T1 T1a
Invasive carcinoma diagnosed only by microscopy
IA
T1a1
IA-1
Stromal invasion less than 3 mm in depth and less than 7 mm horizontal spread
T1a2
IA-2
Stromal invasion 3-5 mm in depth and less than 7 mm horizontal spread
T1b
IB
Tibl
IB-1
T1b2
IB-2
T2
Clinically visible lesion confined to cervix or microscopic lesion greater than T1 a2 or I A-2 Clinically visible lesion 4.0 cm or less in greatest dimensions Clinically visible lesion more than 4.0cm in greatest dimensions Tumor invades beyond the cervix but not to pelvic wall, or to lower third of vagina
T2a
IIA
Without parametrial invasion
T2b
II B
With parametrial invasion Tumor extends to the pelvic wall and/or involves lower third of vagina and/or causes hydronephrosis or non-functioning kidney
T3 T3a
lilA
Tumor involves of lower third of the vagina but no extension to the pelvic wall
T3b
IIIB
Tumor extends to pelvic wall and/or causes hydronefrosis or nonfunctioning kidney
T4
IVA
Regional lymph node metastasis (paracervical, parametrial, internal iliac, external iliac, common iliac, presacral or lateral sacral nodes)
N1 M1
Tumor invades mucosa of bladder or rectum and/or extending beyond the true pelvis
IVB
Distant metastasis
Fig. 20.16 Cervical cancer (arrow, a) causing cervical obstruction and subsequent retention of secretions within the uterus (hydrocolpos, arrow head). Note an enlarged left ovary due to ovarian metastasis from concomitant breast cancer (arrow, b). Volume rendered sagittal and coronal images (4 x 2.5/15). a
b
715
20 Female Pelvis
a
b
c
Fig. 20.17 Invasive cervical carcinoma, (a) Stage liB disease is manifested by streaky, bilateral parametrial involvement without extension to the pelvic wall. (b) Stage IIIB disease is evidenced by parametrial invasion
on the left side and disruption ofthe fat plane bordering the pelvic wall. (c) Stage IV disease is demonstrated by additional involvement of the bladder.
internal obturator or piriform muscles (Fig.20.17b). There may be associated hydronephrosis, which upstages the tumor to stage IIIB, even when there are 2-3 mm intervening fat planes between tumor and pelvic wall. Infiltration of adjacent organs (bladder, rectum) is shown by loss of the adjacent fat planes or by direct evidence of organ invasion, such as asymmetric wall thickening, nodular indentation, or an intraluminal tumor mass (Fig. 20.17c). Iliac lymph nodes with a short axis diameter of 1 cm or more are suspect for malignancy. The overall accuracy in recent studies is reported to lie between 77% and 90% with falsenegative results in cases with micrometastases in lymph nodes smaller than 1 cm. The rounded shape of lymph nodes can be detected with multislice scanning, and may be associated with a higher incidence of malignant involvement. Abnormal para-aortic or inguinal nodes are indicative for stage IV disease.
Histologically, 80-90% are adenocarcinomas that range from well-differentiated to anaplastic grades. Other types include adenosquamous, papillary, serous, and clear-cell carcinomas. They spread via direct extension, lymphatic metastasis, peritoneal metastasis, and hematogenous metastasis. Para-aortic lymphadenopathy may occur without involvement of the pelvic nodes. Peritoneal spread is seen only with papillary and clear-cell types. The lung is the organ most frequently involved in hematogenous spread. The prognosis depends on the histology and degree of cellular differentiation, the depth of myometrial invasion, and the stage. The probability of para-aortic and pelvic involvement increases with the depth of myometrial invasion. When the outer half of the myometrium is infiltrated, lymph node spread is present in up to 40%, involving first the para-aortic and paracaval lymph node groups. In approximately 10% of cases the proximal vagina is infiltrated. Resectability of endometrial carcinoma is determined by the degree of parametrial invasion. As with cervical carcinoma, extension to the pelvic wall signifies irresectability. Irresectable tumors are treated by primary radiotherapy. The tumor is diagnosed by bimanual pelvic examination and fractional endocervical and endometrial curettage. Definitive staging is accomplished at surgery (abdominal hysterectomy; additional salpingo-oophorectomy and
Endometrial Carcinoma
Endometrial carcinoma is the most frequent carcinoma of the female genitourinary tract with an incidence of 28 in 100000 women. Most patients are between 55 and 65 years of age and present with postmenopausal or intermenstrual bleeding. Risk factors include obesity, diabetes mellitus, nulliparity and unopposed estrogen (e.g., tamoxifen) therapy.
Endometrial Carcinoma 717 tumor may be accompanied by a dilatation of the uterine cavity (hydrometra. hematometra, or pyometra) if the cervical orifice is occluded (cf. Fig. 20.16). Extrauterine extension is manifested by reticulonodular infiltration of the surrounding soft tissues with poor delineation of the uterine boundary (Fig. 20.18). Parametrial invasion extending to the pelvic wall is frequently accompanied by hydronephrosis. An associated ovarian mass may represent metastasis or an ovarian thecoma. Extension to the pelvic wall obliterates the fat plane of the obturator and piriform muscles. Invasion of adjacent organs (bladder, rectum) is marked by loss of the intervening fat planes or by direct evidence of organ invasion. Lymph nodes in the para-aortic and pelvic stations with a short axis diameter larger than 1 cm in diameter are suspect for metastatic involvement.
lymphadenectomy in more advanced cases) using the TNM or FICO classifications (Table 20.7). Preoperative radiological staging of endometrial carcinoma is usually reserved for patients in whom clinical diagnosis is difficult (concomitant leiomyomas. adenomyosis) and in those with advanced disease or aggressive histological subtypes. Transvaginal ultrasound. sonohysterography or (preferably) high resolution contrast-enhanced MRI are normally used. CT may be used to evaluate extrauterine extension. but staging accuracy for myometrial invasion is significantly inferior to MRI. CT is however commonly used to plan radiotherapy. •
CT Morphology
On contrast-enhanced CT. the tumor presents as a hypodense mass within a dilated endometrial cavity or in the uterine wall. The uterus may be enlarged. Assessment of myometrial invasion with CT is inaccurate. Tumor involvement of the cervix is characterized by cervical enlargement> 3.5 em with a heterogeneous hypo attenuating mass within the fibrom uscular stroma of the cervix. The
Table
20.7
TNM and FIGO classification RGO
Carcinoma
T1
Tumor
confined
Tumor
lirrued
IA
carcinoma (UICC 1997)
endometrial
Description
Tis
T1a
of
Note that the first lymph nodes to be involved by tumor are the para-aortic and paracaval . nodes before the pelvic nodes are infiltrated,,,,,,,
in situ to corpus
uteri
to endometrium
T1b
IB
Tumor
invades
up to or less than
T1c
1C
Tumor
invades
more than
Tumor
invades cervix
T2a
IIA
7»T2
T2b
II B
T3/N1 T3a
Endocervical Cervical Local
iliA
cells in ascites T3b
III B
Metastasis
T4
IVA
T~~~
M1
IV B
beyond
to pelvic
as specified
and/or
or peritoneal (direct (internal
Fn~aad~~tla~d~rm~~~~
uterus
only '3SSU
spread
serosa
involvement
Vaginal
IIIC
involvement
invasion
regional
Tumor involves
N1
but does not extend
glandular
stromal
and/or
one half of myometrium
one half of myometrium
adnexa
below: (direct
extension
-> or metastasis)
and/or
cancer
washings extension
or metastasis)
iliac, common
iliac, external
and/or
mucosa
bowel
iliac, parametrial,
sacral)
Distart metastasis (excluding metastasis to vagina, pelvic serosa or adnexa-including metastasis to intra-abdominal lymph nodes other than para-aortic or inguinal nodes)
20 Female Pelvis Fig. 20.18 Endometrial carcinoma. Invasion of the cervix and incipient extrauterine extension in stage III disease.
a
b
Recurrent Malignancies of the Uterus
The most important features that influence the incidence of recurrences are the stage and bulk of disease at presentation. Another important factor is the histological differentiation. Early detection is important as additional surgical (pelvic exenteration) or radiation therapies are available that show good results in limited pelvic recurrences. In cervical cancer, 32-35% of patients develop recurrences, usually in the first two years. The vaginal cuff, pelvic wall, and lymph nodes are the most frequent predilection sites for recurrent disease in the pelvis. Due to intensive use of radiation therapy in the pelvis, there is however a trend towards recurrent disease outside the pelvis. CT is more accurate in diagnosing recurrent disease than in primary staging and has a sensitivity of around 90%. In endometrial cancer, the recurrence rate is related to initial tumor stage and varies between 10% and 75%. Sites of recurrence vary with the treatment used, and more pelvic recurrences are seen after surgery than after radiotherapy. Extrapelvic sites include retroperitoneal lymph nodes, and peritoneal, liver, and lung lesions. Therapeutic results are inferior to results from initial treatment and an aggressive approach is advocated. CT is often the method of choice as it is widely available, provides rapid screening of the entire abdomen, has superior spatial resolution and may guide biopsy of suspicious lesions. It has however, limited ability to differentiate (radiation) fibrosis from recurrent tumor. MRI may be the superior diagnostic modality due to
its better soft tissue differentiation and its ability to distinguish between tumor recurrence and fibrosis. However MRI is still more expensive and availability for abdominal applications is limited. The role of PET for recurrent gynecological cancer is evolving. In the follow-up of tumors, a baseline examination (CT or MRI) may be performed 6-8 weeks following surgery or radiotherapy. Subsequent examinations are performed to evaluate response to radiotherapy (downstaging) or if clinical or ultrasound findings raise suspicion of a recurrence. Especially when using MRI, an interval of six months is required after radiation therapy to be able to differentiate between edema and inflammatory response and tumor. •
CT Morphology
CT diagnosis is based on morphology rather than characteristics of contrast enhancement. The morphologic criteria of recurrences of cervical or endometrial carcinoma have the same characteristics as the primary tumor. In cervical cancer, central recurrences can manifest as solid masses with or without necrosis (vaginal cuff, Fig. 20.19), with accompanying hydrometra or hydronephrosis. Lateral pelvic recurrence can appear as irregular pelvic wall masses at the level of the uterus or vagina, or an ovarian tumor, often with urethral obstruction. Parametrial change after radiotherapy or recurrent invasion into rectum or bladder is usually difficult to characterize with CT (Fig. 20.20). Extrapelvic recurrences can manifest as lymphadenopathy, hypovascular liver lesions, peritoneal or omental implants, bowel masses or intestinal obstruction, pulmonary lesions, or de-
Cestational
Fig. 20.19 Recurrence of a leiomyosarcoma on the vaginal stump following hysterectomy.
Trophoblastic
Fig. 20.21 Hydatidiform mole with typical vascularity ofthe trophoblastic villi.
Disease
hyper-
Fig. 20.20 Recurrence of a cervical cancer (arrow) with infiltration ofthe bladder and obstruction of both ureters (arrow heads). Axial (a) and coronal (b) volume-rendered displays (4 x 1/6). b
a
structive bone lesions in vertebral column or pelvis. Recurrent endometrial cancer usually presents as a pelvic mass, or as pelvic or retroperitoneallymphadenopathy. Peritoneal carcinomatosis with ascites is also relatively frequent. The possibility of distant metastases is related to the histological grade or tumor stage at initial presentation. Metastases are commonly located in the mesentery and liver. Gestational Trophoblastic Disease
Gestational trophoblastic disease (GTD) encompasses a variety of diseases such as hydatidiform mole, invasive mole, and choriocarcinoma. Choriocarcinoma is a rare malignancy diagnosed in 1 of 30000 pregnancies. More than 50% arise from hydatidiform mole pregnancies. Vascular
invasion, hemorrhage and necrosis can be present. Local extension and distant metastases to lung, liver, or brain are common. The tumor is treated with methotrexate chemotherapy. Diagnosis and follow-up relies on pelvic examination and serum (3-hCC measurements. The tumor is staged using TNM or FIGO classifications. Transvaginal ultrasound and MRI are used for differentiating the various types ofGTD and to assess the amount of myometrial invasion. CThas a very limited role, but may be used in the evaluation of regional and distant metastatic disease. •
CT Morphology
CT may demonstrate irregular, eccentric hypodense foci in the myometrium or endometrium. Often the uterus is greatly enlarged. Postcontrast scans in the arterial phase show enhancement of vascular intensity in the trophoblastic villous structures (Fig. 20.21). The
719
20 Female Pelvis uterine arteries are invariably enlarged. CT cannot reliably differentiate among the various trophoblastic tumors. Chest or multiphasic liver CT can demonstrate pulmonary nodules or hypervascular liver metastasis.
Vulvar carcinomas are also rare tumors that occur in older patients. Patient" present with a palpable mass, pain, bleeding, or pruritus. Most are squamous cell carcinomas. Treatment is by laser ablation or surgery. MRI is the method of choice for imaging.
Other Primary Malignant Tumors
Primary lymphomas of the ovaries without bone
marrow infiltration or lymph node involvement are very rare. Sarcomas of the genital organs comprise a heterogeneous group and are rare (2-3% of genital malignancies). Patients usually present with abnormal vaginal bleeding. The diagnosis is usually made at operation and histology. The lesions of greatest clinical importance are leimyosarcomas of the uterine wall or ovary (rapid growth compared with leiomyomas), endometrial stromal sarcomas, and mesodermal mixed tumors (e.g., muellerian mixed tumor). Sarcomas have no distinctive imaging characteristics besides their rapid growth. Vaginal carcinoma is rare and usually affects women between 60 and 70 years of age. Almost all tumors are squamous cell carcinomas that most often arise from the upper posterior vagina. Diethylstilbestrol (DES) exposure in utero increases the risk of developing the very rare clear-cell adenocarcinomas of the vagina. Spread of tumor is by direct extension or metastasis to the iliac lymph nodes. MRI is the preferred modality for staging and diagnosis of recurrence. CT may be used for advanced disease or to stage patients who have contraindicationsforMRI (Fig. 20.22).
Metastases
Metastases to the ovaries account for 10% of ovarian malignancies. These occur uni- or bilaterally. The commonest primaries are carcinomas of the gastrointestinal tract, pancreas, breast, and endometrium, and melanoma. Bilateral metastases of a signet-ring carcinoma of the stomach are called Krukenberg tumors and have a very poor prognosis. Ovarian metastases may be predominantly solid or cystic. In patients with generalized malignant lymphoma, the ovaries may be involved in 30% of cases. Tumors are usually large and bilateral. The lymphoma is of the non-Hodgkin type. The vagina may harbor rare metastatic lesions of colon, bladder, and renal cell carcinoma, as well melanoma. These lesions are more frequent than primary vaginal tumors.
Fig. 20.22 Invasive vaginal carcinoma. The club-like expansion of the vagina and its indistinct boundaries on the left side are indicative of extravaginal tumor spread. a
b
Endomyometritis
Infectious and Vascular
and Pyometra
Disease;
Pelvic Inflammatory Disease
Pelvic inflammatory disease (PID) is "a common disease in sexually active women and may be responsible for secondary infertility. Patients with an intrauterine contraceptive device have an increased risk. The most frequent ascending infectious agents are Chlamydia trachomatis and Neisseria gonorrhoeae. Salpingitis due to these microorganisms may rarely be complicated by perihepatic fluid collections {Fitz-Hugh-Curtis syndrome}. A tubo-ovarian abscess occurs as complication of a generalized inflammatory process and fluid retention within the tubes. Pelvic inflammatory disorders are diagnosed by bimanual pelvic examination, laboratory tests, and bacteriologic studies. Transvaginal ultrasound can also be used to detect adnexitis, endomyometritis, and possible complications such as pyosalpinx, tubo-ovarian abscess, cul-de-sac abscess, hydronephrosis, and free intraperitoneal fluid. CT is indicated in patients with suspected complications of inflammatory diseases and in cases that are indeterminate by ultrasound. Abscess
Tubo-ovarian abscess (TOA) and cul-de-sac abscess are severe complications of adnexitis or endomyometritis. Tubo-ovarian abscesses may be multiple or bilateral. The ovaries are not necessarily enlarged, and a typical abscess configuration is not always present. •
CT Morphology
TOA can show an adnexal fluid-filled tubular structure (hydropyosalpinx) or multilocular cystic mass with ill-defined margins. Peri -ovarian edema (acute) and inflammatory infiltration (chronic) are usually present. Often the mesosalpinx is displaced and the contour of the adjacent uterus is effaced. The abscess wall will enhance after contrast administration. A cul-de-sac abscess usually shows an fluid collection with enhancing rim. Surrounding ligaments and fasciae may be thickened by the inflammatory process (Fig. 20.23).
Fig. 20.23 Florid cul-de-sac abscess displays marked rim enhancement after contrast administration. A gas collection is visible near the right pelvic wall.
Endomyometritis and Pyometra
Endomyometritis most commonly occurs after curettage and during the postpartum period. It is characterized by the development of inflammatory infiltrates and small abscesses in the uterine wall. The accumulation of purulent material within the uterine cavity is called pyometra. Ultrasound and MRI are superior to CT for diagnosis. •
CT Morphology
The uterus in endomyometritis appears heterogeneous, and the borders of the uterine cavity are indistinct. After contrast material injection, the affected area may be demarcated as an area of low attenuation within the richly perfused postpartum uterus (Fig. 20.24). Large hemorrhagic areas and placental remnants can also be identified within the uterine cavity. Pyometra appears as an intrauterine fluid collection with low attenuation values that exceed water density (Fig. 20.25). Endometrial carcinoma is a frequent cause of pyometra in older women.
721
20 Female Pelvis
Fig. 20.24 Endomyometritis during the postpartum period (day 10) appears as a hypodense area within the spongy, enlarged uterus.
Fig. 20.25 Pyometra and presacral abscess following proctectomy. The intrauterine collection has caused marked enlargement of the uterine cavity with thinning of the myometrium.
Fournier's Gangrene
gangrene (gas bubbles), erysipelas (no gas), pyoderma gangrenosum, and synergistic necrotizing cellulitis.
Necrotizing
is a rare type of fulminating infection caused by a combination of aerobic and anaerobic organisms. A special form is Fournier's gangrene (necrotizing fasciitis of the vulva), which originates in the anogenital region (often Bartholin's glands). CT is the modality of choice for determining the depth of involvement. •
Iesciitis
CT Morphology
CT reveals extensive gas collections and streaky densities in the perineal soft tissues with phlegmonous extension along fascial planes (Fig. 20.26). Differentiation is required from gas
Treatment consists of immediate, radical debridement of the necrotic tissues.
Ovarian Vein Thrombosis
Septic thrombosis of the ovarian vein occurs almost exclusively during the postpartum period (puerperium). The primary imaging modalities for diagnosing this condition are ultrasound and MRangiography. Other causes of thrombosis are rare, but PID , post-abortion thrombosis, and thrombosis after extensive pelvic surgery with lymphadenectomy have been described. The vein may also thrombose in patients with rare coagulation disorders as part of a systemic involvement. •
Fig. 20.26 Fournier's gangrene. Phlegmonous spread of gas inclusions in the setting of necrotizing fasciitis.
CT Morphology
The main trunk of the ovarian vein, which is in contact with the psoas muscle lateral to the ureter, can be evaluated on contrast -enhanced CT scans. A thrombosed ovarian vein appears enlarged, shows a central low attenuation thrombus and an enhancing vessel wall. In postpartum thrombosis the right vein is affected in 90% of cases.
Pelvic Varices (Pelvic Venous Congestion Syndrome) Fig. 20.27 Pelvic venous congestion with markedly dilated periuterine veins (arrow, a) that drain into an enlarged left ovarian vein (arrow, b).
a
Pelvic Varices (Pelvic Venous Congestion Syndrome)
Pelvic venous congestion usually affects multiparous women and is a frequent cause of chronic, dull pain that is exacerbated by activity. Color Doppler ultrasound is the diagnostic method of first choice. The varices may be treated by transcatheter embolization.
b
•
CT Morphology
CT performed for other indications may show a dilated gonadal vein or dilated veins in the broad ligament or ovarian plexus as the only abnormality (Fig. 20.27).
CT in Pregnancy Low-dose CT has become outmoded as a technique for obstetric pelvimetry. But if a CT pelvic examination is performed in a woman with an unsuspected early pregnancy, and the uterus is scanned by the main beam, the radiologist is required to estimate the dose to the fetus. The volume CT dose index (CTDIvoi) is a rough estimate of the radiation dose to the uterus. This number is printed on the scanner console by most newer CT scanners or can be calculated by the kVp, mAs setting, pitch factor and section collimation SC from the test data of the vendor or routine quality test records for the unit. The radiation risks of exposure to the gravid uterus include fetal abnormalities, mental retardation, and increased risk for developing childhood cancer. For organ doses below 200 mSv, no substantial increase in risk for fetal abnormalities could be shown as compared to the natural risk. For mental retardation, the same holds true up to 1000 mSv if the radiation exposure occurred outside the S_ISlll week of pregnancy.
During this critical phase of brain development, however, a dose in excess of 500 mSv was shown to lead to a slightly higher number of mentally retarded children. The risk of developing childhood cancer starts exceeding the natural risk at 10 mSv, but up to 50 mSv the absolute risk remains below 0.7%. Radiation doses to the uterus applied during diagnostic CT procedures can be reduced well below 20mSv (CTDIvoi= 20mGy) with good CT technique. Even in multiphasic scans the organ dose should never exceed 100 mSv. Nevertheless, a three-stage approach is recommended in cases where exposure of the gravid uterus is assumed to have occurred. If the estimated dose (CTDIvoi) is less than 100 mSv, it may be assumed that there is no proven risk of fetal anomalies or mental retardation in excess of the natural occurrence of theses events (stage 1). There is no need for further dose calculations, and there is no indication for pregnancy termination. If the estimated dose is higher than
723
724
20
Female Pelvis
100 mSv, the radiologist should make a more precise dose calculation based on standard values for equipment- and patient-specific data (stage 2), If the fetal dose is found to be in excess of 200 mSv, the radiologist should again calcu-
late the dose based on individual scanner- and patient-specific measurements (stage 3), The resuits give an indication of the individual risk for the fetus, and elective termination of pregnancy may be discussed in high-risk cases,
21 Male pelvis A.J. van derMolen.
K.-J. Lehmann
26 Examination Technique Congenital Anomalies Ectopic Testis (p. 729) Cryptorchidism (p. 730) Benign Lesions .. Calcifications (p. 730) Benign Prostatic Hyperplasia (p. 731) Cystic Prostatic Lesions (p. 731) Testicular cyst (p. 732) Hydrocele (p. 733) Epididymal Cyst and Spermatocele (p. 733) Varicocele (p. 733) "Enlarged" Seminal Vesicles (p. 733) Benign Scrotal Neoplasms (p. 734) Malignant Tumors Testicular Cancer (p. 734) Testicular Lymphoma and Metastases (p. 736) Prostate Cancer (p. 736) Inflammatory
Processes
Prostatitis (p. 739) Tuberculosis (p. 739) Fournier's Gangrene (p. 740) Postoperative Changes (p. 740) .am
28 729
27 Male Pelvis
Diseases of the spermatic cord, testis, epididymis, and penis are primarily evaluated with ultrasound, CT scanning of the male pelvis is indicated only for lymph node staging or if there is suspicion of spread of a tumor of the prostate or seminal vesicles into adjacent tissues, The key modalities for diagnosing the early stages of prostatic carcinoma are laboratory tests, transrectal ultrasound, fine needle aspiration biopsy, and MRl, preferably using an endorectal or a combined endorectal and phased array pelvic Table 21.1 Tumor
Indications for CT examination of the male pelvis
diagnosis:
Testicular carcinoma: Prostatic carcinoma:
Radiotherapy
planning:
Oncologic follow-up:
a
coil, The role of CT is limited because it is not able to differentiate between normal, hyperplastic, and cancerous glandular tissue, CT is important for nodal staging, the evaluation of extracapsular tumor extension, and for the followup of prostatic and testicular tumors, Characteristic findings in the male pelvic organs, especially the prostate, are frequently noted as incidental findings in pelvic CT examinations (Table 2U).
Essential for nodal staging (atxlomen) and detection q metastases (lungs, liver) Surgical planning for advanced tumor staqes' Staging i1 high-risk patients
Prostatic carcinoma, seminoma Testicular tumors: Prostatic carcinoma:
Regular follow-ups (intially at 3-month intervals) Suspected recurrence; 6-month follow-ups are adequate only for PSA-negative tumors
CT can be used forthese indications, but MRI is preferred
Anatomy Prostate
The prostate is shaped like an inverted cone, with its base along the bladder base and its apex close to the pelvic diaphragm (Fig. 21.1). Normally, the organ measures 4.0-4.5 x 3.0-4.0 x 2.5-3.0 em and is enclosed in a fibrous capsule that is continuous with the fascia of the pelvic diaphragm. Anatomically the prostate is composed of five lobes: anterior, middle, posterior, and two lateral lobes. This description has been superseded by the concept of zonal anatomy of the prostate. The anterior portion is formed by the anterior fibromuscular stroma that forms the prostatic capsule. The inner gland consist of the perirurethral glandular tissue and the transition zone that is located anteriorly in the mid portion and surrounds the proximal urethra above the veromontanum. The outer gland is made up of the central and peripheral zones. The funnel-shaped central zone is situated at the base and surrounds the ejaculatory ducts. The peripheral zone forms the rest of the gland posteriorly, laterally, and at the apex (Fig.21.1c). It surrounds the distal prostatic urethra. MRI can clearly depict the zonal anatomy
of the prostate owing to differences in water content between the peripheral and other zones. To a lesser extent, contrast-enhanced CT can also differentiate these zones. The peripheral zone appears relatively hypoattenuating in relation to the other zones. The paired neurovascular bundles are found posterolaterally at 5 o'clock and 7 o'clock and are important in the staging of prostate cancer. The fascia ofDenonvilliers lies between the posterior surface of the prostate and the anterior rectal wall, The lateral margins of the prostate are formed by the obturator internus muscle superiorly and the crura of the levator ani muscle inferiorly. The periprostatic venous (Santorini's) plexus consists of numerous contorted-vessels of varying size that lies lateral and antehbr'. The prostatic urethra courses centrally through the prostate, making a sharp angulation at the veromontanum which lies in the urethra halfway between prostatic apex and bladder neck. The ejaculatory ducts enter the gland superiorly and insert in the urethra at the veromontanum.
Spermatic Cord Spermatic cord ------.. Obturator
artery and vein
Common femoral artery and vein Inguinal ligament
Obturator artery and vein Spermatic cord
\ \
Common femoral artery and vein
Urethra
" \
Uretralorifice
_----1
Pelvic venous plexus Internal obturator
muscle
Seminal Vesicles
The paired and usually symmetrical seminal vesicles are located superior and posterior to the prostate. They are obliquely oriented convoluted tubules that lie between the posterior surface of the bladder and the rectum (Fig. 21.1). Unilateral aplasia or hyperplasia is extremely rare. The size of the seminal vesicles varies with age and is widest in the fifth and sixth decade. Normally they measure 4.5-5.5 x 2.0cm and ultrasound studies indicate that the seminal vesicles have an average total volume of approximately 8.5 em'. Symmetrically "enlarged" seminal vesicles are usually caused by fluid retention that does not have pathologic significance. The excretory duct of the vesicle joins the ampullary segment of the vas deferens. The seminal vesicles are bordered laterally by numerous small vessels belonging to the periprostatic venous plexus. Calcifications in this region usually represent phleboliths. The angle between the bladder and seminal vesicle on each side is occupied by fatty tissue that is traversed by the ureters on their way to the bladder trigone. Obliteration of this angle may signify invasion by tumor.
Levato ran i m usc ""e-----+--' '-1----1-Internal pudendal _-----.J artery and vein Internal obturator
Ischial nerve
muscle
)k-I--~----\~
Peri urethra I inner zone Transitional zone Urethra
Outer zone (central part) Outer zone (peripheral part) !Ejaculatory duct Fig. 21.1 Sectional anatomy of the male pelvis (a, b) and zonal anatomy of the prostate in benign prostatic hyperplasia (c).
Spermatic Cord
The vas deferens is a continuation of the tail of the epididymis and ascends through the inguinal canal. In the pelvis it follows the pelvic wall and then curves medially and caudally and runs along the superomedial part of the seminal vesicle. After joining with the excretory duct of the seminal vesicle it is called the ejaculatory duct and crosses the prostate to end in the veromontanum. The spermatic cord caudal to the inguinal ligament is clearly visible on CT scans as a structure of soft -tissue density surrounded by an elliptical fatty area and by fascia (Fig.21.1).
727
27 Male Pelvis
The spermatic cord contains a variable offat, vessels, and the vas deferens.
amount
Testis and Epididymis Generally the testes are excluded from the CT imaging volume, and, if possible, they are further protected from x-ray exposure with a gonadal shield which should not extend into the scan field. When included in the scanned range, the testes appear as ovoid structures of homogeneous soft-tissue density with a dense fibrous capsule, the tunica albuginea. This tunica invaginates posteriorly as the mediastinum testis in which the fibrous septa of the testes converge. The seminiferous tubules form a duct
Examination
network at the rete testis which may be dilated as a normal variant. Most normal internal testicular structures have identical soft tissue attenuation and cannot be distinguished with CT. The average testicular size is approximately 3.8 x 3.2 x 2.5 cm'. The epididymides are poorly delineated on CT scans. They are located posterior to the testis and made up of head, body, and tail. They join the ductus (vas) deferens near the posteroinferior testis. Above the epididymis the tortuous pampiniform venous plexus accompanies the testicular artery. Surrounding the testis there is a fluid sheath of 1-2 mm separating the testis from the tunica vaginalis and the layers of the scrotal wall.
Technique
Patient Preparation Adequate bowel distension facilitates the distinction between (collapsed) bowel loops and pelvic lymph nodes. The fluid absorbed from oral administration of 1-1.5 liters of iodinated or barium-containing contrast material also distends the bladder, displacing the loops of small intestine cephalad. lt may be helpful to opacity the rectum with a moderate amount of rectal contrast medium (approximately 100mi of 5-10% diluted CM). An overdistended rectum can compress the pararectal fat and make the scans difficult to interpret. With multislice scans, negative contrast agents such as water or methylcellulose may be used. They offer the advantage of not interfering with CT angiographic displays or volume-rendered images of the soft tissues.
Acquisition Technique The evaluation of extracapsular or extravesicular tumor extension suffers mostly from partial volume effects. Thus, selective examinations of the prostate and seminal vesicles require thinner slices than are used for routine tumor staging and are best preformed using thin-section spiral or multislice CT (Table 21.2). Delineation of tumor spread is highly facilitated by multiplanar reformations in a coronal or sagittal plane.
Due to the high frequency of bone metastases from prostatic carcinoma, images should also be evaluated in a bone window setting whenever suspicious prostatic findings are noted at CT.
Contrast Injection Intravenous contrast is not essential but is helpful for differentiating vessels from mass lesions and lymph nodes. After intravenous contrast administration the zonal anatomy of the prostate can be appreciated more readily but this usually is not sufficient to be of diagnostic value. The zonal anatomy becomes more prominent with higher flow rates and a scan delay of some 30s. Such high flow rates also improve enhancement of the bladder wall and may improve the detection of bladder invasion by prostatic tumors. Prostatic carcinoma and most testicular tumors rarely cause hepatic metastases. Thus, there is no need to optimize the contrast protocol for liver imaging. Opacification of the ureters is particularly helpful in patients with tumors causing urinary obstruction. In most cases, 20-30 ml of iodinated contrast material injected intravenously 5-30 minutes prior to the examination (the
Ectopic Testis delay depending on the degree of obstruction) suffices for opacification of the ureters. After injecting larger amounts of contrast material full distension of the bladder with contrast medium
Table 21.2
729
can lead to artifacts in the regions of the prostate and seminal vesicles (see. Fig. 7.31) or lead to movement artifacts due to patient discomfort.
Protocols for CT scanning of the male pelvis
General Oral contrast
1000-1500
Rectal contrast
Optional: 100 ml (prostatic carcinoma) Optional: 500 ml (testicular tumors)
Patient position Scan range
Supine with elevated arms Pelvis: iliac crest to below ischiac bones from diaphragmatic crura to below ischiae bones Retroperitoneum:
~
Respiratory phase
:==Windowing •... -
ml, administered
Inspiration (spiral CT) Noncontrast CT: Contrast-enhanced CT: Bone (prostatic carcinoma):
-
Scan parameters
l-slice SC/TF/RI
Screnning (lymph nodes)
5/10/5
Small pelvis (tumor staging)
3/5/3
Contrast injection
l-slice
Screening
60 min prior to examination
t t
W/L~400/40 W/L~400/60 W/L~ 1500/500
4-slice' SC
16-slice' SC
axial SWIRl
MPRb SWIRl
2-3
1-1.5
5/4
-
--- -
1-l.25
0.75-1.25
4/3
3/3 cor 4/4 saq
4-slice
16-slice
V/F/D
V+N/F/D
V+N/F/DXc
120/2/70
100 + 50/2/70
100 + 50/2/80
120/4/40
120 + 50/5/40
120 + 50/5/20A
;~
Tumor staging
d
I
SC~ slice collimation (mm), TF~ table feed (rnrnjrotatton), RI ~ recon. increment (mm), f t ~scanning direction, SW ~ effective section width (mm), MPR ~ multiplanar reformats, axial ~ axial sections, cor ~ coronal, sag ~ sagittal CM ~ contrast material, Contrast concentration ~ 300 mg/ml iodine V ~ volume of CM (ml), N ~ volume of saline flush (ml), F~ flow rate (rnljs), 0 ~ start delay, X ~ trigger region a Pitch P ~ TF/(NxSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SW/RI ~ 1-l.5/0.7 ( read: 20A ~ start delay 20 s after reaching 50 HU enhancement in the infra renal aorta d ureters may be opacified by injecting 20-30 ml CM 15-30 min prior to the exam
II Ectopic Testis Testicular ectopia is a rare disorder. The ectopic testis leaves the normal pathway of testicular descent and occupies an abnormal position. The
commonest location is the inguinal pouch; femoral or contralateral hemiscrotum locations are less usual.
27 Male Pelvis
CT Morphology
An oval-shaped mass of soft-tissue density is located at the sites mentioned. Inguinally, this is superficial and lateral to the external ring of the canal.
Cryptorchidism In cryptorchidism, the normal course of descent of the testis becomes arrested at some point (in the inguinal canal in 72%, prescrotally in 20%, and abdominally in 8%). It has an incidence of 0.5-0.8% in boys 1 year and older, is bilateral in one-third of patients, and is more often located on the right side. Cryptorchidism interferes with normal sperm production and predisposes to malignant degeneration (approximately 1 in 20 abdominal testes, 1 in 80 inguinal testes), usually seminoma. Ultrasound or MRI are the diagnostic modalities of first choice, but CT scanning
has also proven to be accurate in localization and is superior in the diagnosis of abdominal testis. Negative imaging is an indication for laparoscopy. CT Morphology
An undescended testis appears as an oval softtissue mass that is located at a typical site along the ureter or spermatic cord. Undescended testes are usually smaller than normal testes (lower perfusion). In general. it is easier to detect and differentiate an undescended testis in the inguinal canal or in the lower pelvis as compared to upper pelvis or lower abdomen, where it has to be distinguished from bowel loops, vascular structures, and lymph nodes. Extrascrotal testes may show heterogeneous soft tissue attenuation values (more sensitive in MRI) due to ischemic changes. A large undescended testis (>2cm) or one containing areas oflow attenuation is suspicious of malignant transformation.
Benign Lesions Benign lesions of the testes and prostate are not an indication for CT, but they are frequently encountered as incidental findings. MRI and ultrasound are clearly superior to CT in differentiating intratesticular from extratesticular lesions and in discriminating solid from fluid structures.
21.3
Spermatic cord:
Calcifications
in the male genital
Diabetes Degenerative
organs
changes
Tuberculosis, syphilis, nonspecific infections Seminal
vesicles:
Tuberculosis, Gonorrhea, Schistosomiasis
Prostate:
Calcified
secretions
Previous transurethral Tuberculosis
Calcifications Calcifications are a common incidental in the male reproductive tract.
Table
resection
finding
CT Morphology
Calcifications in the spermatic cord, seminal vesicles, or prostate may have a degenerative or inflammatory etiology (Table 2 1.3). Small calcifications in the prostate (up to the size of a grape seed) usually result from nonspecific lithiasis in prostatic secretions and are an innocuous incidental finding (Fig.21.2a). Coarse prostatic calcifications may have an inflammatory etiology, occur in focal benign prostatic hyperplasia (BPH), or may develop following a transurethral resection (TUR) of the prostate
(Fig.21.2b). Calcifications in the seminal vesicles are seen in diabetics or following tuberculosis. Stones may develop rarely in the seminal vesicles in patients with urolithiasis and may be of variable size. Small calcifications lateral to the prostate usually represent phleboliths in the periprostatic venous plexus (Fig.21.2b). Small calcifications may be seen in the testis (microlithiasis) and this is believed to be a risk factor for developing malignant disease. Epididymal calcifications are found in 3% of males.
Cystic Prostatic Lesions
Fig. 21.2
Prostatic calcifications.
(a) Calcifications in the ejaculatory ducts in a patient with BPH. (b) Follow-up scan in a patient who underwent previous TU R of the prostate, which yielded malignant cells. CT demonstrates coarse, nonmalignant-type prostatic calcifications in addition to phleboliths.
a
Benign Prostatic Hyperplasia
Benign prostatic hyperplasia is a very common incidental finding in older males and leads to enlargement of the central part of the gland. BPH represents a proliferation of glandular or interstitial elements in the transitional zone and is frequently accompanied by benign prostatic calcifications ("prostate stones"). Focal changes may lead to nodules or adenomyomata. The stage of the disease is determined by clinical symptoms and not by the size of the prostate or the degree of bladder floor elevation. CT is not used in staging BPH, but prostatic enlargement is commonly seen in pelvic CT studies. •
CT Morphology
BPH is manifested on CT by prostatic enlargement (transverse diameter> 5 ern) that is indistinguishable from prostatic adenoma. The peripheral glandular tissue typically not hyperplastic is demarcated as a hypodense zone after
b
contrast administration (Fig. 21.3a). With faster contrast material injection, hypoattenuating and hyperattenuating nodules may be found in the prostate (Fig.21.3b). There are no data yet that correlate these findings with histology. A sufficiently large prostate may form a curved impression on the bladder floor or may lead to subvesical obstruction. The prostate is still sharply delineated with respect to surrounding fat and the levator sling. Cystic Prostatic Lesions
Cystic prostatic lesions are an occasional incidental finding on CT scans. They are often congenital and are evaluated by ultrasonography and if necessary by percutaneous aspiration. Utricular cysts arise from a dilatation of the utricle and are frequently associated with other anomalies of the genitalia. They are 8-10 mm long and communicate with the posterior
Fig. 21.3 Prostatic hyperplasia. (a) The transitional zone of the prostate appears hyperattenuating after contrast administration. The lateral, peripheral part of the gland is defined as a hypoattenuating area on either side. (b) Hypervascular focus in a patient with normal size gland and normal PSA levels. The CT morphology is highly suspicious for an early prostate cancer. Biopsy should be considered. a
b
27 Male Pelvis
urethra. Mullerian duct cysts are remnants of the muellerian duct system and not associated with other anomalies. Patients may present with urinary retention, infection, or stone formation. There is a stalk to the veromontanum and an increased incidence of carcinomas. Ejaculatory duct cysts are rare. They can be congenital or postinflannmatory and result from obstruction. A vas deferens cyst is rare and involves the ampulla (Table 21.4). •
CT Morphology
Utricular cysts lie in the midline, are small and usually teardrop-shaped. Mi111erianduct cysts
are midline, retrovesical, and larger than utricular cysts. Calcification or stones may be seen. Ejaculatory duct and vas deferens cysts are paramedian. CT is not accurate in the differentiation of these cystic prostatic lesions (Fig. 21.4). Testicular cyst
Intratesticular cysts are rare lesions that are usually solitary. They have idiopathic, postinflammatory, or post -traumatic etiologies. They customarily measure 5-7 mm, are often located adjacent to the rete testis, and are often accompanied by cysts in the epididymis. Tunica albuginea cyst make up 10-20% and are peripherTable21.4
Fig. 21.4 CT following open enucleation adenoma demonstrates a hypoattenuating site of the former adenoma.
of prostatic area at the
ally located. Tubular ectasia of the rete testis a normal variant seen in older men, and often bilateral. It is usually associated with spermatocele or epididymal abnormalities. Epidermoids are rare solitary cystic lesions in the testes that can be neoplastic in 1%. Differentiation from other germ-cell tumors is not possible and therefore epidermoids are resected. •
CT Morphology
These are seen as simple well-defined cysts with fluid -like contents in the testicular parenchyma
Cystic prostatic lesions
Congenital Utricular cyst
Usually <10 mm, midline in central zone of prostate; communicates with urethra; associated with other genital anomalies: hypospadia, cryptorchid ism
Mullerian
Large, often extending above the prostate; midline; may contain calcifications; Third to fourth decades; connects with stalk to veromontanum
duct cyst
Ejaculatory duct cyst
Paramedian; rare; may contain calcifications; seminal vesicle
Vas deferens cyst
Paramedian; nal vesicle
very rare; ampullary
dilatation of ipsilateral
region; dilatation of ipsilateral semi?· a-a-
Acquired Retention cyst
1-2 cm in diameter, may occur in any zone; fifth to sixth decades
Prostatic abscess
Fever, irregular shape, septations;
Parasitic
History of echinococcosis
Cystic
infection
degeneration
Carcinoma
with cystic components
Postoperative
changes
fifth to sixth decades
or schistosomiasis
(extremely rare)
Small cysts within area of benign prostatic hyperplasia Transmural from BPH
(BPH)
extension; stage T2 or smaller lesions are indistinguishable
History ofTUR (rare) or adenoma enucleation
"Enlarged" Seminal Vesicles
or peripherally. No contrast enhancement of the cyst wall is seen. Tubular ectasia shows dilated tubules in the mediastinal region. Epidermoids are well-defined, but may have variable densities depending on the cyst content. Hydrocele
Hydroceles are fluid collections between the layers of the tunica vaginalis of the testis and are congenital or acquired. Acquired types may have an idiopathic, inflammatory, traumatic, or radiogenic etiology. They occur less commonly in the spermatic cord. •
CT Morphology
Hydroceles show as masses of fluid attenuation around the testis, variable in size (Fig. 21.5). Thin septa can occasionally be seen. Increased density of the fluid collections may hint towards complications such as infection or hemorrhage. Hydroceles may be a first manifestation of testicular carcinoma.
Epididymal Cyst and Spermatocele
Epididymal cysts are dilatations of tubules. They can be multiple, occur anywhere in the epididymis, and range in size from millimeters to several centimeters. Spermatoceles are retention cysts in
the head of the epididymis that may be multilocular. They are generally less than 1 cm in diameter and have a firm consistency. Both lesions can be bilateral and diagnosed by palpation. •
CT Morphology
Epididymal cysts present as well-defined cystic structures, usually in the epididymal head. Attenuation depends on the content and may be higher in Spermatoceles. There is no contrast enhancement of the cyst wall. Imaging cannot usually differentiate between these two lesions. Varicocele
Varicoceles are dilated and elongated veins involving the cremasteric or pampiniform plexus. They can be idiopathic or result from incompetent valves in the internal spermatic vein. They occur in 10% of healthy and 50% of subfertile men and can be bilateral. Because the left spermatic vein opens into the renal vein, varicoceles almost always occur on the left side. Varicoceles can be diagnosed generally by palpation and are verified by color Doppler ultrasound. They can be an incidental finding on CT. •
CT Morphology
Varicoceles appear as convoluted intrascrotal masses of soft-tissue density, usually located along the spermatic cord. They have the same CT attenuation as blood (isoattenuating to other venous structures) and are best appreciated on postcontrastscans (cf. Fig. 21.5). When a right-sided varicocele is found, it is necessary to exclude obstruction of the spermatic vein by a tumor (usually renal).
"Enlarged" Seminal Vesicles
Fig.21.S Incidental finding ofa hydrocele (arrow head) and dilated veins due to a varicocele (arrow) in a patient who underwent CT for ajoint evaluation of suspected pulmonary embolism and deep venous thrombosis.
The size of the seminal vesicles is highly variable. Symmetrical" enlargement" of the seminal vesicles is usually caused by fluid retention and has no pathologic significance. Larger retention cysts of the seminal vesicles are rare and are best demonstrated by endorectal MRI. Differentiation is required from malignant invasion of the seminal vesicles, especially by prostatic carcinoma.
27 Male Pelvis
CT Morphology
Seminal vesicles that are enlarged due to fluid retention appear nonhomogeneous and often show nodular margins caused by small retention cysts. When large. these cysts show fluid attenuation and may display partial volume effects. The angle between the bladder and seminal vesicles is not obliterated. Malignant invasion of the seminal vesicles typically leads to obliteration of the bladderseminal vesicle angle. Tumor invasion often causes a more homogeneous structure and a soft tissue attenuation similar to that of the prostate. The volume of the seminal vesicles is a useful indicator of potential invasion by prostatic carcinoma. If the vesicles are symmetrical and have a total volume less than 15cm3• there is less than a 20% likelihood that they are involved by prostatic carcinoma.
Malignant
Benign Scrotal Neoplasms
Adenomatoid tumor is the commonest
extratesticular neoplasm. Most arise in the epididymis or spermatic cord in patients between 25 and 50 years old. The tumors are either round and welldefined solid masses or more plaque-like and ill-defined. Lipomas are occasionally seen in the spermatic cord, with characteristic low attenuation. Fibrous pseudotumors are rare lesions of the tunica albuginea, tunica vaginalis, or epididymis. They can be lobulated soft-tissue masses or show circumferential thickening of the tunica. Most are secondary to inflammation or trauma. In 50% there are associated hydroceles.
Tumors
CT is not used as a imaging procedure for the detection of testicular or prostate cancer because of its inability to differentiate between benign and malignant lesions. However. CT is used for the Nand M staging of tumors of the testis. scrotum. spermatic cord. and epididymis. In patients with prostate cancer CT is used in advanced cases to determine the extent of invasion of adjacent organs. For distant metastatic lesions in abdomen and chest. it is more commonly used than MRI.
Testicular Cancer Testicular cancers make up 5% of all male genitourinary tumors and have an incidence of 12 per 100000 men. Most patients with non-seminomatous germ-cell tumors are 15-25 years old. those with seminomas 25-35 years. The tumor also occurs in children 5-10 years of age. Patients usually present with an enlarged testis or a painless palpable mass. There is a strong association with cryptorchidism -10% of testicular cancers occur in undescended testes. Histologically, 95% are of germ-cell origin and are subdivided in seminoma (40%) and the non-seminomatous (60%) tumors like embry-
onal carcinoma (20%), teratoma (carcinoma) (20%), mixed cell (20%). and the rare choriocarcinoma. The other 5% are testicular stromal tumors like Sertoli cell, Leydig cell, and mesenchymal cell carcinomas. Tumors spread by direct extension, and lymphatic and hematogenous metastasis. Germcell tumors metastasize primarily lymphogenously in a predictable sequence that starts with the intra-aortocaval and para-aortic nodes at the renal hilum and descends to the aortic bifurcation. Generally the ipsilateral (80-85%) or ipsi- and contralateral (15-20%) nodes are affected. Primary contralateral metastases are uncommon from the right testis but almost never occur from the left testis. Pelvic node extension signifies infiltration of the epididymis or spermatic cord and inguinal nodes require infiltration of the scrotal skin or prior surgery to scrotum or groin. Distant metastases occur in the lung or liver and are an early route of spread for choriocarcinoma. The tumor is diagnosed by palpation, tumor markers (LDH,AFP, (3-hCG) and (color Doppler) ultrasound. MRI may be used in equivocal cases. Histological confirmation and primaryT-staging using the TNMS classification (Table21.5; tumor markers included) is done at orchiectomy, while
Testku/ar Cancer Table
21.5
TNMS
staging
of germinal
testicular
carci-
- T stage
is classified
after radical
noma (UICC, 1997) Primary
Tumor
orchiectomy PT x
(p T) - otherwise
Primary radical
pTO
tumor
cannot
orchiectomy
T x is used be assessed
has been
No evidence of primary tissue in testis)
pTis
Carcinoma
or no
performed
tumor
(e.g. scar
in situ (intratubular
germ cell
neoplasia) p T1
Tumor limited to testis and epididymis without vascular/lymphatic invasion; tumor may invade tunica albuginea but not tunica vaginalis
pT2
Tumor limited to testis and epididymis with vascular/lymphatic invasion; tumor extending through tunica albuginea with involve-
pT3
Tumor invades spermatic without vascular/lymphatic
pT 4
Tumor invades scrotum with or without vascular/lymphatic invasion
ment of tunica
Regional N1
Lymph
vaginalis cord with cr invasion
Table
with a lymph node mass 2 cm or
mension (para-aortic, preaortic, retro-aortic, interaortocaval, prevacal, para caval, retrocaval, along spermatic vein) Metastasis with a lymph node mass more than 2 cm but no more than 5 cm in greatest dimension or multiple lymph nodes, any mass more than 2 cm but not more than 5 cm in greatest N3
Metastasis
dimension
with a lymph
than 5 cm in greatest Distant
node mass
more
dimension
Metastases
MI
Nonregional lymph node metastases supraclavicular nodes) or pulmonary metastasis
(e.g.,
M2
Extrapulmonary
(bone,
visceral
metastases
IA IB IS IIA II B
T1 T2-4
lie
anyT
IDA IIIB
me
Table 21.7 cell tumors
51
Tumor
AFP
<1(0)
LDH 1.5-10 fold increased;
LDH >10 fold increased;
AFP
hCG>50000ml
AFP
ng/ml;
Intrapelvic and inguinal lymph nodes are considered regional after scrotal or inguinal surgery
anyT anyT
anyT
anyN
anyT
anyN
anyT
anyN
IGCCCG (1997)
MO MO MO MO MO MO M1 M1 M1
prognostic
classification
SO SO
anyS SO-I SO-! SO-I SO-I 52 53 for germ-
T esticularlretroperitoneal primary Pulmonary metastases AF P <1 000 ng/ml; LDH <1.5fold increased HCG <5000mIU/ml
Any primary localization Nonpulmonary distant metastases AnyLDH,AFP
Testicularlretroperitoneal primary Pulmonary metastases AFP 1000-10 (0) ng/ml; LDH 1.5-10 fold increased
and HCG
HCG5000-50000m1U/ml
ng/ml;
U/ml
NO NO NO N1 N2 N3
anyT
localization Pulmonary metastases Any LDH, AFP and HCG
icco
>10 (0)
(UICC,
Seminoma
10000 ng/ml; hCG 5000-50 000 mlU/ml 53
classification
Any primary
hCG<5000mIU/ml 52
TNMS
Prognosis
Intermediate
Markers
LDH <1.5 fold increased;
grouping
Good
liver, brain) Serum
Stage
li'Stage.
less in greatest dimension or multiple Iymi=h nodes, none more than 2 cm in greatest di-
N2
21.6
1997)
Nodes
Metastasis
lymph nodes and distant metastases are staged by CT ofthe chest and abdomen. CT is also used in follow-up and for radiotherapy planning. Prognosis depends on stage group (Table 21.6) and histology at presentation. Prognostic risk profiles. like the one from the International Germ Cell Cancer Collaborative Group (IGCCCG). are of benefit in selecting treatment strategies and follow-up (Table 21.7). Primary therapy for seminoma consists of orchiectomy. Depending on the stage. surveillance. radiotherapy for lymph nodes andlor chemotherapy are added. For non-seminomas orchiectomy is performed initially, with (nerve-sparing) retroperitoneal lymph node dissection only in high-risk patients. Chemotherapy is added for more ad-
Poor
Mediastinal primary Nonpulmonary distant metastases AFP >10 000 ng/ml;
HCG>50000m1U/ml LDH > 10 fold increased
735
27
Male Pelvis
Fig. 21.6 Metastatic testicular tumors, (a) Initial metastasis from a left-sided choriocarcinoma. CT shows a small paraaortocaval lymph node, for which metastatic involvement was confirmed histologically. (b) Metastases from teratocarcinoma appear as large conglomerate mass containing areas of cystic liquefaction.
b
a
vanced tumors. Bulky tumors may require additionallymph node dissection after radiotherapy or chemotherapy. Surveillance requires strict follow-up by tumor markers and CT imaging. •
CT Morphology
The primary tumor is rarely imaged by CT. When the testes are in siru and lie in the scanning field, seminoma presents in the testis as a homogeneous hypo attenuating mass, which may be lobulated. Central cystic necrosis can be seen and the tumor is hypovascular compared to normal testicular tissue. Non-seminomas are more heterogeneous, with necrosis and hemorrhage, and are frequently ill-defined. Para-aortic lymph nodes with a short axis diameter larger than 10mm are suspect for metastatic involvement. Usually there is no difficulty in recognizing the malignant nature of seminomas, which are associated with rapid and pronounced lymph node enlargement. In the case of non-seminomatous tumors, any para-aortic nodal enlargement is considered suspicious for metastasis (Fig.21.6a), and recently a cut-off value as low as 4 mm has been suggested. Teratocarcinomas tend to produce metastases that contain areas of cystic liquefaction (Fig.21.6b).
Testicular Lymphoma and Metastases
In patients more than 50 years of age, lymphoma or metastatic disease is more common than primary testicular cancer. Lymphoma in the testis may be primary extranodal disease or part of systemic involvement. It is the commonest testicular malignancy in this age group. When poorly differentiated, it is often bilateral. Leukemic involvementmay be the first sign of extramedullary disease. Metastases are the second most frequent neoplasms in older patients and are usually multiple and bilateral. The most usual primaries are lung, melanoma, kidney, prostate, and GI malignancies. CT is not used for primary diagnosis. •
CT Morphology
If discovered by chance, testicular lymphoma may show one or more focal hypoattenuating lesions uni - or bilaterally or an enlarged hypoattenuating testis. Leukemia cannot usually be differentiated from lymphoma. Metastases usually show multiple hypoattenuating solid masses. They are often bilateral and the testes may be enlarged. Prostate Cancer
Following treatment in patients with seminoma, residual lymph nodes up to 2cm in diameter are considered nonviable if their size remains unchanged on follow-up scans. CT is also used to detect pulmonary metastases. Except for calcified foci, all pulmonary nodules should be considered suspicious, regardless of their size.
Prostate cancer is the commonest malignancy in men, with an incidence of30-90 per 100 000. The incidence increases with age and most patients present between 65 and 75 years of age. Risk factors include family history, smoking and other environmental factors. African Americans are at higher risk than other groups. Screening in men over 50 (USA) has led to an increased detection with more organ-confined disease and less mor-
Prostate Cancer tality. Management is under debate because of the variability of biologic aggressiveness of the tumor; 75 70 of men over 80 have foci of carcinoma in the prostate, but only 20% of men develop clinical prostate cancer during their lives. More than 95% of the tumors are adenocarcinomas. These can originate in the peripheral (75%), transitional (15%), and central (10%) zones. The other 5% consists of a multitude of epithelial tumors (e.g., squamous cell, endometroid, signet -ring cell carcinomas) and nonepithelial tumors (e.g., carcinosarcoma, rhabdomyosarcoma). The histological specimens are graded with the Gleason score on a scale of 2 to 10, which correlates well with aggressiveness and is a key part of risk assessment. The tumor is often multifocal and spreads through direct extension in the neurovascular bundle and seminal vesicles, along the lymphatics to periprostatic and pelvic nodes, and hematogenously to bone, lung, and liver. Screening and clinical staging is performed using digital rectal examina tion (DRE) and serum prostate specific antigen (PSA) measurements. Clinical staging uses the TNM or American Urological Association (AUA-Whitmore- Jewett) staging systems (Table 21.8). Palpable lesions at DRE or abnormal PSA values (» 2.6-4.0 ng/ml) require
(color Doppler) transrectal ultrasound (TRUS) with directed or screening (6-10) core biopsies. Risk analyses based on clinical stage, number of positive biopsies, Gleason score, and PSA (with refinements like PSA-density, % free PSA and complexed PSA) can predict pathological stage and are used to tailor further investigation and treatment (e.g., Partin's tables). Prognosis is based on tumor grade and stage at presentation. Definitive treatment is not performed if the patient's life expectancy is less than 10 years. Microscopic, low-grade lesions are generally followed by watchful waiting, while patients at high risk for metastasis should undergo work-up before treatment is considered. Radical prostatectomy or interstitial brachytherapy with radioactive seeds is indicated in patients with organ-confined disease. Patients with stage T3 disease or higher are not considered surgical candidates and are treated primarily by radiotherapy, with or without hormonal (androgen ablation) therapy (Table 21.9). The role of imaging is to aid clinical risk assessment and to select patients for definitive treatment. In local stagingTRUS does not perform significantly better than clinical staging with DRE, and its major roles are volumetry and guldance of biopsy. Endorectal MRI is probably
Table 21.8 TNM and ALIA (modified Whitmore-Jewett) staging of prostatic adenocarcinoma (UICC,1997) ITIMM SB ..
I
AUA
Description
TO
No evidence of a primary tumor
Tis
Carcinoma in situ (prostatic intra-epithelial neoplasia)
T1
Clinically inapparent tumor not palpable nor visible by imaging Tumor incidental finding in 5% or less of tissue resected Tumor incidental finding in more than 5% of tissue resected Tumor identified by needle biopsy (elevated PSA)
T1a Tib T1c
T2 T2a T2b T3 T3a
T3b
Al
A2
BO
B2
Tumor confined within prostate Tumor involves one lobe Tumor involves both lobes
C2 C3
Tumor extends through prostatic capsule Extracapsular extension (unilateral or bilateral) Tumor invades seminal vesicle(s)
BI
T4
Tumor is fixed or invades adjacent structures other than seminal vesicles: bladder neck, external sphincter, rectum, levator muscles and/or pelvic wall
NI
D1
Regional lymph node metastasis (periprostatic, internal iliac, external iliac, obturator)
M1
D2
Distant metastasis present (nonregional lymph nodes, bone, lung, liver, brain)
737
27 Male Pelvis
the best technique for local staging (accuracy 75-90%), and is cost effective in patients at intermediate risk for extraprostatic spread. It is used for local staging, status of lymph nodes and also for in tersti tial brachytherapy planning. MR spectroscopic imaging (MRS!) can add functional information and will improve specificity in the near future. New possibilities like USPIO-MRI, Table 21.9 2000)
Treatment decisions by cancer stage (Yu,
1,Stage,_1!
Possible treatmenf'AAIA
1
Tl, Gleason2-4 T1 ,Gleason5-1 0 T2a, low/intermed risk T2a, high risk T2b T3-4 N1-4 M1
Watchful waiting Prostatectomy, XRT Prostatectomy, XRT Prostatectomy or XRT Prostatectomy or XRT XRT or hormonal XRT and hormonal paliiativeXRT, hormonal
XRT = radiotherapy - interstitial brachytherapy or external beam radiation
Table 21.10
a
radioimmunoscintigraphy (ProstaScint) and PET are currently under investigation. CT does not playa primary role in local staging. Its role is the evaluation of spread into adjacent organs and to rule out metastatic disease in lymph nodes, and other sites, in patients at high risk for extraprostatic disease. Helical or multislice CT may improve local staging to some extent, but inherent lack of soft-tissue contrast will continue to be a limiting factor. If a radical prostatectomy is planned, nodal stage needs confirmation by staging lymphadenectomy (Table 21.10). •
CT Morphology
CT findings are unreliable for detecting organconfined tumor (Fig. 21. 7a), especially in the deeper zones of the prostate. Peripheral tumors may show focal or diffuse enhancement after contrast medium. Invasion of the seminal vesicles typically leads to obliteration of the angle between the bladder and the seminal vesicles, asymmetric vesicular enlargement, and an increase in vesicular attenuation as normal secretions are re-
Risk stratification and diagnostic work-up in confirmed prostate cancer (yu, 2000)
Category
Description
Diagnostic work-up
Low risk
Gleason 2-4 and PSA <10 ng/ml and clinical T1 -2
No additional imaging
Intermediate risk
Gleason 2-4 and PSA 10-20 ng/ml and clinical T1-2 Gleason 5-7 and PSA <20 ng/ml and clinical T1 -2
MRI + MRSI, additional pelvic lymphadenectomy for T2 tumors
High risk
Gleason 8-10 and PSA >20 ng/ml or clinical T3-4
Bone scan, CT, immunscintigraphyfor metastatic spread evaluation. If negative, work-up as intermediate risk
b
Fig. 21.7 Prostatic carcinoma, (a) Tumor confined to the prostate, (b) Posterior extracapsular tumor extension on the left side (bladder catheter in place), (c) Inva-
c sion of the bladder by plaque-like tumor. There is complete infiltration of left seminal vesicle (soft-tissue density) and partial infiltration of the right seminal vesicle.
Tuberculosis
placed by tumor (Fig. 21.7 c). Nonvisualization of the fat plane between the bladder wall and seminal vesicle suggests incipient bladder invasion (Fig.21.7c). Stranding of the periprostatic fat combined with ill-defined prostatic margins signifies penetration of the capsule (Fig.21.7b). Evaluation of the prostatic apex is hampered by the adjacent pubic bone. It may be improved by thinsection multislice CT imaging with good quality multiplanar reformations in the coronal and sagittal planes. Perifocal inflammatory served after biopsy.
reactions may be ob-
Lymph nodes with a short axis size larger than 10 mrn are considered
Inflammatory
suspicious
CT sensitivity range is generally below 75% with specificity range of 75-98%. TNM staging requires a staging lymphadenectomy with histological confirmation. CT-guided fine needle aspiration biopsy may be a good and cost-effective alternative with high specificity (over 95%). The obturator, internal iliac, external iliac, and presacral groups are most commonly involved. The common iliac, para-aortic, or even mediastinal nodes are only involved in advanced disease. Given the frequency of skeletal metastases from prostatic carcinoma, CT images with bone window settings are essential. Bony metastases typically incite an osteoblastic reaction (80%) marked by increased bone density, usually with irregular borders. Some metastases are purely lytic (5 %) or mixed osteolytic and osteoblastic (15%).
for malignancy.
Processes
The diagnosis of inflammatory diseases of the male pelvic organs is based on clinical. laboratory, bacteriologic, and ultrasound findings. MRI may be added in puzzling cases. Inflammatory disorders ofthe testes are not evaluated with CT.
Prostatitis Prostatitis is common problem in men younger than 50 years. Acute prostatitis is a retrograde bacterial infection caused by Escherichia coli. Klebsiella, Proteus or Pseudomonas sp. with clinical findings of fever, voiding problems and pain. Prostatic abscess formation is rare and more often seen in diabetics, dialysis patients, or immunocompromised patients. Chronic prostatitis can be bacterial or inflammatory and presents with aspecific symptoms. In all types the seminal vesicles may become involved. Clinical diagnosis rests on DRE, bacterial cultures, or biopsies. Transrectal ultrasound is helpful in the diagnosis and treatment of abscess. CT does not add new information in acute or chronic prostatitis, but is useful in the diagnosis of a prostatic abscesses.
CT Morphology
Acute prostatitis is not clearly distinguishable from a normal prostate or from benign prostatic hyperplasia on CT scans. It usually appears as a symmetric enlargement of the gland, which has approximately the same attenuation as normal or adenomatous prostatic tissue. Chronic prostatitis is usually associated with a small prostate because of scarring and fibrosis of the glandular tissue. Abscesses or foci of liquefaction appear on contrast-enhanced CT as irregular hypoattenuating areas that may show peripheral rim enhancement. The abscess can extend through the pelvic diaphragm into the base of the penis.
Tuberculosis Urogenital tuberculosis may originate in the prostate (granulomatous prostatitis), from which it may ascend to involve the bladder or seminal vesicles or descend to involve the epididymis. CT Morphology
Granulomatous prostatitis may show focal hypoattenuatingareas in the outer gland (Fig. 21.8 a). Concomitant retroperitoneal abscesses are common (Fig.21.8b). Healed urogenital tuberculosis
739
27 Male Pelvis
can leave large areas of amorphous calcifications in the prostate and seminal vesicles. Fournier's Gangrene
Fournier's gangrene is a form of necrotizing fasciitis that usually arises from the scrotum in males and may spread aggressively to involve the perineum, perianal region, abdominal wall, and penis. The differential diagnosis includes erysipelas, gas gangrene, pyoderma gangrenosum, and synergistic necrotizing cellulitis. Treatment consists of immediate radical debridement of the necrotic tissues. CT is the modality of choice for determining the extent of the inflammation. •
CT Morphology
CT shows extensive gas in the soft tissues and stranding of the subcutaneous fat due to the gangrenous spread of inflammation. Postoperative Changes
In a staging examination performed immediately after orchiectomy, the affected side is recognized by the presence of streaky densities and air or fluid in the inguinal soft tissues. These residual findings are no longer seen two to four weeks later. After prostatectomy, the bladder typically occupies a retropubic position and the levator sling is clearly visualized. A moderate amount of
a
b
Fig. 21.8 (a) CT following radical prostatectomy shows atypical band of scar tissue that may be mistaken for the resected seminal vesicles, (b, c) After TUR of the pros-
scar tissue is consistently visible between the bladder and rectum on follow-up CT after radical prostatectomy. Scars can mimic normal seminal vesicles and recurrent tumor and sensitivity ofCT for detection of recurrence is low. Scans taken after the enucleation or transurethral resection (TUR) of prostatic adenoma may demonstrate a hypo attenuating area at the resection site, representing the surgical defect, Coarse calcifications maybe observed (Fig. 21.2b). Following lymphadenectomy, surgical clips are visible along the large retroperitoneal vessels. Lymphoceles may develop, appearing on CT scans as smooth, homogeneous masses of water attenuation, frequently lying extraperitoneally along the pelvic wall. They also may present along the vessels as pseudonodes, with typical low attenuation, that mimic necrotic lymph nodes. The diagnosis is established by chemical analysis of the aspirated fluid. Knowledge of changes in the irradiated pelvis is important to allow for differentiation from recurrent neoplasms. The extent of changes is dependent on the applied radiation dose and the volume of tissue irradiated. Acute radiation leads to interstitial edema and congestion (increased endothelial permeability), while chronic changes after radiation include ischemia, strictures, and fibrosis. The perirectal fascia and the bladder wall are symmetrically thickened, the presacral space is usually widened. These changes are mostly stable after "12weeks. In rare cases, enterovesical fistulae may develop as a complication after pelvic irradiation.
c tate a defect in the prostate may present as a tumor -like lesion or a cyst in the prostate (b). In the excretory phase, the cyst fills with contrast-enhanced urine (c).
22 Lymphatic System A. J. van der Molen, C. Schaefer-Prokop,
Anatomy
A. Leppert
_
Cervical Lymph Nodes (p. 742) Thoracic Lymph Nodes (p. 743) Upper Abdominal Lymph Nodes (p. 744) Lumbar Retroperitoneal and Pelvic Lymph Nodes (p. 746) Examination Technique Criteria for Malignancy Neck(p. 749) Mediastinum (p. 750) Abdomen and Pelvis (p. 750) Differential Diagnostic Considerations Pitfalls
1,1
Benign lymph
Node Diseases
Tuberculosis (p. 753) Histoplasmosis (p. 754) Other Inflammatory Diseases (p. 754) Cystic Fibrosis (p.754) Sarcoidosis(p.754) Silicosis(p.755) Amyloidosis (p. 755) Malignant
lymphoma
Lymph Node Involvement (p. 757) Organ Involvement (p. 757)
(p. 751)
22 Lymphatic System Cross-sectional imaging modalities have all but replaced conventional lymphography in examinations of the lymph nodes (Table 22.1). When cross-sectional modalities are used. the diagnostic evaluation is based primarily on the size of the lymph nodes and to a lesser degree on morphologic criteria. In doubtful cases the diagnosis should be established by biopsy. Only a few cases currently require exploratory laparotomy or mediastinoscopy. High image quality and reproducibility make CT the primary modality for investigating the mediastinal and abdominal lymph nodes. CT can also image the cervical. axillary. and inguinal nodes. but (color Doppler) ultrasound is preferred in these regions because it costs less and provides additional morphologic criteria that aid in the differentiation benign versus malignant ofthese superficial nodes.
Table 22.1
Indications for CT in lymJ=hnode staging
IndicationCT
Region
examined
Head and neck tumors
Neck-',
chest'
Bronchogenic Esophageal
carcinoma carcinoma
Chest (only NSCLC) Chest, upper abdomen
Renal cell carcinoma
Upper abdomen
Gynecologictumors
Abdomen
Prostatic
Abdomen
carcinoma
Testicular
tumors
Abdomen
Germ cell tumors
Abdomen
Malignant
Neck3,b,
melanoma
byCT
chest,
abdomen
Hodgkin's lymphoma, NHL a = ultrasound b = indication NSCLC
=
Neck3,b,chest, abdomen
preferred depends on location
non-small
cell lung cancer
Anatomy Normal lymph nodes have the shape of an ellipse whose long axis is directed parallel to the anatomic structures (mostly vascular) along which the nodes are distributed. The ratio of the longitudinal (maximum) diameter to the transverse (minimum) diameter of a lymph node. called the ur ratio. is normally A 2. Because most vessels run perpendicular to the scan plane, axial CT slices do not display the maximum diameter of lymph nodes. Axial CT, therefore, is not able to provide a reliable measure of the ur ratio. In most regions the minimum diameter of normal lymph nodes is less than 1 em. Spiral CT scanning with overlapping reconstruction intervals is needed to determine the longitudinal diameter of a lymph node and to calculate its LIT ratio. In addition, the minimum diameter more closely correlates with the volume of a lymph node and can be measured more accurately in follow-up examinations because it is less dependent on the scan plane. The CT attenuation oflymph nodes on noncontrast CT scans is equivalent to those of other soft-tissue structures. The node hilum can sometimes be seen to contain a small amount of fat. Normal lymph nodes display moderate, homogeneous enhancement following the intravenous injection of contrast material.
Cervical Lymph Nodes Approximately 300 of the 800 lymph nodes of the body are located in the neck. Although there are many classifications, the most widely used are the ones of the Union Internationale Centre Ie Cancer (UICC, 1997), theAmericanJoint Committee on Cancer (AlCC, 1997) and the American Academy of Otolaryngology-Head Neck Surgery (AAO-HNS, 1991; revised 1998) (Table 22.2). Based on these clinical classifications, an imaging-based nodal classification was introduced by the otolaryngological and radiological communities very recently (1999/2000). This classification proved to give consistent results in early studies and is outlined in Table 22.3 and Fig. 22.1. The jugulodigastric nodes at the mandibular angle, which belong to the upper jugular group (level II), are larger than most other cervical nodes. The jugular chain of nodes (levels II-IV) receives drainage from almost all other lymph node stations of the nasopharynx, oral floor, and facial region. As a result, the enlargement of jugular lymph nodes is as common as it is nonspecific.
Tboracic Lymph Nodes Table 22.2
Lymph node stations in the neck (U1CC and AAO-HNS)
UICC Classification Submental
2
{1997}
1\3 Superior jugular
4
{1998}
Common primary tumor sites
Levell A
Oral cavity, lower lip, tongue
nodes
Level I B
Orbit, submandibular gland, oral cavity, lips, teeth, tongue, anterior nose
nodes
Level II AlB
Tonsils, posterior oral cavity, oropharynx, supraglottic larynx
nodes
Submandibular
AAO-HIMS
and common tumor sites
Middle jugular
nodes
Level III
Glottic and subglottic
5 Inferior jugular
nodes
Level IV AlB
Subglottic larynx, thyroid, breast, lung, stomach
Level V AlB
Nasopharynx, skin of neck, occipital scalp; advanced head and neck tumors
6
Dorsal cervical nodes
7
Supraclavicular
8
Prelaryngeal nodes
9
nodes
and paratracheal
Retropharyngeal
larynx,
Upper chest wall, upper abdomen, neck tumors Level VI
nodes
Subglottic larynx, hypopharynx, esophagus Posterior nose, sinuses, cervical esophagus Frontoparietal external ear
11 Buccal nodes
Face, parotids, nose, eyelids and occipital
advanced thyroid,
nasopharynx,
10 Parotid nodes
12 Retroauricular
hypopharynx
cervical and mid-esophagus,
head and
cervical oropharynx,
scalp, face, eyelids, parotis,
Parietal and occipital scalp
nodes
Thoracic Lymph Nodes
Submandibular gland
--'tt-\.~
Left caroti d ---+-----\!d--I-+--fJ artery Cricoid ---+--+--\-,cartilage Internal ----:.. jugular vein Top of --P+'---"Lmanubrium
Blue Grey
= =
Margins of nodal classification levels Anatomical structures
Fig. 22.1 Lymph nodes in the head and neck region. The numerical labels are defined in Tables 22.2 and 22.3.
Thoracic lymph nodes are usually referred to in visceral and parietal groups. The visceral nodes include the intrapulmonary and mediastinal nodes that primarily drain the lung and mediastinal structures, while the parietal nodes include the internal mammary, paracardiac, and intercostal nodes that drain the chest wall. Most of the parietal nodes are not normally visible on CT. The axillary lymph nodes drain the arm, the lateral chest wall, and the lateral and central portions of the breast. The axillary nodes are divided into three groups based on their relation to the minor pectoral muscle (medial to, level with, and lateral to the muscle) . Intrathoracic (visceral) nodes are usually described as belonging to specific node groups. The average number of mediastinal nodes is around 65. Two classification schemes were commonly used, the American Thoracic Society mapping scheme and the scheme from the Amerlcan.loint Committee on Cancer (AlCC). In "1996, these two schemes were unified and the new classification has been adopted in both the AlCC and the UICC (TNM) classifications (Table 22.4; Fig. 22.2).
743
urruuauio rassa] a41 'paAOldwa S! D a:JlIsq]nw uouoas-unp sssjun sueos D uo paptdap NU AHPWJOU aJP sapou HPWS asaljl 'SJpOU IVUJPO -npl.lJd PUP 'JUJLUV'il!1 toueponpoiodeu 'J!JDJJJ -UVdPJd alp Aq paUIPJp a.lp umuaponp PUP 'pnp aIIq 'seaoued alp aIIllM 'SJPOU (JPOdJI/O.IJSO'il) JPJso'il!.IJd all. Aq pauierp SI tpeurors alII
'sdnOJ'il :J!la.uasaw pup IPAP:J!JpJOP-PJPd 'IPJnD -OJ.aJ '.uawp'illl IPuaponpo.pdaq 'InlOdpad ':J!Ualdspad 'Jp.sp'ilpad alp a.lp uauropqe raddn all. UI sdnors IPPOU .I0rPllI aqi 'punoj aJP sapou lldlllAI D£l auros slAlad pUP uauropqe alp UI
saursu JlwO.PUP Jlay. Aq oi paJJaJaJ aJP sapou leplJJadns pup Pl.OJPd sapa~p Pl.OJPJ IPUJalUIor JPlpaw pue aseq IIn~s JO WJ l U1Y.1M
lea5uAJeydoJla~
squ oi IPlpaw pue aAoqe pue AJalJe PI40JPJ YJPa olleJalel 'ueos lelxP YJPa uo uaas SPaplAPp JO laAal oa lepnPJ JO N IPpnPJ 'umuqnueui
ulaA J1JPydaJoYJeJq JO laAal oi JO do. JO laAal MOlaq sapa.JP Pl.OJPJ uaaM.a8
JPlnJ1APpPJdns sapou [euuseipaui Jopadns
umuqnueur jo do. o. YJJP P10D1D JO urfireui JaMol WOJJsapa.~e Pl.OJPJ uaa~a8
sapou leJaJS1/\
apsnw aualeJs rouaiue Ul5Jew leJaleloJa.sod YllMapsnw P10.sPwoPlapOUJalS JO ~Jeq 5ulPaUUOJ aUll or rouaisod-ueos lelxe YJea UOaplA81nJO laAal or YJJe P10J1D JO Ul5Jew JaMol WOJj apsnui ptciseuroptapourars JO ~Jeq or rouaisod '4JJe P'OJ1D JO Ul5Jew JaMol or aseq IImls WOJj apsnur srnzadan JO afipa .iouaiue oi rouaiue :apsnw aualeJs rouaiue ay. JO uif -reui leJalelOJalSod ay. Y.1Mapsnw Pl0lsPwoPIapOUJalS JO ~JPq 5ulPauuOJ aUll or rouaisod UPJSlelxe y~ea uo aplAep JO laAal or YJJP P1OJ1DJO Ul5JPWJaMol WOJJJO .prorseuropjepourais JO ~Jeq oi rouaisod-tpre P10J1DJO Ul5JPWJaMol or aseq IIn~s WOJj
sapou IPJ1AJaJrpSJOa
apsnw aualeJs rouaiue ay. JO ujfireui leJalPloJa.sod ay. Y.1Mapsnw pjoiseurop -iapourais JO speq 5ulPaUUOJ aUlI or rouaiue 'SapalJe Pl.OJeJ or JPJa.el 'aplAPp JO laAal oi YJJe P10J1D JO Ul5Jew JaMol WOJj
sapou JPln5nf MOl
sapsnw JP.se51P JO saillaq rouaiue oi leJaleloJalSOd sapsnw JPlSe51P JO saillaq .iouaiue JO SU15Jew leipaw uaa~a8 spuejfi Jelnqlpuewqns JO safipa rouaisod 5ulpaUUOJ aUll or rouaiue 'apsnw P10AyoIAWMOlaq pue auoq P10A4aAoqv
JS 11/\ 1/\
sapou /\ laAalJaMol
8/\
sapou /\ laAalJaddn
V/\
apsnw P10.sPwOPlapOUJa.s JO ~Jeq or rouaiue 'YJJP a5PIl.JPJ P10J -ID JO utfireur JaMol oi auoq P10AYJO ApOq JO U1DJPWJaMol WOJj uoueredas auejd leJ e Y.1MulaA Jeln5nf leUJalul JO JOpalSOd rouaisod Jl ulaA Jeln5nf leUJalul WOJJ ajqeredasuj 'ulaA Jeln5nf leuJa.ul o.lelpaw pue 'leJa.el 'JOpalUV ptoiseuropjsp -OUJa.sJO ~Jeq oi rouaiue pue puejfi rejnqipueurqns JO ~Jeq or J011 -aisod 'auoq P10AYay. JO ApOq JO Ul5Jew JaMol oi aseq IIn~s WOJj
d~
sapou Jeln5n[ P1VII
/\
/\1 III
sapou AJOSSaJJe leu Ids tladdn
8 II
sapou Jeln5n~ leuJa.ul Jaddn
VII II
sapou rejnqjpueurqng sapou leluawqns
U01.d!J)sila
-OW aYl oi salwel!wlS ay. a.oN) '(DDDl :666L 'WOS) sapou leJ1AJaJ JOJ uoueoujssep
81 VI I
dnOJDlepON
lilAiI,
(uoueourssep SNH paseq-Buifieun ["ll alqel
Upper Abdominol Lymph Nodes Table 22.4
Station
AJCCjUICC 1996 classification for mediastinal and pulmonary lymph nodes (Mountain,
1997)
Description
N2 nodes:
all N2 nodes lie within the mediastinal pleural envelope
1 R/L
Highest mediastinal nodes
above line at upper level of brachiocephalic
2 R/L
Upper paratracheal nodes
above line tangential to upper margin of aortic arch
3A
Prevascular nodes
3P
Retrotracheal nodes
4R
Right lower paratracheal nodes Left lower paratracheal nodes
right of midline - above line through R main bronchus at upper margin RUL bronchus left of midline - above line through L main bronchus at upper margin LUL bronchus
Subaortic nodes
lateral to aorta or L pulmonary artery - proximal to first branch of L pulmonary artery anterior and lateral to ascending aorta - above line tangential to upper margin of aortic arch
4L
vein crossing the trachea
6
Para-aortic nodes
7
Subcarinal nodes
caudal to carina but not associated with lower lobe arteries or bronchi
8
Para-esophageal nodes
9 R/L
Pulmonary ligament nodes
adjacent to wall of esophagus, left or right of the midline within pulmonary ligament-including those along wall inferior pulmonary vein
Nl nodes:
all N1 nodes lie distal to mediastinal pleural reflection and in visceral pleura
10 R/L
Hilar nodes
11 R/L
Interlobar nodes
directly distal to mediastinal pleural envelope, R along intermediate bronchus between the lobar bronchi
12 R/L
Lobar nodes
adjacent to the distal lobar bronchi
13 R/L 14 R/L
Segmental nodes Subseqrneutal nodes
adjacent to the segmental bronchi around the subsegmen_t_a_1 b_r_o_n_ch_i
--'
R = right, L = left, RUL = right upper lobe, LUL = left upper lobe
is an important pathway between the nodes of the stomach and the deeper nodes that drain the liver and bile ducts. These omental nodes lie within the hepatoduodenal ligament, and the node of the foramen of Winslow is frequently involved and easily seen on CT directly posterior to the portal vein. The hepatic lymph nodes are located close to the hepatic artery in the porta hepatis. The liver is also drained by hepatoduodenal ligament lymph nodes, the internal mammary lymph nodes (Fig. 22.3), and lymph nodes along the anterior and posterior abdominal walls. The central lymph node station for the upper abdominal organs is the para-aortic celiac group of nodes, which also receive afferents from the distal esophagus. The retrocrural lymph nodes connect the nodal groups of the posterior mediastinum with the para-aortocaval lymph nodes.
Table 22.5 Size criteria for mediastinal lymph nodes (adapted for the AJCCjUICC 1996 classification from the ATS 1985 node map; Glazer 1985)
Station
Description
Normal size
2 R/L
Upper para tracheal nodes
$7mm
4 R/L 5
Lower paratracheal nodes
$10mm
Subaortic nodes
$9mm
6
Paraortic nodes
$8mm
7
Subcarinal nodes
$11mm
8R
R para-esophageal nodes
$10mm
8L 10 R
L para-esophageal nodes R hilar nodes
$7mm
10 L
L hilar nodes
$7mm
$10mm
The mesenteric lymph nodes that drain the intestine are arranged in four stations. Normally the first three stations are not depicted on CT
745
746
22 Lymphatic System Fig. 22.2 Lymph nodes in the mediastinum (AJCC-UICC 1996). The numerical labels are defined in Table 22.4.
Trachea ---+-Brach io- ----l---------o',_____+__ cephalic artery
C2:======1rJ
--+t-----=T~~ Bronchus----~,.__jf'7H_ Esophagus
Pulmonary artery
-------rr;:-l=-~-+;::
>~-~~---'--;-
Inferior pulmonary Ligament
a
r----
Bronch us ------I=.X:::l
,::--:::.-v-
--;'Ft'\~~-
Ligamentum arteriosum
PuImon ary artery
~---
Bronchus
b
scans. The group of nodes distributed along the superior and inferior mesenteric arteries is the terminal station for lymphatic drainage from the large and small intestine (Table 22.6). Lumbar Retroperitoneal and Pelvic Lymph
Nodes
Fig. 22.3 Enlarged paracardiac lymph node (of the internal thoracic chain). The anterior, pericapsular portions of the liver are among the structures that are drained by the lymph nodes in this region. This group of nodes should be closely scrutinized in patients with hepatic masses.
The lumbar retroperitoneal lymph nodes consist of five·groups: paracaval, precaval, para-aortic, preaortic, and interaortocaval (Fig. 22.4). The paracaval and para-aortic nodes, which drain much of the lymph from the lower extremities, also serve as the primary nodal station for the kidneys, adrenals, ovaries, and testes (at the level of the origin of the renal vessels) and as a
Lumbar Retroperitoneal Table 22.6
Lymph node stations
lymph node station Retrocrural Gastrohepatic ligament
in the abdomen
(size limits according
Primary tumor/site
lymphoma
os 6 mm
Gallbladder. lung. breast
bile ducts.
Pancreaticoduodenal
Lymphoma. breast
head of pancreas.
Splenic hilum
Lymphoma. leukemia. small intestine. colon. transverse colon
Para-aortocaval
Mesenteric
and retroperitoneal and celiac
Normal
Stomach (lesser curvature). distal esophagus. pancreas. melanoma. colon. breast
Porta hepatis
High preaortic
et al., 1991)
of recurrence size
Lung. mesothelioma. and hepatoduodenal
to Dorfman
Lymphoma.
liver. stomach.
Small intestine.
relay station conveying lymph from the uterine corpus and fallopian tubes. The pelvic nodes are divided into the external iliac. internal iliac. and common iliac groups and each group is subdivided into several chains. The medial chain of the external iliac
tumors.
pancreas.
colon. stomach.
kidney. ovary. testis.
All abdominal
and Pelvic Lymph Nodes
lymphoma. colon. lung.
ovary. ascending
uterine
cervix. prostate
distal esophagus
large intestine.
os 8 mm os 7 mm os 10 mm os 10 mm os 11 mm os 10 mm
lymphoma
os 10 mm
nodes (obturator nodes) has special significance in the detection of nodal metastases from prostatic. bladder. and cervical carcinoma. because these nodes are the primary sites of lymphogenous seeding from tumors in these organs (Table 22.7). The obturator nodes can be visualized by scanning on a level 2-3 ern above the acetabulum near the inner pelvic wall.
Table 22.7 Lymph node stations in the pelvis limits according to Vinnicombe et al., 1995) lymph node station
Primary tumor'
Normal size
Inguinal
Vulva. penis. distal vagina. distal rectum. anus
os10mm
Rectum.
os9mm
Common nodes
nodes
iliac
prostate
External nodes
iliac
Bladder. prostate. proximal vagina. uterus. ovary
os10mm
Internal nodes
iliac
Almost all pelvic organs
os7mm
Prostate. cervix
os8mm
Obturator
nodes
Para-aortic and caval nodes
Fig.22.4 Lymph nodes in the retroperitoneum: 1 celiac nodes. 2 + 3 mesenteric nodes. 4 paracaval nodes. 5 precaval nodes. 6 interaortocaval nodes. 7 preaortic nodes. 8 para-aortic nodes. 9 external iliac nodes. 10 internal iliac nodes.
(size
bladder.
Ovary. testis
os 11 mm
747
748
22 LymphaticSystem
Examination Technique
------------------------------------------~----~
The CT evaluation of lymph nodes is an integral part of tumor staging. In most cases, the imaging requirements for the primary tumor determine the scanning protocol (e.g, bronchogenic carcinoma).
Patient Preparation Adequate opacification of small bowel loops is important to differentiate bowel from lymph nodes in the small pelvis. Thus, the patients should start drinking the oral contrast medium early enough (at least 60 minutes before the exTable 22.8
I General
amination) to ensure ileum loops.
proper opacification
Acquisition Technique Multislice scanning with a narrow collimation may improve morphologic differentiation between benign and malignant nodes but at present this it is not generally recommended unless the primary tumor requires such as scanning protocol. For this reason, 2.5-3.75 mm collimation is a good compromise for 4-slice scanners that will allow for retrospective reconstruction of thinner
Suggested protocols for CT scanning of the lymphatic system
Oral contrast
1000-1500 ml, administered 60-90 min prior to examination (abdomen)
Patient position
Supine with elevated arms (chest, abdomen, pelvis, combined neck + chest) Supine with the arms adducted (neck)
Respiratory phase
Inspiration
Windowing
.---.
Noncontrast CT: Contrast-enhanced
--
CT:
W/L = 400/40 W/L = 400/60 (Neck 300/60)
-
I
Scan parameters
l-slice SC/TF/RI
I
Neck Chest Abdomen (incl. pelvis)" Neck + chestChest + abdomen" N~k to abdomen- d
3/5/3 ~ 5/10/5 t 7/12/6 t 5/10/5 t 7/12/6t two scans
Contrast injection
l-slice V/F/D
4-slice V+N/F/D
16-slice V+N/F/D
Neckc ChestAbdomen (incl. pelvis)" Neck + chestChest + abdomen" Neck to abdomen':"
70/2/20 70/2/20 120/2/60 70/2/20 120/2/40 two injections
70+ 50/2/20 70+ 50/2/20 100 + 50/2/60 70+ 50/2/20 120 + 50/2/50 120 + 50/2/50
50+30/2/30 50+30/2/30 100 + 50/2/60 50+ 30/2/30 120 + 50/2/60 120 + 50/2/50
4-slicea SC 2-H 2-3 2-3 2-3 2-3 2-3
16-slicea SC
axial SWIRl
1-1.5 t 1-1.5 1-1.5 t 1-1.5 1-1.5 1-1.5
3/2 5/4 5/4 3-5/2-4 5/4 3-5/2-4
sc = slice collimation
-
MPRb SWIRl
-
3/4 cor 3/4 cor 4/5 cor 3/4 cor 4/5 cor -2.-4/4-5 co.!:..,
(mm). TF= table feed (mrnjrotation), RI= recon. increment (mm). i t = scanning direction. SW = effective section width (mm). MPR= multiplanar reformats, axial = axial sections, cor= coronal, sag = sagittal CM = contrast material, Contrast concentration = 300 rnqjrnl iodine V = volume of CM (ml), N = volume of saline flush (ml) F= flow rate (mils), D = start delay, X = trigger region Pitch P= TFJ(NXSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners
a
b C
d
of
MPR are reconstructed from a "secondary raw data set" with SWIRl = 1-1.5/0.7 but are rarely necessary place one arm next to body and one above head to reduce artifacts for combined scanning of neck/chest consider interrupting the scan below the kidney for 30-60 s to improve opacification of pelvic vessels
sections if coronal images are required for better morphological analysis. With 8- and 16-row scanners. 1.25-2 mm collimation can be recommended. With all multislice techniques. 4-8 mm thick axial reconstructions serve as the basis of image analysis but 3-5 mm thick coronal or sagittal reformations help further define the shape of suspicious lymph nodes.
Contrast injection The chest and abdomen could theoretically be examined by spiral CTwithout intravenous contrast if there is enough mediastinal and abdominal fat. and if images are reconstructed in a more overlapping fashion and a viewed interactively in a cine display. In most cases this technique will allow nodular structures to be differentiated from tubular ones. While this is very feasible in the chest. it is not generally recommended in the abdomen because it does not detect organ involvement. If the goal of the examination is nodal staging only (Hodgkin's disease, NHL, testicular
tumors), a smaller volume of contrast material may be used (Table 22.8). Vascular opacification makes it easier to differentiate lymph nodes from vessels, but a sufficiently long scan delay has to be used in the upper abdomen to catch the parenchymal phase in the kidneys, and an even longer delay is necessary to ensure a satisfactory contrast enhancement of the inferior vena cava and its side branches. Because the optimum parenchymal enhancement in the liver is earlier than the parenchymal enhancement in the kidneys some compromise is necessary (Table 22.8). If the abdomen is scanned together with the chest or even the neck, one should use the amount of contrast material suggested for the abdomen but time the scan so that the abdominal organs are examined at their proper time. Due to the short scan times with multislice CT, one may consider interrupting the scan above the liver and below the kidneys for a short period to be able to image chest and neck (at about 20-40 s p.i.), the upper abdomen (around 80s p.i.). and the pelvis (> 120s p.i.) at their optimum times.
Criteria for Malignancy While there are no definitive criteria for interpreting the nodal changes observed on CT scans, the combination of size, shape, grouping, and (lack of) contrast enhancement of the lymph nodes can furnish clues to the underlying disease.
Neck In evaluating the lymph nodes of the neck it is important to discriminate between two clinical scenarios, patients with no palpable nodes (NO) and patients with palpable nodes (N -). NO patients are often electively treated by neck dissection if the expected risk of occult metastasis exceeds 20%, based on characteristics of the primary tumor. Imaging may thus reduce the number of dissections in these patients or guide choice of therapeutic procedures. In positive necks, finding additional positive nodes at imaging may guide the type of dissection and adjuvant therapy, depending on the lymph node levels or the side of the neck involved.
Over the years CT has been used extensively for nodal staging, but methodology, definitions, and criteria for malignancy vary widely. While sensitivities of the order of 80-90% have been reported for positive necks, they fall to 40-60% for patients with NO necks. Size: criteria vary and a range of 5-30 mm has been reported, often based on studies encompassing all (NO and N +) patients. Size criteria are especially important for small nodes as other criteria in these nodes are often difficult to evaluate. Based on neck dissections, the best measure of size is the minimal axial diameter. Frequently, 10mm (11mm for the jugulodigastric and 8 mm for the retropharyngeal nodes) is used as the upper limit of normal. As these values are still associated with a low sensitivity for identifying malignancy, a better balance between sensitivity and specificity may be reached by reducing the size limits to a minimum diameter of 8mm (9mm for jugulodigastric nodes), especially in combined populations or positive necks. In patients with a clinically
22 Lymphatic System suspected NO neck, maximum sensitivity is needed and the limits should be reduced to 7mm for jugulodigastric and 6mm for other nodes, giving a sensitivity of 80% and a specificityof60%. Shape: benign nodes are more often oval or fiat and a round shape favors malignancy. Using spiral CT, or preferably multislice CT, the longitudinal/transverse (LIT) ratio can be evaluated and a ur ratio < 2 is associated with a higher suspicion for malignancy. In large nodes, unsharp borders and the obliteration of fat planes may suggest extranodal spread in malignant nodes. This feature is a bad prognostic sign but unfortunately not good for evaluation in small nodes and after recent infection or radiation therapy. Also, imaging can only reveal macroscopic spread with some accuracy and is not suitable for diagnosis of microscopic spread. Grouping: apart from size criteria alone, the grouping of three or more borderline nodes sized 8-10 mm in the drainage region of the primary is suspicious for malignant disease. When size limits are lower, these limits should be changed accordingly. For proper interpretation, a thorough knowledge of the drainage patterns of the primaries is needed (see Chapters). Necrosis: central low attenuation with ring enhancement is usually caused by tumor cells or necrosis. Although this finding is highly specific for malignancy, its incidence is proportional to lymph node size and not often seen in small metastases (NO neck). Moreover, absence of necrosis does not rule out metastasis and it may sometimes be difficult to distinguish it from artifacts, abscesses, or anatomic variants.
Mediastinum In lung cancer staging, a great deal is performed noninvasively. For the mediastinum, CT has limited sensitivity (60-63%) and specificity (7780%) for diagnosis malignant nodes. AFDG-PET is gaining an important place as an adjunct, which can improve nodal staging significantlysensitivities of 79-93% and specificities of 91-94% have recently been reported. Additional confirmation is often established using invasive techniques like cervical or parasternal mediastinoscopy or video-assisted thoracic surgery (VATS). As in other regions, a number of criteria are used in CT of the mediastinum.
Size: as for the neck, it was found that the minimum axial diameter is the most consistent measure for mediastinal lymph nodes. A minimum axial diameter greater than lOmm (l5mm for subcarinal and lower paratracheal nodes) is most frequently used. Based on the older ATS mapping scheme, more differentiated size criteria were published in the mid 1980s (fable 22.5), but these have not really caught on, nor have they been repeated for the new classification of 1996. This probably refiects the fact that noninvasive imaging is used to select those nodes that need additional biopsy. Normal intrapulmonary and hilar nodes usually have diameters of 3-5 mm, while axillary nodes may reach 15 mm in size. Shape: the shape of bronchopulmonary lymph nodes may substantially improve the accuracy for predicting nodal metastases in bronchogenic carcinoma. A convex border towards the lung parenchyma has about a 95% predictive value for malignancy while a concave or straight interface is 95% predictive of a benign node. For other criteria, the same principles as described in the neck section can be applied. Abdomen and Pelvis As in other regions, CT has a limited sensitivity in the detection of nodal metastasis and more detailed information is provided in the individual chapters elsewhere. The criteria used are the same as in other regions, although size criteria are the most important in the abdomen. Size: the minimum axial diameter is the most easy to reproduce and as a general rule a size limit of 10mm is a practical one for the upper abdomen, while in the pelvis a short-axis diameter of 12 mm is frequently considered normal. More differentiated size limits have been published and these are presented along with metastatic spread patterns in Tables 22.6 and 22.7. Size limits are always a compromise between sensitivity and specificity and even a size limit of l Omm can be associated with a sensitivity of only 50% as is demonstrated in Fig. 22.5 for non-seminomas ofthe testis. For other criteria, the same principles as described in the neck section can be applied.
Differential Diagnostic Considerations
o~--------------------------o 5 1015 Size of lymph node in mm
Fig. 22.5 Sensitivity and specificity in the detection of lymph node involvement depends on the selected threshold size value. For nonseminomatous testicular tumors, a threshold of 10 mm provides a specificity of 88% but a sensitivity of only 48%.
Differential Diagnostic Considerations
An apparent lack of nodal regression following radiotherapy or chemotherapy does not necessarily signify a lack of response. Especially with seminomas, a lymph node may show residual enlargement up to 2 cm but may still be devoid oflive tumor cells. After conventional lymphography, the abovementioned size criteria do not apply if CT is performed within 24-48 hours, when a transient increase in nodal size as a result of the lymphangiographic contrast medium is seen. These nodes can be recognized by their relatively high attenuation values. When scanning late (e.g., 3-4 months) after lymphangiography, the nodal size has returned to normal values. The fatty infiltration of lymph nodes may occur in the setting offibrolipomatous degeneration or as a postinflammatory condition. This can lead to corticomedullary differentiation in the affected node, which displays central low density and in extreme cases may appear as a thin, renifonn soft-tissue structure (Fig. 22.6). Lymph node calcifications are most often seen in benign diseases such as tuberculosis, sarcoidosis, and silicosis, but may also develop in malignant diseases such as osteosarcoma or in thyroid carcinoma following radioactive iodine therapy. Centralliquefaction necrosis can occur in metastases, after
Fig. 22.6 Fibrolipomatous degeneration of an axillary lymph node with a typical reniform appearance. The cortex is clearly demarcated from the less attenuating medulla (central low density due to fatty infiltration).
Table 22.9
diagnosis
Lymph node changes: guides to differential
Shape
Oblong (LlT;£2)
Normal lymph nodes, early stages of pathologic involvement
Rounded (LIT <2)
Inflammation, metastasis, lymphoma .; "::AAi~I:;IJA
CT attenuation
Fatty infiltration
Fibrolipomatous degeneration, postinflammatory change
Calcification
Tuberculosis, histoplasmosis, sarcoidosis, silicosis, amyloidosis, metastases from osteogenic tumors, previous irradiation or chemotherapy
Liquefaction
Tuberculosis, staphylococcal infection, metastases, previous irradiation or chemotherapy
Markedly increased density
Previous lymphangiography
Contrast enhancement Homogeneous
Normal lymph nodes, inflammation, lymphoma, hypervascular metastases (thyroid carcinoma, small cell bronchial carcinoma, renal cell carcinoma, carcinoid)
Peripheral
Metastatic squamous cell carcinoma, tuberculosis
Nonhomogeneous
Metastases, amyloidosis
751
22
Lymphatic System
radiotherapy and chemotherapy, and in certain benign conditions such as tuberculosis and staphylococcal infections. Normal lymph nodes, nodes that have undergone reactive hyperplasia, and lymphomas, as well as many metastatic nodes, show moderate, homogeneous enhancement following intravenous contrast administration. Nonhomogeneous enhancement occurs in metastatic lymph nodes and in amyloidosis. To find nonenhancing portions of lymph nodes is quite common in metastases from head and neck tumors and is usually caused by tumor necrosis or cystic degeneration. Although it is a reliable criterion for malignancy, it is not often seen in
small metastases and is sometimes difficult to distinguish from artifacts or anatomic variants. Peripheral enhancement is commonly seen in tuberculosis and in metastases from squamous cell carcinoma. Tuberculous lymph nodes generally have a thicker enhancing rim (» 2 0 % of nodal diameter) than metastatic nodes. As in all areas of diagnostic radiology, the clinical presentation should always be considered in the interpretation of imaging findings. For example, the presence of diarrhea in a patient with segmental thickening of the bowel wall and enlarged mesenteric lymph nodes is more consistent with an inflammatory condition than with neoplastic disease (Table 22.9).
Pitfalls,
Image interpretation in CT requires differentiating lymph nodes from other structures that present a round or elliptical cross-section on axial scans. With spiral CT, overlapping images or secondary reconstructions are useful for distinguishing tubular structures from lymph nodes arranged in a string-of-beads pattern. Nonopacified vessels or vessels that take an atypical course can mimic enlarged lymph nodes, especially in the upper mediastinum or adjacent to the aorta (Table 22.10; Fig. 22.7). If he vessels have been adequately opacified with intravenous contrast material, there should be no difficulty in distinguishing them from lymph nodes. However, some lymph nodes (malignant as well as inflammatory) may show marked contrast enhancement if scanning was started late and some infrarenal veins may not yet be opacified if scanning was started too early. These problems can be solved by using an interactive cine display to distinguish tubular from nodular structures. Muscle tissue has a CT attenuation similar to lymph nodes. The scalene and sartorius muscles may both present similar to lymph nodes on axial cross-sectional images but can be distinguished by their typical location. The diaphragmatic crura can easily be mistaken for lymph nodes, especially if they are thickened and have a marked inferior extension (Fig. 22.8). Again, an interactive cine display or multiplanar reformations will make differentiation easy. The cisterna chyli may be mistaken for an enlarged retrocrural node. It may reach up to
Table 22.10
Enlarged lymph nodes: pitfalls
Scalene muscles Digastric muscle Submandibular gland
Atypical vascular origins from the aortic arch Kinking of the brachiocephalic artery Left superior vena cava Left superior pulmonary vein Aberrant pulmonary vein Azygos vein Superior pericardiac recess Pectoralis minor muscle Abdomen
Thickened crus of the diaphragm Cisterna chyli Venous collaterals at the renal or splenic Lumbar veins Aortic bifurcation, iliac venous confluence Small bowel loops Pelvis Colonic diverticula Appendix Sartorius muse Ovaries
2 cm in diameter and 1-8 cm in length. Differentiation is aided by noting its tubular structure, its typical location (right side of the aorta at the T12-L2 level), and its fluid attenuation.
Tuberculosis Fig. 22.7 A para-aortic blood vessel (in this case a left renal vein crossing behind the aorta) can mimic an enlarged lymph node unless the vessel has been opacified. The tubular structure of the vessel can be appreciated by viewing the images in the cine mode.
b
a
Fig. 22.8 These round masses anterior to the aorta were identified (a) as a thickened, nodular diaphragm crus and (b) as a prominent cisterna chyli by reviewing the slices above and below this section. Multiplanar ref· ormations also improve differentiation, e.g., of this cisterna chyli (c).
Fig. 22.9 In 50% of patients the pericardia! reflection appears as an elliptical structurejust behind the ascending aorta at the level of the left pulmonary artery. Its slightly lower density distinguishes it from the more posteriorly situated azygos lymph node.
Another potential mimic of lymph nodes is the superior pericardial recess (Fig. 22.9), which appears just behind the ascending aorta at the level of the pulmonary artery in approximately 50% of all patients. It can be differentiated from a lymph node as it has broad base connecting to the aorta (bowler hat).
Bowel structures such as the appendix, collapsed bowel loops or intestinal diverticula may also be mistaken for lymph nodes unless optimum bowel opacification has been achieved. While the appendix can be recognized by its tubular structure, diverticula can be identified only by their intimate relation to the bowel wall.
Benign Lymph Node Diseases The spectrum of benign lymph node disorders ranges from inflammatory and granulomatous diseases to storage diseases. CT generally cannot securely distinguish inflammatory diseases from metastatic lymphadenopathy.
Tuberculosis Enlarged lymph nodes are most commonly found in the mediastinum, especially at paratracheal and tracheobronchial sites. As the infection becomes more generalized, lymph node en-
753
22
Lymphatic System
largement may be found throughout the body. The goal of CT scanning. besides defining the intrapulmonary focus. is to determine the extent of the lymph node changes. In patients with an indeterminate pulmonary focus, calcified lymph nodes can direct suspicion toward an old tuberculous infection. CT Morphology
Unilateral hilar adenopathy is detected in 80% of patients and bilateral adenopathy in 20%. Bilateral lymph node enlargement is usually asymmetric. In patients with active tuberculosis, the enlarged lymph nodes have a tendency to form conglomerate masses. Liquefaction of the affected nodes is common, resulting in central low attenuation. In 90% of acute cases, rim enhancement is observed after intravenous contrast administration. The broader rim of these lymph nodes (>20% of the nodal diameter) helps to distinguish them from lymph node metastases due to squamous cell carcinoma. Chronic nodal changes consist of flocculent or homogeneous calcifications.
Other Inflammatory Diseases Enlarged lymph nodes may be seen at the port of entry of the infectious organism or in nodal groups that drain the affected organ. The goal of CT is to differentiate these lymph nodes, which are usually small, from tuberculous and metastatic nodes. CT Morphology
There are no typical morphologic criteria that can differentiate among the various bacterial and viral infections and the various fungal and parasitic diseases. Bacterial pneumonia is usually not associated with mediastinal lymphadenopathy, whereas pneumonias caused by viruses, mycoplasma, or fungi, often present with bihilar adenopathy. Infections with mycoplasma and Pneumocystis cetinii often lead to conspicuous calcifications in the enlarged mediastinal lymph nodes. Other, noninfectious inflammatory changes (e.g., Crobn's disease) and surgical procedures (e.g., thoracotomy) can also lead to mediastinal lymph node enlargement. There is a tendency for lymph node enlargement in patients with
congestive heart disease.
Histoplasmosis Histoplasmosis is a fungal disease that is endemic in central North America (Ohio and Mississippi river valleys) and which has a high prevalence in Middle and South America. The acute or subacute pulmonary form may be associated with generalized mediastinal and hilar lymphadenopathy, while adenopathy is also seen in the disseminated form in patients who are immunocompromised, such as in AIDS. Delayed manifestations include mediastinal granulomas, histoplasmoma, or granulomatous (fibrosing) mediastinitis. CT Morphology
Nonspecific hilar or mediastinal lymph node enlargement or a paratracheallsubcarinal mass with necrosis can be seen. Large nodes may contain popcorn calcifications. Target lesions with central calcifications are pathognomonic for the acute form. Contrast enhancement is often irregular. The mediastinal granuloma manifests as a inhomogeneous, lobulated mass with a thick capsule and irregular septa.
Lymph nodes with inflammatory changes usually appear homogeneous following intravenous contrast administration. However, central liquefaction in lymph nodes, which is common in staphylococcal infections, can lead to central low attenuation with peripheral enhancement.
Cystic Fibrosis Extensive, mediastinal or hilar lymph node enlargement is noted in up to 50% of patients with cystic fibrosis. It should not be mistaken for lymphoma. In nearly all cases, this adenopathy is chronically progressive.
Sarcoidosis The goal detecting mine the tinguish nopathy.
of CT scanning in sarcoidosis, besides intrapulmonary changes, is to deterextent of lymph node changes and disthem from other types of lymphade-
Malignant
•
CT Morphology
Scans usually demonstrate hilar and mediastinal lymphadenopathy chiefly involving the paratracheal and preaortic nodes. Coarse flecks of calcifications develop in the enlarged nodes during the healing stage. Symmetrical hilar involvement is typical and distinguishes sarcoidosis from tuberculosis, which usually shows a unilateral or at least asymmetric pattern of involvement. In cases where the changes are predominantly mediastinal and do not involve the lungs, the differential diagnosis should include malignant lymphoma. This is a likely diagnosis in cases where conglomerate masses have formed. In approximately 75% of cases that have progressed to a chronic stage, enlarged intra-abdominal lymph nodes are also found. Contrast administration does not reveal any changes characteristic of sarcoidosis.
Lymphoma
sis is made from the history and from typical CT findings in the lung. The differential diagnosis should include other forms of pneumoconioses. Asbestosis, for example, is also associated with mediastinal lymphadenopathy but does not lead to eggshell calcifications.
Amyloidosis Amyloidosis is a rare systemic disease characterized by the deposition of amyloid in extracellular tissues. The average age at diagnosis is 55 to 60 years, and males are predominantly affected. A distinction is drawn between primary and secondary forms. The latter is most often secondary to plasmacytoma, rheumatoid arthritis, tuberculosis, or familial mediterranean fever. CT demonstrates a broad spectrum of changes in all organ systems. CT Morphology
Silicosis The goal ofCT scanning in silicosis is to differentiate the affected lymph nodes from other forms of lymphadenopathy. •
CT Morphology
CT demonstrates enlarged hilar and mediastinal lymph nodes with typical fiocculent (eggshelllike) calcifications. Contrast administration does not produce characteristic changes. The diagno-
Malignant
Lymphadenopathy, manifested by nodal enlargement and occasional mottled calcifications, is found in 20% of patients. The enlargement predominantly affects the mediastinal and abdominal para-aortic groups of nodes. The differential diagnosis includes sarcoidosis, silicosis, post-therapeutic lymphoma, and granulomatous diseases. The lymph nodes usually show a nonhomogeneous pattern of contrast enhancement.
Lymphoma
CT has become the imaging modality of choice for the detection, staging, and follow-up of Hodgkin's disease and non-Hodgkin's lymphoma (NHL). Although Hodgkin's disease and NHL have identical CT morphologies, their distribution patterns are dissimilar. A detailed outline of the features that differentiate Hodgkin's disease from NHL is presented in Table 22.11. NHL tends to have a generalized distribution while Hodgkin's lymphoma is often localized. The staging of Hodgkin's disease is based on the number of nodal groups affected, their location in relation to the diaphragm, and the involvement of the spleen and extralymphatic organs (Table 22.12). NHL is staged in similar fash-
ion, except that a distinction is drawn between primary nodal and primary extranodal patterns of involvement. Today, NHL is classified according to the WHO classification published in 1999, which except for minor differences is in agreement with the REAL classification defined in 1994 to unify different European and American classification schemes. Accordingly an attempt was made to standardize the CT morphologic criteria for assessment of therapeutic effects (Table 22.13). This approach, however, uses the long-axis instead ofthe short-axis diameter of the nodes.
756
22 Lymphatic System Table 22.11 Differences between Hodgkin's and non-Hodgkin's lymphoma (NHL)
disease
Table 22.13 Criteria for response to therapy of malignant lymphoma (Cheson 1999)
Involvement
Hodgkin's disease
NHl
Complete remission (CR)
Mediastinum
85%
<50%
Complete disappearance of detectable radiographic and clinical evidence of disease
Lung parenchyma
15%
5%
Para-aortic lymph nodes
25%
50%
Mesenteric lymph nodes
5%
50%
Liver
10%
15%
Diffuse Nodular
++ -+
Regression in size of liver and spleen if previously enlarged
Liver involvement with hepatomegaly
<30%
60%
Complete regression of focal lesions in liver/ spleen or any other organ
Spleen
<40%
>40%
Normalization
Gastrointestinal
lymph nodes 1.5 cm in the largest transverse diameter before therapy ....•'" 1.5 cm
+
tract
Kidney
+
Adrenal gland
+
- = Less common,
Regression of all lymph nodes to normal size: lymph nodes < 1.5 cm in the largest transverse diameter before therapy ....•'" 1 cm
of bone marrow involvement
Complete remission/unconfirmed
(CRU)
The same criteria as for CR but
-
+ = More common
75% regression of volume (sum of the products of largest diameter = SPD) of lymph nodes/masses Indeterminate
bone marrow
Partial remisson (PR) Table 22.12 Staging Arbor classification) Stage I I,
Stage II
II,
Stage III
of Hodgkin's
lymphoma
50% regression of SPD of the six largest (dominant) lymph nodes
Involvement of one lymph node region
No increase in size of lymph nodes, no new sites
Localized involvement phatic organ or site
Regression or constant size of splenic and hepatic nodules
of one extralym-
Involvement of two or more lymph node regions on the same side of the diaphragm Localized involvement of one extralymphatic organ or site and lymph node involvement Involvement of lymph node regions on both sides of the diaphragm
III,
Including localized involvement extralymphatic organ or site
Ills
Including involvement
Stage IV
(Ann
of one
of the spleen
Diffuse organ involvement with or without Imyphatic involvement
All stages are further classified according to the absence (A) or presence (8) of systemic symptoms (fever, night sweats, itching, weight loss)
Bone marrow irrelevant Stable disease (SO) Same criteria as above but less regression No new lesions Progressive disease (PD. non responders) New lesions 50% increase in diameter of lymph nodes previously > 1 cm
~
Organ Involvement
Lymph Node Involvement
Lymph node enlargement is the commonest manifestation of lymphoma on CT scans. Any lymph node station in the body may be affected. •
CT Morphology
Regardless of their location, affected lymph nodes have approximately the same CT density as muscle tissue. Most lymphomas have a homogeneous appearance, but necrotic areas are occasionally seen. Necrotic lymph nodes are more frequently seen with Hodgkin's lymphoma but may also occur after radiation therapy or chemotherapy. Calcifications are rare in untreated lymph nodes and are more common afterradiotherapy. Lymphomas tend to form conglomerate masses (Fig. 22.10). The most commonly affected intrathoracic lymph nodes are the anterior mediastinal, paratracheal, and bronchopulmonary (hilar) nodes. The lymph nodes along the internal thoracic artery and the paracardiac nodes merit special attention, as they are often overlooked despite their high frequency of involvement. A lymphoma is detected in 40% of all patients with enlarged paracardiac nodes. Mesenteric involvement may present with bilateral encasement of the mesenteric vessels ("sandwich" sign, Fig. 22.11). Frequently the nodes are bilateral, are enlarged to more than 2cm in size, and have lobulated margins. Most lymph nodes affected by lymphomatous infiltration display only a slight degree of contrast enhancement (Fig. 22.10). More in-
Fig. 22.10 Retrocardiac nodal conglomerate with encasement of the trachea and main bronchi in NHL. Postcontrast CT demonstrates slight, homogeneous enhancement.
tensely enhancing nodes are occasionally found incertainsubtypesofNHL. Organ Involvement
Primary lymphomas of the internal organs are rare, and most are of the non-Hodgkin variety. Secondary involvement is more common; it is characteristic of higher disease stages, which usually are generalized. For example, the incidence of liver involvement increases from 15% of cases at initial diagnosis to 60% of autopsied patients. The incidence of organ involvement is also increased in lymphomas associated with immunosuppression in patients with organ transp lanta tions. •
CT Morphology
The general criteria for the lymphomatous involvement of an organ are summarized in Table 22.14. Further details are presented in the chapters dealing with specific organs.
Table 22.14 Criteria for organ involvement by malignant lymphoma Disproportionate increase in organ size Hypovascular lesion Interstitial contrast enhancement on delayed scans III-defined margins Diffuse infiltration of surrounding tissues Lymph node involvement independent of typical lymphatic drainage area Involvement of multiple organs
Fig. 22.11 Bilateral mesenteric lymphadenopathy in NHL. The confluent masses may encase the mesenteric vessels, ("sandwich" sign).
757
A
nvart M. Prokop, A. J. van der Molen
Anatomj
.
Examination Technique
0000000000000
General Rules (p. 768) Acquisition Techniques (p. 771) Premedication (p. 776) Contrast Material Injection (p. 777) Imaging Tasks (p. 778) Dose(p.782) Artifacts Congenital
Heart Disease
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Ventricular Septal Defects (p. 788) Atrial Septal Defects (p. 788) Patent Ductus Arteriosus (PDA) (p. 790) Coarctation of the Aorta (p. 791) Transposition of the Great Arteries (TGA) (p, 791) Valvular Anomalies (p. 792) Tetralogy of Fallot (p. 793) Ebstein Anomaly (p. 793) Complex Anomalies (p. 793) Aquired Diseases of the Myocardium Cardiomyopathy (p. 794) Ischemic Heart Disease (p. 795) Cardiac Aneurysms (p. 796) Cardiac Thrombi (p. 798) Arrhythmogenic Right Ventricular Coronary Arteries
0
0
0
0
0
0
0
0
0
0
0
0
Dysplasia 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(p. 799)
0
0
0
0
0
0
0
0
0
0
Congenital Anomalies (p. 799) Calcium Scoring (p. 800) Coronary Artery Stenosis (p. 804) Valvular Disease
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Cardiac Tumors
810
.
Benign Tumors (p. 810) Malignant Primary Tumors (p. 811) Lymphoma (p. 812) Metastasis (p. 813) Pericardia! Pericardial Pericarditis Pericardial "Pericardial IPericardial
[Cardiac
813
Disease Effusion (p. 813) (p. 814) Tumors (p. 815) Cyst (p. 816) Defect (p. 816)
Trauma
.
816
.
817
Penetrating Trauma (p. 816) Blunt Trauma (p. 817) Postoperative
and Postinterventional
Changes
Normal Postoperative Findings (p. 818) Aortocoronary Venous Bypass, Internal Mammary Arterial Grafts (p. 818) Coronary Stents(p.819) Valve Replacement (p. 821) Congenital Heart Disease (p. 821) Heart Transplantation (p. 821) Sternal Dehiscence (p. 823)
The heart is consistently imaged on chest CT examinations, and a variety of abnormal findings may be seen on such standard spiral scans. ECGtriggered and ECG-gated acquisition with multislice scanning, however, allows for motion-free depiction of cardiac structures and even the coronaries. While only a few years ago CT was not considered a primary modality for cardiac imaging, it is now becoming a promising competitor to echocardiography and MRI. Electron beam CT (EBCT), which is also called ultrafast CT (UFCT), is a technique that was specifically developed for cardiac imaging in the 1980s. It has not gained a wide acceptance except for coronary calcium scoring. It used to be the only CT technique that was able to freeze heart motion, and required expensive and dedicated scanners to do so. Despite its short history, multislice CT has gained more rapid acceptance than EBCT ever could. This is probably due to its superior spatial resolution and the fact that theses scanners are general CT units that also can be employed for the heart. In
principle, EBCT is a good and dose-efficient technique for cardiac imaging, in which multislice technology might even be integrated. Even standard spiral and multislice CT offer morphological and functional information about the heart that should be used more frequently when reporting chest CT examinations. CT angiography is sufficient for evaluation of the great vessels and many congenital anomalies but does not allow for more detailed analysis of the heart chambers. This requires ECG-triggering to freeze heart motion. EBCT and multislice CT both offer ECC-triggering as a dose-efficient technique for morphologic analysis of the heart and for coronary calcium scoring. However, spatial resolution along the z-axis is limited and acquisition time is relatively long with present day scanners. Analysis of the coronaries requires coronary CTA, which relies on ECG-triggered EBCTscans or ECG-gated multislice CT. ECG-gating also requires oversampling of data from time intervals outside the desired range and thus increases radiation exposure. The 4D data
Segmental Anatomy ofthe Heart
acquired with ECG-gating can be used to obtain information about cardiac motion and thus can provide functional cardiac parameters (Table23.1). Cardioangiography remains the gold standard for evaluation of the coronary arteries. and for pressure measurements in valvular and congenital heart disease. It is frequently combined with interventional procedures. It provides functional morphological parameters (wall motion. ejection fraction) but has problems with determining the viability and perfusion of myocardial tissue. At present. echo cardiography is the non-invasive workhorse for many cardiac diseases. including septal defects. valvular disease. and motion abnormalities of the myocardial wall. Functional imaging of the myocardial perfusion is based on thallium scintigraphy (SPECT) or PET. MRI provides excellent spatial detail, improved 3D orientation and capabilities of flow quantification, and is well established for imaging of congenital heart disease. The evaluation of valvular disease is possible but more expensive than with echocardiography. The evaluation of motion abnormalities of the myocardial wall as well as perfusion studies and coronary imaging are experimental techniques that hold great promise. At present, CTA only appears superior to MRA for noninvasive imaging of the coronary arteries, but cardiac CT is undergoing extremely rapid development.
Table
23.1
Cardiac
Concomitant
CT: applications
cardiac
findings
Calcifications Coronary arteries Valves and valvular Myocardium Pericardium Abnormalities Dilatation
and indications
on chest
CT
rings
of the
heart
chambers
Thinning of myocardium Old scars/infarction Aneurysms Ventricular
or atrial
thrombi
Tumors Functional
abnormalities
Hypomotility Reduced Present
right ventricular
indications
Coronary
calcium
functio
for cardiac scoring
• Patients with atypical
CT
(under discussion)
chest pain
• Screening moderate
of asymptomatic risk estimates
• Follow-up
of pharmacological
subjects
with
lowl
or nutritional
Interventions Patency of coronary artery bypass grafts Congenital heart disease • Adult patients prior to surgery • Patients with cardiac pacemakers Evolving
indications
Coronary Regional Function:
artery disease (main branches) wall thickness ejection fraction, cardiac output,
regional
motility
Future
potential
Myocardial
for cardiac
applications
viability
after
CT
of cardiac
infarction,
PTA, or
stenting No indications Valvular Disease
for cardiac
disease of the distal
CT
coronary
artery
branches
Anatomy The normal cross-sectional anatomy ofthe heart chambers is shown in Fig. 23.1. Standard crosssectional images include the four-chamber view, long and short-axis views, and views of the right and left ventricular outflow tract. Such images are reformatted from 3D ECG-synchronized data
sets according Fig. 23.2.
to the technique
displayed
in
761
762
23 Heart Cross-sectional Fig. 23.1 anatomy of the heart. Ao AV D1 D1
GCV '" LA LAA
LAD
I
LCA LCX LPV LV MV RA RAA RCA RPV
, ,. •
,.
RV TV
Aorta Aortic valve First diagonal branch Great cardiac vein Left atrium Left atria I appendage Left anterior descending Left coronary artery Left circumflex branch Left pulmonary vein Left ventricle Mitral valve Right atrium Right atrial appendage Right coronary artery Right pulmonary vein Right ventricle Tricuspid valve
4
Segmental Anatomy of the Heart
Segmental analysis is important for comprehensive diagnosis of congenital heart defects, as morphologic and functional heart chambers need not be identical. Segmental analysis requires accurate identification of the atrial situs, ventricular morphology and large arteries, as well as atrioventricular and ventriculoarterial connections. •
Atrial situs
The morphological right atrium is connected to the IVe, while the left atrium has no connections to the IVe. The right atrium contains a larger, triangular atrial appendage that has a wide connection to the atrial vestibule. In the normal atrial situs solitus, the morphological right atrium is located on the right side and the left atrium located on the left side. A complete mirror of atrial situs is referred to as atrial situs in versus. The situs of the viscera is usually in concordance (visceroatrial) with the
atrial situs. The eparterial (over the pulmonary artery) short main bronchus, liver, and IVe are right-sided structures, while the hyparterial (under the pulmonary artery) long main bronchus, stomach, spleen, and abdominal aorta are left-sided. As in the atria, a complete mirror situs is referred to as visceral situs inversus. In situs ambiguus the atrial and/or visceral position is uncertain or cannot be determined. It is diagnosed when symmetry of the main bronchi and pulmonary arteries is present and may occur as bilateral left -sidedness (left isomerism) or right -sidedness (right isomerism). The bronchial situs is usually concordant with atrial situs and may be helpful in difficult cases. While left isomerism is associated with polysplenism, right isomerism shows asplenia. Atrial isomerism is usually associated with complex cardiovascular anomalies. •
Ventricular Morphology
The major morphologic characteristics of the right ventricle include: the presence of a myo-
Segments/
a
b
d
e
Fig. 23.2 Creation of standard sections. An axial section at the level of the mitral valve (a) is used to define the long-axis views though the left (b) and right ventricle (c). An oblique section in the direction ofthe heart
cardial infundibulum between the semilunar and atrioventricular (AV) valves, a tricuspid AV valve, more coarse trabeculation at the septal surface and the moderator band. The left ventricle shows direct continuity of semilunar and atrioventricular valves, a higher attachment of the septal leaflet of the AV valve and two large papillary muscles. Levocardia or dextrocardia (see Fig. 23.41) are defined by the direction of the interventricular axis, with the apex of the heart pointing to the left or to the right, respectively. •
Great Arteries
The great arteries are defined by their branching (pulmonary artery) or the presence of an arch (aorta). Transverse images normally show a left and anterior position of the pulmonary artery. In transpositions, the aorta is anterior and, depending on the position of the aortic valve with respect to the pulmonary valve, D- or L-transpositions are known.
Anatomy
of the Heart
c
f axis creates the four-chamber view (d). Sections perpendicular to this axis yield the short-axis views (e. f). Note the calcification of the posterior mitral valve leaflet.
•
Atrioventricular Connections
The connections of the atria and ventricles are divided into concordant or discordant. When the right (left) atrium is connected to the right (left) ventricle, there is AV-concordance, otherwise it constitutes AV -discordance. AV connections can be further subdivided into usual alignment, criss-crossing (rotating AV connections), overriding-straddling, double-inlet (single ventricle), or absent unilateral AV connection (AV atresia) . •
Ventriculoarterial Connections
The Ventriculoarterial connections can also be divided into concordant or discordant. In concordance the anatomical right ventricle connects to the pulmonary artery and the left ventricle to the aorta, while in discordance it is the other way round (transposition).
763
23 Heart
Heart Chambers The heart has the shape of an ovoid (Figs. 23.1). The left atrium constitutes the posterior portions; the right atrium defines the right border. the right ventricle lies anteriorly. and the left ventricle is situated between the right ventricle and left atrium and defines most of the left border. The outer anatomy can best be appreciated on volume-rendered displays (Fig. 23.3). The right atrium (RA) has an approximately tubular shape that bulges at the level of the tricuspid valve and extends in a straight line to the superior and inferior vena cava. The triangular right atrial appendage (RAA) has a wide-based connection with the atrium. It extends anteriorly and covers portions of the right ventricle. The right ventricle (RV) is located anteriorly immediately behind the sternum. The distance to the sternum may vary depending on interposed lung or precardiac fatty tissue. A direct adherence of the right ventricular pericardium to the sternum may complicate median. sternotomy. The right ventricle shows a more pronounced trabeculation than the left ventricle. which is an important distinguishing criterion for the evaluation of complex congenital heart disease. The normal thickness of the anterior wall of the right ventricle is 5 mm or less. There are three papillary muscles (medial, anterior, posterior) that are connected to the leafiets of the tricuspid valves. A band-like muscle (trabecula septomarginalis or moderator band) crosses the apex of the right ventricle from the anterior wall to the septum. The right ventricular outflow tract (RVOT) has a caudocranial course and often a thicker wall than the rest of the right ventricle (caused by band-like muscle ridges, the parietal band, and supraventricular crest). The RVOT bends posteriorly before it ends at the pulmonary valve. The left atrium (LA) is located in a posterior position anterior to the spine. The distance to the spine depends on the sagittal diameter of the chest and the size of the heart. The esophagus directly abuts the left atrium. Enlargement of the left atrium leads to a reduction of the retrocardiac space, displaces the esophagus to the right, and spreads out the tracheal bifurcation. Normally, there are two pulmonary venous orifices on either side, which collect blood from the upper lobes and seg-
rnents 6, and from the lower lobes and the middle lobe/lingula. Variants of these regions are possible and are of importance in electroablation therapy for atrial fibrillation. The tubular-shaped left atrial appendage (LAA) has a narrow ostium to the left atrium. It extends anteriorly to the left and covers the proximal portion of the left coronary artery. This superimposition requires removal of the left atrial appendage before 3D-visualization of the proximal segments of the left coronary. Like the right atrial appendage it has a lobulated appearance and a relatively sharp tip that is oriented anteriorly. The cardiac septum has a diastolic thickness of between 6mm and 12mm. It is positioned obliquely towards the left anterior chest wall. The left ventricle follows this orientation. Its posterior wall has a normal myocardial thickness of 6-12 mm. The ventricular musculature should display a uniform thickness as far as the attachments of the two papillary muscles (anterior and posterior). The left ventricular outflow tract is located in the center of the heart, in direct vicinity to the LA, RA, and RV. On crosssectional images the aorta arises in a central position and is surrounded by the LA, LV, RVOT, RA and right atrial appendage. The four heart valves consist of a fibrous annulus for stabilization and two to three cusps that provide the valvular function. The tricuspid and mitral leaflets are attached to the papillary cords that are connected to the papillary muscles. The aortic and pulmonary valves consist of cup-shaped cusps with a convexity oriented towards the ventricular outflow tracts. Enlargement of the valvular annulus will ultimately lead to valvular incompetence. Valvular adhesions and calcifications may cause stenosis. Since the aortic valve lies on the axial plane, the aortic valve cusps are imaged in transverse section. The mitral and tricuspid valves are perpendicular to the scan plane, so they are imaged in longitudinal section. The pulmonary valve lies close to the sternum just above the aortic valve. It is visualized in an oblique longitudinal section on axial scan planes. Optimum display of the valves requires sections parallel and perpendicular to the valvular plane. Measurements of cardiac size should be viewed with caution on standard CT sections because there is no precise distinction between diastole and systole, and the cardiac axis is not
Coronary Vessels Table 23.2 Normal end-diastolic cardiac measures derivedfromechocardiography Aortic root Left atrium (LA) Right ventricle (RV) Interventricular septum (IVS) Left ventricular posterior wall (LVPW) Ratio IVS : LVPW Left ventricle (LV)
20-37 mm 15-40 mm <30 mm 6-12 mm 6-12 mm <1.3 39-56 mm
optimally portrayed on axial sections. Size measurements require images along the short and long axis of the heart. These sections can be reconstructed using the technique described in Fig. 23.2 (see also Fig. 23.14). If there are motion artifacts, the diastolic size can be estimated from the widest of multiple contours, and the systolic size from the narrowiest contour. Normal ranges derived from echocardiography are provided in Table 23.2. The diastolic measurements do not vary much with longer duration of data acquisition but systolic measurements systematically overestimate the true size when scans with less temporal resolution are performed.
a
b
Fig. 23.3 Volume rendered views of the heart using viewing angles similar to the standard projection of coronary angiography. (a) 30° ROA (b) lateral, (c) 45°
Coronary Vessels
The normal right coronary artery (RCA) arises anteriorly from the aorta, just above the level of the aortic valve. Its proximal portion is located in the atrioventricular groove between the right ventricle and the right atrium and gives rise to acute marginal branches supplying the anterior RV wall. In 80% it ends in the posterior descending artery that courses anteriorly at the bottom of the heart directly below the septum. Small proximal branches of the RCA supply the sinoatrial node and the AV node. The left coronary artery (LCA) is seen some 1-2 em further, cranial to the orifice of the right coronary, and has a posterolateral origin from the aorta. It branches into the left anterior descending coronary artery (LAD) and the left circumflex coronary artery (LCX). The LAD has a course above the interventricular septum and gives rise to diagonal branches supplying the more anterior portions of the LV wall, as well as to septal branches that supply the septum. The LCX runs in the left atrioventricular groove and gives rise to the obtuse marginal and posterolateral branches supplying the posterior wall of the LV. In 20% of cases it gives rise to the posterior descending artery at the base of the heart. In 10 %, both left and right coronary arteries contribute to the posterior descending artery.
c LAO with 30° craniocaudal angulation. The same abbreviations as in Figs. 23.1, 23.4 and 23.5 are used.
765
766
23
Heart
1 - RCA - Right coronary/proximal 2 - RCA - Right coronary/middle 3 - RCA - Right coronary/distal 4-POA-Posterior desc. artery AM -Acute marginal branch SN -Sinus node branch
5 - LCA - Left main coronary artery 6 - LAD - Left ant. desc'/proximal 7 - LAD - Left ant. desc./middle 8 - LAD - Left ant. desc./distal 9-0-\ - First diagonal branch IO-OZ -Second diagonal branch
Septal .',' branches /:'
". ;::«('.\.::::_:::::::
b
11 - LCX - Left circumflex/proximal 12 - OM - Obtuse marginal branch 13 - LCX - Left circumflex/distal 14 - PL - Posterolateral branch 15 - PO - Posterior desc. branch
Fig. 23.4 Segmental classification of mayor coronary artery branches (American Heart Association 1975).
The coronary arteries are classified according to the American Heart Association (1975) into 15 major segments (Fig. 23.4), which are used in describing the location of coronary disease. There are numerous variants of the coronary anatomy, of which a large branch (intermediate coronary artery) arising in the region of the bifurcation of the left coronary artery into the LAD and Lex is the most important (see Fig. 23.37). The vascular territories of the heart are provided in Fig. 23.5 for the most common type of anatomy. The apex of the heart as well as the superior portions of the septum are supplied by the LAD. In the majority of cases, the LCX supplies the laterobasal portion of the left ventricular wall while the RCA supplies the posterior wall, the basal portions of the septum and the posterior portions of the inferior wall of the heart. In a right-dominant type of vascular supply, the RCA takes over part of the LCX territory, while the LCX takes over part of the posterio- inferior supply to the heart in a left-dominant type. The anterior wall of the right ventricle is supplied mainly by the RCA but small branches from the LAD may also contribute. The coronary veins (Fig. 23.6) run parallel to the main coronary arteries and end in the coro-
Pericardium
a
c
b
Fig. 23.5 Vascular territories of the heart (most common type of anatomy). Variations mainly involve the borders between the LCX and RCA territories (right- or
nary sinus that enters the inferior vena cava about 2-3 cm above the diaphragm and has a small valve at its orifice. The coronary sinus measures some 5-10 mm in width and is located at the posterior base of the heart in the groove between the base of the LA and LV. The right coronary vein runs in the right interventricular groove and enters the coronary sinus together with the posterior interventricular vein [middle cardiac vein). This vein lies parallel to the posterior descending artery. The left anterior interventricular vein, which parallels the LAD, drains into the left coronary vein [great cardiac vein}, which runs in the left interventricular groove and enters the coronary sinus together with the posterior vein of the left ventricle. Anterior cardiac veins are seen along the surface of the right ventricle and drain directly into the right atrium. An aberrant or accessory left superior vena cava will join the left coronary vein as it enters the coronary sinus. Pericardium
The pericardium is visible on CT scans as a thin line that measures between O.8mm and 2mm on thin sections. The pericardium encloses the heart, the main pulmonary trunk, and most of the ascending aorta. The visceral pericardium is separated from the heart by epicardiac fatty tissue that is most pronounced around the coronary vessels. Visceral and parietal pericardium join at the base of the heart and can be differentiated on CT only if there is fluid or air within the pericardial sac.
left dominant type of vascular supply), and the borders between the LAD and LCX territories.
Gev PVLV MeV sev RMV -
Fig. 23.6
Great cardiac vein Posterior vein of left ventricle Middle cardiac vein Small cardiac vein Right marginal cardiac vein
Anatomy of the coronary veins.
Normally, fluid is present only in the upper pericardial recess posterior to the ascending aorta, just proximal to the base of the aortic arch. This produces a typical teardrop-shaped retroaortic density that should not be mistaken for a mediastinal lymph node (see Fig. 22.9). Normally small amounts of fluid, however, may also be found below the aortopulmonary window and anterior to the left ventricle (Fig. 23.7).
767
23 Heart
b
c
Fig. 23.7 The pericardial sac is outlined by contrast material following a tiny accidental perforation of the right atrium in a patient after a thoracic aortic stent graft. Note the pericardial recess anterior to the aortic
arch (a), posterior to the ascending aorta and lateral to the pulmonary artery (b), as well the small amounts of fluid in the pericardial sac lateral to the right atrium and around the ventricles (c).
This fluid distribution may change with patient position, for example, if long-axis views are obtained by tilting the table of an EBCT scanner. A
pericardial effusion is present if the amount of fluid exceeds 50 ml.
Examinatiol
Cardiac motion is rapid and complex and causes substantial artifacts on CT images. Evaluation of heart diseases, however, may be possible even if the are some residual artifacts. Depending on the imaging task, freezing of heart motion may or may not be mandatory. General Rules
For the sharp delineation of cardiac structures, the data used for image reconstruction must be consistent, i.e., obtained during a phase of minimal or no motion. This requires reduction of acquisition time and ensuring little motion during the data acquisition window. Reduction of acquisition time can be achieved by shortening the rotation time of the x-ray beam, half scan reconstruction, and various ECG-gating techniques with multisector reconstruction. Ensuring minimum motion of the heart requires an optimum choice of heart phase, which often can only be achieved by suitable premedication of the patients.
Temporal Resolution
The durations of systole and diastole relate differently to heart rate. Because the duration of systole decreases in an (almost) linear fashion, the duration of diastole decreases much more quickly as the RR interval shortens and the heart rates increases (compare Fig.l.32b). For this reason, (end) diastole offers a good opportunity for imaging only if heart rates are low (/\70bpm). For high heart rates of 90bpm or more, diastole may no longer be optimum, and early systole during the phase of isometric relaxation may be superior. The optimum phase of imaging, however, strongly depends on the duration of the window for data acquisition. In addition, heart motion has a complex three-dimensional pattern that consists of concentric contraction of the ventricles towards the septum, shortening of the cardiac long axis, and a rotational movement of the ventricular wall due to the slightly spiral course of the myocardial muscles. While there is relatively little movement of the cranial portion of the septum
Temporal Resolution
(in the region of the LAD), the right atrioventricular groove usually moves more rapidly. There are substantial variations between patients, depending on anatomy, training status, and preexistent disease. Coronary motion has been derived from EBT data. One study found that mean velocity varied from 22 ± 4 mmls for the LAD, to 48 ± 15 mmls for the LCX, and up to 70 ± 22 mmls for the RCA. Peak velocity in the RCA could exceed 250 mml s. Peak velocities were reached during early systole, early diastole and atrial contraction. The speed closely depends on heart rate and timing during the heart cycle. Interindividual variation is high. This implies that the RCA is most difficult to image and requires either a very short acquisition window or a low heart rate for good results. The time window of minimal motion also varies between the various segments of the coronary tree. The lowest mean velocity is found to be around 40-50% of the cardiac cycle and corresponds to isovolumetric relaxation in end systole. Another time window with low velocity is seen around 80% of the cardiac cycle in diastole just before atrial contraction (Fig. 23.8). With longer acquisition time, the optimum time windows vary and the diastole on average becomes the best phase of imaging. Given that the whole diastole can be used for data acquisition, a lower limit for the temporal resolution (upper limit for the duration of the data acquisition window) can be calculated from the duration of the diastole (see Fig. 1.32b).
Unfortunately not the whole diastole can be used. The phase of atrial contraction should be excluded from the data set, particularly for display of the RCA. This requires an increased temporal resolution. At a heart rate around 80 bpm (750ms per RR), which is found in most patients, the minimum required temporal resolution is well below 200ms. At a heart rate of 60 bpm this number increases to 300 ms, and to as much as 500 ms for 50 bpm. However, even at low heart rates, the quality of images improves with shorter acquisition windows. Heart rate varies physiologically between inspiration and expiration and may change substantially during one acquisition as contrast material is injected. Also, arrhythmias may make it impossible to predict the optimum acquisition time. This can make ECG-triggering difficult because the optimum acquisition window may shift during the scan as the heart rate changes. The situation becomes even more complex because the heart relaxes during early diastolic filling and remains relatively motionless only during end diastole. With high heart rates such a motionless phase is no longer available because the duration of the diastole is so short that it is completely required for diastolic filling, which may even remain incomplete at very high heart rates. The phase of isometric contraction during early systole, and isovolumetric relaxation in end systole, remains relatively constant independent of the heart rate and thus may be more suitable
80
Vi'
E E
Data acquisition window: .------50 ms _____ --100 ms
70
~-400ms -----250 Ins
60
c 0
:,=; 0
50
E
C 40 rc c 0 L... 0
30
rc
20
U
.::: I
Fig. 23.8 Velocity of the coronaries (average values for the major coronary segments) depending on the time window of data acquisition during the cardiac cycle (adapted from Achenbach 2000).
10
a
0
20
40
60
% of cardiac cycle
80
100
769
23
Heart
for data acquisition than late diastole, when higher heart rates are present The end-systolic phase is seen at around 40% of the RR interval (30% with lower heart rates), Unfortunately, substantial variations occur between individual patients that make it impossible to predict the required temporal window for optimum image quality, Independent of patients, however, low heart rates make imaging much easier and yield good image quality more reliably, EEeT allows for a temporal resolution of 30-100 ms, depending on the scanner generation, required spatial resolution and signal-tonoise ratio, A temporal resolution of 80-250 ms can be accomplished with current 4-detectorrow multislice scanners with 0.5 s rotation time, depending on the heart rate, the pitch, the rotation time, and the segmental reconstruction capabilities, With some 8- and 16-slice cardiac scanners, even faster rotation times (0.4-0.42 s) become available and use of narrow section collimation will be the rule, The following general rules give an overview of the most important points to remember. • Low heart rates « 70bpm) should be preferred for data acquisition with 4-and 8slice scanners (consider using beta-blockers) • Short acquisition times of <200 ms improve the display of heart valves and small coronary arteries • Optim um tigger delays depend on heart rate and duration of the acquisition window (with substantial inter individual variations) For short acquisition windows « 200ms), triggering at 40-50% or 80% of the cardiac cycle yield best results, • For longer acquisition windows (s 200ms), optimum tigger delays at low heart rates vary between 40 and 80%,
Table
23.3
Scan
range depending
Scan Range and Scan Duration Scanning of the heart should b'e performed within one breath hold, The scan range varies depending on the imaging task (Table 23,3) and may have to include the heart up to the supraaortic vessels for patients with congenital heart disease or may focus on just the region of the heart including coronary orifices for evaluation of coronary artery disease (some l Zcm). For coronary artery disease, a test scan at the top of the scan range can be recommended to make sure that no portion of the coronaries is seen, so that the coronary arteries are completely covered, Scanning of the heart is performed with the breath held in inspiration, In principle, the duration of breath holding should not exceed 30 seconds but this may not be possible with ECG-gating techniques on 4-slice scanners, The actual scan duration will vastly depend on the heart rate of the patient and the employed technique (pitch, number of sectors used for reconstruction) and requires between 30s and 40s, With 8- and 16-slice scanners, scan duration can be reduced to well below 30 s, In any case, it is of paramount importance to inform the patient about the procedure and the importance of the breath hold, and to test the breath-holding capability of the patient prior to the actual scan, A period of hyperventilation prior to the scan as well as nasal oxygen can vastly improve the ability of patients to hold their breaths even for longer periods of time, It is also advisable to instruct them not to gasp for air "internally" (i.e, make involuntary diaphragmatic movements without actually breathing) close to the end of the scan, which would also seriously impair evaluation of the heart chambers and the coronary arteries, If possible the patients should be trained not to perform a Valsalva maneuver during breath-holding because this influences heart rate,
on indication
Indication Congenital
artery
bypass
Coronary
artery
disease
a in adults
Base of heart to 3 em above
heart disease
Coronary
graft patency
lLR1g apex
Base of heart tojust above aortic arch Base of heart to (1 em below)
trachea!
bifurcation
20-25 15-20 10-12
em3 em em
Acquisition Techniques Fig. 23.9 Principle of EBCT: an electron beam is accelerated in a giant vacuum channel and deflected by magnetic coils to sweep up to four consecutive semicircular tungsten targets. Note that there is only a 210° rotation of the x-ray beam that enters the patient from below.
Deflection coils
Focussing coils
1---
Electron
Detector rings
H---X-rays ~-rr---'+'--'-'
'------Target
Vacuum pumps
rings
Electron beam --------'
Acquisition Techniques Electron Beam Computed Tomography (EBCT)
Electron beam CT scanning, also called electron beam tomography (EBT) or ultrafast CT (UFCT), replaces the conventional rotating x-ray tube with an electromagnetically deflected electron beam that generates x-rays "on site" and thus can achieve millisecond scanning speeds. An electron beam scanner is similar in its functional principle to a giant x-ray tube. Various scanner geometries have been designed, but only one has found practical use (GE- Imatron). An electron beam is produced at the "head" end of the scanner, accelerated at 130kV, and electromagnetically deflected within a vacuum chamber to strike a series of four parallel tungsten targets (A, B, C, and D) arranged in a semicircular ring (210°) around the patient (Fig. 23.9). The system can be programmed to scan one of the tungsten target rings in 50ms or lOOms (with the newest scanner generation this can be performed in 30 ms and 60ms, respectively). An additional 8ms (at 50ms) or 16ms (at lOOms) is needed to rescan the same target ring or switch from one target ring to an adjacent ring. When the beam strikes the target, it generates a. fan-shaped x-ray beam with a 30° angle of divergence. Mounted opposite the target rings are two parallel, semicircular (240°) tungsten detector arcs, each containing 1878 detector elements. Only the second ring is used for high-resolution imaging. The focus is elliptically shaped and rather large.
The electron beam has a constant milliamperage (630mA), so the applied dose is always proportional to the scan time. At the shortest scan time of 50 ms per slice, the tube output is low (31.5 mAs) and image noise is correspondingly high. Images are mainly used for functional studies. At lOOms, the image quality is sufficient for coronary calcium screening and coronary CTA (63 mAs) in most patients. The gantry cannot be angulated but the patient couch can be tilted upward or downward and from side to side. This can be used to obtain long and short axis views of the heart. •
Single Slice Mode
In the single slice mode of operation, the electron beam scanner is used like a conventional CT scanner. This mode utilizes lOOms sweeps along a single target ring (C). Increasing the number of sweeps (up to 17) will increase exposure time and patient dose but will also improve signal-to-noise ratios. The collimation can be varied from 1.5 mm to 3 mm and 6 mm, and only the high-resolution detector is used. For most cardiac applications, the 3 mm single slice mode with lOOms acquisition time is recommended. The 3 mm sections are also most dose-efficient because the 1.5mm and 6mm sections use a post-patient collimation that approximately doubles radiation exposure. The single slice mode provides two acquisition features that resemble sequential scanning and spiral scanning in conventional CT. Step volume scanning (SVS) is step-andshoot procedure, which can be combined with ECG-triggering. At a predefined phase of the
771
23 Heart
cardiac cycle (as a percentage of the RR interval or at a fixed time after the R wave) a lOOms sweep is obtained. After translating the patient by a predefined table feed the scan is repeated in the next cardiac cycle. In order to cover the heart (some 12cm). 40 sections and thus 40 heart cycles are required. Scanning is faster with higher heart rates (30s at 80bpm versus 40s at 60bpm). A new feature is available with the most recent models that allows for acquiring three images at various trigger points during an RR interval before moving the table. This three-phase electron beam angiography was developed to optimise imaging of the coronaries even in patients with arrhythmias. In continuous volume scanning (CVS) , data are acquired while the patient is moved continuously through the scan plane, very much as in spiral CT. Unlike spiral CT, the table feed per acquisition is small and z-interpolation is not used. This mode is generally not used for cardiac imaging. Multislice
Mode
In multislice sequence scanning, the electron beam sequentially sweeps along target rings A through D in 50ms each, with a time delay of 8 ms between each sweep. Because both detector rings are exposed, two 8 mm sections are acquired simultaneously during each sweep. Because of the distance between the target rings, there is a gap of 4 m m between consecutive sweeps (image pairs). The multislice mode thus covers some 8 cm of volume with eight sections in a quasi-simultaneous fashion (total acquisition time 224ms) without movin~ the patient through the scanner. Only 256 and 3202 matrices are used because of the increased noise at 50ms/31.5mAs and a low number of projections in this mode (detector elements are combined for readout to improve signal-tonoise ratio). A maximum of 128 images (= 16 cycles of eight images each) can be acquired with current scanners. The flow mode is a variant of the multi-slice mode in which the same region is sequentially rescanned at every heartbeat or every second beat (triggered by the R wave of the ECG). The transit of a bolus of contrast medium can be analyzed to evaluate flow characteristics (e.g., detect right-to-left shunts) and to assess myocardial pel-fusion.
In the cine mode, a single target is repeatedly scanned for 50ms with an 8ms interscan delay. This technique can produce multiple images during one cardiac cycle. The use of both detector rings permits simultaneous imaging of two 8 mm sections. After one cardiac cycle or a specified number of scans has been completed, the next target is scanned. This can provide scanning rates as high as 17 sections (or image pairs) per second. The scanning sequence is triggered by the R wave [Q%RR interval) of the ECG. The patient table must be tilted to obtain longaxis views of the heart. The examination should be performed during the plateau phase of contrast enhancement to obtain optimum delineation of the cardc chambers. A time-resolved 8cm long (4D) volume can be acquired, without moving the patient, in as little as four heart cycles. The cine mode is best suited for obtaining functional morphologic information (cardiac output, ejection fraction). Both flow and cine modes can also be performed using detector 2 and target C only. The flow mode can be used to determine circulation times (test bolus). The cine mode, using a highresolution setting, can be used to find the best timing (%) within the RR interval for subsequent coronary eTA.
Single and Multislice CT Non-gated
Scanning
Non-ECG synchronized spiral CT scanning (as for standard chest applications) can provide sharp images of the heart for those table positions in which most projectional data are acquired during diastole. Whether this is at all possible depends on the heart rate and the rotation speed of the scanner. In single slice spiral CT, data is sampled from two 360' rotations for 360' Ll interpolation, and from 360' plus the fan angle for 180' Ll raw data interpolation. This means that the total acquisition time window for 1 s scanners is 2 s for 360' Ll and some l.l5s for 180' Ll interpolation. In other words, each image on a 1 s scanner includes data from more than one heart cycle and therefore would never be able to provide a sharp representation of the heart. This is not absolutely correct, however, because the raw data interpolation assigns smaller weighting factors to projections that are farther away (both spa-
Single and Multislice CT
a
b
Fig. 23.10 Cardiac motion during a non-ECG synchronized spiral scan leads to a kymographic effect that reflects local pulsatilitywithin the scan plane. Compare the effect in a patient with a right-sided tension pneumothorax(a)toa patientwith bradicardiaand hypertrophic cardiomyopathy (b). Both acquisitions were obtained with 4 x 1 mm collimation and 6 mm table feed.
tially as well as temporally) from the reconstructed plane. With both interpolation schemes, 75% of data used for reconstruction are acquired during 50% of the acquisition time window. For this reason, it is possible to obtain fairly good images for those sections in which the highest weights are assigned to projections obtained in diastole (diastole-weighted images), as long as the heart rate of the patient is low (around 60 bpm or less, see Fig. 1.32b). Pulsation will cause serration of contours on multiplanar reformations along the z-axis, creating a kymographic effect (Fig. 23.10, see also Fig. 7.34) that is an indicator of regional wall motion. The frequency of serrations depends on scanning speed (table feed TF/rotation time RT) and heart rate (f). Faster table speed will spread out the serrations over larger ranges and may ultimately completely obscure this kymographic effect. This is the case with the use of thicker sections, higher pitch factors, and, in particular, dual or multislice CT scanning. It is most pronounced for 16-slice scanners. With a 16 x 1.5mm collimation and a pitch P of 1.5, these scanners cover 72 mm per second, which means that the whole heart is acquired in less than two heartbeats. The-relatively artifact -free-volume that is acquired during diastole extends over several em, and so does the artifact -prone regions that are acquired during mid systole. Since no ECG-gating is used, it is a random event that
determines which portions of the heart will be artifact -free and which ones will suffer from artifacts. Pulsation effects in systole are only reduced with faster rotation speeds. A higher pitch factor has the advantage that the serrations are spread out and may be less disturbing on volume-rendered images of the heart and great vessels. At the same time, the various structures of the heart may only be captured on a few sections in diastole or may be completely missed if they are scanned during systole. If no ECG-triggering or gating is available, low pitch factors may be a suitable way in some patients to obtain enough diastoleweighted images to assess heart morphology. Reconstructing thicker sections from multislice CT raw data will average projections over more than one heart cycle and will thus yield images with relatively reliable outer contours but superimposition of the various internal contours of the heart. This technique is not recommended for interpreting the heart on non- ECG synchronized multislice CT image data sets. •
Prospective ECG Triggering
Prospective ECG triggering allows for sequential acquisition of multiple simultaneous sections, depending on the number of detector rows available. A prospective trigger is derived from the ECG signal to initiate scanning during a phase of little heart movement. The resolution is 100 ms for EBCT and more than 250 ms for the various single and multislice CT scanners. With single slice and multislice CT scanners, optimized temporal resolution is achieved by applying a partial scan technique (see Chapter 1, p. 40), which requires 180 rotation + fan angle. Since the fan angle for displaying the central portions of the scan field of view is low, rebinning of the data results in an effective scan time that is only slightly above 50% of the tube rotation time (RT) for the center of the scan field and increases for areas that are at a larger distance from the center of the scan field (effective scan time up to 60-70% of RT). There are differences in how partial scanning is implemented by the various manufacturers, which results in differences in temporal resolution between vendors. Various vendors use rebinning, others use a reduced fan angle limited to the area of the heart (FOV < 240 mm). The prospective trigger is derived from the ECG at a user -selected time within the RR inter0
773
774
23 Heart Fig.23.11 Types of ECG triggering and gating. Note the effect of a changing heart rate on the acquisitionwindow.
absolute delay
absolute reverse I
relative delay RRintervall Heart rate
I
I
I
I
I
I
I
I
I
1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
900 ms : 850 ms : 800 ms : 750 ms : 750 ms : 67 71 75: 80 : 80 :
val. It is always initiated by the R peak of the ECG but uses information from the preceding RR interval in order to calculate the trigger delay. The trigger can be selected by the user as a percentage (relative, usually between 40% and 80%) of the RR interval or as a fixed time after (absolute delay) the preceding R peak or before the next R peak (absolute reverse) (Fig. 23.11). Variations in heart rate make prospective triggering difficult because the position of the next R-wave (required for relative or reverse trigger delays) can only be estimated from the preceding RRinterval. Because of the shorter scan times in EBCT, triggering will usually be somewhat later than with single or multislice CT. The scan is usually triggered at 70-80% of the RR interval but may have to be started earlier (40%) at higher heart rates. With multislice CT and low heart rates, some 60-70% of the RR interval appear best suited. For higher heart rates, 30-40% of the RR interval may yield better results. In EBCT, a single section is acquired during each consecutive heartbeat and the table is moved in the time between consecutive scans. In general, a 3 mm collimation is applied, which lets the heart be acquired within some 30-40 s. In multislice CT a 4 x 2.5mm collimation is used (with Toshiba scanners, 4 x 2 mm or 4 x 3mm are available) and every second heart cycle is used for data acquisition (see Fig. 1.31). The table feed chosen is identical to the total collimation (4 x 2.5 mm = 10mm) or slightly less (8 mm) for overlapping sections. The heart can then be covered within some 24 heartbeats. With 16-slice scanners, 16x1.25 mm (GE), 16x1.5mm (Philips/Siemens), or 16xl-2mm (Toshiba) collimation are used, and the scan du-
ration may be reduced to 8-15 heart beats despite better spatial resolution. Almost motion-free images of the heart are obtained up to a heart rate of 60 bpm. Administration of beta-blockers is advisable to slow down the heart rate and prolong the diastole whenever such motion-free images are required (e.g., for the evaluation of morphologic abnormalities of the heart chambers). Up to 80 bpm the artifacts are still low enough to allow for coronary calcium scoring. It is therefore rare to use beta-blockers for this indication. In order to find out whether the temporal resolution is sufficient with triggering, the exposure time should be less than 40% of the RR interval (Table 23.4). Prospective ECG-triggering is used as the standard technique in EBCT, both for coronary calcium scoring as well as for coronary CT angiography. With multislice CT, prospective ECGtriggering is mainly applied for calcium scoring, but can be very useful also for evaluating heart morphology, e.g., in patients with congenital Table 23.4 Suggested tion of heart rate
temporal
resolution
as a func-
Temporal resolution
Heart phase
500ms
diastole
50 bpm
400ms
diastole
60 bpm
300ms
diastole
70 bpm
200ms
diastole
80 bpm
150ms
diastole
90 bpm
120ms
late systole
100 bpm
100ms
late systole
Heart rate
40 bpm
120 bpm
•
late systole
Single and Multislice CT
heart disease or with aneurysms, thrombi, or tumors, The advantage of triggering over gating is the lower radiation dose, because all data that are acquired are also used for image reconstruction, The differences in dose between EBCT and multislice CT are minor for this indication (see below), •
Retrospective
EGG Gating
In retrospective ECG gating, a continuous spiral scan is acquired with simultaneous ECG-recording (see Fig, 1.32), In order to obtain enough projectional raw data during each part of the cardiac cycle, oversampling with a low pitch factor (P'" =0,8-2) is required, which will necessarily increase radiation exposure to the patient The pitch is adapted to the heart rate, with higher pitch for higher heart rates (Table 23,5), The maxim urn pitch can be calculated according to the following formula: P*«Nl)x RT/(TRRX m) in which TRR is the maximum RR interval that is expected during the scan, N is the number of active detector rows, and m is the number of sectors for multisector reconstruction (m !EN), The user can prospectively or retrospectively select a gate during the RR interval from
Table 23.5 Suggested volume non of heart rate in ECG-gated
pitch factors as a funcmulti slice reconstruc-
tion.
art rat 40bpm 50bpm 60 bpm 70bpm 80 bpm 90 bpm 100 bpm 120 bpm
0.8-1.0' 0.9-1.3' 1.0-1.5' 1.0-1.5' O.B_1.Sb 1.0-1.Sb 1.0-1.Sb do not perform exam
Volume pitch factors
4.0' 4.0-4.8' 4.0-5.6' 4.0-6.0b 4.0-6.0 b 4.0-6.0b 4.0-6.0b 4.0-6.0b
(= slice pitch, helical pitch)
are provided for 0.5 s scanners. For faster I slower rotation RT, the numbers for P* have to be multiplied by RT/O.Ss. Pitch P ~- P" / number of detector rows Within the given range, numbers for P" vary be-
tube
tween manufacturers. a single sector reconstruction b multisector reconstruction
preferable preferable
which data for image reconstruction are used (Fig. 23.11). Gating can be performed with a relative approach (time delay as a fraction of the heart cycle) or with absolute approaches (either a fixed delay after the preceding R peak-absolute, or before the next R peak-absolute reverse). There are now more advanced techniques that use dual parameter cardiac gating, which takes into account that the phase of the slowest heart motion is more complexly related to the heart rate than defined by fixed triggering. This technique integrates heart rate, trigger latency, compliance parameters, and scan window length (effective scan time) to ensure that the same cardiac phase is imaged in every heart beat. The program can better adapt to dynamic changes in heart rate during the scan and also reduces the sensitivity to cardiac arrhythmia. Changes in heart rate occur as soon as contrast material is injected into the patient (unless beta-blockers have been administered). Various implementation techniques have been suggested, including single sector and multisector reconstruction, which use data either from one or multiple heartbeats. Multisector reconstruction yields the shortest effective scan time (theoretical minimum == 65 ms for 4 sectors and O.Ss rotation time) but also needs the highest dose to accomplish this goaL In principle, multislice CT could yield a higher temporal resolution than EBCr.. All techniques allow for overlapping reconstruction of sections, thus improving 3D imagmg of the heart. The various reconstruction techniques have been described in more detail in Chapter 1 (p.41). The pitch factor has to be adapted to the heart rate to ensure gapless data acquisition. With multisector reconstruction an even lower pitch has to be used (Table 23.5). Multisector reconstruction only works if new data can be acquired during every consecutive heartbeat. At unfavorable heart rates, exactly the same sectors will be acquired at a given rotation time, thus making multisector reconstruction much less efficient (see Fig.1.32b). For these reasons, some manufacturers allow for changes in rotation time in order to be able to scan successive sectors even at otherwise unfavorable heart rates. The problem with this approach is that the heart rate changes substantially during contrast material injection, unless beta-blockers are used. In addition, multisector reconstruction
775
23 Heart
substantially increases radiation exposure and is more sensitive to arrhythmia and changes in heart rate because data from consecutive heartbeats are used for data reconstruction. For this reason. it appears good practice to apply betablockers to stabilize the heart rate during contrast injection and to reduce it in order to allow for single or dual sector reconstruction in the majority of patients. For sector reconstruction, adaptation of the rotation time is a helpful technique.
Premedication All substances that increase heart rate should be avoided. The commonest one is caffeine in the form of coffee or various soft drinks, or as part of various combination drugs, e.g., against headache. Atropine and nitroglycerine should also be avoided (Table 23.6). Beta-blockers are a suitable tool to stabilize and reduce the heart rate. Most patients will present with a heart rate of some 80 bpm, which is particularly unfavorable for 0.5 s multislice scanners and dual sector reconstruction (see Fig. 1.32 b). Scanning patients without premedication and without adaptation of the rotation time to their heart rate will yield suboptimum results in a large number of cases. When contraindications have been ruled out, beta-blockers are recommended for all patients with a heart rate above 70 bpm, and may be advantageous in coronary CTA in patients with even lower heart rates because they reduce variations of heart rate during contrast injection. The goal is to reduce the heart rate to 60 bpm and below. With EBCT or multisector reconstruction with 16-slice scanners, reasonably good results are obtained up to a heart rate of 90 bpm and betablockers are far less often required. Many cardiac patients, however, will be on beta-blocker therapy as a standard medication applied by their cardiologist. However, the beneficial effect of beta-blockers in this patient group, is only of limited nature if they have taken their medication more than 3 hours before. Beta-blockers can be applied safely in the vast majority of normal patients and those with heart disease, unless the contraindications listed in Table 23.6 apply. In case of doubt it is recommended that the patient's cardiologist is consulted.
Table 23.6 Patient preparation and use of beta-blockers (modified from Becker et aI., 2001) Avoid
the following
substances
prior to cardiac
CT
Caffeine Atropine
Nitroglycerine Theophylline (IV), N-butylscopolamine Contraindications
(IV)
for beta-blockers
Bronchial asthma AVblock Severe hypotension Overt heart failure (consult cardiologist) Known intolerance to beta-blockers Beta-blockers
Obtain informed consent Metoprolol tartrate 50-100 mg orally, 30-90 rrm before scan, Esmolol (Brevi bloc) iruial bolus 500 u,mol/kg within 1 minute IV maintenance 50-1 00 u,mol/kg/mindirectly before scan
Beta-blockers can be applied orally, and then usually reach their peak blood concentration after some 30-90 minutes. As an (expensive) alternative, very short acting beta-blockers (e.g., esmolol) can be infused intravenously (best via an infusion pump), which will give their maximum action immediately and will lose it shortly after they are discontinued (Table 23.6). With this approach, monitoring of vital functions, ECG, and blood pressure is mandatory to avoid overdosing and adverse reactions.
Contrast
Material
Injection
For the evaluation of cardiac morphology, homogeneous enhancement of the cardiac chambers is important while coronary CTA requires good enhancement of the coronary arteries. These goals are not necessarily identical because there often is differential enhancement between the right and left heart. Coronary CTA
Coronary CTA has conflicting requirements. On the one hand, a high enhancement is the prerequisite for adequate evaluation of small vessels' on the other, distinction of calcified
Contrast Material Injection
plaques and vessel lumens becomes difficult if enhancement is too high. High intravascular contrast is especially important if image noise is high in obese patients or in low-dose scanning techniques. Partial volume effects. which are larger with thicker sections (3mm with some EBCT techniques. 2mm with some multislice scanners) may be partially compensated by higher contrast enhancement. Calcified plaque may be detected even on coronary CTA studies but calcifications need to have a higher attenuation than the contrast-enhanced vessel lumen. As can be derived immediately from the threshold level of 12ll HU that is commonly used for quantification of coronary calcium, coronary CTA will always need to use a higher threshold and therefore yield smaller scores. The differences will almost be negligible for higher calcium scores but in the low range, differences will be major. The difference between scores will be less with thinner sections (less partial volume effect). Some authors suggest aiming at a vascular enhancement of 250 HU for a compromise between calcium detection and vascular opacification. Monophasic contrast material injection appears most suitable for coronary CTA because the gradual increase of the height of the contrast plateau over time can be used to optimize enhancement of smaller vessels with craniocaudalscanning. Conrrasr material concentration is a controversial issue. Some authors have suggested using higher contrast concentrations (> 350 mg/ ml iodine) to improve enhancement at a given fiow rate and contrast volume. More highly concentrated contrast material, however, will be more viscous and always requires the use of a saline chaser in order not to lose contrast material in the injection veins towards the end of the scan. In addition, there will be major high-contrast streak artifacts in the right ventricle (see Fig.23.16b) which may seriously impair evaluation of the right coronary artery. For this reason, we suggest using a standard concentration of 300mg/ml iodine and adjusting the contrast volume and flow rates as needed. Scan duration with current EBCT and 4-row multislice scanners is in the range 30-45 s. It is substantially reduced with 8- and 16-slice scanners. A scan duration of 30-45 s requires an injection time that is in the same range. For this
reason, 150-200 ml of contrast material will have to be injected if a flow rate of 4mlls is to be maintained. Some authors therefore suggest lower flow rates in the range 2-3 mlls, which require contrast volumes of only 90-140 ml but also yield lower vascular contrast (see Table 23.7). Bolus timing can use either bolus triggering techniques or a test bolus injection. Bolus triggering should use the ascending aorta for a trigger region with a threshold of 50 HU and a trigger delay of 5 s. Bolus triggering has the advantage of being fast and quite reliable. Test bolus injection using 10-15 ml of contrast material followed by a 2Il ml saline flush at 4 mlls has the advantage that the actual scan delay can be tailored more precisely. We suggest extending the time to peak by 20% and using this figure as a start delay to compensate for the slower rise of contrast enhancement with reduced cardiac output, which would result in a prolonged timeto-peak of the enhancement curve. In addition, the maximum enhancement during this bolus test gives an indication of cardiac output (higher enhancement is due to less dilution and therefore a reduced cardiac output) and may help decide about the flow rate and contrast volume (lower flow rates with higher test bolus enhancement) . Our suggestion is to limit the total amount of contrast given to 2.5 mllkg body weight (0.75 g iodine/kg BW) if renal function is not impaired. Proper hydration of patients (e.g., oral fluids on the day of the examination) is advisable. The duration of contrast material injection should match the scan duration. We suggest a flow rate of 4 mlls, which should be altered depending on the scan duration (lower with longer duration), the size of the patient (higher rates with larger patients), and-if availablemaximum enhancement during test bolus injection (lower rates and lower contrast volumes with higher enhancement). Cardiac
Morphology
Compared to coronary CTA, a lower but more constant enhancement level through the heart is required for the evaluation of cardiac morphology. Monophasic contrast material injection is not ideally suitable for evaluating cardiac morphology because it leads to a gradual increase in intravascular contrast over time, until a maximum is reached and enhancement
777
23 Heart
rapidly decreases thereafter. This effect will occur at different points in time for the right and left side of the heart and will therefore make homogeneous enhancement of the heart chambers difficult, if not impossible, In addition, there is a rapid contrast uptake in the myocardium, which will decrease contrast between lumen and wall as the scan progresses, Biphasic injection protocols are more suitable for overcoming this problem, To reduce the high-contrast streak artifacts in the right atrium, contrast material concentration may be reduced while maintaining the flow rate during the second phase of contrast injection, This is cumbersome with single barrel injectors but is easily possible with double-barrel injectors by simultaneously injecting contrast material and saline at suitable flow rates, To maintain enhancement in the right side of the heart the injection time has to be prolonged by the right heart transit time of some 5-15 s. depending on cardiac output Because cardiac output is hard to estimate, we suggest determining the required injection time by increasing it over the scan duration by 10-15 s. The first phase of the biphasic injection should last some 10-15 s. A flow rate of 2,5-3 mlls is generally sufficient The duration of the second phase is calculated from the injection time, and a flow rate of 1.5-2 mlls is used, A saline chaser bolus of some 30ml is advisable but not mandatory because of the relatively low flow rates (Table 23,7), Imaging Tasks Cardiac Morphology
For the assessment of cardiac morphologic, ECG synchronization is necessary as soon as the ventricles have to be evaluated, There are not yet definitive recommendations as to which triggering to use but setting the trigger at 50% of the RR interval appears to be well suited for multislice CT and 70-80% for EBCT, given the heart rate is below 70 bpm. For the aorta, pulmonary artery, and the pulmonary veins ECG synchronization is usually not necessary, which means that single slice and multislice spiral CT provide good results, They are even preferable to ECG-synchronized techniques because radiation exposure is lower. Suggested protocols are given in Table 23.7, ECG triggering using EBCT or multislice CT is the preferred technique for evaluating the
morphology of the heart chambers, including the assessment of ventricular aneurysms, cardiac thrombi, cardiac tumors, and congenital heart disease, Suggested parameters are also provided in Table 23.7, Coronary Calcium Scoring
Since CT is very sensitive at detecting high-density structures, numerous techniques for the detection and quantification of coronary calcium have been suggested that range from non ECCsynchronized single slice scanning to ECG-gated multislice scanning and ECG-triggered EECT scanning. EBCTis the classical technique for coronary calcium scoring, The single slice mode with 100 ms scan time and 3 mm collimator setting is chosen, Data are acquired during every sequential heartbeat with an ECG trigger set at 40-80% of the RR interval. The standard was considered to be 80% but recently the 40% setting was shown to provide the best reproducibility, Images are reconstructed with a 512' matrix and a field-of-view (FOV) of220-260mm, Single slice spiral CT scanning without ECG gating can demonstrate coronary calcifications but the reproducibility is less than with ECG synchronization, In addition, it may be hard to rule out minor calcifications as a sign of developing atherosclerosis and potentially unstable plaques, Non-ECG synchronized dual slice spiral CT is a technique that has been widely studied for coronary calcium scoring, In principle, the same problems as with single slice spiral CT apply, Suggested parameter settings are also provided in Table 23.7, ECG-triggered multislice CT scanning should be the standard for multislice CT evaluation of coronary calcium, For this indication, a trigger level is set at 50% RR interval (may vary with higher heart rate), Most reproducible results are obtained if the heart rate is below 70 bpm. Beta-blockers are generally not applied but their use may improve reproducibility of measurements. Retrospective ECG-gating can further improve reproducibility with multislice CT scanning, Low dose settings are available, but the dose is still higher than with ECG-triggering techniques (by a factor of2 to 3), ECG-gating has also been suggested for single spiral CT scanning, The coronary calcium scores correlate well with those from EBCT for a threshold of 130 HU
Table 23.7
Suggested protocols for CTscanning of the heart
General Patient selection
No coronary CTAin patients with severe arrhythmias or frequent extrasystolic beats No coronary CTAin severely dyspnoic patients Consider MRAin young individuals or children (radiation exposure)
Preparation
Avoid caffeine, atropine, nitroglycerine Proper hydration (orally or intravenously) if coronary CTAis planned Beta-blockers for coronary CTAif heart rate> 70 b[)m
Oral contrast
None
Patient position
Supine with elevated arms
Scan range
Always as short as possible, tailored to the clinical problem (see Table 23.3)
Field of View
Standardized, e.g. 26 cm
Respiratory phase
Inspiration
Window setting
Calcium scoring: CTangiography:
Scan parameters
EBG SCfTF
W/L=400/40 W/L= 500/150 (adjust as needed) 4-sliceG SCfTFa
SW/RI
16-sliceG SCfTF'
SW/RI
Calcium scoring. cardiac morphology No ECGsynchronization Prospective ECGtriggering Retrospective ECGgating C
b
3/3 t
2-3/11-16 t 2-3/8-12 t 2-3/2-4 t
3/3 3/2 3/2
1-2/23-46 1-2/16-32 1-2/4-8/1-2
t t
b
3/3 t
1-2/6-12 t 2-3/8-12 t 2-3/2-4 t
1/0.7 3/2 3/2
0.5-1/8-16 t 1-2/16-32 t 1-2/4-8/1-21
1-2/0.7 2/1 1-2/0.7
b
3/2 t
1-2/4-8 t 1-2/1.3-2.5
2/1 1-2/0.7
0.5-1/8-16 1 0.5-1/2-5 1
1-2/0.5-1 0.5-1/0.3-0.7
t
2/1 2/1 1-2/0.7
Bypass grafts No ECGsynchronization Prospective ECGtriggering Retrospective ECGgating C CoronaryGA Prospective ECGtriggering Retrospective ECGgating C Contrast injection
Scan duration
1
monophasic V+N / F / DXd
Coronary CTA/ bypass grafts
20s 30s 40s ;,: 50 s
100 + 50/5/8 150 + 50/5/5 150 + 50/4/5 150 + 50/3/5
Cardiac Morphology
lOs 20s 30s 40s ;,:50s
70 + 50/5/10 A
biphasic V,/F,+V,/F,/DX
A A A A 50/5 + 30/3/8 A 50/5 + 60/3/5 A 50/5 + 90/3/5 A 45/4.5 + 105/2.7/5 A
SC= slice collimation (mm), TF= table feed (rnmjrotatlon), RI = recon. increment (mm), L = scanning direction, SW::::effective section width (mm). MPR= multiplanar reformats. axial = axial sections. cor= coronal. sag = sagittal CM = contrast material, Contrast concentration = 300 mglml iodine (optional 400 mglml iodine for 16-slice) V=volume of CM (ml), N =volume of saline flush (ml) F=flow rate (mils), D=start delay. X=trigger region • Parameters are given for 0.5 s rotation time RT. For faster I slower RT, the numbers for P' have to be multiplied by b sequential scanning r for retrospective gating the table feed (pitch) has to be adjusted to the heart rate (see Table 23.5) dread: 5-10 A= start delay 5-10s after reaching SO HU enhancement in the aorta e biphasic injection, e.g. 5015 + 30/3/8 A => SOml CM at 5 mils flow followed an additional SOml saline flush injected at 3 mils may be beneficial.
d.e
by 30 ml CM at 3 mils, delay 8 A d:
RT/0.5 s.
II
23 Heart
but, for low calcium scores, differences in assessment of patient risks may occur, The data are evaluated either using the Agatston score or by measuring total coronary calcium mass, The Agatston score is the traditional measure for coronary calcium, which uses a threshold of 121)HLJ to define calcified portions of the coronary arteries, The total area of each calcified region (larger than 1 mm' or 1-3 pixels) is weighted with a multiplication factor depending on the peak density in this region and summed to yield a score that is a representation of the total amount of coronary calcium (see also Fig, 23,34), Risk stratification based on the Agatston score are provided below in Table 23.11 (p, 803), Newer techniques involve the determination of the total volume of the calcified plaques or the total coronary calcium mass (derived from the plaque volume and the mean plaque density), Total calcium mass can be individually calibrated by using a calcium hydroxyapatite phantom that is placed in the scanning mat underneath the patient Total calcium mass appears to be most reproducible because it is less susceptible to partial volume effects than other measures of coronary calcium, In clinical practice, various programs are available that automatically calculate scores, The user has to identify potential plaques, assign them to the various portions ofthe coronary artery tree, and place ROls around them for further evaluation by the automatic program, In general a threshold of 121)HU is used, but a threshold of 90 HLJ has been suggested for non-EBCT evaluation, Coronary
CT Angiography
Coronary CT angiography is the most daunting imaging task in CT Coronary CT angiography using EBCTscanningwith ECG-triggering at 40-80% RR interval (depending on the heart rate of the patient) can cover the whole heart if overlapping scans with 3 mm collimation and 2 mm table feed are performed, The lOOms single slice mode is used, Proper training of patients is necessary, to ensure that they can hold their breath for the scanning duration of some 40-50 heartbeats, If the proximal portions of the coronary arteries only are to be evaluated at higher spatial resolution, a 1.5 mm collimation with 1.5 mm table feed has been suggested (trigger at 80% RR interval), This technique covers two-thirds of the heart in 40-50 sections and is feasible for patients not overly obese,
Three-phase electron beam angiography was developed to optimize imaging of the coronaries in patients with arrhythmia, With this technique, three sections are acquired during each heartbeat (minimum time difference 58 ms between phases), With multislice CT scanning, retrospective gating is mandatory, Multisector reconstruction yields the best results but requires markedly increased patient exposure, A collimation of 1-2 mm (depending on the scanner) can be applied with a pitch factor P between 0,2 and 0,375 (P*=0,8- 1.5 for 4-slice scanners, see Table 23,5), The optimum setting for ECG-gating depends on the heart rate of the patient For heart rates below 65, a gating at 60% of the RR interval usually yields sufficient results with single sector reconstruction (also less dose), In case of movement artifacts, further images at other intervals have to be reconstructed (start with 50% and 70 %, add more if necessary). For heart rates above 65 bpm, dual or multiple sector reconstruction has to be chosen, and the pitch has to be reduced accordingly (higher dose), Images should be reconstructed with 50% overlap relative to the effective section width and a FOV A 250 mm to ensure optimum 3D-reformations, It has been shown that the right coronary artery and the left coronary artery have different time intervals at which they are best demonstrated (40-50% for RCA, 50-60% for LAD, 60-70% for LCX at 250ms duration of data acquisition and 4-slice scanning), With higher heart rates, reconstructing more than one data set rarely can be avoided, A percentage reconstruction setting may not always be optimum, Some researchers suggest an absolute reverse setting with 550ms for the RCA and 450ms for the proximal LAD and LCX, Newer techniques involve dual parameter settings that take into account that the optimum phase at which images should be reconstructed shifts with heart rate (e.g. HeartBeat-RT, Philips), This technique improves visualization of the heart even in patients with (moderate) cardiac arrhythmia, Image Postprocessing Detection of stenoses on axial images is often problematic, Various postprocessing techniques have been suggested that simplify this task. The first step, however is to select the data set with the
least motion artifacts for the vessel segment of interest. Ifheart rate is low enough, one data set only may suffice for these postprocessing efforts. Good results are obtained with curved planar reformations (CPR) along each vessel segment (Fig. 23.12), preferably in at least two perpendicular planes to be able to detect and quantify stenoses. Vessel tracking can be used successfully if little motion artifact is present. Similar results are possible with manual tracking although this is a more time-consuming procedure. Standardized 5 mm wide thin-slab MIP reconstructed in an overlapping fashion every 2.5mm provide an alternative evaluation tool that can be more easily correlated with coro-
nary angiography if three MIP data sets are reconstructed that correspond to the standard angiographicviews: 30° RAO, 45°LAO, and45° LAO with 30° craniocaudad angulation (Fig. 23.13). Table 23.8 gives an indication which vessel segment can best be evaluated on which of these projections. One has to keep in mind, however, that these standard views in coronary angiography are frequently adapted to the individual anatomy, so that one-on-one correlation may only be possible if the precise angiographic angulations are known. Volume rendering techniques (VRT) are helpful for demonstrating of the outer shape of the heart chambers and great vessels and for displaying the anatomy of the coronary arteries.
b
c
a Fig. 23.12 CPR of the coronary arteries improves stenosis quantification. Comparison of a longitudinal section of the LAD parallel to the heart surface (a), perpendicular to the heart surface (b), and three transverse sections perpendicular to the vessel axis (c). Note that
a
b
Fig. 23.13 Optimum projection angles for thin-slab MIP of the coronary arteries for correlation with coronary angiography. Sections are prescribed on an axial image at the level of the mitral valve (see Fig. 23.2).
the eccentric lesion is well seen on (a) but is almost completely missed on (b) due to its eccentricity, which can best be appreciated on the transverse sections. The location of these sections is indicated on (a) and (b).
c Compare 30° RAO(a),45° LAO (b), and 45° LAO with 30° craniocaudad angulation (c). The numbering of vessel segments is in accordance to Fig. 23.4.
23 Heart
Identification of the various coronary branches is frequently easier on VRT images than on cross-sectional images or MIP displays (see Fig. 23.3). Some manufactures provide presets that allow for selective display of the heart (excluding the chest wall and pulmonary vasculature) , and even selective displays of the coronary artery tree have been proposed. Virtual angioscopy of the coronaries is a technique that can be successfully used to impress clinicians, but practical applications are few (e.g. evaluating complex branching patterns or coronary aneurysms). In addition, virtual angioscopic images suffer from blooming effects that will make calcified plaques look much larger than they really are. Assessment of Cardiac Function
Assessment of cardiac function requires at least two data sets (systole and diastole). While the diastolic image is reconstructed at 60-80% of the RR interval (depending on the temporal resolution of the chosen imaging technique), the end systolic image is obtained at 20-30% ofthe RR interval. In-detail analysis of heart movement requires reconstruction of multiple image sets during the whole RR interval in 5-10% steps. For analysis of cardiac functional parameters there are a number of commercial software packages becoming available for CT use (e.g. CTMASS, GE and CT CFA, Toshiba). The most important basic functional indices are summarized in Table 23.9. It should be noted that most norTable 23.8
mal ranges are derived from MRI data. Because the temporal resolution of cine- MRI is better than multislice CT, variations may be higher since endocardial contours are more difficult to draw on CT data and tend to blur more during phases of rapid movement. This can be expected to improve with 16-slice scanners and better multisector reconstruction algorithms. If a dedicated CT software package is not available, a simpler approach could be considered (arealength method ) that uses a short axis view at the base of the papillary muscle (roughly half the distance between the apex and the mitral valve annulus) to evaluate the cross-section of the ventricular lumen and determine its short and long axis diameters a and b (Fig. 23.14). A long axis view is used to measure the length of the ventricle c from the apex to the mitral valve annulus. A simple estimate of the ventricular volume based on the simplified assumption of an elliptical shape of the ventricles is: V=(a.b.c)/2
Data from systolic images (10-20% of the RR interval) and diastolic images (60-80% of the RRinterval) can then be used to estimate functional parameters. Dose
Dose has become an issue of substantial controversy between EBCT and multislice CT manufacturers. The discussion is complex because the users can vary dose arbitrarily and the recommendations of the manufacturers have been
Suggested projection angle for the various coronary artery segments (Becker et al., 2001) 30 RAO
45° LAO
---------45
LAO-30
interventricular groove
atrioventricular groove
long axis
Orientation
along LAD
between RCA and LCX
long axis
Segments through plane
RCA(1,3) LCX(11)
PDA (4) LCA(5) LAD (6-8)
RCA(2) LCX(13)
Segments in plane
RCA (2) PDA (4) LCA(5) LAD (6-8) LCX(13)
RCA (1-3) LCX(11) OM(12)
RCA(1,3) PDA (4) LCA(5) LAD (6-10) LCX(11) OM(12)
CC
LAD = left anterior descending coronary artery (6-10), LCA = left coronary artery (5), LCX = left circumflex coronary artery (11. 13), OM = obtuse marginal branch (12), PDA= posterior descending coronary artery (4), RCA = right coronary artery (1-3)
b
a
c
Fig.23.14 Simple functional data can be derived from the short axis view in diastole (a) and systole (b) by measuring the diameters of an ellipse centered on the
ventricular lumen. The longitudinal diameters are estimated from the corresponding long axis or four-chamber views.
Table 23.9
in cardiac function
Overview of basic quantitative
measurements
Definition
Ventricular Volume V (ml)
L (A.ndo- Apap) x SW
Myocardial Volume MV (ml)
L (A.pi - A.ndo + Apap) x SW
Myocardial mass (g)
1.055
Stroke Volume SV (ml) Ejection fraction
EF (%)
analysis based on short-axis views.
Normal Ranges a
Parameter
RVES54::':: 21 RVED138::':: 40
LVES40::':: 14 LVED121 ::'::34
RV 44::':: 10
LV 150 ::'::37
c
RV 46::':: 11
LV 158 ::'::39
c
VEO- YES
RV 84::':: 24
LV 82::':: 23
(VEO- YES)I VED
RV 61 ::'::7
LV 67::':: 5
x
b
MV
Regional wall motion
CLEO- CLES
Regional wall thickening
(WTEs- WTED) I WTED
Regurgitant volume
SVLV-SVRV
b
RV= right ventricle; LV= left ventricle; ES= end systolic measurement; ED = end diastolic measurement SW = section width of adjacent short axis views covering the whole ventricle A.ndo= area within endocardial contour; A.pi = area within epicardial contour; Apap= area of papillary muscles CL= centerline drawn between endocardial and epicardial contours WT = regional wall thickness measured as the distance between epi- and endocardial contours perpendicular to CL a Based on MRI data from Lorenz, J Cardiovasc Magn Reson, 1999 b Approximation based on Simpson's rule; contours preferably drawn on workstation (quantitative analysis software) C Includes ventricular septum
continuously reduced over the past years, both for EBCTas well as multislice CT.When comparing dose requirements, one has to consider image quality as well. Many of the studies published so far are flawed and have to be viewed with caution. The following aspects, however, are well established (Table 23.10): EBCTuses only a 210° sweep of the x-ray beam from below the patient (see Fig.23.9), which should eliminate direct exposure of the
breasts. Exposure of the thyroid is much less of an issue because the gland is usually kept outside the scan range and therefore is not directly exposed. At an identical CT dose index (CTDI), EBCT will have a slightly lower radiation exposure to the breasts than multislice CT. However, calculations of the effective dose in females did not show a smaller increase over males with EBCT as compared to multislice CT. Techniques that lower the exposure during anterior projec-
•
23
Heart
Radiation exposure in cardiac CT (12 em scan length). Note that the numbers for EBCT are fixed Table 23.10 63 mAs per sweep possible) but numbers for spiral and multi slice CT depend on the actual choice of parameters may vary substantially. Technique Coronary Calcium
"t-snce
CT
16-sliceCT EBCT Cardiac
SC
TF
kVp
mASeff
CTDlvol
prospective retrospective prospective retrospective prospective
triggering gating triggering gating triggering
4X2.5 4X2.5 16X1.5
10 3.7 24 6.7
120 120 120 120 130
36 133 36 133 63 f
2.9
prospective retrospective prospective retrospective prospective
triggering gating triggering gating triggering
120 120 120 120 130
60 200 60 200 63
4.8 16.0 4.0 13.3
3
10 3.7 24 6.7 3
retrospective retrospective prospective
gating gating triggering
4X1 16XO.75 3
1.5 3.7 2
120 120 130
300 300 63
29.4 24.0
16x1.5
3
10.6 2.4 8.9
and
0.6 I 2.2 1 0.5 1.9 1
Morphology
4-sliceCT 16-slice
ECG synchronization
scoring
(only
CT
EBCT
4x2.5
4X2.5 16x1.5 16x1.5
1.0 I 3.3 I 0.9 1 2.9
11!
1.01
CoronaryCTA 4-slice CT 16-sliceCT EBCT
SC= section collimation (mm); TF= table feed (mm); mASeff=mAs/P; E = effective dose (mSv) for males; dose is 20% higher for females The mAs and kVp settings refer to Siemens scanners (CTDI values Higher dose is needed for obese patients. Dose reduction by ECG-controlled tube current modulation With low kVp settings, further dose reci.Jction by 20-50% Ca-mass
has to be used instead
ofAgatston
= =45%. is possible
score for Ca scoring
tion angles of a tube rotation have been discussed for (non-cardiac) multislice CT but have hot yet been implemented. Older EBCT scanners with Xe detectors had a quantum efficiency that was some 30% less than the solid-state detectors used in multislice CT scanners. However, solid-state detectors are now also available for EBCT. EBCT scanners use post-patient collimation for 1.5mm sections and then increase dose by a factor of 2 as compared to the 3 mm collimation. Non-gated single slice, dual slice, and multislice spiral CTyield less noise at a given radiation exposure (CTDl) than sequential scans, if high enough pitch factors are chosen (see Fig. 5.12). This holds true in comparison to EBCT as well as to sequential single or multislice CT Thus, non-gated spiral scans (such as in CT A) in general require less dose for a given image quality than EBCT. ECC-triggered scans using EBCT or sequential multislice or dual slice CT use all acquired projection data for CT image reconstruction and
CTDlvoi=
volume
CTdose
6.1 5.0 ·1.54
I
I
index (mGy)
from www.impactscan.org).
in slim patients.
because
Lklder such conditions,
of higher attenuation
in calcified
total
plaques.
therefore should be similar in terms of dose (CTDl) requirements. Differences in image noise result from differences in detector efficiency, geometric efficiency, and from the choice offilter kernel for image reconstruction. Higher resolution kernels will cause more image noise (see Fig. 5.11). Thinner sections will increase image noise at a given exposure dose. This holds true for multislice CT as well as for EBCT Since most EBCT protocols use 3 mm sections while m ultislice protocols use 4 x 2.5 mm or even 4 x 1 mm collimation, noise with multislice CT may be similar (4 x 2.5 mm) or substantially higher (4 x 1 mm). At constant mAs settings, 1 mm sections in 4-slice scanners suffer from an approximately 30% higher dose because of decreased geometric efficiency as compared to 2.5mm sections (see Fig. 5.5 b). In 16-slice scanners, geometric efficiency is much higher; thus dose is not substantially increased for thin sections. As mentioned above, 1.5mm EBCT sections result in a 100% dose increase over 3 mm sections.
Fig. 23.15 Dose reduction in cardiac CT: ECG-adapted dose modulation reduces the mA settings during the systolic phase of the heart cycle. The time interval A TM at which the nominal tube current remains at 100% is chosen larger than the reconstruction time window in order to be able to retrospectively adjust the position of the reconstruction window. Since ATN is independent of the heart rate, the amount of dose reduction decreases with higher heart rates.
Tdel. mA , , Vl
100%
fl.T N
Reconstructed image data
,-, , ,
01 C
B « Q.) Vl
E Q.)
'-
20%
0%
, 'T del, Recon
Electronic noise from the detector is another limiting factor for low-dose applications and thin sections. Substantial differences between manufactures and scanner models appear to be present. In general, 16-slice scanners suffer less from electronic noise because of improved detector electronics. ECC-gated multislice CT uses continuous data acquisition and retrospectively selects the data used for image reconstruction. The higher the temporal resolution of this technique relative to the RR interval derived from the heart rate, the more projectional data have to be abandoned, and the less dose-efficient is the scan. If a 250ms reconstruction window is chosen at a heart rate of60bpm (1000ms RR interval), only one fourth of the data will be used. Hence, the dose at identical image quality is some four times higher than with an ECG-triggered multislice or EBCT technique. With multisector reconstruction of a 80 ms image at a heart rate of 90 bpm (667ms RR interval), the dose at
identical image noise has to be over eight times higher than with ECG-triggered techniques. The advantages of such techniques are a higher spatial and temporal resolution and the ability to reconstruct multiple images during the heart cycle, which would make more efficient use of the dose. Dose modulation techniques (Fig. 23.15) have been suggested that still use retrospective gating but reduce the rnA settings during systole, so that dose is reduced but the ability to retrospectively adjust the reconstruction window is retained with the possibility to reconstruct systolic images (oflower quality) for evaluating cardiac function. While these techniques are still experimental, some manufacturers have already implemented a reduction of the scan field of view to the center of the gantry (e.g., Toshiba). Together with proper positioning of the patient this technique can be used to reduce dose to the chest wall and breasts.
Artifacts •
Pulsation
Artifacts
Pulsation is responsible for incongruence during data acquisition that leads to blurring of structures, if motion during the phase of acquisition is minor, and to marked distortion, double contours, or complete lack of fine detail, if motion is major. While large structures such as the heart, aorta, and pulmonary artery can usually still be evaluated despite pulsation effects, more
detailed analysis of internal structures in the heart chambers (such as the endocardium, papillary muscles, and valves) and the coronary vessels becomes impossible. Strong pulsations can seriously hamper the evaluation of the aortic root and aortic arch (especially the region of a suspected ductus arteriosus). Such aortic pulsations are especially large in younger individuals because of higher elastic-
23 Heart
ity of the aorta. In such patients, alternative techniques such as MRA or ECG-triggered CT acquisitions should be considered. Pulsation artifacts can be used to estimate the motion of a cardiac structure or great vessel (see Fig. 23.10). For the coronaries, pulsation is the main reason for inconclusive examinations. With retrospective gating, image quality may be recovered by reconstructing data sets at various time points during the cardiac cycle but there is a considerable number of patients in whom evaluation of one or more coronary arteries (most frequently the RCA) is impossible with 4slice scanners. The number of conclusive examinations is considerably improved at low heart rates and with 8- and 16-slice scanners. •
Arrhythmia
Arrhythmia is a major cause for suboptimum results with prospectively triggered EBCT or multislice CT scans. Retrospective gating may compensate for changes in heart rate but extrasystolic beats can lead to incomplete or abnormal contraction that will cause images that are not in synchrony with the reset of the data set. Arrhythmia may be responsible for pseudostenoses of the coronary arteries or may obscure stenoses that otherwise would remain within the scan volume (Fig.23.16a). Patients with severe arrhythmias or frequent extrasystolic beats are not good candidates for ECG synchronization. While CT may
a
b
Fig. 23.16 Artifacts. Arrhythmia (a) may simulate stenoses or vessel interruption (arrow). Note the loss of the distal portion ofthe RCA due to breathing (arrow head). High contrast streak artifacts from the right atrium
adequately evaluate cardiac morphology, coronary CTA frequently does not provide diagnostic results. Coronary CTA should therefore not be attempted in these patients. •
Respiratory Artifacts
Artifacts due to respiratory motion during data acquisition are the second commonest problem with ECG-synchronized CT (in EBCT as well as multislice CT). Because of a total scan time that varies between 30s up to 50s with current scanners, proper patient training is extremely important so that breathing at critical moments is avoided. For larger ranges, such as bypass evaluation, caudocranial scanning may be advantageous, so that the heart is imaged first and regions like the upper mediastinum that are less sensitive to breathing are scanned last. Breathing artifacts can simulate vascular discontinuities, stenoses, or aneurysms depending on the direction and amplitude of the motion. Respiratory artifacts are easily recognized by the appearance of steps or undulations in the contours of the skin surface, the chest wall, or the heart. Artifacts appear at precisely the same table position. •
High-Contrast Artifacts
Highly concentrated contrast material in the superior vena cava can cause substantial streak artifacts that obscure the RCA and may make evaluation of the atria, the right ventricle, and the RCA difficult (Fig.23.16b). Streak atifacts caused
c simulate irregularities of this ectasia of the RCA (b). Contrast enhancement of the myocardium as well as loss of enhancement of the right heart may occur towards the end of the scan (c).
by a high concentration of contrast material in the superior vena cava (SVC) can mimic intimal flaps in the ascending aorta (see Fig. 24.5). Mixing of non-enhanced blood (from the IVC) and highly concentrated contrast material (from the SVC) occurs in the right atrium and causes the display of the atrial septum and tricuspid valve to deteriorate. It may be hard or even impossible to detect a right atrial thrombus under these conditions. High-contrast artifacts may be used as a diagnostic clue to detect a R-L shunt. e.g .. in patients with sinus venosus atrial septal defect (ASD; see Fig. 23.21). Using diluted contrast material reduces the effects somewhat. while highly concentrated contrast material (> 350 mg/ml iodine) and high flow rates (> 3 mlls) increases the artifacts. Inhomogeneous
Contrast
Because of sequential enhancement and de-enhancement of the right and left heart chambers, scanning at an early phase will yield less enhancement of the left heart, while later during the scan, contrast enhancement in the right heart starts to drop. If the timing of contrast material injection is not adjusted specifically to take account of these effects, evaluation of portions of the heart chambers may be difficult or even impossible. In addition, there is continuous contrast material uptake in the myocardium that will reduce the contrast between lumen and wall as the; scan progresses (Fig.23.16c).; As ;a consequence, 3D evaluation of the heart chambers or ventricular volumetry may be hampered. Partial Volume Effects
Partial volume effects are particularly troublesome for small details such as the more peripheral coronary artery branches. While an effective section thickness of some 3 mm (such as in EBCT or ECG-triggered multislice CT) is able to display the proximal coronary arteries, smaller and more peripheral vessels will be lost. Partial volume effects also affect the detection and apparent density of coronary calcifications. Because the Agatston score was developed for 3 mm thick EBCT sections, a similar section thickness should be chosen with other techniques as well, if present reference values are to be used for comparison.
Blooming of coronary calcifications is more severe with thicker sections, and thick sections also hamper the evaluation of soft plaques. Evaluation of coronary artery stents is only possible for stents that are markedly larger than the effective section width, which makes it difficult to use coronary CTA for more distally placed coronary stents. Image
Noise
Image noise poses a substantial problem in obese patients if low-dose protocols are used. This is independent of the scanning technique and affects EBCT as well as multislice CT scanning. Since EBCT does not allow for increasing the mAs settings per sweep (63mAs/lOOms), however, there is little opportunity to compensate for image noise, while multislice CT can simply apply a higher mAs setting. Image noise will lead to more substantial variations in pixel values (CT numbers). As a consequence for coronary calcium scoring, more pixels will exceed the thresholds of 130 HU and may be erroneously counted as a coronary calcification (compare Fig. 23.35). Using filters that require more than 2-4 of these pixels to be connected reduces the number of false positive classifications but also reduces the sensitivity for small calcified plaques. For coronary CT A, image noise can substantially hamper the evaluation of smaller arteries and may simulate wall irregularities due to random variations in pixel numbers. Superimposed
Structures
Cardiac veins will be less opacified than the coronary arteries if craniocaudal scanning is performed and contrast material timing was successful. Coronary veins, however, may be hard to distinguish from coronary arteries if both structures are opacified and the continuity of the coronaries is not adequately displayed due to pulsation effects. This mainly concerns the left circumflex branch that runs parallel to the left coronary vein, and diagonal and marginal branches that have a similar course as oblique ventricular veins. The left atrial appendage aimost invariably superimposes over the origin of the left coronary artery and the proximal portions of the LAD and LCX. The RCA runs in the right atrioventricular groove and may be superimposed by these structures. The diaphragm obscures the
23
Heart
undersurface of the heartland the sternum superimposes over the anterior surface of the heart. Proper editing with removal of these structures is necessary prior to 3D rendering. Automated techniques that separate the heart from the chest wall have been suggested and
even techniques that remove the ventricular lumen and isolate the coronaries are being developed. Thin-slab MIP are an alternative for evaluation of the coronaries (see Table 23.8, Fig. 23.13).
lenital Heart Disease
Standard spiral CT scanning is sufficient to evaluate those types of congenital heart disease that mainly involve the great vessels. However, it is only able to detect large abnormalities in the heart and lacks the ability to provide accurate quantitative information on these anomalies. ECG-synchronized scanning using EBCT or multislice CT can successfully evaluate cardiac morphology' but radiation exposure is a limiting factor in younger patients and makes CT unsuitable as a primary diagnostic study. Echocardiography and MRI are the techniques of choice for such indications but CT can be useful in patients with cardiac pacemakers (or other contraindications to MRI) and in adult patients with congenital heart disease. Ventricular Septal Defects
Defects of the ventricular septum (VSD) are the commonest congenital heart disease and account for some 30% of cases. Most are isolated but in approximately 20%, a VSD is associated with other anomalies. Depending on the size and location of the defect, most VSD close spontaneously in the first years of life. Functional problems of VSD are related to the size of the L-R shunt, which may even progress to Eisenmenger syndrome with severe pulmonary hypertension and shunt inversion. CT is not the technique of choice to detect or quantify VSD, but VSD may be a finding that is associated with other congenital heart defects that are evaluated with CT. •
Fig. 23.17 AV channel type ventricular septal defects (VSD) in a one-year-old child with D-TGA (low-dose scan: 80 kVp, CTDlvoi = 0.6 mGy).
Conal {supracristalj VSD (5- W%) are located
just underneath the pulmonary valve and the commissure between the right and left aortic valve cusp. The right aortic valve cusp may herniate into this VSD and lead to aortic valve insufficiency. A muscular VSD is located entirely in the sepal myocardium. Often multiple defects are present that close spontaneously in the first years of life. These defects are rarely examined by CT. The atrioventricular canal type [AV canal, endocardial cushion type, posterior) VSD is located close to the mitral valve and is best seen on axial sections. It is rarely an isolated defect. For this reason, it is frequently encountered on CT scans for other congenital heart disease (Fig. 23.17). ,
CT Morphology
Membranous VSD are most frequent (> 75%) and involve the membranous portion of the septum close to the right and posterior coronary valve cusp. They can best be appreciated on short -axis views as a defect in the posteriocranial portion of the septum.
Atrial Septal Defects
_
Atrial septal defect (ASD) is the commonest congenital cardiac anomaly in adults, and is often diagnosed late in adulthood because of its benign course. Females are four times more
During normal embryonic development, a thin fibrous septum primum forms as the left side of the later atrial septum with a physiologic opening (ostium secundum) close to its superior portion, followed by a muscular septum secundum on the right side that has a large physiologic defect in its mid-portion the foramen ovale (Fig. 23.18). The resulting valve-like opening present during fetal life closes after birth as the left atrial pressure rises and presses the fibrous portion of the septum primum over the foramen ovale. In most individuals the foramen ovale becomes permanently obliterated but in some 25% of cases it remains patent and may allow R-L shunting if there is a rise in right atrial pressure. It may be responsible for paradoxical emboli after acute pulmonary embolism. An ostium secundum ASD is most frequent (» 60 %) and represents a fenestration or absence of the foramen ovale flap of the septum primum. It may be associated with pulmonary valve stenosis, tricuspid atresia, a hypoplastic left heart, and an interrupted aortic arch. An ostium primum ASD (30% of cases) is a defect of the AV endocardial cushion that did not close during embryonic development. It is located inferior and ventral to the foramen ovale. It is associated with endocardial cushion defects (AV canal), cleft mitral valve and anterior fascicular block. A sinus venosus ASD is a defect of the atrial septum close to the entrance of the superior vena cava. It is associated in over 90% with an
aberrant return of the right upper lobe pulmonary vein. A Common atrium is an extreme form of ASD with very large defect of the septum primum and secundum. Only a small remnant of the septum is seen as a ridge along the roof of the atrium near the junction of the SVc. Lutembacher syndrome is the rare combination of ASD and mitral stenosis. The magnitude of the L-R shunt through an ASD depends on the size of the defect, the relative compliance of the ventricles and the relative resistances in the pulmonary and systemic circulations. The L-R shunt causes diastolic overloading of the right ventricle, up to three to six times increased pulmonary blood flow, and may finally result in shunt reversal {Eisenmenger syndrome}. As in VSD, CT is not the technique of choice for detecting or quantifying ASD, but ASD may be an incidental finding or associated with other congenital heart defects that are evaluated with CT. CT is well suited to demonstrate the aberrant pulmonary venous return in sinus venosus ASD. •
CT Morphology
A patent foramen ovale or an ASD may be suspected if the aorta enhances simultaneously or even earlier than the pulmonary artery on test bolus injections or on CTA of the pulmonary arteries. In patients with a patent foramen, the aortic enhancement drops again after an initial peak (Fig. 23.19). This peak is caused by the sudden rise in right atrial pressure and temporary
SVC-----
Pulmonary
Septum ---JJ primum Septum -r-rrr-rw; secundum
~~Abel
/'
pulmor
~
IVC Normal
Fig. 23.18 Embryonic development of the atrial septum. A thin fibrous septum primum forms as the left side of the later atrial septum with a physiologic opening (ostium secundum) close to its superior portion, followed by a muscular septum secundum, on the right side, that has a large physiologic defect in its mid-portion, the foramen ovale (modified from Dahnert, 2000).
Ostium primum defect
~ Sinus venosus defect
Ostium secundum defect
790
23
Heart Fig. 23.19 Paradoxical enhancement of the aorta is indicative of a patent foramen ovale and can be seen on CTA for pulmonary embolism. Early enhancement of the aorta (a) is followed by a drop in enhancement a few seconds later (b). Note the lack of pulmonary venous enhancement on the first scan (arrow), indicating that the contrast passed to the aorta via a R-L shunt. (c) Excellent display of the membranous portion of the atrial septum in a patient with known patent foramen ovale. No opening is seen as long as the left atrial pressure exceeds the right atrial pressure.
a
b
c
R-L shunting at the beginning of contrast material injection. It is followed by the later regular aortic enhancement. CT cannot directly visualize a patent foramen ovale (Fig.23.19c). An ostium secundum ASD is located in the central portion of the atrial septum and is best appreciated on axial sections (Fig. 23.20). An ostium primum ASD is located in the inferior portion of the septum and is also well visualized on axial sections. Because of the L-R shunt, the injected (highly concentrated) contrast medium only enters the left atrium if there is a large defect in the septum. Due to the increased blood flow there is marked dilatation of the right ventricle and the pulmonary arteries which not only involves the main branches but also the periphery. Paradoxical interventricular septal motion may be seen on (ECG-gated or consecutive non-ECG synchronized) sections of the interventricular septum. Sinus venous ASD occurs high in the atrial septum, near the entry of the superior vena cava and is frequently associated with anomalous connection of the right upper lobe pulmonary vein to the junction of the superior vena cava and right atrium (Fig.23.21a). In this type of ASD, contrast material from the IVC may enter the left atrium in ajet that can be demonstrated on axial sections as well as volume rendered images (Fig.23.21 b, c). Right ventricular overload may only be minor. Patent Ductus Arteriosus
The ductus arteriosus is a vessel leading from the bifurcation of the pulmonary artery to the aorta lust distal to the left subclavian arte .
Normal closure of the ductus immediately after birth is due to the sudden increase in oxygen tension that accompanies ventilation. It may present as isolated finding or be associated with other malformations (e.g., coarctation, VSD, pulmonary stenosis, or aortic stenosis). CT may demonstrate a patent ductus arteriosus (PDA) but is not the technique of choice to define shunt volume and pressure differences. PDA may be visualized in patients examined for aortic coarctation. •
CT Morphology
CT demonstrates a connection between the proximal descending aorta and the pulmonary artery bifurcation or proximal left pulmonary artery. This connection may be hard to visualize on single-slice CT because of marked pulsation
Fig. 23.20 Large ostium secundum atrial septum defect (ASD) in a patient with suspected chronic pulmonary embolism. Note the increased noise level and ring artifacts in this low dose multislice study (CTDlvoi= 2.4 mGy).
a
b
c
Fig. 23.21 Sinus venosus ASD with an associated aberrant pulmonary venous return from the right upper lobe (a) that enters the right atrium at the site of the small
defect (arrowheads). Note that there is ajet phenomenon (arrow) as highly concentrated contrast material enters the left atrium (b. c).
effects in the affected regions. With multislice CT, scanning is faster and artifacts are more spread out and thus allow for better detection of even small PDA (Fig. 23.22). Especially in patients examined for the presence of aortic coarctation, one should also search for a (small) PDA. Large PDA are obvious and are associated with enlarged pulmonary arteries and a dilated ascending aorta and aortic arch. In older patients, the PDA may calcify.
Transposition of the Great Arteries (TGA)
Coarctation of the Aorta
Coarctation of the aorta is discussed in detail in Chapter 24 (p. 872). Concurrent cardiac anomalies are rare with localized coarctation but are common with tubular hypoplasia.
Fig. 23.22 Small patent ductus arteriosus in a patient with an aortic coarctation (arrow).
The term transposition identifies a group of malformations that are characterized by abnormal relationships between the cardiac chambers and the great vessels. With complete transposition (D-TGA) , the aorta arises from the right ventricle and the pulmonary trunk emerges from the left ventricle. The position of the atria, ventricles, and valves is normal. This results in two separate circulations with some communication that is needed to sustain life. Patients present with a patent foramen ovale and PDA, or VSD, or another more complex cardiac anomaly. The creation or enlargement of an intraatrial communication provides better mixing of systemic and pulmonary venous blood. It may be created surgically by a Blalock-Hanlon procedure or by balloon septostomy during cardiac catheterization (Rashkind procedure). Corrective repair is accomplished by an intra-atrial baffle that directs the systemic venous return to the left ventricle and the pulmonary venous return to the right ventricle (Mustard and Senning operation). In patients with a VSD and a pulmonary stenosis, a corrective switch operation (Rastelli procedure) is performed to transpose the aorta back to its correct anatomic position on top of the aortic valve. The pulmonary valve is resected and substituted by an extracardiac prosthetic conduit to replace the pulmonary artery and valve. A corrected transposition of the great arteries (L-TGA) is characterized by an inversion of
23 Heart
the morphologic left and right ventricle with the AV valves following their respective ventricles. It may be associated with a membranous VSD, pulmonic stenosis and dextrocardia. In a double outlet right ventricle (DORV), most of the aorta and the pulmonary artery arise from the right ventricle. The aorta straddles a VSD. CT plays a role in adult patients with cardiac pacemakers, particularly for non-invasive detection of complications of corrective surgery (baffle stenosis). •
CT Morphology
L-TGA
A P
=
aorta
= pulmonary artery
DORV
double outlet right ventricle TGA = transposition of great arteries =
Typical patterns of the relationship of the ascending aorta and pulmonary artery are provided in Fig. 23.23. Fig. 23.23 Schematic representation ofthe position of In D- TCA, the aorta is located anterior (and the aorta and pulmonary artery in various forms of frequently to the right) of the pulmonary artery transposition of the great arteries (modified from Dahnert, 2000). (Fig. 23.24). The right ventricle has a normal trabecularization but is markedly hypertrophied with a thickened ventricular wall. After a Mustard or Senning procedure, there may occur a The commonest form ofsub aortic stenosis is stenosis of the systemic venous inflow that is the idiopathic hypertrophic form, also known as directed towards the mitral valve (see hypertrophic cardiomyopathy. The lesion conFig. 23.57). To avoid high-contrast artifacts, it is sists of a membranous diaphragm or fibrous advisable to use diluted contrast material for ring encircling the left ventricular outflow tract just beneath the base of the aortic valve. answering this specific diagnostic question. In L-TCA, the aorta holds the position norSupra valvular aortic stenosis is a localized or mally occupied by the pulmonary artery and diffuse narrowing of the ascending aorta, origivice versa. The aorta is connected to the mor- natingj ust above the level of the coronary arterphologic right ventricle that is located posteri- ies at the superior margin of the sinus of Valorly, in the position normally occupied by the salva. In contrast to other forms of aortic stenoleft ventricle. It is characterized by a marked sis, the coronary arteries are subjected to the wall hypertrophy and trabecularization. The elevated pressure and are often dilated and tortuous. Adherence of the free edges of the aormorphologic left ventricle is located anteriorly and is connected to the pulmonary artery. tic cusps to the sites of supravalvular stenosis In DORV, various types are distinguished, may interfere with coronary arterial inflow. Valvular pulmonic stenosis usually presents depending on the relative position of the aorta and pulmonary artery (Fig. 23.23). There is a in childhood and causes a dilated main and left concomitant VSD that can be easily detected on pulmonary artery. Sub valvular pulmonic stenosis may be infunaxial sections. dibular (such as in tetralogy of Fallot) or sub infundibular caused by hypertrophied anomalous muscle bundles (usually associated with VSD). Valvular Anomalies Supra valvular pulmonic stenosis is caused by A Valvular aortic stenosis occurs in 4 % of narrowing of the pulmonary trunk, pulmonary patients with congenital cardiovascular defects. bifurcation, or more distal arteries. It may be asMost children are asymptomatic, but the lack of sociated with various congenital heart defects as symptoms does not preclude the presence of well as with Williams- Beuren syndrome, Ehlersmoderately severe obstruction. Left atrial en- Danlos syndrome, and post rubella syndrome. CT is not the technique of choice for the largement, concentric left ventricular hypertroevaluation of congenital valvular disease unless phy and poststenotic dilatation of the ascending peripheral pulmonary stenoses are suspected. aorta are common findings.
a
Fig.23.24 Transposition of the great arteries (TGA). In D-TGA the aorta arises from the morphologic right ventricle (mRV) and the pulmonary artery from the mor-
•
c
b
CT Morphology
See below, p. 807.
phologic left ventricle (mLV) (a, b). Typical 3D representation obtained from a single-slice spiral CT data set (3/5/2) (c).
lar to the ventricular septum. The infundibular pulmonic stenosis is also well visualized on these sections. ECG-triggering is suggested for optimum demonstration of the anatomy.
Tetralogy of Fallot
Tetralogy ofFallot is characterized by an obstruction ofthe right ventricular outflow tract (usually an infundibular pulmonic stenosis), right ventricular hypertrophy, a VSD, and an aorta overriding the interventricular septum. It may be accompanied by a bicuspid pulmonic valve, a stenosis of the left pulmonary artery, a right aortic arch, and anomalies of the coronaries and the chest skeleton. If combined with an ASD it is termed pentalogy ofF allot. A trilogy ofF allot consists of a pulmonary stenosis, right ventricular hypertrophy, and a patent foramen ovale. Palliative treatment consists of a BlalockTaussig shunt between a subclavian artery and a pulmonary artery, a Pott operation with anastomosis of the left pulmonary artery to the descending aorta, or a Waterston-Cooley procedure with anastomosis of the right pulmonary artery to the ascending aorta. Definitive treatment consists of correction of the VSD and resection of the obstructing tissue from the RVOT. CT may be used in patients with pacemakers to define morphology but it is not suited to provide functional information. •
CT Morphology
A VSD with the typical overriding aorta is best seen on semicoronal reformations perpendicu-
Ebstein Anomaly
In this rare anomaly there is abnormal morphogenesis of the tricuspid valve. Portions of the septal and posterior leaflets of the tricuspid valve are attached lower than normal to the ventricular wall rather than to the atrioventricu1ar ring (atrialization of the RV inlet). The valve is incompetent, the foramen ovalc is patent, and the right ventricle variably hypoplastic. •
CT Morphology
Ebstein anomaly presents with an extremely large right atrium that extends partially into the superior portions of the right ventricle. Because of tricuspid insufficiency the SVC and IVC are enlarged and contrast material injected via an arm vein may retrogradely enter the IVC and liver veins. The functional right chamber is small and displaced inferiorly (Fig. 23.25). Complex Anomalies
Complex anomalies usually represent combinations of the defects described above and may be associated with various other congenital diseases.
23 Heart
Fig. 23.25 Ebstein anomaly with typical atrialization of the right ventricle due to distal insertion of the tricuspid valve and consecutive massive dilatation of the right atrium and bulging of the atrial septum (a). Note the narrow pulmonary arteries (b). There was reflux of contrast material into the dilated hepatic veins (c), a sign that is indicative of impaired right ventricular function.
a
•
b
CT Morphology
Analysis of complex anomalies is based on the segmental anatomy of the heart (see p. 761). Morphologic and functional heart chambers need not be identical. Levocardia and dextrocardia (see Fig. 23.41) are defined by the direction of the heart axis. The morphologic right chamber is characterized by its increased trabeculation and longer outflow tract. The right atrium is defined by its connection to the inferior vena cava. The bronchial anatomy can be helpful in defining left or right isomerism because the bronchial situs is usually concordant with the atrial situs. Complex anomalies of the great vessels include totally anomalous pulmonary venous re-
turn, truricusarteriosus, pulmonary atresia, pulmonary vein atresia, hypoplastic left heart syndrome, and interruption of the aortic arch. Atrioventricular anomalies are manifold. They include concordant and discordant anomalies, unilateral AV atresia, double-inlet single ventricle, and unusual alignment with crisscrossing or overriding great vessels. Ventriculoarterial anomalies include transpositions, atresia and great vessel straddling over a VSD. Atresia of the pulmonary artery is associated with dilated systemic (usually bronchial) supply. Concomitant valvular incompetence is important but very hard to evaluate by CT.
Acquired Diseases oftheMyoca Valvular disease, cardiomyopathies, and coronary heart disease are associated with known morphologic changes in the cardiac chambers. CT is not indicated as a primary imaging study but may assume increasing importance in assessment of these disease entities. Cardiomyopathy
Congestive cardiomyopathy can be caused by ischemic coronary artery disease, myocarditis, endocardial fibroelastosis, metabolic diseases (glycogenosis, mucolipidosis, mucopolysaccharidosis), and muscular dystrophies, and is seen in children of diabetic mothers. Hypertrophic
obstructive
cardiomyopathy
(HOCM) can be due to autosomal dominant
transmission causing idiopathic hypertrophic subaortic stenosis (IHSS) and asymmetric septal hypertrophy (ASH). Symmetric concentric hypertrophy is a rarer entity that may be diffuse, apical or midventricular. Restrictive cardiomyopathy can be caused by amyloid, glycogen deposits, or as a consequence of hemochromatosis. It may also be due to constrictive pericarditis. Echocardiography is the technique of choice for evaluation of these entities. •
CT Morphology
Congestive cardiomyopathy is characterized by
cardiomegaly with global heart enlargement the (mild) left atrial enlargement is not sociated with enlargement of the left atrial pendage. There is global hypokinesis with
but asapre-
Ischemic Heart Disease
duced pulsation artifacts on axial sections or MPR of non-ECG-gated scans (Fig.23.26a). The ventricular wall is not thickened. In hypertrophic obstructive cardiomyopathy (HOCM) size ofthe ventricular cavities is normal to small with a marked increase in wall thickness (Fig.23.26b). IHSS and ASH are characterized by a markedly asymmetric thickening of the ventricular septum as compared to the free left ventricular wall. Restrictive cardiomyopathy caused by amyloid or glycogen deposits is characterized by left ventricular wall thickening and reduced LV systolic and diastolic function (Fig.23.26c). In amyloidosis, pulmonary congestion and pulmonary deposits of amyloid may be present (patchy infiltrates). Constrictive pericarditis is characterized by a thickened (> 2 mm) and often extensively calcified pericardium. Ischemic Heart Disease
Cardiovascular diseases are the most frequent cause of death in the western world. Classical risk factors include age, (male) gender, total cholesterol levels, increased LDL, decreased HDL-cholesterollevels, high systolic blood pressure, treatment for hypertension, cigarette smoking, and diabetes. Newer independent risk factors include increased fibrinogen, increased factor VIlc, elevated homocysteine, CRP, and increased PAI-1. These factors can be used for risk stratification. The classical risk factors have
a
b
Fig. 23.26 Cardiomyopathy. Congestive cardiomyopathy with massive dilatation ofthe heart chambers and thinning of the LV wall (a). Hypertrophic cardiomyopathy with substantial thickening of the left ventricular
been established as a tool for deciding between various management options ranging from dietary changes to statin therapy or the need for further testing. Acute ischemic heart disease includes new onset of cardiogenic chest pain, myocardial infarction, and sudden cardiac death. Myocardial infarction and sudden death are the most frequent initial presentations of cardiovascular disease. It has been shown that the majority of these patients do not have a significant coronary stenosis (» 50%) prior to their event, which underlines the need for early detection by risk assessment and screening of appropriate patient groups. . .• , , : . /Chronic ischemic heart disease leads to changes in myocardial perfusion, myocardial kinetics, and ultimately myocardial morphology. As a consequence of acute infarctions, wall thinning, aneurysm, and scar formation, and various forms of dyskinesia of the myocardium may result. Ischemic cardiomyopathy is the consequence of repeated acute or chronic ischemic injury to the heart. ECG, stress tests, and (dobutamine) thallium scintigraphy or MRI are the techniques of choice for evaluating patients with increased risk factors or symptomatic ischemic heart disease. CT at present has no role in acute coronary syndromes but may be used to rule out extracardiac diseases in patients with atypical angina. Coronary calcium scoring by CT (see below, p. 800) allows for risk stratification but it is not yet clear whether it provides estimates
c wall even in diastole (b). Restrictive cardiomyopathy due to amyloid deposition (c). Note the hypoattenuating subendocardial region (arrowheads) and an increased thickness of the myocardium.
795
23 Heart
that are independent from classical risk factors. In patients with chronic ischemia, CT holds the potential to demonstrate scars, wall thinning, dyskinetic wall segments, and ventricular aneurysms. These findings may also be accidentally detected by conventional (non ECG-synchronized) CT but are commonly underreported. Although perfusion studies for myocardial viability have been performed with CT, scintigraphic techniques and MR currently are more suitable for this imaging task. •
CT Morphology
Acute infarction will lead to ischemia, myo-
cardial damage, and ultimately scar formation. On late scans (10-40 minutes) after contrast material application, the damaged tissue has been shown to become hyperdense because of increased interstitial trapping of the contrast agent 0, " ,. _ , A, .• , Formation of subendoeardial scars can be appreciated as a hypo attenuating rim in the endocardial region of the ventricles (Figs. 23.27). Subendoeardial, but more frequently transmural infarction leads to focal thinning of the ventricles (Fig.23.26b) with regional disturbance of wall movement. Ultimately, aneurysms or thrombi may form (see Figs. 23.29). The various portions of the myocardium can be assigned to the different coronary vascular territories (see Fig. 23.5) but there is some insecurity at the base of the heart depending on whether the right, left, or both coronary arteries I
a
•••
'-
••
,,_
•••
,,;
_
b
Fig. 23.27 Ischemic scars lead to thinning of the ventricular wall following transmural infarction or to subendoeardial hypoattenuating regions after contrast application, (a) Massive thinning of the ventricular septum and apex (transmural scar) due to infarction of the
supply the posterior descending coronary artery (PDA). In general, however, the LAD supplies the anterior left ventricular wall including the superior portions of the septum, the roof of the left ventricle, the apex, and the anterior papillary muscle. The left circumflex artery supplies the lateral wall and the posterior papillary muscle. The right coronary artery supplies the right heart, and in the majority of patients (via the PDA) also the inferior left ventricular wall and the inferior portions of the septum. Disturbed wall motion may be appreciated by functional analysis of ECG-gated data sets, reconstructed m various -cardiac phases; akinetic wall segments show no movement, hypo kinetic segments have a reduced wall motion, dyskinetic segments show a paradoxical systolic expansion, and asynchrony describes a disturbed temporal sequence of contraction. On VRT images, thin portions of the wall appear more translucent and bright on both systolic and diastolic images (Fig. 23.28). Cardiac Aneurysms
Congenital left ventricular aneurysms are rare and occur in young black adults. The sub mitral type leads to a bulge at the left middle or upper cardiac border while the subaortic type may be small and hard to visualize or lead to massive cardiac enlargement due to aortic insufficiency. The risk of rupture is usually low.
c LAD. Note the subendoeardial extent of the scar in the common territory of LAD and LeX (arrows), (b) Transmural scar in the territory of the LAD with persisting subendoeardial hypodensity. (c) Subendoeardial scar at the anterior wall of the right ventricle.
Card'iacAneurysms 797 Fig. 23.28 Hypokinesiaofthe anterior wall of the left ventricle demonstrated on volume rendered images in systole (a) and diastole (b). The thin portions of the myocardium reveal the bright intraventricular contrast on both, systolic and diastolic images. Images were acquired at 20% and 1O%ofthe RR-interval.
b
a
Acquired left ventricular aneurysms are a complication of transmural myocardial infarction or, rarely, Chagas disease. True aneurysms locally involve the ventricular wall, which becomes noncontractile and dyskinetic. False aneurysms (pseudoaneurysms) are due to a left ventricular rupture, which is contained by fused layers of visceral and parietal pericardium (after pericarditis) as well as extra cardiac tissue. It may be caused by transmural myocardial infarction or trauma and has a high risk of delayed rupture. Echocardiography and cardangiography are the primary imaging modalities but ECG-gated multislice CT may offer similar information.
a
b
Fig. 23.29 Cardiac aneurysms. (a) Aneurysm of the cranial portion of the left ventricle. Concomitant pericardial effusion, (b) Ischemic cardiomyopathy leading to formation of an aneurysm of the apex of the heart, (c)
•
CT Morphology
A true cardiac wall aneurysm often causes only a slight bulge in the heart contour, but there is always marked, localized thinning of the myocardium compared with adjacent areas (Fig. 23.29). On ECG-gated scans, reconstruction in systole demonstrates a characteristic paradoxical expansion. Rarely, a rim of calcium is seen in the fibrotic wall. There is always a wide communication with the heart chamber (no neck). A cardiac pseudoaneurysm may have a globular appearance with a mouth that is smaller than the actual diameter of the aneurysm. It is typically located at the posterolateral/diaphragmatic wall of the left ventricle. Due to delayed filling the contrast enhancement may not be identical to that of the ventricular lumen.
c Extremely calcified aneurysm ofthe left ventricle (4 x II 6). Note the internal mammary graft (arrow) and the reduced pulsation of the aneurysmatic portions of the ventricle.
23 Heart
Cardiac Thrombi
Cardiac thrombi can form at the site of myocardial injury due to infarction (or, rarely, tumors), and in regions of reduced or turbulent blood flow. The latter may be the consequence of tricuspid or mitral stenosis (or other valvular disease) and then most commonly affects the atria, including the atrial appendages. Thrombi may also form in regions with reduced wall motion, and thus may coat the internal surface of ventricular aneurysms. Cardiac thrombi hold an increased risk for peripheral embolization, either into the systemic circulation (from the left heart) or into the pulmonary arteries (right heart). In fact, the presence of peripheral emboli should prompt the search for a source within the heart or the
Fig. 23.30 Intraventricular thrombi, (a) Wall-adherent thrombi frequently form in the region of a myocardial infarction. (b) Round thrombi in the left ventricle may be a source of small peripheral emboli. (c) This large thrombus presenting as a mass in the right ventricle was found incidentally on conventional spiral CT of the ab-
a
b
Fig. 23.31 Intra-atrial thrombi, (a) Thrombus in the left atrial appendage (LAA) seen on a low-dose multislice CT (CTDlvoi= 2.2 mGy) of the pulmonaryvessels, (b) The patient also had an embolic occlusion ofthe left iliac artery seen on the corresponding CT venogram. (c. d)
aorta, or for an open foramen ovale (paradoxical emboli) . Echocardiography is the present technique of choice. CT is superior for detecting thrombi in regions that are hard to assess by transthoracic echocardiography. •
CT Morphology
Cardiac thrombi frequently form broadly walladherent hypoattenuating structures that are best detected after contrast administration. This is the case in the ventricles, in particular in areas of (old) infarction or ventricular aneurysms. Older thrombi may partially calcify (Fig.23.30a). Calcifications may also be seen in the regions of scar tissue, in particular in an endocardiallocation.
domen. Note that the myocardium on this portal phase scan is nearly isoattenuating to the heart chambers, (d) Large embolus in the right atrium that extended through the right ventricle and ended in the main pulmonaryartery.
c
d
Thrombus in the right atrial appendage (arrow) that is hard to see during the arterial phase (c) but better appreciated on a delayed phase scan (d) in this patient with an anastomotic leak (arrow head) after aortic surgery.
Congenital Anomalies
Some thrombi detach and then may appear as a mass-like intraventricular filling defect that can simulate a tumor (Fig.23.30b). As opposed to tumors (compare Fig.23.44b), thrombi do not take up contrast even on delayed images after 2-5 minutes. /\ ' Wall-adherent thrombi in the atria are less frequent. Thrombotic material may accumulate in the atrial appendages, especially in the left atrium in patients with marked atrial dilatation secondary to mitral stenosis (Fig. 23.31). Concomitant infarctions of brain or internal organs as well as occlusion of peripheral arteries may be seen (Fig.23.31b). Arrhythmogenic Dysplasia
Right Ventricular
Arrhythmogenic right ventricular cardiomyopathy or dysplasia may lead to life-threatening ventricular arrhythmias. There is a congenital disposition toward fatty infiltration of the myocardium of the right ventricle leading to changes in conduction (left bundle branch block) and development of arrhythmogenic zones. Typical locations include the ventricular apex, the RV outflow tract and the inferior wall below the tricuspid valves. The dysfunction may lead to variable ventricular dilatation and dysfunction with reduction of ejection fraction.
Fig. 23.32 Arrhythmogenic right ventricular dysplasia (ARVO) with fatty infiltration of the right ventricular wall (arrowheads) and a pericardia I effusion.
•
CT Morphology
CT can show abundant epicardial adipose tissue, RV dilatation, prominent trabeculation of the right ventricle with low attenuation, and a scalloped appearance of the free wall (Fig. 23.32). Intramyocardial fat deposits may be seen in advanced cases when image quality is good. Cine imaging and functional studies may reveal akinetic or dyskinetic areas with bulging in diastole.
Coronary Arteries, Congenital Anomalies
An anomalous left coronary artery from the pulmonary trunk may lead to postnatal infarction or, if sufficient collaterals to the right coronary artery exist, to a flow reversal in the LCA. With massive collaterals, there will be a L-R shunt with volume overload of the heart. Anomalous coronary arteries with a common trunk may take a course between the aorta and the right atrium or between the aorta and the pulmonary outflow tract. Such arteries may be at an increased risk for injury during median sternotomy. ECG-synchronized CT (EBCT or multislice CT) plays a role in those cases in which the precise location is hard to determine on coronary angiograms.
A coronary artery fistula is defined as connection between the coronary arteries and other heart structures. Most commonly (>90%) the fistula drains into right heart structures such as the right ventricle, atrium, pulmonary trunk, coronary sinus, or superior vena cava. As a result, a L-R shunt develops with concomitant cardiomegaly only in large shunts. CT may define the precise anatomic location although cardangiography remains the technique of choice. Muscular bridges are most frequent in the proximal portion of the LAD but may be seen almost anywhere in the coronary artery tree. They lead to a characteristic systolic "milking" on coronary arteriograms with compression and stenosis of the affected vessel.
799
23
Heart
•
CT Morphology
CTA demonstrates a dilated main coronary artery with an anomalous connection. An anomalous left coronary artery arising from the right coronary artery may either take a path to the left between the pulmonary outflow tract and the aorta to the LAD and then the LCX (Fig. 23.33), or it will run between the right atrial appendage and the aorta to the LCX and then the LAD. Similar anomalies are possible if the right coronary artery arises from the left ostium. Two separate ostia for the LCX and the LAD are possible. Coronary artery fistulae may be demonstrated by ECG-synchronized CT. They present as tortuous dilated vessels that are easily seen on axial sections. It is important to determine into which structures the fistula drains. Volumerendered images provide a good anatomic overview. Muscular bridges present as a (short) portion of the coronary artery that is surrounded by soft-tissue dense material (myocardial muscle). The intramuscular course can best be proven by MPR perpendicular to the vessel course. The proximal LAD is most frequently affected. If systolic images of sufficient quality are available as well, thinning of the affected portion of the vessel may be noted.
Calcium Scoring
Calcification in the arterial wall is part of the development of atherosclerosis and is absent in normal vessels. Coronary calcium as an indicator of atherosclerotic coronary artery disease (CAD) has been known for along time. EBCThas substantially increased the sensitivity of radiological methods for the early detection of coronary calcium and therefore has been used both as a screening tool in asymptomatic patient groups as well as in symptomatic patients, to establish the presence and estimate the risk of coronary artery disease, and thus to estimate the risk of acute cardiovascular events. There is an ongoing discussion as to the clinical value ofEBCT and its use in screening and following patients with CAD. Recently, single, dual, andmultislice (spiral) CThave also been suggested for this purpose. Atherosclerotic (cholesterol) deposits have been demonstrated in a small portion of adolescents and young adults with an increasing prevalence with age. In some individuals, coronary calcium may be seen already in the second and third decades of life but it is not an inevitable aspect of aging. Calcium hydroxyapatite is thought to precipitate in atherosclerotic arteries via an active mechanism that is similar to bone formation and breakdown. The amount of coronary calcium is closely related to total plaque burden, although calcifications represent only the tip of the iceberg. The volume of calcified plaques has been estimated to make up 20% of total plaque burden. There may even be individual patients with no calcifications but vulnerable plaques that are prone to rupture. •
Fig. 23.33 Common right coronary artery trunk with left coronary artery running between the aorta and the right ventricular outflow tract (arrowheads).
CT Morphology and Technical Aspects
TheAgatsron score is traditionally used to express the burden of calcification. It is based on a sliceby-slice analysis of EBCT sections. An arbitrary value of 130 HU was chosen to separate calcifications from high-density pixel noise. This threshold was based on late 1980s EBCTtechnologyto be two standard deviations above the CT number of blood. A (vendor-specific) pixel filter can further eliminate singular noise-related pixels by determining the minimum number of connected pixels (usually 1 -4 pixels, or 2 mm') that are included in the evaluation. With current evaluation software, candidate plaques are highlighted on the images (Fig. 23.34). On each section, the user places an ROI around a group of plaques that is located in
Calcium Scoring Fig. 23.34 Agatston score. Each plaque is encircled by a region of interest (ROI). The calcium score of each plaque is calculated from the area of all pixels above 130 HU multiplied by a weighting factor that depends on the maximum density of the plaque. The Agatston score is the sum of all individual plaque scores.
801
Agatston Score: Ag
= L: Area. x C ROI,
I
I
Cofactors Cj: c, = 0 if CTj,max < 130 HU Cj = 1 if CTj,max = 130 199 HU c, = 2 if CTj,max = 200 299 HU Cj = 3 if CTj,max = 300 399 HU Cj = 4 if CTj,max ;:: 400 H U
the course of a coronary artery, and the program calculates the area on each section and multiplies it by a weighting factor that depends on the maximum pixel number in the ROI (Fig. 23.34). The sum of individual scores yields the total calcium score (Agatston score).
is not yet commonly employed and does not work well for obese patients (Fig. 23.35 b). While there is an excellent correlation between single (and dual) slice CT for a threshold of 130 HU. values for 90 HU differ slightly more. A calcium volume score (CVS) has been developed by Callister (1998) to improve repro• Artifacts ducibility over the Agatston score. It uses interCare has to be taken not to include calcifications polation between sections and determines the volume of calcified plaques. The CVS was used of the aortic bulb at the orifice of the right or left coronary artery (Fig. 23.35 a). Calcifications successfully to demonstrate regression of calof the mitral valve annulus should not be miscified plaques in patients treated with HMGtaken for LCXcalcifications (Fig. 23.35 c). CoA reductase inhibitors. Motion artifacts should be included in the The total coronmy calcium mass uses volumetric and density information and a calievaluation ROI to improve reproducibility (Fig.23.35 c). bration phantom of hydroxyapatite (as in bone Image noise is a limiting factor for EBCT. mineral density measurements) to calculate the With spiral CT scanners, a lower threshold of actual mass of the calcified plaques. It should be least influenced by technical factors such as sec90HU yields a higher accuracy (increased sensitivity for the detection of calcium at the cost of a tion thickness and partial volume effects and slightly reduced specificity), but this technique has a high reproducibility.
a
b
Fig. 23.35 Artifacts for calcium scoring: mural calcium in the aortic wall at the orifice of the coronary arteries must not be included (a). Image noise in obese patients is especially cumbersome if the threshold for calcium detection is reduced from 130 HU to 90 HU (b). Pulsa-
c tion artifacts can blur a plaque (c) but should nevertheless be included in the evaluation ROI to improve interobserver variation. Note the calcification in the region of the mitral valve (arrow).
•
23
Heart
New CT workstations provide the traditional Agatston score as well as newer volumetric measurements (CVS, coronary calcium mass) independent of the CT technology used for data acquisition. They also correct for differences in section thickness and examination technique and therefore hold the potential to further improve reproducibility of measurements. Comparison
of CT Techniques
With the advent of spiral and multislice CT scanning there is now a heated argument about the most suitable CT technique. There is, however, an excellent correlation between the various CT techniques, independent of whether ECG synchronization was used or not. The most important differences arise in patients with a low calcium score, and regarding the reproducibility for follow-up examinations. Motion artifacts may be accepted to a certain degree for quantification of coronary calcium (Fig.23.35c) but are disturbing if high reproducibility is required for follow-up examinations. EBCT has the highest temporal resolution of lOOms. The temporal resolution with single or dual slice spiral CT is above 500ms. With 0.5 s multislice CTscanners, temporal resolution is 250-320 ms with prospective triggering and can be further reduced with retrospective ECG gating. Image noise reduces the accuracy for detecting and quantifying small amounts of coronary calcium (Fig.23.35b) and also reduces reproducibility of such measurements. EBCT uses a fixed setting of 63 mAs that assures low radiation exposure but may be insufficient for more obese patients. The various other CT techniques can adapt mAs settings to the clinical requirements and thus ensure reasonably low noise levels at similar radiation exposure for slim patients. This individual adaptation, however, is not always used in present practice. In EBCT the noise problem has been tackled with repeat examinations and averaging the results, which increase radiation exposure as well. EBCT uses sequential 3mm sections. This may reduce reproducibility of measurements if calcified plaques are sectioned differently at follow-up and the Agatston score is used. Plaque volume (CVS) and mass are more robust measures for this purpose. Spiral techniques can use overlapping reconstruction, which5 also can improve reproducibility.
Radiation exposure with EBCT is low but can be kept in the same range with other CT techniques if scan parameters are adjusted accordingly. Only retrospectively gated (single or multislice) techniques inherently require higher radiation dose (see also p. 784). Availability of spiral and multislice CT is more widespread and there are less hardware costs involved with EBCT. To ensure maximum reproducibility, the same techniq ue (scanner, acq uisition techniq ue, kVpsettings,reconstructionkemel,FOV,filtered back projection methods, and scarmer calibration) should be used for follow-up. Data Interpretation
Prevalence of calcifications increases with age. Interpretation of calcium scores may be based either on absolute calcium scores (Table 23.11) or on scores relative to a standard male and female population group (Fig. 23.36). Despite various controversies, the following consensus statements from the 1996 report of the writing group of the American Heart Association (AHA) have been further substantiated over the years: • Negative EBCT makes presence of atherosclerotic plaque very unlikely. • Negative tests occur in a majority of patients with normal cardiac angiograms. • Negative tests are unlikely in the presence of significant luminal obstructive disease. • Negative tests (Ca score ~O) may be consistent with a low risk of cardiovascular events in the next 2-5 years. • High Ca scores may be consistent with moderate-high risk of cardiovascular events in next 2-5 years. • Positive EBCT confirms presence of a coronary atherosclerotic plaque. • The greater the amount of Ca, the greater the likelihood of occlusive CAD, but there is no one-to-one relationship and findings may not be site-specific. • Total amount of Ca correlates best with total amount of atherosclerotic plaque, but true plaque burden is underestimated.
1<
Calcium Scoring Table 23.11 1999)
Guidelines for interpretation of the Agatston calcium score in asymptomatic patients (Rumberger,
Ca score
Plaque burden
Probability CAt
o
no plaques
very low
Recommendation very low
reassure
1-10
minimal plaque
very unlikely
low
discuss guidelines for primary prevention of CAD
11-100
mild plaque
mild or minimal stenosis likely
moderate
counsel risk factor modification
101-400
moderate plaque
CAD highly likely
moderately high
institute risk factor modification exercise testing
> 400
extensive plaque
high likelihood of significant stenosis
high
aggressive risk factor modification exercise test or pharmacological stress test
CAD = coronary artery disease; CV = cardiovascular Total coronary calcium mass '== Ca score x 0.83
2000 1000
90% __ --75% 50%
100
25%
10 1+---r-~--~--~--~--+---r-35-39 70+ 40-44 50-54 60-64
Age Fig. 23.36 Normal values for coronary calciumscores in asymptomatic men (Rumberger 1996).
The following issues are controversial and remain under discussion: There is conflicting evidence as to whether extensive calcification is associated with more mature stable plaques (with less risk for acute events) or whether the risk of acute cardiovascular events is actually further increased. While some studies report that more extensive calcification is found in patients with stable angina than in those with unstable angina or acute infarction, a recent report found annual event rates (for myocardial infarction or coronary death) of 25% in patients with extremely high calcium scores (> WOO). Calcium scores > 100 have been shown to be highly predictive for separating patients who experience cardiovascular events (angina, infarction, PTCA/
surgery) from those who do not, during 2-5 year follow-up. While a calcium score of zero is highly predictive of angiographically normal or near-normal arteries, it does not rule out the presence of vulnerable plaques, especially in young highrisk patient groups. Up to 7% of patients with acute events have been found to have zero calcium scores. This is in accordance with angiographic findings in patients who by chance underwent angiography prior to a cardiovascular event; in this group over 60% of individuals had no significant (» 50%) coronary artery stenosis. •
Indications
There is little consensus about firm indications for coronary calcium scoring. General screening of asymptomatic individuals is currently not recommended. For asymptomatic subjects with risk factors (see ischemic heart disease, p. 795), however, calcium scoring may help identify individuals who benefit from aggressive risk management or further testing. In a large metaanalysis the risk ratio for coronary calcium in regard to infarction or death was found to be only slightly above conventional risk factor assessment (e.g., Framingham). At present, no hard evidence exists that coronary calcification is an independent predictor for cardiac events. Calcium scoring is not recommended for routine use, but can be justified in selected asymptomatic patients (intermediate risk, elderly) when standard risk assessment is insufficient to direct therapy plans. Various large studies (CARDIA, MESA) are under way to determine which com-
803
23
Heart
parative measures are additive to traditional risk factor models. For diagnosis of CAD in symptomatic patients, a large meta-analysis (ACC/AHA consensus) showed a pooled sensitivity of 91-92% and specificity of 49-51 % for the detection of significant stenoses, with a predictive accuracy of around 70% for EBCT when compared to angiography. Although a different cut point may alter these figures, it was concluded that calcium scoring is not superior to other well-validated tests like exercise ECG, stress echo card iography, or perfusion scintigraphy. The low specificity, in particular, may lead to unnecessary work-up and patient anxiety. In patients with atypical chest pain and a low pretest probability, calcium scoring may identify those who are highly unlikely to suffer from CAD. Calcium scoring has been successfully used to assess progression or regression of CAD. High reproducibility of results is important (see above, p. 802). Volume (CVS) and mass scores appear superior to the classical Agatston score. Coronary calcium has been shown to be reliable for distinguishing ischemic from nonischemic cardiomyopathy. Cost Effectiveness
Conducting effective cost-benefit trials is difficult. In model calculations by Rumberger et aI. (1999), EBCT was shown to be cost effective, compared to angiography, treadmill testing, stress echocardiography, and stress thallium scans, in specific subsets of the general population. A testing pathway utilizing quantification of coronary calcium by EBCT as an initial noninvasive approach minimized direct costs, and maximized cost effectiveness in ambulatory patients with low to moderate prevalence for obstructive coronary artery disease (70%). Coronary angiography as the first and only test proved most cost-effective in patients with a high prevalence (> 70%) of disease. The data, however, were deemed insufficient, as yet, by the ACC/AHA expert consensus document to warrant widespread screening efforts. More cost-benefit analyses are considered necessary.
Coronary Artery Stenosis Coronary artery stenosis is associated with coronary ischemia but there are substantial numbers of individuals in whom no significant stenoses could be demonstrated on a coronary angiogram prior to an acute cardiac event. Absence of high-grade stenosis, therefore, does not rule out a high plaque burden, particularly instable plaque prone to rupture. Correlation of high-grade stenosis and acute cardiac events is poor. High-grade stenosis, however, is frequently associated with symptoms of chronic coronary ischemia and usually requires treatment by PTCA, stent graft placement, or surgical revascularization. EBCTand multislice CTA have been found to provide a sensitivity of between 60% and over 90% for the presence of significant stenoses, depending on the size of the vessels. Stenosis detection is more accurate for proximal than for distal stenoses or for small caliber vessels. If these numbers can be improved, CTA may be used as a non-invasive tool to rule out coronary artery disease (CAD) in populations with a low to moderate prevalence, such as asymptomatic patients with high cardiovascular risk factors, unspecific exercise test findings, or patients with atypical chest pain. Coronary CTA is not yet suited to following patients with known CAD, typical angina, or obvious myocardial ischemia at exercise testing, because of a high incidence of marked calcifications that hamper CT evaluation of stenoses. CT Morphology
CT angiography is not sensitive enough to grade coronary stenoses precisely because of the small size of these vessels and considerable problems with pulsation effects. For this reason, a grading scheme has been suggested that uses five categories: Normal o 0% 1 1-49% Nonobstructive disease 11 50-74% Significant stenosis 1lI 75-99% High-grade stenosis Occlusion IV 100% Suspicious areas should be analyzed in at least two perpendicular planes (Fig. 23.37). The suspected stenosis then has to be classified and attributed to a particular segment of the coronary tree (see Fig. 23.4). In addition, the nature of the
Coronary Artery Stenosis
a
c
b
Fig. 23.37 Long, moderately stenotic segment of an intermediate, coronary artery, a third branch arising at the bifurcation of the LCA into the LAD and LCX (8X1.25/2.5), seen on CPR and transverse sections
through the stenosis (a), oblique thin-slab MIP (b) and a volume-rendered display (c). This stenosis was due to a soft plaque and lead to a vascular occlusion 3 month later with massive left ventricular infarction.
stenosis should be stated with respect to plaque composition (see below). Blooming of calcified plaques is a problem that can be reduced by using wide window settings for image analysis. Extensive calcified plaques are often present in areas of remodeling and cause no significant stenoses until the very late and severe stages of atherosclerosis. These areas may be impossible to evaluate with current CTA techniques. Regions with motion artifacts should be excluded from the evaluation (see Fig. 23.16). With retrospective gating, additional data sets can be reconstructed and may be less affected by cardiac motion (see above, p. 774).
•
Table 23.12
Plaque
Morphology
The normal arterial wall measures 0.1 mm and is not visible on CTA. An overview of the various plaque types is given in Table 23.12. Calcified or hard plaques have a density well above fibrous tissue (90 H). Calcified nodules present as small dots of calcium surrounded by dense soft tissue (Fig.23.38a, b). They are rarely associated with acute events in coronary thrombosis and are considered to represent atheromas that did not rupture but directly calcified instead. Extensive coronary artery calcifications without any non-calcified plaque are rarely the locations for significant stenosis.
Plaque morphology (Stary, 1995; Becker 2001)
Plaque
AHA
calcification
classification
noncalcified portions
shape
remodeling
'50HU
well-defined
positive
V
no may be'present
'70HU
well-defined
positive/negati
trhrombus
VI
may be present
'40HU
irregular
high-grade ste or occlusion
oFibrocalcified
VII
large calcification, across vessel
"100HU or absent
irregular
stenosis
.Calcified nodule
VII-VIII
small round nodule or calcification along vessel wall
"100HU or absent
irregular
positive
JPibrotic lesion
VIII
no
irregular
positive/negative
I&Atheroma Thin fibrous cap atheroma
IV
plaque
AHA I - III cannot be visualized by CT
100 HU
805
23 Heart
a
b
c
Fig.23.38 Calcified coronary plaques (4x1/1.S). Small round deposits are due to calcified nodules (a, b), which rarely cause stenosis (arrows). Larger, often circumferential, and longer calcifications and soft tissue
material are caused by fibrocalcified plaques (arrowhead) that are frequently associated with stenoses (b. c). Wide window settings are required for evaluation.
Fibrocalcified plaques are characterized
well below 1 mm. They are rarely detected by CTA. Atheromas contain a large lipid (cholesterol) core and have been described as displaying a CT attenuation of 50 HU or below. CT cannot differentiate a fibrous cap from a lipid core. Atheromas are wall-adherent and may be focal or present as a long region of hypoattenuating, irregular wall thickening (see Fig. 23.37). Remodeling may make the lumen look unaffected, despite considerable amounts of plaque burden, but significant stenoses due to atheromatous lesions may be present, even in patients with a coronary calcium score of O. Fibrous plaques are denser and have been associated with a CT attenuation around 90 HU. They are commonly associated with stenoses.
by longer segments of irregular calcifications associated with dense fibrous tissue (some 90 H) on coronary CTA. They are considered to represent a consequence of plaque rupture with negative remodeling, causing scar tissue and stenosis. These plaques and the associated stenoses are particularly hard to evaluate by CTA (Fig.23.38c) because of "blooming" effects that let the calcifications appear larger than they really are (caused by limitations of spatial resolution) . Soft plaques are non-calcified and may consist of preatheroma, atheroma, fibroatheroma, or fibrous plaques. Preatheromas are thin walladherent layers of soft plaque with a thickness
a
b
Fig. 23.39 Coronary thrombi are hypodense intraluminal filling defects and are always associated with highgrade stenoses or occlusion. The vessel wall is seen as a
c hyperdense rim (a). Occlusion of a segment of the LAD demonstrated on longitudinal (b) and horizontal CPR (c).
Valvular Disease Fig. 23.40 Coronary aneurysm of the LAD in a patient with EhlersDanlos syndrome. Location and calcifications are best demonstrated on VRT (b) while the thrombosed portions are best seen on CPR (a).
a
Fibroatheromas have intermediate CT attenuation, between 50 HU and 90 HU. Coronary thrombi can be suspected if an artery has a dark center and a bright rim (Fig. 23.39). They are always associated with high-grade stenoses or occlusion. The density of the thrombus may vary depending on its age and overlaps with the various types'of soft plaques. Coronary Artery Aneurysm and Dissection
Coronary artery aneurysms are associated with a rast number of systemic diseases including Ehlers-Danlos or Marfan syndrome, SLE, panar-
b
teritis nodosa, Takayasu arterities, Kawasaki disease, or generalized fibromuscular dysplasia. They may be mycotic or posttraumatic but can occur also in atherosclerosis. They represent a focal dilatation of a coronary artery and may be associated with wall calcification, thrombi, stenoses or occlusion. CTA is superior to other imaging modalities (with the exception of intravascular ultrasound) for displaying not only the lumen but also calcification and thrombi in the aneurysm (Fig. 23.40). Coronary artery dissection is a rare complication of type A aortic dissection and is due to a retrograde extension into the coronary orifices. Focal dissections may be due to cardangiographic interventions or surgery.
Valvular Diseasi
Aortic regurgitation may be caused by intrinsic
aortic valvular disease (congenital bicuspid valve, rheumatic endocarditis, bacterial endocarditis, a myxomatous valve in Marfan syndrome, breaks or leaks in a prosthetic valve), by dilatation of the aortic annulus (aortic aneurysm, dissection, syphilitic aortitis, rheumatic diseases, and Marfan syndrome), or by laceration of the aortic valve. Aortic stenosis can be caused by a congenital supravalvular, subvalvular, or valvular stenosis (most frequently the bicuspid valve), by rheumatic valvulitis (almost always combined with mitral valve disease), or senile calcific valve degeneration.
Mitral stenosis is more frequent in females
(8: 1) and is mainly due to rheumatic heart disease. Rarely, an obstructing mass may cause mitral stenosis. Atrial fibrillation is common in these patients. Thrombi forming in the left atrial appendage may cause systemic emboli. Mitral regurgitation in the adult is most frequently caused by mitral valve prolapse, rheumatic heart disease, or dysfunction or rupture of a papillary muscle. Rarely it may be caused by a left atrial myxoma, a split mitral valve leaflet in endocardial cushion defects, or a strongly calcified annulus (especially in older women).
807
23 Heart
Pulmonary stenosis is frequently associated
with various congenital heart defects but may also occur as an isolated finding. Acquired stenoses or pulmonary valve regurgitation are rare. Tricuspid insufficiency is usually a consequence of right ventricular dilatation together with increased RV pressure due to a right ventricular obstruction or severe pulmonary hypertension. Rarely may it be caused by bacterial endocarditis in drug addicts, or by rupture of a papillary muscle. Tricuspid stenosis is almost always due to rheumatic disease, and is then invariably associated with mitral valve stenosis. Rarely it may be caused by a right atrial tumor (myxoma, carcinoid tumor) or by SLE. Valvular disease is evaluated primarily by echocardiography, and by angiography for pressure measurements. MRI is becoming an increasingly important modality. Even standard CT is able to demonstrate indirect signs of valvular disease. ECG-synchronized techniques hold some promise in directly visualizing the valvular cusps and leaflets. •
CT Morphology
Indirect signs of valvular disease are summarized in Table 23.13. Calcifications of the valves have to be differentiated from calcifications of the valvular ring. Aortic regurgitation causes enlargement without thickening of the left ventricle, a large amplitude of LV motion, an increased ejection fraction, and an increased pulsation of entire aorta (best seen on MPR from non- ECG synchronized data). The aorta is normal in intrinsic
Table 23.13 I
•••
; •• ',{
Indirect signs of (compensated)
valvular disease 1\
/If, ~ J\J\J\J\J\J\
val vular disease. In patients with dilatation of the aortic annulus (e.g., Marfan syndrome), aortic regurgitation can be indirectly diagnosed if the aortic root is >5cm wide (>80% probability), and >6cm (always). The aorta in these patients is enlarged, often calcified, and elongated resulting in a tortuous descending aorta. In aortic stenosis, there is frequently a poststenotic dilatation of the ascending aorta and a thick-walled left ventricle (concentric LV hypertrophy) of either normal or enlarged size. In the compensated state there is a hyperdynamic contraction of the LV. Calcifications of the aortic valve are a good indicator of aortic valve stenosis in adults >30 years indicating a gradient above 50mm Hg (Fig. 23.41). Calcification of the aortic valve annulus is frequently present. Only on ECG-gated images may a decreased aortic valve area be visible « 0.8 ern'). Subvalvular aortic stenosis may present as a "1-2mm thin membranous stenosis (type I), a thick collar-like stenosis (type II), an irregular fibromuscular stenosis (type III) , or an excessive thickening of the upper portion of the ventricular septum causing a tunnel-like stenosis of the left ventricular outflow tract (tunnel sub aortic stenosis, type IV). For CT evaluation, ECG-triggering or gating is important. Subvalvular aortic stenosis may be due to these fixed anatomic stenoses, or due to a functional narrowing during systole in asymmetric septal hypertrophy (ASH) , idiopathic hypertrophic subaortic stenosis (IHSS) or hypertrophic obstructive cardiomyopathy (HOCM). These entities are characterized by normal to small, right and left ventricular cavities, and an asymmetrically thickened
1\
PA Aortic
"
regurgitation
LA
LV
Ao
I
II
-II I
-H
Aortic stenosis Pulmonary
regurgitation
Pulmonary
stenosis
f'['
/I
ii
H
T
ii
Mitral regurgitation Mitral stenosis Tricuspid
regurgitation
Tricuspid
stenosis
f
i
/\
/I
/\
'[
I
Ii
""""-
TT
RA= right atrium, RV" right ventricle, PA = pulmonary artery, f = dilatation, /I f = marked dilatation, H = hypertrophy
LA = left atrium,
<_"
LV = left ventricle,
Ao= aorta
Valvular Disease Fig. 23.41 Patient with dextrocardia (a) and aortic valve calcifications and stenosis (b).
b
ventricular septum that is thicker than the free left ventricular wall. ., . Supravalvular aortic stenosis may be caused by a localized hourglass narrowing just above the aortic sinuses, a fibrous membrane just above the sinus of Valsalva, or a diffuse tubular hypoplasia of the ascending aorta and the branching arteries. The aortic diameter should be below 20 mm in the stenotic segment. It may be associated with peripheral pulmonary stenoses, valvular and subvalvular aortic stenosis, and Williams-Beuren or Marfan syndrome. Since there is increased pressure in the coronary arteries, they dilate and become more tortuous with early secondary atherosclerotic changes. Valvular pulmonic stenosis is associated with a dilated main and left pulmonary artery.
a
b
Fig. 23.42 Subvalvular pulmonic stenosis with reduction of diameter ofthe pulmonary outflow tract on axial images (a) as well as a lateral MIP (b). Supravalvular pulmonic stenosis (c) of the main pulmonary arteries (ar-
Subvalvular pulmonic stenosis can be dem-
onstrated on axial sections but is better evaluated on lateral or coronal thin-slab MIP or MPR (Fig.23.42a, b). The infundibular and subinfundibular types are both associated with VSD. Supravalvular pulmonic stenosis involves any portion of the central pulmonary arteries. The main pulmonary arteries can be evaluated on axial sections but more peripheral arteries require volume-rendered or thin-slab MIP displays for optimum detection of stenoses (Fig.23.42c). The stenosis may be localized with poststenotic dilatation or present as a long tubular narrowing. Mitral valve stenosis is characterized by calcification of the mitral valve leaflets. Calcifications of the mitral valve ring are non-predictive
c rowheads) in a patient with TGA who underwent a Rastelli procedure with an aortic homograft and (now calcified) pulmonary allograft.
809
23 Heart
because they occur in old age. Direct demonstration of thickened or partially fused leaflets may be possible with ECG-gating but echocardipgraphy is the superior modality. There is massive enlargement of the left atrium (>5cm increase risk of atrial fibrillation and atrial thrombus) associated with LA wall calcifications and displacement of the esophagus toward the right. The left atrial appendage may be enlarged (unless there is retractile clot). Hypertrophy and dilatation of the right ventricle and resulting tricuspid insufficiency, and signs of pulmonary hypertension with dilatation of the central pulmonary artery segments, may be present. Associated pulmonary findings include interstitial and alveolar pulmonary edema in decompensated disease. Mitral insufficiency leads to simultaneous dilatation of the left atrium and ventricle. CT can quantify the size of both chambers, and ECG-gated techniques allow for evaluation of the left ventricular ejection fraction. CT can directly demonstrate a calcific mitral valve ring. The direct evaluation of the valve leaflets is difficult even on ECG-gated CT. Tricuspid stenosis is associated with substantial dilatation of the right atrium but a normal size RV Due to concomitant mitral valve stenosis, a dilated left atrium is frequently pres-
Fig. 23.43 Secondarytricuspid insufficiency due to dilatation of the tricuspid valve ring in a patient with chronic thromboembolic pulmonary hypertension and massive dilatation and hypertrophy of the right ventricle.
ent. Tricuspid insufficiency leads to a marked dilatation of the right atrium and right ventricle. Secondary tricuspid insufficiency can be suspected if the diameter of the annulus of the valve measures above 5.0cm. In tricuspid valve disease, the IVC and SVC are dilated and injected contrast material flows from the SVC down into the IVC and the liver veins (Fig. 23.43, compare alsoFig.23.25c) .
Cardiac Tumors
Primary cardiac tumors are rare and occur in 0.6% of patients in autopsy series. Metastases to the heart are approximately 40 times commoner than primary tumors, most of which (more than 70%) are benign. However, because benign lesions may also lead to disturbance of blood flow, emboli, or arrhythmia, these lesions may still lead to significant morbidity. Echocardiography is still the primary imaging study. Ifechocardiographic findings raise suspicion of a cardiac tumor, MRI should be performed. Most cardiac tumors found on CT scans are detected incidentally during a tumor staging examination. However, CT, especially with ECGgated multislice technique, is becoming more important as a secondary means of investigation. It allows better detection of calcification and fat.
Benign Tumors
Several types of benign tumors can arise in the heart: myxoma, fibroma, lipoma, lymphangioma, paraganglioma, and papillary fibroelastoma. Papillary tumors cannot usually be visualized with cross-sectional imaging. Cardiac Myxoma
Myxomas are the commonest cardiac tumors, accounting for 50% of cases. They manifest in adulthood and are more frequent in women. They may occur on their own (> 90%) or as part of the autosomal dominant Carney complex (myxomas, hyperpigmented skin lesions, extracardiac neoplasms) in younger patients « "10%), in which case they can be multiple. Patients can
Malignant Primary Cardiac Tumors Fig. 23.44
Left atrial myxoma.
EBeT ofa calcified lesion (a). Spiral CT of accidentally detected tumor arising from the atrial septum (b).
a
b
present with nonspecific com plaints like fatigue, weight loss, and arthralgia, but may also show signs of hemodynamic obstruction or embolization of tumor thrombi to the CNS or peripheral vasculature.
sites. Only a handful of cardiac lymphangiomas have been described and these occurred primarily in children. Patients present with palpitations or arrhythmias, and can have lymphangiomas at other sites in the body.
•
•
CT Morphology
The most frequent location (80%) of the tumor is the interatrial septum (fossa ovalis), more commonly on the left side than on the right side (Fig. 23.44). There is often a narrow base of attachment, and the tumor may prolapse through the cardiac valves. On precontrast scans, the tumors have heterogeneous low attenuation and calcifications are common. After contrast administration, intense enhancement can be seen, but more often enhancement is heterogeneous. Other Benign Tumors
All other tumors are very rare. Fibromas primarily affect children and are associated with arrhythmias. There is an increased incidence in Gorlin syndrome (nevoid basal cell carcinoma). Lipomas can occur at all ages and are frequently asymptomatic, but may give arrhythmias. The tumors arise from the endo- or epicardium and both intracavitary and pericardiallocations have been reported. Paragangliomas can be functioning or non-functioning and can occur in young adults that present with hypertension and catecholamine overproduction. They arise from the cardiac paraganglia and are most often located in the left atrium. The primary means of detection is MIBG-scintigraphy, and they can be synchronous with tumors at other extra-adrenal
CT Morphology
Adequate opacification of the cardiac chambers is required. Tumor is most clearly differentiated from normal myocardium during the parenchymal phase of enhancement (40-90 seconds after the start of the contrast injection). Fibromas are large homogeneous masses that are either well-defined or infiltrative. Most occur in the ventricles. They commonly show dystrophic calcifications. Little or no contrast enhancement is the rule. Lipomas are broadbased masses, with characteristic fat attenuation, that grow either in cardiac chambers or in the pericardial space. They do not enhance after contrast administration. Paragangliomas usually present as large, circumscribed, heterogeneous, well-enhancing masses (see Fig. 10.10). They have a broad base against the atrial wall. As in other locations, they show intense contrast enhancement. To date, CT morphology of lymphangiomas has not been described in the literature. Malignant Primary Cardiac Tumors
Only 25% of the rare primary cardiac tumors prove to be malignant. They are usually sarcomas, with angiosarcoma, undifferentiated sarcoma, rhabdomyosarcoma, leiomyosarcoma, and osteosarcoma the most frequently reported subtypes.
23 Heart
Angiosarcoma is the most frequent subtype, occurring in 35-40% of cases. Patients can be of any age, but most angiosarcomas occur in middle-aged men. Prognosis is poor as the tumor may be asymptomatic for a long time. Due to its location, the tumor may cause rightsided heart failure or tamponade. Leiomyosarcomas occur in the fourth decade and have a predilection for the left atrium resulting in cardiac output failure. This subtype can also originate in the pulmonary arteries, often in association with the pulmonary valve, or veins and involve the heart secondarily. Rhabdomyosarcomas can be either of the embryonal or adult forms. They are the commonest cardiac malignancy in children, and show variable clinical signs. Primary osteosarcomas usually arise in the left atrium and are commonly very aggressive. Liposarcoma is extremely rare. Undifferentiated sarcomas are malignant neoplasms without specific histologic features, with a variable prevalence, and most commonly result in pulmonary congestion, because of a left atrial location, with a tendency to involve the heart valves.
Leiomyosarcomas arise from the posterior wall of the left atrium and may invade the mitral valves. They are lobulated, irregular masses with low attenuation. Osteosarcomas show variable calcification, ranging from speckles to dense deposits. Osteosarcoma may be confused with left atrial myxoma, especially if no dense calcifications are seen and because of their left atrial location. Liposarcomas may arise anywhere within the heart with varying appearance of pericardial involvement. The tumor is usually large and multilobulated with little or no macroscopic fat. All sarcomas show infiltration that may involve the atrial septum, pericardium, or large pulmonary vessels. Contrast enhancement is variable and often inhomogeneous.
•
•
CT Morphology
Angiosarcomas grow large before being detected. The tumor has a predilection for the right atrium with pericardial involvement. Two distinct types are known: either a well-defined, inhomogeneous mass protruding in the right atrium, or an infiltrative mass along the pericardium. The latter may obliterate the whole pericardial space with tumor debris, which may appear at CT as pericardial effusion or thickening. Heterogeneity is often due to hemorrhage and necrosis. There may be an associated pericardial effusion. Angiosarcomas are highly vascular and enhance strongly but inhomogeneously. Undifferentiated sarcomas are large, irregular intracavitary lesions with low attenuation. Tumors are most often located in the left atrium. Tumor infiltration may resultsin irregular thickening of the myocardium. Rhabdomyosarcomas are smooth or irregular masses, and central necrosis is common. They are frequently multiple and tend to involve the cardiac valves. They also may invade the pericardium but usually in a nodular form as opposed to the sheet -like spread of an angiosarcoma.
Lymphoma
Primary cardiac lymphoma has to be differentiated from the commoner type of cardiac spread of non- Hodgkin lymphoma (mostly aggressive Bcell lymphoma) with an increased incidence for the latter in immunocompromised patients. CT Morphology
Pericardial effusion or thickening may be the only finding. Variable morphologic types are described ranging from a circumscribed polypoid mass to a more infiltrative type of lesion (Fig. 23.45). The density and contrast enhancement characteristics are also rather variable.
Fig. 23.45 ment.
Cardiac lymphoma with pericardia I involve-
Pericardia) Effusion
a
c
b
Fig. 23.46 Ventricular metastasis from large-cell bronchial carcinoma (a). Breast cancer metastasis to the atrial septum with infiltration ofthe pulmonary vein (b).
Invasion of the left superior pulmonary vein by thoracic metastases from a soft-tissue sarcoma (c).
Metastasis
ferior vena cava up to the right atrium (level 4 thrombus).
In malignant tumors of the heart, the frequency of metastatic lesions is 25-40 times more common than primary cardiac sarcoma. These tumors occur late in the metastatic process and the prognosis is invariably poor. The most frequent primaries include lung and breast cancer, melanoma and both Hodgkin's and non-Hodgkin's lymphoma. The heart usually becomes involved by retrograde lymphatic extension to the epicardium. Other pathways include hematogenous spread, and direct or transvenous extension. A cardiac location is especially frequent for metastasized melanoma. Renal cell carcinomas involve the inferior vena cava in 4-10% of cases. Tumor thrombus may extend beyond the intrahepatic course of the in-
Although CT is an excellent tool to visualize pericardial disease, echo cardiography is the diagnostic modality of choice for many standard indications. CT is important for evaluation of complex situations and tumor staging. Pericardial Effusion
Pericardial effusion is defined as more than 50 ml of pericardial fluid. The main differential diagnoses are listed in Table 23.14. Pericardial effusions may cause reduced cardiac pulsations and may eventually lead to cardiac tamponade.
•
CT Morphology
In lung and breast cancer, firm nodules may be found in the walls of the cardiac chambers but most lesions are in a pericardial location with possible intrapericardial growth (Fig. 23.46). Malignant pericardial effusions are common. In melanoma, multiple masses can be found in the myocardium. Mesothelioma has a strong tendency to involve the pericardium, much less frequently the myocardium. Pericardial effusion is not a reliable sign of pericardial involvement since effusion can be also due to radiation therapy, drug-induced pericarditis, infection, or be idiopathic.
Cardiac tamponade is caused by fluid in the pericardial sac and consequent compression of the heart chambers with insufficient diastolic filling of the ventricles. •
CT Morphology
CT is very sensitive for the detection of pericardia} fluid especially anterior to the right ventricle and posterior to the left ventricle. Little pockets of fluid (width < 5 mm) next to normal regions of thin pericardium are normal but become abnormal if the fluid encircles the heart. Measurement of the CT attenuation within this
814
23 Heart Table 23.14
Differential diagnosis of pericardial effu-
sion lymphatic fluid «10
H, even negative HU)'
Congenital Obstruction of lymphatic drainage (hilum. SVC) Tumors Cardiothoracic surgery Serous fluid (:5 18 HU)' Congestive heart failure Hypoalbuminemia Radiation therapy
Infection: viral. pyogenic. TBC Chronic> acute uremia, hemodialysis Rheumatoid arthritis, acute rheumatic fever, SlE Hemorrhagic effusion (> 30HU)'
Acute myocardial infarction Rupture of the ascending aorta (dissection. aneurysm) Rupture of the pulmonary trunk Trauma (blunt/penetrating) Coagulopathy. anticoagulants Cardiac surgery. catheterization Tumors Chemotherapy * No sharp transitions between given HUlimits.
fluid can give suggestions as to the potential etiology of an effusion (Fig. 23.47, Table 23.14). One has to be careful to avoid measurement errors due to pulsation-induced inhomogeneities. Indirect signs of cardiac tamponade include decreased pulsation artifacts in the ascending aorta (indicating low cardiac output) and a widened superior vena cava (indicating decreased right heart filling). Heart size may enlarge rapidly. On ECG-gated acquisitions there is evidence of a diastolic collapse of the RV and cyclic collapse of either atrium.
Pericarditis
Acute pericarditis
can be caused by tuberculosis or viral (Coxsackie B) infections. by chronic renal failure. rheumatoid arthritis. and radiotherapy to the mediastinum. Constrictive pericarditis may develop as the consequence of acute pericarditis and is characterized by fibrous thickening of the pericardium. which leads to restriction of heart motion and decreased filling of the ventricular chambers. It affects males more often than females (3: 1). Concentric neoplastic involvement of the pericardium leads to similar symptoms.
beware of pulsation-induced measurement errors
b
Fig.23.47 Uremic pericardial effusion (a). Hemorrhagic pericardial effusion with impending pericardial tamponade (b). Iatrogenic esophago-pericardial fistula
after esophagoscopy (c). Note the contrast material and air in the pericardial sac after oral administration of iodinated contrast material.
Pericardia! Tumors
CT is excellently suited to visualize pericardial morphology in constrictive pericarditis and can aid in pre-surgical evaluation prior to pericardial fenestration or pericardectomy. •
CT Morphology
A thickened or enhancing pericardium together with a pericardial effusion suggests active pericarditis (Fig. 23.48). Adhesions between parietal and visceral pericardium cannot be directly demonstrated by CT but can be suspected if there is an atypical distribution of pericardia I fluid. In constrictive pericarditis the pericardium is thickened to 2 mm or more, often containing extensive linear or plaque-like calcifications. The calcifications are most commonly seen over the right ventricle, the posterior surface of the left ventricle, and in the atrioventricular groove. 3D rendering visualizes the site of the most extensive calcifications prior to surgery (Fig. 23.49a, b). Secondary signs involve dilation of the SVC and IVC, and reflux of contrast into the liver veins and coronary sinus. There is flattening of the right ventricle and bulging of the intraventricular septum towards the left. Pleural effusion and ascites are signs of decompensation. Pericardial Tumors
Benign pericardial neoplasms include teratoma, leiomyoma, hemangioma, and lipoma. Malignant neoplasms include mesothelioma, sarcoma, metastases from lung and breast cancers, lymphoma, leukemia, and melanoma. Metastases or continuous tumor extension into
Fig. 23.48 Infectious pericarditis with pericardial empyema and enhancement of the pericardium.
the pericardium are much commoner than primary tumors. CT is excellent for defining the extent of benign tumors and for demonstrating mediastinal involvement in malignant disease, but echocardiography and MRI often are better suited for demonstration of myocardial infiltration. •
CT Morphology
CT demonstrates a mass adjacent to or within the pericardial sac (Fig. 23.50). A pericardial effusion is more often seen in malignant disease, in which case it is often hemorrhagic (>30HU). Thin-section imaging and ECG-triggering improve delineation of the mass relative to the heart chambers. An intact epicardiac fat line makes myocardial involvement highly unlikely. Malignant pericardial tumors may lead to encasement and stenosis of coronary arteries (see Fig. 10.10).
Fig. 23.49 Constrictive pericarditis after surgical fenestration of the pericardium anterior to the right ventricle. Note the residual calcifications of the pericardium posterior to the left ventricle (a) and at the base of the heart (b). There is substantial dilatation of the atria and increased pulsation of the pulmonary outflow tract (arrows).
a
b
815
23
Heart
Fig. 23.50 Pericardial sarcoma infiltrating the region ofthe left atrial appendage.
Pericardial Cyst
lies (VSD, PDA, mitral stenosis). The smaller foraminal defects are distinguished from complete absence (bilateral, left -sided, or diaphragmatic pericardial aplasia) oflarge portions of the pericardium. Acquired defects are usually caused by surgical procedures to the heart or pericardium (pericardectomy). Pericardial defects are commonly asymptomatic but may be associated with nonspecific intermittent chest pain and cardiac symptoms, as well as positional discomfort. A large foraminal defect requires surgery because herniation of the left atrial appendage, left atrium, or left ventricle may occur. CT is excellently suited to directly demonstrate the pericardium as well as secondary effects due to herniation of cardiac structures. •
CT Morphology
Small foraminal defects will show no abnormalPericardial cysts present as a paracardiac mass, ity other than a short interruption of the periusually in the right or left costophrenic angle. cardial line on axial sections. With moderate Pericardial cysts in the mediastinum are rare. foraminal defects there may be a focal bulge in -the area of the right ventricular outflow tract, See also Chapter 10, Mediastinum (p. 373). main pulmonary artery, or the left atrial ap• GT Morphology pendage. Large defects may lead to herniation of CT demonstrates a smooth, round or ovoid mass cardiac structures or lung, e.g., insinuation of in typical location with a diameter of up to 8 cm lung between the aortic arch and the pulmonary that has attenuation values of 20-40 HU, and artery. A diaphragmatic pericardial aplasia may occasionally higher. lead to interposition of lung between the heart and the left hemidiaphragm. Absence of the anterior portions of the pericardium will cause an increased distance between sternum and the Pericardial Defect heart in the supine patient (lack of the sterCongenital pericardial defects are caused by a nopericardic ligament). Complete absence may failure of pericardial development that may be lead to levoposition of the heart. Patients with associated with bronchogenic cysts, diaphragpneumothorax and pericardial defects may matic hernia, sequestration or cardiac anoma- develop pneumopericardium.
Cardiac
Trauma
Penetrating Trauma
Penetrating injuries to the heart have a high mortality and less than 20% reach the hospital alive. The clinical presentation can be quite deceptive and can range from hemodynamic stability to profound shock with cardiac arrest. The presence of pericardial effusion should raise suspicion of penetration but the classic signs of pericardiac tamponade like Beck's triad (muffled heart tones,
hypotension, and distended neck veins) or Kussmaul's sign (jugular venous distension on inspiration) are usually absent. The commonest causes are stab or gunshot wounds. The anterior located ventricles are at greatest risk, whereas the left atrium is only rarely involved. Stab injuries result in perforation of the pericardium or myocardium that may quickly seal but later the effects of pericardial tamponade seriously compromise car-
Postoperative and Postinterventional Changes diac function. Survival rates may be as high as 60-65% with appropriate therapy. Gunshot injuries usually result in large myocardial and pericardial injury. more often with hemorrhage in the left hemithorax. Survival rates of only 10-15% are reported. Diagnosis is usually made with echocardiography as part of a focused trauma ultrasound protocol. CT is agood alternative in stable patients and has the advantage that it can also evaluate the other vulnerable structures of the thorax and abdomen and that it can show the knife or bullet tract. Unstable patients should proceed directly to emergency thoracotomy with intensive fluid resuscitation and cardiorrhapy. •
CT Morphology
CT can show as little as 25-50 ml pericardial fluid. usually located anteriorly to the right ventricle. Signs of pericardial tamponade (see above. p. 814) have to be excluded. There may be hemothorax. usually on the left side. An abnormal contrast leakage into the pericardium is a direct sign of myocardial injury.
phragmatic pericardium. Pericardial ruptures can be isolated or be associated with cardiac herniation. Dislocation of the heart harbors the risk of incarceration of myocardium or strangulation of a coronary artery. CT performed for blunt chest trauma may demonstrate the lesion. More than 80% of patients with myocardial lacerations after blunt chest trauma die acutely. Myocardial lacerations are present in 0.2-2.0% of patients that reach the hospital. Early diagnosis and emergency surgery can result in survival rates of 50-80%. Although the rupture may involve any part of the myocardium including the atrial appendages. the most frequent location is the right atrium. Patients usually present with a widened mediastinum or with left hemothorax on chest x-ray. Diagnosis is usually made by transesophageal echo cardiography. CT of the thorax in such patients is commonly performed to rule out vascular mediastinal injury and may be life-saving if myocardial rupture is detected. However. the findings are often subtle and easily missed if the radiologist is not actively searching for such a lesion. CT Morphology
Blunt Trauma
Cardiac contusions or infarcts rarely show abnormalities on CT. In pericardial ruptures, the thin
Blunt cardiac injury is usually caused by a direct blow or a deceleration injury in a motor vehicle accident. Relatively more cases are seen following sternal injuries. Cardiac contusions are frequent but rarely manifest clinically. Infarction of myocardial tissue is rare. It is commonly caused by injury to the coronary artery (dissection. thrombosis. or embolism of plaque material). Presently. there is no indication for CT for suspected cardiac contusion. Pericardial ruptures are rare. They most frequently involve the pleuropericardial or dia-
line of the pericardium may be interrupted, and pneumopericardium, cardiac dislocation, or an empty pericardial sac may be seen. In myocardial lacerations, CT will show hemopericardium with or without associated cardiac tamponade. There is mediastinal hemorrhage. Focal areas of abnormal enhancement of the myocardium or (subtle) abnormal contrast leakage into the pericardium are direct signs. A (usually left-sided) hemothorax indicates concomitant injury to pericardium and pleura.
Postoperative
and Postinterventional
CT is the imaging modality of choice for the evaluation of local complications in the mediastinum and after sternotomy. For the heart, the preferred modalities for evaluating postoperative and postinterventional complications are echocardiography, cardiac angiography, and MRl, depending on the clinical question.
Changes CT can provide a minimally invasive procedure for assessing the patency of an aortocoronary venous bypass but still has trouble evaluating bypass stenosis. CT is not used to evaluate cardiac valves or septal defects, but it can be used in the follow-up of complex surgically corrected heart defects in patients with cardiac pacemakers and in cases difficult to evaluate with ultrasound.
817
23 Heart
Normal Postoperative Findings
Retrosternal fluid collections, small hematomas, and air inclusions are considered normal findings during the first three weeks after thoracic operations on the heart and great vessels, and these changes regress with passage of time. Often it is difficult to distinguish normal from pathologic findings in a single examination. Pleural effusions, deficient aeration of the posterobasal lung, and compression atelectasis are consistently observed. Pericardial effusions are not a normal finding. In cases where a median sternotomy was performed, the osteotomy line should not be visible, although it is normal to find areas ofimpaction or slight displacement.
The role of CT is to demonstrate graft patency. This task does not require ECG synchronization. Evaluation of graft stenoses may be possible for ACVB but is very difficult to achieve even with EBCT and gated multislice CT techniques. •
CT Morphology
Aortocoronary venous bypass grafts (ACVB) are usually performed in the middle third of the ascending aorta, although higher or lower anastomoses (e.g., the brachiocephalic trunk) are feasible. Bypasses to the left coronary artery are routed over the pulmonary trunk. Internal mammary arterial bypass grafts using the left (LIMA) or right internal mammary artery (RIMA) are more difficult to evaluate due to the multiple clips that are used to occlude the small side branches.
Since the diameter of a venous bypass is 3-7 mm, ACVBs can generally be visualized with CT (Fig. 23.51). A bypass that does not opacity after intravenous contrast administration is occluded, whereas a stenosed bypass will still show complete opacification. Stenoses most frequently occur at the site of anastomosis. Stenoses at the aortic anastomosis often can be evaluated even without ECG synchronization. Stenoses of the distal anastomosis, however, require triggering or gating techniques. Because of the long scan range of up to 20 em, evaluation with EBCT is difficult, and multislice CT with 4-slice scanners either requires fast gating techniques (P * = 3) or thicker sections (4 x 2.5-3 mrn), which limit spatial resolution. Mural calcifications may develop in the bypass over time without causing occlusion or stenosis. A narrow, nonenhancing band of softtissue density 1-5 mm wide along the course of an expected ACVB signifies a chronic occlusion (Fig.23.51). With LIMA and RIMA bypass grafts, patency is usually detectable, but the lumen is often very small (1-4 mrn). Detection of stenoses is not yet
a
b
Aortocoronary Venous Bypass, Internal Mammary Arterial Grafts
Fig. 23.51 Aortocoronary venous (ACVB). Volume rendered display of grafts from the ascending aorta to the segment and the distal LCX (a). Chronic
bypass grafts normal venous distal LAD, D1 occlusion of an
c
ACVB to the LCX (arrow), and patent LIMA bypass to the first diagonal branch (arrow head) and ACVB to the LAD (wide arrow) (b. c).
Catheter Ablation
ofArrhythmogenic
Foci
a
b
Fig. 23.52 Volume-rendered display of the interior of the chest wall with RIMA and LIMA (a). LIMA and RIMA
bypass and radial artery transplant in a patient with multiple coronary stenoses (16 x 0.75/3.6) (b).
reliably possible. Preoperative imaging can demonstrate the anatomy of the internal mammary arteries (Fig. 23.52).
ing is mandatory for morphological evaluation of stent patency but even then, blooming effects will make evaluation difficult. Wide window settings are helpful in such situations. Morphological analysis relies on longitudinal and transverse sections using MPR and CPR capabilities (Fig.23.53). Multisection flow studies with EBCT have been shown to indirectly detect stent obstruction by demonstrating malperfusion of the dependent regions.
Coronary Stents
Coronary stents are frequently used to treat coronary stenosis primarily or to salvage acute complications of PTCA. Nevertheless, the risk of restenosis or occlusion is still high with most stents. CTA has a limited spatial resolution for the evaluation of stents in smaller more distal vessels but may be sufficient for the evaluation of proximal stent grafts in larger vessels. Perfusion studies with EBCT have shown promising results. At present, however, there is no hard indication for CT. •
CT Morphology
Any type of ECG-synchronized CT study is able to precisely localize a stent. Thin-section imag-
Catheter Ablation ofArrhythmogenic
Transcatheter ablation of arrhythmogenic foci in the left atrium may be performed in patients with severe atrial fibrillation. For this purpose a loop-shaped electrocoagulation wire is placed in the orifices of the pulmonary veins, thus interrupting the electrophysiological connection between the pulmonary veins and the atria. The anatomy of the pulmonary venous orifices is
Fig. 23.53 Coronary artery stent near the ostium of the RCA in the region of a stenosis due to a calcified plaque. Axial (a), longitudinal (b) and perpendicular MPR (c). Only larger stents in the more proximal portions of the coronary arteries can be evaluated (16XO. 75/ 3.4).
a
Foci
b
c
820
23
a
Heart
c
b
Fig. 23.54 Ablation of arrythmogenic tissue at the pulmonary venous ostia. (a) Display of pulmonary venous anatomy is possible from low-dose spiral scans (4 x 1 /6,
2.2 mGy). (b) Follow-up with occlusion ofthe left upper lobe vein. (c) Note the pulmo-pleural venous anastomosis in this patient (arrow).
highly variable, and cardangiography only provides projection images that provide little information about the sagittal diameter of the veins, thus making the procedure time-consuming and risky because overcoagulation may lead to stenosis or even occlusion of a pulmonary vein. Low-dose multislice CT is sufficient to provide the necessary anatomical information about the left atrium and the pulmonary veins. Gating should not be performed because it requires an unnecessarily high dose for this purpose. Even triggering is not necessary because of only minor movement artifacts in the vast ma-
jority of cases. Contrast material injection is helpful but not mandatory, and can be combined with a Iow-kvp scanning technique.
a
b
Fig. 23.55 Perforation of a right ventricular pacemaker into the epicardial fat (a.) Malposition of a pacemakerthatwas no noted onthe PAradiograph. The
•
CT Morphology
Volume rendering after removal of the aorta is best suited to provide an overview of the venous anatomy (Fig.23.54a), and multiplanar crosssections though the orifices of the pulmonary veins can be used to provide the measurements necessary for the choice of loop size. Follow-up may demonstrate stenoses or even occlusion of one or more pulmonary veins.
c pacemaker entered the coronary sinus (b) and ended in the posterior vein of the left ventricle (c).
Heart Transplantation 821 Due to segmental venous congestion, an increased lung density and various grades of interstitial edema may be seen. In chronic occlusion, calcifications may occur (Fig.23.54b), and collateral vessels to other segments or even the chest wall may form (Fig. 23.54 c). Pacemakers
CT can be used to define venous cardiac anatomy prior to complex pacemaker placement. It can prove penetration of a pacemaker wire through the myocardium, even in difficult cases that cannot be resolved by fluoroscopy (Fig.23.55a). CT can also depict malposition of pacemaker wires (Fig.23.55b), but this is usually an incidental or concomitant finding. Dislodged catheter or wire fragments in the pulmonary circulation can be precisely located.
Congenital Heart Disease
Patients with surgically repaired heart disease require long-term follow-up, which generally is done with echocardiography. Many cases additionally require MRI for the follow-up of complex defects (transposition of the great vessels, atretic valves, single ventricle, tetralogy of Fallot, etc.) and aortic anomalies (coarctation, duplicated arch, etc.). In most cases CT can also resolve common postoperative questions on the anatomy and morphology of the vessels about the heart (Table 23.15, Figs. 23.57 and 23.58), but image quality may be degraded by pulsation artifacts and there may be radiation concerns if follow-ups are maintained over a period of years. As a result, CT is reserved mainly for use in patients with cardiac pacemakers and in emergency situations. Heart Transplantation
Valve Replacement
Cardiac valves that contain metal are a source of CT artifacts. It is rarely possible to detect perivalvular leaks (Fig.23.56a) or endocarditic valve lesions. Usually these are incidental or concomitant findings on CT examinations performed for other indications. The modalities of choice are echocardiography and cardiac angiography. CTA is well suited, however for detection and mapping of anastomotic leaks after valve replacement (Fig.23.56b).
Cardiac transplantation is an accepted treatment for end-stage cardiomyopathy or coronary artery disease. Improvement in immunosuppression and early diagnosis of rejection has resulted in improved survival rates in the order of 60-80%. In orthotopic transplantation the native heart is removed and the donor heart is joined to the recipient atria, aorta, and pulmonary artery. Heterotopic transplantation is much less often performed and in this procedure the donor heart is placed in the right hemithorax and connected to the recipient in such a way that the donor heart provides the bulk of left-
Fig.23.56 Complications after aortic valve replacement. Paravalvular leak (a). Anastomotic leak with pseudoaneurysm (b). a
b
23 Heart Table 23.15 Complex heart defects: surgical procedures and diagnostic issues
I Surgical
technique
Goal
Procedure
Diagnostic issues
Mustard and Senning procedures for atrial inversion
Reverse blood flow at the atrial level to palliate complete transposition of the great arteries (TGA)
Intra-atrial baffle is inserted to connect the SVC and IVC to the subpulmonary ventricle (mLV) and to channel LV blood to the systemic ventricle (mRV)
Stenosis at thejuncti the SVC or IVCwith systemic venous atrium or at thejunction ofthe pulmonary veins with the pulmonary venous atrium
Fontan procedure
Segregate the pulmonary and systemic circulations in patients with anatomic or functional univentricular heart (e.g., single ventricle, tricuspid atresia, etc.)
An anastomosis is created between the SVC, IVC or right atrium and the pulmonary arteries
Anastomotic stenosis, intraatrial thrombi
Blalock-Taussig shunt
Increase pulmonary blood flow (e.g., in tetralogy of Fallot or pulmonary atresia)
An anastomosis is established between the subclavian artery and the ipsilateral PA ora plastic shunt is interposed between the subclavian artery and the PA
Shunt patency, peripheral pulmonary arterial stenoses (area of anastomosis)
Coarctation repair
Bypass or remove the aortic constriction
Multiple options include patch grafting, resection with end-to-end anastomosis, and extracardiac bypass
Exclude aneurysms in the ascending aorta or at the surgical site, exclude recurrentstenosis, confirm by-", pass patency
Rastelli procedure
Bypass or remove pulmonarystenosis, e.g., in TGA with VSDand (sub)pulmonic stenosis
Includes insertion of an extracardiac conduit (e.g., a homograft) between the subpulmonary ventricle and PA
Conduit stenosis
IVC = inferior vena cava, SVC =superiorvena cava, mRV = morphologic right ventricle, mLV = morphologic left ventricle, PA = pulmonary artery, TGA = transposition of great arteries
a
b
Fig. 23.57 Correction of D-TGA by a Mustard/Senning procedure. A baffle in the atria (arrowheads, c) directs the blood from the systemic circulation to the morphologic left (subpulmonic) ventricle, and the blood from
c the pulmonary veins to the morphologic right (subaortic) ventricle. A stenosis of the baffle at the region indicated by the arrows (a, b) is a typical complication.
Sternal Debiscence
a
Fig. 23.58 An arterial switch procedure for correction of transposition of the great vessels results in an abnormal anterior postion of the pulmonary artery but successful separation of the right and left circulation. Low-dose scan (80kVp, CTDlvoi= 0.6 mGy) in a 1-year-
side cardiac output, while the native heart provides the right -side output. Complications are common and often related to the surgical procedure. Non-surgical complications include infection with or without empyema and mediastinits, acute rejection, accelerated graft atherosclerosis, Ieukencephalopathy, and £ost-transplant lymphoproliferative disorder (PTLD). CT is an excellent technique for detection of these complications (see also Chapter 10, Mediastinum). •
c
b
CT Morphology
The normal postoperative morphology in orthotopic transplantation may show a high main pulmonary artery and a large space between the recipient SVC and the donor ascending aorta. There may be a change in caliber between donor and recipient aorta and a waist in the atria at the corresponding anastomosis (Fig. 23.59). The remnant of the donor SVC is medial to the recipient SVC and the donor aorta. In heterotopic transplantation there is obvious right -sided cardiomegaly with anastomoses between both atria, aortae and pulmonary arteries.
old infant (a). In a Fontan operation, the superior vena cava is anastomosed to the right pulmonary artery, bypassing the right ventricle (b. c). Note the lack of a SVC in (b) and the highly concentrated contrast material entering the right pulmonary circulation in (c).
CT is the imaging technique of choice for
this'entity. •
CT Morphology
Sternal dehiscence is seen most clearly on curved coronal reformations. CT demonstrates intrasternal gas bubbles or liquid pre- and retrosternal masses that show peripheral enhancement (abscesses). If a sternal fistula is suspected, the fistulous tract is injected with dilute (" 1:10) contrast material to detect osseous involvement and contrast extravasation into the retrosternal space. These findings are best appreciated on sagittal reformations.
Sternal Dehiscence
Sternal dehiscence following a median sternotomy is defined as a gap of at least 2 mm persisting at the osteotomy site. Sternal dehiscence leads to sternal instability and predisposes to sternal infection, which is diagnosed clinically.
Fig. 23.59 Typical indentation of the left and right atrial anastomosis after heart transplantation. Note the dilatation of the right atrium with secondary tricuspid insufficiency.
824
23
Heart
Mechanical Support of Circulation. Artificial Heart
Various technique have been recently developed for internal or external mechanical support of end-stage heart failure. They include intraaortic ballon pumps, mechanical tube systems that
a
b
Fig. 23.60 Artificial heart implanted in the same patient as in Fig.23.29c. The connections to the atria and ventricle are best demonstrated on MPR (a) while VRT provides a good overview ofthe position ofthe Tef-
connect the heart to an external pump or completely implantable systems (artificial hearts) . CT is a valuable tool to detect complications such as pseudoaneurysms, perforation, hematoma or infection (Fig. 23.60). ECG gating is rarely required.
c Ion sheets placed around the cardiac tubes (b). Note the mediastinal hematoma demonstrated by the axial sections (c).
24 Vascular System M. Prokop, C. Engeike
Anatomy
.
Pulmonary Vessels (p. 827) Aorta (p. 828) Aortic Side Branches (p. 829) Principal Systemic Veins (p. 830) Portal Venous System (p. 831) Examination
Technique
Aqusition Technique (p. 834) Contrast Injection (p. 838) Evaluation and Documentation
_
832
(p. 844)
848
Artifacts Pulmonary Vessels
_
851
Developmental Arterial Abnormalities (p. 851) Developmental Venous Abnormalities (p. 853) Pulmonary Thromboembolism (p. 854) Pulmonary Emboli of Other Etiology (p. 862) Pulmonary Hypertension (p. 864) Arteriovenous Shunting (p. 866) Pulmonary Aneurysms (p. 867) Vasculitis (p. 868) Neoplasia (p. 869)
870
Aorta Anatomic Variants (p. 870) Coarctation of the Aorta (p. 872) Aortic Aneurysms (p. 873) Aortic Rupture (p. 878) Traumatic Aortic Injury (p. 880) Dissection of the Aorta (p. 880) Aortic Stenosis and Aortic Occlusion Aortitis/Arteritis (p. 890) Neoplasia (p. 891)
(p. 888)
Continue
/\
826
24 Vascular System
Abdominal Anatomic Living
and Peripheral
Variants (p.
Renal
Donors
Arteries
BK
,¥¥KN
898)
(p. 899)
Artery Stenosis (p. 900) Celiac and Mesenteric Stenosis (p. 904) Acute Mesenteric Ischemia (p. 904) Peripheral Arterial Occlusive Disease (p. 906) Renal
Aneurysms
(p.906)
Venous System Variants (p. 908) Thrombosis of Major Veins (p. Portal Hypertension (p. 911) Neoplasia (p. 913)
908
Anatomic
910)
Trauma
915
Pulmonary Vessels (p. 915) Aorta (p. 915) Neck Vessels (p. 916) Arterial Bleeding (p. 917) Venous Trauma (p. 917) Postoperative Changes
.
918
and Follow-up
921
Postoperative Findings (p. 918) Malpositioned Catheter Systems (p. 918) Anastomotic Complications (p. 919) Arterial Dissection (p. 919) Stenoses and Occlusions (p. 920) Infected Prosthesis (p. 920) Normal
Angiographic
Interventions:
Patient Selection
Planning of Angiographic Interventional Procedures (p. Follow-up after Angiographic Procedures (p. 921) Planning of Endovascular Aortic Repair (p. 922) Follow-up after Endovascular Aortic Repair (p. 926)
cr angiography (CTA) has revolutionized vascular imaging. Vessels smaller than I mm in diameter can be visualized with single slice CTA, and vessels 2 mm or more in diameter can be diagnostically evaluated. The acquisition of 3D data volumes makes it possible to generate "angiographic" views that facilitate anatomic orientation, With the advent of multislice CT A, near isotropic resolution becomes available for most applications and allows for evaluation of even smaller vessels. Multislice scanning makes CTA less technically challenging and more robust. It reduces scan time and the vulnerability to mo-
921)
tion artifacts, and may be used to substantially decrease the required contrast material volume to as little as 50 ml in selected cases. The advantages ofCTA over arterial angiography (DSA) include substantially lower invasiveness, less cost, less radiation exposure, and better patient tolerance. Diagnostic advantages include the simultaneous visualization of vessel wall and lumen, and the ability to studyvascular anatomy from arbitrary viewing angles using only one data acquisition. CTA even allows for projections (such as caudocranial) that cannot be obtalned with conventional angiographic techniques. Although CTA can replace diagnostic
Anatomy intra-arterial DSA for many indications, the poorer spatial resolution in comparison with intra-arterial DSA is a limiting factor in patients in whom diagnostic evaluation: of small vessels is required (Table 24.1). Ultrasound with its various Doppler techniques may be preferred over CTA in regions that are easily accessible by sonography or when flow information is required. It also holds advantages whenever bedside imaging is to be preferred. Contrast-enhanced magnetic resonance angiography (MRA) offers competitive results to CTA, although spatial resolution is often lower than with multislice CT scanning. The lack of ionizing radiation of MRA is important in young patients, for vascular screening, and for patients requiring repeated follow-up examinations. In patients with impaired renal function, contrast-enhanced MRA requires less (and therefore less nephrotoxic) contrast material. Dynamic MRA examinations offer therapeutically relevant additional information in only a relatively small number of cases. Flow quantification with MR may gain increasing importance for therapeutic decision-making. In acute lifethreatening diseases such as aortic aneurysm, aortic rupture, or pulmonary embolism, CTA is the procedure of choice owing to the short examination time, easier patient monitoring, and less reliance on patient compliance.
The anatomy of the Vascular system is well known from angiographic studies. Crosssectional anatomy, however, provides the opportunity to trace blood vessels to and from their target organs and appreciate their relationship to surrounding structures. The normal ranges of vascular calibers of the major vessels are reviewed in Table 24.2. Pulmonary Vessels
The pulmonary trunk divides at a Y-shaped bifurcation into the left and right pulmonary arteries. The diameter of the pulmonary trunk should not exceed that of the ascending aorta, and the diameters of the pulmonary arteries should be less than the diameter of the de-
Table
24.1
Indications
CT angiography Aorta
for vascular
Hepatic
with CT
All lesions affecting the aorta or wall: arch anomaly, coarctation, aneurysm, occlusion,
Renal arteries
imaging
(eTA) its /\
dissection, stenosis, arteritis, trauma
Suspected stenosis, evaluation living renal donor work-up
for
PTA,
arteries
Preoperative vascular anatomy; detection of stenosis or occlusion after
Mesenteric
Chronic
liver transplantation
I
ischemia,
eTA),
acute
ischemia
arteries
(multislice
Carotid arteries
Suspected dissection
Runoff vessels
Peripheral arterial in non-dabetics
Pulmonary artery
Pulmonary embolism, A V mal formations, arteritis, chronic thromboembolic pulmonary hypertension,
Vena cava, portal vein
Suspicion of thrombosis, tumor invasian (equivocal ultrasound findings)
Postoperative
Complications: hemorrhage, infection, bypass thrombosis, anastomoticj
congenital
aneurysms
stenosis,
aneurysms,
occlusive
disease
anomalies
aneurysm
scending aorta. The pulmonary arteries taper uniformly towards the periphery. All artery segments branch dichotomously into two subsegmental arteries. The only exception is segment 6 which gives rise to three subsegmental arteries. There are, however, numerous variants of this branching pattern with up to four main branches and occasional smaller branches coming from a segmental artery. Two pulmonary veins pass from each lung to the left atrium of the heart. In their central portions the upper lobe veins run directly anterior to the pulmonary arteries, while the lower lobe veins course medial to the arteries (Fig. 24.1). Variants of the pulmonary veins are frequent and become important for planning
827
24 Vascular System Table 24.2 vessels
electroablation in atrial fibrillation because the size of the pulmonary venous orifice may vary substantially between patients, and aberrant vessels may be hard to detect on retrograde angiography. If the vein draining segment 6 (posterioapical lower lobe) runs posterior to the upper lobe bronchus and intermediate bronchus it may simulate a pathologic bronchopulmonary lymph node on non-enhanced scans. This vein enters the left atrium directly from above. Within the lung parenchyma, the pulmonary veins and arteries can be distinguished by the fact that the arteries run parallel to the bronchi.
Normal average diameters of major blood
Pulmonary trunk?
22-35 mm"
Right pulmonary artery
20-23 mm
Left pulmonary artery
18-21 mm
Pulmonary veins
14-16mm
Aortic root
30-38 mm
Distal ascending aorta
28-36 mm
Proximal descending aorta
24-28 mm
Distal descending aorta
22-26 mm
Suprarenal abdominal aorta
20-25 mm
Infrarenal abdominal aorta
18-23 mm
Celiac artery
5-10 mm
Superior mesenteric artery
5-10mm
Renal arteries
5- 9mm
Inferior vena cavl
•
Aorta
The aortic root appears as an elliptical expansion (bulb of the aorta) that contains three smaller dilatations called the aortic sinuses. Pulsations
22-28 mm
_~R.BCV ~_~BCA T
Es
VCS •
L. BCV
L. SeA
AZV
Pericardium
ILAD
Fig. 24.1 Cross-sectional anatomy of the mediastinal vessels. BCAlBCV= brachiocephalic artery and vein SCA=subclavian artery Ao = aorta RA/LA= right/left atrium RVOT = right ventricular outflow tract RV/LV = right/left ventricle RCA/LCA= right/left coronary artery LCX= leftcircumflex branch LAD = Left anterior descending artery Es= esophagus T = trachea AZV = azygos vein PV= pulmonaryvein
Anatomy
in this area produce characteristic crescentshaped double contours or density variations in the ascending aorta that should not be mistaken for an aortic dissection (see Fig. 24.5). The aortic arch is most commonly defined as the portion of the aorta extending from the origin of the brachiocephalic trunk to just behind the origin of the left subclavian artery. An elongation in elderly individuals or in patients with aneurysms of the descending aorta may push the aortic arch to an anterior position. As a result, the proximal portion of the descending aorta may form the most apical point of the aortic convexity. This so-called "aortic hump" (see Fig. 24.46) should not be confused with the aortic arch. The descending thoracic aorta has a markedly smaller diameter than the ascending aorta, and its diameter should decrease very little as the vessel descends to the diaphragm. The abdominal aorta begins distal to the aortic hiatus and tapers slightly at the origin of the infradiaphragmatic side branches to the iliac bifurcation. •
Aortic Side Branches
The brachiocephalic trunk is the first supra-aortic branch. It is followed by the left common carotid and the left subclavian arteries (Figs. 24.1, 24.2a). There are numerous variants, which are described later on p. 870 and p. 892. The left subclavian artery commonly arises from the apex of the aortic arch. The subclavian arteries ascend through the thoracic inlet and then turn downward toward the axilla. As axillary arteries they descend posterior to the accompanying veins, their positions varying somewhat with the position of the arms. The vertebral arteries originate from the apex of the subclavian arteries. A left subclavian artery that arises some distance posterior to the aortic apex may indicate a coarctation or other aortic arch anomaly. The subclavian arteries vary in their origin; the most important variation is an aberrant right subclavian artery that arises from the back of the aortic arch below its apex and passes to the right side behind the trachea and the esophagus causing dysphagia lusoria (see Fig. 24.41). The carotid arteries bifurcate at the level of C2-C4. Tortuous courses of the supra-aortic arteries are quite common in elderly individuals.
Fig. 24.2 Schematic diagrams of the aorta (a) and the venous system of the chest and abdomen (b).
The proximal portions of the bronchial arteries' intercostal arteries, lumbar arteries, and other small aortic branches can be demonstrated even with single-slice CTA technique, which may help in planning interventional procedures. With multislice CT scanning, such arteries can' be followed further into the periphery. Bronchial arteries are rarely identified past the region of the main bronchi. Dilated bronchial arteries (> 2 mm) are common in pulmonary hypertension or other causes of abnormal collateral circulation of the lung. Only with excellent vascular opacification can the anterior spinal artery be identified in some patients when thin-section scanning is employed, but DSA remains the standard for the evaluation of the spinal cord arteries. Identification of the phrenic or suprarenal arteries is usually possible with multislice scanning. The testicular and ovarian arteries can be identified, given sufficient contrast enhancement and thin-section scanning, as
829
24
Vascu far System
structures that run parallel to the ureter. The gonadal arteries commonly arise from the aorta but gonadal arteries that originate from the left, and rarely from the right, renal artery may assume importance for living renal donors. On single- or multislice CTA the celiac and superior mesenteric arteries can easily be identified as separate structures. The division of the celiac trunk into the splenic and hepatic arteries displays a typical swallowtail configuration. The left gastric artery appears as a smaller vessel that commonly runs upward from this division. In cases where the celiac trunk takes a descending preaortic course, portions of the diaphragm may abut the trunk anteriorly, causing a stenosis with a transverse oval shape on axial views (see Fig. 24.80). Anatomic variants of the upper abdominal arteries are found in approximately 50% of patients (see below, p. 898). With thin-section multislice CT, the various branches of the celiac and superior mesenteries artery can be analyzed in more detail (Fig. 24.3). The renal arteries generally arise just caudal to the origin of the superior mesenteric artery at the level between LI and L2. Multiple renal arteries are present in approximately 20% of cases. The inferior mesenteric artery arises from the frontal or left anterior aortic wall between the third and
Celiac artery Common Proper hepatic hepatic artery artery
Left gastric artery
Short gastric arteries
fourth lumbar vertebral bodies. The aortic bifurcation lies approximately at the level ofthe L4-L5 intervertebral space. Tortuous iliac arteries may appear as multiple segments on axial CT sections. The various internal iliac side branches can best be demonstrated on 3D-rendered images but small branches such as the uterine or pudendal arteries are too narrow to detect pathologic changes. The arteries of the arms and legs are easily displayed on volume rendered images but maximum intensity projections (MIP) after bone removal are often more useful for diagnostic evaluation because they follow the anatomic display provided by arterial DSA. •
Principal Systemic Veins
The right jugular vein, right brachiocephalic vein, and superior vena cava (SVC) lie on a vertical axis that is perpendicular to the scan plane (Fig. 24. 2b). The left brachiocephalic vein passes over the front of the aortic arch and terminates in the vena cava. A retro-aortic variant may also occur and should not be mistaken for lymph nodes. The azygos-hemiazygos venous system establishes a collateral pathway in response to occlusion of a portion of the vena cava. The veins, which flank the aorta, are highly variable in
Gastroduodenal artery
Cystic artery
Inferior mesenteric artery
3astrofuodenal artery Superior supra duodena arteries
Gastroepiploic artery
;"
Appendicular artery
Fig. 24.3
Superior mesenteric artery
Normal anatomy of the celiac and mesenteric arteries.
Sigmoid arteries
Anatomy their diameters but normally do not show abrupt caliber changes on adjacent scans (differential diagnosis: paraesophageal lymph nodes). There are extensive connections to the spinal venous plexus. and small connections also to the (left) renal vein. The azygos vein opens into the SVC above the right main bronchus. If an azygos lobe is present. the vein will pursue a characteristic course through the upper lobe of the right lung. The intercostal veins Thl-Th4 drain into a right and left superior intercostal vein. which enters the azygos vein on the right and may either enter the accessory hemiazygos vein on the left or pass across the aortic arch into the left brachiocephalic vein giving rise to an aortic "nipple" on PA chest radiographs. In the abdomen. the azygos and hemiazygos veins continue into a network oflumbar and spinal veins. The inferior vena cava (IVC) is situated slightly to the right of the midline. passing through the right hemidiaphragm. Its intrahepatic segment is compressed into a triangular or lanceolate shape by the right lobe and the caudate lobe of the liver. The three principal hepatic veins (right. middle. and left) are accompanied by a number of accessory veins. One of these veins opens directly into the IVC between hepatic segments 5 and 6 and can be important for planning atypical hepatic resections. The veins of the caudate lobe are small and enter directly into the IVC (this accounts for the hypertrophy of the caudate lobe seen in Budd-Chiari syndrome). The renal veins enter the IVC at about the level of the first lumbar vertebral body, the left renal vein crossing in front of the aorta on its way to the vena cava. A retroaortic left renal vein usually passes obliquely downward. It is single in up to 2.5% of the population. In up to 9% of the population there is a duplicate renal vein, with one vein passing behind and one in front of the aorta. The early opacification of the renal veins
after contrast infusion
can cause
pseudothrombus artifacts within the IVC to appear at the level of their terminations. Early images show a marked decrease in CT attenuation caudal to the renal veins. There are small collaterals from the left renal vein to the lumbar venous system as well as via the adrenal veins to the splanchnic venous system. Such collaterals attain significance in IVC obstruction or portal hypertension.
The pelvic venous confluence lies just below the aortic bifurcation. The left common iliac vein passes behind the iliac arteries, where it presents an oblong cross section. Generally the main pelvic veins run posterior to the arteries. Familiarity with the vascular anatomy of the bifurcation region is important in nodal staging, because veins may not have enhanced if scanning was started too. early, and pathologic lymph nodes may enhance on delayed scans. Scans in the portal venous perfusion phase, which is commonly used for abdominal examinations, may display pseudothrombus artifacts at venous bifurcations. This is caused by differential enhancement of veins draining proximal and distal portions of the pelvis or lower extremities; the internal iliac veins and profunda femoris veins enhance earlier than the external iliac or superficial femoral veins (see Fig. 7.39). Portal Venous System
The portal vein arises from the junction of the superior mesenteric and splenic vein. On reaching the liver hilum, the portal vein divides into a right and left main branch (see Fig. lU). The left portal vein turns a short distance forward in its terminal portion, defining the boundary between the lateral segment of the left hepatic lobe (subsegments II and III) and subsegment IV. The right portal vein divides into an anterior branch, which supplies hepatic subsegments V and VIII, and a posterior branch, which supplies segments VI and VII. Numerous variants may occur, especially in the mode of division of the right portal venous branches. Some of these variations, such as a trifurcation of the portal vein or a left portal vein arising from the anterior branch of the right portal vein, may provide important information for planning partial liver resection.
The inferior mesenteric vein runs parallel to the inferior mesenteric artery in the small pelvis and lower abdomen, then turns upward in front of the left renal vein and drains into the splenic vein several centimeters proximal to the venous confluence. Large numbers of collateral pathways to the systemic circulation exist, via gastroesophageal and anorectal veins, paraumbilical veins, and retroperitoneal veins (liver capsule, common bile duct, splenorenal ligament, omentum and abdominal wall, mesenterico-ovarian). The size of the portal vein should not exceed "13mm, the diameter of the splenic
831
832
24 Vascular System vein and the superior mesenteric vein should remain below 10 mm unless portal hypertension is present.
Examination Tedhnique CfA relies on thin-section spiral imaging and optimum timing of contrast material injection to ensure proper opacification of the target territory. To obtain the best possible spatial resolution along the z-axis, all the spiral scan parameters should be optimized to produce the smallest possible effective slice thickness in singleslice CT. While z-axis resolution is a limiting factor with such single-slice techniques, near isotropic imaging is possible with multislice techniques. Signal-to-noise ratios become a more important limiting factor with multislice scanning because of low detector signals at thin sections. A section width between 1.25 mm and Table 24.3
3 mm suffices, depending on the scan length and the size of the patients. Smaller sections may be technically feasible but should be avoided for most body applications because of excessive noise or dose. In general, the scanning technique with multislice CfA becomes more simple and robust. The contrast protocol must be individually adapted to achieve a sufficiently high and constant level of vascular opacification throughout the examination (see Figs.24.6 and 7). This is done by following the steps outlined below (recommendations for scanning protocols are summarized in Table 24.3):
Suggested protocols for vascular imaging with CT
General Preparation
Not necessary
Oral contrast
None or negative contrast agents (water)
Patient position
Supine with elevated arms (standard) [Carotids: arms parallel to the body
Scan range
As short as possible: tailored to the clinical problem
Respiratory phase
Inspiration (standard) Shallow resplration (patients that are unable to hold their breath)
Window setting
Precontrast scan: a angiography:
Scan parameters
l-slice SqTf/RI
W/l-400/40 W/l= 500/150 (adjust as needed) 4-sliltea SC
16-slicea SI;:
axial SWIRl
MPRb SWIRl
5J~0-30 1 c
2-2.501
1.25-2 t
5-7.5/5
-
Pulmonary vessels (standard) (dyspneic patients)
3/5/2 t
1'-1.25 t 2-2.5 i
0.75-1.25 t 1-2 i
1-1.25/0.7 1-3/1-2
-
Thoracic aorta Abdominal aorta Thoracoabdominal aorta Carotid arteries Renal arteries (stenosis)
3/5/2 t 3/5/2 t 5/10/3 t 1/3/1 1 d 2/3/~ t 3/5/21 2/3/~ 1 3/5/2 t 3/5/2 t d
2-2.5 i 2-2.5 1 2-2.51 1-1.251 t-r.zs i ~-1.25t 1-1.251 1-1.251 2-2.5 t
1-1.5 t 1-1.5 t 1-1.51 0.5-1.25 t 0.5-1.251 0.5-1.25 t 0.5-1.251 0.5-1.25 t 0.5-1.25 t
4/3 4/3 4/3 3/3 2/2 3/3 3/3 3/3 3/3
3/3 3/3 3/3 2/2 2/2 3/3 3/3 3/3 2/2
Pre-contrast
(donors)
0;
Hepatic arteries Mesenteric arteries lower extremity arteries
parasag cor CPR CPR cor cor cor CRR
Continue ~
Examination Table 24.3
Technique
833
Continue
Scan parameters
l-slice SC/TF/RI
Lower extremity veins Portal vein Superior vena cava Inferior vena cava
5/10-15/10 3/5/2 ! 3/5/2 ! 5/8/2 !
Contrast injection
4-slice" SC
t
2-2.5 2-2.5 1-1.25 2-2.5
i ! t !
Scan duration
16-slicea SC 1-1.5 1-1.5 1-1.5 1-1.5
t ! i !
axial SWIRl
MPRb SWIRl
5/5 4/3 4/3 4/3
-
monophasic V+N/F/D
biphasic V,/F, + V2/F2/DX.'
3/3 cor 3/3 cor 3/3 cor
f
Arterial CTA
105 205 305 405 ,",50s
70 + 50/6/10 A 100+50/5/8A 150 + 50/5/5 A 150+50/4/5A 150+50/3/5A
50/5 + 30/3/8 A 50/5 + 60/3/5 A 50/5 + 90/3/5 A 45/4.5 + 105/2.7/5 A
Pulmonary arterial CTA
55 105 205 305
40+20/4/5 P 60 +40/4/5 P 80+ 50/4/5 P 120 + 50/3/20
-
40/4 + 30/3/5 P 40/4 + 60/3/5 P
120+60/5/30A
-
60:90/5/15h 120 + 60/2/150 -/-/3 min i
-
Portal venous CTA CT venography (CTV) direct enhancement indirect enhancement indirect enhancement
(chest/abdomen) (legs in pulmonary
-
- .-
embolism)
SC= slice collimation (mm). TF= table feed (mm/rotation), RI= recon. increment (mm), t t = scanning direction, SW = effective section width (mm), MPR= multi planar reformats, axial = axial sections, cor = coronal, sag = sagittal CM = contrast material, Contrast concentration = 300 mglml iodine (optional 400 mglml for 16-slice scanning) V=volume of CM (rnl), N =volume of saline flush (ml), F=flow rate (mils), D = start delay, X = trigger region a Pitch P = TF/(NxSC) should be chosen close to 1.5 with 4-slice scanners, 1.3-1.5 with 16-slice scanners b
MPRare reconstructed from a "secondary raw data set" with SWIRl = 1-1.5/0.7
C
MIP/VRT routine display mode; CPRfor stenoses suspected on axial sections, MIP or VRT sequential CT
d e
or 0.5-0.8/0.5
localized examinations only (region of suspected pathology) read: 5-30A (5P) = start delay 5-30 s (5 s) after reaching 50 HU enhancement in the aorta (pulmonary artery)'
r basic injection, e.g. 50/5 + 30/3/5A-?50 ml CM at 5 mils flow followed by 30 ml CM at 3 mils, delay 5A' 50 ml saline flush at 2.7-3 mils optional 9 h
choose lower dose and kVp setting (CTDI""r = 80-100 kVp, depending on patient size and CT scanner) dilution 60 ml CM: 90 ml normal saline
r scan in combination with CTA of the pulmonary vessels, total start delay 3 min
• Perform (discontinuous) pre-contrast scans in cases with suspected hemorrhage, mural hematoma. • Determine the scan length L by means of a scanogram, which may be supplemented if necessary by low-dose noncontrast scans. • Administer a test bolus to determine the scan delay time D. The test bolus may be omitted if bolus triggering is available, i.e., the scan is automatically initiated when intravascular contrast appears.
• Determine the scan parameters as a function of the scan length and scan duration TI. • Define the contrast protocol, taking into account the patient's body weight « 2.5 mg iodine/kg body weight) and any impairment of renal function. • Evaluate the images interactively on the viewing monitor, and construct maximum intensity projections (MIP), shaded surface displays (55D), or volume-rendered images.
I
24 Vascular System Patient Preparation
The patients should refrain from eating a few hours before the examination but otherwise no special preparation is needed unless there is known deterioration in renal function or an otherwise increased risk for contrast material application (see also Chapter 3, p. 85). No positive oral contrast material should be employed. This is especially important to remember ifCTA is performed as part of another examination, for example a biphasic liver CT. Acquisition Technique Pre-contrast Scans
Pre-contrast scans are only required in patients with suspected hemorrhage, and mural hematoma in suspected aortic dissection. Rarely, it may also be helpful in suspected aortitis. In most cases, however, such precontrast scans offer only little additional information over those that can be obtained from the contrastenhanced CT angiograms. Because the abnormalities extend over larger regions, discontinuous scanning (e.g., 5mm sections every 20mm) with single-slice CT or low-dose, thick-section scanning with multislice CT (e.g., 4 x 5 mm detector configuration, 7-10 mm section width) usually suffice. Pre-contrast scans are rarely necessary for determining the scan range. In single-slice CT, however, long scan ranges require the use of thicker collimation and thus reduce spatial resolution, or require increased scan time and thus need more contrast material and make motion
Table 24.4
artifacts more probable. For -this reason, precontrast scans can help decide whether an aneurysmatic dilatation of the aorta involves both chest and abdomen, or whether CTA of one of these regions suffices. Scan Length
Scan length is no limiting factor in multislice CT, but is critical for spatial resolution in singleslice CTA. However, keeping the range to a minimum reduces scan time, allowing for thinner collimation and thus better spatial resolution, with reduced motion-related artifacts, contrast material requirements, and patient dose. The cranial and caudal extent of the acquisition can usually be determined on the scanogram for most vascular structures (Table 24.4). Scan Duration and Breath Holding
CT angiography is performed during an inspirational breath hold. As a rule, the duration of breath holding should not exceed 30 seconds. It is helpful to have the patient hyperventilate briefly (i.e., perform several forced inhalations and exhalations) before each scan. Informing the patient about the procedure and the importance of the breath hold is an important part of a successful CTA examination. In critical patients, the technologists should try the breath holding maneuver before the actual scan and watch the patient's abdomen. If it moves during the breath hold phase, the patient has to be instructed again. In particular, movement artifacts may occur if the patient relaxes the diaphragm
Scan range for various vascular structures
111111111111111111111111111111111111111111111111111111111~rn~
Multislice CT
Carotids
Below skull base tojugulum (12 ern)
Vortex or 2 cm above skull base to aortic arch
Pulmonary arteries
Diaphragm to aortic arch
Whole chest Diaphragm to aortic arc (dyspneic patients,L_
Thoracic aorta H
"
From below aortic root to above aortic arch (including brachiocephalic trunk)
Whole chest
From above celiac artery to iliac bifurcation (internal/external) or inguinal region
From above celiac artery (ThIO/IT) to inquinal reqion
_
BT'
Abdominal aorta
"1
'
co,
.
, .
Acquisition
Scanning Parameters
during the breath hold or if there are involuntary diaphragmatic movements towards the end of the scan phase. The images obtained during the test bolus series are a good indicator of patient compliance. If the patient is not able to hold the breath for sufficient time. the scan duration should be reduced. This is feasible for most emergency examinations but reduces diagnostic evaluation of smaller vessels because of the need to employ a thicker collimation. If this is not possible. no attempt should be made to force a breath hold examination. In such cases, the patient will continue breathing anyway, resulting in major breathing artifacts. lt is better to have the patient hyperventilate before the scan and then breathe shallowly. This is especially important in patients with suspected pulmonary embolism. In single-slice CTA, the scan duration can be increased to 40-60 seconds by first scanning the part of the imaging volume that is susceptible to respiratory artifacts (e.g., the upper abdomen). After 30 seconds have elapsed, the patient is instructed to resume breathing while the less motion-sensitive part of the volume is imaged. This technique can be used in examinations of the abdominal aorta and renal arteries as a means of reducing the slice thickness. In multislice CTA, scan duration can be substantially reduced for many indications. This should not make the technicians less careful in instructing the patients. A sudden breath hold command in an unprepared patient will often result in substantial breathing artifacts either during the beginning or the end of the scan.
Table 24.5 RT ls ls ls 0.75s 15 0.55 0.4s RT «tube
Single
16 rotation
time, TI
TI
Interpolation
255 245 305 305
360' LI 180'LI 180'LI 180'LI
305 205 85
cone beam
=
scan time,
180'LI 180'MLI
SC
=
Slice
Parameters
Standard parameters can be used in most CT vascular examinations. The parameters listed in Table 24.3 are for single-slice scanners with 1 second tube rotation that permit at least a 50 second scan duration. These parameters may have to be adjusted for the available scanner technology, since different technologies substantially affect the z-axis resolution (Table 24.5) and the available scan length. In very obese patients in whom the maximum mAs settings do not suffice, the use of a thicker collimation may be necessary to improve the signal-tonoise ratio. As described in Chapter 4 (p. 117) it is always desirable to keep the slice collimation SC as small as possible and use a pitch P close to 2. A pitch of 3 may be used for selected indications such as scanning of the carotid arteries or the pulmonary arteries with 1 mm collimation. A 1800 interpolation algorithm and a soft convolution kernel improve the z-axis resolution and reduce image noise. For optimum results of multiplanar reformations and 3D reconstructions, the images should be reconstructed with a high degree of overlap (reconstruction increment RI ss section widthSW/2; Fig. 24.4). Dual
Slice
Parameters
With dual slice scanners, the z-axis resolution can be substantially improved as compared to single slice scanners. In principle, the collimation could be split in half without altering the scan duration, but often it more advisable to re-
CT of the thoracic aorta: effect of scanner technology Detectors
Technique
(scan length 24 em) on z-axis resolution
SC
TF
10mm 4mm 3mm
lOmm lOmm 8mm 6mm
2mm Imm O.5mm
6mm 12mm
5mm
slice collimation,
Pitch
P
p"
12.8mm 6.4mm 5.1 mm 3.8mm
8mm
TF «table
feed, SW
SW
1.5 1.5 =
section
24 width
2.6mm 1.3mm O.5mm (FWHM)
835
24 Vascular System
b
a
Fig. 24.4 Influence of overlapping image reconstruction on the image quality of CTA of the renal arteries: SCITF/RI= 2/2/2 (a) and 2/2/1 (b). Note the display of the small, partially stenosed renal arteries on the left side.
duce the scan duration as well and chose a slightly wider collimation (Tables 24.3, 24.5). This reduces breathing artifacts and makes it possible to use less contrast material. A pitch P of 2 or slightly less (P" < 4) should be used for most indications together with a 180 interpolation algorithm. Only in very obese patients in whom the maximum mAs settings have been reached may it be necessary to use a pitch P of 1 (P*=2) together with a 360 interpolation to decrease image noise. Under these conditions, the same collimation as with single slice CT has to be used. The coverage and z-axis resolution are still equal to that of single slice CTA with pitch 2 but signal-to-noise ratio is better at identical mAs settings. 0
0
•
Multislice Parameters
With 4-slice CT scanning, the number of scanning protocols is substantially reduced. Only in very obese patients may it be necessary to use a section collimation above 2.5mm. Most frequently, volumetric imaging can be applied in the neck, chest, or abdomen (Table 24.3). With current 4-slice scanners, detector configurations of 4 x I mm or 4 x 1.25 mm collimation will be used. Volumetric imaging is important if small vessels, such as aortic side branches or peripheral pulmonary arteries, have to be evaluated. For the aorta (chest, abdomen, or thoracoabdominal), fast spiral scanning with 4 x 2.5 mm collimation (4x2 mm on Toshiba units)
usually yields sufficient results and allows for substantial reduction in scan duration and contrast material volumes. In patients with aortic dissection and abdominal symptoms, however, volumetric imaging (4x1-1.25 mm) may improve evaluation of abdominal side branches but requires 40-50 s scan duration, even with 0.5 s 4-slice scanners. Volumetric imaging with 8- and 16-slice scanners is possible with a collimation of 8 x 1-1.25 mm or 16 x 0.5-1.25 mm which substantially reduces scan duration by up to a factor of 4 (see more details for 8- to 16-slice scanners in Chapter 4 and Table 4.6). Fast spiral scanning using an 8 x 2.5mm collimation with 26mm table feed allows for covering the chest or abdomen in 4-8 s. A 16 x 1.5 mm collimation with a 36 mm table feed will achieve the same in 3-6 s, and a 16 x 2mm collimation with 44mm table feed will require 2-5 s. Such extremely fast scan durations substantially improve evaluation of dyspneic patients and allow for reduction of contrast material volumes. Timing of contrast material, however, will have to be adapted accordingly.
Dyspneic patients with suspected pulmonary
embolism can be examined within some 5s
using fast spiral scanning with 4-slice scanners and 4 x 2.5 mm collimation if only the central portion of the pulmonary arterial system is covered from the top of the lower half of the diaphragm to the top of the aortic arch. In general, however, near-isotropic imaging is preferred for the pulmonary vessels. Proper breath-hold instructions, are mandatory. If breath holding is impossible, shallow breathing should be attempted. Current 8- and 16-slice scanners require no modifications of scan protocols relative to normal patients. For the peripheral vessels, fast spiral scanning with 4-slice units is able to cover the abdominal aorta and the legs (some 120cm) in 40-64 s, depending on the rotation speed of the x-ray tube. With 0.5s rotation, near isotropic imaging becomes feasible within 75-80 s scan duration if (N x SC/TF =) 4 x 1/8 or 4 x 1.25/7.5 protocols are used. Contrast opacification of the arteries usually poses no problem, despite long scan times, because the bolus geometry widens on its way to the feet. There is increasing venous overlay, however, as scan duration increases. Using 8- and 16-slice scanners the scan duration can be reduced by up to a factor of 4, even with
Acquisition
high spatial resolution. If scanning is too fast, however, the bolus may be passed by the scan and distal vessels may be inadequately opacified. For this reason, a reduction of pitch may be advisable to keep scan durations in the range of20-30 s.
CT Perfusion Imaging (CTP) CT perfusion is firmly established only for the brain (see below, p. 895), although applications elsewhere in the body have been suggested for a long time and commercial software is now available. Multiple scans in rapid succession (or even during continuous radiation) are obtained during the injection of a contrast material bolus. The various sections must be registered to obtain the enhancement curve (time-density curve) of each voxel within the organ of interest. For this reason, the organ of interest must be free of (respiratory) motion during data acquisition. The enhancement curve is then used to calculate various perfusion parameters (see below). Perfusion depends on the velocity and volume of the blood flow through large capacity vessels as well as through the capillary vascular bed. CTperfusion usually applies first-pass evaluation using kinetic models that assume that the contrast material is non-diffusible, and neither metabolized nor absorbed by the tissue bed studied. A simple model, which uses the slope of the enhancement curve to determine regional blood flow (RBF) , is most robust. It relies on the Fick principle: change in enhancement (C) in a ROI is proportional to the blood flow x difference in concentration of contrast material between feeding artery (Co) and draining vein (Cv):
dCWdt=RBFx (Ca(VCv(V) This model requires a delta impulse to work best, which means that a sufficient amount of contrast material (e.g., 50ml) has to be injected within as short a time as possible (i.e., using large bore 14 G venous cannulae and a flow rate of8-1 Omlis) . Since the absolute concentration of contrast material in the artery is not known, only relative numbers can be obtained for RBF. The time to peak (TIP) is easily established. Deconvolution models provide both qualitative and quantitative data and allow for slower
Technique
injection rates (50ml at4-5mlls). They use the time-density curve in a feeding artery (which has to be contained in the imaging volume) as the input function that is used to deconvolute the tissue time-density curve to determine RBF, regional blood volume (RBV) , and mean transit time (MTT). Scanning at 80kVp is suggested for both techniques because it provides superior contrast at a lower dose. Scan duration is usually 45-50 s for the brain.
Dose Issues With optimum contrast opacification of the target vessels, wide window settings can be used and a higher image noise can be accepted as compared to standard CT indications. At the same time, reconstruction of thick multiplanar reformations may reduce the contrast of small vessels because of increased partial-volume effects. Thus, there is a delicate balance between dose reduction, acceptable image noise, and vessel contrast. In general, higher doses are necessary for near-isotropic imaging, while less dose can be applied for fast volume protocols. In standard-sized patients, a CTDIvoi of 5-6 mGy will suffice for the chest, while 8-15 mGy will be required for the abdomen. In more obese patients with larger body cross-sections, substantially higher dose may be necessary. In the chest and in slim patients, lowering the kVp settings is a technique to increase vascular contrast because of the higher attenuation of iodine at lower x-ray energies. For CTA of the pulmonary arteries, the use of 100 kVp instead of 140 kVp will, on average, increase vascular opacification by some 80-100 HU for identical injection parameters, and thus allow for scanning the chest with a eTDIvoi of 3-4 mGy. The resulting effective dose will be in the range of 1.5-2 mSv. A tube voltage of 80kVp holds an even higher potential for dose savings in slim individuals and children (effective dose less than 1 mSv) but it requires substantially higher mAs settings in more obese patients because of less penetrating radiation at low kVp settings. Such low settings therefore should be avoided in obese individuals. In the abdomen, x-ray attenuation is also higher, and reducing the kVp settings will only be advantageous in slim patients and allow only for minor dose reduction.
24 Vascular System Adaptive tube current modulation and zaxis modulation is an excellent technique for further dose reduction by some 10-30%, depending on the body region, It should be used wherever possible, For similar image quality 16slice scanners require some 20-30% less dose than 4-slice scanners,
Contrast Injection Contrast injection is critical for CTA procedures, Optimization becomes more important for multislice scanning, especially when decreased volumes of contrast material have to be used, It 'is important to obtain a sufficiently high degree of vascular opacification throughout the imaging volume and to display the vessel of interest with as few artifacts as possible, As a general rule of thumb, small vascular calibers, in-plane vessels, and greater slice thickness require a higher degree of vascular opacification to ensure that the vessels can be evaluated despite partial- volume effects, Only a moderate level of enhancement (» 100 HU) is needed in studies that are limited to large vessels (such as the thoracic aorta or the deep veins of the trunk) or vessels that run perpendicular to the scan plane, such as the carotids and the vessels of the pelvis and lower limb, Contrast volume and flow rates are less critical in such examinations, Good vessel opacification, however, will allow for using less patient dose during data acquisition, For evaluation of suspected pulmonary embolism and for the arteries of the upper abdomen, spatial resolution is critical, In addition, vessels may run parallel to the scan plane and are therefore more susceptible to partial volume effects, For these indications, optimum vascular opacification is mandatory, A similar requirement exists in regions where 3D reconstructions are desired, Good results have been .achieved with the protocols shown in Table 24.3,
Positioning and Venous Access For cubital venous access in CTA, a sufficiently large-bore intravenous catheter « 18 gauge) should be used to enable a high contrast flow rate, The injected arm should be stationary and well extended to prevent catheter kinking and
reduce venous compression effects, Right cubital venous access is preferred for thoracic CTA, because this will avoid streak artifacts caused by the approximately horizontal course of the left brachiocephalic vein, Good results are also achieved by positioning the left arm above the head and the right arm parallel to the body and using right cubital venous access for the contrast injection, This technique reduces compression effects in the area where the brachiocephalic vein crosses the ribs, Central venous lines should be avoided if flow rates above 2mlls are required, The only exceptions are large bore catheters such as for dialysis or hemoftltratton. It is always advisable to test catheters or check with manufacturers to determine the maximum flow rate and pressure that is still safe,
Contrast Volume and Flow Rate The goal of contrast injection in CTA is to cause a constant and sufficiently long phase of vascular enhancement during which the actual data acquisition can take place, Vascular enhancement for CTA of the arterial system is an off-equilibrium effect that is driven by the steady inflow of contrast-enhanced blood from the injection veins via the pulmonary circulation into the aorta, Soon after the injection stops, the vascular enhancement decreases as well. Vascular enhancement depends on dilution effects and is therefore strongly influenced by cardiac output Patients with a high cardiac output, such as young adults, suffer from more dilution, and thus less enhancement, than patients with a reduced cardiac output, such as an acutely ill patient in shock. Recirculation effects will cause a continuous increase of contrast enhancement during the plateau phase, leading to maximum enhancement close to the end of the scan, For most standard indications a monophasic contrast injection is sufficient The user has to choose the injected volume V and the flow rate F as well as the scan delay D, These three parameters are adapted to the clinical imaging task and the resulting scan duration (Fig, 23,5; Table 24,3), The results can be further improved, however, if a saline flush is used to better utilize the injected contrast material. More homogeneous enhancement is gained by a
Contrast Injection biphasic injection technique or even an exponential decrease of flow over time. Using mathematic modeling or test bolus injections. a predefined level of contrast enhancement independent of the individual circulatory state can be achieved with good approximation (see below. Individualization of Contrast Material Injection). For most standard applications. however. the following rules have to be considered.
Saline Flush
The injected contrast material can be used to better advantage if the injection veins are flushed by a bolus of normal saline, injected immediately after the contrast material using similar flow rates. Such a saline flush pushes the contrast material forward and thus increases the length of the contrast plateau. It can also be used to save contrast material without jeopardizing vascular contrast, and to increase vascular opacification if combined with higher flow • A larger contrast volume will prolong the rates. It is particularly important when small plateau phase of enhancement. amounts of contrast material are injected, and • Enhancement increases over the duration therefore becomes essential for short multislice of the plateau phase due to recirculation CT acquisitions. In addition, a saline flush will effects (monophasic injection). A reduce the amount of artifacts in the superior • Hence. larger volumes will increase maxi-A 1 vena cava and brachiocephalic injection veins mum enhancement. when chest CTA acquisitions are obtained in a • Maximum enhancement will occur close caudo-cranial scanning direction (see Fig. 24.13). to the end of the plateau phase. In a single barrel CT injector, layering is • Vascular enhancement increases as more necessary so that contrast material is injected iodine is injected per second. first, followed by normal saline (see procedure • This can be achieved either by higher flow in Chapter 3, p. 99). Because it speeds this prorates or higher contrast concentration. cedure up, we suggest pointing the syringe • Higher flow rates decrease the duration of downwards and pulling up saline first when the plateau phase. loading the injector, followed slowly by contrast • The flow rate F should be at least 3 ml/s. material. Such a technique will not provide a better 4-5 ml/s. for all indications in sharp fluid-fluid level within the syringe but a which high intravascular contrast is resmall mixing zone at the interface between the quired. fluids. This procedure will have to be repeated • Higher contrast concentration increases! for each injection. viscosity and requires pre-warming of the" Double-barrel injectors are presently encontrast material (incubator). tering the market and make a saline flush tech• The duration of the contrast plateau may nique much easier to perform; no refill is necesalso decrease with higher contrast concensary between patients, and preparation of contration unless a saline flush is used to push trast injection is much faster. In addition, such the (high-viscosity) contrast material forinjectors allow for multiphasic injection protoward. cols. The contrast injection can be further op• As a rule of thumb. the volume of contrast timized if normal saline is injected before the material should be equal to the scan duramain contrast bolus as well, because it will open tion (s) x flow rate (mlls). up the injection veins and will fill small'col • For short scan duration (AlOs; multislicel lateral veins with saline, thus creating a more CT). higher volumes, higher flow rates or sharply defined enhancement curve. higher contrast concentrations, and a saIt is not yet clear how much saline has to be line flush are needed. injected for optimum results. Based on our clinical experience, we suggest 40-60 ml. When determining the proper scan delay, any saline that The risk of contrast extravasation with higher is left in the injection lines has to be taken into flow rates can be reduced with pressure-limited account. This is no problem when bolus triggerinjectors, which stop contrast injection when a ing techniques are used but may be problematic sudden pressure increase occurs, or with 30when scan delay is derived from a test bolus incalled extravasation detection accessories. jection.
839
24 Vascular System has shown, however, that the best results are achieved by increasing this delay time by some 10-20% when determining the definitive scan delay D because a larger amount is injected during the actual CTA data acquisition (Table 24.6).
Scan Delay Although correct timing of the contrast material bolus is critical for optimum results of CTA, individual timing is not always mandatory but may avoid occasional failures to obtain adequate vascular opacification. It is particularly important in emergency patients with decreased cardiac output and prolonged circulation time. Individual timing is based on a test bolus injection or bolus triggering.
Bolus Triggering
Bolus triggering (or bolus tracking) is a more elegant technique than a test bolus, because the contrast bolus itself is used to initiate the scan. Most manufacturers equip their scanners for bolus triggering. This involves performing monitor scans with a substantially reduced dose at a suitable table position (trigger region) and defining a region of interest (ROl) in which vascular enhancement is to be measured. Once a sufficient level of enhancement has been reached (determined visually or by a threshold attenuation value), the table moves to the beginning of the scan range, and the actual spiral acquisition is initiated.
Test Bolus Injection
A test bolus of 10 (to 20) ml of contrast material is injected intravenously by rapid manual injection or power injection at 4-5 ml/s. and several low-dose scans are performed at a reference level that is chosen close to the beginning of the scan range. Using 80-100 kVp further decreases dose and improves contrast enhancement of these test scans. Scans are acquired at 1-2 second intervals. depending on the expected scan delay. Even these test scans should be initiated after a certain delay time of usually 8-12 seconds following the start of the test bolus injection. The test scans are evaluated by identifying the image that displays maximum opacification in the region of interest (e.g.. the aorta). The time delay At of this image from the start of the injection determines the scan delay. Experience Table 24.6
Some scanners offer a continuous fluoros-
copy mode for monitor scans. This mode should be avoided in favor of scans every 1-2 s because it may increase patient dose substantially. There is an additional delay between the time at which the threshold level is reached and the time the actual data acquisition is started. This delay varies from scanner to scanner and may be as high as 7 s. This makes it necessary to use
Test bolus and bolus triggering
Test bolus Reference
region:
Level of pulmonary Cranial
Contrast
protocol:
Scanning
protocol:
Scan delay D:
border
V/F!D=
bifurcation
(thorax)
of scan volume
(abdomen)
10/5/8-12
5 mm slice thickness, sequence: 20-30 s Calculated opacification D=AtX".2a
minimal
from the time ofthe target
dose, scans acquired
at 1 to 4s intervals,
interval At from the start ofthe structure
injection
duration
of
to maximum
Bolus triggering Reference Contrast
region: protocol:
Scanning Scan 3
delay
protocol: D:
Ascending
aorta
or left ventricle
No test injection Slice thickness intervals Scanning
add 5-10 s for short scan duration
same as for spiral scanning,
initiated
5-7
S3
after attenuation
and large contrast
volumes
minimal
dose, scans acquired
has increased (16-slice
scanners)
by 50 HU3
at 1 to 2 s
Contrast Injection a low threshold level to determine contrast arrival in the target region, and may even make it advantageous to place the trigger region upstream. For example, a trigger level in the ascending aorta may be used for CTA of the chest, or a level in the descending aorta for CTA of the abdomen. Too short a delay between the initiation of the scan and the actual start of the spiral acquisition is also not advisable because patients have to breathe in and hold their breaths during this short additional period of time. Proper patient instruction and a scan delay of 4-5 s after reaching the trigger level is advisable. With fast multislice acquisitions using 8and 16-slice scanners, however, a higher trigger level should be set or the start delay after reach-
ing this level should be increased. The rationale behind this approach for very short acquisition times is to delay scanning until a relative peak enhancement is reached (Fig. 24.5). Some systems allow for optimization ofROI placement during the monitor scans to be able to correct for malpositioning. If placement of the trigger ROI is difficult it can be advisable to manually start the scan as soon as contrast material arrives in the target region. •
Venous Imaging
Examinations of the venous system may be performed during the phase of direct vascular enhancement or during the recirculation phase. In either case it is unnecessary to individually determine the scan delay.
,0,.HU
,0,.HU
400
400 300 -
Enhancement "plateau" grows
200
200 Optimum scan window 120 ml eM 4mljs
100
100
O~~--~~~--~--~~------~ o
20
40
60
80
20
400
400
300
300 120 ml CM 4ml/s
80 Time
80m1CM+ 50 ml NaCI 4ml/s
200 100
100
o~~--~------~----~------~ 40 20 60 80 o
Time
Fig. 24.5 Effect of contrast material (eM) injection protocol on arterial enhancement (simulated data, compare also Fig. 3.3). With monophasic injection protocols the enhancement increases over time because of recirculation effects (a). The optimum time period for scanning (scan window) is reached when the enhancement exceeds a predefined target value. Increasing the flow rates increases the maximum
60
,0,.HU
,0,.HU
200
40
Time
enhancement
but
O~-e~-L
o
20
L-
40
~
60
~
80 Time
shortens the scan window (b). If the contrast volume is reduced (e.g., because of short scan duration with 16slice scanners), only a short period of peak enhancement is attained and the maximum enhancement is less than with larger volumes (c). A saline chaser bolus (saline flush) has a similar effect as increasing the contrast volume (d).
841
24
Vascular System
during the phase of direct material should be well diluted (1 :5) to avoid high-contrast streak artifacts that would hamper vascular assessment. A scan delay of lOs is sufficient during this phase (see Fig. 24.89). For imaging during the recirculation phase, there is usually an equilibrium of venous and arterial enhancement, and the optimum scan delay is 40-60 s for the superior vena cava, some 100 s for the inferior vena cava, and some 200 s for the legs. The latter is important when an examination of the peripheral veins is added on top of aCTA of the pulmonary arteries (CT venography). For this particular indication, and additional 2-3 min delay period should be added after the pulmonary CTA has ended (see Fig.24.87d), and scanning from the periphery up into the abdomen yields best results. For examinations
enhancement, the contrast
Individualized Contrast Material Injection Ideally, the contrast material injection should provide a predefined and constant enhancement level that is guaranteed over the whole duration of the scan. At present, there are two approaches that come close to this goal (see also Chapter 3, p. 103). Instead of using a monophasic contrast injection, both vary injection rates over time to achieve the desired contrast enhancement curve ("bolus shaping" techniques). Mathematical
Modeling
Mathematic modeling uses a complex pharmacological flow model to calculate the required volume and flow rate from individual patient parameters (sex, weight, and height) and various assumptions regarding cardiac output, organ perfusion, and other physiologic parameters. These parameters are used to determine the enhancement over time for any arbitrary volume and flow rate of contrast material. In addition the technique allows estimation of the amount of contrast material that has to be injected to obtain a predefined enhancement plateau. U sing this technique it can be demonstrated that an exponential reduction of the injection speed yields a constant enhancement plateau if the flow rate is adjusted to the individual patient (Fig.24.6b). In addition, the model predicts that the maximum enhance-
ment does not increase linearly with flow rate but that the gain in enhancement becomes less with higher flow rates because saturation effects due to the limited total blood volume. The technique has not yet been integrated into a commercially available product and relies on a number of assumptions, e.g., regarding cardiac output, which may vary between patients. The technique works for the many patients and allows for an individual increase or decrease of the injected contrast material volume. Test Bolus Technique
This technique assumes that the body responds in a linear fashion toward the injected contrast material. This response can then be described as an individual patient function that describes how the injected contrast is transformed into a particular enhancement curve. The patient function is derived from the test bolus injection and includes information about first pass as well as recirculation effects. With this patient function the individual response to a given contrast material injection can be calculated. Conversely, the patient function can be used to determine the flow rates required to obtain a predefined enhancement curve (Fig. 24.6 c). With the current prototype system, a 16 ml test bolus is injected, followed by 40 ml of normal saline. The resulting time-density curve is then calculated for the target vessel or the aorta at that anatomic level. The time-density curve is then used as the basis of determining the patient function. Then the required enhancement level (e.g., 250 HU) and the duration of the plateau phase is entered into a computer program, and the computer uses the patient function to calculate a biphasic injection protocol that best fits the desired enhancement. Such a protocol is based on a smaller volume and a fast flow rate first, followed by a larger volume at a lower flow rate. It could be shown that such a technique yields more homogeneous contrast enhancement over the desired scan phase than a standard monophasic injection. However, there is still some variation as to the actual enhancement level that is reached (up to 50HU). This may be due to the fact that the assumption of a linear response is not strictly accurate because of the limited total blood volume and resulting saturation effects, or that some of the test bolus is pooled in small veins.
Contrast Injection ~HU
~HU
400
400
300 -
300 _ 105 ml eM
200
200
100
exponentially
100
120 ml eM 4 m1/s
o~----------~--~--~----~ o
20
decreasing flow
40
60
o~-=----------------------~ 40 o 20 60 80
80
Time
Time
b Fig. 24.6 Effect of bolus shaping techniques on contrast enhancement, (a) Monophasic injection with a constant flow rate leads to an increase in arterial enhancement over time. (b) A patient-specific exponential decrease in enhancement can achieve a real enhancement plateau, (c) Individualized biphasic injection techniques substantially improve homogeneity of enhancement as compared to monophasic injection and can be implemented on most CT injectors.
~HU 400 300 200 40 ml eM 100
6.7ml/s + 76 ml eM 2.8 mils
O~~--------~------~----~ 80 o
20
40
60
Time
c
Even where such individualization techniques are not available, biphasic injection is still advantageous. As a rule of thumb, some 50 ml can be injected first at 5 mlls, followed by an amount of contrast material that is chosen so that the total injection time (including the lOs of the first phase) is identical to the scan duration and is injected at a flow rate of 3 mlls (see Table 24.3). Contrast Material in Patients with Reduced Renal Function
With single slice CTA, the volume of contrast material is rarely below 100ml, and is frequently about 150ml or more. Iodinated contrast media are known to be nephrotoxic, and require proper hydration of patients (see Chapter 3, Patient Preparation and Contrast Media Application). Patients with compensated renal
failure should not be examined by CTA unless other imaging modalities, such as contrast -enhanced MRA or duplex ultrasound, do not provide sufficient information. In such cases, proper patient preparation is required to keep the risk to a minimum. Multislice scanning using a fast spiral acquisition with 4 x 2 mm, 4 x 2.5 mm, or 4 x 3 mm collimation reduces the scan duration for the chest or abdomen to 10-12 s, and to 4-6 s for the central pulmonary vessels or the renal arteries. As a consequence, the volume of contrast material may be reduced substantially and often may be kept below 50ml (see Table 24.3). In properly hydrated patients, such low volumes of iodinated contrast material are usually well tolerated. Gadolinium compounds (MR contrast agents) are an alternative for patients with severe allergies to iodinated contrast material who cannot be examined by ultrasound or MRA.
843
24 Vascular System However, they give less enhancement at the centricity of some lesions such as stenoses, same injection rates (comparable to 100- aneurysms, and injuries. Interactive multiplanar 150mg/ml iodine). The total amount of these reformations (MPR) are best suited for this task but again scrolling speed is an important issue. agents required for CTA can be quite high, which In addition, no loss in spatial resolution should may also lead to nephrotoxic effects in patients be accepted when the sectional plane is with severely impaired renal function. changed interactively. Longitudinal sections help identify the length and shape of a lesion Evaluation and Documentation while sections perpendicular to the vessel's Depending on the clinical imaging task, evalua- course help to define the vessel diameter pretion of CTA is primarily based on axial sections cisely and thus the degree of stenosis or the size (aortic dissection, pulmonary embolism) or on of an aneurysm. Curved planar reformations (CPR) along the various 3D displays (aortic aneurysm, pulmonary vascular anomalies). The various 2D and vessel's course are much more comfortable to 3D processing techniques are also discussed in evaluate than standard multiplanar reformations (Fig. 24.7). To use this feature effectively, Chapter 2. the workstation must provide a means of scroll• Axial Sections ing through the data set when defining the Evaluation of axial sections requires an interac- curve on the reference sections. In addition, intive cine mode because of the large number of teractive corrections of the points that define sections, which vary between 80 and 400 for this curve should be possible because positionsingle slice CTA, and between 150 and 1500 for ing errors may frequently occur in more commultislice CTA. The reviewing workstation must plex vascular situations. An immediate update be capable of scrolling through these large data of the resulting CPR further improves handling. sets rapidly. Scrolling speed should be intui- Still, such a process can be quite time-consumtively and interactively adjustable so that fine ing. For this reason, technicians should be detail can be evaluated more slowly while large trained to perform standard tasks such as CPR through the renal arteries, so that the radioloranges can be covered rapidly. Not all current systems are equally suited for this task, es- gist can focus on reporting these images instead pecially when large multislice data sets are in- of having to struggle with technology. One has volved. Interactive viewing is important for the to keep in mind, however, that off-center longievaluation of small vessels that cross the scan tudinal sections can simulate vascular stenosis. plane several times, and for evaluating the ves- CPR is mandatory for evaluation of stents or arteries with heavy wall calcifications such as pesel wall and vessel lumen. ripheral arteries or the carotids, especially in • Multiplanar and Curved Planar diabetics and patients with (secondary) hyperReformations parathyroidism (see Fig. 24.15). Vessel tracking software, which semiautoIf abnormalities are suspected, evaluation in a matically finds the centerline of a vessel second plane is mandatory because of the ec-
c
a
b
Fig. 24.7 Curved planar reformation along both renal arteries. Anteroposterior view (a), caudocranial view
(b), and section perpendicular to the vessel axis in the region of an eccentric stenosis (c).
Evaluation and Documentation without much user interaction, is now commercially available (see Fig. 2.44). Most programs require the user to place a proximal. distal. and possibly some intermediate points on the vessel. and then automatically track the vessel between these points. Such vessel tracking software makes it feasible to evaluate the vessel of interest interactively with respect to wall changes (soft versus hard plaques. aneurysms). measurements of the true length of an abnormality. or relationship of the vessel to surrounding structures (e.g .. pancreatic carcinoma). A feature that allows the imaging plane to rotate around the centerline of the vessel. thus making it possible to evaluate eccentric abnormalities. is particularly helpful (Fig.24.7 a. b). Stretching the vessel out can simplify length measurements but causes the anatomic overview to deteriorate. Vessel cross-sections perpendicular to the centerline substantially improve the grading of stenosis or evaluation of other abnormalities (Fig.24.7c). In addition. software tools are available that automatically calculate the local vascular diameter and display the location and degree of maxim um vessel stenosis or dilatation. Such tools are most reliable for the aorta but may fail for smaller vessels. especially in cases with increased image noise (obese patients). crossing or branching vessels (errors in determining the vessel wall). or limited spatial resolution (too thick a collimation). Thick CPR may be helpful in obese patients and for eccentric stenoses. By increasing the thickness of the CPRjust above the diameter of a vessel. the same effects as in conventional angiography or DSA occur: the local density of a vessel decreases as it gets thinner. This will happen close to the vessel wall but it will also affect stenotic segments, which will appear less dense even in vessels with eccentric stenoses and thus can be more readily appreciated (see Fig. 2.15). In addition, this technique will more readily appreciate irregularities of the vascular lumen, even when not cut perpendicularly. Another beneficial effect is the reduction of image noise. •
Maximum Intensity Projections
Maximum intensity projections (MIP) are an excellent tool for providing angiography-like images of the vascular system (see Chapter 2). They optimize contrast between a vessel and the background, and are therefore excellently suited for displaying small and peripheral ves-
sels that are otherwise hard to demonstrate by 3D techniques. For most applications, prior editing of the data set is necessary to remove all high-attenuating vascular or skeletal structures, or bowel loops containing positive contrast material that obscure the vessels of interest (Fig. 24.9). MIP are well suited to display simple vascular anatomy in which only few vessels overlap (e.g., see Fig. 2.12). For such applications the main disadvantage of MIP, namely the inability to differentiate between foreground and background, does not playa role. A 3D impression can be gained by reconstructing images from various projection angles that can then be displayed in a cine mode. This is the case for the retroperitoneal vessels (aorta and renal arteries). the mesenteric vessels (excluding the aorta by editing procedures, see Fig.24.74a), the thoracic aorta (excluding the pulmonary vessels), the carotid arteries (separate displays for each side, see Fig.24.69a), and the peripheral arteries (unless there are massive calcifications). They are not particularly well suited for displaying the whole abdominal vasculature, the pulmonaryarteries (unless targeted MIP are used), complex anomalies of the aortic arch, or vascular malformations. MIP suffer from an increase in background density when the projected volume is top wide. Overlying contrast-enhancing organs may superimpose over smaller vessels, and even image noise may hamper evaluation of small arteries. For this reason, thin-slab MIP (targeted MIP) of the vessel of interest often provide superior results (Fig. 24.8). In many cases, no prior editing of the skeletal structures will be necessary with this technique. Curved thin-slab MIP provide an excellent overview of branching vessels, for example in the renal hilum, and are thus superior to CPR for this purpose (Fig. 24.9). Calcifications are clearly delineated in MIPs, and this can be helpful in planning vascular interventions or anastomotic surgery in patients with severe atherosclerosis. On the other hand, calcifications superimposed on the arterial lumen can make it impossible to quantify the degree of stenosis. The original axial sections or CPR should be used for evaluating these cases (seeFig.24.82b) . Editing used to be a complex and time-consuming process that hampered the use of (nontargeted) MIP in clinical practice. Software is
845
846
24 Vascular System
a
b
Fig. 24.8 Oblique thin-slab MIPs parallel to the course of the renal arteries require no prior bone removal. The thickness in AP direction can be somewhat smaller
(10mm, a) than in the caudocranial direction (20mm, b).
a
b
Fig. 24.9 After editing the skeletal structures, there are still mesenteric vessels superprojecting over the kidneys on AP projections (a). Best display of renal vessels is possible with curved thin-slab MIP (b). The location of this thin slab is indicated on a caudocranial MIP that was tilted in a way to separate aorta and vertebral column (c)
Eva/uatf'on and Documentation
now available that does this semiautomatically. Many such programs require a seed point to be placed on the structures to be removed and sometimes also the structures of interest. The program then automatically removes unwanted structures with no or very little further interaction (see Fig. 2.43). Such programs are particularly helpful in the chest, abdomen, and for peripheral run-off vessels. •
Shaded Surface Displays
Shaded surface displays are best suited to demonstrate complex 3D-relationships, for example at the aortic arch or abdominal vasculature. They are helpful for preoperative planning, and may be used routinely in the evaluation of thoracic aneurysms (without a large thrombus), coarctations and other aortic arch anomalies, and in the follow-up of extra-anatomic vascular bypass surgery. SSD require a high level of vascular enhancement for good results. Generally there is no need for editing of the image data set prior to reconstruction of a SSD. Most processing tools, from simple cutting function to semiautomated techniques for bone removal, can be applied to the reconstructed 3D object. A useful rule of thumb for the optimum 3D imaging of in- plane and through-plane vessels is that the vascular diameter should be at least 1.5 times the slice collimation. The optimum threshold level depends on numerous factors but should be set at approximately 40-50% of the contrast between a vessel and its surroundings. Common artifacts mimic stenosis or vascular occlusion on SSD. This is due chiefly to partial-volume effects, which are especially pronounced in small vessels that course in plane. As the threshold for rendering is increased, the image depicts fewer vessels but displays an increasing number of pseudo-stenoses and occlusions. On the other hand, setting the threshold too low leads to the underestimation of stenoses (see Fig. 2.21) and produces band-like distortions of small vessels in the direction of the scan (Fig. 24.10). Calcifications can mask stenoses, because they appear to enlarge the vessel. Multiple thresholds can be used together with different colors to distinguish calcifications from true vessel lumen. However, such a technique is not very accurate in depicting the real vessel diameter because of blooming effects that make a calcification look larger and exag-
Fig. 24.10 Effect of threshold level on the shaded surface display (SSD) of a contrast-filled infusion line.
gerate stenoses. Image noise will cause "flying pixels" and is the main reason why SSD does not provide optimum results with thin-section multislice imaging. Volume rendering is the superior technique for this purpose. •
Volume Rendering
Volume rendering has become the standard rendering technique for many CTA indications, especially when multislice scanning is used. Compared to surface shaded displays it is less vulnerable to threshold effects and is easily able to display calcified plaques and lumen with different colors. It has the advantage ofM1P that it retains 3D information as well as density information but requires less editing. However, MIP may be superior when evaluation of more peripheral vessels is required. An appropriate color scale can be used to better differentiate between vessels of varying density (such as arteries and veins, see Fig. 2.31). A color scale may also be used to color code density ranges, which is particularly helpful for the pulmonary parenchyma or the kidneys (see Fig.2.38a). Editing with bone removal and removal of overlying structures may further improve the visualization of small detail (see Fig. 2.43) but is often not essential. Thin-slab rendering is an easy tool to free the data set
24 Vascular System from overlying structures editing. •
Virtual
without
much further
Angioscopy
Virtual angioscopy relies on the same principle as other virtual endoscopic techniques with the exception that the high-density areas (contrastenhanced vessel lumen) have to be rendered transparent. This is possible most easily with volume rendering and properly placed trapezoids for soft tissue (100% opacity below 100150 HU. 0% opacity above 150-200 HU). and for calcifications or metal stents (0% opacity up to maximum CT number in enhanced vessel. 100% opacity for CT numbers some 200-400 HU larger). Virtual angioscopy frequently suffers from noise-related vascular irregularities (see Fig. 24.60). and from difficulties in finding the correct opacity settings. especially in small vascular structures. Vessels that are in direct contact to the vessel of interest may blend and leave a connecting hole in the vascular wall. Virtual angioscopy rarely adds diagnostic information over other 2D or 3D-rendering techniques. In a few cases it may improve the assessment of
stent graft localization relative to branching arteries. Volume rendering with the opacity settings used for virtual angioscopy may often be superior for this purpose. especially when the vessel is cut open perpendicular or longitudinal to the course of the lumen. Documentation
Given the large number of sections generated by spiral scanning. it is impractical to document all images on film. With the increasing use of PACS systems. image documentation on hard copy loses importance. In such situations. only the pathologic findings may be recorded on film or even on paper prints. In a non-PACS situation. however. documentation on hard copy remains important. For single slice CT. it is advisable to document every second to fourth image (spaced 4-8 mm apart) on hard copy. For multislice CT, reconstruction of thick sections (5-10 mm) can be advocated. In any case. all relevant 3D images (MIP, SSDor VRT) together with CPRofthevessels of interest have to be printed because they provide the information pertinent to the further treatment of the patient.
Artifacts Artifacts may be caused by movement (pulsations. respiration). the selected scan parameters. or the rendering technique (MIP. SSD). An awareness of artifacts is important because they can simulate pathology. Pulsation
Artifacts
Artifacts occur when pulsations lead to changes in the diameter. configuration. or position of an opacified structure. The results are double contours. streak artifacts. jagged margins along the z-axis, and a spurious increase or decrease in CT attenuation. Pulsation artifacts can alter the diameter or the configuration of the heart. ascending aorta. and aortic arch. Double contours of the aorta can create the appearance of multiple lumina with different degrees of opacification (Fig.24.11a. b). Strong pulsations can seriously hamper the evaluation of the aortic root and aortic arch. Indeed. one should be cautious in
selecting young patients for CTA of the thoracic aorta, as the elasticity of that vessel can lead to extreme pulsation artifacts. Pulsation- induced position changes may be seen in the abdominal aorta, the renal arteries, and especially in renal graft arteries (Fig.24.11c). This artifact leads to saw-tooth irregularities in vessel margins. Pulsations in the craniocaudal direction can mimic stenosis in inplane vessels and when a high table feed is used. ECG-gating techniques may be used with multislice scanning to reduce pulsation effects. In addition, the faster table speed with multislice scanning, in particular 16-row scanners, will spread out artifacts over wider ranges along the z-axis. They will therefore appear less disturbing on 3D-rendered images (Fig. 24.12).
Artifacts
a
b
Fig. 24.11 Pulsations of the ascending aorta lead to characteristic double contours ofthe vessel wall (a) that can mimic a dissection (b). Positional changes during pulsation causes typical serrations, even on multislice
c CT data ofthe abdomen (c). Note the undulations ofthe abdominal wall due to shallow breathing. They have no additional influence on the display of the mesenteric vasculature.
The key factors in eliminating respiratory artifacts are optimum patient instruction and practice in effective breath holding. •
High-Contrast
Artifacts
Highly concentrated contrast material in the left brachiocephalic vein can cause significant streak artifacts that obscure adjacent structures. Streak artifacts caused by a high concentration of contrast material in the superior vena cava can mimic intimal flaps in the ascending aorta (Fig. 24.13), and they degrade the quality of 3Da
b
Fig. 24.12 Pulsation is spread out along the Z-axis with fast scanning using 16-row scanners (16 x 1.5/36) (a) and is thus less disturbing than with a 4-row scanner (4Xi/6)(b) .
•
Respiratory Artifacts
Artifacts due to respiratory motion during the spiral scan can simulate vascular discontinuities, stenoses, or aneurysms depending on the direction and amplitude of the motion. Respiratory artifacts are easily recognized by the appearance of steps or undulations in the contours of the skin surface or organs that move with respirations (see Fig.24.11c). Spurious vascular changes appear at precisely the same table position.
a
b
Fig. 24.13 High contrast artifacts in the superior vena cava can simulate dissection in the aorta (a). Such artifacts can be substantially reduced when a saline flash is used together with caudocranial scanning (b).
849
24 Vascular System rendered views. In examinations of the carotid arteries or thoracic vessels, an effort should be made to avoid these artifacts by using a saline flush and a scanning direction that captures the inj ected veins at the end of the scan. Artifacts due to dental implants will cause similar effects in the neck, particularly close to the region of the carotid bifurcation (see Fig. 24.69). Since the artifacts arise further ventral to the vessel they are frequently less disturbing than contrast-material induced artifacts. They can be reduced by placing the patient's head in such a way that the line of the teeth runs as closely parallel to the scan plane as possible. •
Flow-related
Artifacts
Differential opacification of vascular territories is due to differences in blood velocity, length of a vessel, vascular resistance (arterial phase) or variations in tissue transit times (venous phase). They may cause mixing of well-enhanced and less-enhanced blood that may simulate thrombi. This effect is most pronounced in regions of venous confluences (see Fig. 7.39) but can also be found in the pulmonary arteries (see Fig. 24.24 a, b). Layering of contrast material can occur in a vessels with a low blood velocity because contrast material is denser than blood (Fig. 24.14). •
Partial Volume
Effects
Small vessels that course in-plane are subject to marked partial-volume effects (see Fig. 2.21), especially in cases where a wide slice collimation has to be used. These effects cause a marked
decrease in the attenuation of small vessels on axial sections. As a consequence, small vessels that are poorly opacified may not be visualized on MIP or SSD. Small vessels that course through the imaging plane are clearly defined and show markedly higher contrast than inplane vessels. It should be noted, however, that partial-volume effects can cause segmental stenoses of these vessels to be missed. Partialvolume artifacts frequently affect imaging ofthe pulmonary, renal, and splanchnic arteries in single slice CTA examinations. With multislice CTA such artifacts are substantially reduced and occur only with fast scanning techniques that require thicker sections. •
Spiral Artifacts
The raw data interpolation process with spiral scanning has been shown to cause undulations of the vessel contour even in stationary vessels. In clinical practice such effects rarely affect image interpretation but they may become important with automated analysis of vessel diameters and stenosis quantification. Artifacts increase with larger pitch factors for both single and multislice scanning. •
Calcifications
and Vascular Stents
Metallic vascular stents and calcifications of the vessel wall have a substantially higher attenuation than the opacified vascular lumen. Due to
I
I
a a
b
Fig. 24.14 Contrast material is denser than blood and may for a fluid-fluid level in vessels with slow flow. (a) superficial femoral arteries; (b) abdominal aortic aneurysm.
b
Fig. 24.15 Metal stents will lead to "blooming" effects, which 'lead to underestimation of the true vessel lumen at standard CTA window settings (550/150) (a), wide wide window settings (1500/500) provide a more realistic view (b). Similar effects are seen with calcifications that are directly adjacent to the perfused vessel lumen.
Developmental
the limited spatial resolution of the CT system, both in-plane as well as along the z-axis, stents and calcifications will appear blurred and thus cause a halo of intermediate density CT numbers around them. This "blooming" effect will lead to an overestimation of stenoses if the calcification or stent borders directly to the vessel lumen and standard CTA window settings are used (Fig.24.15a). Due to partial volume effects on the vessel, this depends not only on the density of the stent material/calcified plaque but also on the spatial orientation of the vessel (parallel or perpendicular to the scan plane). Using a specifically adjusted narrow window setting is advocated for this purpose but this technique is cumbersome and requires readjustment for each calcified vessel segment. A simpler technique is visual analysis of the vessel using a very wide window setting that makes sure that the stent or plaque does not appear completely white (Fig. 24.15b, see also Fig. 7.30c. d). •
3D Rendering
Arterial Abnormalities
Table 24.7 Artifacts seen in eTA with various 2D and 3D imaging techniques
• Simulated stenoses with off-center cut-lines
A
MIP • • • •
Increasing background density with thick VOl Calcification superimpose vessel lumen No foreground-background differentiation Eccentric stenoses may appear normal
SSD • No differentiation between lumen and calcified plaque • Degree of stenosis threshold-dependent VRT • Noise may cloud pathology
As discussed above the various rendering techniques suffer from specific artifacts that are summarized in Table 24.7 and are discussed more extensively in Chapters 2 and 7.
Pulmonary Vessels
The main indication for pulmonary CTA is suspected acute pulmonary embolism (PE), for which it is superior to ventilation perfusion scintigraphy. With multislice scanning, even peripheral vessels (up to the sixth to eighth order) can be adequately evaluated. CTA is also an excellent tool for diagnosing developmental abnormalities, arteriovenous malformations, and for ruling out chronic thromboembolism as a cause of pulmonary hypertension. CTA has thus become the primary tool for the diagnostic work-up of the pulmonary vasculature. Developmental Arterial Abnormalities
Developmental pulmonary artery abnormalities in children should be evaluated with MRA whenever possible. If CTA is necessary, low dose techniques with reduced kVp (effective dose <1-2mSv) can be employed in most patients;
CTA may be preferred because of substantially faster scan times, which allow for less sedation in smaller or non-compliant children. In addition, excellent 3D displays of the whole pulmonary vasculature as well as the lungs, mediastinum, and chest wall become available. Congenital aneurysms of the pulmonary arteries are usually located in the central pulmonary artery system. They are associated with other pulmonary or cardiovascular abnormalities and in particular often result from blowout by systolic jets just distal to a stenotic pulmonary valve. Pulmonary artery coarctation is characterized by pulmonary artery stenoses in variable number, severity, and anatomical distribution. It is often associated with pulmonary atresia and congenital cardiac defects. It can be related to intrauterine rubella infection or may be associated with Ehlers- Danlos or Williams- Beuren syndromes.
851
24 VascularSystem Proximal interruption of one main pulmonary artery refers to the obliteration of a proxi-
mal segment of one pulmonary artery, which can reconstitute distally or remain occluded into the lung periphery. It should be differentiated from absence or aplasia of a pulmonary artery, in which no pulmonary artery can be found and which is accompanied by a mediastinal shift to the side of the hypogenetic lung. The pulmonary blood supply is maintained via bronchial or intercostal collaterals, an anomalous artery originating from the ascending aorta (right lung), or descending aorta (left lung), the subclavian or brachiocephalic artery, and, rarely, infradiaphragmatic collaterals. Left-sided interruption is associated with cardiovascular anomalies such as tetralogy of Fallot. Right -sided interruption is often associated with left-to-right shunt or pulmonary arterial hypertension. Adult patients within the second group may present with extensive collateralization and hemoptysis.
Anomalous origin of the leftpulmonary artery
from the right pulmonary artery can result in obstructive effects on the trachea and right main bronchus, a feature described as pulmonary sling. Early recognition of the abnormality is crucial as surgical correction is feasible. The peripheral lung vasculature may be affected by oligemia. Congenital systemic arterial supply to the lung is commonly found in bronchopulmonary sequestration. Information about the vascular supply is important prior to embolization or surgery. CTA is able to evaluate the pulmonary and the systemic arterial systems in one examination but evaluation of the venous return may require scanning in a second phase.
pulmonary artery straightforward.
aneurysms
is
normally
Pulmonary artery coarctation is character-
ized by short focal pulmonary artery strictures without wall thickening or perivascular abnormality of the mediastinal fatty tissue. Congenital cardiac abnormalities may be associated. In patients with proximal interruption of one main pulmonary artery the affected pulmonary artery segment either cannot be identified or appears as a narrow non-opacified band. The distal branches of the pulmonary arteries may be patent, when there is sufficient collateral vascularization, or may be obliterated by chronic thrombosis. The bronchial arteries are dilated. Collateralization by intercostal arteries is associated with web-like angiomatous pleural vascular malformations that can induce hemoptysis. The affected lung is commonly hypoplastic and can display small peripheral bullae (Fig. 24.16). In patients with pulmonary sling the atypical course of the left pulmonary artery posterior to the trachea is already evident on axial sections but can be best displayed by BD-rendering techniques (Fig. 24.17). In congenital systemic arterial supply to the lungs (pulmonary sequestration) the systemic vessel originates most often from the inferior thoracic or superior abdominal aorta and less frequently from supra-aortic, chest wall, pericardiophrenic, or abdominal aortic side branches (see Fig. 9.6). The aberrant artery runs to the abnormal lung segment. This abnormal
Acquired systemic arterial pulmonary supply
can be found in various conditions including aspergilloma, TBC or other inflammation, neoplasia, after trauma or surgery, and in chronic pulmonary arterial or venous obstruction. It is commonly associated with hemoptysis and bronchiectasis. CT angiography can be valuable in these patients to exclude bronchial artery aneurysms or a Rasmussen aneurysm in TBe. DSA and embolization is required for treatment of significant hemoptysis. •
CT Morphology
Congenital aneurysms typically present as a
fusiform dilatation of a central pulmonary artery (trunk, left or, less frequently, right main arteries). The differentiation from peripheral
Fig. 24.16 Proximal interruption of the pulmonary artery (2x2. 5/10) with complete obliteration of the segmental arteries but extensive collateralization by bronchial arteries (arrows).
Developmental
Venous Abnormalities
Fig. 24.17 Pulmonary sling syndrome. The left pulmonary artery passes behind the trachea to the contralateral side and causes a typical narrowing of trachea and esophagus (5/5/2, first generation spiral CT, 1992). (a) Axial section, (b) shaded surface display (SSD).
a
lung segment is ventilated in cases of intralobar sequestration, usually with venous drainage to the left atrium. It is collapsed in patients with extralobar sequestration, usually with drainage to a systemic vein. Developmental Venous Abnormalities
Anomalous unilateral single pulmonary veins (AUSPV) with normal connection to the left atrium are rare. Confusion with other conditions like pulmonary varices, or venous aneurysms, parenchymal pulmonary masses, partial anomalous pulmonary venous return (PAPVR) , the scimitar syndrome, or pulmonary arteriovenous malformations may occur on chest radiographs but these entities are easily distinguished by CTA. Isolated AUSPV does not require treatment.
Partial anomalous pulmonary venous return (PAPVR) and the scimitar syndrome refer to
anomalous connections of pulmonary veins to the systemic circulation that are associated with left -to-right shunting. Often an occult congenital pulmonary venous abnormality can be suspected already from a chest film. CTA as well as high quality MRA yields excellent morphologic results, while time-resolved or phase contrast MRA techniques provide additional functional information. Congenitalpulmonary vein stenosis (PVS) can occasionally be occult if only one pulmonary vein ostium is involved. Similar to the pulmonary veins in lung transplants it can be complicated by retrograde pulmonary vein thrombosis.
b
•
CT Morphology
Anomalous unilateral single pulmonary veins demonstrate a large single tortuous vessel draining the entire lung to the left atrium.
Partial anomalous pulmonary venous return
may be easily missed on routine evaluation of axial sections but is readily evident if one searches for characteristic patterns. On the right side, the anomalous vessels enter the superior vena cava or the right atrium directly. On the left, the anomalous vessels drain in to an aberrant venous structure that is usually located lateral to the aortic arch in a position similar to a left superior vena cava. This vessel drains into a dilated left brachiocephalic vein (see Fig. 9.7). In rare cases, the draining vein may have other positions such as medial to the aortic arch. The anomalies are best identified and classified on volume-rendered displays (Fig. 24.18). The remaining normal pulmonary veins can be assessed best on these displays when the descending aorta and the vertebral column are removed by editing procedures. Identification of a pulmonary vein stenosis requires thorough evaluation of the width of the veins in the vicinity of the pulmonary venous ostia into the left atrium. Substantial stenosis; may lead rarely to signs of venous congestion with increased size ofintrapulmonary veins and interstitial fluid retention. Iatrogenic pulmonary vein stenosis or occlusion may occur as a consequence of electro ablation treatment of atrial fibrillation (see Fig.23.54b).
853
24 Vascular System
a
.
b
Fig. 24.18 Partial anomalous pulmonary turn (PAPVR) in which the left pulmonary into an atypical mediastinal vein (arrows) through the aortopulmonary window into
Pulmonary
c venous revein drains that passes the left bra-
Thromboembolism
Acute Pulmonary Embolism
Acute pulmonary thromboembolism is a potentially life-threatening complication of peripheral deep venous thrombosis (DVT). A certain number of patients with massive pulmonary embolism do not receive any imaging because they die before the diagnosis is suspected or they are treated on suspicion and are too unstable for off-site diagnostic procedures necessitating transportation. Patients who are referred to imaging have to be in a sufficiently stable clinical condition. The role of imaging is to confirm or rule out the diagnosis of acute pulmonary embolism (PE), and, ifpossible, to establish an alternative diagnosis. This is important because up to 60% of patients have a false positive clinical suspicion ofPE. In addition, deep venous thrombosis cannot be ruled out clinically because less than one-third of patients with DVT of the lower extremity are symptomatic. Blood tests such as D-dimer have a high negative predictive value (although they do not completely rule out PE) but are rather unspecific, and positive results may be associated with a host of diseases that are associated with clotting activity. Imaging procedures, therefore, play a central role in therapeutic decision making about such patients.
chiocephalicvein (4 X 1 /8, 80 kVp, CTDlvoi < 3 mGy). (a) Axial sections, (b) coronal MPR (c) volume-rendered displays (VRT) after removal of the descending aorta.
Chest radiography is normal in many patients with pulmonary embolism or may just demonstrate mild effusion and plate-like atelectasis. Pulmonary infarcts are rare and occur in only 10-15% of cases. Direct signs of PE are rare and not very reliable. Chest radiography, therefore, mainly plays a role in establishing alternative diagnoses, such as pneumonia or pneumothorax, and in deciding upon which further diagnostic tests to perform. Transthoracic or transesophageal echocardiography is used in suspected massive PE (unstable patients) for the detection of acute right heart overload. They are not used in stable patients because they lack the sensitivity to rule out peripheral emboli. Pulmonary angiography is considered the radiological gold standard for demonstrating acute PE. Because of its invasive nature, however, it is rarely requested in most institutions' even for those patients with indeterminate imaging findings. In addition, animal studies suggest that the sensitivity and specificity of pulmonary angiography is similar to high-quality CTA for detecting the presence of pulmonary emboli. Using pulmonary angiography as a gold standard in clinical studies, therefore, may underestimate the true accuracy of other techniques such as CTA. Ventilation-perfusion scintigraphy (V/Q scanning) used to be the first -line test in stable patients. V/Q. scanning provides probability esti-
Pulmonary Thromboembolism mates for .the presence of acute PE. Large studies in an unselected population demonstrated that the number of high-probability positive or negative findings is quite low (together in the 30-35% range), Even modern scintigraphic techniques could improve these numbers only slightly, As a consequence, most clinicians would treat according to their clinical judgment and not based on imaging findings, Far better results are obtained in patients who were previously healthy and without signs of cardiopulmonary disease on chest radiographs, Highprobability results are achieved in over 80 "/0 of such patients, Because most of these ventilation scans are normal, perfusion (Q,) scanning alone suffices and provides a good diagnostic option in this patient group, Conversely, the number of high-probability results in patients with known cardiopulmonary disease or pathologic abnormalities on chest radiographs is extremely low, This is the group that profits most from alternative imaging techniques such as CTA Gadolinium-enhanced MR angiography has been suggested as a test for acute PE. It offers direct visualization of emboli without invasiveness or ionizing radiation, The accuracy was reported to be high but large patient studies are not yet available, Potential drawbacks are the limited access to the (potentially unstable) patient, the higher susceptibility to breathing artifacts, an examination volume that usually does not cover the whole chest, and insufficient display of the surrounding parenchyma for establishing potential alternative diagnoses, CT angiography is presently the workhorse for the evaluation of stable patients with suspected PE. It is the first-line modality in patients with an abnormal chest radiograph or history of cardiopulmonary disease, and can be used as an alternative to perfusion scanning for all other patients, Interobserver agreement is substantially higher than for V/Q, scanning or pulmonary angiography, The number of indeterminate diagnoses is below 10%, Specificity is well above 80% for single slice CTA, while sensitivity varies depending on the study design and the patient group examined (Table 24,8), False negative findings may be due to breathing artifacts and peripheral emboli, which are beyond the resolution of single slice scanners, Isolated peripheral emboli are rare in a general patient group (3-7%) but are more frequent in patients with indeterminate ventilation- perfusion scans
Table 24.8 Accuracy of single-slice the diagnosis of acute PE
Author
Sensitivity
Remy-Jardin.
1993
van
1993
Rossurn.
Mayo,1997
van
CT angiography
199)
Rossurn.
Specifkity
75
91%
75m
149
94%
82C
142
87%
95%
123
75%
~H !"
111941
ESTIPEP,
1998
401
88%
Baqhaie.
1998
370
96%
100% 96%
Kirn,1999 ANTELOPE, Qanadli.2000
1999
110
92%
617
69%
(157
90%
in
94o/c1
(30%), Their clinical importance is a matter of controversy, Multislice CTA with 1 mm sections substantially improves detection rates of subsegmental emboli and evaluation of more distal pulmonary arteries (see Fig, 24,20), It yields higher interobserver agreement than with thicker sections, A One of the biggest advantages of CTA, however, is that it can establish an alternative diagnosis in up to 30% of patients with suspected PE. It allows for direct visualization of the thrombotic material as well as of signs of right heart overload, It has the potential of distinguishing massive from non-massive PE. Technical factors, however, are criticaL Breathing artifacts should be reduced as much as possible, and it is advisable to have dyspneic patient breathe shallowly instead of attempting a breath hold maneuver. Best results, however, rely on a breath hold examination, Breath hold duration can be increased by proper patient instruction and supplemental oxygen, Multislice scanning with 4 x 2-2,5 mm or 16 x 0,75L25mm collimation can reduce scan time to below 5 s if only the central pulmonary arteries from 2cm below the pulmonary veins to the aortic arch are examined, Up to now, no significant change in diagnostic accuracy has been reported when increasing the scan range to include the whole chest Keeping the scan range small and reducing the tube voltage to 100 kVp (in slim patients even SOkVp) can substantially lower radiation dose to well below 2 mxv, Using such scan protocols, CTA in pregnant women has a substantially lower uterine radiation exposure than V IQ, scanning, Lower kVp settings also
855
856
24 Vascular System
a
b
Fig. 24.19 Acute pulmonary embolism, (a) Large riding embolus in the pulmonary artery bifurcation, (b) Embolus riding at the bifurcation of the right in the pulmonaryartery (801
improve intravascular enhancement especially of small arteries that suffer most from partial volume effects. Together with a narrow collimation of 3 mm for 1 s single slice scanners, 2-3 mm for 0.75s scanners, and 1-2 mm for dual and multislice scanners, the number of subsegmental arteries that can be evaluated grows substantially and should further improve the diagnostic accuracy of CTA. CTA also offers a "one-stop shop" approach for evaluation of thromboembolic disease by adding a venous phase scan of the abdomen and lower extremities (CT venography). The scan is started some 3 minutes after the end of the pulmonary CTA using the already injected contrast material. This technique can detect peripheral venous thrombi at least as well as ultrasound. Low-dose scanning should be mandatory and may be achieved by thick sections and low CTDIvoi settings. Discontinuous scanning, e.g., with 4-5 mm sections every 2cm is also acceptable, e.g., using 4x1-1.25 mm collimation merged to 4-5 mm thick sections. The clinical value, however, is not yet firmly established because only few patients with suspected PE (3% in a large study) present with an isolated deep venous thrombosis without associated PE. Thus substantial numbers of patients have to be exposed to radiation to detect those individuals in whom clinical consequences arise. For this indication, duplex ultrasound should probably be preferred.
c
d
Segmental emboli in the left lower lobe arteries (3/5/2). (d) Peripheral embolus with distal vessel occlusion on a thin-slab MIP (1/3/1).
•
CT Morphology
Direct signs of pulmonary embolism are intraluminal filling defects and lack of enhancement of a pulmonary artery. Acute emboli get trapped either at pulmonary artery bifurcations [riding emboli} or in peripheral arteries that are smaller than the embolus. Such riding emboli may not only occur at the pulmonary artery bifurcation (Fig. 24.19a) but also at the segmental
a
b
Fig. 24.20 Peripheral emboli on a low dose multislice CTA (80 kVp, CTDlvoi < 3 mGy, 4x1/6) reconstructed with 1.25 mm section width (a) and 3 mm section width (b). The embolus rides on a subsegmental bifurcation and is more easily distinguished from partial volume artifact on the thin sections.
Pulmonary Thromboembolism Fig. 24.21 Older clots may be suspected when they are waIl-adherent and show signs of resorption (here: three weeks old, 4 x "II 6) (a). Mosaic perfusion can be appreciated on color-coded volumerendered displays (b). Note the darker colors indicating reduced CT attenuation and thus, less perfusion.
b
a
or subsegmentallevel (Fig.24.19b). Long emboli from peripheral veins may involve more than one artery, may coil up in a larger vessel, or they may become fragmented and lead to a shower of small emboli in multiple lung segments (Fig. 24.19c). The diameter of the embolus, and thus the level of vascular obstruction, corresponds to its site of origin. More peripheral emboli tend to partially or completely occlude the affected vessel. Such emboli can often be better evaluated with thinslab MIP because a longer portion of a peripheral vessel is displayed so that the boundary of contrast enhancement is more easily appreciated (Fig.24.19d). The effect requires thin-section imaging and high intravascular contrast, which is more easily achieved with multislice techniques (Fig. 24.20). Complete occlusion of a vessel by fresh emboli is possible but residual perfusion in the periphery is more usual. Secondary thrombosis of this residual lumen may occur, leading to a moderate enlargement of the vessels with secondary thrombosis of a distal portion of the artery. In such a case, mosaic perfusion with reduced attenuation in the affected segment will occur. Mosaic perfusion may also be secondary to high-grade obstruction because of vasoconstriction (Euler- Liliestrand effect) of the affected pulmonary artery segment (Fig.24.21b; see also Fig. 24.28). The mosaic pattern is more easily appreciated in the pulmonary parenchymal phase some 30s after the start of contrast injection and 5-15 s after opacification of the pulmonary arteries. Color-coding can improve detection of such abnormalities (Fig. 2.21 b). In a matter of days the clot usually adheres to the vascular wall (Fig.24.21a). Early segmental arterial vasoconstriction distal to the site
of embolism can persist to the chronic phase as segmental or sub segmental vascular asymmetry (see Fig. 24.28). Normal bronchial arteries take over the blood supply to the embolized lung areas immediately. Pulmonary infarcts, which can complicate various pulmonary vascular disorders including PE, can occur after hours if the collateral supply by bronchial arteries is insufficient (Fig. 24.22). The CT morphology reflects the appearance on the conventional radiograph: ground-glass opacity can precede the typically wedge-shaped segmental consolidation. Larger infarcts may contain varying amounts of air. Regions within nonaerated lung that are suspicious for infarction display decreased contrast enhancement. They occur in up to 15% of acute PE cases. Cavitation is frequently observed in
a
b
Fig. 24.22 Pulmonary infarct with a wedge-shaped, non-enhancing soft tissue mass with small irregular air inclusions. Axial section (a), coronal MPR (80 kVp, CTDlvoi< 3mGy, 4x-1/6) (b).
857
24 Vascular System
septic infarcts, rarely in bland infarcts. Occasionally, pulmonary infarcts are associated with pneumothorax. •
Artifacts
and Pitfalls
The commonest factors leading to false positive and false negative results of CTA are summarized in Table 24.9. Breathing is the single most important factor that can render a CTA of the pulmonary arteries useless. Breathing can abruptly decrease vascular contrast and should not be mistaken for intraluminal filling defects (Fig.24.23a). Breathing artifacts may be hard to identify with a soft tissue window setting but they are readily Table 24.9 Pitfalls in eTA for acute pulmonary embolism False positive findings I • Breathing artifacts I • Pulsation artifacts S • Partial volume effects j • Differential enhancement of arterial territories j • Hilar lymph nodes ; • Unenhanced pulmonary veins JFalse negative findings • • • • •
Partial volume effects Breathing artifacts Pulsation artifacts High-contrast artifacts Poor vascular enhancement,
frig. 24.23 Breathing may s~ulate endoluminal clots (a), but typical artifacts can be identified on lung window settings (b). Partial volume effects may simulate
apparent as distortion of anatomic structures with a lung window setting (Fig.24.23b). Breathing may decrease the density of obliquely oriented vessels to an extent that makes them impossible to evaluate. Pulsation predominantly affects the lung parenchyma of the lingula and left lower lobe. It aggravates partial volume effects in obliquely oriented vessels. Pulsation appears as double contours, decreased enhancement simulating thrombus, or vessel beading. False negative findings may be the result of inconsistent visualization of smaller arteries in the affected segments. Partial volume effects make the apparent enhancement of a vessel decrease as the vessel becomes smaller in the periphery. Vessels with a more horizontal course are more susceptible to these effects. Mistaking such partial volume artifacts for a real embolus will lead to false positive findings (Fig.24.23c, d): However, comparing the density of a vessel to its more distal portions or to a vessel of similar size and orientation toward the scan plane will allow a more confident diagnosis to be made. Partial volume effects are also responsible for lack of visualization of more peripheral artery branches and may cause false negative diagnosis. Too thick a collimation will substantially impair the diagnostic accuracy of CTA. For good quality MPR, the collimation should not exceed 3mm.
cthrombdtic material (c) and r@quireclose scrutiny ofthe adjacent sections for differentiation (d).
Pulmonary Thromboembolism
a
b
c
d
Fig. 24.24 Differential enhancement between various portions of the pulmonary vascular system (inflow effects) may be seen during the early phases of pulmonary arterial enhancement (4 x 1/6) (a). Such inflow effects may simulate intraluminal thrombi (b). An increase in pressure within the chest cavity may cause inflow obstruction of contrast material and thus, insufficient
enhancement. Inthis patient (c. d), an unexpirienced operator performed the scan during expiration in a caudocranial direction. The initially good enhancement dropped as the patient tried to hold the breath and performed a Valsalva maneuver. Note that contrast injection had continued throughout the lower ha If of the
Poor vascular enhancement aggravates the problems with other artifacts, in particular partial volume and breathing effects, and may result in nondiagnostic examinations. The main reasons are a low injection rate, wrong bolus timing, or too Iowa contrast volume. Vascular enhancement can be substantially improved by using low kVp settings, high flow rates (> 3 mll s), and a saline flush to optimize utilization of contrast material (see Fig. 24.5). Differential enhancement of the arterial lumen between various segments of the pulmonary arterial tree can be seen even at identical table positions. It occurs when there is increased flow due to peripheral abnormalities, or decreased outflow, and may even be seen sporadically without apparent reason. It is most frequently encountered in early scans, especially when (fast) multislice scanning is employed (Fig.24.24a, b). Usually, however, the density of the affected vessel is higher than that of soft tissue, making it easy to distinguish it from a true embolus. High-contrast artifacts from the injection veins may make evaluation of arteries in or close to the mediastinum difficult. Artifacts can be reduced with caudocranial scanning direction and the use of a saline flush. Contrast delivery through catheters that are positioned in the pulmonary artery (Swan-Gflnz catheters)
will cause major artifacts and will leave one lung without sufficient contrast enhancement. Anatomic pitfalls include lymph nodes (Fig.24.25a) that are seen in typical locations at the origin of upper lobe vessels, alongside lower
a
b
Fig. 24.25 Anatomic pitfalls, (a) Hilar lymph nodes, especially close to the right upper lobe vein, may simulate central emboli but can be differentiated by their typical location (arrows) and, in case of doubt by MPR. Note the embolus on the left occluding the artery to segment 6. (b) Non-opacified pulmonary veins may simulate non-enhanced arteries and thus embolus. They can be identified by their typical anatomical course.
859
24 Vascular System lobe arteries, and at the origin of branching segmental arteries and can be mistaken for central emboli. If in doubt, coronal or sagittal MPR will help. Isolated wall-adherent thrombi at these positions are rare in acute PE. With caudocranial scanning, the enhancement of the pulmonary veins can be lower than in the pulmonary arteries and may be mistaken for intraluminal clots (Fig.24.25b). The veins can be differentiated from the arteries using their anatomic position (medial to the lower lobe arteries), their continuity to the left atrium, and their lack of proximity to the bronchi. Chronic Thromboembolic Hypertension
Pulmonary
Within six weeks to several months most emboli will resolve (60-65%) or recanalize (2530%). Some 15% of patients with incomplete resolution of thrombus after 10 months will develop secondary chronic thromboembolic pulmonary arterial hypertension (CTEPH). CTEPH develops if a sufficiently large crosssectional area of the pulmonary arteries is occluded or suffers from reduced perfusion due to vasoconstriction. The clinical diagnosis of chronic thromboembolic pulmonary hypertension is difficult owing to nonspecific symptoms and lung function tests. Progressive occlusion of the pulmonary vascular bed will cause pulmonary hypertension. Often the patients present with pulmonary hypertension after progressive occlusion of the pulmonary vascular bed caused by medial hypertrophy, intimal fibrosis, and arterialization
of pulmonary arterioles. Most patients have a substantial right ventricular hypertrophy and may even suffer from secondary tricuspid incompetence due to dilatation of the ventricle and the valvular ring. Causative treatment is possible if central thrombotic material in the pulmonary arteries can be removed by pulmonary arterial thromboendarterectomy, CTA plays a crucial role in differentiating CTEPH from other causes of pulmonary hypertension and in defining the presence of wall-adherent material in the proximal pulmonary arteries. Single slice CTA is always complemented by pulmonary arteriography but multislice CTA has the potential to become the sale diagnostic modality for this patient group. •
CT Morphology
The presence of intraluminal thrombotic material is virtually diagnostic for CTEPH. The central pulmonary vessels enlarge, while more peripheral recanalized vessels may have a reduced size with eccentric, smoothly marginated residual thrombus adhering to the vessel wall (Fig.24.26a). Thrombotic material can display dystrophic calcification (Fig.24.26b). Peripheral stenoses may form during recanalization ofvessels. They are best appreciated on volume-rendered views or thin-slab MIP (Fig.24.27a, b). Intraluminal webs can be found as a residuum of emboli (Fig.24.27 c). Chronically occluded vessels have a substantially reduced size and are associated with mosaic oligemia. Mosaic oligemia (mosaic perfusion) is a typical feature of chronic PE and is characterized by areas of reduced lung attenuation and re-
Fig. 24.26 Chronic thromboembolic pulmonary hypertension (CTEPH). (a) Post-thrombotic material at the vessel wall is pathognomonic (80 kVp, CTDlyoi <3mGy.4x 1/6). (b) Large thrombi may eventually calcify (wide arrow). Note the recurrent embolus in the left lower lobe artery. a
b
Pulmonary Thromboembolism
a
c
b
Fig.24.27 CTEPH (4x1/6). Peripheral stenoses are best appreciated on volume-rendered images (a) or on thin slab MIP (b). Small thread-like webs may remain in
the pulmonary arteries after resorption ofthe rest ofthe thrombus (c).
duced vessel diameter due to hypoperfusion and the vasoconstriction effect (Fig. 24.28). Noninvolved areas appear denser than hypoperfused areas and may be mistaken for areas of pathologic ground-glass opacity. Mosaic oligemia is virtually diagnostic of chronic pulmonary embolism, even without additional demonstration of intra -arterial thrombus. Lung areas with normal perfusion pattern do not exclude the presence of proximal intravascular embolic material in the supplying artery. The pattern is best appreciated on color-coded images (Fig.24.28b). Scars or nodular masses may be seen as a consequence of old pulmonary infarction. Such old infarcts may present as a cavitating nodule
(Fig.24.28c). Plexiform lesions (see Fig.24.34b) are rarely seen in CTEPH. Bronchial arteries frequently hypertrophy in chronic PE and can often be seen running alongside the occluded artery, especially when thin-section scanning is performed (Fig. 24.29). However, bronchial artery aneurysms and extensive chest wall or infradiaphragmatic collateralization, as in pulmonary artery interruption, have not been reported in PE.
a
b
Fig. 24.28 CTEPH. (a) Mosaic perfusion is associated with thin vessels in hypoperfused and thick vessels in hyperperfused areas, (b) Color-coded maps of the pulmo-
•
Differential Diagnosis
Wall-adherent apposition thrombus may form in the central pulmonary arteries of extreme cases of primary pulmonary hypertension or Eisenmenger's syndrome because of slow or tur-
c nary parenchyma best demonstrate the docal differences in perfusion (4 x 1 /6). (c) Old infarcts may convert to scar tissue or become cavitating.
861
862
24 Vascular System Fig. 24.29 Bronchial arteries are typically dilated in chronic pulmonary hypertension (4 x 1/6) (a). This can be also appreciated on volume-rendered displays (b).
a
.b
elusion occurs when platelets are damaged by whipping of blood with larger amounts of air. Talcum, cellulose, and srarch embolism can be seen in chronic drug abusers after venous injection of tablet drug forms that contain such materials as filler components or with the use of cotton as filter material. Embolization to pulmonary arterioles and capillaries causes obstruction and often thrombosis with transient pulmonary hypertension, chronic pulmonary hyPulmonary Emboli of Other Etiology pertension, and even sudden death. Pulmonary mercury embolism can occur acPulmonary artery tumor emboli are encountered cidentally from venous sampling with mercuryat autopsy in up to 30% of patients with undersealed syringes or in patients attempting suicide lying malignancies such as renal cell carcinoma, with venous injection. The diagnosis is generally hepatocellular carcinoma, choriocarcinoma, breast, gastric and prostate carcinoma, and made on plain film radiography and does not remalignant melanoma. Embolism occurs pre- quire CT. Coifs or other foreign bodies occasionally dominantly into small or medium-sized peripheral arteries, sometimes inducing pulmonary in- embolize into the pulmonary circulation as a refarction. Tumor embolism to central vessels is sult of attempted treatment of peripheral arteriassociated with a higher mortality rate. Al- ovenous malformations or systemic venous intervention. CT angiography is used to localize though the presence of lymphangitis carcinomatosa is common, tumor embolism may be the the coil in the chest. only evidence of metastatic disease. Septic emboli are discussed in Chapter 9, • CT Morphology CTA in massive tumor embolism demonstrates p.318. CT is highly sensitive to the presence of irregularly dilated or beaded arteries with vascular occlusion (Fig.24.30a) and sometimes small amounts of intravascular air in brachiocephalic vein, heart, or pulmonary artery. Most evidence of pulmonary infarction. frequently it is secondary to accidental injection Intravascular mercury presents as a stark hyperattenuating material, which may remain of minute quantities of air during intravenous at the bottom of the right ventricle or may be injections of fluids or even from merely placing found in small caudally located peripheral pulan IV cannula. It may also be found as a complication of surgery, delivery with placenta previa, monary artery branches (Fig.24.30b). Intravascular air appears as a strongly hyand some gynecological examination procedures. The (fatal) fibrinous terminal vessel oc- poattenuating small bubble and most frequently bulent flow in these arteries. In such cases, however, there is massive enlargement of the central pulmonary arteries (see Fig.34.a). Moderate mosaic perfusion may also be seen in other causes of pulmonary hypertension. It usually is less pronounced than in CTEPH (Fig. 24.28).
Pulmonary Artery Thrombosis Fig. 24.30
Tumor emboli rarely
may demonstrate intravascular growth with polycyclic enlargement of peripheral pulmonary artery branches (a). Note the normal size of the bronchi (arrowheads). Concomitant Iymphangitic carcinomatosis may be present. After suicidal IV injection of mercury, extremely hyperattenuating material is seen in the periphery of the pulmonary arteries (b).
a
a
b
Fig. 24.31 Small amounts of air in the pulmonary outflow tract (arrow) or left brachiocephalic vein are quite common after intravenous injection (a). More than 1 ml of air form a fluid level, here in the left brachiocephalic vein and the superior vena cava. This finding rarely results in symptoms unless there is a R-L shunt (b).
can be found at the highest (most anterior) portion of the main pulmonary artery, or the left brachiocephalic vein (Fig.24.31a). Formation of an air-blood interface can occur from air volumes of more than 1 ml (Fig. 24.31b). Pulmonary Artery Thrombosis
Pulmonary artery thrombosis is the result of a complete obstruction of a central or more peripheral pulmonary artery branch. It can occur as a complication of congenital pulmonary artery interruption or obstructing processes of the pulmonary artery wall, such as Takayasu arteritis or pulmonary artery sarcoma. Most
frequently it is the consequence of an obstructing pulmonary embolus. Massive parasitic embolism is a rare cause of pulmonary artery thrombosis. Direct infiltration, obstruction, and consequently the thrombosis of the distal portions of a pulmonary artery, may be caused by bronchogenic carcinoma although obstruction of pulmonary veins is more common. Pulmonary artery thrombosis without arterial obstruction or classical pulmonary embolism is a rare complication of various disorders such as thrombophilia (e.g., autoimmune or due to SLE), nephrotic syndrome, or pneumonia, especially TBe. The prognosis of acute main pulmonary artery thrombosis is poor and frequently results in sudden death. •
CT Morphology
CTA demonstrates a lack of enhancement of the affected pulmonary artery segments and resulting narrowing of the affected vessels in the chronic stage. Depending on the amount of collateral circulation by bronchial arteries, peripheral vessels may be patent or completely occluded. In acute PE there is lack of enhancement even of the distal vessels in an affected segment. While reduction of diameter is readily appreciated on lung window settings in the chronic stage, demonstration of acute thrombosis relies on thin-section scanning with multislice CT.CTA is able to demonstrate a bronchogenic carcinoma (see Fig.9.16c), pulmonary sarcoma (see Fig. 24.40), Takayasu arteritis (see Fig. 24.39), or congenital pulmonary artery interruption (see Fig. 24.16) as the reason of pulmonary artery thrombosis.
863
24 Vascular System Pulmonary Vein Thrombosis
Pulmonary vein thrombosis is caused by obstruction of a pulmonary vein due to pulmonary vein stenosis, stenosis of the anastomosis after surgery, or due to tumor invasion or compression. In lung transplant patients it can be associated with subacute or acute graft rejection. •
CT Morphology
There is a lack of enhancement of affected segments (Fig. 24.32), often associated with signs of pulmonary venous congestion and edema. In the acute setting, there may be interstitial fluid or even alveolar edema with centrilobular opacities or even consolidation. In the chronic setting, collaterals may develop (see Fig.23.54c) . Pulmonary Hypertension
Pulmonary hypertension can be the result of a large number of disease entities (Table 24.10). CTA is particularly important for identifying individuals with recurrent pulmonary embolism and CTEPH, who can be successfully treated by thromboendarterectomy. CTA and HRCT can be used to differentiate between the various disorders that cause pulmonary hypertension. The diagnosis of primary pulmonary hypertension (PPH) is made by excluding any clinical or radiological alternative etiologies.
Fig. 24.32 Pulmonary vein thrombosis following attempted correction of an aberrant pulmonary venous return from the right upper lobe (arrow) in the same patient as in Fig. 9.7. Correction ofthe aberrant left pulmonary veins was successful.
Table 24.10 Findings that help in differential diagnosis of patients with chronic pulmonary arterial hypertension Wall-adherent material Chronic thromboembolic pulmonary tensions Severe PPH (apposition thrombus) Pulmonary artery sarcoma Intraluminal calcifications Chronic thromboembolic pulmonary hyper tension Very longstanding severe pulmonary hypertension Eisenmenger's syndrome Idiopathic infantile arterial calcification (IIAC) Dilated central and peripheral vessels Increased pulmonary blood flow with L-R shunt
J
Dilated pulmonary veins, intestinal or alveolar edema Mitral stenosis Mediastinal fibrosis or tumor mass Pulmonary vein stenosis Pulmonary venoocclusive disease Peripheral pulmonary AV shunting Hepatopulmonary syndrome Chronic pulmonary schistosomiasis Diffuse or focal decreased lung attenuation" Chronic obstructive lung disease Pulmonary emphysema Cavitating lesions in granulomatous infections" Vasculitis , II Signs of interstitial lung disease or granulor tous infections Interstitial fibrosis Sarcoidosis Granulomatous infections Pneumoconiosis Vasculitis Alveolar capillary dysplasia Pulmonary capillary hemangiomatosis Veno-occlusive disease
Plexogenic arteriopathy is a vascular hyperplasia that affects small to medium-sized pulmonary arteries. It is one of several responses of the pulmonary arteries to continued stress from chronic severe pulmonary arterial hypertension. The vascular lesions have no significant connections to pulmonary veins that would allow for arteriovenous shunting. Plexogenic arteriopathy can occasionally be detected on CT.
Pulmonary Hypertension 865 •
pertension. The turbulent flow in these arteries
CT Morphology
Regardless of the presence and nature of an underlying pathology, the basic radiological pattern of chronic pulmonary arterial hypertension includes central pulmonary arterial dilatation, tapering of peripheral pulmonary arteries, and right heart enlargement (Fig. 24.33). The correlation of the pulmonary artery diameter on CTA to the degree of pulmonary hypertension is nonlinear. A main pulmonary artery diameter of s 29 mm at its widest point has a positive predictive value above 95%, and a distal pulmonary artery trunk width exceeding the diameter of the ascending aorta has a positive predictive value above 90%. The distal vessels may be large, normal, or reduced in caliber. Massive enlargement of the central pulmonary arteries is found in primary pulmonary hy-
may lead to formation of apposition thrombus that can be hard to differentiate from CTEPH (Fig.24.34a). Mosaic perfusion, however, is usually much less in PPH than with CTEPH. In patients with secondary pulmonary hypertension a variety of CTA and HRCT features can provide information about primary disorders and facilitate their differential diagnosis. Table 24.10 lists the CT-morphological criteria that aid in the exclusion of PPH and the differential diagnosis of disorders with secondary pulmonary hypertension. Plexogenic arteriopathy is characterized by dilated small tortuous peripheral arteries without evidence of a large AV shunt as in pulmonary 'arteriovenous malformations (Fig.24.34b) .
Fig. 24.33 Pulmonary hypertension. (a) Massive dilatation and thickening together with trabecularization of the right ventricle, (b) Marked enlargement of the central pulmonary artery and a sudden drop in vessel caliber to the periphery, best appreciated on thinslab MIP.
a
b
a
b
Fig. 24.34 Primary pulmonary hypertension may lead to massive dilatation of the central pulmonary arteries with formation of apposition thrombus (4 x 1 /6) (a). In the periphery of the pulmonary artery system, small tortuous vessels may become visible which represent plexogenic arteriopathy 0/3/1) (b). .
24 Vascular System morphology and flow characteristics prior to coil embolization of larger communications. Contrast -enhanced echocardiography or microsphere nuclear imaging can provide information about the shunt volume during rest and exercise. The role ofCTA is not yet defined. The hepatopulmonary syndrome is characterized by pulmonary hypoxemia in the presence of advanced liver disease and is due to arteriovenous microshunts that resemble spider angiomas. The patients present with gradual onset of hypoxemia and dyspnea that is commonly more severe in upright position.
Arteriovenous Shunting
Pulmonary arteriovenous malformations (pAVM) are commonly associated with Osler-WeberRendu disease. Many pulmonary AVM are discovered incidentally in adulthood. Patients are frequently asymptomatic and present with a suspicious nodule or abnormal vessel on chest radiographs. Untreated or occult pulmonary AVM can increase in size over time, with increasing shunt volumes inducing hypoxemia, together with an increased risk of pulmonary hemorrhage, stroke, or AVM -rupture. CTA has an important role in screening for pulmonary AVM with a sensitivity above 95%. CTA provides 3D rendering of complex malformations and thus facilitates planning of transarterial embolization. CTA is especially helpful in malformations with more than one feeding vessel. A diameter of the feeding artery of 3 mm or more is regarded as an indication for embolization. Diffuse pulmonary arteriovenous shunting can occur in pregnancy, Osler-Weber-Rendu disease, complex cardiac malformations, proximal pulmonary artery interruption, polysplenia syndrome, liver disease, and in chronic schistosomiasis. The patients present with severe hypoxemia and commonly have a history of systemic embolism with neurological complications. Spontaneous regression after pregnancy has been reported. DSA is required to assess the
CT Morphology
Spiral scanning without the use of intravenous contrast material may already be able to demonstrate the pulmonary malformation, which consists of a, frequently tortuous, feeding artery and an enlarged draining vein (Fig.24.35a, see also Fig. 9.38). Aneurysms are common in the region of the actual AV shunt. Pulmonary AVM enhance simultaneously with the pulmonary arteries. Complex AVM have more than one feeding artery or draining vein and are best evaluated by SSD or VRT. Multiple AVM are a common finding. In order not to miss these small lesions, volumerendered images (after removal of the chest wall) are particularly helpful (Fig.24.35a). In addition, the cine viewing of axial or multiplanar
b
a Fig. 24.35
•
Pulmonary
arteriovenous
malformations
(arrows) in a patient with Osier's disease. Some AVMs were embolized (arrowheads). VRT after removal ofthe posterior chest wall (4 x 1/6) (a). Macroscopic diffuse arteriovenous shunting with dilated veins in a young
previous healthy pregnant woman (ultra low dose CT,
80kVp, CTDlvoi=O.6mGy, 4x1 /8, thin-slab MIP) (b). Hepatopulmonary syndrome with multiple tiny peripheral shunts in a patient with long-standing liver disease (2/4/2, thin-slab MIP) (c).
Pulmonary Aneurysms sections demonstrates disproportionally large vessels in the lung periphery, or vessels with an increased tortuosity. Demonstration of an AV shunt in these regions confirms the diagnosis of AVM. Shunts between the bronchial arteries and pulmonary veins are extremely rare causes of pulmonary hemorrhage. Such findings are very hard to demonstrate but may be approached by thin-section multislice scanning and volume rendering. Bleeding typically presents as areas of patchy ground -glass opacities in patients with hemoptysis, or even consolidation in patients with massive hemorrhage.
Macroscopic diffuse arteriovenous shunting
appears as massively dilated veins, sometimes associated with conglomerate areas of small web-like dilated arteries including the lung periphery (Fig. 24.35 b). CT A in hepatopulmonary syndrome demonstrates arcade-like dilated subpleural vascular branches, predominantly within the lower lobes (Fig.24.35c). A negative CTA and HRCT do not exclude the presence of diffuse AV shunting or hepatopulmonary syndrome. Pulmonary Aneurysms
Congenital aneurysms are typically located in the central pulmonary artery system. They may be induced by systolic jets from a stenotic pulmonary valve. Other pulmonary or cardiovascular abnormalities may be present. Infection is an important pathogenetic factor in (peripheral) pulmonary artery pseudoaneurysm formation, and occurs with pulmonary aspergillosis, and rarely with tuberculosis (Rasmussen aneurysm) or syphilis. Once demonstrated on CTA, early treatment is indicated because rupture is usually fatal.
Bronchial artery aneurysms can be id-
iopathic or occur in association with pulmonary artery aplasia, Ehlers-Danlos syndrome, trauma, infection, silicosis, and vasculitis including Behcet's disease, Hughes-Stovin syndrome, and polyarteritis nodosa.
Iatrogenic pulmonary artery pseudoaneurysms, commonly induced by Swan-Ganz
catheters, are prone to rupture and, like infectious aneurysms, require early treatment (by embolization) .
Postoperative pulmonary artery aneurysms
can be due to anastomotic leaks or sub-optimal correction of the pathological hemodynamics at the pulmonary outflow. They do not always require immediate correction. •
CT Morphology
CTA in congenital aneurysms demonstrates a fusiform dilatation of a central portion of the pulmonary arteries (Fig. 24.36). Associated peripheral pulmonary artery stenoses may be present. If caused by a stenotic pulmonary valve, the aneurysm involves the most anterior portion of the pulmonary trunk. Infectious pseudoaneurysms may be found within an area of consolidation representing pneumonia (Fig. 24.37). They may develop rapidly and are prone to rupture as the pneumonia resolves. Massive hemorrhage also can cause consolidation but there are usually other morphologic signs of pulmonary hemorrhage present (centrilobular and geometric groundglass pattern) in the periphery of such a con-
Primary vasculitic pulmonary artery aneurysms are found in giant cell arteritis, poly-
arteritis nodosa, Behcet's disease, and the Hughes-Stovin syndrome. Distal pulmonary thromboembolism caused by the aneurysm can induce pulmonary infarcts and aneurysmal rupture. Aneurysmal pulmonary artery disease is a well-recognized complication of cystic arterial medial necrosis and Marfan's syndrome.
Fig. 24.36 Large aneurysm of the main pulmonary artery demonstrated by SSD (2/4/2) in a 3-year-old patient with double outlet right ventricle (DORV).
867
868
24
Vascular System Fig. 24.37 Infectious pseudoaneurysm in a consolidated lung area (a) that ruptured as the pneumonia resolved. Multiple pulmonary pseudoaneurysms (b) that developed within one week after catheter-induced
b
a
solidation, and the patient complains of hemoptysis. Other causes of peripheral pulmonary aneurysms may be hard to differentiate. Thrombosis of the aneurysm can occur in vasculitic syndromes, which can be complicated by distal embolism and pulmonary infarction. Small peripheral aneurysms may present as small enhancing nodules on thin-section CTA. Bronchial artery aneurysms are readily identified when involving the mediastinal portions of the bronchial arteries (Fig. 24.38). In the periphery they may be hard to differentiate from pulmonary artery aneurysm.
Vasculitis
Takayasu arteritis and granulomatous /temporal or giant cell) arteritis can affect the central
pulmonary artery system. In the acute stage there is inflammatory wall thickening; in the chronic stage, smooth fibrotic narrowing or occlusion with arterial thrombosis and distal pulmonary infarction may occur. CTA is excellently suited to demonstrate not only the luminal but also wall changes. Low-dose examination techniques are generally sufficient. Various vasculitic diseases of intermediate and small-size vessels may affect the pulmonary arteries. These include polyarteritis nodosa (PAN), Microscopic polyangitis, Wegener's disease, Churg-Strauss syndrome, and systemic lupus erythematosus (SLE). They are not evaluated by CTA but by HRCT and are thus discussed in Chapter 9 (p. 328). •
a
septicemia.
b
Fig. 24.38 Multiple bronchial artery aneurysms in Ehlers-Danlos syndrome (4 x 1/6).
CT Morphology
Takayasu arteritis and granulomatous arteritis cannot be distinguished morphologically. Arterial wall thickening indicates vasculitic activity and can disappear after immunosuppressive therapy. Stenosis of the affected pulmonary segments causes peripheral pulmonary perfusion asymmetry (Fig.24.39a) and leads to thinning of peripheral vessels that can be excellently demonstrated by thin-slab MIP or volume rendering. Arterial thrombosis may be a complication in more extensive disease (Fig.24.39b, c). Non-enhancing, linear or wedge-shaped peripheral consolidations may be seen as a consequence of distal pulmonary
Neoplasia
a
c
b
Fig. 24.39 Takayasu arteritis of the pulmonary arteries. Note the narrowing and thickening of the pulmonary arterial wall on thin slab MIP (2/4/2) (a). More ex-
tensive disease shows a thread-like stenosis of one pulmonary artery (4 x 1/6) (b). Note the reduced pulmonary vascularity on the VRT image (c).
infarction. Pulmonary granulomatous arteritis can (rarely) cause aneurysms that are prone to rupture.
extend into the contralateral pulmonary artery or into the right ventricle. Secondary thromboembolic events are common and may cause the presenting clinical features. CT and MR are well suited to demonstrate both the endoluminal as well as extraluminal tumor components.
Neoplasia
•
Primary Malignancies
Leiomyosarcoma or angiosarcoma of the pulmonary arteries generally develops in the main or central pulmonary arteries, often in close relationship to the pulmonary valve. The tumor can
a
b
Fig.24.40 Angiosarcoma of the right pulmonary artery (4x1/6). (a) The right pulmonary artery is completely obliterated by contrast-enhancing tissue. Note the non-enhancing apposition thrombus that ex-
CT Morphology
A sarcoma of the pulmonary artery appears as a lobulated mass within the vessel lumen that can be distinguished from thromboembolism by the presence of contrast uptake in the tumor (Fig.24.40a) . The tumor may locally invade the
c tends into the left pulmonary artery (arrow), (b) Recurrent tumor after thromboendarterectomy demonstrates mediastinal involvement on sagittal sections, (c) Hematogenic spread to the right upper lobe.
869
24 Vascular System
mediastinum or the lungs (Fig.24.40b). In about 50% of cases pulmonary artery sarcomas spread endoluminally to the periphery with a centrifugal growth pattern. This may be com plicated by obliteration of the vessel with secondary pulmonary artery thrombosis or by hematogenic metastases to the lungs (Fig.24.40c). Differentiation from metastatic tumor emboli from an unknown primary tumor may be difficult. However. dilatation of peripheral arteries. as seen in tumor embolism, is not a common feature.
Tumor Invasion Although central pulmonary artery tumor invasion can occasionally be seen in stage IV bronchogenic cancer or in metastatic disease, the preoperative identification of arterial wall infiltration presents a major diagnostic problem. Intravascular or transesophageal ultrasound is probably superior to CTA. Pulmonary veins are frequently involved in bronchogenic cancer or metastatic tumor. The
risk of systemic metastatic spread will increase with pulmonary venous infiltration. In advanced cases segmental obstruction of pulmonary drainage can occur, which can be complicated by retrograde thrombosis. CT Morphology
Extrinsic compression of the pulmonary arteries is frequently observed in advanced stages of bronchogenic carcinoma or other malignancies infiltrating the hila or mediastinum. Circumferential tumor encasement of the pulmonary artery is associated with a higher wall infiltration rate. CTA can directly demonstrate involvement of pulmonary veins with intraluminal filling defects or occlusion of the vessel. Signs of venous obstruction with segmental or lobar edema and altered contrast dynamics, as in pulmonary venous stenosis, can be indicators of secondary pulmonary venous thrombosis, which is demonstrated on CTA by the failure of the venous lumen to opacity (see Fig. 23.46 c).
Aorta Young patients should be carefully selected for CTA of the thoracic aorta, because pulsation artifacts can seriously degrade the quality of the images. MRA may provide a better alternative, especially in children, although low-dose scanning is feasible in these instances. For the abdominal aorta, multislice CTA is superior to MRA with respect to spatial resolution and complete coverage of the abdominal cross-section. CTA is the preferable imaging modality for all aortic emergencies.
Anatomic Variants A duplicated aortic arch is a rare anomaly that encircles the trachea and esophagus like a ring. One arch is frequently smaller and may be present only as a cord-like rudiment. The anomaly may cause dysphagia, stridor, and recurring bouts of pneumonia, or it may be clinically silent and thus become an incidental finding.
A right aortic arch occurs in approxhnately 1-2% of the population. The supra-aortic branches may either branch out in a mirrored fashion with formation of a left brachiocephalic artery, or according to the following order: left common carotid artery, right common carotid artery, right subclavian artery, and left subclavian artery. Such a left subclavian artery runs behind the trachea and esophagus and can cause symptoms of dysphagia lusoria similar to an aberrant right subclavian artery. If the distal portion of the left aortic arch is preserved during embryonic development, an aortic diverticulum olKommerell will persist at the origin of the left subclavian artery. An aberrant right subclavian artery (arteria lusoria) arising as the last branch of the aortic arch runs behind the esophagus in approximately 80% of cases and can cause dysphagia. Congenital hypoplasia of the aorta (e.g., in Williams-8euren syndrome) is a very rare anomaly, which is often associated with a su-
Anoton-uc Variants
pravalvular aortic stenosis, stenosis of major vessels (carotids, brachiocephalic, and renal arteries) as well as pulmonic stenosis. The role of CT is to detect a specific vascular anomaly that will account for the observed symptoms. Three-dimensional renderings and special contrast protocols are necessary only in preoperative studies. The principal competitive imaging modalities are MRI (to investigate symptoms) and Gd-enhanced MRA (for 3D rendering) . These procedures are preferred in youngpatients. •
CT Morphology
Variants of the. aortic i arch can be diagnosed easily following contrast administration. A sub-
3
clavian artery that arises more posterior than normal (in continuation of the descending aorta) is suggestive of an anomaly such as aortic coarctation or an aberrant right subclavian artery (Fig. 24.41). The retroesophageal artery in dysphagia lusoria is frequently visible on unenhanced scans, and it is clearly distinguishable from the esophagus after intravenous contrast administration. An aortic diverticulum appears as a circumscribed, asymmetric, aneurysm-like protrusion from the distal aortic arch (Fig. 24.42). Aortic hypoplasia is presumed to exist if the diameter of the descending or abdominal aorta is less than 1 cm in an adult patient.
Fig. 24.41 Aberranttight subclavian artery (dysphagia lusoria) with typical pattern on axial sections (a). The anatomic relationships are best appreciated on VRT (b).
a
b
a
b
c
d
Fig. 24.42 Aortic diverticulum of Kommerell: rightsided aortic arch, retroesophagealleft subclavian artery (mirror image of aberrant right subclavian artery), and
concomitant dilatation at the origin of the subclavian artery from the aorta (3/5/2, SSD).
871
24 Vascular System Three-dimensional reconstructions are optimum for elucidating complex anatomy. Leaving anatomic markers, such as the spinal column, in the data set may improve spatial orientation. Coarctation of the Aorta
Coarctation is a congenital disorder of the aorta with an unfavorable spontaneous prognosis and an average life expectancy of some 35 years. Tubular hypoplasia of the aorta (infantile preductal type) involves a long segment of the aorta distal to the origin of the brachiocephalic artery and may have a high-grade stenosis in the descending aorta proximal to the ligamentum arteriosum. Since it is diagnosed in early childhood, CTA is used mainly for the follow-up after surgical repair. MRI and Gd- MRA are preferred for this application, as they do not entail exposure to ionizing radiation. Localized coarctation (adult postductal type) is characterized by a localized constriction at, or distal to, the level of the ligamentum arteriosum. It is diagnosed later in life, often as an incidental finding. Coexistent cardiac anomalies are uncommon. CTA is an excellent tool for displaying the complex spatial relationships of collateral vessels but MRA is a potent competitor.
a
b
Fig. 24.43 Coarctation of the aorta, (a) Localized type with high-grade stenosis in the descending aorta and abundant collaterals (4 x 1/6, VRT). (b) Substantial in-
•
CT Morphology
A left subclavian artery that arises posteriorly in continuity with the descending aorta is suggestive of coarctation, as it is present in more than 70% of these patients (Figs. 24.43-44). Other indicators of coarctation are enlarged intercostal arteries and large mammary arteries. Collateral vessels are especially pronounced if the aortic arch is occluded. Collaterals are best displayed on volume-rendered images (Fig.24.43b). Their localization may become important before surgical procedures. Tubular hypoplasia is characterized by concomitant hypoplasia of the aortic arch (usually between the origins of the left carotid artery and subclavian artery) or the proximal descending aorta. In localized coarctation there is only a localized, often septal, stenosis at or around the level of the ligamentum arteriosum (Fig.24.43a). Poststenotic dilatation of the aorta is frequent. Longitudinal reformations or 3D representations are essential for the accurate assessment of septal stenoses (Fig.24.44a, b). Pseudocoarctation is an elongation of the aorta with kinking at the level of the ligamentum arteriosum (Fig.24.43c). In the follow-up after surgery, it is important to detect aneurysmatic complications and re-stenoses. Small saccular aneurysms are more prone to rupture than fusiform aneurysms. They can best be appreciated on 3D-rendered images.
c tercostal collateral formation (4 x 1/6, VRT). (c) Pseudocoarctation (arrow) with ectasia ofthe leftsubclavian artery (arrowhead).
Aortic Aneurysms
a
b
c a
Fig. 24.44 Surgical repair of aortic coarctation. Recurrent membranous stenosis after patch plasty (3/5/2), which mimics aortic dissection on axial sections (a) and is better evaluated on SSD (b). Concomitant aneurysm
of the subclavian artery in a patient after bridging of
Attention should be given to possible or accompanying aneurysms of the supra-aortic branches (Fig.24.44c). Aneurysms of the ascending aorta may be found in older patients.
depends on individual findings, and even a small lesion may warrant surgical repair. Endoluminal repair by vascular stent grafts is becoming an important alternative in patients who fulfill certain selection criteria (see also Tables 24.18-19, p. 923). The frequency of atherosclerotic aneurysms increases with age. Aneurysms are mainly located in the descending or infrarenal aorta. Extensive disease, however, may involve most of the thoracoabdominal aorta. Most aneurysms are fusiform, only some 20% are saccular. Rupture is the main cause of death but a substantial number of patients die from other causes related to their atherosclerosis. Penetrating ulcers are a consequence of ulcerative aortic plaques. They develop after destruction of the intima and may form saccular aneurysms with a high propensity for acute hemorrhage or may progress to aortic dissection. Mycotic aneurysms are due to infection that either spreads contiguously from surrounding tissue to the aorta, or is due to septicemia. Staphylococcus aureus and Salmonella are the most frequent organisms causing mycotic aneurysms. Predisposing factors include bacterial endocarditis, IV drug abuse, immunocompromised patients, severe atherosclerosis, and aortic injury due to trauma, surgery, or catheterization. The ascending aorta is most frequently affected but the visceral, intracranial, and peripheral (extremity) arteries may also be involved. Mortality is high if untreated.
Aortic Aneurysms
The most frequent cause of aortic aneurysms is atherosclerosis (up to 80%) with hypertension, followed by trauma (up to 20%). Rare causes include syphilis, fungal infection, aortitis (Takayasu, granulomatous arteritis, and many other inflammatory diseases), cystic medial necrosis (Marfan and Ehlers- Danlos syndrome), and congenital anomalies. Aortic valve stenosis may cause a poststenotic aneurysm of the ascending aorta. A distinction is drawn between true aneurysms, characterized by dilatation of the entire arterial wall, and false aneurysms with a disrupted wall that is contained by adventitia, clotted blood, and perivascular connective tissue. False aneurysms are usually associated with a confined rupture of the aortic wall. Aneurysms are classified morphologically as fusiform (circumferential) or saccular. The critical diameter of a fusiform aneurysm for elective surgical repair is frequently chosen at 5cm in the descending and abdominal aorta, and 5.5-6 cm in the ascending aorta. In case of a saccular aneurysm, the indication for surgery
coarctation by an aortic tube graft (3/5/2) (c). Note the tubular dilatation of the proximal left subclavian artery (arrowheads) .
24
Vascular System
Syphilitic aneurysms may develop in untreated patients after an incubation period of more than 10-15 years. Syphilitic periaortitis and mesaortitis predispose for aneurysm formation. mainly in the ascending aorta and the aortic arch. Inflammatory aneurysms are characterized by perianeurysmatic inflammatory tissue and fibrosis. similar to retroperitoneal fibrosis. No infectious organism is found. Media degeneration due to Marfan or Ehlers- Danlos syndrome is the most frequent cause of aneurysms in the ascending aorta. Later in life. aneurysms may form in all segments of the aorta and in smaller arteries. There is a substantially increased risk of aortic dissection in these patients. , Traumatic pseudoaneurysms form in patients surviving contained traumatic rupture of the aorta. They are the commonest form of thoracic aneurysm in young patients. The aneurysms tend to enlarge over time and may rupture years after the acute event. Thoracic aneurysms may present as a mass on chest radiographs or may become symptomatic with compression of adjacent structures such as the esophagus, the recurrent laryngeal nerve (hoarseness), the vena cava, pulmonary artery, or the trachea and main bronchi. Substernal pain or pain between the shoulders or in the back are present in some 25% of patients. Thoracic aneurysms are frequently associated with abdominal aortic aneurysms and coronary heart disease. Rupture of a thoracic aneurysm is much more dangerous than that of an abdominal aneurysm because there is little tissue that can contain the bleeding. Abdominal aortic aneurysms (AM) become more common with severe atherosclerosis and old age. They are most frequently located in an infrarenal position making them amenable to endoluminal repair by stent graft placement (requirements see Table 24.20). Extension to the renal artery orifices or even more proximally requires surgical treatment and reimplantation of the main branching arteries into the graft. AM often extend distally into the iliac arteries. Concomitant aneurysms of the visceral and renal arteries, the common, internal, and external iliac arteries, and the femoral arteries are common. AM are associated with stenoses of the celiac and superior mesenteric artery, as well as with occlusion of the inferior mesenteric and lumbar
arteries (usually due to mural thrombus). The risk of rupture increases with aneurysm size. Perianeurysmal inflammation and flbrosis may surround abdominal aneurysms and can lead to ureteral obstruction as in Ormond's disease. CTA and Gd- MRA are the imaging procedures of choice for the diagnosis of thoracic and abdominal aortic aneurysms. Arterial DSAis unnecessary unless information is needed about the coronary arteries or cerebral, spinal, or peripheral blood flow. CTA is not useful in the diagnosis of aortic valve incompetence. CT Morphology
There are no generally established size limits for diagnosing a fusiform aneurysm because the diameter of the aorta increases with age. In the abdominal aorta, an aneurysm is diagnosed if the diameter is increased by a factor of two over normal aortic segments or if the transverse diameter exceeds 3-4 cm. In the chest, a diameter of 5 cm in the ascending aorta and 4 cm in the descending aorta can be considered an aneurysmatic dilatation. The measurement of the aortic diameter includes any wall calcifications or mural thrombi that are present. Measurements should always be perpendicular to the course of the aorta and thus should rely on multiplanar reformations more than on axial section (Fig. 24.45). lfMPR are not available, the minimum diameter is often a more reliable indicator than the maximum diameter on these sections. Pulsation artifacts may complicate measurements in the ascending aorta, especially in younger patients. Since the indication of surgery in saccular aneurysms is not based on size alone, additional longitudinal CPR or volume-rendered images are mandatory for optimum evaluation (Fig.24.45c). Three-dimensional representations are particularly useful for surgical planning while CPR are most important for measurements prior to stent graft placement (see Fig. 24.96). Volume rendering gives a unique opportunity of simultaneously dis playing calcifications and vessel lumen as well as thrombus (Fig.24.47b). The involvement of vessel origins in the aneurysm complicates surgical treatment. It is therefore important to determine the precise relation of the supra-aortic arteries to an aneurysm of the aortic arch (Fig.24.46b) or the relation of the abdominal side branches to an aneurysm of the abdominal aorta (Fig.24.47a).
Aortic Aneurysms
a
b
c
Fig. 24.45 Aneurysm measurements. The minimum diameter on axial sections is more reliable than the maximum diameter (a). Correct measurements require MPR perpendicular to the centerline of the aneurysm (b) or
curved reformations through the centerline for length measurements (c). Note that the relative orientation of the sections (a) and (b) is indicated on (c).
Slow and turbulent flow of contrast medium can lead to inhomogeneous enhancement of the aortic lumen. Mural thrombus is a frequent finding and will cause underestimation of the size of an aneurysm on 3D representations unless there is calcification of the
aneurysm wall (Fig. 24.47 a). The thrombus itself may partially calcify. Irregularities and exulceration of the mural thrombus may indicate an increased risk for distal emboli, in particular during angiographic procedures. Hyperattenuating layers within an otherwise hypoat-
a
b
Fig. 24.46 Thoracic aortic aneurysms. (a) Giant, partially thrombosed saccular aneurysm of the aortic arch, causing displacement of the pulmonary vessels. Note the relationship ofthe aneurysm to the supra-aortic vessels (3/5/2). (b) Typical aneurysm in the proximal as-
c cending aorta in a patient with Marfan syndrome (4x2.5/15). (c) Large aneurysm of the descending aorta with kinking and formation of an aortic hump (4X2.5/15).
875
876
24 Vascular System Fig. 24.47 Abdominal aortic aneurysms. (a) Shaded surface displays only demonstrate the perfused lumen. Note the bilateral renal artery stenosis (3/5/2). (b) Volume rendering can display soft tissue as well as perfused lumen. Note the thrombosed portion of this aneurysm after surgical repair (3/5/2). Note that there is a recurrent aneurysm due to an anastomotic leak into the old aneurysm sac.
a
b
tenuating mural thrombus may be an indicator for fresh thrombus formation and thus recent growth of the aneurysm (see Fig.2449b). The most important information that CTA should provide is listed in Table 24.11. In addition to spatial information for treatment planning it is important to detect complications, such as (contained) rupture, or compression ofmediastinal or abdominal structures. Fistulation into adjacent organs is a rare but severe complication. Aortobronchopulmonary fistula leads to pulmonary hemorrhage with lung consolidation. Aortoenteric fistula causes hyperattenuating clots in the bowel lumen (mostly in small bowel, but often best seen in the colon), as well as direct contrast extravasation into the affected bowel segment, which is often the horizontal portion of the duodenum. Concomitant vascular disease is quite frequent. Therefore it is important to evaluate the main branching vessels for the presence of stenoses or aneurysms (Fig. 24.47 a). It should be noted that accessory renal arteries are present in 25% of the population. Occlusion of the inferior mesenteric artery (IMA) is quite frequent with collateralization via the arch of Riolan. In patients with patent IMA and lack of good collateralization, left colonic ischemia may develop as a complication of surgery, if no reimplantation of this artery was performed and if there is concomitant stenosis of the superior mesenteric or celiac artery. Involvement of the iliac arteries (aneurysms, stenoses) is an important factor in
Table 24.11
Checklist for aortic aneurysms
Thoracic
Diameter Signs of rupture Involvement of the aortic root or aortic arch Longitudinal extension toward the diaphragm Aortic elongation: kinking, hump Compression of superior vena cava esophagus trachea and central bronchi Aortobronchopulmonary fistula Abdominal
Diameter Signs of rupture"AAA! (confined perforation is not uncommon) Penetrating ulcer? Inflammatory aneurysm (perianeurysmal fibrosis) Length Distance from the aneurysm to the origin of the? renal arteries (proximal neck?) Involvement or stenosis of splanchnic arteries Involvement of aortic bifurcation and iliac arteries Retroaortic left renal vein
deciding whether to implant a tubular or yshaped prosthesis. Elongation of the aorta is associated with longstanding arterial hypertension and substantial arteriosclerosis. Elongation of the descending aorta shifts the aortic arch anteriorly
Aortic Aneurysms 877 and forms an aortic hump (Fig.24.46c). This hump should not be mistaken for the aortic arch because surgical treatment is much more complex for arch aneurysms than for aneurysms of the descending aorta. Elongation of the distal descending aorta just cranial to the thoracoabdominal junction may cause massive kinking of the aorta. Similar findings are seen in the proximal abdominal aorta. All these regions are predilection sites for fusiform aortic aneurysms that frequently involve multiple aortic segments. The etiology of an aneurysm can sometimes be suspected from its location and morphology. Aneurysms of the ascending aorta are most frequently caused by degenerative disease (e.g., Marfan syndrome Fig. 24.46 a), or have a mycotic or syphilitic etiology. Aneurysms of the descending aorta in a young patient are most likely post -traumatic, while other aneurysms of the descending and abdominal aorta are primarily caused by atherosclerosis. Penetrating aortic ulcers cause eccentrically located aneurysms at any location within the aorta although the descending and abdominal aorta are
a
b
Fig. 24.48 Traumatic pseudoaneurysm of the descending aorta at a typical location (3/5/2) (a). Mycotic aneurysm of the ascending aorta at the site of aortic
more frequently involved (see Fig.24.S7) A penetrating ulcer can be suspected if the aneurysm has overhanging edges or a focal dissection. Traumatic pseudoaneurysms typically occur in the concavity of the aortic arch, distal to the origin of the left subclavian artery, and usually bulge anteriorly. When an aneurysm is detected at this site, a history of trauma should be elicited. Aneurysms are usually calcified and may contain thrombus. The findings on axial sections can resemble aortic dissection (Fig. 24.48a). Syphilitic aneurysms preferably involve the ascending aorta and the aortic arch, and more rarely the sinus of Valsalva or the pulmonary artery as well. Saccular aneurysms of the ascending aorta are common, but fusiform dilatation may also occur. Combined forms with saccular aneurysms protruding from a fusiform aneurysm are possible. Frequently there is extensive calcification in the aneurysmal wall. Signs suggesting a mycotic etiology are an ascending aortic aneurysm with a saccular configuration (Fig.24.48b). During the acute phase of infection, aneurysms tend to grow rapidly.
c cannulation in a heart transplant patient (3/5/2) (b). Inflammatory abdominal aortic aneurysm with a rim of periaortic inflammatory tissue (c).
24 Vascular System
They may lead to an interrupted mural ring of calcification. Associated findings may include mediastinal lymph node enlargement and. rarely. gas in the periaortic tissues. Abscess formation or osteomyelitis adjacent to he aorta may be present. Inflammatory aneurysms occur in the abdominal aorta. There is thickening of the aortic wall and perianeurysmal fibrosis that may extend into the retroperitoneum and can involve the adjacent organs. Characteristically. three distinct layers surround the perfused lumen: mural thrombus. aortic wall with calcifications. and perianeurysmal soft tissue. The soft-tissue rim surrounding the aorta often shows little contrast enhancement during the arterial phase (Fig.24.48c). Enhancement is best appreciated on delayed scans acquired during the interstitial phase (>90s). Possible secondary effects are ureteral obstruction or the involvement of intestinal loops. Aneurysmal dilatation of the aortic lumen in a chronic aortic dissection determines the indication for surgical treatment (see Fig. 24.59). Thrombosis in one of the lumina can create the appearance of a simple aneurysm with a thrombotic rim.
Aortic Rupture The most important complication of aortic aneurysms is rupture. which is defined as seepage of blood through clefts in the aortic wall. The larger the diameter of the aneurysm, the greater is the likelihood of eventual rupture. In the abdomen, the risk grows from 10% in aneurysms smaller than 4cm to 60% in aneurysms larger than 10cm. Elective surgery is indicated if the aneurysmal diameter exceeds 4-5 cm in the descending and abdominal aorta, and 5-6 em in the ascending aorta. Saccular aneurysms have a higher risk of rupture, and there is no absolute size limit that governs the decision to operate. Surgical mortality rises substantially in patients with ruptured aneurysms. Thoracic aneurysms may rupture into the mediastinum but can also rupture into the pleural space or pericardium, which makes such an event much more dangerous than in abdominal aortic aneurysms. Abdominal aneurysms commonly rupture into the retroperitoneum, which contains the hemorrhage to a certain degree. Only if the peritoneum itself ruptures,
the blood may freely fiow into the peritoneal cavity, leading to a high risk of sudden death. Rarely can an aneurysm rupture into the gastrointestinal tract causing massive gastrointestinal hemorrhage (aorto-duodenal fistula), or into the inferior vena cava leading to rapid cardiac decompensation. CT is indicated for suspected aortic rupture, provided the patient is not referred for immediate surgery based on ultrasound findings. Depending on the clinical setting, noncontrast CT should be followed by CTA in hemodynamically stable patients for the purpose of surgical planning. In the abdomen, CTA is important to define whether insertion of a bifurcated prosthesis or reimplantation of the renal arteries is required. In the chest, CTA will be able to better demonstrate the relation to the supraaortic branches. If contrast material is used, the examination should be completed as expeditiously as possible to avoid an unnecessary delay in surgery. Whenever possible, however, scan delay should be determined individually because the circulation time may be substantially prolonged (to beyond 60s) in patients with shock. As soon as it is clear that contrast enhancement ofCTA is sufficient the patient can be taken off the CT table. To speed up evaluation, it may be advisable to reconstruct fewer images first with larger spacing to determine the major findings. Later an overlapping secondary raw data set can be reconstructed, from which 3D representations for fine-tuning of the surgical procedure can be derived. CT Morphology
Impending rupture 'nay
be indicated by intraruural hematoma, a crescent-shaped hyperattenuating rim of clotted blood within the aortic wall on non-contrast CT. Hyperdense portions within a pre-existing hypodense mural thrombus are an indicator of fresh thrombus formation and thus recent growth of an aneurysm (Fig.24.49b). Furthermore, nose-like protrusions of perfused lumen into a clotted region suggest a higher risk for rupture. Small amounts of stranding around an aneurysm indicate minor hemorrhage and can be the precursor to aortic rupture. Contained rupture occurs mainly in the abdominal aorta and is characterized by local hemorrhage or acute pseudoaneurysm formation (Fig.24.49a). Sometimes interruption of a
Aortic Rupture
a
c
b
Fig. 24.49 Aortic rupture, (a) Contained rupture with interruption of the mural calcification and localized thrombosed pseudoaneurysm.(b) Precontrastscan of a retroperitoneal rupture of an AAA. Note the high at-
tenuation portions within the mural thrombus of the aneurysm indicating fresh thrombotic material (arrowheads). (c) Retroaortic contrast extravasation due to an acute rupture.
wall calcification may be detected at the site of the pseudoaneurysm. Active bleeding is present if contrast extravasation can be observed (Fig.24.49c). Rupture of the thoracic aorta is manifested by hyperattenuating or streaky opacities in the mediastinum and by hemorrhagic pericardial or pleural effusion (Fig. 24.50). The CT numbers of hemorrhagic effusions may be increased to above 30 HU, depending on the blood concentration in the fluid. In patients with hemorrhagic pericardial effusion (Fig.24.50a), there is an increased risk of cardiac tamponade due to
progressive intrapericardial hemorrhage. Despite the urgency ofthe situation, it is sound practice to evaluate a suspected rupture with contrast -enhanced scans whenever possible in a clinically stable patient, since there may be a dissection that was not visible on unenhanced scans, and the extent of the intimal flap may require modification of the surgical approach. Hemorrhage from the ascending aorta may enter the perivascular connective tissue of the pulmonary artery and then may extend into the interstitium of the lungs (see Fig.24.55b) .
a
b
Fig. 24.50 Hemorrhagic pericardial effusion with a density identical to that of heart muscle (a). Pleural hemorrhage with CT numbers of 40-50 HU in a patient
c with a ruptured acute type B dissection (b).Coronal reformation (4 x 1/6) (c).
879
24 Vascular System Table
24.12
Signs of aortic perforation
Intrathoracic • Hemorrhagic pleura! effusion • Streaky mediastinal densities adjacent to the aorta • Pericardia! effusion (caution: pericardial tamponade) • Mediastinal hematoma Intra-abdominal • Stranding of para-aortic tissue • Para-aortic hematoma • Hyperattenuating ascites
Rupture of the abdominal aorta appears on unenhanced CT as increased attenuation or stranding of the retroperitoneal fat surrounding the aorta (Fig.24.49b. c). Frequently this obscures the vascular margins. Free intraperitoneal fluid is rare. but its presence signals a very unstable situation in which the rupture is no longer confined by the peritoneum. Acute hemorrhage need not be hyper attenuating but may be isoattenuating relative to muscle because of a lack of clotting or separation between blood cells and serum. It may even be hypoattenuating if the hematocrit is low because of plasma expanders. If signs of ru pture are noted in a symptomatic patient (Table 24.12). immediate surgery should be considered. Traumatic Aortic Injury See below, Trauma, p. 915
Dissection of the Aorta Acute Dissection Aortic dissection is a separation of the aortic intima and adventitia. There is a primary intimal tear that lets circulating blood gain access to the media and thus split the aortic wall in two. Aortic dissection has an incidence of approximately 3:1000. The prognosis depends critically on the site of the primary intimal tear and the extent of the intimal flap. A dissection that involves the ascending aorta can become acutely life-
threatening due to acute aortic valve incompetence or perforation into the pericardium with acute tamponade. Risk factors for aortic dissection are arterial hypertension in patients with atherosclerosis (over 60% of cases), Marfan syndrome (some 15% of cases) , aneurysms, aortic valvular disease, trauma, idiopathic medial necrosis, and Ehlers-Danlos syndrome. Today the Stanford classification (Fig. 24.51) is most widely used, which distinguishes dissections that involve the ascending aorta (type A) from dissections that are confined to the descending aorta (type B). Untreated acute type A dissections have a mortality of approximately I % per hour and therefore require urgent surgical treatment. The older DeBakey classification is still of value , as it distinguishes the more surgically and prognostically favorable Stanford type A dissections confined to the ascending aorta (DeBakey Type 11) from dissections that extend beyond the aortic arch (DeBakey Type 1). Acute type B (DeBakeyType 111) dissections have a more favorable prognosis and are often treated conservatively. Patients complain of a sharp, tearing, intractable chest pain that is primarily anterior in type A dissection, and posterior in type B disease. As the dissection progresses, the pain may extend from the anterior chest to between the shoulder blades, down to the lower back, or up into the neck and head. Secondary symptoms depend on complications of the dissection. Acute aortic valve incompetence with regurgitation as well as hemodynaruic shock may substantially reduce cardiac output. Further complications include rupture into the pericardium with pericardial tamponade, and rupture into the pleural space, both of which are frequently fatal and require immediate surgery. The false channel has a propensity to dilate and to compress the true lumen or the orifice of a side branch vessel. Compression or occlusion of major aortic branches in the neck, abdomen, and pelvis may lead to cerebral infarctions, renal failure, mesenteric ischemia, or acute claudication. Rupture into the right ventricle, left atrium, superior vena cava, or pulmonary artery is very rare and usually fatal. Rupture or leakage of blood into the mediastinum is more frequent (see Fig. 24.56). Compression and occlusion of coronary arteries may lead to myocardial infarction. Because the wall of the false channel is very thin, aneurysmatic dilatation is frequent and may require surgery.
D/ssectf'on
of Aorta
Fig. 24.51 Classificationof aortic dissection. In the widely used Stanford classification, dissections are classified as type A or B depending on the involvement of the ascending aorta.
Type I
Type II Type A
The technique of eTA is critical for good results. Cardiac output is frequently reduced and can prolong the arrival time of contrast material to beyond 60s. Bolus timing (test bolus injections or bolus triggering) is therefore mandatory. Thin collimation should be used if side branch involvement is suspected. While this is not critical with multislice scanning, single slice CTA may not provide sufficient resolution for subtle assessment of side branch involvement if chest and abdomen have to be scanned together. Two separate scans may be more advisable in such situations. CTA has almost completely substituted for angiography in the diagnosis and evaluation of complications in patients with suspected aortic dissection. Transesophageal echocardiography (TEE) has a role only in suspected acute dissection as a bedside test. MRA is a good alternative in patients with chronic dissection but patient monitoring is problematic in the acute stage and makes MRA not very convenient for this patient group.
•
Type III
(DeBakey)
Type B
(Stanford)
CT Morphology
Non-contrast CT is important because it most clearly demonstrates intramural hematoma, mediastinal hemorrhage, and hemorrhagic pericardial or pleural effusion, as well as calcifications within the aortic lumen. Table 24.13 provides a checklist of the most important questions that have to be answered by imaging. CTA is required to detect the dissection membrane. In the ascending aorta, this intimal Table 24.13
Checklist for aortic dissection
• • • • • • •
'SjBlIIIIBlBBBB
Stanford Type A or Type B? Signs of rupture? Mural hematoma? Thrombosis of true or false lumen? Compression of true or false lumen? Aortic valve incompetence? Concomitant involvement (dissection) of aortic side branches • Stenosis, compression, or occlusion of aortic side branches (including the coronaries) • Impaired perfusion of abdominal organs • I ocation. of iotimal tear (entry, re-entry)
BBSB8Saa'SBascuSiiBalaaeB""
881
24 Vascular System
a
b
Fig. 24.52 Pulsation affects in acute type-A dissection. Rapidly moving membranes in ascending aorta lead to multiple contours and may make identification of intimal flap difficult (a). Pulsation in the descending aorta leads to double contours of the flap (b).
membrane is subject to marked pulsation. Since it is very thin, it moves rapidly and is therefore displayed at multiple positions within the same axial section. Such fresh intimal flaps are often hard to delineate close to the aortic root but they become more clearly visible in the distal ascending and descending aorta, where a fresh membrane frequently displays a double contour (Fig.24.52a, b). Direct visualization of the intimal flap should always be possible, given good enough contrast enhancement. The sensitivity and specificity of CTA for this purpose is well over 95%, given proper examination technique. The primary entry site in type A dissections is often visible but its size is hard to evaluate because of marked pulsation artifacts. Pulsation can be suppressed with ECG-gated data acquisitions during multislice scanning. A primary entry is typically located at the right anterior wall of the ascending aorta in the region where the blood jet from the aortic valve hits the wall. The primary entry in type B dissections is frequently located in the descending aorta just beyond the left subclavian artery. Dissection usually progresses in an ante grade fashion further distally until it ends at a distal re-entry. Retrograde dissection may occur and can involve the aortic arch from a more distal entry, or involve the proximal descending aorta from a more distal entry, e.g., from just above the thoracoabdominal junction.
True and false channels usually show a helical pattern that rotates around the aortic axis (Fig. 24.53). In type A dissections the true lumen lies posteriorly in the ascending aorta, spirals upwardly so that it gives rise to the supra-aortic arteries, and then moves to an anterior position in the proximal abdominal aorta, where it commonly gives rise to the celiac and superior mesenteric artery. At the level of the renal arteries the true lumen is usually positioned on the right and gives rise to this renal artery while the left is frequently at least partially perfused by the false channel. There are, however, marked interindividual variations, and differentiation of true and false channel should never be based on position alone. The distal re-entry in dissections
a
b-e
Fig. 24.53 Volume rendering of a type B dissection (a). Note the thrombosis with little perfusion (arrow) in the proximal false lumen (b), the beak sign (arrow) identifies the false lumen (c), detachment of the membrane from an arterial orifice leaves a hole (arrowhead) at the level of a renal artery originating from the false lumen (d). The dissection extends into the left iliac artery (e).
Dissection of Aorta
involving the abdomen is often locat.e? in the left iliac artery (some 80%). In addition to a main distal re-entry there are multiple smaller holes in the membrane whenever the intimal flap unattaches from an arterial orifice (Fig.24.53d). Calcifications ofthe whole wall act like glue. They do not allow for se~aration of the aortic wall into two layers and will change the typical pattern of dissection (Fig.24.5~a). Aortic dissection at an usual location should raise the suspicion of a penetrating ~ortic as the etiology. Such dissections are first localized but can dissect in an antegrade or retrograde fashion and involve large portions of the aorta. The true lumen and false channel can be distinguished by traci~g the. true. lumen from .the beginning of the dissection In the ascending aorta (type A) or following the co~ti~uity of the normal aorta (type B). Rarely, t.he intimal m~mbrane is completely destroyed In the ascending aorta or there are multiple channels (Fig.24.54b). In such cases, the following secondary signs can be taken into account:
=:
• Due to slow flow the false channel frequently has a larger cross-section. • The true lumen is compressed in some 80% of cases. • The false channel has a higher propensity to thrombose (see Fig. 24.62 a). • Slow blood flow leads to delayed opacification of the false channel at the begin-
a
b
Fig. 24.54 Calcification of th~ aortic wall (arro~) acts like glue and may induce atypical shap.es of the dissecting membrane. Note the high contrast In the false cha~nel in this late image from a chest CTA study (a). Multiple channels may form in some patients with Marfan
ning of the scan (Fig.24.53b) a~d prolonged opacification towards Its end (Fig.24.54a). ~ The false channel may form an acute angle ("beak" sign, Fig.24.53c) at the ju.nction between intimal membrane and aortic wall. When one lumen wraps around the other lumen in the aortic arch, the inner lumen is invariably the true lumen. The outer wall of the false channel is thinner than the outer wall of the true channel. Aortic cobwebs (bands of elastic fibers) may span the false channel (Fig. 24.54c). Because atherosclerotic calcifications are primarily located in the i?tima, oute~ ~all calcification and eccentnc flap calcification are indicative of the true lumen (in "acute dissection).
Complications such as rupture,
~emorrhag~, aneurysm formation, or malperfusion of aortic branches are excellently visualized by CTA. However, aortic valvular regurgitation cannot be directly visualized by CTA but ectasia of the aortic valve annulus is an indirect sign. Ultrasound is best suited to confirming the suspected diagnosis. Involvement of the coronary arteries is also hard to demonstrate by CT unless ECG gating is performed. . . Perforation, generally involving the thoracic aorta, is manifested on CT scans by a most
c syndrome (b). Aortic (arr?w) c~b~ebs are strands of tissue from the aortic media that Indicate the false channel (c). Note the almost complete detachment ofthe intimal flap from the aortic wall (t = true lumen).
883
884
24 Vascular System Fig. 24.55 Mediastinal hemorrhage along the connective tissue layer around the pulmonary arteries (arrow) (a) that extends into the lung interstitium (b). Note the concomitant intramural hematoma (arrowheads) in the ascending and descending aorta.
a
b
frequently left-sided hemorrhagic pleural effusion and streaky mediastinal opacities or hematoma. A hemorrhagic pericardial effusion that has a CT attenuation above 20HL] (see Fig.24.50a) strongly suggests an incipient perforation of the aorta into the pericardial sac, with a high risk of tamponade. Fluid with approximately the same CT attenuation as muscle tissue is compatible with fresh, not yet clotted blood (see Fig.24.50b). Blood may spread along connective tissue surrounding the pulmonary artery into the perivascular lung interstitium (Fig. 24.55). Aneurysms may form rapidly because of the substantially reduced stability of the aortic wall. The ascending and descending aorta are most frequently involved. In the long run, they may become markedly elongated. The same size criteria hold true as for atherosclerotic aneurysms. Fake negative and false positive findings may be due to excessive pulsation artifacts (see Fig. 24.13). Inadequate contrast opacification has to be avoided. A thrombosed false channel should not be mistaken for a mural thrombus. The aortic valve leaflets and the normal sinus of Valsalva must not be mistaken for dissecting membranes. In general, distinction is simple by following the suspicious findings over multiple sections. Pulsations involve the aorta symmetrically on either side, and come and go in a regular fashion depending on the heart rate and table speed of the scanner.
Intramural Hematoma
Intramural hematoma is supposed to be due to subintimal hemorrhage into the aortic wall from rupture of the vasa vasorum, without the presence of an intimal tear. Frequently however, similar findings with hemorrhage in the aortic wall can be seen if there is an intimal tear that forms an entry site but no sufficient distal reentry. Such a situation leads to thrombosis of the blood that leaked into the aortic wall. Both entities are morphologically almost indistinguishable. By analogy with aortic dissection, type A and type B intramural hematoma are distinguished. Although both types can heal spontaneously, type A intramural hematoma has a less favorable prognosis than type B disease. In some 80% of cases it develops into a overt dissection, leads to rupture or pericardial tamponade. Conversely, 80% of type B hematomas have a benign course and can be treated conservatively. •
CT Morphology
Non-contrast CT is excellently suited to detect the typically crescent-shaped hyperattenuating areas in the aortic wall (Fig.24.56a), which may be harder to identify on contrast -enhanced scans (Fig.24.56b). On CTA, small entries (or re-entries) can be detected in some patients without overt perfusion of the false channel (Fig.24.56c). If there is continuous intramural bleeding, the thickness of the wall increases and compresses the true lumen.
Dissection of Aorta
a
c
b
Fig. 24.56 Intramural hematoma with hyperattenuating rim in the aortic wall (a). Intramural hematoma is harder to appreciate after contrast material injection
(b). Small perfused areas at the orifices of the intercostal arteries in athrombosed false channel (c).
Penetrating Atherosclerotic Ulcer
•
Penetrating atherosclerotic ulcer has been described as a separate entity since 1986. It is an ulcerated plaque penetrating the internal elastic lamina causing hematoma formation within the media. It can be viewed as a focal chronic dissection. Most patients are of advanced age and hypertensive. They are frequently asymptomatic, but may present with acute chest or abdominal pain. Ulcers may grow and extend longitudinally resulting in a "double-barreled" or thrombosed dissection, or they may penetrate through the media into saccular pseudoaneurysms. Pseudoaneurysm formation is seen in 25% of patients, with increased risk of aortic rupture (8%). Distal embolization is another common complication. Differential diagnosis includes aortic dissection, aortic aneurysm with irregular thrombus, and intramural hematoma. Distinction from intramural hematoma and dissection has important therapeutic consequences as therapy for ulcers (surgical interposition graft, or short endovascular stent graft) differs from that for dissection. CTA has the benefit of direct visualization of thrombosed portions in pseudoaneurysms, calcifications, vessel lumen, and vessel wall, and is preferable to DSA. In patients that are not operated on, careful CT follow-up is advised, especially during the first month.
CT Morphology
The findings with penetrating atherosclerotic ulcer overlap with those from intramural hematoma and dissection. The classic form results in focal involvement with a localized hematoma in the aortic wall. This can occur anywhere, but is most commonly found in the middle or distal descending thoracic aorta and best seen on noncontrast CT. Other distinctive features include the presence of associated ulcerated plaque (may be multiple) with or without calcifications, focal aortic wall thickening, or enhancement and sparing of the region just distal to the left subclavian artery origin. Pseudoaneurysms due to penetrating atherosclerotic ulcer have to be distinguished from true saccular aneurysms (Fig.24.57a). Calcification of the outer wall of the aneurysms suggests a true saccular aneurysm. Focal dissections at unusual locations (outside the ascending or proximal descending aorta) suggest a penetrating ulcer as their etiology (Fig. 24.57). They may be only a few cm long but can involve longer sections. They may even develop into a full retrograde dissection that extends up to the left subclavian artery. Side branch involvement is rare and occurs only in extensive cases. Such cases are virtually impossible to distinguish from classical type B dissections unless long-term follow-up is available that demonstrates the gradual progression of the dissection.
24 Vascular System Fig. 24.57 Penetratingulcer presenting as pseudoaneurysm (a) or focal dissection (b).
a
b
Chronic Dissection
•
In chronic dissection a neointimallayer forms in the false channel and leads to a thickening of both the intimal flap and the outer wall of the false channel. Pulsation effects are much less prominent than in acute dissection, resulting in better delineation of the flap (Fig. 24.58a). Thus, direct visualization of the entry tear is far more easily accomplished in chronic dissections than in acute cases. Even small entries and re-entries can be detected. Mural thrombi may accumulate in the dilated false channel as well as the true lumen (Fig.24.58b). Wall calcifications may also form in the false channel. Attention has to be paid to the aortic diameter because of gradual formation of aneurysms that .may involve large
Chronic aortic dissections are those older than 2 weeks. Most untreated chronic dissections are type B disease but, occasionally, untreated patients with type A dissections may survive the acute stage. Most chronic type A dissections are seen in surgically treated patients with repair of the ascending aorta. Malperfusion of aortic side branches and growing aneurysms are the main complications that occur in this patient group. Rarely, a second event may take place, which may lead to an extension of an already existing dissection or may even form another lumen in the aorta.
a
b
Fig. 24.58 Chronic aortic dissection, (a) The thickness of the intimal flap increases in chronic dissection, (b) Mural thrombus may form in the true or false channel. Note the rupture ofthis abdominal aneurysm in chronic type B dissection, (c) Chronic ischemia ofthe kidney due
CT Morphology
c
d to intra-aortic compression of an intimal flap that did not detach from the renal arterial orifice, (d) Multiple channels may form in the aorta. Note the involvement oftheSMA.
Dfssectfon of Aorta
a
b
d
c
Fig. 24.59 Chronic aortic dissection type B with aneurysmatic dilatation of the whole descending aorta and formation of an aortic hump (3/5/2). Note the
thrombus in the false channel and the compression of the true channel (arrow, c). There is formation of an aortic hump (SSD, d).
portions of the aorta (Fig. 24.59). Large aneurysms may be present within the course of a dissection without evident intimal flaps at the site of the aneurysm (Fig. 24.60). Malperfusion of organ arteries may also occur in chronic dissection and can lead to chronic organ ischemia due to compression of a lumen or direct compression of the intimal flap (Fig.24.58c). Acute dissection on top of a pre-existing chronic dissection will again cause a thin intimal membrane that moves rapidly with pulsation.
Malperfusion
Malperfusion is a serious complication of acute and chronic aortic dissection. All branching arteries including the coronaries can be involved. There are morphologic and hemodynamic factors that lead to malperfusion. Imbalance between blood velocities, inflow and outflow from the false channel, as well as compression of one channel or obstruction of the branch vessel by intimal flaps (Fig. 24.61) can all cause clinically relevant organ malperfusion. •
a
b
Fig. 24.60 Chronic aortic dissection type B with formation of an AAA. In the aneurysm, no flap can be distinguished. There are double-barreled vessels proximal and distal to the aneurysm (a). Virtual angioscopy demonstrates the beginning of the intimal flap distal to the aneurysm and a small calcification in the flap itself (b).
CT Morphology
The most serious complication is the thrombosis of the true lumen due to massive compression and outflow obstruction, or a thrombosis of the false channel with marked compression of the true lumen (Fig.24.62a). Some patterns may look quite harmless because CT will display opacification of all vascular beds that are at all perfused (Fig.24.62b) despite high-grade obstruction of the blood flow. Differences between left and right -sided enhancement of the kidneys or marked reduced organ perfusion may be readily apparent on CT. Intimal flaps that have not detached from the artery orifice may lead to perfusion of an artery from the true lumen despite it being situated on the side of the false channel (see Figs.2458c and 24.94a). In such situations, stenosis of these tube-like flaps may occur within the aortic lumen. In addition, as these flaps detach from the orifice, they may fold back and
887
888
24
Vascular System
a
b
Fig. 24.61 Side branch perfusion in aortic dissection. (a) Normal situation with distal re-entry, (b) No or insufficient distal re-entry leads to bulging of the false channel and compression of the true channel (from Borst, 1995).
lead to obstruction and even thrombosis of a branch vessel. Dissecting membranes extending into branch vessels may cause no harm if there is a distal re-entry but may lead to distal stenoses by an enlarged sack -like false channel (Fig. 24.61). Such a constellation has to be actively searched for because it is easily overlooked. It is best appreciated on CPR from multislice data sets. Malperfusion of the bowel may occur with involvement of one or more splanchnic arteries, especially when combined with pre-existing stenoses (Fig. 24.63).
Fig. 24.62 Complications of aortic dissection, (a) Complete thrombosis of the false lumen with massive stenosis of the true lumen, (b) High grade compression of the true channel by a dilated false channel. Note the malperfusion of both kidneys. There is the danger of malperfusion of the visceral organs, (c) Side branch involvement with extension of the intimal flaps into the left renal artery.
Aortic Stenosis and Aortic Occlusion
Aortic stenosis may occur in older patients as a consequence of severe atherosclerosis. It is caused in young patients by congenital hypoplasia (e.g., supravalvular aortic stenosis in patients with Williams-Beuren syndrome) or midaortic syndrome. The latter has an unknown etiology but is probably related to fibromuscular hyperplasia. Letiche syndrome includes an acute occlusion of the aortic bifurcation, usually caused by acute thrombosis at a pre-existing site of atherosclerotic stenosis. An embolic pathogenesis is less common. It is important to determine the begin of the terminal aortic occlusion
Fig. 24.63 Acute aortic dissection type B with highgrade stenosis of the celiac artery (arrowhead), extension ofthe intimal flap into the SMA (arrow), and stenosis of the inferior mesenteric artery (arrowhead) (4x 1/ 6, VRT).
Aortic Stenosis and Aortic Occlusion 889 relative to the renal arteries, to confirm or exclude renal artery stenosis, and to assess the feasibility of treatment with a bifurcated aortic graft. CTA provides excellent information about the presence of large calcified plaques or other wall changes in atherosclerotic stenosis, and is able to describe the 3D relationships in patients with midaortic syndrome. CTA can diagnose Leriche syndrome, but only multislice scanning provides the performance to exclude concomitant renal artery stenosis and to assess the peripheral arteries. Gd-enhanced MRA requires modern equipment to assess multiple levels
Fig. 24.64 Midaortic syndrome in a 15-year-old female with a stenosis of the SMA and a large anastomosis of Riolan (low-dose scan, 80 kVp, 2.8 mGy, 4 x 1/6, VRT) (a). The reduced aortic diameter, the high-grade stenosis of the left renal artery, and the occlusion of the right renal artery are best appreciated on a PA view (b). The axial section demonstrates little difference in diameter of aorta and SMA in another patient (c).
from the aorta to the peripheral arteries. Only under these conditions is MRA a diagnostic alternative. •
CT Morphology
CTA easily differentiates soft and hard atheromatous plaques as the cause of atherosclerotic aortic stenosis. In a heavily calcified aorta, wide window settings have to be used. The longitudinal extent of disease can best be appreciated on curve planar reformations. MIP are not helpful for this purpose. In patients with midaortic syndrome the midabdominal aorta is reduced in diameter and
b
a
c
Fig. 24.65 Acute Leriche syndrome with occlusion caused by an embolus straddling the aortic bifurcation (5/8/2) (a). Chronic Leriche syndrome with complete thrombosis of the aortic lumen up to the renal arteries (4 x 1 /6, VRT). Note occlusion of the left renal artery and the extensive collateral formation and the distal reconstitution of flow in the superficia~ femoral artery (b).
a
b
24 Vascular System may be as small as the superior mesenteric artery. In Williams-Beuren syndrome other portions of the aorta may be hypoplastic. In midaortic syndrome it is common to find involvement of the renal arteries and splanchnic vessels. Extensive collateralization with a hypertrophic artery of Rio Ian is present if the splanchnic vessels are involved. The stenoses are most pronounced in the proximal vessel segments [Fig. 24.64). Inflammatory para-aortic changes are not observed. Volume-rendered displays provide best results. The terminal aortic occlusion in acute Leriche syndrome is usually located just above the bifurcation (Fig. 24.65 a). In chronic cases the aorta becomes filled with clot to the level of the renal arteries (Fig.24.65b. c). Thrombotic and embolic occlusions can be distinguished by detecting an intraluminal embolus. which appears as a partially occlusive filling defect. Embolic occlusions are typically associated with a milder grade of aortic sclerosis.
Aortitisl Arteritis
Takayasu aortitis is a rare disorder that leads to progressive occlusion of the thoracic and abdominal aortic branch vessels. In the acute inf1arumatory stage. a granulomatous infiltration of the vessel wall occurs that leads to nonspecific signs of fever. weight loss. myalgia. and arthralgia. In the later fibrotic stage. constriction of the vessel wall with stenosis. occlusion (lack of pulse) or aneurysm formation may occur. According to location Takayasu arteritis is classified into four types (Table 24.14). If there is clinical suspicion of aortitis. CTA should be performed before arterial DSA, as it can sensitively detect the wall changes even in the initial stage of the disease. Granulomatous {temporal or giant cell} arteritis is a systemic granulomatous vasculitis that can affect any artery of the body but is more frequent in the supra-aortic arteries. particularly affecting the temporal artery. It occurs in patients> 50 years of age and is frequently associated with polymyalgia rheumatica of the shoulders and hips. Unlike Takayasu arteritis. the acute stage lasts only for 1-3 weeks and is associated with flu-like symptoms and often a unilateral headache. The hallmark is a rapid erythrocyte sedimentation rate (ESR >40mm/hr).
Table /I
24.14
Classification
ofTakayasu
Type I
Aortic arch and supraaortic
Type
Descending thoracic abdominal branches
g
Type III
arteritis arteries
+ abdominal
aorta +
Combination oftype I and type II
Concomitant aneurysmal disease of the pulmonary artery and dilatation of the aortic root with aortic valve incompetence has been reported. Syphilitic aortitis occurs in stage III of the disease after an incubation period of more than 10-15 years. It is characterized by periaortitis and mesaortitis. which predispose for aneurysm formation. mainly in the ascending aorta and the aortic arch. There is secondary injury of the intima. leading to accelerated calcific atherosclerosis. Polyarteritis nodosa is a rare vasculitis involving systemic small and medium sized arteries and arterioles. All organs may be involved but kidney, heart, and liver are most frequently affected. CT Morphology
Takayasu arteritis leads to thickening
and enhancement of the vessel wall. The vessel wall may be seen to be hyperattenuating already on precontrast scans. In the acute stage, it takes up contrast substantially, which can already be seen in the arterial phase of CTA but is more prominent in later phases (which also allow for a better differentiation from atherosclerotic wall thickening). Inflammatory wall changes may be detected even if the DSA is normal (Fig.24.66a). Takayasu aortitis commonly involves the aorta, the pulmonary arteries, the left subclavian (Fig.24.66b), left common carotid, and brachiocephalic artery in the chest, and the celiac, superior mesenteric, and renal arteries in the abdomen. Rarely, smaller arteries are involved. Proximal stenoses of the supra-aortic branches are typical findings. The stenoses may be long and smooth but can also be short and irregular. Skip lesions are frequent. There is abundant collateralization in the chronic stage. Fusiform aortic ectasia and aneurysms may be found. Granulomatous arteritis cannot be distinguished morphologically from Takayasu arteritis
Neop/asia
a
c
b
Fig. 24.66 Takayasu aortitis with significant wall thickening in a patient with no angiographic evidence of supra-aortic stenosis (a). Takayasu arteritis with highgrade stenosis and wall-thickening of the subclavian
artery (4 x 1/6) (b). Granulomatous arteritis of the superior mesenteric artery and its side branches (4x 1/6)
but it can be suspected if arteries are involved that are rarely affected by Takayasu disease (Fig.24.66c). It is commonly bilateral and symmetric, and there is a lack of atherosclerotic changes to account for the vascular narrowing. The common carotid artery is usually not involved. , , Syphilitic aortitis also causes aortic wall thickening with increased contrast enhancement due to inflammation. A thin pencil-like calcification of the intima of the ascending aorta is characteristically seen late in the disease, but it can be superimposed by thicker atherosclerotic calcifications. Aneurysms may form that are most commonly saccular and involve the ascending aorta and the aortic arch, and also, more rarely, the sinus of Valsalva or the pulmonary artery. Polyarteritis nodosa is characterized by small saccular aneurysms of 1-5 mm size, which can sometimes be detected by thin-section CTA. Aneurysms can be found in the intrarenal or intrahepatic branches on early arterial phase scans (only possible with multislice scanning) before there is substantial parenchymal enhancement. Aneurysms in mesenteric side branches are more easily detected in (moderately) obese patients. The typical luminal irregularities and stenoses of small arteries are
very hard to detect on CTA. In the kidneys multiple cortical infarcts;may be present.
(c).
Neoplasia
Aortic tumors are rare. The majority are mesenchymal tumors, most notably leiomyosarcoma, angiosarcoma, or malignant fibrous histiocytoma. CT Morphology
Aortic tumors may present initially as proximal occlusions of the aortic branches in cases where tumor growth is confined to the aortic wall. Nodular lesions of soft-tissue density in the aortic wall extending into the para-aortic tissue can suggest the correct diagnosis. The differential diagnosis includes retroperitoneal fibrosis (Ormond disease), inflammatory aneurysms, and aortitis. These conditions are much more common than sarcomas.
891
24 Vascular System
The scan range with single slice CTA is limited but evaluation of focal pathology such as stenosis of the carotid bifurcation or evaluation of thoracic outlet syndrome is nevertheless possible. Multislice CTA offers advantages whenever large ranges have to be covered, such as for the evaluation of the carotid arteries from the aortic arch to the intracranial circulation, which now also allows for the diagnosis of tandem stenoses. Anatomic Variants
Anatomic variants of the supra-aortic branching pattern have also been described above (p. 870). A retroesophageal course of the left or right subclavian artery can cause dysphagia lusoria. Knowledge about a direct origin of the vertebral artery from the aortic arch may be helpful for selective catheterization. An abnormally high or low carotid bifurcation as well as the presence of massive carotid kinking or coiling may be important for preinterventional planning. •
CT Morphology
An aberrant subclavian artery and other variants can be identified by tracing the vessels from the aortic origin into the neck or axillary region. In many cases the vessels display a typical arrangement that can be recognized even on noncontrast scans (see Fig. 24.41). The common carotid artery bifurcates between the level of C2 and C4. Abnormally caudal
or cranial branching patterns are well appreciated on curved reformations or thin-slab MIP and are important to detect prior to surgery. Kinking is due to elongation ofthe carotids leading to sharp bends within the vessels (Fig.24.67a), while coiling is an even more massive elongation that causes a full rotation of the vessel, usually in the distal extracranial portion of the vessel. Kinking of the vertebral artery may lead to bony erosion (Fig.24.67b) due to pulsation effects. Such findings are easily overlooked on axial images. Patients are at increased risk of vascular damage during chiropractic manipulations. Carotid Stenosis
Carotid artery stenosis is a common complication of atherosclerosis. Higher grades of carotid stenosis are associated with an increased risk of stroke. Large studies (NASCET, ECST) could show that symptomatic patients with a stenosis benefit from surgical endarterectomy. Even asymptomatic patients with carotid stenosis have been shown to benefit from surgery: the ACAS trial (1995) found a reduced stroke rate of 1% with surgery as compared to 2% with medical treatment but had less influence on clinical practice than the NASCETtrial. The area of the carotid bifurcation is most commonly affected. The stroke risk is increased in patients with ulcerated plaques that facilitate formation of thromboemboli. The size of infarc-
Fig. 24.67 Bilateral carotid artery coiling (4 x 1/6, VRT) (a). Bone erosion by vertebral artery kinking (4X1/6,VRT) (b). a
b
Carotid Stenosis 893 tions depends on intracranial collateral circulation, which may be impaired if portions of the circle of Willis are hypoplastic or stenosed. Single and multislice CTA are excellent for detecting carotid stenosis in the region of the carotid bifurcation and assessing the need for surgical treatment, but the proximal and intracranial portions of the carotid arteries can only be adequately evaluated with multislice CTA. CTA is in direct competition with color duplex ultrasound and MR angiography as a minimally invasive alternative to diagnostic angiography. •
CT Morphology
Because the carotid arteries run perpendicular to the scan plane, maximum opacification is unnecessary for their evaluation. An interactive review of the axial images is still important, however, because MIPs and SSDs may be seriously degraded by superimposed calcified plaques. Thin-slab MIPs and curved planar reformations (CPR) with a slice thickness of 3-10 mm are best for documentation purposes (see Fig. 24.69). Quantification of stenoses is most precise if transverse sections or curved planar reformations are generated perpendicular to the longitudinal axial of the artery. Such CPR are especially important in the bony segment of the carotids, where most other visualization techniques fail. Precise positioning of the centerline of the CPR within the vessel is essential for stenosis quantification because an eccentric cut may simulate stenosis. Vessel tracking may help to automatically track the vessel course but it usually fails for the petrous segment of the carotid arteries. Stenosis quantification in CTA is usually based on the (angiographic) NASCET criteria, which use the diameter of a normal distal segment of the internal carotid artery as a reference for calculating the degree of stenosis (Fig. 24.68). A 70% cutoff is employed for distinction between significant and non-significant stenosis but newer results (Cochrane Reviews 2000) suggest that even 50% stenoses benefit from endarterectomy. The ECST criteria for stenosis quantification are based on sonography and use the whole diameter of the carotid lumen (including any plaques) at the site of stenosis as a reference for calculating the degree of stenosis (Fig. 24.68). According to new results, a 70% cutoff is employed for distinction between significant and non-significant stenosis.
Carotid stenosis (%) NASCET:
100-(a/dx1 00)
ECST:
100-(a/bx1 00)
Fig. 24.68 Quantification of carotid using NASCET and ECST criteria.
artery
stenosis
Fibromuscular dysplasia of the carotid artery is a very rare entity and can be suspected in a younger individual if typical beaded stenoses or formation of small aneurysmatic dilatations of the vessel can be observed. CT attenuation is a good predictor of plaque composition (Fig.24.69a); lipid plaques have the lowest density (below 50 HU), fibrous plaques have densities between 50 HU and 150 HU, and calcified plaques have the highest densities (over 150 HU, often more than 1000 HU). Small plaques, however, are susceptible to partial volume effects. Ulcerated plaques present as marked irregularities on the surface of a soft plaque (Fig.24.69b). Calcified plaques make evaluation of stenoses difficult unless transverse sections or curved planar reformations are generated and a wide window setting is used. Concomitant aneurysmatic dilatation of the carotid (Fig.24.69d) may increase the risk of stroke if there is irregular thrombus formation within the aneurysm sack. It is important to evaluate not only the region of the carotid bifurcation but also more distal segments, as well as the petrous segment of the carotid artery, because such distal segments are hard to evaluate by ultrasound (Fig. 24.70). The same holds true for stenoses close to the origin of the common carotid artery. Tandem stenoses are present if there are stenoses outside the carotid bifurcation as well, usually located either at the origin of the common carotid artery or in the distal or intracranial segments of the carotids (Fig. 24.70).
894
24 Vascular System
a
b
c
fibrous plaque, (b) High-grade carotid artery stenosis with an ulcerated plaque (4 x 1/6, CPR). (c) long subtotal stenosis of the ICA involving also the petrous portion of the vessel (d) Thin-slab MIP of an aneurysm and stenosis of the internal carotid artery.
Fig. 24.69 Internal carotid artery (ICA) stenoses, (a) The plaque density may provide a clue about plaque composition. In this patient there are hypodense «SOH U) regions considered to be lipid plaques and denser regions (50-90 HU) considered to represent
a
b
c
Fig. 24.70 Findings outside the region ofthe carotid bifurction (4 x 1/6, CPR). (a) Tandem stenosis ofthe bifuraction (arrowheads) and the petrous segment of the ICA(arrows). (b)Corresponding axial sections, (c)Stenosis at the origin ofthe left carotid artery, (d) Occlusion of
d
d
e the proximal right common carotid artery, (e) Incidental finding of an intracranial aneurysm and hypoplasia of the A1 segment of the left anterior cerebral artery (VRT).
Stroke Imaging
Associated intracranial abnormalities include intracranial stenoses of the main cerebral arteries as well as aneurysms or hypoplasia of these vessels (Fig.24.70e). Y
present as multiple areas of focal narrowing or aneurysmal dilatation. Similar findings can be seen in vertebral artery dissection, but spatial resolution may become a limiting factor for this indication.
Spontaneous Carotid Dissection
Internal carotid artery dissections occur in young or middle-aged patients and are a major cause for stroke (5-20%) in the young population. Some 70% involve the cervical and petrous portions of the vessel. while 20% affect the cervical part only. In spontaneous (non traumatic) dissections. there is either no precipitating factor or they can occur after trivial trauma such as rapid head turning or during normal sporting activity. The pathogenesis is usually unclear. but underlying arteriopathy like hypertension. fibromuscular dysplasia ("15-20% of cases). cystic medial necrosis. Marfan syndrome. or Ehlers-Danlos syndrome may predispose to dissection. Patients present with symptoms of headache. neck pain, vascular bruits, Homer's syndrome, or focal neurological deficits. Stroke is often delayed for days to weeks when the dissection remains extracranial. The most common extracranial site is the internal carotid, a few cm beyond the bifurcation. Anticoagulation therapy is usually effective, and 85% of patients with minor symptoms do well and only 5% suffer severe strokes. Refractive cases may require surgery, such as an extracranial- intracranial bypass. Vertebral artery dissections are half as common as those involving the internal carotid artery. They are less common in the mid cervical region where the artery is protected by the transverse processes, and usually occur around the CI-C2 region (two-thirds of cases). In up to 20% there may be asymptomatic dissections in other arteries as well. •
CT Morphology
Carotid dissection may present as a smooth, long stenotic segment that may extend intracranially ("string" sign, see Fig. 24.91). Usually the false channel is thrombosed and may appear hyperdense with a crescent shape. A thin rim of enhancing vasa vasorum may be seen around the intramural hematoma. If the false channel is perfused, diagnosis is easy with CTA. Fibromuscular dysplasia as a causative factor may
Stroke Imaging Stroke is the third leading cause of death in most industrial countries. Acute cerebral ischemia accounts for more than 75% of cases. Hemorrhagic brain infarction and acute intracerebral hemorrhage are responsible for about equal numbers of the remaining patients. New therapeutic concepts based on early fibrinolytic therapy of patients with ischemic stroke require a short time to definite diagnosis for optimum therapeutic results. This can be achieved by SPECT, PET, xenon CT, or MRI but probably as well (or even quicker) with the following combination ofCT techniques: 1. Precontrast CT; 2. CT perfusion imaging (CTP) of the brain; 3. CTA of the carotid arteries including the brain. The precontrast scan is used to rule out hemorrhage or signs of manifest infarction. The perfusion scan allows for distinction between infarcted regions and hypo perfused areas (penumbra) that may recover after fibrinolytic therapy. The CTA can detect potential causes for the acute ischemia, such as a high-grade stenosis, an ulcerated plaque, or a (partially thrombosed) aneurysm. In addition, demonstration of an occluded large intracranial vessel suggests that the patient may benefit from intra-arterial thrombolysis. By demonstrating normal perfusion CTP can help to exclude stroke mimics like TIA. migraine, or seizures. For this reason, CT may become an important part of the work-up of hyperacute stroke (3-6 h after onset). CT Morphology
Hyperattenuating areas on the precontrast CT indicate hemorrhage or hemorrhagic brain infarction. No further examinations need to be performed because fibrinolytic therapy is not an option for these patients. The classic signs of ischemic infarction, such as a well-demarcated hypoattenuating region with a CT attenuation slightly below that of normal brain substance, as
895
24 Vascular System
well as loss of gray-white differentiation or locally decreased gyration also suggest areas that are infarcted and also do not profit from lytic therapy. Normal or near-normal findings on the precontrast scan warrant a pel-fusion scan. The dynamics of contrast enhancement are acquired at a representative level within the brain. The level usually just cranial to orbital roof usually includes parts of the anterior, medial, and posterior cerebral arteries, as well as the basal ganglia. The thickest possible sections are used (10 mm for single slice CT, and 2 x 10 mm or 4 x 8mm for multislice CT). Automated programs . for CT perfusion imaging can calculate regional cerebral blood volume (CBV), cerebral blood flow (CBF), mean transit time (MTT), and timeto-peak (TTP). The normal CBF is 50-60 mlilOOg tissue/ min. CBF is reduced in neonates, but increases during early childhood (peaks age 7) and then lowers again until adult values are reached during adolescence. Dysfunction is present if CBF falls below 20mlll00g/min. Neuronal death has been reported to occur if CBF is lower than 10-12mlilO 0 g/ min. SPECT data would show that there is an increased risk for hemorrhage following thrombolysis if the CBF in the ischemic core falls below 35% of normal cerebellar flow. Lack of blood volume and blood flow indicates an area of manifest infarction. Reduced blood volume and blood flow suggest lack of sufficient collateral supply, which very likely
a
b
Fig. 24.71 Perfusion imaging in a stroke patient with a near normal precontrast scan 2 hours after the incident (a). The map of mean transit time (MTT) best shows the affected area (b). The cerebral blood flow map (CBF)
will progress to manifest infarction, while regions with near-normal blood volume but reduced blood flow can be found in areas with good collateral supply (penumbra) that will most probably benefit from fibrinolytic therapy (Fig. 24.71). A similar pattern, however, may be found in patients with chronic cerebral ischemia, although reduction in blood flow is not common in these patients although the arrival time is prolonged. CBV images are a better predictor of the final minimum infarct size, while CBF and MTT images overestimate the size of the infarction. Demonstration of the occlusion of a large intracerebral vessel (either on axial source images or thin-slab MIP) suggests that the patient may profit from intra-arterial thrombolysis if the acute event occurred less than 6 hours earlier (Fig. 24.71). Evaluation of the extracranial circulation has been described above. Thoracic Outlet Syndrome
Thoracic outlet syndrome describes a compression of nerves, veins, and arteries along the course from the chest to the arm. The most frequent congenital causes are a widened or abnormal insertion of the anterior scalene muscle (scalene syndrome), followed by a cervical rib (only 5-10% of patients have symptoms), an unusually straight course of the first rib with narrowing of the costoclavicular space, or a scalenus minimus muscle extending from the trans-
c
d
demonstrates a corresponding hypoperfused zone (arrowheads, c) that is larger than the region with reduced cerebral blood volume (d), which is at increased risk for permanent ischemia.
Thoracic Outlet Syndrome verse process of the seventh cervical vertebra to the first rib, inserting between the brachial plexus and the subclavian artery. Acquired causes are a supraclavicular mass or lymphadenopathy, fracture of the clavicle with exuberant callus formation, or muscular compression of the artery in the pectoralis minor tunnel. Patients present with pain in hand or forearm that increases with elevation of the arm. Paresthesias of the hand are very common. A hyperabduction maneuver will frequently obliterate the radial pulse. A Raynaud phenomenon may be present in up to 40 % of patients. CTA may be performed instead of angiography to demonstrate the site of compression. To be most effective, however, scans in a neutral position and after appropriate postural maneuvers are required. The maneuver that induces most symptoms orchanges in pulse wave or arterial pressure should be used. The Adson maneuver is usually most successful as it tests the compression by the anterior scalene muscle; the patient holds a deep inspiration while the neck is fully extended and the neck is turned to the ipsilateral and contralateral side. The costoclavicular maneuver tests the compression between the clavicle and the first rib and uses an exaggerated military position with the arms at 90° and the shoulders pressed back. The hyperabduction maneuver tests the compression by the humeral head and pectoralis minor muscle while the arm is moved from a neutral position to 180° over the head.
•
CT Morphology
CT angiography demonstrates a focal stenosis or occlusion of the subclavian artery during the proper postural maneuver (Fig. 24.72). One should watch out for mural thrombus (which increases the risk of distal embolization), poststenotic dilatation, or even aneurysms of the subclavian artery, and associated venous obstruction or even thrombosis. Obstruction of the subclavian vein is a frequent incidental finding during intravenous injection of the contrast material and is characterized by substantial flow of highly contrasted blood from the injection vein into cervical and chest wall collaterals while there is marked narrowing or even obstruction of the venous lumen as it crosses between the clavicle and the rib.
Fig. 24.72 Thoracic outlet syndrome. Compression of the subclavian artery during postural maneuver (4 X II 6).
JVbdominal and Peripheral Arteries
Thin-section CTA is necessary for the adequate evaluation of the abdominal aortic side branches. This limits the examination volume with single slice scanning. Multislice CTA allows for excellent displays with near-isotropic resolution while being able to cover the whole chest
or abdomen in a single breath hold. CTA is able to substitute almost completely for diagnostic arterial angiography for the evaluation of aortic side branches but color duplex ultrasound and (Gd-enhanced) MRA are potent competitors.
897
24 VascularSystem Anatomic Variants
Anatomic variants are incidental findings in CT examinations of the abdomen. They assume importance when they cause symptoms or have a bearing on planning of surgery or interventional radiological procedures. 1 , Anatomic variants of the arterial supply to the liver are found in up to 50% of patients. They are typical incidental findings in arterial phase CT examinations of the abdomen. They become diagnostically relevant for planning partial liver resection or interventional radiological procedures. Variants of the superior mesenteric artery, other than a common origin with the celiac artery, are exceedingly rare. Superior mesenteric artery syndrome is caused by compression of the duodenum by the SMA in thin individuals. The nutcracker syndrome is due to compression of the left renal vein and may present as painless hematuria in slim patients. •
CT Morphology
Variants of the upper abdominal arteries can usually be detected on standard arterial phase scans that are part of a biphasic or multiphasic study of the upper abdomen. The radiologist should be alert for the following variants of the arterial supply to the liver: • Branches of the celiac trunk arising separately from the aorta (Fig. 24.73 a).
a
b
Fig. 24.73 Anatomic variants of the hepatic blood supply, (a) All three branches of the celiac trunk arise from the aorta, (b) Accessory right hepatic artery (here
• A vessel running posterior to the portal vein (Fig.24.73b) may be an accessory right hepatic artery, a replaced right hepatic artery, or a replaced proper hepatic artery. Accurate identification requires tracing the vessel into the liver and toward its origin (from the superior mesenteric artery, the aorta, or directly from the celiac trunk) . • A vessel coursing between the caudate and left lobes of the liver superior to the porta hepatis may be an accessory or replaced left hepatic artery arising directly from the left gastric artery (Fig. 24.73 c). • An accessory right hepatic artery that arises from the gastroduodenal artery presents as a second vessel just below the right hepatic artery. Such vessels are easily missed and require thin-section imaging. They are best seen on thin-slab M1P (1-2 ern thickness) parallel to the course of the proper hepatic artery. A good overview of major findings is gained by coronal thin-slab M1P of some 5-6 cm thickness that are placed immediately anterior to the aorta and tilted slightly so that there is a selective display of the celiac and mesenteric branching pattern (Fig. 24.74). Volume-rendered displays are a good alternative that even allows for semitransparent display of the surrounding tissues. Atypical branching patterns of the hepatic arteries are important for evaluation of living
c from the superior mesenteric artery) showing a typical retroportal course, (c) Accessory left hepatic artery arising from the left gastric artery.
Living Renal Donors
Living Renal Donors
Fig. 24.74 Thin-slab MIP of a displaced right hepatic artery from the SMA and an accessory left hepatic artery from the left gastric artery (a).
hepatic donors. In particular, the presence of multiple arteries supplying the right or left lateral segments and segment IV are of importance in deciding about a suitable donor. Segment IV can receive branches from the right or left hepatic arteries and directly from the hepatic arterial bifurcation, which thus becomes a trifurcation. High flow rates and thin-section imaging are required because this imaging task pushes the limit of spatial resolution with CTA. Accessory renal arteries occur in up to 25% of individuals. Usually one vessel is dominant on either side. However, multiple small vessels may be present, and accessory arteries may originate even above the superior mesenteric artery (exceedingly rare) or the iliac vessels (some 1% of cases). Main renal arteries enter the kidney via the hilum, while polar arteries enter the renal parenchyma directly. Such variants are important for interventional radiological dilatation of renal artery stenosis and for the work -up of living renal donors. The gonadal arteries usually arise from the aorta just below the level of the kidneys but rarely may originate from the renal artery (more common on the right) . Early branching of a normal sized artery or multiple or abnormal arteries or veins may make harvesting a kidney from a living renal donor difficul t. / In slim (young) patients, the duodenum or the left renal vein may be compressed between the aorta and the superior mesenteric artery leading to superior mesenteric artery syndrome or nutcracker syndrome, respectively.
The demand for renal transplantation is growing rapidly and has led to an increasing number of living-related donors. Advances in surgery now allow for minimally invasive explantation techniques with low morbidity and shortened hospital stays for the donor. Precise preoperative planning is mandatory, however, because the overview of the intraoperative situs is limited and unexpected findings, such as accessory vessels, may cause major complications. CT has become the imaging technique of choice for evaluating potential renal donors but MR may assume an increasing role in the future. CT must rule out pre-existing kidney disease and should precisely map the preoperative vascular anatomy. It is used to decide which kidney to explant and to plan the surgical procedure. •
CT Morphology
The examination consists of a pre-contrast CT scan to rule out calcifications and calculi, aCTA to demonstrate the arterial (and venous) anatomy, an nephrographic phase scan to exclude parenchymal disease and tumors, and a KUB radiograph immediately afterward to demonstrate the morphology of the renal pelvis and the course of the ureter. The pre contrast and nephrographic phase scans can use less dose and thicker sections than the CTA, which should have a high spatial resolution to demonstrate even small accessory vessels. The main criteria for evaluation are listed in Table 24.15. It is important to rule out concomitant parenchymal disease, tumors, cysts, and renal calculi. CTA is used to evaluate whether a normal arterial and venous anatomy is present, and to describe all anatomic variants that may lead to surgical complications. The extrarenal branching pattern of arteries and veins (length of common segment) should be noted, and the distance and length of accessory vessels relative to the main arteries or veins should be measured. It is important to distinguish between main renal arteries that enter the renal hilum and polar arteries that directly enter the parenchyma (Fig.24. 75a). A high spatial resolution is needed to demonstrate gonadal or suprarenal arteries that may arise from the renal arteries. Diseases of the renal arteries, such as stenosis, fibromuscular dysplasia, or aneurysms have to be ruled out.
899
900
24 Vascular System Table 24.15
Checklist for living renal donors
Renal arteries
Accessory arteries Renal artery stenosis or aneurysm Early branching Atypical course Gonadal arteries
tonal veins Accessory veins • Early branching • Retroaortic/circumaortic renal veins Gonadal and lumbar veins Renal parenchyma
• Atypical position, congenital anomalies (e.g. horseshoe) • Asymmetric size • Signs of chronic parenchyma! damage • Tumors Collecting
system
• Renal calculi/calcifications' • Ureteral anomalies • Tumors
On the venous side, the presence and number of lumbar veins that drain into the renal vein (mainly present on the left) as well as adrenal veins should be documented (Fig.24.75b). Noting the location, size, and potential duplication of gonadal veins helps to simplify the surgical procedure and reduces the risk of postoperative hemorrhage.
Renal Artery Stenosis
Renal artery stenosis (RAS) is an uncommon cause of arterial hypertension (approximately 5%). Because it is a potentially curable condition, early detection is important. Clinical symptoms may appear at a young age and consist of severe hypertension, hypertensive crises, disturbance in diurnal rhythm, a rapid rise in blood pressure within a few months, progression of hypertension following successful medical therapy, and an increase of retention parameters in response to ACE inhibitors. The commonest cause of RAS is atherosclerotic disease (65-70%). Angioplasty has been considered the treatment of choice for these hypertensive patients but newer longerterm studies from the Netherlands suggest that optimum medical treatment is equally effective. There is also an ongoing discussion about the positive effects of angioplasty or stenting on the preservation of renal function in these patients. Thus, the final role of noninvasive techniques (CTA, MRA, ultrasound) for diagnosis of renal artery stenosis will depend on whether interventional therapy is considered or not. Fibromuscular dysplasia is the next commonest cause of renal artery stenosis (some 25%). It affects younger patients and is more prevalent in women (3: 1). Angioplasty is usually a very effective treatment. Six separate forms have been distinguished (Table 24.16), of which medial fibroplasia is commonest (601~%).
. .....
The sensitivity and specificity of CTA are well above 90% for the detection of significant
Fig. 24.75 Important findings during work-up of living renal donors, (a) Small polar artery entering the parenchyma directly. (b) Retroartic left renal vein with branching prior to entering the IVC. a
b
Renal Artery Stenosis Table 24.16 Classification of the renal arteries
of fibromuscular
dysplasia
1. Intimal fibroplasia (1-2%) Circumferential band, poststenotic dilatation main + segmental arteries, often bilateral, children
+ young
adults
CT Morphology
2. Medial fibroplasia (60-700/0) String of beads and aneurysms, branches
=
alternating areas of stenoses' mid + distal renal arteries +
3. Medial hyperplasia (5-15%) Long smooth tubular arteries and branches
narrowing,
main
rena!"
4. Perimedial fibroplasia (20%) Long irregular stenosis, beading, no aneurysms (width" unaffected segments), distal main renal artery
5. Medial dissection (5-10%) False channel in media, aneurysm main renal artery + branches
formation
6. Adventitial fibroplasia «1 %)
~
and patient-related factors (obesity, bowel gas). Its diagnostic yield is substantially reduced under unfavorable conditions. Scintigraphy with ACE inhibitors may yield additional functional information but is inferior in bilateral disease.
I
Long segmental stenosis due to periarterial proliferation offibrofatty tissue, main renal artery large branches
Indirect signs can already be appreciated on standard CT examinations. Differences in size (right kidney 2 cm smaller than left kidney, or left kidney 1.5cm smaller than right kidney). cortical thickness (on corticomedullary phase images), level of enhancement, and time to excretion between the two kidneys suggest unilateral significant renal artery stenosis. Effective interpretation of the actual CTA examination is based primarily on 3D views such as volume-rendered displays (VRT) or curved thin-slab MIP (Figs. 24.76, 24.77; see also Fig. 24.11). While VRT are superior for the main renal arteries, MIP yield better results for small intrarenal branches. In order not to miss eccentric plaques, anteroposterior views as well as caudocranial views should be used. For optimum display, multiple 3D images should be generated that are rotated in 10 steps around the longitudinal and a horizontal axis. Whenever a suspicious finding is encountered, it should be confirmed by cross-sectional images. This can either be done by interactive analysis of axial sections in a cine display, or by generating curved planar reformations (CPR) through the vessel segment of interest (see Fig. 24.7). Cross-sectional images are mandatory whenever calcified plaques are present. In such situations, a wide window setting is necessary not to overestimate a stenosis. If the crosssectional images are disregarded and only the MIP are considered, the diagnostic accuracy of CTA declines to 80-90%. The accurate quantification of renal artery stenoses is very hard due to partial volume effects, but CTA has proven accurate in the grading of stenoses on a five-point scale; no stenosis (grade 0, <30%), nonsignificant (grade I, 30-50%), moderate (grade II, 50-70%). highgrade (grade III, 70% to < 100%), and occlusion (grade TV). Only in very rare cases will CTA overor underestimate stenoses by more than one grade. Poststenotic dilatation is a valuable indirect sign of a significant (high-grade) stenosis (Fig.24.76a). Collateral vessels also indicate a (long-standing) high-grade stenosis or an occlu0
stenoses. For fibromuscular dysplasia sensitivity is lower at 75-80% for single-slice CTA. Given the low prevalence of renal artery stenosis. CTA should not be used for the indiscriminate screening of all hypertensive patients. In individuals with a high index of clinical suspicion, arterial angiography should be considered as the first-line modality, because it can be followed immediately by angioplasty or stent placement in patients with significant stenoses. CTA. however, may improve planning of intervention even in such patients because it allows for more subtle characterization and localization of the stenosis. CTA is an effective screening tool in patients who have low to moderately high clinical suspicion of renovascular hypertension. Because of the very high negative predictive value (above 95%), a normal CTA virtually excludes a significant stenosis, making further investigations unnecessary. Competitive modalities are Gd- MRA and color duplex sonography. Both procedures yield results comparable to those of CTA with optimum technique and may be performed in place ofCTA. Color duplex ultrasound, however, strongly depends on the skills of the examiner
901
24 Vascular System
902
a
b
Fig. 24.76 Single slice eTA (2/3/1) of renal artery stenosis. (a) High-grade stenosis of the left renal artery. Chronic parenchymal damage accounts for the extended peripheral visualization of the segmental arteries. (b) Eccentric soft plaque in the right renal artery thatwas poorly defined by angiography. (c) Axial MPR in a patient with fibromuscular dysplasia. (d) Stenotic kinking of the renal artery following renal transplantation.
sion. Opacification of the renal artery distal to an occlusion is typical because of the long scan duration that causes enhancement of all perfused arterial structures on CTA. With multislice techniques, an even more precise quantification may be performed using 10% steps. Automated techniques for vessel analysis can aid in stenosis quantification but results should always be checked on sections perpendicular to the course of the artery because image noise or a renal vein that is in immediate contact to the artery may simulate focal stenoses or aneurysmatic dilatation. Calcified hard plaques are distinguished from hypoattenuating soft plaques that are detectable within the course of a, vessel (Fig.24. 76b). Generally it is necessary to image the vessel in a second plane due to the possibility of an eccentric stenosis (Fig.24. 76b). The distinction between and ostial and truncal stenoses may be of importance for planning angiographic interventions. CTA is superior to angiography for making such a distinction. An ostial stenosis often appears only as a thin, hypoattenuating band in continuity with the aortic wall or may be caused by an aortic plaque extending into the renal artery ostium. A truncal stenosis is localized in the renal artery distal to the ostium. Fibromuscular dysplasia may present as a unifocal stenosis in the middle and distal third of the main renal artery (Fig.24. 77a), multifocal stenosis with string-of-pearls appearance (Fig.24. 76c), or aneurysm formation. The right renal artery is most commonly affected but bilateral disease is seen in up to two-thirds of cases. It is characteristic that the proximal third of the main renal artery is spared (in over 95%). Renal artery branches are rarely affected without concomitant disease in the extrarenal arteries. Such cases of isolated intrarenal fibromuscular dysplasia are extremely hard to detect even with multislice techniques. Thin-section techniques are mandatory for correctly classifying the disease, and cross-sectional images often provide better detail than the various 3D-rendering techniques (in particular MIP). Peripheral stenoses are difficult to detect unless there is a good contrast between the opacified vessels and the renal parenchyma. This is rarely the case with single slice CTA (because of the longer scan duration) but can almost always be achieved with multislice scan-
Renal Artery Stenosis
a
b
c
d
Fig.24.77 Multislice eTA (4 x 1 /6) of renal artery stenosis. (a) Fibromuscular dysplasia with stenosis and aneurysm in a young woman, (b) Multiple renal emboli (arrow heads) and truncal stenosis of an accessory renal artery (arrow), (c) Anterior origin with stenosis of an ac-
cessory artery that was considered normal on regular 15° LAO projection during angiography. (d) Two stenosing soft plaques (arrows) in the sole remaining renal artery of a patient with chronic Leriche syndrome.
ning. Thin-slab MIP or thick MPR (5-10 mm thickness) are the techniques of choice for evaluating these vessels. One has to be careful not to mistake pulsation effects or image noise for real stenoses. Artifacts may be due to a large number of factors, the most important of which are partial
volume effects (too thick a collimation), insufficient contrast enhancement (wrong timing, slow flow rates), superimposing renal veins (too long a start delay), or the display technique (see above, p. 848). Insufficient breath holding may simulate a stenosis or aneurysm on multi planar reformations or 3D images (Fig. 24.78). Such
a Fig. 24.78 Respiratory artifacts. A renal artery aneurysm simulated by inspiration during the scan is characterized by "dilatation" only along the z-axis (an-
b teroposterior MIP, a), and not in the scan plane (b). Stenoses may be simulated by expiration during the scan (c).
903
24 Vascular System
breathing artifacts can be detected by checking for jagged renal contours at the level of a detected stenosis. eTA is superior to arterial angiography for the quantification of eccentric stenoses (unless 3D rotational angiography is performed). for distinguishing between hard and soft plaques. and for differentiating ostial from truncal stenoses. eTA can thus help to decide whether to perform angioplasty (PTA) or primary stenting (hard plaques. ostial stenoses). eTA can demonstrate an anterior origin of a renal artery from the aorta and thus may detect an ostial or truncal stenosis of this artery that can be missed on angiography because of superimposition by the aorta or due to a malposition of a selective catheter distal to the stenosis. eTA is a valuable tool for the preinterventional assessment of vascular status. as it can reliably demonstrate variants such as accessory vessels or an unusual anterior or posterior shift in vascular origins (Fig.24.77c). eTA has poor accuracy in the detection of intrarenal stenoses. but treatable lesions can be adequately evaluated prior to PTA.
Celiac and Mesenteric Stenosis
Abdominal angina is a postprandial
abdominal pain some 15-20 minutes after food ingestion. It is considered to be due to gastric steal. which diverts blood from the intestine. This intermittent mesenteric ischemia is caused by severe stenosis of splanchnic arteries without adequate collateralization. Patients have a food aversion. complain of weight loss. and can develop malabsorption or bowel strictures. Stenosis of the celiac artery is a common incidental finding that is usually caused by a transverse ligament spanning the left and right crus of the diaphragm. just below the level of the thoracoabdominaljunction. This stenosis is typically exacerbated by inspiration and causes no symptoms. Association with an increased rate of pancreatitis or complications of the liver transplantation has been discussed. Stenosis of the superior or inferior mesenteric artery are also relatively frequent incidental findings in patients with severe arteriosclerosis. They are asymptomatic unless adequate collateral supply can no longer be maintained between the superior and inferior mesenteric ar-
teries through the Riolan anastomosis or between the superior mesenteric artery and celiac artery through gastroduodenal vessels. eTA is well suited for the detection of pro ximal stenoses and associated collateral pathway formation. It is less accurate in detecting the peripheral stenoses but accuracy with thin-section muItislice scanning has improved a lot. With its ability to demonstrate vascular calcifications, eTA is an effective imaging procedure when used prior to surgical revascularization of patients with chronic mesenteric ischemia. It is important however, to use a negative contrast agent such as water for bowel distension. CT Morphology
In abdominal angina patients, combined severe stenoses of at least two of the three vessels supplying the bowel (celiac, superior, and inferior mesenteric artery) are present. More rarely, successive proximal and peripheral stenoses may be found in such patients. A lateral MIP or VRT (the width of the VOl covers only the aorta) is best for demonstrating proximal stenoses of the celiac trunk and superior mesenteric artery (Figs.24.79a, 24.80b). Coronal thin slab MIP of some 3-6 cm width that are placed parallel but anterior to the aorta provide the best anatomic overview (see Fig. 24.74). Volume-rendered views are a good alternative (Fig.24.79c) to demonstrate more peripheral stenoses as well as collateralization via the arc of Riolan. Dilated collaterals display a typical pattern on axial sections (Fig.24.79b). Stenoses due to diaphragmatic trapping can be diagnosed in a celiac trunk that runs straight downward from its aortic origin with an elliptical distortion of the normally round cross-section of the artery. The trunk displays a normal diameter in the AP projection but shows marked narrowing on lateral projections.
Acute Mesenteric Ischemia Acute mesenteric ischemia is a life-threatening event that can be caused by embolic disease, thrombosis of a stenosed vessel segment in atherosclerosis, aortic dissection, vasculitis, direct trauma, intravascular coagulation, hypoperfusion due to shock, hypovolemia, or endotoxins. Venous ischemia may be due to thrombosis of the mesenteric veins, usually in
Acute Mesenteric Ischemia
a
c
b
Fig. 24.79 Chronic mesenteric ischemia with stenosis of the celiac trunk and an occlusion of the superior mesenteric artery (SMA) have induced collateralization through a Riolan anastomosis (4 x 1/6, VRT) (a). The
stenoses are well seen on a lateral VRT (b). Note the occluded stent in the SMA. The mesenteric vessels are best demonstrated by a MIP image from which the aorta has been removed (c).
a
c
b
Fig. 24.80 Stenosis of the celiac trunk due to diaphragmatic trapping results in a typical elliptical distortion in its proximal portion (a) and a U-shaped course on the
lateral VRT (b). Celiac artery occlusion in an asymptomatic patient (4 x 1/6, VRT) (c).
young patients after surgery. Strangulation of the mesenteric vessels by volvulus and various other conditions has been discussed in Chapter "15,Gastrointestinal Tract. Mesenteric angiography used to be the first -line technique for demonstrating suspected acute mesenteric ischemia but the increased spatial resolution and its ability to detect alternative diagnoses has made multislice CTA a good alternative in many patients. Single slice
CTA, however, is less suited because it may miss more peripheral lesions. •
CT Morphology
Pre-contrast scans can directly demonstrate a hypodense acute intravascular thrombus. CTA will show the site of obstruction and any residual perfusion distal to the obstructing lesion (Fig. 24.81). Good patient cooperation with no breathing artifacts is mandatory for good results
905
24 Vascular System
Fig. 24.81 Acute mesenteric ischemia with a fresh embolus in the superior mesenteric artery (arrowheads).
in the periphery. In patients with vasoconstriction the arterial vessels are extremely small, and even the proximal portions of the superior mesenteric and celiac arteries may be involved. Associated findings include thickened bowel wall due to edema, reduced bowel opacification as compared to other segments, and gas within the bowel wall (pneumatosis, see Fig. 15.35) or in the mesenteric or portal veins indicating infarction. Ascites may be present in the region of the affected bowel. Focal or diffuse bowel dilation is a nonspecific sign. Involvement of the splanchnic vessels is a potentially life-threatening complication in patients with aortic dissection. Compression of the true lumen by a dilated false channel may occur in the aorta or if the dissection membrane extends into the superior mesenteric artery. It may result in global or segmental malperfusion with subsequent mesenteric infarction (see Figs. 24.62. 24.63). Peripheral Arterial Occlusive Disease
Single slice CTA is appropriate for localized investigations of the iliac vessels (infected prosthesis, aneurysms, etc.) in cases where color duplex sonography is unrewarding and arterial angiography is problematic, especially in cases where femoral arterial access cannot be established (vascular prosthesis). Multislice CTA has extended the range that can be covered with good spatial resolution from the abdominal aorta down to the feet. Even 1 mm sections are
feasible within a reasonable scan time. Scanning speed is not critical in 4-row scanners but may become an issue with 16-row scanners in patients with unilateral obstruction of the vascular bed and consecutive marked asymmetry in contrast arrival. CTA has a higher spatial resolution than MRA, especially if 1 mm sections are obtained. However, postprocessing is still quite cumbersome and has prevented CTA of the peripheral vessels from entering everyday clinical practice. Patients with massive calcifications, such as those with diabetes or on hemodialysis, are not good candidates for CTA of the peripheral arteries. Modern automated techniques for bone removal or vessel tracking will substantially simplify data evaluation and can make CTA a potent alternative to MRA or DSA. •
CT Morphology
The CTA study is evaluated by the analysis of MIP and curved planar reformations (Fig. 24.82). An interactive review of axial sections requires very fast workstations because of the large number of sections (between 800 and 1200 for multislice CTA, depending on the scanning technique). Hard and soft plaques, aneurysms, and calcifications are easily distinguished. CPR are particularly important to display the vessel lumen in patients with wall calcifications that otherwise superimpose on the lumen on MIP displays. A wide window setting (see Fig. 24.15) has to be used to be able to assess the vessel lumen in calcified stenoses. Aneurysms
Aneurysms of the aortic side branches are occasionally discovered incidentally during CTA or during the arterial phase of an upper abdominal examination. They are particularly common in patients with atherosclerosis, aortic aneurysms (splenic artery), or coarctation of the aorta (subclavian artery). Renal artery aneurysms are frequently congenital and are associated with stenoses. Pseudo aneurysms may form after penetrating trauma, surgery, or as a consequence of pancreatitis (superior mesenteric artery, splenic artery, gastroduodenal artery and side branches).
Aneurysms
907
Fig. 24.82 Peripheral eTA with multiple stenoses and segmental occlusions of the runoff vessels. The best overview is gained by MIP (a), curved planar reformations are best for detailed evaluation of stenoses, especially for eccentric and calcified plaque (b). Note that there is superprojection of calcium over the vessel lumen on the MIP, which can be resolved byevaluating the CPR (c). The stent lumen cannot be securely evaluated with CPR even with a wide window setting (d). Stent occlusion can be expected because of the occlusion of the superficial femoral artery directly proximal to the stent (d). The CPR (b) was created after semi-automated vessel-tracking of all major leg vessels. Note that all vessels are displayed simultaneously by this prototype software (Fleischmann 2002).
a
b
c
d
908
24 Vascular System
a
b
Fig. 24.83 Large renal artery aneurysm with a stenosis (arrow) of the renal artery as it exits the aneurysm (2/3/1, SSD) (a). Multiple mycotic aneurysms (2/3/1,
VRT) (b). These lesions involved numerous abdominal vessels and developed over a three-week period in a
CTA can aid in choosing the optimum treatment (surgery versus interventional radiology) and in planning the best approach to the vessel.
cifications, and they may produce compression effects. On detecting a renal artery aneurysm, the radiologist should assess its relationship to the intrarenal side branches and look for accompanying stenoses, which frequently are intrarenal (Fig.24.83a). MIP views are excellent for the diagnosis of simple aneurysms, but the more complex the anatomy, the greater is the value of 3D reconstructions (Fig.24.83b). Selective removal of the aorta will allow for subtle 3D analysis of the vessels feeding and draining the aneurysm (see Fig. 2.32).
•
CT Morphology
Saccular aneurysms of substantial size are readily detected on axial CT scans. Small aneurysms and fusiform dilatations in regions that contain multiple vessel cross-sections (such as the upper abdomen) are easily missed unless volume-rendered images are used for primary review of multislice CT data sets. The aneurysms may contain mural thrombi and cal-
Evaluation of the venous system is possible with portal phase or delayed phase imaging. It is common to note incidental findings during the course of a CT examination performed for other reasons. Secondary sections are helpful for the optimal depiction of venous anatomy. Anatomic Variants
Asymmetry of thejugular veins is a normal finding.
A persistent left superior vena cava is found in 0.3% of all patients. In most cases the right superior vena cava is also present. It is very rare
patient with fungal sepsis.
to find a left superior pulmonary vein draining into a left superior vena cava. Varianrs of the inferior vena cava (IVC) are found in over 3 % of patients and usually involve the infrarenal segment. An accessory or replaced left inferior vena cava drains into the left renal vein. Azygos continuation is a rare syndrome (incidence < 1 %) based on the aplasia of one segment of the IVe. When the hepatic segment is aplastic, the hepatic veins drain directly into the right atrium. The azygos and hemiazygos veins receive blood from the iliac and renal veins. A retroaortic left renal vein occurs in some 2% of the population as an isolated vessel ,and
Anatomic Variants 909 may be present as an accessory vessel in up to 5%. A circumaortic left renal vein consists of a venous collar around the aorta « 2%). The intrahepatic branch pattern of the portal veins is subject to numerous variations, and so the Couinaud classification (see Fig. 11.1)provides only a general orientation. Mapping of the intrahepatic vascular anatomy can be useful prior to partial hepatic resections. Congenital anomalies of the portal vein include a congenital agenesis of the portal vein or of major portal venous branches, a double portal vein or a prepancreatic portal vein, which is frequently associated with congenital malformations and situs inversus. As a rule, anatomic variants of the venous system are an incidental finding. An awareness of these variants is important because they can mimic pathologic lymph nodes. Moreover, the venous collaterals that develop in response to aplasia of the infrarenal vena cava can simulate a mediastinal mass. CT provides definitive clarification. •
CT Morphology
/lpersistent left superior vena cava appears on CT as an ovoid structure that may show very little enhancement or may enhance intensely, depending on the site of the contrast injection and the phase of the scan. It is located to the left of the aortic arch, anterior to the main trunk of the left pulmonary artery, and drains into the coronary sinus. Confusion with lymph nodes is
a
b
avoided by tracing the course of the vessel (Fig. 24.84a, b). There is a multitude of venous variants in the abdomen that can be distinguished by following the inferior vena cava (or its substitute) upward towards the heart and downward into the pelvis. With duplication of the infrarenal vena cava, the veins ascend on each side of the aorta, the left vein terminating at the level of the renal hilum (Fig. 24.84d). A left-sided inferior vena cava also enters the left renal vein. A retroaorti c left renal vein (see Fig. 24.7 5c) can be isolated or be combined with a normal left renal vein, or with variants of the inferior vena cava. Many retroaortic left renal veins take a downward course before crossing the aorta and entering the lVe. On non-contrast scans it should not be mistaken for a lymph node. Multiple or bifurcated renal veins are quite common and gain importance in living renal donors. A circumaortic renal vein causes a venous collar at the level of the renal hilum. When azygos continuation is present, portions of the inferior vena cava cannot be visualized. The hepatic segment is most commonly involved (Fig.24.84c). The abdominal and especially the thoracic portions of the azygos and hemiazygos veins are markedly dilated and in some cases may exceed the aortic diameter. The anomaly can mimic esophageal varices. The paravertebral and spinal venous plexus are dilated. Portal variants include trifurcation of the portal vein (Fig.24.85a) and a right anterior
c
d
Fig. 24.84 Persistent left superior vena cava (a, b). Azygos continuation with a dilated azygos vein and absence of the intrahepatic inferior vena cava (c, d). Duplication oftheinfravenal IVC (4 x 1/6, coronal MPR) (d).
24 Vascular System Fig. 24.85 Portal variants. Trifurcation of the portal vein (a). A hypotrophied left liver lobe with small left portal vein (b).
b
a
branch ansmg from the left portal vein, or a right posterior branch arising directly from the portal vein. The latter two variants can lead to surgical complications if errors occur during hilar dissection, f
Congenital anomalies of the portal vein have
to be differentiated from acquired anomalies of the portal veins due to atrophy of hepatic lobes. In acquired diseases, there is usually a small remnant of the portal vein in the atrophied segment (Fig. 24.85), while the affected portal venous branch should be absent in congenital disease.
contrast-enhanced scans, fresh thrombi tend to be denser in their central portion. Enhancement of the venous wall may be present (Fig.24.87d). Contrast uptake in an intravenous filling defect should raise the suspicion of a tumor infiltration or tumor thrombus (see Fig. 11.31). As the thrombus gets older, the vessel becomes either recanalized and may retain a walladherent (hypo dense) thrombus, or it remains occlude and begins to shrink in diameter until it ends up as a fibrous band in long-standing occlusion (Fig.24.86b). Collateral formation is an indicator of an older process with venous stenosis. Such collaterals may form varices or even
venous aneurysms.
Thrombosis of Major Veins
Venous thrombosis is a common incidental finding in CT examinations. Primary evaluation with CT is indicated in cases where ultrasonography or venography have not yielded satisfactory results (e.g., in determining the extent of a vena cava thrombus). CT and MRI furnish similar results, but MRI is slightly better for demonstrating thrombi within the lesser pelvis. •
CT Morphology
Fresh venous thrombi are hyperattenuating on noncontrast scans, and become hypoattenuating as they become older and organize (Fig.24.86a). Following contrast administration they appear as filling defects that may occupy the center or periphery of the venous lumen (Fig. 24.87). Fresh thrombi tend to expand the vessel lumen until its maximum capacity is reached, and are only then able to completely occlude it. Even on
The incomplete mixing of opacified and nonopacified blood can mimic a filling defect, creating the appearance of an intraluminal thrombus (see Fig. 7.39). Pseudothrombus artifact is typically observed in the suprarenal inferior vena cava, the proximal superior mesenteric vein, the main trunk of the portal vein, and occasionally in the common iliac vein and common femoral vein. This problem can be avoided by a sufficient delay before initiating the spiral scan. Thrombophlebitis is characterized by the same criteria as venous thrombosis but is combined with edema and stranding in the surrounding tissue (Fig.24.86c). Rarely can gas inclusions be seen in patients infected with gasforming organisms.
Port-of Hypertension
c
b
a
Fig.24.86 Young patient with chronic occlusion ofthe right iliac vein, fever and right lower quadrant pain. There are fresh hyperattenuating thrombi (arrow) in dilated varices (a). The occluded iliac vein shows, fibrous
transformation (arrow, b). Note the stranding and edema (arrowheads, c ) around the affected varicous collaterals.
Deep Venous Thrombosis
3-20 mGy. depending on the examination technique.
Deep venous thrombosis (DVT) is the most frequent cause for pulmonary embolism. CT venography for the evaluation ofDVT only plays a role in conjunction with examination of the chest for suspected pulmonary embolism. The interpretation is based on the general criteria described above (Fig.24.87d). Sensitivity and specificity are similar to (compression) ultrasound. Radiation exposure is an issue, especially in young males because of direct exposure of the gonads to a radiation dose of some
a
b
Portal Hypertension
Portal hypertension commonly causes dilatation of the portal and splanchnic veins. No significant dilatation, however, needs to be present if there are sufficient venous collaterals. In rare instances, there is the widening of the portal venous system without portal hypertension. For this reason, one should always check for signs of
c
d
Fig. 24.87 Venous thrombosis in the superior vena cava (a), inferiorvena cava (b), mesenteric confluence (c), and both femoral veins (d).
911
24 Vascular System collateral circulation or associated findings in patients with widened splanchnic veins and suspected portal hypertension (see also Chapter 11, p. 462) Collateral pathways between the splanchnic system and the systemic veins may form at various levels. The junction between protective and resorptive epithelium in the region of the esophagus and anus form a first group of collaterals. The paraumbilical veins drain portal venous blood directly into subcutaneous veins at the umbilicus. In the retroperitoneum and peritoneum various pathways may form, involving genital veins, collaterals between the mesentery and lumbar veins or the abdominal wall, diaphragmatic veins, and even veins within scar tissue after abdominal surgery. Finally there is the possibility of natural shunts forming between the splenic or gastric veins and the left renal vein. Such shunts may be large enough to bring down the portal pressure to almost normal, even though there may be markedly reduced hepatopetal flow in the portal vein. Large portosystemic shunts, however, may lead to encephalopathy or septicemia due to intestinal organisms. CT Morphology
Direct morphologic signs of portal hypertension are the dilatation of the portal vein (> 13 mm), the splenic vein (> 1 Omm) or the superior mesenteric vein (Fig.24.88a). On pre-contrast CT scans, varices may be mistaken for non-contrast-filled bowel loops, enlarged adrenal glands, or other space-occupying lesions. Varices may exceed the aorta in diameter (see Fig. 10.13). Following intravenous contrast injection, varices can be identified but scanning too early may mean that there is incomplete enhancement of the splanchnic venous system. Esophageal varices are the commonest complication of portal hypertension. They form a collateral pathway from the splanchnic system via the fundus of the stomach and the mucosa of the esophagus to the azygos system. The anterior branch of the left gastric vein drains into veins located in the esophageal wall, which leads to thickening of the esophageal wall with marked contrast uptake in the central portion of the esophagus close to the esophageal lumen. The posterior branch drains into the azygos system via paraesophageal veins, which form mul-
tiple tubular contrast-enhancing structures within the paraesophageal fatty tissue (see Fig. 10.13). There are communications to the esophageal veins via perforating veins that are rarely seen on CT. Paraesophageal varices may escape sclerotherapy and lead to early recurrence of esophageal varices. Large paraesophageal varices may cause mediastinal masses and form varicose aneurysms (see Fig. 10.13). Cardiophrenic collaterals may be relatively small (average size dmm). Following the esophageal veins down into the abdomen will demonstrate gastric collaterals, usually the coronary vein or short gastric veins (see Fig. 11.52). Anorectal varices cause a connection between inferior mesenteric vein and the rectal vein and via anorectal varices into the internal iliac venous system. While rectal varices often cannot be appreciated on CT. pararectal varices can be easily detected. As a result of the anorectal shunting, the inferior mesenteric vein or the internal iliac veins may be enlarged. Collaterals between the inferior mesenteric and the internal iliac veins may form colonic varices that are much rarer than rectal varices. Collaterals along the old path of the obliterated umbilical vein will cause enlargement of one or more paraumbilical veins. These paraumbilical veins leave the liver in the direct extension of the proximal portion of the left main portal venous branch and form a collateral pathway to the superficial superior and inferior epigastric veins (Fig.24.88b). Collaterals extend to the chest wall and down to the groin, forming a caput medusae. Paraumbilical varices are seen in patients with intra- or post-hepatic obstruction of portal flow such as in liver cirrhosis or chronic Budd-Chiari syndrome. Large subcutaneous varices are quite common and may complicate laparotomy. Collaterals between the splanchnic veins and the retroperitoneum or the abdominal wall, include connections between the liver capsule and the diaphragm, within the spleno-renal ligament, between the omentum and veins of the abdominal wall, lumbar veins, or veins in scars of previous laparotomies. Even paracaval anastomoses may be present. Collaterals to pulmonary veins have been described. In females, a pathway between the mesenteric veins and the ovarian vein may form. Collaterals around the common bile duct may be life-threatening at endoscopic cholecystectomy.
Neop/asia
a Fig. 24.88
c
b Portal hypertension, (a) Dilated mesenteric
vein and spontaneous splenorenal shunt (arrowheads) indicated by a markedly dilated left renal vein. Note the portal vein occlusion (arrow) by HCC (b) Thin-slab MIP of
Focal occlusion of a splanchnic vein may be caused by tumors or thrombosis and leads to segmental portal hypertension with local collateral formation (Fig.24.88c). A spontaneous splenorenal shunt is a relatively frequent finding in long-standing portal hypertension. It forms between the portal venous system and the left renal vein either directly from the splenic vein, or via diaphragmatic, pancreatic, left adrenal, or gastric veins. The typical CT finding is a dilated left renal vein. In some patients a short connection between the splenic vein and the left renal vein can be found (Fig.24.88a), while in others collaterals in the left upper abdomen may be so tortuous that they are hard to follow on cross-sectional images. In these cases, display of images in a cine mode or using MIP or volume rendering techniques will help. Neoplasia
Leiomyosarcomas of venous structures are rare
but more frequent than arterial sarcomas. They most commonly arise in the regions of the superior or inferior vena cava and lead to a well-perfused mass that obstructs the veins. CT can demonstrate the mass and establish its precise relationship towards adjacent organs prior to surgery.
paraumbilical collaterals (4x1/6). (c) Volume-rendering of gastroepiploic collaterals in a patient with splenic vein occlusion due to pancreatic cancer (3/5/2).
More frequent, however, is extrinsic compression of the major veins by a retroperitoneal or mediastinal mass, or direct infiltration of the veins and spread via intravascular tumor growth. Extrinsic compression is frequent in lymphoma or lymph node metastases, bronchogenic cancer, liver masses, adrenal and renal masses. Direct tumor infiltration of the vena cava is common in bronchogenic cancer, and in large retroperitoneal lesions. Indirect extension into the inferior vena cava (IYC) is often caused by renal cell carcinoma (via the renal veins) or hepatocellular carcinoma (via the liver veins). CT requires a long scan delay but then offers excellent characterization of disease. The main collateral pathways form the superior to the inferior vena cava involve the azygos system, the paraspinal venous plexus, the anterior chest wall, mediastinal veins, but rarely the hepatic or splanchnic venous system. Flow in the azygos system may be reversed if there is obstruction at the junction into the superior vena cava (SYC). Collaterals are well depicted by CT. •
CT Morphology
The suspicion of a venous leiomyosarcoma should arise in large tumors at unusual locations along the course of the superior or inferior vena cava. The masses often contain marked neovascularity with a large number of small to medium-sized vessels (Fig.24.89a). Infiltration
913
24 Vascular System
a
b
Fig. 24.89 Tumor involvement of the inferior vena cava (IVC). (a) Sarcoma of the IVC (arrow) that contains an abundant arterial blood supply (arterial phase, 4 x 1/ 6). It is hard to distinguish from an adrenal mass. (b) Infiltration of a recurrent renal cell carcinoma into the IVC
info their cava erals
adjacent organs may occur directly or via draining veins. Secondary signs of vena obstruction with various types of collatmay be present. Extrinsic compression causes a slit -like highgrade obstruction of the vena cava. The tumor is smoothly marginated relative to the opacified lumen (Fig.24.89c). Infiltration of the vena cava wall is proven by intra caval tumor masses or by a nodular or irregular margin between tumor and vessel lumen (Fig.24.89b), which may be hard to demonstrate in larger masses. The likelihood of infiltration grows with the circumferential contact between tumor and vein. Contact in less than 90° of the circumference makes infiltration unlikely, while it is highly likely with more than 270° of contact. There is a gray zone in between. Indirect infiltration can occur if a renal cell carcinoma infiltrates the renal veins, or a hepatocellular carcinoma infiltrates the he-
c (direct femoral contrast material injection), (c) Intrahepatic collateral pathway via an accessory inferior right hepatic vein in a patient with recurrent adrenal cancer and compression ofthe intrahepatic IVC (direct femoral contrast material injection, 3/5/2).
patic veins andspreads to the inferior vena cava. Tumors may extend into the right atrium. Tumor thrombus can be distinguished from apposition thrombus by its contrast uptake and heterogeneity (see Figs. '11.31,18.29). Collaterals from obstruction ofthe SVC form via the azygos veins, veins of the anterior chest wall, mediastinal veins (pericardiophrenic veins), and veins of the posterior chest wall that drain into the azygos system via intercostal veins. There are communications to the IVC via diaphragmatic veins and the renal veins, which have connections to the paraspinal venous plexus. Rarely are the communications via the anterior chest wall to paraumbilical veins into the left portal vein (see Fig.ll.SOa). Collaterals for the IVC mainly involve paraspinal veins and the azygos system. Rarely may there be connections via the splanchnic veins or via accessory hepatic veins (Fig. 24.89c).
Trauma Pulmonary Vessels Central pulmonary artery rupture with active hemorrhage into the chest is often fatal before the patient reaches hospital and CTA is limited to very few situations. Less severe major pulmonary vascular injury can result in pseudoaneurysm formation, which can be associated with distal thromboembolism. Traumatic pulmonary artery dissection is exceedingly rare. CTA is only indicated if the patient is stabilized but can confirm the diagnosis and establish the best site for surgical access through the poly traumatized chest wall. In addition, it can exclude concomitant aortic injury. •
CT Morphology
Similar to aortic injury, active contrast extravasation indicates the site of rupture (see Fig. 10.30). False aneurysm formation may occur at the site of a central transection. Peripheral traumatic microaneurysms smaller than 1.5 mm in size are difficult to detect with CTA in the presence of surrounding lung contusion. They may be complicated by pulmonary hematoma formation. Accompanying injuries of the thoracic aorta and-especially in children-the supra-aortic vessels should be carefully excluded.
Aorta
Acute traumatic aortic injury (AT AI) is seen in 3/1000 patients with blunt chest trauma. About 10-20% make it to the hospital but when untreated 90% of these patients die within 4 months. Survival with emergency treatment may be as high as 60-70%. Traumatic laceration of the aorta is commonly due to sudden horizontal deceleration, usually during a motor vehicle accident. In more than 90% of cases the injury occurs at the level of the aortic isthmus and ligamentum arteriosum. Injuries of the aortic arch with avulsion of the brachiocephalic artery, the ascending aorta immediately above the aortic valve (see Fig.24.54b), or the descending aorta are much
less frequent. Injuries of the abdominal aorta are usually due to direct penetrating trauma (Fig.2 4.90c). A complete tear will lead to exsanguination before the patient can reach the hospital. Incomplete tears can involve the intima and media, and will cause formation of a pseudoaneurysm that is contained by adventitia, periaortic connective tissue and clotted blood. Some 50% of such pseudoaneurysms will rupture within 24 hours. CTA has replaced angiography as the primary diagnostic tool in patients in whom traumatic aortic injury is suspected from the trauma mechanism or the chest radiograph. CT Morphology
Non-contrast CT will demonstrate mediastinal hemorrhage. Only in 20% of patients with mediastinal hemorrhage is aortic laceration present on subsequent CTA; most other patients suffer from hematoma due to fractures or injuries to smaller arteries or veins. CTA has been proven to have a high sensitivity and specificity for the detection of aortic laceration. It most frequently demonstrates a (false) aneurysm at a typical location in the proxtmal descending aorta with the aneurysmal sac oriented anteriorly and to the left (Fig.24.90a; see also Fig.24.48a). However, other locations such as the aortic arch, the ascending aorta immediately above the aortic valve, or the descending aorta may also be possible. There is frequently a flap-like protrusion of aortic wall elements (intimal tear) into the aortic lumen at the proximal or distal end of the injury. A post-traumatic dissection extending to distal portions of the aorta is quite rare. Direct injury due to penetrating trauma may cause similar findings but may occur anywhere along the aorta (Fig.24.90b). Most patients with such thoracic aortic injuries do not reach the CT scanner because of massive uncontained hemorrhage, but in the abdomen, pseudoaneurysms may form because of the retroperitoneal location of the injury. Attention should be given to concomitant injuries to other organ systems along the path of the injury (Fig.2 4.90c).
24 Vascular System
c
a
b
Fig. 24.90 Aortic injury, (a) Traumatic aortic laceration at the typical site in the proximal descending aorta with formation of an intimal flap. (b) Shotgun injury of the abdominal aorta with formation of a pseudoaneurysm
and concomitant vertebral injury (3/5/2). (c) Rare case of a penetrating trauma by a knife entering the descending aorta causing only minimum hemorrhage.
Neck Vessels
They differ from nontraumatic dissections by the fact that the common carotid is more frequently involved, patients present more often with cerebral ischemia, dissection results more often in complete obstruction, and response to therapy is less satisfactory. Blunt trauma to the vertebral artery with stretching of the vessel may result in dissection. The most frequent sites are at C1-C2, C6, and the craniocervical junction. Patients are frequently asymptomatic and rarely develop cerebral ischemia or stroke. Injury from penetrating trauma is often the result of fractures of the transverse processes or facet dislocations. Patients often present with spinal cord or nerve root damage. When the foreign body does not traverse the path of the artery, injury is unlikely. Color duplex ultrasound can be used as a first -line technique in blunt or penetrating trauma of the carotids but is not suited for investigation of zones I and III. Multislice CTA has the advantage of being able to depict the whole course of the carotid and vertebral arteries. It can evaluate nonvascular injury to surrounding soft tissues and spinal fractures in the same examination. It has the benefit of depicting the trajectory of penetrating foreign bodies.
Traumatic injuries may result from either blunt or penetrating trauma to the neck. Motor vehicle accidents are the commonest cause of blunt trauma. Other causes include falls, direct blows, or strangulation. Blunt trauma may result in dissection or thrombosis. Penetrating trauma, such as a stab or shot wound, more frequently causes vascular injuries. Penetrating carotid trauma in children is frequently due to an intraoral trauma when a child falls with a foreign object in the mouth. For penetrating carotid injuries, the neck is divided in three zones, which can be accurately examined by CTA: zone I below the cricoid cartilage, zone II between cricoid and mandibular angle, and zone III above the mandibular angle. Most penetrating injuries lead either to occlusions or to formation of pseudoaneurysms, while dissections are much less frequent. Severe acceleration-deceleration forces in the carotid arteries can produce dissections; severe hyperextension combined with rotation stretches the artery over the lateral masses of C1-C2, while sudden hyperflexion can compress the artery between the mandibular angle and the spine. Traumatic carotid dissections may be clinically silent in 50% of patients directly after trauma and only manifest after 12-24 hours.
Venous Trauma
is particularly frequent after cardio-angiographic procedures at the site of arterial and venous access in the groin. CTA has the potential to substitute for diagnostic angiographic procedures because it allows for demonstration of the hematoma, accompanying injuries to the skeletal system or to internal organ, as well as the site of active bleeding. In patients with a continuing blood loss, however, angiography with subsequent embolization of the affected artery will be abetter choice. • Fig. 24.91 Carotid injury following a motorcycle accident. There is focal vessel laceration and a traumatic carotid dissection. The intimal flap lead to a large media infarction.
•
CT Morphology
Traumatic dissection presents similar to nontraumatic dissections (see above, p. 895). Pseudoaneurysms have variable appearances with irregular vessel contours, focal dilatation with or without surrounding hematoma, or active extravasation in a soft -tissue hematoma. Intimal flaps may be present and cause intermittent malperfusion of the brain (Fig. 24.91). Occlusions show a lack of intravascular enhancement with or without proximal narrowing from a concomitant dissection. Comparison with the contralateral side is crucial for diagnosis. Arterial Bleeding
Arterial injury outside the neck usually leads to substantial bleeding at the site of injury with large hematoma formation that can extend into accompanying structures. Complete transection or the presence of an intimal flap may lead to occlusion of the affected vessel. Long-term sequelae are pseudoaneurysm formation or arteriovenous shunts in cases with simultaneous trauma to arteries and veins. The latter
CT Morphology
As opposed to venous bleeding, the hematoma in arterial bleeding is usually extensive and often has a convex outer contour. The site of active bleeding should be searched for in the region of the hematoma. A small spot of brightly opacified blood outside the vessel lumen is indicative of the site of bleeding. The size of the paravasation indicates the amount of active bleeding. This spot will enlarge if a second scan is added in a venous phase. Such venous phase images are especially helpful to rule out injuries to abdominal organs in patients with blunt abdominal trauma. Venous Trauma
Venous bleeding becomes dangerous only if it bleeds into a body cavity, or if there is consequent venous compression and stasis with resulting thrombosis of the affected vein. •
CT Morphology
Hemorrhage associated with venous bleeding is usually less extensive and often causes streaky densities in the surrounding tissue. The associated vein may be compressed or thrombosed. Active bleeding is almost impossible to detect by CT because the bleeding site becomes compressed by the ensuing hematoma and thus spontaneously closes.
917
24 Vascular System
CT is the imaging modality of choice for evaluating many postoperative and postinterventional complications. The only exception is the heart, where the preferred modalities are echocardiography, cardiac angiography, and MRI, depending on the nature of the study. Normal Postoperative Findings
Stranding in fatty tissue, air bubbles, small hematomas, and seromas are among the normal changes seen during the immediate postoperative period. Ectopic gas should disappear by 3-4 weeks postoperatively, hematoma should be completely resolved by 2-3 months. As normal hematomas also may develop a rim of enhancing granulation tissue, the distinction between infected and bland hematoma may be impossible by CT. Prosthetic materials such as Dacron and Gore-tex are iso- or hyperattenuating to the soft tissues. Thus, it may be difficult to identify aortic prostheses on CT scans. Only Teflon, which is used to encase and stabilize anastomoses, is easily identified by its relatively high CT density. Vascular prostheses tend to assume a straight, extended position in the body. Slight angulation between two prosthetic segments or within a
a
b
Fig. 24.92 Vascular grafts, (a) Replacement of the ascending aorta and proximal arch shows typical angulation atthe site ofthe anastomosis (3/5/2). (b) lIiacorenai bypass following an iatrogenic aortic dissection caused by unsuccessful dilatation of the left renal artery
prosthesis is normal, provided it does not cause significant luminal narrowing (Fig. 24.92 a). Venous bypass grafts are generally 4-7 mm in diameter, and therefore abrupt caliber changes may be seen at sites of anastomosis with smaller vessels (Fig.24.92b). A similar effect is seen with synthetic bypass grafts used in the splanchnic vascular system. Patch grafts tend to undergo slight dilatation. To simplify anastomoses in subsequent operations on the aorta (e.g., for aortic dissection or multiple aortic aneurysms), the graft may be sutured into the aorta in such as way that its distal end projects freely into the aortic lumen ["elephant trunk'} while the actual anastomosis is performed several centimeters proximal to that level. This makes it possible to attach another, more distal prosthetic segment without having to open the original anastomosis (Fig.24.92c) . Malpositioned Catheter Systems
Malpositioned catheters can be directly demonstrated by CT. This has the advantage that the catheter as well as the surrounding tissues can be seen. It is advisable to follow each catheter and line that is present on a CT scan from out-
c (3/5/2). There is a caliber difference between the graft and the renal artery, (c) A free distal end of an aortic graft within the aortic lumen is called an elephant trunk. It is used to provide an easier access for a proximal anastomosis in case of further aortic replacement.
Arterial Dissection
side the body to the tip of the probe within the patient. CT can identify catheters that are positioned in the wrong vessel, have perforated into the surrounding tissue, and have caused hemorrhage, Venous h'nes can enter various side branches, may be localized in an artery, or may have perforated into the soft tissue of the neck, or mediastinum, or into the pleural space, One has to watch out for concomitant pneumothorax, hematoma, or (infused) fluid in the pleura or even the pericardial sac. Concomitant thrombosis of the access vessel, or apposition thrombus at the catheter tip or around the course of the catheter are important findings that may complicate removal of the line, A central venous line should end in the distal superior vena cava, Positioning the arm above the head may lead to a downward movement of the catheter tip if the line was placed via a cubital vein. Arterial catheters or intra-aortic balloon pumps may have caused perforation (with hemorrhage and potential stenosis), thrombosis, dissection, or occlusion of the access vessels,
Anastomotic Complications
Table
24.17
Complications
of vascular
grafts
and
by-
passes Acute
anastomotic
• Anastomotic
leak
pseudoaneurysm
• Perigraft perfusion • Confined perforation • Suture-line failure with hemorrhage Chronic
anastomotic
• Anastomotic Hemorrhagic
leak
aneurysm
(pseudoaneurysm)
complications
Local hematoma Retroperitoneal
or mediastinal
hematoit]
Cardiac tamponade Hemorrhagic
pleural
effusion
Hemoperitoneum Infection • Infected • Abscess • Mycotic • Fistula
prosthesis aneurysm formation
(trachea,
bow;
Malperfusion • Dissection (due to cannulation) • Vascular occlusion (thrombosis) • Stenosis (anastomosis) • Organ infarction or bowel)
(especially
brain,
kidney,
Dehiscence of the suture line can lead to a broad range of complications (Table 24.17), A confined perforation develops into a pseudoaneurysm (Fig,24,93a), which is also a typical long-term complication, The extent and site of the leak is often well appreciated on 3D reformations, Multiplanar reformations are preferred if concomitant thrombotic changes are present. Perigraft perfusion occurs when the native vessel is wrapped around the vascular graft and a faulty suture line allows blood to enter the space between the prosthesis and the wrap (Fig,24,93b), Usually only a small part of the space is perfused, while most of it is thrombosed, Perigraft perfusion can lead to compression of the graft or may develop into a confined or free perforation, An unconfined extravasation leads to local hematoma formation or to active bleeding into the mediastinum or retroperitoneum. Grave complications can result from the development of a hemorrhagic pleural or pericardial effusion (ranging to cardiac tamponade) or hemoperitoneum.
Arterial Dissection Arterial angiography, vascular cannulation in heart surgery, intra-aortic balloon pumps, and the insertion of monitoring probes or aortic stent grafts can cause iatrogenic dissection of the aorta or great arteries (see Figs,24,92b and 24,98a), Typical signs of dissection are seen as far as the re-entry point, which often occurs at the level of the major abdominal aortic branches or the descending aorta, Secondary problems consist of dissection-related perfusion deficits or infarction due to thrombosis or luminal compression.
919
920
24 Vascular System
a
·b
c
Fig. 24.93 Complications of aortic replacement, (a) Anastomotic dehiscence with development of a false aneurysm in the chest wall following replacement ofthe ascending aorta, (b) Peri graft perfusion due to a distal anastomotic leak following thoracoabdominal aortic re-
placement. Blood entering the native aorta that was wrapped around the graft has caused luminal compression (3/5/2). (c) Graft infection following replacement of the ascending aorta. Note the small air bubble at the site of infection.
Stenoses and Occlusions
Infected Prosthesis
Stenoses most commonly occur at anastomotic sites and may be primary (overtightened suture) or secondary (fibrosis, thrombotic deposits). Vascular occlusions are a serious complication that, when acute, can cause severe perfusion deficits in the region supplied by the occluded vessel. Anastomotic sites are predisposed to occlusion if there are changes in coagulation parameters, impaired distal outflow, competitive afferents, or the slowing of blood flow due to the greater diameter of the bypass relative to the anastomosed distal vessels. CTA is helpful in detecting a stenosis or occlusion (see Fig.24.98b), but it is not useful for evaluating functional parameters such as distal outflow or competitive afferents. The lesions are most clearly appreciated on axial sections and on CPR that follow the longitudinal course of the vessel.
Perigraft infection leads to fever, chills, and leukocytosis, as well as local swelling in the groin or even spontaneous cutaneous drainage of putrid material. An aortoenteric fistula may present as acute or chronic, or even occult gastrointestinal bleeding together with sepsis. CT cannot reliably distinguish a periprosthetic infection from normal postoperative changes during the immediate postoperative period. It can, however, support a presumptive diagnosis when interpreted within the context of clinical findings. Perigraft infection is suggested by the peripheral enhancement of periprosthetic fluid collections, gas bubbles, and diffuse periprosthetic enhancement (Fig.24.93c). More than 5mm between the graft and the surrounding wrap beyond the seventh postoperative week are suspicious for infection. Ectopic gas may indicate a fistulous connection between graft and bowel structures, or superinfection with gas-producing organisms. Focal bowel wall thickening may accompany a aortoenteric fistula.
Follow-up
Angiographic
Interventions:
Patient Selection
CTA has become the standard technique for selection of patients for endovascular aortic repair and is useful also for planning other interventional procedures because it provides information, not only about the vessel lumen but also about the vessel wall and the (non-distorted) three-dimensional relationship of the various vascular structures, It is an excellent tool for the minimally invasive follow-up of angiographic interventional procedures, especially for aortic stent grafts and complications after angioplasty. For multiple long-term follow-up examinations, MRA should be considered as an alternative in younger individuals, and color duplex ultrasound is an effective alternative in regions that are easily accessible by ultrasound,
Planning Angiographic Interventional Procedures CTA is rarely performed for the sole purpose of planning angiographic interventional procedures, The main exceptions are aortic stent grafts (see below) and the three-dimensional assessment of complex aneurysms (see Fig, 2,32), CTA performed primarily for diagnostic purposes also provides valuable information for planning angioplastic procedures, for the decision to perform PTA or primary stenting, for demonstrating a distorted portal anatomy in massive cirrhosis prior to TIPS procedures, and for the embolization of tumors, malformations or aneurysms. A normal CT angiogram almost completely rules out the presence of significant stenoses or aneurysms if the technical quality of the examination is adequate (good contrast enhancement, no major breathing artifacts), CTA is able to distinguish non-calcified from calcified plaques, which usually respond less well to balloon angioplasty and are more prone to complications such as focal dissection or rupture, In calcified plaques, primary stenting can be an alterative, Ulcerated plaques at the site of intervention or along the access route may increase the risk of embolic complications, Again primary stenting of ulcerated stenoses (in the carotids combined with the use of protection devices) is an option (see Fig, 24,69),
afterAngiographic
Procedures
and Follow-up
Unusual anatomy may make an intervention more time-consuming, CIA can demonstrate the level of origin of such an atypical vessel (e.g, accessory renal artery, anterior origin of a renal artery, parallel origin of two renal arteries at the same level) and can even give information about the angle at which it arises from the aorta (see Fig,24.77d), This information may be used for selection of proper catheter types for selective angiography, Accessory or replaced hepatic arteries may complicate arterial embolization of the liver. Accessory left hepatic arteries that arise from the left gastric artery can be missed on angiography but their presence is easily established by the arterial phase of dual phase liver CT (see Figs, 24,73, 24,74), Feeding and draining vessels of (pseudo) aneurysms or AV malformations can be excellently displayed by CTA prior to embolization (see Fig,24,35), Thin-slab MIP or volume-rendered displays after removal of superimposing vessels are best suited for this purpose, CTA can exclude stenosis or excessive kinking of the access route and thus help decide about the site of a femoral access or whether to use a brachial or axillary access route instead, Follow-up after Angiographic Procedures The formation of hematomas, pseudoaneurysms, or arteriovenous shunts at the site of the femoral access can easily be detected by color duplex ultrasound and does not require CTA CTA, however, is an excellenttool fordetection of other acute complications after angiographic procedures unless there are signs of renal failure, in which case ultrasound or MRA are to be preferred, A postinterventional decrease in hematocrit levels should prompt the search for an intraabdominal hematoma, which often may be caused by bleeding underneath the inguinal ligament into the psoas muscle, retroperitoneum, or anterior abdominal wall (injury to the inferior epigastric arteries), In addition, the site of intervention should be evaluated to rule out perforation and active bleeding, While the search for a hematoma does not require injection of contrast material, vascular complications
24 Vascular System such as active paravasation, dissection, stenosis, emboli, or vascular occlusion requires performing a CT. In cases of vascular occlusion, signs of organ ischemia have to be searched for. In patients after arterial embolization, CT can detect dislodged endovascular coils or demonstrate lipiodol-containing embolization material outside the target region. CTA can be used to evaluate the position and patency of vascular stents (Fig. 24.94). Accurate assessment requires the use of thin collimation. Possibly it is best with thin-section multislice scanning. Patency is confirmed indirectly by finding normal opacification of the vascular segments just proximal and distal to the stent. Occlusion of a stent usually leads to thrombosis of the vessel segments distal to the stent, depending on the amount of collateralization. Instent stenosis is best evaluated on curved planar reformations through the centerline of the stented vessel or on sections perpendicular to the vessel course. Luminal narrowing is commonly circumferential in cases ofneointima formation but may be eccentric in cases of mural thrombi. Nitinol stents have a lower CT attenuation than stainless steel devices and thus suffer less from artifacts and blooming effects. These blooming effects are worst with Strecker Tantalum and Passager stents. Blooming lets the stent appear larger than it is in reality and causes artificiallumen narrowing. As with mural calcifications, wide window settings improve evaluation of the stent lumen.
a
b
Fig. 24.94 Renal artery stents. (a) Stent placed in the left renal artery after an aortic dissection to maintain perfusion from the true lumen (2/3/1). (b) Stent malposition. The stent projects far into the aortic lumen, and a soft plaque distal to the end of the stent causes
Chronic complications that can be detected by CTA include formation of pseudoaneurysms, usually at the site of vascular access, or at the site of vascular injury along the access vessels, or in the target region of vascular interventions. In addition, re-stenosis or dislocation of metallic stents or covered stent grafts with recurrence of aneurysms can be demonstrated. Stent migration may cause reappearance of aneurysms or cause distal stenosis. Chronic dissections may form after endovascular injury (see Fig.24.98a). Dissections are especially common in the iliac vessels in cases with increased tortuosity. They may remain focal or extent into abdominal aorta. Planning Endovascular Aortic Repair
Endovascular aortic prostheses (stent grafts) are used for minimally invasive treatment of abdominal aortic aneurysms, thoracic aortic aneurysms, penetrating aortic ulcers, and type B aortic dissections. These devices consist of endoluminal stents covered with a synthetic fabric to exclude the surrounding aneurysm or false channel in aortic dissection from perfusion and to promote thrombosis. In simple cases, a single tubular graft is sufficient but more complex cases require extension of the stent by a second graft, or the use of a bifurcated graft to extend the stent into the iliac arteries. CT has an essential role in the preinterventional planning and postinterventional follow-
c
d
additional luminal narrowing within the artery (2/3/1). (c. d) Intima hyperplasia in a renal artery stent (4 x 1/6). A wide window setting is required for optimum evaluation.
Planning Endovascufar Aortic Repair
up of these procedures. The criteria vary substantially between thoracic and abdominal grafts.
Table
24.18
• Aneurysm
Thoracic
Aorta
The mdicarions for stenting of the descending aorta are steadily increasing. Most indications are similar to those for surgical repair but since the morbidity associated with the stenting procedure is less. even patients who are unfit for surgery may be treated. Best results are obtained with penetrating aortic ulcers or saccular aneurysms that can be completely excluded from the circulation. Longer aneurysmatic portions of the descending aorta can also be treated with stent grafts but usually more than one graft will be necessary. Treatment of type B aortic dissections is performed by placing a stent graft over the proximal entry. Treatment is indicated when complications such as malperfusion of side branches. compression of the true lumen. progressive aneurysmatic dilatation. or distal progression of dissection occur. Contraindications include infected aneurysms or surgical grafts. aneurysms involving the arch or the celiac artery origin. and type A dissections, as well as manifest aortic rupture. The prerequisites for stenting have to be checked by CTA (Table 24.18). There has to be a suitable proximal and distal anchoring zone for the stent graft, which should measure at least "1.5 cm. The diameter of the proximal and distal anchoring zones must fit the diameter of the implanted stent graft. Usually the proximal anchoring zone is located distal to the left subclavian artery origin but "overstenting" of this vessel may be feasible if a surgical transposition to the left carotid artery is performed prior to stent graft placement, of if a sufficient natural collateral circulation is present via retrograde flow in the left vertebral artery (see Fig. 24.97 a). Reverse tapering of the "normal" vessel segments proximal or distal to the aneurysm necks should not exceed 0.3-0.4 em diameter per em length because this reduces the graft friction substantially and may lead to graft migration (Fig.24.95a). Care should be taken in a patient with a large saccular aortic aneurysm and a wide neck (aortic wall defect) combined with a short proximal graft-anchoring zone. In such situation the blood flow can pull the lateral graft Sideways into the aneurysm neck. The minimal slippage
Thoracic
Preinterventional
aortic
stent graft checklist
planning:
or dissection
selection is entirely
criteria located
in the
descending aorta • A segment s 10-15 descending
aorta
mm of distal arch/proximal can be used as anchoring site
• Minimum aortic diameter is less than 30 mm to 35 mm (depending on the device) • Celiac artery is unaffected • No infection or aortoenteric fistula • No severe (circular) atherosclerosis of the proxi-
mal anchoring zone • No severe aortic angulation anchoring
zone
Complexity measures?
in the
>60°
• Tortuosity of access access device
of the procedure,
• Aneurysm or dissection origin of left subclavian
proximal
"
rcuo
may be corrected
by
concomitant
extends artery
right up to the
• Documentation of a patent vertebral and right subclavian circulation in patients with planned stent graft covering the left subclavian artery origin • Mural thrombi • Vascular calcifications Determining
the dimensions
ofthe
stent
• Width of stent: diameter ofthe narrowest segment ofthe proximal and distal descending aorta (anchoring sites) • Proximal and distal diameters ofthe proximal and distal anchoring zone • Length of proximal anchoring zone between the most distal supraaortic branch and the proximal aneurysm neck • Length ofstent: plus more than Postinterventional
total length of dilated segments 10-15 mm proximal and distal follow-up
• Diameter of the aneurysm • Endoleak present? Type of leak? Sustaining vessel? • Perfusion of aortic side branches? • Fracture of stent material
~,»»
can render the proximal graft fixation incompetent and result in a proximal type 1 endoleak (see Fig.24.99a). A similar mechanism has been thought responsible for graft migration in the distal descending thoracic aorta in cases with very large fusiform aneurysms. The choice of stent graft length is not critical if the distal third of the aorta above the dia-
923
924
24 Vascular System Fig. 24.95 Reverse tapering of an atherosclerotic aneurysm of the descending aorta (a). There is a rapid decrease in aortic diameter immediately above the level of the celiac artery ("). which makes the resulting short distal anchoring zone highly unstable. Substantial kinking of the descending aorta in another patient makes stent graft placement impossible (4 X 2.5/15) .
.b
a
phragm displays a normal aortic (true) lumen. Multiple grafts can be used to achieve the desired length. However, positioning a graft junction into the outer curve of a tortuous aneurysm should be avoided because of the risk of graft module disengagement with a second-
arytype III endoleak (Table24.19). Withalonger covered segment of the descending aorta the potential risk of paraplegia increases if the anterior spinal artery is excluded from the circulation. The origin of this artery is still hard to detect by CTA. In selected cases, however, even the
Table 24.19 Abdominal aortic stent graft checklist Preinterventional planning: selection criteria Aneurysm is entirely infrarenal More than a 15 mm segment of the infrarenal aorta has a normal diameter (to ensure stable attachment) Minimum aortic diameter is less than 25 mm to 30 mm (depending on the device) Iliac bifurcation is unaffected on one side (no absolute prerequisite) No infection or aortoenteric fistula No severe (circular) atherosclerosis ofthe proximal anchoring zone No severe aortic angulation in the proximal anchoring zone >60 Tortuosity of access route may be corrected by access device 0
Complexity of the procedure, concomitant measures? Involvement ofthe iliac arteries (stenoses or aneurysms) Renal artery stenoses Accessory renal arteries Stenosis of the celiac or superior mesenteric study Mural thrombi Vascular calcifications
Determining the dimensions of the stent • Width of stent: diameter of the narrowest infrarenal aortic segment • Length of stent: total length of dilated segments plus 15 mm proximal and distal B-» Width of Y extensions: diameter of common and external iliac arteries on either side
'ostinterventional follow-up • • • •
Diameter of the aneurysm Endoleak present? Type of leak? Sustaining vessel? Narrowing of the extensions? Fracture of stent material
Planning Endovascular Aortic Repair origm of the celiac artery has been covered without serious consequences, although such a procedure is not recommended on a general basis. •
Abdominal Aorta
Aortic stent grafts were first used for abdominal aortic aneurysms and there is now extensive clinical experience available for this procedure. CTA is the technique of choice to evaluate whether a patient is a suitable candidate and to obtain length and size measurements for individual fitting of the graft. As a prerequisite for stenting, it is determined whether the device can be attached stably at its proximal end without jeopardizing the renal arteries, and whether adequate lower extremity arterial inflow can be maintained distally. (Generally, one of the two internal iliac arteries should be preserved.) This requires an entirely infrarenal location of the aneurysm, a sufficiently long (10-15 mm) infrarenal aortic segment of normal diameter « 25 mm or larger, depending on the available stent), and an iliac bifurcation that has a normal caliber on at least one side. Reverse tapering of the proximal anchoring zone should not exceed 0.3-0.4 cm diameter per cm length because this reduces the graft friction significantly and may lead to distal migration and recurrence of the aneurysm.
Individual fitting of the stent graft requires determining the width of the proximal end of the stent, the length of the stent, and if necessary the width and length of the iliac extensions. The measurements are performed on MPRs that are orthogonal to the vessel axis (for determining vascular diameters) and parallel to the vessel axis (for determining the length of the aneurysm). Standard measurements should include the distances shown, for example, on the standard evaluation form of the EVARmulticenter trial (Fig. 24.96). It should be noted that stents placed in very large perfused aneurysms tend to lie against the posterior wall of the aneurysm, if the distal anchoring is stable, so that measuring the length of the device at the center of the aneurysm (vessel axis) will yield values that are too large. Conversely, insufficient distal anchoring in such aneurysms will cause the stent graft to lie against the anterior aneurysm wall, so that measuring the length of the device at the center of the aneurysm (vessel axis) will yield values that are too small. •
Evaluation of Abdominal and Iliac Arteries
The inferior mesenteric artery is always occluded by the stent. Concomitant stenoses of the celiac and superior mesenteric artery may thus increase the risk of abdominal ischemia and should be excluded or treated prior to stent
~I Level of the rellilLilrt~n~~ Does AAA exhibit Yes 0 soccular element? No 0 Are patent lumber vessels Yes 0 demonstrated? No
Fig. 24.96 Standard measurements prior to abdominal aortic stent graft placement (European multicenter trial for aortic aneurysm repair, EVAR).
925
suprarenal
diameter
CD
_ top neck diameter CD neck length CD bottom neck diameterCD
0 0 0
0
I
24 VascularSystem grafting. Stenoses of the renal arteries may be a cause of hypertension and thus should be approached by angioplasty or stenting before or after endovascular repair of the aorta. In some centers the presence of "prominent" lumbar or inferior mesenteric arteries is seen as an indication for prophylactic embolization, as these vessel origins are difficult to access after abdominal aortic repair. In the presence of iliac artery involvement a graft extension into the external iliac artery may be required and the sizes of external and internal iliac arteries should be obtained. Before the internal iliac artery is covered, collateral supply from the inferior mesenteric, pelvic, and profunda femoris territories should be investigated and stenoses of these vessels should be excluded. •
Access Vessels
The left subclavian artery is commonly approached with a diagnostic flush catheter. Hence, CTA of the thoracic aorta should give information about stenosis or occlusion of its origin, which is important for the choice of a contralateral approach. The femoral and iliac arteries serve as the main access vessels. Diameter measurements and exclusion of significant stenosis are therefore obligatory. Current introducer diameters measure up to 22-24 F (7.3-8.0mm). The sizing of iliac arteries is of particular importance, especially in slender patients (e.g., those from east
a
b
Fig. 24.97 Successful stent graft placement for a proximal descending aortic aneurysm. The patient remains asymptomatic despite occlusion of the left subclavian artery (arrow) by the stent (4 x 2.5/15) (a). Abdominal
Asia) , to determine the requirement of an iliac conduit for introduction of the delivery system. Increased tortuosity of the access vessels may increase the risk of dissection or perforation (see Fig.24.98a). For this reason, the side with less curvature should be used for femoral access or the patient should be excluded from endovascular repair if the tortuosity is too marked. Tortuosity measurements are automatically available with some automated vessel tracking programs. Follow-up after Endovascular Aortic Repair
CTA is currently the preferred modality for follow-up after endovascular repair. It has a higher accuracy in the detection of endoleaks than color duplex ultrasound and is not as susceptible as MRA to metal artifacts from the stent material. A delayed scan is necessary to detect endoleaks with slow flow that may be missed on typical arterial phase CTA scans. At present there is no final consensus about follow-up intervals, but in many institutions CTA is performed 3, 6,12, 18, and 24 months after the procedure and yearly thereafter (Fig. 24.97). Under such conditions, Gd-enhanced MRA may be considered as a long-term alternative. After successful endovascular repair associated abnormalities affecting the aneurysm wall or periaortic fat tissue (e.g., stranding in
c
d aortic aneurym 3 months (b) and 18 months after stent graft placement (c). Pressure measurement by direct puncture of the aneurysm sack in a patient with aneurysm growth despite overt endoleak (d).
Follow-up after Endovascular Aortic Repair
covered perforation) should disappear. Even the periaortic reaction in inflammatory aneurysms has been reported to decrease over time. Fracture ofstent material can best be appreciated on conventional plain radiographs of the stent area, but MIP or volume-rendered displays from multislice CT data may also demonstrate such events. Stenoses of the stent lumen may occur in aortoiliac grafts, especially at juncture sites between the various components of the graft. Dissection of the access vessel during graft placement is usually associated with excessive vascular tortuosity (Fig.24. 98a). Thrombosis of the stent lumen (Fig.24.98b) is most frequently caused by mechanical obstruction, either by the stent itself or by dislocated coating material. In very large aneurysm, the stent graft may migrate to the anterior portion of the aortic lumen (Fig.24.98c), especially when the distal anchoring site is not stable. Such a migration may cause massive type I endoleaks and has an increased risk of aneurysm rupture. The diameter of the aneurysm is measured at follow-up, and serial measurements should not show an increase in size. However the mode of size assessment is important. Rarely, a decrease in aneurysm diameter may be associated with an increase in size, and thus volume, and a further predisposition to rupture. Successfully treated aneurysms or dissections should thrombose completely and shrink in size over a period of months to years (Fig. 24.97 b, c). An increase in size should prompt the search for an endoleak, which may be treated interven-
a
b
Fig. 24.98 Complications after stent graft placement, (a) Dissection of the left iliac artery along the path of graft insertion is most frequent in tortuous vessels, (b)
tionally. If no leak is detected despite a growth over time (endotension), an occult leak must be suspected. Gd -enhanced MRA may be an alternative in such a situation because it is more sensitive to small amounts of contrast material in the aneurysmal sac. Four types of endoleaks are distinguished that cause residual perfusion of the aneurysm or dissection (Table 24.20; Fig. 24.99). Early technical failure or graft migration will cause type I endoleaks at the proximal or distal anchoring points. Disconnection at a modular graft junction with disengagement of two components Table 24.20 ment Type
Types of endoleak after stent graft placeDescri ptionjj
Treatment
'·,it"
Type I
proximal or distal graftrelated leak
always
Type II
non graft-related leak from retrograde perfusion via an aortic side branch
often
Type III
graft -related leak at a modular graft junction (disengagement of the two components)
always
Type IV
early leak without evidence of any source, attributed to porosity of the graft material
rarely
Endotension
enlargement of the aneurysm without evidence of endoleaks by imaging
always
c thrombosis ofthe right leg of a graft, (c) Anterior deviation ofthe graft position in a giantAAA may lead to type III endoleaks and then requires re-stenting.
24 Vascular System
a Fig. 24.99 I endoleak doleak due in a patient
b
c
d
Endoleaks after stent graft placement. Type at the distal attachment site (a). Type II ento a nonoccluded inferior mesenteric artery with a horseshoe kidney (b). Type II endo-
leak not detected on the CTA (above) but on the delayed scan 3 minutes after the end of the CTA (c). Type III endoleak after disengagement of two stent grafts in a tortuous descending thoracic aneurysm (d).
will cause a type III endoleak (Fig.24.98d). Such an endoleak may appear in large saccular or fusiform aneurysms ifthe graft is pulled into the aneurysm, thus retracting from a graft junction. Type I and type III endoleaks require urgent reintervention. Vessels that are perpetuating the outflow from the aneurysm are the reason for type II endoleaks (Fig.24.99b). Typically these are lumbar vessels, the inferior mesenteric artery, or accessory renal arteries that were not
occluded during the primary procedure. CTA is then used in planning the reintervention, e.g., embolization of these side branches, if the endoleak persists for a longer period of time. Slight residual perfusion without apparent cause may be found immediately after the procedure and has been attributed to early ("physiological") porosity of the graft material. Such type IV endoleaks should disappear within 4 to 6 weeks.
25 Musculoskeletal System C. Schaefer-Prokop,
U. v. Smekal, A.j. van der Molen
Anatomy
,
_
Spinal Column (p. 931) Upper extremity (p. 934) Pelvis and lower extremity (p. 936) ExaminationTechnique
_
General (p. 937) Spinal Imaging (p. 941) Skeletal Imaging (p. 941) Poly trauma Patients (p. 943) CT Arthrography (p. 945) Torsion Assessment (p. 945) Qua ntitative CT Bone Densitometry
(p. 946)
Spinal Trauma
_
Upper Cervical Spine (p. 948) Lower Cervical Spine (C3-C7) (p. 951) Thoracolumbar Spine (p. 952) Trauma
to the Pelvis and Extremities
Upper Extremity (p. 956) Ribs and Sternum (p. 961) Pelvis (p. 962) Lower extremity (p. 967) Stress Fractures (p.975) Intervertebral
Disc Herniation
and Degenerative
Disease of the Spine
Inter vertebra I Disc Lesions (p. 975) Spinal Stenosis (p. 980) Infection
and Inflammatory
Processes
Infectious arthritis (p. 982) Osteomyelitis (p. 983) Infectious Spondylitis and Spondylodiscitis (p. 983) Sacroiliitis(p.984 ) Heterotopic Ossification (Myositis Ossificans) (p. 984) Soft-tissue Infections (p. 985) Bursitis(p.986)
Bone
986
Tumors
Primary
Bone
Skeletal
Metastases
Soft-Tissue
Tumors
Postoperative Metallic Nonunion
Tumors
(p. 986) (p. 992) (p. 994)
Changes
Implants
and Arthrodesis
Osteonecrosis
and
994
Complications
(p. 994) (p. 995)
(p. 996)
Musculoskeletal imaging relies primarily on plain films and MRI for diagnosis, while the use of high-resolution ultrasound is increasing, Computed tomography has seen a revival with the introduction of helical and multislice CT techniques, CT is the technique of choice for detailed evaluation of skeletal trauma; the use of multiplanar reformations and 3D rendering has revolutionized trauma diagnosis, especially when complex fractures are present or articular involvement is suspected, It is also used for complications in fracture healing, for selected bone tumors, in infectious diseases, to guide musculoskeletal interventions and for some miscellaneous indications such as quantifying rotational deformities before and after corrective surgery (Table 25.1), MRI has assumed a leading role for imaging the intervertebral disc space and its relationship to the myelon but imaging of the lumbar spine is still a frequent indication for CT scanning, CT offers the advantages of its ready availability and the better delineation of skeletal structures, With multislice scanning, CT may assume a new role for this purpose because of its superior 3D spatial resolution, Quantitative CT (QCT) of the spinal column is still among the best of many techniques that have been devised for bone densitometry, Although QCT has been discredited in recent years due to improper use, it can still furnish important quantitative information in the follow-up of osteoporosis.
Table 25.1
Indications
for musculoskeletal
CT imaging
Trauma Screening of poIytraumatized Complex fractures
patients
Fractures in anatomically complex areas Fracture detection in patients with equivocal films Preoperative Nonunion Osteonecrosis Intervertrebral Herniated Foramina! Joint
planning
and postoperative
plain
follow-up
disk disease
disk (lumbar spine) stenosis (lumbar spine)
disease
Shoulder
instability
(CT arthrography)
Miscellaneous All traditional
indications
fortomography
Quantitative CT Rotational deformity CT as a second-line Cervical
and thoracic
imaging
tool
spine in degenerative
eases (MRI first) Bone tumors (radiography, MRI) Soft-tissue tumors (MRI, ultrasound) Arthritis, osteoarthritis (radiography, sound)
dis-
MRI, ultra-
Anatomy The emphasis in musculoskeletal CT is on bone imaging. MRI and ultrasound are more suitable for imaging the soft tissues. Because bone contains calcium compounds. most notably hydroxyapatite [Caw(P04)6(OH)2]. bone has a high x-ray attenuation and contrasts sharply with surrounding soft-tissue structures. Since bone minerals have a higher effective atomic number than water. the CT attenuation of bone is greatly influenced by the tube voltage and prefiltering. The cortex of the long bones may exceed 3000 HU, while the attenuation of cancellous bone depends strongly on the bone mineral content and the proportion of fatty marrow. The measured attenuation declines with increasing age, and negative values may even be found in patients with severe osteoporosis or fatty infiltration of the bone marrow. CT sections display a summation of all the cancellous bone trabeculae that are located within the section. As a result, fissures in the cancellous bone are much more difficult to define than in the cortex. Cancellous bone comprises only 20% of the bony skeleton, and its large surface area makes it particularly susceptible to the effects of metabolic abnormalities. QCT can provide quantitative information on the bone mineral content of the cancellous portion of the vertebral bodies, femoral neck, or distal radius. The bone marrow, which consists of red hematopoietic marrow in all bones of the newborn, is almost completely transformed into yellow fatty marrow as the individual matures. By adulthood, hematopoietic marrow is found only in the epiphyses of certain tubular bones, in short bones, and in flat bones (e.g., sternum, iliac crest). Material of soft-tissue attenuation within the bone marrow in an adult is considered pathologic (Table 25.2). In patients with a high bone mineral content, the attenuation of
Table
25.2
Soft-tissue
density in the medullary
cavity
the trabeculae will cause partial volume effects that may make it impossible to detect cellular infiltration of the fatty marrow. For this reason, diagnosis of bone marrow infiltration is the domain ofMRI. Vascular canals and persistent apophyses or ossification centers in the growing skeleton should not be mistaken for fracture lines. The presence of marginal sclerosis or the absence of soft-tissue changes, such as hematoma or joint effusion, make it unlikely that a fracture is present. The periosteum provides a matrix for new bone formation during normal bone growth and in response to bony injuries. The periosteum is visible on CTscans only if it is thickened (due to reparative processes) or elevated (by hematoma ortumor). Muscle tissue has a CT attenuation of approximately 50 HU. Muscular atrophy and fatty infiltration lead to an increase in the fatty tissue located between muscle bundles. Thus, a decrease in the CT attenuation of muscle is almost always a pathologic sign reflecting prolonged immobilization, a neuromuscular, or a metabolic disorder. In particular, poliomyelitis causes a typical pattern in which the muscle volume is partially retained but muscle attenuation is similar to that of fat. Articular cartilage, ligaments, menisci, and the glenoid rims of joints have moderately higher attenuation values than muscle tissue (70-90 HU). They are difficult to evaluate with CT unless CT arthrography with intra-articular contrast instillation is performed. Joints are surrounded and shaped by a variable amount of structural fat, which functions as a shock-absorbing cushion during movements (elbow) and may assist in forming the boundaries of the joint cavity (knee joint). Only this peri- and intra-arterial fat shows a distinct contrast to surrounding soft tissues on unenhanced CT scans.
of bone Physiologic (children, adolescents) Chronic blood loss Disturbance Hematologic
of hematopoiesis disorders (leukemia,
Spinal Column lymphoma)
Storage diseases Tumor invasion
(metastases,
bone tumors)
The spinal column normally consists of 7 cervical, "12 thoracic and 5 lumbar vertebrae. In a substantial number of patients transitional
25 Musculoskeletal System lar facets; on axial images the anterior facet is the superior facet of the lower vertebra, while the posterior facet is the inferior facet of the upper vertebra (Fig. 25.1). At the cervical level the orientation of the facets is in the coronal plane. The inferior facets are continuous with the laminae that form the posterior arc and the spinous process. The neural foramina are oriented in a 45' oblique plane and accommodate the eight pairs of nerve roots. Each pair of roots exits cephalad to the corresponding vertebral body.
anomalies exist, especially in the lumbosacral region. They are often difficult to evaluate on lateral scout views and correlation with plain films is important for adequate assignment of the level of abnormalities. For thoracolumbar fracture classification based on Cl, the threecolumn concept of Denis is commonly used (Table25.3). Cervical
Vertebrae
The atlas and axis have a different embryological origin from the lower cervical vertebrae. The atlas (Cl) does not have a vertebral body but instead is composed of a ring with an anterior tubercle to which the anterior longitudinalligament inserts. The bulky lateral masses form the articulation with the axis and the occipital condyles. Rotational movement is possible in the midline atlantoaxial joint that articulates with the dens. The dens is held in apposition by the transverse ligament that encircles it. The normal atlantodental space is < 3 mm in adults and <5mm in children. The axis (Cz) consists of a large corpus that is fused to the dens. A fusion defect may result in an os odontoideum. The lateral masses form the axially oriented atlantoaxial articulation cranially, and a coronally oriented facet joint with e3 caudally. The lateral masses of Cl to Cf contain transverse foramina for the vertebral arteries and veins, C e3 to C] all have a common origin and similar anatomic features. Like other vertebrae they a have a ventrally located body (corpus) that connects via short pedicles to the articular pillars. The vertebral bodies increase in size from e3 to Cl, and each has a lateral uncinate process. These processes form the uncovertebral joint, which is located ventromedial to the neural foramen. The articular pillars contain the articu-
Table 25.3
Three-column
model
Anatomic Anterior
Central
Posterior
column
column
column
of spinal
anatomy
Thoracic
lumbosacral
(Denis,
1983)
Optimum CTvisualization
components
Vertebral Posterior
body posterior margins part of anulus fibrosus
Posterior
longitudinal
Posterior
portions
Sagittal
and pedicles
and coronal
reformations"
Axial sections Sagittal reformations
ligament
of vertebral
flava interspinous
Vertebrae
Because of transitional vertebrae, there are between 4 and 6 lumbar vertebrae. The vertebral bodies make up the largest part of the vertebrae and thick pedicles form posterosuperiorly. The neural foramina run in a coronal plane, and the nerve roots exit caudad to the corresponding body. The lateral masses form the facet joints that lie in a progressively sagittal plane, except at L5-S1 where the configuration is again near
Anterior 2/3 of vertebral body Anteriorior 2/3 of anulus fibrosus Anterior longitudinal ligament
Ligamenta Supra-and
Vertebrae
The corpus of the thoracic vertebrae is larger and has large, obliquely orientated transverse processes with articulating facets for the ribs. At the lateral margin of the corpus there is another special fovea for the head of the rib. The pedicles are positioned posteriorly on the superior part of vertebral body and the neural foramina are located laterally and run in a coronal plane. The nerve root pair exits caudad to the corresponding vertebral body. As in the cervical spine, the orientation of the facet joints is in the coronal plane. Posteromedially, the laminae extend from the transverse processes and fuse in the midline at the spinous process.
arch and facet joints
Axial sections Sagittal
ligaments
and coronal
reformatior
Anatomy
coronal. The laminae are flat and do not overlap as in the thoracic spine. Transverse processes extend laterally and are more prominent at LI and L5. They contain a processus costarius that serves for muscle attachment. The sacrum is made up of 5 fused vertebrae, but there may be residual disc spaces in between. It courses in an oblique coronal plane, but in some patients this may change to a near axial orientation (sacrum acutum). It contains 3 dorsally located crests, and 4 pairs of neural foramina are located laterally to the laterally situated intermediate crests. •
Nerve Roots, Ligaments Intervertebral
and
discs
The nerve roots pass through the lateral recess of the spinal canal and emerge from the neuroforamina in root sheaths that are separated from the rest of the dural sac by a layer of epidu-
ral fat (Fig.25.1a). Rarely, two consecutive nerve roots may emerge from the same neuroforamen {conjoined roots'}. These doubled root sheaths are larger than normal (differential diagnosis: nerve sheath cyste I'arlov cyst, see Fig.25.51b), usually fill most of the lateral recess, and cause narrowing of the epidural fatty layer. The spinal nerves leave the spinal canal through the cranial portion of the neuroforamina, where the diameter of the neuroforamen is largest. This can be best appreciated on reformations in a parasagittal plane perpendicular to the course of the nerve roots. The neuroforamina are bounded above and below by the pedicles, posteriorly by the interarticular part of the vertebral arch, and anteriorly by the vertebral body. A number of ligaments stabilize the spine. The anterior longitudinal ligament runs anteriorly to the vertebral bodies and is important for
-Basivertebral vein
----\-~-m
Psoas muscle
~~---
Internal vertebral venous plexus
~---:--'~~---'---
L4 nerve root Duralsac -
Transvers process
--
L4 spinous -----=----tprocess
Quadratus lumborum
muscle
Erector spinae muscles
a ~~""li";;=------
Anterior tudinal
---\---\\------
longiligament
Intervertebral
Anulus fibrosus ---~
disc
Dural sac Cauda equina L4/ L5 fa cet
---'--------,f------P~\.
joint
\
------;--'----;---
Superior articular process of L5
'-----
Inferior articular process ofL4
'----
Ligamenta flava
Epidural fat ----Fig. 25.1 Schematic cross sections through the fourth lumbar vertebral body (a) and the L4/L5 intervertebra I disc space (b).
L4 spinosus process
b
L5 nerve root
933
25 Musculoskeletal System stability following fracture. The posterior longitudinal ligament runs over the posterior surface of the vertebral bodies and forms the anterior boundary of the spinal canal and a barrier for disc pathology. The flava ligaments connect two laminae laterodorsally. The ligamentum flavum is important in degenerative processes. when it may hypertrophy. Other ligaments are normally not seen with cross-sectional techniques. The intervertebral discs of the lumbar spine (70 ± 30 HU) range from 5-13 mm in height. The L5-S 1 intervertebral disc space is normally the narrowest. The intervertebral disc comprises a soft central portion, the nucleus pulposus, which is composed of collagen fibrils and mucoprotein gel and lies eccentrically and somewhat posteriorly; and a firm fibrocartilaginous ring, the annulus fibrosus, which surrounds the nucleus pulposus and is reinforced by the anterior and posterior longitudinal ligaments. On axial sections the intervertebral discs have a reniform shape that is concave posteriorly (Fig.25.1b). However, a flat posterior surface of the L4-L5 disc and a slightly convex surface of L5-S1 is normal. The discs become progressively thinner in the thoracic and cervical spine and assume a transverse elliptical shape. Standard CT visualization of these regions is often unsatisfactory due to partial volume effects from the vertebral end plates. CT display improves with thin-section multislice CT. In the cervical and thoracic spine 0.5 mm sections provide exceptional image quality but require an increased dose with 4-detector-row scanners. Sagittal sections from m ultislice CT data sets are able to display the intervertebral disc in a similar fashion to MRI but spatial resolution is superior and may allow for better delineation of the nerve roots, especially when sagittal oblique sections through the foramina are reconstructed. Unlike MRl, an individual adaptation of cut planes to the anatomic situation is possible after data acquisition has been performed. The posterior margin of the vertebral bodies bears the canal of the basivertebral vein. The transvertebral veins extend anterolaterally from the basivertebral vein, forming a Y-shaped bifurcation that can occasionally be seen within the vertebra (Fig. 25.1 a).
Large root sheaths or root cysts (isoattenuating to the dural sac) should not be confused with a herniated disc (hyper attenuating to the dural sac) (see Fig.25.51b), and the intraspinal venous plexus at the posterior margin of the vertebral body should not be misinterpreted as disc tissue. The venous plexus is isoattenuating to muscle tissue on unenhanced CT scans, usually has a distinct quadrangular shape, and can be traced along the posterior margin of the vertebral body to its connection with the basivertebral vein (Fig25.~a),
Upper extremity Shoulder
and Upper Arm
The humeral head has a smooth, round shape when imaged in axial sections at the level of the coracoid process. On lower sections at the level of the capsular attachment just above the anatomic neck, the humeral head is flattened and occasionally slightly concave on its posterior aspect. The bicipital groove is visible anteriorly at this level. appearing as a depression between the greater and lesser tuberosities. It is highly variable in its shape and depth. The groove occupies the anteromedial quadrant in internal rotation and the anterolateral quadrant in external rotation. The glenoid articular surface has a marked degree of physiologic anteversion. The cartilage lining the glenoid terminates at the glenoid labrum. CT arthrography may demonstrate a small notch at the anterior and posterior attachment of the glenoid labrum, which should not be mistaken for a tear. The anterior and posterior portions of the labrum have a triangular shape in axial section. Typically the anterior part is sharply tapered while the posterior part shows a broad-based, almost rounded shape (with a considerable range of anatomic variation). The anterior attachment oftheJoi'nC capsule is variable and extends a variable distance toward the neck of the scapula (Fig. 25.2). Posteroinferiorly, thejoint capsule attaches directly to the labrum or to the junction of the labrum and bone. The humeral attachment of the capsule is located between the anatomic and surgi-
Upperexlfemfty
II
III
Fig. 25.2 Schematic diagrams illustrate the variable anterior attachment ofthe shoulderjoint capsule (CT arthrography). Type I is attached on to or nearthe labrum, while types II and III attach further medially on the scapular neck. Reportedly, the type III attachment predisposes the shoulder joint to anterior instability.
cal neck, with the result that CT arthrography normally does not demonstrate contrast material distal to the anatomic neck except for a small amount surrounding the long biceps tendon. The upper part of the capsule forms the subscapular recess, which is traversed by the tendon of the subscapularis muscle. The inferior extension of the capsule forms the axillary recess. Occasionally the subacromial bursa communicates with the subdeltoid bursa, but neither of these bursae communicates with the interior of the joint. In the upper arm, the midportion has a protuberance where the deltoid inserts. During the course the humerus is round, but the medial side is relatively flattened. There may be lobulated marrow spaces. Distally the humerus becomes flat and ends in the elbow at the medial lateral epicondyles and the curved trochlea and capitellum. Just above this, the deep olecranon fossa accommodates the olecranon process when the arm is fully extended. •
Elbow and Forearm
In the elbow, the ulna contains a deep fossa that articulates with the trochlea, and anteriorly it extends in the coronoid process. The fossa has a
central groove that should not be mistaken for a fracture. The radial head is a slightly concave disc that articulates with the capitellum. The radial neck tapers slightly before the radius forms the posteromedial tuberosity for the biceps attachment. The medial and lateral epicondyles are important attachment sites of the flexors and extensors of the forearm and wrist respectively. With the arm in the supinated (anatomical) position, the proximal radial and ulnar shafts lie completely parallel, and allow for good evaluation of the proximal radioulnarjoint. It is important for ease of evaluation that pronated positions are avoided as much as possible. In the diaphysis, the interosseous membrane can be seen connecting the two bones, which have a triangular shape in cross-section. Distally the radius broadens in all directions to form the radiocarpal joint, and the ulna becomes rounded and ends in the styloid process medially. •
Wrist
The wrist is a complex three-dimensional structure that requires routine coronal, sagittal, and oblique planes for proper evaluation. In the distal radioulnar joint, the radial surface is concave medially to accommodate the convex distal ulna. This configuration should be maintained throughout pronation and supination. On the dorsal surface of the radius, the tubercle of lister is readily seen, which separates groups of extensor tendons. On the lateral side the radius ends in a triangular styloid process, which also extends dorsally creating a volar tilt of the radiocarpal articular surface. The distal radius contains two fossae (scaphoid and lunate) for the radiocarpal articulation, while the distal ulna articulates via a complex of ligaments and a fibrocartilage disc (triangular fibrocartilaginous complex) with the proximal carpal row. The carpus consists of two rows of articulating bones (remembered by the mnemonics PITRILUSCA proximally and HACATITUM distally), which form the carpometacarpal joints distally. In the sagittal plane, the longitudinal axes of radius, lunatum, and capitatum normally form a straight line. In the axial plane, the characteristic mediovolar hook of the hamate is easily appreciated. Together with the more volarly located pisiform bone this makes up the bony part of the carpal tunnel.
935
25 Musculoskeletal System Pelvis and Lower Extremity •
•
Pelvis
The pelvis is composed of the sacrum, which articulates at the sacroiliac (SI) joints with the two hemipelvic bones. These pelvic bones are each comprised of the iliac wings, the ischium, and the pubis and meet at the symphysis anteriorly. The SIjoint has a parallel synovial portion (anteroinferior third) and an inverted V -shaped ligamentous portion (posterosuperior two-thirds), and is best evaluated on oblique axial and coronal MPR. The acetabulum is formed by the ilium superiorly, the ischium posteroinferiorly, and the pubis anteroinferiorly. On axial views two oblique articular facets and a central fossa can be appreciated. Functionally, the periacetabular portion of the iliac bone is comprised of an anterior pillar and a posterior pillar, which transmit forces from the sacroiliac j oint through the acetabulum to the femoral head (Fig. 25.3). This arrangement is most clearly appreciated on axial scans through the acetabular roof and through the central fovea of the femoral head. Fracture lines and intra -articular fragments are often seen more clearly on sagittal and coronal reformations, while volume-rendered 3D displays (with or without exarticulation of the femoral head) are useful for fracture mapping and preoperative planning.
Axial scans ofthe/emomf head at the level of the fovea demonstrate a central stellate figure reflecting the trajectorial arrangement of the bony trabeculae (the asterisk sign). The trabecular pattern can be seen in longitudinal section on coronal reformations (see Fig.25.31a). At the junction of the trochanteric region and femoral neck, there is a relative lack of a trabecular architecture (Ward's triangle), which makes this portion of the femoral neck particularly susceptible to fractures. The greater and lesser trochanters are seen laterally and ventromedially as bony protuberances. The femoral shaft has a variable oval shape that is flattened posteromedially in its proximal part and laterally in its more distal part. The posterior linea aspera protuberance serves as the insertion of the adductor muscles and as the origin of the medial and lateral vastus muscles. The femur flattens distally and forms the medial and lateral femoral condyles separated by the intracondylar notch. •
b
Fig. 25.3 Surface-shaded views of a hemipelvis demonstrating the acetabulum (a) and the quadrilateral plate (b). Anterior collumn = 1, posterior collumn = 2.
Knee and Lower Leg
The femoral condyles articulate with the tibial plateau. The patella glides in the femoral trochlear groove anterior to the condyles. The fibula is located posterolaterally to the tibia and is connected to it by an interosseous membrane. The ligaments and menisci are hyperattenuating structures that are not well seen with CT unless CT arthrography is performed (see Fig. 25.36). •
a
Hip and Upper Leg
Ankle
The ankle is formed by the distal tibia and fibula which form a hinge into which the talar dome fits. The talus articulates with the calcaneus inferiorly as the subtalarjoint, and with the navicular anteriorly. The subtalar joint has three separate facets, the posterior, middle, and anterior. The middle one is the articulation of the talar head with the sustentaculum tali on the medial calcaneus. Lateral to this is the cone-shaped sinus tarsi, which contains fat and ligaments. The Chopart line describes the line between talus/calcaneus and the rest of the foot, while the Lisfranc line connects the various tarsometatarsal joints. The entire tarsus is an anatomically complex region in which the joint spaces can be dif-
General ficult to visualize on standard radiographic projections. When thin-section spiral or multislice CT scanning is performed. high-quality coronal and sagittal images can be reconstructed in any plane from an isotropic axial data set acquired with thin collimation. These reformatted images can be made perpendicular to all the separate joint spaces so that eachjoint can be evaluated in two planes (fable 25.4).
Table
25.4
Evaluation
of the
tarsal
joints
in various
planes
Calcaneocuboid joint (CCJ) Talonavicular joint (TNJ) Metatarsus Coronal
and forefoot
(Lisfrancjoint)
sections
Anklejoint (talocruraljoint) Anterior
and posterior
subtalarjointswith
taculum Lateral and medial malleoli Peroneus tendon and flexor tendon Sagittal
susten-
of great toe
sections
Anklejoint (talocruraljoint) Subtalarjoint angle)
(compression
fracture,
tuber-joirt
Calcaneocuboid joint (CCj) Talonavicular joint (TNJ) Lisfrancjoint
Examination
Technique
General An optimum examination ofthe skeletal system in orthopedic and trauma settings requires the use of thin slices that are tailored to the region of the scan (Tables 25.5. 25.6). Spiral acquisition with thin slices allows for generating good quality MPRs to display the region of interest on multiple planes. The short scanning time helps to reduce motion artifacts in patients who are in pain. Patient positioning must be comfortable. taking into account limitations due to pain or cast immobilization. For the proximal joint. scanning is usually limited to the axial plane. In the distal joints like wrist or ankle. non-axial positions can also be achieved. Immobilization with straps. blankets. or similar is important to obtain motionless studies. Plaster casts need not be removed but mAs settings should be increased to accommodate for the higher attenuation in the plaster cast. With single-slice spiral CT, the collimation is set to 1-3 mm, depending on the body region
and the scan rage that has to be covered. The pitch factor should preferably be 1.5 or higher. A pitch as high as 3 can be used with a 1 mm slice collimation, e.g., for scanning the knee or the whole cervical spine. Compared to thicker collimation, it will yield improved z-axis resolution but will also introduce a slightly greater number of stair-step artifacts. It is thus indicated only when a large range has to be covered that cannot be scanned with a lower pitch. The reconstruction interval should be approximately onehalf of the section width to permit optimum MPR and 3D reconstructions. With multislice CT, 1-1.5 mm collimation (4- to 16-slice scanners) should be used for most indications in the musculoskeletal system. As an exception, thicker sections (4-16 x 2 mm or 4-8 x 2.5 mm) may be used for scanning the shoulder, hip, lumbar spine, or pelvis, especially in obese patients. For the highest possible resolution in the cervical spine, or in smaller joints like the ankle, knee, elbow, and wrist, as well as in the hands or feet, section collimation may be reduced to 0.5-0.75 mm. This will generate an
937
938 25 Musculoskeletal System Table 25.5
Suggested
protocols
for CT scanning
of the musculoskeletal
system
Position
Supine with elevated arms. comfortable and pain-free position Spinal CT or QCT: knee flexed (over padding) Cervical spine: arms adducted to body; downward traction may be applied if needed Shoulder: other arm over head. scan region close to center of gantry CT arthrography of shoulder: arms adducted to body. scans acquired in internal and external rotation Hand or elbow: prone position may be used; arms directed cephalad. scan region placed in center of gantry. palms up Foot or knee: scan region placed in center of gantry; oblique position (knee flexed) - not mandatory
Scan range
Spine: Pelvis: Joints:
Respiratory
phase
affected + adjacent segments Acetabulum only or complete bony pelvis Main fragments. if possible complete fracture
Shallow respiration Sternum: inspiration
Windowing
Bone: In osteoporosis: Soft tissues (noncontrast): Soft tissues (postcontrast):
Scan parameters
single-slice S€/TF/RI
multislice SC
scanning P'
1-2/3-4/1 1/2/0.7 2/3-4/1 2/3-4/1 2/3-4/1
0.75-1.25 0.5-0.75 0.75-1.25 0.75-1.25 0.75-1.5
>1 <1 >1 >1 <1
Pitch < with short scan range
Shoulders (CT arthrography)
2/4/1.5 3/4/2 2-3/3-5/2 3/3/1.5
0.75-1.25 1.25-2.5 1-1.5 1.25-2.5
>1 >1 <1 <1
Slim patients Obese patients High dose scans. dose modulation Obese patients
Extremities: Elbow Handjwrist Knee Foot Long bones
1/2/1 0.5/1/0.5 1/2-3/1 0.5-1/1-2/0.5-1 3/5-6/2
0.75-1.25 0.5-0.75 0.75-1.25 0.5-0.75 0.75-1.25
<1 <1 <1 <1 >1
MPR orthogonal to anatomic structures Ultra high resolution kernel Higher pitch with longer scan range Ultra high resolution kernel Include adjacent joint
2/4/1.5
0.75-1.25
>1
Slim patients. longitudinal dose modulation
3/4.5/1.5 5/10/5
1.25-2.5 0.75-1.25
<1 >1
1.25-2.5
<1
0.75-1.25 1-1.5 1.25-2.5 1.25-2.5
> 1 >1 >1
Spine: Cervical spine C1/C2 only Thoracic spine Lumbar spine Disc herniation Pelvis/shoulder: Pelvis/acetablulum
Poly traumatized Cervical spine
W/L = 2000/400 (increase W as needed) W/L= 1500/200 (reduce W as needed) W/L=300/30 W/L=400/60
Remarks
Place neck in isocenter of gantry
Pitch < with short scan range Pitch < with short scan range High dose scans. thicker sections in obese patients
patients:
lower C-spine Chest + Abdomen
Other: Tumors: Soft-tissue infection CT torsion assessment Quantitative CT
2-5/4-8/2-4 5/10/5 5/10/5 8-10/-
Separate scan in obese patients
Slim patients. longitudinal dose modulation Obese patients; reconstruct SW= 5-7.5 firstl
SC depends on scan range smaller SC for short scan range Proximal and distal low-dose scans Single low-dose scan through each of L1-L3 follow manufacturer's
recommendations
General 939 Contrast injection Abscess. DD hernation/ scar Tumor staging
1-slice
4-slice
V/F/D
V+N/F/D
16-slice
V+N/f/DX'
120/2/80
120+50/2/80
120 + 50/2/90
120/4/40
120 + 50/5/40
120 + 50/5/20A
Rl= recon. increment (mm). j! :: scanning direction, width (mm). MPR= rnultiplanar reformats, axial = axial sections, cor= coronal, saq= sagittal eM::: contrast material, Contrast concentration = 300 rnqfrnl iodine V "volume of eM (ml), N =volume of saline flush {ml} F=flow rate (mils), 0 = start delay, X = trigger region J Pitch P=TFf(NxSC) with 4-slice and 16-slice scanners b MPR are reconstructed from a "secondary raw data set" with SWIRl'" 1-1.5}0.7 or 0.5-0.8/0.4 c read: 20A = start delay 205 after reaching 50 HU enhancement in the aorta SC:: slice collimation
Table 25.6
(mm).
TF= table
feed (mm/rotation).
Suggested MPR and 3D protocols for musculoskeletal Axial SW/RI
SW '" effective
section
CT
Coronal SW/RI
Sagittal SW/RI
3D
Remarks
1/1.5
VRT'
Higher pitch with longer scan range
1/1
1/1.5
1/1.5
VRT'
Adjust axial sections to plane of atlas and axis
Thoracic spine
2/3
2/3
2/3
VRT'
Coronal CPR only for burst fractures
Lumbar spine Disc herniation
2/3 3/3
2/3
2/3
VRT'
Coronal CPR parallel to spinal canal
3/4
2/3
3/4
VRT'
Sagittal only for sacrum fractures
2/2
2/2
2/2
VRT
3D display of hemipelvis or exarticulated joint
3/2 2/2
2/2 2/2
3/2 3/2
VRT'
Coronal parallel to scapula
Spine: 1.5/2
Cervical spine C1/C2 only
Axial parallel to disk spaces
3/3
Pelvis/shoulder: Pelvis/ Acetabulum Shoulders (CT arthrography)
Sagittal for subacromial space
Extremities: Elbow
1.5/2
1/1.5
1/1.5
VRT'
MPR orthogonal
Hand/wrist
0.5-1/1
0.5-1/1
0.5-1/1
VRT'
Axial sections not mandatory
Knee
2/3
1.5/3
1.5/3
VRT'
Axial parallel to tibia plateau; 3D joint exarticulation
Foot
0.5-1/1
0.5-1/1
0.5-1/1
VRT'
MPR orthogonal
2/3
2/3
VRT'
Longitudinal
1/1.5
1/1.5
VRT'
Coronal only for atlas and axis lesions
2/3
VRT'
Long bones Polytraumatized
to anatomic structures
to anatomic structures
MPR with adjacent joint
patients:
Cervical spine
1.5/2
Chest + abdomen
7.5/7.5
Spine Ribs and sternum
Immediate review of organ injuries
2/3
VRT
Axial sections only for involved segment Separate VRT for anterior and posterior chest
VRT'
MPR for longitudinal
Other: Tumors:
5/5
2/3
2/3
extent
a 3D display only necessary from complex dislocation of fragments. VRT is preferred but SSD is also possible MPR require prior reconstruction of an overlapping thin-section data set (secondary raw data set) The number of MPR sections can be limited by focusing on the involved area (even for axial orientation)
I
25 Musculoskeletal System almost isotropic data set that enables the highest quality MPR in any direction. It is important to position the region of interest exactly in the center of the gantry aperture to reduce cone beam artifacts. A very th in collimation, however, requires an increase in mAs settings for most applications and will provide substantially increased dose to the patient in 4-slice scanners. Only with 8- or 16-slice scanners will the radiation exposure be significantly reduced, making ultra th in scanning an option for all low absorption regions in the musculoskeletal system. The musculoskeletal system is one of the few exceptions where lower pitch factors can be advocated for multislice CT. A pitch factor P of 0.7-0.9 (P*=2.5-3.5 for 4-row scanners) will lead to less step artifacts with thin-section scanning (see Figs. 2.9 and 4.7). Reconstruction increments should be chosen to be 50-60% of the reconstructed section width. For fast image review (such as in poly trauma patients), nonoverlapping axial sections of 2-3 mm thickness may be reconstructed first, but we prefer to use a "secondary raw data set" of overlapping thinsections to reconstruct a problem-adapted set of 1-3mm thick MPR images in the axial, sagittal, or coronal plane for image analysis. When scanning at I mm collimation or higher, linage quality of MPR can be optimized using the principle of obliqUity. The extremity should be positioned so that the scanning plane is oblique to the joint surfaces of interest, and 45' obliquity is considered optimal. To maximize in-plane resolution, the joint should be placed in the isocenter of the gantry. If bone is the only tissue of interest, the exposure dose can be substantially reduced in patients who are not obese, because the bone window setting (width of 1000-2000 HU) will substantially reduce visualization of image noise. Adaptive dose modulation is highly recommended for the shoulder or pelvis. and z-axis modulation of exposure dose should be used whenever regions with throng variations in xray attenuation have to be examined, such as the neck and chest or chest and abdomen. The use of a high-resolution algorithm will improve the quality of axial sections and MPRs, but a standard or soft algorithm should be used for shaded surface displays or volume rendering to reduce image noise. For the shoulders and the pelvis, kernels with less high resolution should be employed for axial and MPR images because
of increased image noise in these high-absorption regions. For optimum spatial resolution, small fields of view that focus on the region of interest can be recommended. If available, a 1024' matrix may be used with a larger field of view.
MPRviews are tailored to the specific problem, and most orthopedic and trauma cases will require evaluation on a second plane (Table 25.6). The MPR should be adapted to the individual position of the joint and not to the patient table. Coronal curved planar reformations are helpful for the spine paralleling the course of the spinal canal. Three-dimensional images are important for the reporting of CT findings in patients with complex fractures and in cases where anatomic details are obscure. Such images are also valuable for treatment planning and teaching purposes. They should be obtained routinely for the evaluation of acetabular fractures. Nowadays, volume rendering is the preferred technique for the musculoskeletal system, yielding improved visualization or better demonstration of subtle findings that may alter management in a significant number of cases. In patients with metallic implants, linages should be obtained with high kilovoltage (140 kVp), high mAs, and thin sections. The attenuation of metal is reduced at high kVp and produces less beam hardening. Thin sections reduce partial volume effects and therefore also diminish streak artifacts from beam hardening. In addition, they allow for reconstructing multiplanar reformations that suffer less from streak artifacts than the original images (see Fig. 25.67). Higher effective mAs settings can be accomplished with dual-slice and multislice CT by using overlapping spirals with pitch factors P< I. Artifacts from gold (teeth) are highest. Titanium implants produce the least artifacts, and are therefore Increasingly favored by surgeons. If available, special algorithms for metal artifact correction (e.g., fast iterative deblurring, extended CT scaling) should be used (see Fig. 25.68).
Skeletal Imaging
Spinal Imaging In trauma patients, scanning en bloc is generally preferred, which provides good quality MPR that is required for routine evaluation. For the cervical spine, collimations of 0.5-2 mm are employed, for the thoracic spine 0.75-2 mm, and for lumbar spine, 1-2.5 mm. The scanned regions are normally focused on a limited segment of interest, although CT is gaining importance in high-risk patients as a primary modality for evaluating the entire region of the cervical spine. With multislice CT, using thin-section scanning of longer spinal segments becomes feasible. In the setting of evaluating poly trauma patients, images of the cervical or thoracolumbar spine can be generated from the same data used for evaluation of the neck or chest and abdomen. Spiral and multislice CT evaluation of the spinal canal and intervertebral discs requires a high dose and thin sections. The standard technique uses a gantry angulation parallel to the studied disc space as determined from a lateral scanogram, but this may cause problems with distortion of MPR on some scanners. Scanning of the whole region en bloc is often easier and has the advantage that reformations can be adjusted to be exactly parallel to the disc space, even in scoliotic patients. Such a technique, however, increases the radiation dose to the patient. In patients with lumbar radicular complaints, the scan should cover the segment at the level of emergence of the suspicious spinal nerve as well as the segment directly above it so that the intraspinal nerve root can be evaluated. Scans at the cervical level should cover the suspicious segment along with the two adjacent segments. The coverage of each segment should extend from the inferior border of the upper pedicle to the superior border of the lower pedicle and should include the neuroforamina and lateral recess. !f a herniated disc or other abnormality is detected (e.g., isolated sequester), the full longitudinal extent of the lesion should be evaluated. In the detection of cervical disc herniation, intravenous contrast administration to enhance the venous plexus can result in a higher sensitivity for detection. Image reconstruction uses a small field of view to ensure optimum spatial resolution. In trauma patients, reconstruction with a high-res-
olution (bone) algorithm is normally used, but additional reconstructions with a standard algorithm are needed for evaluation of the paras pinal soft tissues. When scanning for disc disease, a standard to slightly smoothing reconstruction algorithm is used to keep image noise low. Additional high-resolution algorithms can be used for better evaluation of the bony canal and neural foramina. Sagittal or parasagittal reformations should always be reconstructed if spiral or multislice acquisitions are performed. By creating reformations with a thickness of 1-1.5 mm in the cervical spine and 2-3 mm in the thoracic or lumbar spine, image noise can be reduced substantially. Noise from the shoulders can seriously degrade the quality of CT images at the level of the cervicothoracicjunction. For this region, adaptive dose modulation and new 3D interpolation filters for image reconstruction (adaptive filtering) can substantially improve image quality and reduce dose requirements. For adaptive dose modulation, the maximum mAs setting should be increased by a factor of > 1.5-2 (depending on the manufacturer's implementation) to ensure a sufficient exposure dose during the lateral projections, while the program itself will substantially reduce exposure dose (often by a factor of> 10) for the projections in an AP direction that require less dose because of the reduced patient diameter in this direction (see Fig. 5.7). !fdose modulation is not available, MRI is a better primary modality for imaging in this region. Alternatively, sequential axial CT or longer rotation times (1.5-2.0 s) allowing higher mAs, can be used in the shoulder region.
Skeletal Imaging Upper Extremities In the shoulder, scanning should cover the range from the upper acromion to a few cm below the lower edge of the glenoid or as far as the tip of the scapula. To avoid movement, the distal part of the arm can be immobilized with sand bags, pillows, or tape. Adaptive dose modulation should be used whenever possible (see above) and thicker sections of 2-3 mm can be reconstructed (from the primary or secondary raw data set) to keep image noise within limits. With spiral and multislice techniques, MPR can
941
942
25
Musculoskeletal System
When an evaluation using MPRs from an be performed in planes that match the oblique isotropic dataset with the wrist in a neutral planes used in MRI. The abduction-exoration (ABER) position may improve the evaluation of position does not answer the specific clinical question, special positions from conventional CT the anteroinferior labrum in CT-arthrography. as suggested by Stewart may be useful. The scaVolume-rendered 3D is very useful in complex areas like the scapula or sternoclavicular joint. phoid long-axis view (wrist oblique with ulnar deviation) is optimum for scaphoid fractures In the elbow, scanning can be performed with the arm at the side, but more often it is and the oblique SIT projection (oblique with 45° performed with the arm overhead in slight flex- overpronation and ulnar deviation) is helpful ion (less dose requirements, better image qual- for evaluation the scaphoid-trapezium-trapeity). As this position is tiring, careful immo bili- zoideum (STT) complex (Fig. 25.4). When zation and adequate padding is required. When detailed evaluation of the distal radio-ulnar flexed positions are used it is important to joint is the question, examination of both wrists generate a (nearly) isotropic data set using thin is beneficial. slices and small pitches from which truly axial MPRs for the upper and lower arm-can be generated. Pelvis and Lower Extremities In the wrist, the advent of multislice CT with high quality MPRs has eliminated most of In the pelvis, complex trauma often requires detailed evaluation both of the pelvic wings and the need for cumbersome specialized positioning of the wrist. However, especially in single- acetabulum. For single-slice spiral CT scanning, 2/4/1.5 protocol yields excellent slice spiral CT, the quality of MPR may not be a (SC/TF/RI=) results but requires some 50 s scan duration to good enough to obviate scanning in multiple planes in some instances. Direct coronal or cover the whole pelvis. With older scanners this sagittal positions with the arm overhead are may require cutting the scan range in half (above the acetabulum). Multislice CT scanning relatively easy to perform, but immobilization of the entire pelvis is possible in less than 20s. and padding is crucial for optimal results.
Scaphoid long-axis view
Oblique SST projection
Overhead I prone
Overhead prone
45 ° overpronation Wrist in ulnar deviation
Scan plane
Fig. 25.4 Positioning of the wrist for scaphoid long-axis view and oblique STT projection (Stewart, 1992).
Po/ytroumo
With 4- and 8-slice scanners, 4-8 x 2.0-2.5 mm collimation provides sufficient information in most cases and is the preferred technique for obese patients. Optimum image quality, however, can be achieved by using 1-1.5 mm collimation on 4- to 16-slice scanners, and by creating a secondary raw data set from which further images are reformatted. In addition to axial sections, multiplanar reformations in two perpendicular planes through the fractured sacrum or acetabulum help define fragment position and fracture classification. The latter is vastly simplified if 3D reconstructions (SSD or VRT) of the injured hemipelvis or of the sacrum are performed. For complex pelvic fractures, we suggest always reconstructing 3D views in radiological standard projections (ala view, obturator view, inlet and outlet views, as well as AP, PA, and lateral proj ections). Exarticulation of the femoral head provides excellent visualization of the fractured acetabular surface (see Fig. 25.29). The evaluation of the complex three-dimensional anatomy of the ankle and foot also benefits from thin-section isotropic data sets and multiple MPR, especially in the calcaneus and talus (see Table 25.4). Orthogonal MPR in three planes are also important for complex knee trauma. Scanning with 0.5-1.25 mm collimation is best suited for this purpose. For the knee and the hindfoot, volume rendering improves anatomical insight and aids presurgical planning for treatment of complex fractures. Exarticulation of the femur or tibial head is simple with region growing procedures and is highly appreciated by the surgeons.
Poly trauma
Patients
CT is gaining an increasing role in the management of patients with multiple trauma. CT used to be a second-line modality outside the head, but spiral and multislice techniques have made it possible to use CT as a screening tool in highrisk poly trauma patients. The prerequisite for such an approach is a trauma unit with a close collaboration between radiologists and the surgeons who are responsible for the patients. Patient management and outcome can be substantially improved if a fixed diagnostic procedure is determined, which is followed for every patient who meets predefined inclusion criteria. These rules will vary from institution to
Patients
institution since scientific evidence for optimized procedures is not available for all clinical conditions. In addition, the speed of the examination is essential for deciding which techniques to use but it depends strongly on the available scanner technology. Older spiral CT units will therefore only be employed for specific questions such as screening for injuries of the cervical spine or screening for thoracoabdominal injuries, while the most recent generation of 16-row multislice CT units will have the potential to scan the whole patient, head to toe, with 1.25-2 mm sections in some 25 seconds. The latter approach may even make it possible to avoid conventional radiographs of the extremities and may thus be a technique that is performed shortly after stabilizing the patient. Longitudinal dose modulation, however, is mandatory in order not to expose the patient to unnecessary radiation. Patients with suspected head trauma will be referred to CT in the time immediately following clinical stabilization. If an additional chest or abdominal trauma is suspected, a long sccmogram should be performed first, which gives information about malpositioned chest tubes, life-threatening pneumothorax, and massive intra-abdominal air, which warrant immediate intervention. An additional lateral scanogram may provide information about major spinal injuries. After the (conventional) head scan, an examination of the total cervical spine has been shown to be well above 90% accurate in detecting fractures as compared to some 80% sensitivity for conventional radiographic series of the cervical spine. A cross-table lateral radiograph of the cervical spine may be performed first. CT can be time and cost effective if the criteria in Table 25.7 are met. For CT, the upper cervical spine from the occiput to above the shoulders should be examined at 120 kVp with less dose and thinner sections (2/4/1.5 protocol with spiral CT, 4x 116 protocol for 4-slice CT). A second scan of the lower cervical spine through the shoulders to Th4 should follow with higher dose and 140 kVp. For thin patients the same protocols as for the upper cervical spine can be used but for obese patients, a 3/4.5/1.5 protocol can be advocated with spiral CT, and a 4 x 2-2.5/ 6-7.5 protocol with 4-slice CT. With spiral-CT, additional thin-section CT of selected regions may be needed in 5% of cases. If available, longi-
943
25 Musculoskeletal System tudinal dose modulation should be used, which increases dose for the shoulders and allows for including both regions in one scan. CT screening of patients with a high risk for chest and abdominal injuries involves a spiral scan of both regions with a 5110/5 protocol. Post-contrast scans improve the detection of abdominal hematoma, which may appear isoattenuating to liver parenchyma (see Fig. 16.9). In addition, contrast-enhanced CT helps detect aortic injuries, active bleeding, and parenchymal injuries to the liver, spleen, kidneys and pancreas. Hematoma may appear relatively hypodense on post-contrast scans but measurement of CT numbers still allows for secure identification of hemorrhage. For this reason, only contrast-enhanced scans of the chest and abdomens are required. For suspected parenchymal damage, a portal phase scan with a protocol of 100/3/60 is sufficient, but suspected aortic injuries or active bleeding require an additional arterial phase scan. This scan may focus only on the chest or the abdomen, depending on the findings at chest radiography (or the scanogram) and abdominal ultrasound. A radiologist should be present as the images are reconstructed, and should immediately review them for the presence of life-threatening complications (hemorrhage, organ injuries, pneumothorax) as well as for obvious spinal and pelvic injuries. If there are no complications requiring immediate treatment, additional scans with thinner sections may be added for frac-
Table 25.7
Clinical decision parameters for primary
use of CT in cervical injury mechanism High speed
motor
spine
injury
(Hanson,
2000)
parameters vehicle
accident
(" 35 mph or
55 km/h combined impact) Crash with death at motor vehicle accident Fall from height (s- 10 feet or 3 m)
scene
Clinical patient parameters Significant
closed
hemorrhage
head
injury
or intracranial
on head CT
Neurological symptoms or signs referred cervical spine Pelvic or mufiple extremity fractures
to the
Anyone criterion places the patient in a highrisk (>5%) group and selects patient for primary cervical spinal CT
tures of the spine or pelvis. Under such conditions, the patients can be on and off the table in less than 10 minutes. To speed up patient positioning and reduce the risk to the patient, a radiolucent support should be used in the emergency room that can be moved on to the CT table together with the patient. With multislice CT scanning, thinner sections can be used even for scanning the chest and abdomen. Depending on the size of the patient, a 4xl to 4x3.75 mm (16xl-2mm) collimation may be used. Suspended respiration is usually not required. Longitudinal dose modulation is highly advantageous in these patients but a high enough dose setting for the shoulders and pelvis has to be chosen. If the scanner allows doing so, even the cervical spine can be included when a 4 x 1 mm collimation with high enough dose is available for a scan duration of some 60 seconds. It is important, however, to first reconstruct thick sections from the raw data (5-7.5 mm) with identical increments to have a quick feedback about potential life-threatening complications. If thin-section scanning is performed, the patient can leave the table immediately after the scan is finished because all information can be gained from the raw data set. If thicker collimation is chosen for data acquisition (see Table 1.8), additional thin-section scans may become necessary in selected patients. After the patient is off the table, thinner overlapping sections (secondary raw data set) should be reconstructed that serve as the basis for sagittal reformations of the thoracic and lumbosacral spine (see Figs. 25.16 and 25.25). These images are excellently suited for detection and display of spinal injuries, in particular fractures that involve the middle column with the posterior surface of the vertebrae. Additional MPR are reconstructed wherever required. Volume-rendered images of the ribs and sternum can improve fracture detection (see Fig. 25.24). Volume-rendered images of the pelvis give an excellent overview of major fragments and the type of pelvic injury (see Fig. 25.30). They can probably substitute for conventional radiographs of the pelvis in these patients.
Torsion IAssessment
TorsionAssessment
CTArthrography
CT provides a standardized and reliable means With the increasing role of MR arthrography, the role of CT arthrography is diminishing. It is for measuring torsion deformity in long bones (Fig. 25.5). With the increasing use of inmost often used in the shoulder for evaluating the glenoid labrum and j oint capsule in patients tramedullary fixation in long tubular bones with shoulder instability. The examination (e.g., medullary nailing), CT has assumed a starts with double-contrast arthrography of the greater role in torsion assessment after fracture shoulder, in which contrast material (approxi- treatment and prior to corrective osteotomy. CT assessment of the torsion of a long bone mately 2-3 ml of a 300 mg IIml compound, more if diluted) and air (12-15 ml) is injected requires scanning of the proximal and distal into the inferior third of the joint capsule from joint regions. Several contiguous 8-10 mm sections or short spiral acquisitions through the an anterior or posterior approach. Iso-osmolar nonionic contrast material is to be preferred be- proximal and distal bone ends are sufficient. cause there is less dilution due to osmotic ef- Only a very low dose (CTDlvoi <2mGy) is refects if the patient has to wait longer than quired for torsion assessment. In more complex cases, a low-dose spiral or planned between arthrography and CT scanmultislice scan of the whole bone can be perning. This study is immediately followed by the formed with the additional option of displaying CT examination, which consists of two scans the full anatomic range with multiplanar reforwith internal and external rotation of the af- mations or 3D displays to evaluate dislocations fected arm. Alternatively, placing the patient in or axis deviations. Care has to be taken that the longitudinal a prone 45° oblique position has been suggested to ensure a homogeneous spread of the air to axis of the bone is perpendicular to the scan plane. Given the large range of inter- and inthe anterior and posterior compartments. When imaging the hip or knee, it has been traindividual variations, torsion deformity is shown in MRI that applying traction to the leg assessed by comparison with the opposite side. with a weight attached to the ankle can improve Average values stated in the literature are useful results in evaluating the acetabular labrum or only as guidelines; in the case of the femur, for example, the standard deviation is as much as cartilage.
Femur
Tibia
Humerus
(blue)
Line through center of femoral head and neck axis
Line parallel to posterior border of tibia condyles
Line through center of humertal head and bicipital groove
Aecond axis (black)
Tangent to posterior border offemoral condyles
Normal values
20° to 25°ofanteversion (AT) (range: 6 °-40 0)
Transverse axis of ankle joint (center of pilon to center of distal tibiofibular joint) 30° to 35° retroversion (RT) (range: 16°-50°)
Line midway between tangents to anterior and posterior trochlear borders 60° to 70° retroversion (RT) (range: 40° -80°)
Pathologic
> 15° discrepancy
> 15° discrepancy
<40° of retroversion
between sides
between sides
First axis
Fig. 25.5 Assessment of bone torsion with CT.
945
25 Musculoskeletal System ±20°. More than a 15° discrepancy relative to the opposite side is considered pathologic for the lower extremities. Quantitative CT Bone Densitometry
Bone densitometry with quantitative CT (QCT) is based on a single axial scan (SC=8-10mm) through the center of each of the first, second, and third lumbar vertebral bodies (Fig.25.6a). The average CT numbers measured in the cancellous bone of the vertebrae are calibrated by comparison with a calibration phantom with a known mineral content (Fig. 25.6b). The calibration material may consist of calcium hydroxyapatite or k2HP04. When the patient is positioned on a calibration phantom, care has to be taken to minimize the air space between the patient and phantom so that the measurements are not distorted (e.g., the knees are supported and the hips are flexed to straighten the lumbar lordosis). The manufacturer's recommendations should be followed closely to avoid measurement errors. Recently, continuous data acquisition with spiral or multislice CT has been suggested using thin-section imaging. This technique holds several promises because the whole vertebra is scanned; measurements are more reproducible, local variations in bone mineral content can be detected, and more sophisticated evaluations,
such as finite-element analysis, become possible that allow for a better estimate of individual fracture risk. The radiation dose in QCT is reduced by using the lowest possible kVp and mAs settings. With volumetric imaging, dose is increased as compared to standard QCT. The CT attenuation increases linearly with the bone mineral content and therefore allows for quantitative assessment of bone mineral content. Water attenuation (OHLJ) has been arbitrarily assigned as the zero point for this scale. This induces measurement errors because the soft tissue component of vertebral cancellous bone is not identical to water but is determined by varying amounts of fat, hematopoietic marrow, and soft tissue components of the bone matrix. As a result, the presence of fat marrow in the measurement region leads to an underestimation of bone mineral content with QCT, while the presence of hematopoietic marrow leads to overestimation. Because of the calibration process, kilovoltage has very little effect on the measured bone mineral content, but tube voltage does affect the CT attenuation of fat and thus the degree to which marrow alterations influence measured bone mineral content. The fat-induced error in standard singleenergy QCT (SEQCT) can be reduced byemploying a dual-energy technique (DEQCT) in which the difference in the CT numbers at low and high tube voltages is calibrated, rather than the
CCT BHO EVALUATION O. G.
PLOT 43 years
Female
Aver~
cor-res
240
180
'20
60
o DATABI
a
b
Fig. 25.6 QCT. Lateral survey view of the lumbar spine (a), axial scan through the L3 vertebral body (b), and a graphic plot of QCT results (c). The scans through L1 -L3
c and ROI selection are performed semiautomatically. A calcium hydroxyapatite phantom (Siemens) was used for calibration in this case.
Quantitative CT Bone Densitometry CT numbers themselves (see Chapter 4, p. 130). Despite its theoretical advantages, however, DEQCT is rarely practiced nowadays due to its substantially higher cost and higher radiation exposure.
QCT is feasible only if the scan plane can be placed precisely through the center of the vertebral body. It is not feasible in patients with significant scoliosis or with osteoporotic impression fractures and areas of increased cancellous bone density. The scan plane and ROI for densitometry should be defined semiautomatically to ensure reproducible results (Fig. 25.6). Follow-up requires a standardized technique and specified parameters such as tube voltage, filtration, and ROI selection. If possible, the same type of scanner should be used for all follow-up examinations. When swttching to a new system, crosscalibration has to be performed first in order to assure correct follow-up measurements. QCT is appropriate only if the examination will have therapeutic implications.
Spinal Trauma With trauma to the spinal column, CT is the diagnostic method of choice for the assessment of severity of injury. CT is much more sensitive for delineation of fractures or dislocations after spinal trauma than conventional radiographs. CT is frequently indicated to avoid underestimation of injury (e.g., involvement of the posterior margin of the vertebral body), to correctly classify the injury as stable or unstable and chose the proper treatment, and to avoid misinterpretation of lesions that are difficult to assess on radiographic projections (e.g., dislocation of the articular masses or dislocation of fragments). In patients with suspected instability or narrowing of the spinal canal. CT examination is warranted for surgical planning. If there is suspicion of spinal cord damage, however, MRI should be performed. MR is also indicated for the evaluation of the disco ligamentous complex and has been suggested as a superior tool for the evaluation of the cervical spine.
CT Morphology
Dependent on the trauma mechanism and the major direction of force, various types of vertebral fractures are differentiated (e.g., flexion, extension, compression, rotation, and translation, Table 25.8). The major questions to be answered by CT refer to the stability of the fracture and the assessment and quantification of compression effects on the myelon andlor nerves. Injuries to the spinal column most commonly involve the cervical spine and the thoracolumbar junction. The AO classification of spinal fractures is outlined in Tables 25.9 and 25.10. Flexion and extension injuries predominate in the cervical spine, while compression injuries are commonest at the thoracolumbar junction. Fractures of the upper cervical spine comprise a separate category (see Figs. 25.8-25.11). Fractures are often associated with prevertebral soft tissue swelling. On a sagittal MPR the normal soft tissues in adults (children)
947
25 Musculoskeletal System Table 25.8
Spine injuries according to mechanism
II Hyperflexion
injuries
Hyperflexion sprain (anterior subluxation) Hyperflexion dislocation Dens fracture Wedge fracture Chance fracture Bilateral facet joint dislocation Flexion teardrop fracture Clay shoveler's fracture pHyperextension
injuries
injuries
Burst fracture Jefferson fracture Rotation
injuries
Rotatory subluxation Unilateral facet joint dislocation (+ hyperflexion) Pillar fracture (+ hyperextension) Pedico-Iaminar separation ATranslation
Upper Cervical Spine Occipital Condyle Fractures
These are uncommon fractures that require significant forces. CT is the most accurate means of identification, and although previously underdiagnosed, are now seen more frequently (Fig. 25.7). Anderson and Montesano describe three types. Type I are compression fractures from axial loading, type II are linear occipital fractures that extend into the condyles, and type III are unstable avulsion fractures due to lateral bending and rotation. These fractures can be associated with lower cranial nerve deficits.
Hyperextension dislocation Isolated C1 arch fracture Isolated laminar fracture Hangman's fracture Extension teardrop fracture (compression
subsidence, bone graft displacement, cancellous bone resorption, or inflammatory changes. As fracture healing progresses, the contours of the fractured vertebrae may become smooth due to remodeling.
injuries
Uncinate process fracture
measure maximally 7 mm (7 mm) at the base of C3 and 22 mm (14 mm) at the base ofC6. In a postoperative setting CT is used to evaluate the spinal axes, including the alignment of the vertebral bodies, and to check for residual narrowing of the spinal canal. It is used after internal fixation procedures and after bone grafting to detect the faulty placement of implants (especially transpedicular screws), the fracture or displacement of implants, vertebral body
Suboccipital Injuries (CO-C1)
Atlanto-occipital dissociation results from distraction and shearing of the ligament complex between skull and cervical spine, which is often fatal. Survival has been described, and CT with coronal and sagittal MPR can show an increased distance between basion and dens (» 12mm pathologic) with associated soft -tissue swelling. Atlantoaxial rotational fixation is an irreducible subluxation that can be post -traumatic. This is most commonly seen in children with torticollis. CT with reformations can unequivocally diagnose this entity and allow measurement of the rotation angle directly (Fig. 25.8). There are four types with increasing severity; types II and higher are associated with a rupture of the transverse ligament and an increased atlantodentalinterval.
Fig. 25.7 Occipital condyle fracture (4 x 1 /6). Axial sections display the fracture (a) but spatial orientation is superior on coronal reformations (b). a
b
Upper Cervical Spine Fig. 25.8
Direct measurement of
atlantoaxial rotation on axial CT sections (4 x 0.5/1.5) allows for assessment of rotation subluxation (a) in this child with torticollis and atlantoaxial-axial rotational fixation best demonstrated on 3D reconstructions (b).
b
a
C1 Fractures
Fractures of the atlas (C 1) comprise 3-13% of cervical spine fractures and are divided in to five subtypes, depending on the course of the fracture lines (Table 25.9). The Jefferson fracture (type III) is an unstable fracture that is caused by an axial force (blow to the vertex of the head)
Table 25.9
that is transmitted through the head and the occipital condyles on to the atlas. In more than 40% it is associated with a fracture of C2. The characteristic bilateral displacement of the lateral masses ofCl beyond the margins ofC2 is best appreciated on coronal reformats, and a summated overhang > 7 mm is a definitive sign of transverse ligament rupture (Fig. 25.9). Sagit-
Upper cervical spine fractures
I C1 fractures
Type I
fracture of the anterior arch
frequently associated with a dens fracture
Type II
fracture of the posterior arch
most common fracture
Type III
Jefferson fracture
unstable fracture symmetrical bilateral fracture of the anterior and posterior arches, always associated with a disruption ofthe transverse ligament no displacement offragments anterior displacement and angulation of the anterior arch, additional anterior and cranial displacement of the posterior arch
type I type II type III Type IV
fracture of the lateral mass
Type V
fracture of the transverse process
C2 Dens fractures-Anderson-D'Alonzo
Type I
Fracture of the tip of the dens (usually stable)
Type II
Fracture through the base of the dens (unstable)
Type III
Fracture through the base of the dens and body of C2 (stable)
C2 Hangman's
fractures-Effendi
Type I
Solitary fracture of the axis arch with dislocation <3 mm (stable)
Type II
Arch fracture with disc rupture and ventral dislocation of the body of C2 >4 mm and/or angulation >110 (unstable)
Type 111
Arch fracture with disc rupture and luxation ofthe C2-C3 facet joints (unstable)
25
Musculoskeletal
System Fig. 25.9 C1 fractures, (a) Isolated fracture of the atlas arch on the right side. (b) Jefferson fracture type I with concomitant fracture of the articular facet.
a
b
tal reformats provide important information as to whether only the anterior arch is anteriorly displaced and angulated (Type II) or whether the anterior and posterior arch are displaced leading to a potentially devastating narrowing of the spinal canal. C2 Fractures
Dens fractures occur in 10-15% of cervical spine
fractures and are caused either by a hyperflexion trauma (anterior displacement of the dens and anterior subluxation of C1 and C2) or, less frequently, a hyperextension trauma (posterior displacement of the dens and posterior subluxation of C1 and C2). Fracture types can be classified according to Anderson and D 'Alonzo (Table 25.9) but because of the rarity of type I fractures other authors simply divide these fractures in high versus low dens fractures. While type I and III fractures are stable, type II is unstable (Fig. 25.10). It is a transverse fracture through the base of the odontoid and is the commonest C2 fracture. Due to dislocation ofC1
with respect to C2 and compression' of the medulla, these fractures may be lethal in 25-40%.
The fusion of the os terminale of the dens with the body of the axis can be delayed or may completely fail and will result in the presence of an odontoid bone. This should not be misinterpreted as type I or II fracture. A rounded and smooth contour of the os odontoideum and a compensatory hypertrophy of the anterior arch of C1 is suggestive of a developmental disorder rather than a fracture. A large os odontoideum can also be the consequence of a pseudarthrosis after previous C2 trauma.
Hangman's fracture or traumatic spondylolisthe-
sis (5-10%) is caused by hyperextension and distraction and describes bilateral fractures through the pedicles (or the arches) of the axis (C2) with anterior displacement of the body and angulation. These fractures can be divided according to Effendi in three subtypes, based on
Fig. 25.10
Anterior displacement
of the dens axis in an Anderson type II fracture (4x1/6). Sagittal view (a), volume-rendered display (b). Note a fracture component that involves the vertebral body as well.
a
b
Lower Cervical Spine (C3-Cl)
Fig. 25.11 Hangman's fracture type II with slight rotation and ventrocaudal dislocation of the body of C2 (unstable) that is better appreciated on the sagittal reformats (b, c) than on axial sections (a) Note the involvement of the intervertebral foramen on the right side (c).
a
subluxation patterns (Table 25.9). The extent of C2-C3 subluxation is best appreciated on sagittal reformats (Fig. 25.11).
b
CT adequately demonstrates the frequent displacement of fragments into the spinal canal. Articular Mass Injuries
Lower Cervical Spine (C3-C7)
Traumatic lesions of a cervical articular mass are infrequent, yet potentially unstable and require C3-C7 Fractures internal fixation. CT is the diagnostic method of Teardrop fractures that are caused by flexion choice but MRI should be preferred if accomtrauma are located in the lower cervical spine panying disc or ligament injuries are suspected. Dislocation of the cervical spinal facet (C5-C6) in more than 70% of cases. Flexion teardrop fractures are extremely unstable and may joints (locked facets) is the result of an extreme flexion of the head and neck. It frequently oclead to a dorsal subluxation of the vertebral body due to the rupture of the anterior longi- curs bilaterally and is an unstable condition due to disruption of the posterior ligament complex. tudinal ligament. There is often a significant swelling of the prevertebral tissues due to In general, the ligamentum flavum, the interhemorrhage. Teardrop fractures must not be un- spinous, and the supraspinous ligaments are inderestimated, despite the minor osseous lesion volved. There is anterior displacement of the with avulsion of a small triangular fragment of upper vertebra leading to an anterior dislocathe anteroinferior surface of the body, as there tion of the inferior articular facet of the upper are often midsagittal fracture lines through the vertebra to the front of the superior articular corpus. If the height of this fragment is smaller facet of the lower vertebra. Unilaterally locked than its sagittal length, this is suggestive for the facets are combined with a malrotation of the presence of an unstable flexion teardrop frac- upper vertebra into a slight oblique location (flexion and rotation trauma). ture as opposed to the stable extension teardrop The locked facet joints, and the anterior and fracture. The clay shoveler's fracture is the result of a rotational displacement of the upper vertebra flexion trauma that leads to an avulsion fracture are both best appreciated on sagittal reformats of the spinous process of C6 or C7 without in- (Fig. 25.12). Sagittal reformats also show the obvolvement of ligaments. It is thus a stable frac- literation of the intervertebral foramen by the dislocated articular mass. Axial scans demonture and is usually not examined with CT. Cervical burst fractures are a result from ex- strate the inverted order of the facets with the plosion of the vertebral body after the inter- triangular articular mass of the lower vertebra in the posterior location instead of the normal vertebral disc herniates through the fractured end plate following axial loading injury. There is anterior location (reversed hamburger bun sign). Note that axial sections of normal facet joints alusually a sagittal fracture line and comminution of fragments. The stability is dependent on the ways show a maximum of two pieces of bone. involvement of the posterior vertebral margin. Any additional bone fragment indicates a frac-
952
25
Musculoskeletal System
a
b
c
d
Fig. 25.12 Unilateral locked facets (0.5/1) as a consequence of hyperflexion/rotational injury. Note the empty facets on axial sections (a, b) and compare the
sagittal MPR of the riding facet joints tralateral side (d).
ture (triple image sign). Four fragments indicate a unilateral biarticular fracture (quadruple image sign). Axial and sagittal planes are complementary for outlining fractures of the articular masses and dislocations of fragments.
stabilizing rib cage, and the change from kyphosis to lordosis in this region. It is a common site of fractures associated with calcaneal fractures after falls orjumps from height. Hyperflexion and compression mechanisms are the main causative factors in this segment, with vertebral body compression fractures the most frequent entity (Figs. 25.13-25.15).
Thoracolumbar Spine Thoracic Spine
Because of the stability provided by the thoracic cage, thoracic spine fractures are relatively uncommon. They can occur in severe trauma, and are frequently associated with neurologic complications because of the large size of the spinal cord relative to the small spinal canal. Because of the natural kyphosis, hyperflexion type injuries are commonest, often resulting in fracturedislocations. Because of the high energy needed, noncontiguous fractures elsewhere in the spine or in the sternum are not uncommon and should be looked for. Normal variants can cause confusion in this region. There is a physiologic wedging in the lower thoracic spine (wedge ratio 0.80-0.90) and thin pedicles can mimic an increased interpedicular distance. Thoracolumbar Junction and Lumbar Spine
The thoracolumbar junction (ThI2- L2) is a vulnerable site for fractures because of the increased mobility of this segment, the lack of a
-
(c) to the con-
Fracture Classification
Based on CT findings, Denis developed his three-column system (see Table 25.3) in the early 1980s. Thoracolumbar fractures were classified in four types: compression fractures, burst fractures, seat -belt injuries and fracturedislocations. Each of these types was further subdivided into three or four SUbtypes. There has been criticism about this model because his three column definition does not match anatomically or biomechanically and the importance of soft -tissue lesions was underestimated. This led to the AO classification in the mid 1990s (Table 25.10), which used a two-column model and formed three categories: -A-vertebral body compression - B-anterior and posterior element injury with distraction, and - C-anterior and posterior element injury with rotation. Each category is further subdivided into nine SUbtypes using the classic AO 3-3-3 scheme.
Tnoracofumbar Spine Table 25.10 AD classification fractures (Magerl et al.. 1994)
of thoracolumbar
spinal
Table 25.11 trauma
Signs
of
instability
following
spinal
Vertebralbodycompression
Anterior column (sagittal reformations)
A1: Impaction fracture A2: Split fracture A3: Burst fracture
• More than a ser/o decrease in vertebral body height • More than 10° angulation of the vertebral body in the cranial or caudal direction compared with adjacent segments • Tearing of the anterior longitudinal ligament (teardrop sign)
Anterior B1: B2: B3:
and posterior
element
with distraction
Flexion-distraction, predominantly ligamentous Flexion-distraction, predominantly osseous Hyperextension shear injury
Anterior
and posterior
element
Central column tions)
(sagittal and coronal reforma-
• Contour irregularities in the posterior margin of the vertebral body (caution: basivertebral vein) • Decreased height of the posterior marginofthe vertebral body • Displacement of the posterior margin of the vertebral body • Asymmetry or separation of the pedicles
with rotation
C1 : Type A with rotation C2: Type B wrh rotation C3: Rotational shear injuries
Posterior column (sagittal and coronal reformations, also reformations on the plane of the facet joints) Fracture
Stability
In his original model, Denis defined three types of instability, which for CT was simplified to the rule that if any two or all three pillars are damaged (as seen on CT), it is assumed that the spine is unstable, For spinal surgeons however, stability of the spine means addressing several kinds of stability: mechanical, neurological and long-term stability, The AO classification defines instability for each of the classification patterns, including bony and disco ligamentous components, For this reason, MRI has been advocated as the most appropriate technique for use with this classification, For CT, the more simplified stability definitions are still commonly used and the criteria for the anterior, central, and posterior columns are summarized in Table 25.1L Multiplanar reformations are essential for a complete evaluation of spinal trauma (Figs, 25.13-25.15), Involvement of the middle pillar is never an isolated finding and always causes an unstable fracture, Unstable fractures frequently require surgery, while stable fractures have not necessarily to be operated, In the lower thoracolumbar spine a vertebral fracture is considered unstable with • involvement of the dorsal border of the vertebral body (middle column) • a post-traumatic kyphosis of more than 30 0
• Separation and divergence of the spinous processes • Fractures involving the pedicles, laminae, or facet joints • Lateral displacement ofthe articular processes • Subluxation of the facet joints with <50% congruence of articular surfaces • Dislocation (and possible locking) ofthe facet JOints
• an association with fractures num, and • a fracture dislocation,
of ribs or ster-
The Denis model is only partially applicable to the cervical spine and upper thoracic spine, In the cervicothorecic spine, a vertebral fracture is considered unstable with: • horizontal dislocation between two adjacent vertebral segments of more than 3.5 mm, and • an angulation of more than II between two adjacent segments, 0
Fracture Types
Wedge compression fractures occur after hyperflexion and axial loading stress, The anterior body becomes compressed, but the posterior part of the body and posterior ligament remain
953
954
25 Musculoskeletal
System
a
c
b
Fig. 25.13 Unstable burst fracture of L1 (2/3/1) with narrowing of the spinal canal and lateral recess (b. arrow). The interpedicular distance is increased, and the left facet joint is in a subluxated position (a). Sagittal re-
a
c
b
d
Fig. 25.14 (a, b) Chance fracture (AO classification: B2) ofTh 11 (2/3/1). Note the horizontal fracture component extending through the body, arc, and spinous process, (c. d) Flexion-distraction injury of C6 in a patient with ankylosing spondylitis (2/3/1). Note the almost coronal orientation of the primary CT images due to extensive ankylosis and kyphosis (c).
d
formatted images show the decreased vertebral height. the narrowing of the spinal canal (c) and left recess (d), and involvement ofthe spinous and articular processes.
intact. They are considered stable fractures. With minor anterior compression, usually no further CT evaluation is needed. Burst fractures are the result of hyper flexion and axial loading and also involve the posterior body and posterior longitudinal ligament. The compressed disc usually herniates into the vertebral body and typically there is a fractured posterosuperior fragment. Retropulsed fragments in the spinal canal often cause neurological symptoms. Often, there are also associated posterior column fractures or facet dislocations. As the posterior and/or middle column is ruptured these represent unstable fractures. They are most commonly located in the thoracolumbarjunction (Fig. 25.13). A Chance fracture (seat belt fracture) is a distraction injury that causes a horizontal splitting of the vertebral body, pedicles, the lamina, and the spinous process (Fig. 25.14). In the classic type, there is no dislocation or damage to the ligaments. It is most frequently located in LI to L3. There are several subtypes of seat belt fractures that involve the ligaments to a variable extent and that are unstable fractures.
Thoracolumbar Spine
a
Fig. 25.15 Torsion-distraction injury (AO classification: C2) ofT11 (2/3/1). The 3D image shows the rotation and impaction ofT1 0 upon T11 (a). The lateral view of the spinal canal (b) and the coronal reformation on
Combination of a flexion trauma with one or multiple additional forces (rotation, distraction, anteroposterior or posteroanterior shear stress) will result in fracture-dislocations or in failure of all three columns (Fig. 25.15). Hence these fractures are always unstable and frequently associated with severe trauma of the myelon and neurological symptoms.
a
c
b
b
Fig. 25.16 Sagittal reformations of a traumatic compression fracture with sharp angulation of the fractured end plate (a). Note the intravertebral disk prolapse with vacuum phenomenon. Severe osteoporosis causing im-
the plane ofthe facet joints (c) show that while the facet joints are dislocated, the spinal canal has not been narrowed (patient had no neurologic symptoms).
9
Malignant versus Traumatic versus Osteoporotic Fracture
It may be difficult to distinguish benign vertebral injury, due to osteoporosis or trauma, from malignant (metastatic) causes, especially in elderly patients. Sagittal and coronal reformations are essential for differentiation (Fig. 25.16). Traumatic injury to the spine is usually associated with detectable fracture lines. Pure compression fractures may sometimes be hard to distinguish from osteoporotic or malignant
c pression of the vertebral end plates (fish mouth vertebrae) and a wedge-shaped compression (b). Metastatic bone destruction (arrows) with a wedge-shaped compression of L1 (c).
955
25 Musculoskeletal System
fractures. Typical traumatic compression will be associated with sharp angles, as opposed to a more concave shape in osteoporotic fractures, and an irregular shape, frequently associated with bone destruction, in malignant fractures. Vertebral fractures in osteoporotic patients are most commonly located in the thoracolumbarjunction. The vertebrae have a characteristic shape (wedge-shaped or fish-mouth vertebra) with involvement of the both, the upper and lower end plates. Frequently, multiple vertebrae are involved. There may be an intravertebral vacuum phenomenon. In the subacute or
chronic state, reactive sclerosis is seen and there is none or only mild « 1 ern) soft tissue widening. In patients with metastatic involvement, the density of the vertebra is heterogeneous and areas of frank bone destruction may be present. The laminae are frequently involved in a destructive process. Solitary fractures of vertebrae in atypical locations are suggestive of a malignant etiology if no adequate trauma history is given. A soft -tissue mass is almost invariably present. It is best appreciated by MRI.
Trauma to the Pelvis and Extremities Upper Extremity Shoulder Trauma
In cases of acute shoulder trauma, CT is used only for surgical treatment planning in complex fractures of the proximal humerus or scapula that cannot be adequately evaluated on conventional radiographs. CT can also be helpful in differentiating between a fracture and mere luxation of the sternoclavicular joint. Scapular fractures are relatively uncommon and account for 3-5% of shoulder fractures. They usually occur in men 30-40 years of age and are the result of major trauma. They are commonly associated with rib fractures, pulmonary contusions, pneumothorax, or clavicular fractures. Depending on their anatomical location they can be divided into fracture of the body, neck, or intra-articular glenoid fractures. Complications include arterial injuries and brachial plexus lesions. Fractures of the proximal humerus are common and are most frequent in older patients with osteoporosis. The fractures can be two-, three- or four-part fractures, with or without associated dislocations. As therapy is usually functional, CT is only indicated for the younger individual in whom reconstruction of comminuted fractures is planned. Complications include avascular necrosis, frozen shoulder, and late instability.
•
CT Morphology
In scapular fractures, body and spine fractures have the highest incidence of associated thoracic injuries. Fracture lines are often transverse or have stellate patterns through the body. Neck fractures usually run longitudinally. Stability is determined by associated lesions of the clavicle or acromioclavicular ligaments, and unstable fractures require surgery. The most important consideration in scapular fractures is glenoid involvement. It is important to check for fracture line displacement and intra -articular loose bodies. Multiplanar reformations perpen-
Fig. 25.17 Complex fracture of the scapula without involvement of the glenoid. Volume-rendered display (4 x 1 /5.5). Due to a thin cortex, the humeral neck appears translucent.
Upper Extremity
1-part-fracture
3-part -fracture
2-part-fracture
D fl
4-part -fracture
.
~.,~. ,
t;[l1)
.
no / minor dilocation no / minor axis deviation
major tuberculum dislocated
dislocated tuberculum
dilocation of anatomic / surgical neck
impacted Fig. 25.18
/'>.7both tubercula dislocated
r=
minor tuberculum dislocated
comminuted
Neer classification of humeral fractures (modified from Greenspan 2000).
dicular to the articular surface of the glenoid are recommended. 3D volume-rendered reconstructions can improve anatomic orientation in complex fractures (Fig. 25.17). With comminuted fractures of the humeral head, attention should be given to the position
and fragmentation of the articular surface, intra-articular fragments, and the greater and lesser tuberosities. Fractures follow the physeal lines that divide the proximal humerus in four parts (the articular part of the head, the lesser and greater tuberosities, and the shaft), and are classified according to Neer (Fig. 25.18); assignment to the fracture type is based only on the dislocated fragments (c-Ic m dislocation, >45 axis deviation). There may be significant subluxation of the articular head fragment, especially in four-part fractures. Multiplanar reformations in two planes parallel to the humeral shaft and 3D volume-renderings should be routinely obtained for preoperative planning (Fig.25.19). 0
Shoulder Instability
a
b
Fig. 25.19 Substantially dislocated three-part fracture of the humeral head (4 x 1/5,5),
Shoulder dislocations are the commonest dislocations of any of the great joints and 96% are anterior. They are induced by a combination of indirect forces, such as abduction, extension, and external rotation in anterior dislocations. CT
957
25 Musculoskeletal System arthrography is used to evaluate chronic A Hill-Sacks lesion should be identified on shoulder instability (anterior, posterior, or mulhigh sections that pass through the coracoid tidirectional) following recurrent dislocation. process. This avoids confusion with the norThe improved quality of multiplanar imaging mal flattening of the humeral head contour planes with multislice CT, in particular, offers that is visible at lower levels. the opportunity of obtaining similar imaging projections to MRI. The increased tissue differentiation of MRI, however, leaves it still as the Injuries of the labrum are classified as partial superior modality for most indications (rotator tear (stage I), complete tear without dislocation cuff tears, impingement, shoulder instability). (stage II), complete avulsion with dislocation Even with MRI, the highest diagnostic accuracy (stage III) and complete tear with dislocation is achieved with direct MR arthrography, which and laceration of the capsule (stage IV). Dependrequires the injection of intra-articular contrast ing on the severity of the injury, the lesion of the material. For evaluating the rotator cuff, ultraglenoid labrum appears as a focal area of consound can be excellent but the learning process trast material imbibition (partial tear, no alterais long, at about 50-100 arthroscopically con- tion of the external shape of the labrum), as a firmed examinations. complete contrast -filled tear, or as a complete Sternoclavicular joint dislocations are rare avulsion with a thinned or shortened remaining but difficult to analyze with radiography and CT area and a dislocated fragment. (Fig. 25.20). Inis a primary modality for diagnosis. Anterior, jury to the anteroinferior labrum is frequently rather than posterior, dislocations are much accompanied by a bony avulsion. This traumatic more usual. CT with multiplanar reformations detachment of the labrum is classified as a cartilaginous or osseous Bankart lesion. allows differentiation from medial clavicular fractures and can show secondary mediastinal Injuries of the anterior and posterior lalesions in posterior dislocations. The latter may brum are best appreciated on axial sections. Inrequire additional contrast -enhanced CT stud- juries of the superior and inferior labrum used ies. to be a domain of MRI, but semicoronal MPR based on thin-section multislice data sets also • CT Morphology offers promise for these entities but there is Chronic shoulder dislocation is commonly as- little clinical experience available to date. Lesions of the superior part of the labrum, sosociated with a Hill-Sachs lesion (depression fracture) on the posterolateral aspect of the nor- called SLAP lesions, are classified depending on mally smooth humeral head. the anteroposterior extent of the injury (type II), dislocation of the fragment (type III), or additional involvement of the biceps tendon (type IV). With later additions to the original four types, nine types now exist. (Table 25.12).
Fig. 25.20 CT arthrography in a patient with recurrent anterior shoulder dislocation, (a) CT arthrogram in external rotation demonstrates a Hill-Sachs lesion on the posterolateral margin of the humeral head (arrow), (b) View in internal rotation reveals a Bankart lesion of the anterior labrum. The posterior labrum bears a normal notch at itsjunction with the articular surface (arrowhead). Type II capsular attachment.
a
b
Upper Extremity Table 25.12
Classification
of
SLAP
lesions
in
the
shoulder Original
Type I Type II Type III Type IV
classification
Frayed and degenerated labrum Detachment of biceps tendon with or without fraying Bucket-handle tear of the superior labrum Bucket-handle tear of the superior labrum extending in the biceps tendon
Subsequent
additional
performed with the humerus rotated internally and externally. Injury to the posterior capsular attachment and dislocation of the long biceps tendon are more easily detected. A rupture of the capsule is marked by the extravasation of contrast material, and a rotator cuff tear allows contrast material to enter the subacromial bursa. These collections require differentiation from extra -articular iatrogenic collections of contrast material that result from the percutaneous puncture.
classifications
Type V
Anteroinferior Bankart lesion with superior extension into the biceps anchor Type VI Unstable radial or flap tear with separation of the biceps anchor Type VII Extension of SLAP lesion beneath the middle GHL Type VIII Superior labral tear extending posteriorly with extensive labral detachment Type IX Complete concentric avulsion of the labrum from the glenoid rim
Elbow
Approximately 6% of all extremity fractures and dislocations involve the elbow. In adults, 90% of distal humeral fractures involve the elbow joint, while this is rare in children. CT can be beneficial in the preoperative evaluation of bicondylar intra-articular fractures (AO type C), T- or yfractures of the condylar complex or complex Monteggia or Essex-Lopresti fracture-dislocations. However, CT was rarely performed in cliniThe small sublabral recess located between cal routine due to the lack of mobility of the the bone and the anterosuperior labrum traumatized joint that is needed for adequate should not be misinterpreted as ~ SLAP lesioA~;tSi positioning. Thin-section spiral and multislice CT resolve this problem as isotropic datasets An anterior avulsion of the joint capsule can be with very thin slices allow high quality reformadifficult to diagnose because the mode of cap- tions in any given plane (Fig. 25.21). CT can also sular attachment varies anatomically (see be used for double-contrast CT arthrography for Fig. 25.2). Therefore the examination should be the evaluation of intra -articular loose bodies or
a
b
Fig. 25.21 Luxation fracture (2 x 0.5/1.5) of the elbow with radial head fracture and small intra-articular fragments (arrows). Note the direct contact oftrochlea and
c radius, as well as the step in the trochlear articular surface indicating a cartilage laceration (arrowheads).
959
960
25 Musculoskeletal System osteochondrosis of the capitellum (Panners disease). Wrist and Carpus
In the majority of extra-articular wrist fractures, such as the Colles or Smith fractures, there is no added benefit in performing CT. It should be reserved for intra -articular fractures, especially those that are associated with communition. Examples are the Barton, reversed-Barton or Hutchinson fractures, pilon-type distal radial fractures and complex fracture-dislocations such as the Essex-Lopresti fracture-dislocation. Also, injuries to the distal radioulnar joint can be detected with higher sensitivity. While positioning of the fractured wrist may be difficult, the advent of spiral and multislice CT enables good quality imaging in all three planes reconstructed from data sets using a 1 mm or smaller collimation. Carpal injuries can also reliably imaged with CT and the commonest indications include fractures of the scaphoid, or other individual carpal bones, and complex carpal luxations. Some 70% of all carpal fractures involve the scaphoid. These fractures are considered unstable when the fracture is complete or dehiscence > \ mm between fragments is present. Surgical therapy is indicated is these cases. Other indications are assessment of fracture healing (scaphoid), postsurgical conditions, and rotational deformities, while the previous role in the detection of occult fractures has been almost completely taken over by MRI. •
CT Morphology
In distal radial fractures, the intra-articular Barton-type shear fractures are best evaluated on sagittal reformations. In the Barton fracture the fracture displaces a dorsal fragment (hyperextension injury) while in the reversed- Barton it is a volar fragment (hyperflexion). Hutchinson or Chauffeur's fractures run obliquely through the base of the radial styloid. Associated fracture lines can run anywhere in the epiphyseal or metaphyseal regions of the radius. Pilon fractures are comminuted T-type fractures extending in the radiocarpal joint with complex fracture lines. In all these, the evaluation of the radiocarpal j oint surface and step-off thereof are the most important (Fig. 25.22). The Essex-Lo-
a
b
Fig. 25.22 Persistent pain in a patient after osteosynthetic repair of a reversed Barton-type fracture (2xO. 5/1.2). Note the small intra-articular fragments (arrow) on the coronal MPR (a), and a small step in the articular surface on the sagittal MPR (b).
presti fracture dislocation represents a comminuted fracture of the radial head and neck with associated rupture of the interosseous membrane and dislocation in the distal radioulnar joint. In scaphoid fractures, the best plane of evaluation is the oblique long axis plane, and positioning in ulnar deviation is helpful. Fracture lines are transverse or oblique and usually involve the waist. There is an increased risk of avascular necrosis of the proximal pole, identified as collapse or increased bone density. CT is also indicated if other carpal fractures are clinically suspected despite a normal radiograph. Non-dislocated fractures (Fig. 25.23), lunate fractures, hamate fractures (hook), and fractures of the trapezium-trapezoideum complex benefit from CT evaluation. CT is mandatory in the preoperative planning of severe fracture-luxations of multiple carpal bones after axial loading injuries. In the hand, CT is beneficial in showing the exact arrangement of fracture fragments in intra-articular luxation-fractures of the base of first metacarpal, such as Bennett or Rolando fractures.
Upper Extremity
a
b
Fig. 25.23 Oblique fracture of the capitate bone that could not be detected on conventional radiographs
Fig. 25.24 Rib fractures in a poly traumatized patient are easily located on volume-rendered images
(2XO.5/1.2).
(4X2.5/15).
Ribs and Sternum
ties. Resection of a rib is often performed while leaving part of the periosteum in place. This may result in typical irregular ossification and of the periosteal remnants. Fresh rib fractures present as sharp lines and are frequently missed on CT if they run almost parallel to the scan plane. Only with thin section multislice scanning can the rib cage be evaluated properly. VRT give an excellent overview offractures of the ribs (Fig. 25.24) but thick MPR (3-5 ern) of the anterior and posterior portions of the ribs in a coronal direction and thick sagittal MPR (Scm) of the lateral ribs also demonstrate pathologic abnormalities.
CT is not the primary technique of choice to evaluate trauma to the ribs or sternum. Conventional radiographs are more suitable for the ribs, and ultrasound may provide an initial work-up for suspected sternum fractures. However, ribs and sternum are always depicted on a chest CT, so should be reported on. •
CT Morphology
Healed rib fractures are a frequent incidental finding at chest CT. They present as a focal fusiform widening and sclerosis of the affected rib, and rarely may be associated with deformi-
a Fig. 25.25
b (a) Sternal fractures are best displayed on
MPR (4 x 2.5/15). Note the retrosternal hematoma (arrowheads) and the con concomitant distraction injury of the thoracic spine (arrows), (b) Incidental finding of
c an anatomic variant, a sternal foramen that may assume
significance if sternal puncture is attempted (4 x 1/6, VRT). (c) Complete ankylosis of the costovertebraljunctions in ankylosing spondylitis (4 x 1/6, VRT).
961
25
Musculoskeletal System
Sternal fractures can be excellently displayed with thin-section spiral or multislice CT. The best results are gained from sagittal (Fig.25.25a) or thick coronal curved planar reformations (3-10 mm) parallel to the body and the manubrium of the sternum Other sternal abnormalities can best be visualized using volume rendered displays (Fig.25.25b, c).
but vertical stability, and type C injuries with a combined vertical and rotational instability (Table 25.13). Involvement of the acetabulum is always classified as a type C3 injury. Associated injuries from pelvic fractures include pelvic hemorrhage from pelvic veins or branches of the internal iliac artery branches, visceral injuries, and lower urinary tract injuries such as rupture of the membranous urethra. CT
Pelvis Pelvic Ring In the evaluation of pelvic trauma, fractures involving the pelvic ring are distinguished from fractures of the acetabulum. CT is frequently mandatory for diagnosis of complex injuries and planning appropriate surgical interventions. Instability of the pelvis is commonly defined by two kinds of displacement, rotational and vertical. Rotational stability is provided by the symphysis and anterior sacroiliac ligaments, while vertical stability depends on the iliolumbar and posterior sacroiliac ligaments. Severe compression trauma or vertical shearing injuries generally disrupt the bony pelvic ring at a minimum of two sites, the anterior ring (ischial and pubic bone) and the posterior ring (sacrum and sacroiliac joints). The relatively common isolated fracture of the anterior pelvic ring by direct trauma is not a primary indication for CT. Three basic disruptions patterns are recognized: sagittal compression, lateral compression, and vertical shear. The classification most commonly used is that of Tile, in which pelvic injuries are differentiated into stable type A injuries, type B injuries with rotational instability
Table
25.13
Fracture No
Pelvic fracture type
pelvic
Anterior
ring disruption
compression
classification
Morphology
Anterior compression leads to "open book" or external rotation deformity. Small forces show vertical obturator ring fractures and rupture of anterior sacroiliac, sacrotuberal. and sacrospinous ligaments. With higher forces separation of the symphysis occurs as well diastasis in the sacroiliac (SI) joint because of rupture of posterior sacroiliac ligaments. Lateral compression leads to internal rotation deformity leading to coronal obturator ring fractures with overriding fragments and crush fractures of the sacrum or the iliac ala. Vertical shear forces lead to a combination of (vertical) fractures of the anterior ring with posterior injuries like longitudinal sacral fractures or SI -diastasis from ligamentous ruptures (Fig. 25.26). There is characteristic superior displacement of the involved hemipelvis. Involvement of the neuroforamina and spinal canal is a particular concern in patients with sacral injuries (Fig. 25.27). Oblique coronal reformations in the plane of the sacral tilt are helpful in describing the course of a transforaminal fracture line (Fig.25.27a). Sagittal reformations are in many cases essential for diagnosing a transverse fracture. Three-dimensional representations give a good overview of the extent of an injury.
(Tile, 1996)
Classification
Features
Stability
Type A
Stable
A1
apophyseal
A2
stable ring fractures
avulsions
Type 81
External
Lateral compression
Type 82
Internal
rotation
Vertical shear
TypeC
Vertical
shear injuries
Complex
Complex
Variable
Re rotationally. V = verticalIy, Pe posteriorly
rotation
or open book injuries
R unstable; V, P stable
injuries
Runstable;V, R V, Punstable Variable
Pstable
while the anterior sacroiliac joint capsule is disrupted (type II, see Fig. 25.30). A complete unilateral (type III) or bilateral sacroiliac dislocation (type IV), with or without bony involvement, is usually associated with extensive softtissue injuries and plexus lesions. Fractures of the iliac bone or sacrum may show intra-articular extension. A severe anteroposterior compression of the pelvis (open book mechanism) typically leads to a complete bony and/or ligamentous disruption of the dorsal pelvic ring resulting in vertical instability (type C injury). Fig. 25.26 Vertical shear injury (type C) with a rightsided transforaminal fracture, rupture of the symphysis, and a left-sided anterior pelvic ring fracture (2/4/2, SSD). Low-dose scan in a pregnant woman (third trimenon). Note the spine and femoral bones of the fetus (arrows).
The extent of sacroiliac injuries is frequently underestimated in projection radiography and requires CT for diagnosis.
Acetabulum
Fractures of the transverse process of the lower lumbar spine are frequently associated with severe unstable pelvic injuries (type C). Injuries of the sacroiliac joint are classified into four types. Intra-articular gas inclusions indicate distortion of the sacroiliac ligaments without widening of the joint space (type I). This finding, however, can also be seen in degenerative j oint disease ofthe sacroiliac junction and should be considered only when extensive and unilateral. Separation of the sacroiliac joint appears as a unilateral widening of the joint space relative to the opposite side. Frequently the posterior ligaments are intact
Acetabular fractures are caused by axial or lateral forces that are transmitted to the acetabulum through the femoral head. CT is important for the classification of acetabular fractures and the planning of treatment. Accurate classification is important to select surgical candidates and to guide the most appropriate treatment. The widely accepted classification is that by Letournel and ]udet, which divides ten fracture patterns into five elementary and five associated patterns (Tables 25.14, 25.15). The five commonest patterns include more than 90% of fractures and these are the two-column, transverse, T-shaped, transverse with posterior wall, and posterior wall types (Fig. 25.28).
Fig. 25.27 I ntraforaminal fracture of the sacrum on the right side (2/4/2). Semicoronal reformation along the longitudinal axis of the sacrum (a). Three-dimensional reconstruction (SSD) in which the overlying anterior pelvic ring has been partially removed to demonstrate the sacrum (b).
a
b
25 Musculoskeletal System Table 25.14 Classification of acetabular (after Letournel and Judet): basic forms
fractures
! Posterior wall fracture
• Transverse
• Frequent impaction ofthe acetabular rim, weh incongruence to reduced femoral head tissue
with incarcerated
and demonstrable
capsule
intra-articular
frag-
ments
~;~Qsterior column fracture • Starts above the sciatic notch, descends through the acetabular roof, and termin.ates in the ischiopubic ramus • Fracture line passes
through
the quadrilateral
plate
; Anterior wall fracture ;
• Uncommon
I •
/\
Extends forward acetabular fossa
fe AnteIiorcolumn s
and downward through the
fracture
extends from the superior pubic ramus into the iliac segment of the anterior pillar
• Fracture
• The higher the fracture emerges on the anterior iliac wing, the greater the area of involvement of
the anterior acetabularjoint I Transverse
I? •
f
fracture
with
through the acetabular ring
of the posterior
rim
• Often associated
fractures
T fracture
• Most common fracture • May involve any portion
acetabular
Table 25.15 Classification of acetabular (after Letournel and .Judet): combined forms
surface
Posterior column/posterior
vertical extension fossa into the obturator wall fracture
• Fracture pattem similar to an isolated posterior pillar fracture • Acetabular rim is often displaced into the gluteal muscles Transverse/posteriorwall • • • •
fracture
Common fracture type Pattem similar to the transverse fracture Highly variable size of acetabular rim segment Acetabular rim fragment may be juxta- or infratecta I
Anterior column verse fracture
(or wall)/posterior
hemitrans-
• Anterior part of fracture typically runs below the anterior superior iliac spine through the acetabular fossa • Fracture terminates in the superior pubic ramus or passes through the obturator foramen and terminates in the ischiopubic ramus • Transverse component starts in the acetabular fossa and runs posteriorly through the posterior pillar
fracture
Transverse
fracture
line extending
pillars and dividing the acetabulum rior and inferior
through both into a supe-
portion
CT Morphology
The classification of acetabular fractures can be a difficult task on axial scans. A good method is to reconstruct 3D images in which the femoral head is subtracted to give a direct view of the acetabulum, but this requires proper editing software. A method that is simpler and often satisfactory is to image the quadrilateral plate, i.e., the medial aspect of the iliac bone facing the lesser pelvis (Fig.25.29a). The fracture lines visible in this area will, in general, reflect the main fracture pattem in the acetabulum although lesions of the anterior and posterior acetabular wall will not be demonstrated by this technique. The 3D imaging findings are then correlated with the axial scans to ensure that small fracture lines are not missed. The axial images at
Two-column
fracture
• Complex fracture in which the acetabulum is completely dissociated from the sacroiliacjoint, and the anterior and posterior pillars are separated from each other • The posterior part of the fracture runs downward from the greater sciatic notch, and multiple 'secondary' fractures may occur in the quadrilateral plate • The anterior part of the fracture extends completely through the iliac wing, which may contain 'secondary' fracture lines • The acetabulum, quadrilateral plate, and obturator ring may show localized ares of comminution • The fracture is usually associated with central dislocation of the femoral head
the roof (tectum) of the acetabulum can also be useful since wall fractures pass obliquely anterior or posterior to the femoral head, transverse fractures pass sagittally through the tectum, and colunm fractures pass coronally through the tectum.
Anterior column fracture
Posterior wall fracture
T fracture
Two-column fracture
Fig. 25.28 Classification of acetabular fractures, viewed from the lateral aspect (acetabulum) and the
a
b
Fig. 25.29 Anterior collumniposterior hemitransverse fracture (4 x 1i 6, SSD) that is easily classified using a volume-rendered view of the quadrilateral plate (a). Compare with an exarticulated view of the acetabulum (b).
It is also helpful for treatment planning to produce multiplanar reformats and 3D displays (Fig. 25.30) that simulate the anteroposterior projection, and projections in the obturator (anterior-oblique, fractured side elevated by 45 0) and ala positions (anterior-oblique, healthy side elevated by 45°). These projections demonstrate the load-bearing portion of the acetabular roof (Matta arc) in multiple views (Fig. 25.31). Conservative treatment may be considered if no
medial aspect (quadrilateral plate). All the fracture types listed in the tables are illustrated.
fracture lines cross the acetabular roof (tectum) or there is no incongruence within this part of the joint. Multiplanar and 3D reformations can also be used to generate images that correspond to the inlet and outlet views of radiography. In addition to fracture classification, it is important to note the position of the femoral head in the acetabulum and look for intra-articular fragments, lesions of the femoral head (depressed fractures), and associated soft-tissue injuries. An area of direct contact between the bony femoral head and an acetabular fragment indicates severe cartilage damage and is a poor prognostic sign (Fig. 25.32). Sagittal and coronal reformations are indispensable for evaluating complex fractures. Avulsion fractures of the pelvis are not a primary indication for CT, but may be an incidental finding in trauma patients. The apophyses especially are common locations for acute and chronic avulsion injuries. These are more frequent in adolescents and the most usual sites include the ischial tuberosity (hamstring muscles), anterior superior iliac spine (sartorius), anterior inferior iliac spine (rectus femoris), and the lower edge of the symphysis (adductor muscles).
966
25 Musculoskeletal
a
System
b
c Fig. 25.30 Complex pelvicfracture (4 x 1 /6, VRT) with an open book mechanism, anterior diastasis ofthe right sacroiliacjoint. rupture of the symphysis, bilateral fractures of the anterior pelvic ring, and a two-column fracture of the acetabulum. AP view (a), PA view (b), inlet view (c), obturator view (d), and ala view (e).
d
a
e
Fig. 25.31 Multiplanar reformations parallel to the femoral neck (a) and perpendicular to the femoral neck through the region of the acetabulum (b) best demonstrate the involvement of the acetabular roof.
b
a Fig. 25.32 Transverse fracture combined with posterior acetabular rim fracture and subluxation of the femoral head. The 3D images identify the fracture type
c (a, b), and the coronal reformation (c) shows a point of direct contact between the femoral head and acetabulum, indicating severe cartilage damage.
Lower Extremity Femoral Head
Fractures of the femoral neck or transtrochanteric femur fractures rarely require CT. Fractures generally lie in the axial or sagittal planes, thus multiplanar reformations or 3D volume-rendering are needed for adequate evaluation of fracture fragments. MRI is far more sensitive for occult femoral neck fractures. A luxation of the femoral head only occurs after a heavy trauma (e.g., motor vehicle accident). The luxated femoral head has to be repositioned immediately to reduce the risk of femoral head necrosis. Frequently this has to be done as a surgical open reposition maneuver because parts of the labrum, the joint capsule, or an isolated intra -articular fragment preclude closed repositioning. A Pipkin fracture is defined as a ventrocaudal femoral head fracture due to posterior luxation of the femoral head caused by axial loading on the femur with a flexed hip (dashboard injury). The femoral head spontaneously repositions in the acetabulum in many cases. Four types have been described. In type I the femoral head fracture lies below the fovea capitis (Fig. 25.33); in type II the fracture also involves the fovea; type III is a combination of I or II with a medial collum femoris fracture, and in the frequently occurring type IV there is an associated fracture of the dorsocranial acetabular rim. In patients who have undergone complex. joint reconstructions, CT can be useful for evaluating the congruence of the articular surfaces and the placement of the implants. The radiologist should examine the load-bearing portions of the acetabulum for any steps or discontinuities « 2 m m for an optimum result) and make
certain that there are no screws projecting into the interior of the joint. Lower Extremity Knee and Lower Leg
The rationale for using CT in the evaluation of knee trauma is the fact that conventional radiographic projections tend to underestimate the severity of bony injuries. A depression of the tibial plateau must measure at least 5 mm to be visible on projection radiographs. In patients with symptoms but negative radiographic images, MR is recommended for diagnosis of a bone bruise or occult fracture lines. Axial deformity and internal disruption (cruciate ligaments and menisci) are of greater importance in knee injuries than incongruence of the articular surfaces. The diagnostic method of choice for evaluation of the internal knee structures is MRI. However, recently a Belgian group has shown promising results with highresolution, single contrast, multislice CT arthrography for internal derangement. Spiral or multislice CT is superior for evaluation of complex bone injuries of the (supra) condylar femur or tibial plateaus with displaced or multiple fragments. CT offers the option of constructing reformatted images in arbitrary planes and generating direct 3D views of the tibial plateau. An isolated injury of the femur is generally not an indication for CT. Fractures of the proximal tibia may be classified as sprain-type fractures or plateau fractures, according to the mechanism of the injury. Sprain-type fractures are subdivided into fracture-dislocations and internal knee disruptions, which are associated with a bony avulsion or marginal depression resulting from the ligamentous injury (Fig. 25.34). Fractures of the tibial plateaux result when the knee is struck by a moving vehicle and therefore they are also called fender or bumper fractures. In spiral fractures of the tibial shaft, rotation measurements with CT may aid in the presurgical planning of complex cases. •
a Fig. 25.33
b Pipkin type I fracture ofthe femoral head.
CT Morphology
The distal femur is involved much less frequently than the tibial plateau. Monocondylar fractures more commonly involve the lateral condyle of both the femur and tibia, and the ma-
967
25 Musculoskeletal System Fig. 25.34 Classification of plateau fractures (a) and sprain fractures (b) of the upper tibia.
Plateau fractures
Split fracture (P1)
Depressed fracture (P2)
Bicondylar frakture (P4)
Depressed split fracture (P3)
Sprain fractures
\ Split fracture (01)
b
Marginal depression (04)
Marginal avulsion fracture (03)
jority are compression-type injuries (e.g., tibial plateau fractures). A fracture that involves only the medial tibial plateau is usually the result of extensive torsional trauma with involvement of the capsule and ligaments (spraining-type fracture). Even when the bony fragments are quite small in a tibial plateau fracture, associated injuries to ligaments and menisci have to be considered depending on the trauma mechanism. In the case of plateau fractures, it is important to evaluate the degree of plateau depression and the displacement of the fragments.
a
Condylar fracture (02)
b
Fig. 25.35 Depressed split fracture (type P3) of the tibial plateau (1/2/1, MPR). The depression has led to comminution of the lateral articular surface. The longitudinal split (arrowheads) is not displaced (a, b).
Four fragment comminuted fracture (05)
Usually, surgery is indicated for patients with articular depression of 2-3 mm, but orthopedic surgeons may also operate with depression of 3 mm. Coronal and sagittal reformations are essential for this purpose (Fig. 25.35), while axial sections are best for defining the fracture lines in the plateau itself. When bony ligament avulsions are found (e.g., intercondylar eminence), the torn ligamentous structures should be identified by correlating CT findings with the clinical presentation and the site of the avulsion. In the case of fracture-dislocations, it is important to
c
d
Depressed tibia plateau fracture (type P2; c, d). Note the improved spatial resolution with submillimeter scanning (2 X 0.5/1.2, MPR).
Lower Extremity
a
c
b
Fig. 25.36 CT arthrography in a patient after anterior cruciate ligament plasty (2 x 0.62/1.25). Note the excel-
lent display ofthe cruciate ligaments and the small tear in the lateral meniscus (arrow).
define the fracture pattern, assess the involvement of the intercondylar eminence, and determine the number and location of displaced fragments. Lipohemarthrosis is an indirect sign for an intra-articular fracture. It causes the so-called FBI (fat -blood interface) sign that represents a fluid- fluid level between hypoattenuating fat and iso- to hyperattenuating blood. This sign is best appreciated on axial sections or sagittal reformations in soft tissue windows. CT arthrography offers an excellent, albeit invasive tool for evaluation of the menisci, ligaments and tendons (Fig. 25.36) although MRI will remain the technique of choice for most patients.
and may occur as two-part or three-part fractures. A pilon fracture is a comminuted intra-articular fracture of the tibia and constitute about 5% of all tibia fractures. They can be classified in three types, ranging from nondisplaced to heavily comminuted fractures, depending on the magnitude of force.
Ankle
CT in the ankle and foot focuses mainly on complex or comminuted fractures of the ankle, calcaneus, talus or midfoot. Ankle fractures are in general no indication for CT evaluation. Only fracture-dislocations or comminuted fractures may benefit from CT, usually before operative therapy. Effective imaging is possible for Tillaux, triplanar, or pilon fractures ofthe distal tibia. The biplanar fracture of Till aux is a variant of a Salter- Harris III injury and occurs after eversion trauma. It occurs in older children in which the medial part of the physis is more fused than the lateral. Triplanar fractures are a variant of Salter-Harris IV injury,
•
CT Morphology
In the Tillaux fracture, there is a vertical epiphyseal fracture in the lateral aspect of the ankle. The lateral fragment is displaced laterally by the intact tibiofibular ligaments. Usually the more proximal interosseous membrane is ruptured. Triplane fractures show a vertical (sagittal) fracture through the epiphysis associated with an axial fracture through the physis and an oblique fracture through the posterior metaphysis. This separates an inferolateral fragment from the intact tibia shaft and medial malleolus (Fig. 25.37). When the malleolus is also completely separated, a three-part fracture exists. Pilon fractures often result from a combination of axial compression and torsion injury. Type I shows a nondisplaced cleavage fracture extending into the tibiotalar joint, type II moderate comminution and incongruity of the articulation surface, and type III gross comminution with shattering of the articular surface. The latter may be a devastating injury leading to early functional impairment and severe secondary osteoarthritis.
969
970
25 Musculoskeletal
System
a
c
b
Fig.25.37 Triplane fracture with epiphyseal and epimetaphyseal involvement.
Calcaneus
Calcaneal fractures comprise 60% of all major hindfoot injuries. Most result from falls, and are a result of axial loading. Frequently, there are associated compression fractures of the thora-
columbar spine or other extremity fractures. The calcaneus has a complex anatomy consisting of six surfaces and with four articulating surfaces to the talus and cuboid. Numerous classifications have been devised for calcaneal fractures (see example in Fig. 25.38), but essen-
2 fragments 0
5
Nojoint involvement
I
Subluxationj 4 fragments
1 joint
1joint
5 fragments
•
-:».« ~
•
4
4
4
*
E
~ 1 joint
2joints
Fig. 25.38 Zwipp classification (1989) of calcaneal fractures, based on the number offragments and articular involvement. The main fragments are numbered as in Table 25.16. S= shear fracture, G=joint depression
2 joints
3 joints
fracture, E= tongue-type fracture, various surgical approaches are based on this classification: 0 medial approach, • bilateral approach, * extended lateral approach.
Lower Extremity
tially they are divided in extra-articular and intra -articular fractures. Extra -articular fractures (25%) are shear fractures caused by adduction of the forefoot from an equinus position (e.g., shoes with high heels). Articular involvement may still occur in this type of fracture (e.g., the calcaneocuboid joint), but the important subtalar joint is not affected. The fractures may involve the anterior portion of the calcaneus (often with involvement of the calcaneocuboid joint), the sustentaculum, or the superior part of the tuber calcanei. Intra-articular compression fractures are more common (75 %) and always involve the facets of the subtalar joint. A vertically directed force drives the lateral process of the talus into the Gissan angle of the calcaneus leading to a primary oblique coronal fracture. With greater forces secondary fracture lines to the posterior facet, calcaneocuboid j oint or tuber calcanei lead to comminution. Stress fractures occur in joggers and runners but also may be seen in older patients with osteoporosis. They frequently become clinically apparent only some 10-14 days after the incident. CT is an excellent tool for detection of such fractures, especially with negative or ambiguous plain films. The goal of surgery is the reconstruction of the subtalar joint and reconstruction of the
a
tuber -joint angle (Bohler' s angle). Since the sustentacular fragment frequently remains in its anatomical position, it is used as the initial fragment from which surgical reconstruction of the calcaneus begins. The calcaneocuboid joint is of small importance. Accurate surgical planning requires identifying the principal fragments and evaluating the involvement of the posterior subtalar joint, the anterior subtalar joint (anterior and middle facets), and the calcaneocuboid joint. •
CT Morphology
Extra -articular fractures are frequently oblique and divide the calcaneus into posterolateral and anteromedial fragments (Fig.25.39a). Nondisplaced fractures may appear only a/\ a linear density. CT is not always indicated, as these fractures can be treated conservatively. In intra -articular fractures, the "primary" oblique fracture line extending into the posterior subtalar facet splits the calcaneus into two main fragments, an anteromedial sustentacular fragment and a posterolateral tuberal fragment. A third fragment with the anterior process may be split from the anteromedial fragment. Secondary fracture lines may extend posteriorly to the subtalar joint (joint depression type, Fig.25.39b) or to the tuber calcanei (tongue
b
Fig. 25.39 Types of calcaneal fractures (2 x 0.5/1.2). Shearfracture (a), joint depression fracture (b), tonguetype fracture (c)
c
971
972
25 Musculoskeletal
System
a
b
Fig. 25.40 Bilateral joint depression type of calcaneal fracture (0.5/1) with lateral bulge and impingement of the peroneal tendons (arrows) on the right (a). Stress
fracture of the calcaneus with typical sclerosis and extra-articular location in an osteoporotic patient (b).
type, Fig.25.39c), and divide the posterolateral fragment apart into a cranial fragment with the posterior joint facet and a caudal fragment with the tuber calcanei. With both fracture types the .posterior articular facet may be further fractured and is always rotated and depressed into the tuberal fragment. Damage to the posterior articular facet is often severe and is best appreciated on coronal CT images (Fig.25.40a). With a tongue-type (beak) fracture, Bohler's angle is usually negative due to rotation of the jointbearing fragment, although there is relatively 'little articular damage. The j oint depression fracture is a more severe type that is associated with more extensive articular damage and is occasionally associated with an almost normal Boehler's angle (Fig.25.39b). An intra -articular calcaneal compression fracture consists of these four main fragments in more than 50% of cases, and comminution may be present. Approximately 20% of cases have three main fragments, and an equal percentage have five main fragments. The key fragment between the talus and calcaneus is the sustentacular fragment, which usually retains its relative position owing to the powerful medial talocalcanealligaments. A descriptive classification has been devised for calcaneal fractures based on the number of main fragments and the extent of articular involvement (Table 25.16). Consideration has to be given to displaced fragments that may cause incarceration or trapping of tendons and accompanying structures.
Table 25.16 Descriptionof calcaneal fractures in terms of main fragments andjoint involvement (listing corresponds to frequency of involvement) Main fragments: 1. Sustentacular fragment 2. Tuberosity fragment 3. Posterior facet fragment 4. Anterior facet fragment 5. Anterior process fragment Articular involvement: 1. Posterior subtalar joint (posterior facet)" 2. Calcaneocuboidjoint 3. Anterior subtalarjoint (middle and anterior facets)
Compression fractures of the calcaneus always produce a lateral bulge, which may cause entrapment of the peroneal tendons (Fig.25.40a). Abutment is a severe static condition caused by the impingement of the fractured calcaneus on the lateral malleolus as a result of compression and transverse widening of the bone. On the medial side, the tendons of the long digital flexors may become entrapped beneath the sustentaculum. Post -traumatic spur formation or subcutaneous fragments can lead to severe pain. Stress fractures present as a vertical or oblique fracture line that frequently is extra-articular but may have an intra-articular component. Sclerosis is a common and characteristic finding (Fig.25.40b).
Lower Extremity
Talus
Tatar fractures are the second commonest injury to the tarsus, but only occur in 0.3% of all extremity fractures. Talar neck fractures (65 %) result from forced dorsiflexion against the anterior tibia (foot -brake injuries). Talar body fractures (25 %) comprises transchondral fractures of the dome, shear fractures, crush injuries and avulsions of the posterior or lateral processes. The rest are talar head fractures (10%), resulting from forced plantar or dorsiflexion mechanisms and involving the talonavicular joint. •
CT Morphology
A
In talar neck fractures, a coronal fracture line separates the body of the talus from the anterior process. Hawkins divides these into three types depending on the amount of force: Type I nondisplaced neck fracture, Type II displaced neck fracture with subtalar subluxation, and Type III fracture with dislocation in the ankle and subtalar joints. With increasing distraction, there is an increasing gap between the fragments with progressive malalignment. Subtalar dislocations are more
a
Fig. 25.41 Fracture of the talar dome (2 x 0.5/1.2), demonstrated on sagittal MPR (a), and shaded surface displays after removal ofthe tibia (b) and total exarticulation of the talus (c).
often medial, but the lateral ones are associated with greater disability. There may associated fracture lines extending into the talonavicular joint. These fractures are associated with a high incidence of avascular necrosis as the critical anatomic sling of vessels lies directly beneath the neck. Secondary osteoarthritis is also common after these fractures. A special kind is the aviator's astragalus fracture-dislocation which commonly occurred in fighter pilots during World War I after nose-down crash landings. Fractures of the talar dome are infrequent but can be excellently displayed using CT with multiplanar reformations and exarticulation of the bone (Fig. 25.41). Lateral lesions occur when the ankle is forcibly inverted in dorsiflexion, and medial lesions while in plantar flexion. They are best demonstrated on coronal MPR. Posterior process avulsion fractures are usually only seen with CT and usually involve the lateral tubercle (Shepherd's fracture). Fragment excision is often needed to prevent chronic pain syndromes. Transchondral fractures of the dome are chip-fractures, commonly associated with ankle ligament ruptures (Fig. 25.42). Osteochondritis dissecans is thought to represent a chronic stress injury involving the talar dome. Both conditions are best seen with coronal MPR.
b
973
974
25 Musculoskeletal
System Fig. 25.42 Missed transchondral chip-fracture of the distal tibia (2 x 0.5/1.2) with demonstration of an osteochondral body (a) and a corresponding defect in the lateral aspect of the distal tibia (arrow). The size of the defect can best be appreciated on the axial sections (b). Note the osseous changes after plasty of the fibula-talar ligaments.
a
b
Midfoot
Foot iuxations are the most severe traumatic events in the midfoot and these can involve the Chop art joint (midtarsal) or, more commonly, the Lisfranc j oint (tarsometatarsal). These result from forced plantar flexion whereby the midfoot folds over the forefoot, and can be caused by parachute jumping or by motor accidents when the forefoot gets trapped under the clutch or foot-brake. Lisfranc injuries are divided into homolateral or the more severe divergent types, depending on the metatarsal IIII -cuneiform relation. In the diagnosis of these complex injuries, and especially of calcaneal and talar fractures, CT is essential for preoperative planning as it can define fracture lines and articular involvement that are not apparent on conventional radiographs. CT is also useful for evaluating residual deformities and malalignment following complex trauma, and in the timely planning of subsequent surgical procedures such as arthrodesis. It has substituted for conventional tomography for the evaluation of fracture healing (Fig. 25.43). Spiral and multislice CT with narrow collimation yields an (isotropic) axial data set that can be processed to generate high quality reformations in an axial plane parallel to the axis of the forefoot, sagittal images parallel to the metatarsals (a slight cone-shaped angulation of the planes can be helpful), and coronal images perpendicular to the long axis of the calcaneus. These imaging planes allow for optimum evaluation of all foot joints in at least two perpendicular imaging planes.
a
b
Fig. 25.43 Fracture of the first metatarsal bone with formation of a pseudarthrosis despite surgical repair (arrows).
•
CT Morphology
The Lisfranc (tarsometatarsal) fracture luxation is best evaluated with an oblique axial plane parallel to the long axis of the midfoot, and by coronal sections perpendicular to this. In the commoner homolateral type, the bases of the second to fifth metatarsal are luxated laterally, while the first metatarsal retains a normal relation with the first cuneiform. In the divergent type the first metatarsal is luxated medially as well. Classically, there is an avulsion fracture of the base of the second metatarsal or tip of first cuneiform, but there may be more communition with multiple small fractures at the bases of the metatarsals. These luxations can be complicated by associated cuboid fractures.
Intervertebral
Stress Fractures Stress fractures can be divided into fatigue fractures (fractures resulting from overload on normally structured bone) and insufficiency fractures (resulting from normal load on pathologic bone). Fatigue fractures are a common problem in sports medicine and almost 80% of these are related to running or fitness training. These should be differentiated from stress reactions that have no cortical break and only bone edema or bony remodeling. Insufficiency fractures can be seen in osteoporosis (steroids),
Table 25.17
Regions at risk for stress (fatigue)frac-
tures
High risk Femoral
neck
Anterior mid-tibia Navicular bone ci the Sesamoids Intermediate Femoral
foot
risk
shaft
Medial malleolus Proximal
metaphysis
of metatarsal
V
Low risk Sacrum
Distal fibula Calcaneus Distal metatarsal
Intervertebral
II-V
Disc Herniation
Disc
Lesions
after radiation, or in bone that is infiltrated with malignant tumor. Common sites for stress fractures are the sacrum, femur, tibia, navicular, and metatarsals (Table 25.17). While stress reactions and ligamentous injuries are best imaged with MRl, stress fractures can 'be diagnosed with either modality. CT Morphology
In the sacrum, stress fractures show sclerotic bands, linear fracture lines, or a combination of both that typically run longitudinally in the sacral ala. The insufficiency fractures usually show the classical H-pattern (Honda sign), while fatigue fractures in athletes (rare in this location) do not. With insufficiency, there may associated fractures in pubic rarui. In the femur and tibia, stress fractures are usually the result from overuse in sports. They can involve the femoral neck or diaphysis, medial tibial plateau, tibial diaphysis, or medial malleolus. In the femoral neck, these can be located medially (compression) or laterally (tension). In the/oof, stress fractures of the navicular occur in young performance athletes and typically show a sagittal fracture line involving the central third. CT is important as it guides duration of treatment, which can be as long as 4 months. Less frequently, fatigue fractures can occur in the calcaneus, talus, or cuboid (see Fig.25.40b). March fractures of the metatarsals are usually managed without CT.
and Degenerative
Disease
of the Spine
Intervertebral Disc Lesions
Standardized
Injury to the intervertebral disc and the discovertebral junction can result from acute trauma or from subtle subclinical, often endogenous, injury. Depending on the direction, an anterior, intravertebral (ventrocaudal or ventrocephalad) or posterior herniation of disc material is differentiated.
According to the recent 2001 NASS/ ASSRI ASNR consensus report (Fardon and Milette, 2001), lesions of the intervertebral disc are described as bulging, herniated disc, son sequestration (Fig. 25.44). The term herniated nucleus pulposus (HNP) is imprecise and should be avoided. Bulging is reserved for a generalized extension of disc material beyond the vertebral apophyses occurring in more than 50% of the disc circumference. Disc herniation can be further classified as a protrusion or an extrusion. A disc protrusion
Terminology
25 Musculoskeletal System
Protrusion
Extrusion
Extrusion
is ruptured, but a sequestrum may also move subligamentously. Discrimination from an extrusion requires the definition of the discontinuity, which is sometimes difficult to assess. Migration of disc material can occur either with or without a sequester. Regarding localization, in the transverse plane the terms central, subarticular, foraminal, and extra foraminal are suggested. In the sagittal plane, location is defined as discal, infrapedicular, pedicular, or suprapedicular. Disc Degeneration
No migration
Subligamentous migration
Sequestration
000 Protrusion
Extrusion
Asymmetric bulging disc
Fig. 25.44 Terminology for classification of intervertebral disc disease.
means that disc material extends into the inner part of the annulus fibrosus, leading to a protrusion of the intact outer part of the annulus fibrosus beyond the dorsal contour of the vertebral body. The greatest distance between the edges of disc material beyond the disc space is less than the distance between the edges of the base in the same plane. Protrusions can either be focal (base < 25% of disc circumference) or broad-based (25-50% of circumference). A disc extrusion is associated with herniation of nucleus pulposus material through a tear in the annulus fibrosus and mostly (but not always) through a ruptured posterior longitudinal ligament. In at least one plane the distance between the edges of disc material beyond the disc space is greater than the distance between the edges of the base in the same plane (Fig. 25.44). A sequestrum refers to an extrusion with disc material that is seen separated from the parent disc and is dislocated upwards or downwards into the spinal canal (Fig. 25.44). Most often, the posterior longitudinal ligament
With aging, disc degeneration results in desiccation, narrowing of the disc space, diffuse disc bulging, annular tears of the annulus fibrosis, defects and sclerosis of end plates, and osteophyte formation. Spondylosis deformans refers to changes that occur with normal aging, which affect primarily the annulus fibrosus and ring epiphysis. Findings include fibrosis of the disc with preservation of height, anterior and lateral osteophytes, and, in a minority of people, annular tears. Intervertebral osteochondrosis refers to pathologic disc degeneration involving the nucleus pulposus and vertebral end plates. Findings include thinning of disc with frank bulging, end-plate erosions with osteosclerosis, a central vacuum phenomenon, posterior osteophytes and extensive annular tearing (see 25.48 b). Disc Herniation
Because of its superior soft-tissue resolution, MRI is the preferred method of diagnosis for disc pathology, and even more so in the examination of the cervical and thoracic spine. CT is still frequently used for the lumbar region and is superior for detection of osseous abnormalities. Studies in asymptomatic groups have shown that disc bulging and protrusions are common features, which may not be clinically relevant. Extrusions are infrequent and sequestrations do not occur, so these are always pathologic and also usually symptomatic. Correlation of imaging findings with clinical pain symptoms is often difficult, and pain may be either localized in the back or radiate to the buttocks or legs. Disc herniations are commonest in
Intervertebral Disk Lesions the L5-S1 and L4-L5 interspaces, and their frequency is less at the LI-L2 level. The so-called failed back surgery syndrome (FBSS) occurs in 25-40% of patients after surgery of the spine. Reasons for FBSS are a residual or recurrent extrusion or hypertrophic epidural fibrous tissue. The differentiation is based on morphologic criteria and contrast enhancement, the latter being more conspicuous in MRI. •
CT Morphology
A disc protrusion appears as a smoothly marginated, broad-based bulge at the level of the disc space that displaces the epidural fat layer and may indent the dural sac anteriorly. A foraminal (lateral) disc protrusion may extend into the lower part of the neuroforamen without compressing the spinal nerve that courses above it. A disc protrusion is usually associated with narrowing of the intervertebral disc space and is sometimes associated with a vacuum phenome-
non resulting from pathologic degenerative disc changes. With CT, accurate differentiation from disc extrusion can only be made when sagittal reformations are used, and therefore thin-slice volume data sets are important for improved diagnosis (Fig. 25.45). A disc extrusion leads to a circumscribed. convex, more lobulated displacement of disc material at the level of the disc space. A clinically significant extrusion leads to displacement or compression of the dural sac, nerve roots, or spinal nerves (Fig. 25.46). An extrusion may be confined beneath an intact posterior ligament, but, most commonly, rupture occurs posterolaterally because the lateral portions of the ligament are relatively thin and weak. This subarticular (posterolateral) disc herniation is seen most commonly in the lumbar spine (L4-5 and L5-S1) and is often associated with dural sac or nerve root compression (Figs.25.46a, b; 25.47 d). On the other hand, midline herniations usually remain subligamentous. CT imaging,
Fig. 25.45 Disc protrusion on an axial reformation parallel to the intervertebral disc space (a) and on a sagittal MPR (b).
a
b
b Fig. 25.46 Schematic representation of the commonest forms of intervertebral disc herniation, illustrated here for the L4-L5 interspace, (a) Posterolateral herniation: the most usual form, causing com pression of the intraspinal nerve root (here L5). (b) Foraminal herniation: an extensive herniation com-
presses the spinal nerve within the foramen (here L4) and also the intraspinal nerve root (here L5). The rare lateral extraforaminal herniation compresses only the spinal nerve, (c) Medial herniation: this form presents clinically with lumbalgia and alternating or bilateral radiculopathy.
978
25 Musculoskeletal System
a
b
Fig. 25.47 Iritervertebral disc herniation (disc extrusion). (a) Typical posterolateral disc extrusion causing compression of the right S1 nerve root (L5-S1 interspace). (b) Lateral extrusion with extraforaminal compression of the right L4 spinal nerve (L4-L5 interspace).
however, is inaccurate in differentiating between sub ligamentous and transligamentous herniations. , ' /\ The herniation may extend upward or downward from its site of origin, or the herniated tissue may separate from the rest of the disc and form a sequestrum in the spinal canal. The epidural fat at the level of the extrusion or sequestrum is displaced. The degree of compression or displacement of the dural sac and nerve roots caused by a ruptured disc depends on the size of the extrusion. The extrusion itself is hyperdense to the musculature, but its CT attenuation can vary greatly as a result of calcifications and a vacuum phenomenon (Fig. 25.4 7). Calcifications within the herniated material are considered to indicate a chronic lesion. Anterior disc herniation leads to elevation of the anterior longitudinal ligament and subsequent formation of peripheral osteophytes. It does not have any therapeutic consequences (Fig.25.48a) . Intra vertebral disc herniation occurs into the ventrocaudal as well as ventrocephalad direction. The latter is less common and produces an abnormality described as "limbus vertebrae"; herniation of disc material into a vertebral body at the site of attachment of the annulus fibrosus to the body's rim separates a small triangular fragment of bone, which should not be mistaken for an acute fracture, an infectious spondylitis, or the secondary ossification center of the apophysis in growing patients. Reactive bone
c
d
(c) Foraminal extrusion on the right side with a vacuum phenomenon (L4-L5 interspace), (d) Calcified posterolatera! disc extrusion on the right side (L3-L4 interspace),
a
b
Fig. 25.48 (a) Anterior disc extrusion leads to osteophyte formation but has no therapeutic consequence. Note the intravertebral herniations through the end plates (Schrnorl's nodes) are characteristic for Scheuermann's disease, (b) Intervertebral osteochondrosis with severe disc degeneration, chondrocalcinosis of the discs (arrowheads), and a vacuum phenomenon in a region of erosive osteochondrosis (arrow).
sclerosis adjacent to the defect indicates a chronic process. The adjacent disc space is typically narrowed and a vacuum phenomenon may be seen in the disc space. ' Intravertebral disc herniation through the vertebral end plate is known as Schmorl's node (Fig.25.48a). Involvement of more than three consecutive thoracic vertebrae by Schmorl's nodes is described as Scheuermann's disease. In this disease, a wave-like outline of adjacent
Intervertebral Disk Lesions
become secondarily vascularized and may then show a peripheral enhancement. A sequestrum and space-occupying fibrous tissue may be differentiated by contrast enhancement in images obtained 5-10 min p.i. in association with morphologic criteria. Knowledge of the surgical or interventional procedure is required for proper assessment of A soft disc herniation has to be differentiated postoperative findings. With laminectomy varifrom a so-called hard disc herniation because able parts of the dorsal arc including the ligait has therapeutic implications. While the mentum flavum are removed. With laminotomy former is (acutely or subacutely) herniated the defect is much smaller. In the early postdisc material with soft tissue attenuation, the operative phase, edema and hematoma lead to latter shows as secondary calcified disc mateincreased attenuation of the soft tissue. Late rial and osteophytic rims (Figs.25.47d; sequelae are a variable amount of fibrous tissue 25.48 a). A hard disc herniation is considered or-depending on the surgical technique- interto indicate a chronic process which leaves ponated fatty tissue (Fig. 25.49). Discectomy is open whether the clinical symptoms are due the almost total resection of the disc usually to compression by osseous or by disc material performed after laminectomy. A discectomy or by both. may be combined with a fusion operation of the two adjacent vertebra using an interponated In failed back surgery syndrome (FBSS) the differ- bone fragment (autograft or allograft). Frequent late findings after any type of spinal operation entiation between a residual or recurrent extrusion and hypertrophic epidural fibrous tissue is are advanced degenerative changes in the facet based on morphologic criteria and contrast en- joints due to the decreased height of the disc hancement. Residual or recurrent herniated disc space after intervention and the altered mematerial has usually the same attenuation as the chanical stress of the spine in adjacent segments. Facet joint degeneration leads to a varidisc in place, has a lobulated or polypoid shape, able degree of spinal canal stenosis. A pseudoand shows a continuity with the disc. Fibrous of the arachnoid tissue on the other hand has a variable shape meningocele (herniation and unsharp borders. Fibrous tissue shows a through a dural tear), or infection (arachnoiditis), are rarer complications, representing inmeasurable, and sometimes nonhomogeneous enhancement after intravenous contrast applidications for MRI examinations. cation, with a stronger enhancement in the ventral epidural fibrous tissue, especially in the first Pitfalls: care has to be taken not to misinter6 months postoperatively. Disc material, on the pret the intervertebral veins or a hypertroother hand, shows no contrast enhancement up phied ligamentum flavum, which is to IOmin p.L, and very little enhancement after frequently also calcified in patients older than 30-40 min p.i. due to diffusion (Fig. 25.49). 50 years, as herniated disc material. Rarely does chronically herniated disc material
vertebral end plates has been described and one may find a sagittal elongation and anterior wedging of the vertebral bodies, and a kyphotic curve ofthe thoracic spine (juvenile thoracic kyphosis). Most of these changes can be detected most easily on sagittal reformations.
Fig. 25.49 Failed back surgery syndrome (FBSS). There is a partial left hemilaminectomy after microsurgery of a massive disc herniation at the level L3/4. The slight residual left-sided protrusion of the L3/4 disk is hyperattenuating on the precontrast scan (a). The contrast uptake in fibrous scar tissue (arrows) is best seen after 5-10 min (b).
a
b
25 Musculoskeletal System Spinal Stenosis
Narrowing of the bony spinal canal can have numerous causes, but the commonest mechanism is degenerative changes in the facet joints and spinal ligaments that cause narrowing of the lateral recesses or neuroforamina. If the changes are bilateral, they lead to concentric spinal stenosis. The lordotic portions of the cervical and lumbar spine are most commonly affected, and the caudal motion segments are sites of predilection. Congenitally short pedicles, shortened laminae, sagittal orientation of the facet joints, and common nerve root origins are predisposing factors for spinal stenosis. Spondylolisthesis is an anomaly characterized by defects in the interarticular part of the vertebral arch, while pseudospondylolisthesis is characterized by degenerative changes in the facet joints. Both conditions lead to anterior displacement of the affected vertebra and may cause spinal stenosis. CT is the technique of choice for detecting and quantifying spinal stenosis. •
CT Morphology
Normal values for the diameter of the bony lumbar spinal canal are shown in Table 25.18. Studies indicate that the lateral recesses of the spinal canal measure more than 5 mm in their sagittal dimension between the posterior margin of the vertebral body and the anterior margin of the superior articular process. A value less than 2 mm is definitely pathologic. The lumbar neuroforamina normally display a craniocaudal extent of 5 mm or more. Normal ligamenta flava have a thickness of 4-5 mm. The critical surface area of the spinal canal for spinal stenosis has been cited as 0.75 em", Narrowing of the spinal canal may result from degenerative changes (osteoarthritis) in the facet joints (spondylarthrosis), hypertrophy of Table 25.18
the ligamenta flava, disc protrusion, or spondylophytosis. The resulting canal often assumes a trefoil shape. The primary signs of osteoarthritis of the facet joints are joint space narrowing with subchondral sclerosis, hypertrophy of the articular processes and capsuloligamentous structures, and subluxation and vacuum phenomena in the facet joints (Fig. 25.50). Narrowing of the cervical spinal canal may be caused by hypertrophy of the uncovertebral joints (uncovertebral osteoarthritis). Rarely, a synovial cyst of the facet joints may cause narrowing of the lateral recess. Such synovial cysts represent intraspinal, extradural extensions of the joint capsule of arthrotic facet joints, usually occurring at the levels from L3 to S1. The CT attenuation of the cyst is highly variable and can range from hypo attenuating (airor fluid-filled) to hyperattenuating (partially calcified). The diagnosis is suggested by the posterolateral location of the lesion in the spinal canal and a demonstrable connection with the joint space (Fig. 25.51). Doubts are resolved by contrast -enhanced CT or MRI, which reveal a typical annular pattern of enhancement of the synovia. Nerve compression in the neuroforamina does not occur until the upper portion of the foramen has become narrowed. This may be caused by any of the changes noted above or by posterolateral spondylophytes on the vertebral body margin. The craniocaudal extent of the neuroforamina can be best demonstrated on parasagittal MPR, which are well suited for detecting and differentiating the underlying cause of narrowing of the foramina (Fig.25.50b). Pseudospondylolisthesis always leads to narrowing of the neuroforamina and the spinal canal because the entire vertebra, including the spinous process, moves forward leading to a step-off at the interspace below the level of the slipped vertebra. By contrast, there is no obliga-
Normal dimensions of the bony spinal canal in bone-window CT (Ulrich, 1980)
Sagittal diameter
Axial surface area
LI
17mm (14-25)
2.4cm2 (1.7-3.5)
L2
16mm (13-28)
20mm (15-28)
2.4 ern" (1.5-3.5)
L3
15 mm (13-27)
20mm (16-30)
2.2cm2(1.5-3.5)
L4
16 mm (13-27)
22 mm (17-35)
2.4cm2 (1.6-3.9)
Imm(12-3i
27 mm (19-40),
t2.9 ern" (1.5-5.6)
,WI:S>!,'"
,
In tervertebral Disk Lesions
a
a
b
Fig. 25.50 Spinal stenosis caused by bilateral short pedicles, a sagittal orientation of the facet joints, spondylarthrosis, hypertrophy of the ligamenta flava, and intervertebral disc protrusion (a). Stenosis of the cervical neuroforamina is mainly caused by hypertrophied (osteoarthritic) facet joints (b).
tory narrowing of the intervertebral foramina with true spondylolisthesis; while the body, the pedicles, and the superior articular process slip anteriorly, the spinous process, laminae, and the inferior articular process remain in their normal position due to the defect in the interarticular part of the vertebral arch. The step-off is located at the interspace above the level of the slip. Spinal stenosis may be caused by anterior vertebral body displacement in spondylolisthesis, by the osteoarthritic changes in pseudospondylolisthesis, or by bony hypertrophy at the site of the defect in spondylolysis (Fig. 25.52). The grading of spondylolisthesis, as proposed by Myerding, is based on the amount of forward displacement and is quantified in 25% steps of the sagittal diameter of the vertebral body (1=0-25%, 11=25-50%, 111=50-75% and IV =75-100% displacement) .
b
Fig. 25.51 Synovial cyst of the L4-L5 facet joint on the left side (a). Cyst of the L3 nerve root sheath on the left side (b).
Pitfalls:
• The osseous defects and hypertrophy in spondylolysis should not be mistaken for the facet joints located below the defects. • The detection of disc tissue between the vertebral body margins in spondylolisthesis should not be misinterpreted as a herniated disc. In doubtful cases sagittal MPR clarifies the anatomic situation.
Fig. 25.52 Spondylolysis with bilateral defects of the pars interarticularis (4 X 1/6).
a
b
981
Infection and Inflammatory Processes
MRI is generally superior to CT for evaluating in- • CT Morphology flammatory diseases because it can assess the Septic arthritis leads to destruction of the subextent and activity of the process while also de- chondral bone plate, followed by ill-defined tecting associated edema. This applies to ar- j oint contours and erosive changes that can thritis (including sacroiliitis) , osteomyelitis, culminate in bone destruction (Figs. 25.53; spondylitis, and spondylodiscitis. Exceptions are 25.54a). joint effusion and surrounding soft diagnostic or therapeutic CT-guided interventissue swelling are accompanying features. Airtional procedures involving the axial skeleton. bubbles may suggest gas-forming microor-
ganisms (Klebsiella, Serratia).
Infectious Arthritis
Infectious arthritis can be pyogenic (bacterial) or non-pyogenic (TB, viruses, fungi). The commonest organisms cultured in septic (pyogenic) arthritis are Staphylococcus aureus, Haemophilus influenzas, and Klebsiella and Serratia. Infectious agents may invade the joint directly through a penetrating wound, via a surrounding soft-tissue infection, or hematogenously. Tuberculosis and other mycobacteria (like Mycobacterium avium-intracellulare) may lead to granulomatous synovitis. Septic arthritis needs rapid therapeutic action to avoid severe joint damage, especially in children. CT is not commonly performed forjoint infection. After conventional radiography, scintigraphy (bone or leukocyte scans) or contrast-enhanced MRI are the methods of choice.
a
b
Fig. 25.54 (a) Acute osteomyelitis of the talus with septic arthritis in the surroundingjoints. Note the bone erosions (arrowheads) and subchondral osteoporosis (wide arrows) as well as the faint bone sclerosis due to
Fig. 25.53 Sepfic'arth'ritis of the sacroiliac Joint in a IV drug user. Irregular joint margins with complete destruction of the subchrondral bone. Note the abscess (arrows), the beginning osteomyelitis (arrowhead) and the signs of periostitis.
c incipient talar necrosis (arrows), (b) Tuberculous osteomyelitis ofC1 with bone destruction and abscess formation. (c) Chronic osteomyelitis with sclerosis ofthe right pubic bone.
Infectious Spondylitis and Spondylodiscitis Osteomyelitis Infection of the bone and medullary cavity is regularly seen in daily practice. Infectious agents may directly invade the joint in the setting of trauma or surgery. Other routes are via a surrounding soft-tissue infection or hematogenous spread. In children and in the adult spine the hematogenous route is the most usual way for Staphylococcus aureus infection to spread. When present. the physis acts as a natural barrier to spread. Although nowadays frequently replaced by MRI. CTcan be a useful adjunct to radiography as it can detect cortical destruction and periosteal reaction. as well as the formation of an involucrum (shell of living new bone over dead bone). the tract of a cloaca (opening for discharge of sequester). and sequestrated dead bone. Spiral or multislice CTenables high-quality MPR or 3D reconstructions for optimal visualization. •
CT Morphology
In acute osteomyelitis, initial findings include soft-tissue swelling, osteoporosis, and medullary resorption with indistinct borders (Figs. 25.53; 25.54). Later this progresses to cortical lucencies and periosteal reaction, with subperiosteal abscess formation and creation of an involucrum. This sequence is particularly seen in children. In adults cortical erosion and fracture can occur. Reparative changes are prominent in chronic osteomyelitis (Fig.25.54c). Irregular sclerosis around osteolytic areas, solid periosteal reactions, and the possible formation of a Brodie's abscess or cortical sequester, may be seen, with or without sinus tracts originating in cloacae. Chronic Recurrent Multifocal Osteomyelitis This is a rare chronic form of aseptic osteomyelitis from unknown cause that primarily affects children and adolescents. The disease is seen as a distinct entity but there are known associations with bony lesions from psoriasis, palmoplantar pustulosis, and acne conglobata as well as overlap with features of Carre's sclerosing osteomyelitis and the SAPHO syndrome. The
disease has a chronic'intermittent can take more than 15 years.
course
that
CT Morphology
Chronic recurrent multifocal osteomyelitis is manifested by single or multiple metaphyseal lesions that occur predominantly in the tibia, femur, clavicle, and fibula. The lesions are osteolytic and eccentric and are surrounded by rims of bone sclerosis. Periosteal new bone formation is variable. Infectious Spondylitis and Spondylodiscitis Spondylitis is a hematogenous infection that starts in the anterior medullary cavity of a vertebral body, eventually penetrates the cortex, and spreads along the paravertebral ligaments, or it may penetrate the vertebral end plates and involve the disc space (Spondylodiscitis). It is most frequent in the lumbar spine. Most frequent causative microorganisms are S. aureus, Pseudomonas aeruginosa (drug addicts). Streptococcus (endocarditis). H influenza (meningitis), and Mycobacterium tuberculosis. CT Morphology
In the early stage, the disc space shows a decreased height and decreased density. Within 1-3 weeks paras pinal soft-tissue swelling and osteolytic areas appear in the adjacent vertebral end plates (Fig. 25.55). Rarely, the infectious process destroys a single vertebra or part of the vertebra without invasion of the disc. In cases that respond to antibiotic therapy, serial CT examinations show increasing sclerosis of the cancellous bone. The process may culminate in the fusion of adjacent vertebral bodies (block vertebrae). In other cases the vertebral body may be destroyed, resulting in vertebral wedging and gibbus deformity (most typical for tuberculous Spondylitis). Paravertebral abscesses may gravitate downward along the psoas sheath and spread beneath the inguinal" ligament (gravitation abscess, cold abscess). Tuberculous Spondylitis cannot be differentiated definitively by imaging. Clues to a tuberculous origin are involvement of more than two vertebrae, sparing of disc spaces (Spondylitis), thoracic spine localization, primary spread via the longitudinal ligament with a resulting large
25 Musculoskeletal System Fig. 25.55 Spondylodiscitis with complete destruction of the end plates ofT6 and Th7 (a). Postoperative Spondylodiscitis with asymmetric erosion and bone sclerosis (b). Note the drill holes after explanation of the osteosynthetic material.
a
b
paras pinal abscess that may also, extend Into the epidural space, gibbus deformation, and eburnation (diffuse sclerosis of the vertebral body surrounding osteolytic areas). The most usual complication of tuberculous spondylitis is compression of the thecal sac and the spinal cord with resulting paraplegia. Sacroiliitis
MRI has become the most important modality for diagnosing sacroiliitis, as it furnishes information on inflammatory joint changes and can distinguish active inflammation (joint effusion, synovitis, medullary edema) from inactive chronic inflammation (fatty involution of the medullary cavity in a setting of fibrosis or sclerosis). The role of CT is limited to the detection of bony abnormalities. Conventional tomography is considered obsolete. •
Fig. 25.56 Sacroiliitis due to ankylosing spondylitis. Irregular sclerotic bands and erosive defects are present in both sacroiliac joints.
scan). A more circumscribed sclerosis is seen in osteitis condensans ilii, common in middle-aged active females. The dominant feature of septic sacroiliitis (confirmed by serology) is joint destruction (see Fig. 25.53).
CT Morphology
The early stage of sacroiliitis is characterized by j oint surface irregularities (intra -articular new bone) and microfractures in the cancellous bone. Further progression leads to erosive changes, irregular joint space narrowing, and subchondral sclerosis (Fig. 25.56). Partial or complete ankylosis of the sacroiliac joint occurs as an end stage or late sequel. Differentiation is required from osteophytes and ligamentous calcification due to degenerative changes involving the anterior and posterior joint margins (axial scans) or the superior and inferior joint margins (semicoronal
Heterotopic Ossification (Myositis Ossificans)
Heterotopic ossification can have various causes but presents similar morphologic features. The primary form (generalized progressive myositis ossificans, Munchmeyer's disease) is very rare. Secondary forms are most often caused by trauma, surgical procedures (e.g., total hip replacement), or prolonged immobility due to head trauma, spinal cord trauma, or-long-term ventilation. As a rule, surgical treatment is indicated only if there is severe restriction of move-
Soft-tissue Infections ment or significant deformity. The goal of surgical treatment is complete resection of the heterotopic ossification. Surgery should be deferred until the lesion has "ripened" (approximately 6 months after trauma or 12-18 months following head trauma or long-term ventilation). The differential diagnosis should include paraosseous osteosarcoma, which usually shows central rather than peripheral ossification. •
CT Morphology
Preoperative CT should let the surgeon know how far the lesions have "ripened", i.e., the degree of calcification of the lesions, the periand/ or para -articular location of the heterotopic foci, and their relation to muscles, nerves, and blood vessels. Early lesions are often hypoattenuating to surrounding muscle. Later they calcify from the periphery toward the center while also increasing in size. The lesion is assumed to be "ripe" when it has become completely ossified. Long-term ventilation and head injuries are typically associated with para -articular ossification, while direct trauma tends to induce periarticular ossification, which is usually localized to the area of the trauma (Table 25.19). The most commonly affected joints are the shoulder, elbow, hip, and knee. Three-dimensional reconstructions are helpful in defining the extent and relative location of the ossified areas (Fig.25.57).
Table 25.19
Common sites of heterotopic ossification
• May occur anywhere about the elbow after trauma, usually on the medial or lateral side with frequent involvement ofthe ligaments; on the ulnar side, the ulnar artery and nerve are frequently encased by new bone • Posterior ossification is most common following* head injury; the posterior aspect of the elbow is the most common site in general for heterotopi] ossification in head-injured patients A
Shoulder
• No site of predilection after trauma; frequent involvement of coracoacromial ligament • A common site after head injury is the inferomedial area below the shoulderjoint; pseudojoint without ankylosis Hip
• Ossification in the abductors may occur after total hip replacement or trauma; the foci may enlarge and involve the femoral neck • In head-injuries patients: anteromedial (anterior superior iliac spine -> teres minor) or anterolateral (anterior superior spine -> teres major) periarticular ossification or inferomedial paraarticular ossification adjacent to teres minor Knee
No specific location; the quadriceps tendon is common site of post-traumatic ossification, an medial ossification (similar to Stieda-Pellegrini sign) is common after head trauma • Surgical intervention is usually not required
Soft-tissue Infections
Fig. 25.57 SSD demonstrates extensive heterotopic ossification along the left sacrotuberal ligaments and about the right hip joint; following; long-term ventilation.
Soft-tissue infections can be primary or are associated with j oint infection, osteomyelitis, or diabetes. Primary infections include abscesses, pyomyositis, cellulitis, or fasciitis. Abscesses can be caused by direct inoculation from penetrating wounds or by hematogenous spread from distant sites. Infection is preceded by a phlegmonous phase before it is walled-off and liquefaction occurs. Pyomyositis is uncommon in temperate climates, but is more often seen in the tropics or in immunocompromised patients. Usually the lower extremity is involved with formation of multiple small abscesses. Cellulitis is a phlegmonous infection of the superficial soft tissues with thickening of the skin, subcutaneous fat and fasciae. Necrotizing fasciitis is a
985
25
Musculoskeletal
System
rare, rapidly progressive infection of the deep fasciae and surrounding muscle with massive necrosis rapidly leading to septic shock. In the extremities, ultrasound and MRI are the methods of choice. There is no primary role for CT, but soft -tissue changes may be seen frequently in studies done for bony abnormalities. CT is used in examinations of the trunk (mediastinum, retroperitoneum, pelvis; see the respective chapters). Bursitis
Bursitis is not an indication for CT scanning but may be an incidental finding that can be mistaken for a cystic mass. •
CT Morphology
Bursitis manifests as a fluid-filled mass in a periarticular location that is characteristic for eachjoint. A iliopsoas bursitis may mimic a cystic mass but is seen in a characteristic location medial and cranial to the hip joint within the iliopsoas muscle (Fig. 25.58). Differentiation
b
a
Fig. 25.58 Iliopsoas bursitis presenting as a fluiddense zone within the iliopsoas muscle in a patient with subchrondral cysts in severe osteoarthritis of the hip joint.
from joint recesses and synovial cysts is only possible by their para-articular location and the lack of demonstration of a communication with the joint.
iibne Tumors" Primary bone tumors are relatively uncommon lesions. Diagnosis is still primarily based on patient characteristics and plain radiographs. Additional imaging is mainly done with radionuclide scanning and MRI. CT has a limited role but may be used to assist in the diagnosis of selected bone-forming tumors, to evaluate bone destruction, to assess skeletal stability, and to guide percutaneous ablative therapies. Skeletal metastases in adults are far commoner than primary bone tumors and occur predominantly in the axial skeleton (skull, spinal column, ribs, pelvis) due to the extensive paravertebral plexus of valve-less veins. The bones are imaged in every CT examination, and bone-window scans should be obtained whenever there is clinical suspicion of an osseous infection or neoplasm. Equivocal findings are resolved by CT-guided percutaneous procedures for cytological, histolpgical,. or microbiological sampling.
Primary Bone Tumors
Plain radiographs continue to be the mainstay of the differential diagnosis of bone lesions. CT is indicated if there is uncertainty as to the presence of matrix calcification or ossification, or if it is necessary to detect incipient erosion or define the extent of gross destruction. MRI is the modality of choice for defining the soft -tissue component and detecting invasion of the medullary cavity. The staging of bone tumors (according to Enneking) is based on the histological type and grading, the definition of tumor extent, and the diagnosis of metastases (lung or bone). A differentiation is made between intra compartmental and extracompartmental localization of tumors. Intracompartmental tumors are: • soft tissue lesions confined to the subcutaneous fat; • parosseous lesions that neither infiltrate the bone nor the muscles;
Primary Bone Tumors • intraosseous lesions that have not penetrated the cortical bone; and • soft-tissue tumors that lie within one muscle compartment and have not crossed the surrounding fascia Extracompartmental tumors are: • all tumors that have penetrated the predefined anatomic borders of fascia or cortical bone, or lie within different muscle compartments; • all tumors that infiltrate nerves or vascular structures; and • all tumors that have invaded joints. In addition there are some anatomic areas that are primarily extracompartmental because they lack predefined anatomic borders. These areas are the middle and back-foot, middle hand, popliteal fossa, femoral fossa, and the axilla. MRI is the method of choice for staging of benign and malignant bone lesions. While MRI is frequently not able to define the type of lesions (which is based on radiographic criteria and histology), it plays an important role in differentiating between tumors and infarcts or osteomyelitis. Dynamic contrast -enhanced studies may help in differentiating benign from malignant cartilaginous tumors or in assessing response of bone tumors to chemotherapy. CT has similar potential for the differentiation of intra- versus extracompartmental lesions. Matrix calcifications in a lesion are best illustrated by CT and are useful in differentiating a chondrogenic tumor from an osteogenic tumor. CT can delineate the thin bony wall of an aneurysmal bone cyst, and it can also demonstrate the diffuse permeative destruction and extraosseous soft-tissue component associated with Ewing sarcoma. In these cases CT provides typical morphologic imaging features and in addition makes an essential contribution to the assessment of bone stability. Enostosis
Enostoses or bone islands are common incidental findings that have no pathologic significance. They may occur anywhere and are observed at all ages.
b
a
Fig.25.59 Enostoma in the calcaneus (a). Bone island in L2 (b: 4x1/6, curved coronal MPR). The very dense, homogeneous sclerosis is a typical finding.
•
CT Morphology
Bone islands (Fig. 25.59) appear as sharply demarcated sclerotic foci that are usually located in cancellous bone (ubiquitous in osteopoikilosis). They can be round or ovoid and are usually aligned with the long axis of the trabeculae. Some lesions can be difficult to distinguish from osteoblastic metastases on CT scans but are nonenhancing on MR images and do not show tracer uptake on radionuclide scans. Osteoid Osteoma
Osteoid osteoma most commonly presents in the second and third decade with pain that is typically worse at night and relieved by aspirin. Most instances occur in the shafts of the femur and tibia, but lesions are also common in the hands or feet. Spinal occurrence (approximately 5%) is manifested by a painful scoliosis in which the tumor is located in the posterior elements on the concave side of the curve. Although surgery is the therapy of choice, a newer application for CT is the guidance of locally ablative therapies. Good results have been described with thermocoagulation, interstitial laser photocoagulation, and radiofrequency ablation (see Chapter 6, p. 181).
987
25 Musculoskeletal System
•
CT Morphology
Osteoid osteoma presents as a small nidus of vascular osteoid tissue approximately 0.5-1.5 cm in diameter (the actual tumor), which may have a sclerotic center. It is often, but not always, surrounded by a dense, homogeneous zone of reactive sclerosis. The nidus shows intense contrast enhancement, especially during the early phase, although there is no need for contrast administration in most cases. The CT diagnosis of osteoid osteoma requires thin sections (1-3 mm). For lesions in long tubular bones, a wide window setting (4000 HU) is needed to detect the hypoattenuating nidus within the high-density cortical bone (Fig.25.60a). . ... The cortical type of osteoid osteoma is the most frequent and exhibits intense reactive sclerosis around a nidus located in the cortex (Fig.25.60a). More than one nidus may be present (multifocal osteoid osteoma, very rare). The medullary type is characterized by a densely calcified nidus within the medullary cavity (Fig.25.60b), which is surrounded by a halo of hypervascularized osteoid tissue. In the subperiosteal type, the nidus is calcified and there is little periosteal response and no reactive sclerosis. The periarticular type is located close to a joint and shows little if any evidence of reactive sclerosis (Fig.25.60c; see also Fig. 6.13). Differentiation is required from osteomyelitis with a sequestrum and from Brodie's abscess, in which the central hypodensities are often
a
b
Fig. 25.60 Osteoid osteoma in the femoral shaft (cortical type), displaying a central nidus surrounded by a dense sclerotic zone (a). Medullary osteoid osteoma in the articular process of L2 with a calcified nidus (b). Sub-
larger and irregular. In addition, Brodie's abscess is located close to the growth plate and often demonstrates a serpiginous tract. Both entities often display heterogeneous sclerosis. In the spine the osteoid osteoma may simulate a sclerotic metastasis but demonstration of the (calcified) nidus is pathognomonic. Enchondroma, Osteochondroma, and Chondrosarcoma
Enchondroma, osteochondroma, and chondrosarcoma are cartilaginous tumors. Chondromas may arise within a bone (enchondromas) or on the surface [periosteal chondromas}. They occur in the meta- or diaphysis of tubular bones, most commonly in the small bones of the hand (phalanges and metacarpals). Osteochondromas (osteocartilaginous exostoses) occur on the outer bone surface and usually involve the metaphysis of long bones particularly in the vicinity of the knee and shoulder. With both tumors, pain is a sign of malignant transformation with the development of a secondary chondrosarcoma. Enchondromatosis is characterized by multiple enchondromas {Ollier's disease}. The disease has a strong preference for a single side of the body and leads to knobby swellings of the hand, forearm, or legs. The risk of malignant transformation is much higher. The association of enchondroma to sis with soft tissue hemangiomatosis is called Maffucci's syndrome.
c periosteal osteoid osteoma of the sixth thoracic vertebra with moderate calcification of the nidus and almost no reactive scleroses (c).
Primary Bone Tumors
Chondrosarcomas most commonly occur in
the pelvis and femur (together accounting for approximately 40% of cases). They can be divided in conventional, juxtacortical', or mesenchymal types. Low-grade sarcomas may be difficult to differentiate from enchondromas in contrast to clearly malignant high-grade tumors. Dynamic contrast -enhanced MRI is the diagnostic method of choice for this differentiation. •
CT Morphology
All three tumors are characterized by flocculent, curvilinear, or stippled areas of matrix calcification. Most enchondromas are associated with popcorn-like matrix calcifications. The softtissue component may occupy the entire diameter of the medullary canal. While the lesions are frequently completely radiolucent in short bones, they sometimes demonstrate extensive calcifications in long bones. A typical finding is the scalloping of the inner cortical margin, since the cartilage grows in a lobular pattern (Fig.25.61a). Periosteal chondromas, mostly occurring at the proximal humerus, may lead to a painful swelling and may erode the cortex in a saucer-like fashion, producing a solid buttress of periosteal new bone. Chondromas in long tubular bones show a markedly higher incidence of malignant transformation than those in short tubular bones.
a
b
Fig. 25.61 Enchondroma with typical intramedullary location, popcorn-like calcifications, and sharp margins (a). Osteochondroma of the s" rib with a,wide connec-
Osteochondroma
has a cancellous bone structure at its base that is similar to that of the adjacent host bone. Typically, there is an uninterrupted merging of the cortex of the host bone with the cortex of the osteochondroma. In addition, the medullary portion of the lesion and the adjacent bone communicate (Fig.25.61b). The cartilaginous cap of the lesion displays typical foci of matrix ossification. Enlargement of the cartilaginous cap in adults or a cap that is more than 2cm thick on CT scans and has irregular margins should raise suspicion of malignant change (in less than 1 %). More frequent complications may arise from the fact that the lesion compresses adjacent nerves, vessels, or bones. Multiple osteochondromas are present in hered-
itary multiple exostosis syndrome. Low-grade chondrosarcomas typically show
an eccentric expansile type of growth and display marked matrix calcification, thickening of the cortical bone, and endosteal scalloping. The higher-grade tumors are distinguished by a concentric growth pattern, little matrix calcification, necrotic areas «40HU), cortical destruction, and an extensive soft tissue mass (dedifferentiated chondrosarcoma, mesenchymal chondrosarcoma, Fig.25.61c). Periosteal c1wndrosarcoma has the same morphologic and pathologic features as central chondrosarcoma except for its localization on the surface of the bone. It may be indistinguishable from a periosteal osteosarcoma.
c tion to the bone and chest deformity (b). Chondrosarcoma with destructive growth and a large soft tissue component (c). Note the intratumoral calcifications.
989
25 Musculoskeletal System Giant Cell Tumor Giant cell tumor, also known as osteociastoma, is approximately twice as common as aneurysmal bone cyst and is the second commonest benign spinal tumor after hemangioma, Around 60% of the lesions are localized in the long bones, almost always at the articular end of the bone (proximal tibia, distal femur, distal radius, and proximal humerus), Giant cell tumor may undergo malignant transformation iin 5-10% of cases. CT Morphology
The typically expansile lesions are purely lytic, without a sclerotic rim or any periosteal reaction (compare Fig, 6,8), The cortical bone is eroded, giving the tumor a malignant appearance, although its histology is benign, Giant cell tumors are characterized by highly vascularized tissue containing giant cells of osteoclast type, The tumor has a CT density of 20-70 HU and shows marked enhancement after contrast administration (40-60 HU), Internal septation is visible in rare cases, The precise extent of the soft-tissue component is sometimes difficult to distinguish from reactive edema; MRI is superior for this purpose, Treatment usually consists of wide resection with implantation of an allograft or endoprosthesis, CT is also used for monitoring the incorporation of the graft into the normal bone, and for follow-up because of the high incidence of recurrences, Recurrent lesions exhibit malignant transformation, especially after radiation therapy,
Aneurysmal Bone Cyst Aneurysmal bone cysts occur in children and young adults (90% of patients are younger than 20 years) and most commonly arise in the metaphyseal area of the femur and tibia or in the posterior elements of the vertebrae (transverse and spinous processes, neural arches), There is no risk of malignant transformation, They can develop as a primary lesion or as a result of cystic changes in a pre-existing lesion (chondroblastoma. osteoblastoma, giant cell tumor, fibrous dysplasia),
CT Morphology
An aneurysmal bone cyst causes marked bone expansion (blowout) and cortical thinning, Internal septa are frequently observed, CT may demonstrate fluid levels within the loculi in bedridden patients, The soft-tissue component has a CT density of 60 -70 HU and is well vascularized, showing intense enhancement after contrast administration, A typical and important feature, well demonstrated by CT, is the integrity of the cortex, Recurrence after treatment (surgical removal and bone grafting) is frequent
Hemangioma Hemangioma is the second commonest benign vertebral lesion and is usually an incidental finding, The incidence increases with age, The most usual sites are the spine, especially the thoracic spine, and the skull, CT Morphology
The typical pattern of coarse, predominantly vertical, linear striations produced by the rarefied trabeculae creates a stippled pattern on axial scans through the affected vertebra, Generally the shape and size of the vertebral body are unchanged, The hemangioma may completely fill the vertebra, or involvement may be limited to small areas (Fig, 25,62), Fatty involution is typical, thus differentiation from an isolated focus of osteoporosis may be difficult If clinical symptoms are present, the lesions often show marked contrast enhancement An intraspinal component should be excluded prior to operative treatment or radiological intervention (embolization), In the tubular bones, hemangiomas show a lacelike pattern, and honeycombing as they do in the spine,
FibrousDysplasia Fibrous dysplasia is the most commonly found benign tumor-like lesion of bone, It is usually confined to a Single bone (monostotic) but occasionally is more Widely distributed (polyostotic), Asymptomatic lesions are usually detected incidentally during middle age, while symptomatic forms are usually diagnosed during childhood, The monostotic form most frequently in-
Primary Bone Tumors 991 Fig. 25.62 Vertebral hemangiomas. (a) Typical incidental finding of a thoracic lesion with a coarsened, rarefied, vertically oriented trabeculae that produce a stippled pattern on axial sections. There is extensive fatty infiltration of the medullary cavity, (b) Smaller focal lesion in the lumbar spine.
b
a
volves the femur, especially the femoral neck, the tibia, and the ribs. The polyostotic form has a striking predilection for one side of the body. The pelvis is frequently involved followed by the long bones, the skull ribs, and proximal femur. Sarcomatous transformation is extremely rare.
Pseudotumors
There are numerous entities that may simulate a bone tumor. In cases in which plain films are inconclusive, CT can help identify intraosseous ganglia, brown tumors of hyperparathyreoid-
A
•
CT Morphology
The spectrum of CT appearances is dependent on the lesion's proportion of osseous-to- fibrous content. Lesions with greater osseous content are more sclerotic while those with greater fibrous content are more radiolucent with a characteristic milky or smoky ground-glass appearance. Thus the morphology ranges from pure lytic areas and ground-glass densities to irregular bands of sclerosis. The ground-glass densities have attenuation values of 70-150HU. The cortex may be expanded and thinned or may show compensatory thickening (Fig.25.63a). Involvement of the skull base usually produces a homogeneous, well-circumscribed area of sclerosis and expansion with resultant narrowing of involved foramina. The differential diagnosis includes meningioma, which typically occurs in older patients and has a markedly enhancing soft-tissue component. The most frequent complication of the polyostotic form is the recurrent pathological fracture with resulting deformity.
a
b Fig. 25.63 (a) Typical ground-glass lesion of fibrous dysplasia involving the iliac wing. (b) Hemophilic pseudotumor due to recurrent bleeding with bone destruction.
25 Musculoskeletal System
ism, encystified bone infarcts, posttraumatic myositis ossificans, various infectious lesions and hemophilic pseudotumors (Fig.25.63b).
types can be better circumscribed. There are periosteal reactions, typically in the form of sunburst pattern or a Codman's triangle. Parosteal types typically show a lobulated tumor that grows outward from the cortex.
Osteosarcoma
Osteosarcoma is the second commonest primary tumor of bone. There are many different types, such as conventional, juxtacortical, and telangiectatic Osteosarcoma. The tumors most often present in the second or third decade, but lower-grade types can also be seen in the fourth to fifth decade. Most types are relatively more frequent in men. The most usual locations are the femur, tibia, and humerus, with most osteosarcomas occurring in the vicinity of the knee. Microscopically, they can be primarily osteoblastic, chondroblastic, or fibroblastic. Telangiectatic variants contain large cystic spaces filled with blood. ]uxtacortical tumors proliferate outward from the cortex. The parosteal type is typically of lower grade than other types. MRI is generally superior to CT in diagnosis and follow-up, but CT may be used to assist in diagnosis and staging of the tumor.
Ewing's Sarcoma
Ewing's sarcoma is a highly malignant neoplasm in children and adolescents. Conventional radiographs and MR are the techniques of choice for evaluation of diaphyseal tumors, while CT is well suited for describing the extent of bony destruction in flat bones such as the ribs, pelvis, and scapula. •
CT Morphology
CT demonstrates an ill-defined diffuse permeative or moth-eaten bone destruction with a malignant (aggressive) type of periosteal response and a large soft-tissue component. Skeletal Metastases
Most tumors show a mixture of osteosclerosis and osteolysis, although the telangiectatic type is primarily lytic (Fig. 25.64). The tumors are usually poorly defined, but the lower grade
Metastases are the most frequent malignant skeletal lesions. Approximately 25% of all primary malignant tumors are associated with skeletal metastases. Lesions in cortical bone are more conspicuous than in cancellous bone and can be detected at an earlier stage. While CT is better for delineating lesions in cortical bone,
a
b
Fig. 25.64 Osteosarcoma. (a) Telangiectatic lesion of the proximal tibia (3/5/2) with bone destruction and a
large soft tissue component, sacrum (4 x 1/6).
•
CT Morphology
(b) Osteosarcoma
oftt/e
Skeletal Metastases
a
c
b
Fig. 25.65 Metastases. (a) Metastatic breast cancer with destruction and pathologic fracture of L2, compression of the spinal canal and a substantial anterior soft tissue component, (b) Diffuse osteoblastic metastasis from prostatic carcinoma. There is extensive
concomitant osteolysis of the left arch of the atlas with incipient displacement of the dural sac. (c) Enhancing intrapelvic metastasis from thyroid carcinoma. The bone destruction is more extensive than the visible area of contrast enhancement.
MRI is superior for diagnosing metastases in cancellous bone that have not destroyed the trabecular architecture. While the determination of the type of a skeletal metastasis, whether it is predominantly sclerotic, lytic, or of a mixed type (Fig. 25.65) cannot furnish a specific diagnosis, it can help to narrow the differential diagnosis in patients with an unknown primary tumor (Table 25.20). CTis superior to MRI in assessing the stability of the affected bone. As rule of thumb , long bones become unstable when more than 50% of the cortex is destroyed. In the spine, the vertebra is considered unstable when the dorsal border of the body is destroyed, when the height of the body is reduced, and when MR imaging shows bone marrow infiltration within the entire vertebral body even if the cortical bone is still intact.
After radiation, hormone treatment, or chemotherapy primary lytic lesions may become sclerotic.
Table 25.20
Incidence of skeletal
Unlike inflammatory processes, spinal metastases spare the intervertebral disc space. If the posterior elements of a vertebral body have been destroyed, attention should be given to the extent of an accompanying intraspinal soft-tissue component (Fig.25.66). Less than IQ% of skeletal metastases are solitary lesions.
metastases
sderettc
lytic
20%
60%
Prostatic carcinoma
(35 %)
75%
10%
15%
Bronchial carcinoma
(10%)
5%
75%
20%
<1%
90%
10%
<1%
55%
45%
<1%
95%
Renal carcinoma Thyroid carcinoma Colon carcinoma
(25%) (8 %) {1-2%}
993
25 Musculoskeletal System
Lipoma and Liposarcoma
Lipomas are frequent incidental findings and generally do not have pathologic significance. Liposarcoma is the second commonest softtissue sarcoma and most occur in the proximal, lower or upper extremity. While high-grade lesions have a significant soft-tissue component, well-differentiated lesions may primarily contain fat and may be difficult to differentiate from atypical lipomas. • b
a
Fig. 25.66 Destruction of the posterior vertebral elements (a) with spinal infiltration by a large cell lymphoma (b).
Soft- Tissue Tumors
Most soft -tissue tumors are not an indication for CT scanning. MRI can better delineate these lesions, and ultrasound is preferred for imaging superficial tumors in the extremities. CT may be used in the diagnosis of fat -containing tumors.
CT Morphology
Benign lipomas appear as homogeneous, wellcircumscribed masses that have negative attenuation values and displace adjacent structures. Infiltrating lipomas grow between muscle fibers but retain their homogeneous fat attenuation. They should remain confined to a single compartment and are sharply defined on all sections, even though they cause a very irregular pattern by separating muscle fibers. Liposarcoma should be suspected if the tumor contains components of soft-tissue density, is located deep in the soft tissues, and exceeds Scm in diameter. Infiltration into surrounding tissues, especially into adjacent compartment is characteristic of malignancy (see Fig. 16.26).
Postoperative Changes and Complications Metallic Implants
CT is increasingly preferred in situations where conventional x-ray tomography was formerly used. It is excellently suited for precisely describing the positions" of screws and other metallic implants, as well as for evaluating nonunion or other complications in patients after osteosynthetic procedures. The quality of postoperative CT examinations for evaluation of the placement and position of metallic implants is compromised by high-contrast streak artifacts that arise from plates and screws. The extent of the artifacts depends on the implant material (gold »steel »titanium) and the size and orientation of the metal structures. Artifacts are most pronounced on axial scans and are less obvious on MPR. Artifacts can be reduced by narrowing the slice
thickness, increasing the kVp and mAs, and using a wide window setting (width S:4000HU). Artifact suppression algorithms (e.g. by GE, Siemens, Toshiba), however, are the most effective tool for obtaining high-quality CT sections. •
CT Morphology
Periarticular fixation screws should be inserted
to the subchondral level, and any screw that projects above the bone contour into the cartilage is considered malpositioned. Transpedicular screws should be clear of the spinal canal. The tips of cortical bone screws should penetrate the cortex. For the evaluation of implant material damage, an extended CT scale and a very wide window setting are mandatory.
Nonunion and Arthrodesis 995 Streak artifacts due to beam-hardening and photon starvation can mimic fracture lines on MPRs (Fig. 25.67) or simulate gas collections resulting from an infection. Axial images identify the "gas collections" or "fracture lines" as artifacts by showing that they are in direct continuity with artifacts around metallic implants. Artifact -reducing software is offered by some manufacturers and provides a good tool for almost complete elimination of streak artifacts (Fig. 25.68). However, it requires a substantial increase in computing time for reconstruction of corrected, images, ,,,, ";
Fig. 25.5). CT can excellently describe the position and form of fragments in patients with
pseudarthrosis.
A plaster cast need not be removed prior to the CT scan. CT images reformatted on two planes perpendicular to the nonunion are necessary for the optimum evaluation of the fracture margins and gap. •
CT Morphology
The osseous gap following fracture treatment or arthrodesis should be bridged by bone and undergo complete consolidation within 6-9 months. Sclerosis and residual gaps are symptomatic of nonunion. The most frequent cause is deficient fracture stabilization leading to the development of a hypertrophic nonunion. An inadequate blood supply, or metabolic abnormalities, may lead to deficient bone formation and may cause an atrophic nonunion with a residual interfragmental gap and no sclerosis. Bacterial infection leads to an infected nonunion with signs /\f, osteomyelitis. Malunion is assessed by conventional radiographs. CT has a role in determining rotational deformities (see
During normal fracture healing or after arthrodesis' periosteal and endosteal calluses form, which are fibrous initially and thus of soft-tissue density. As the healing process progresses, the callus becomes more sclerotic. Finally there is radiographic union with osseous bridging, and various degrees of thickening of the cortex, and medullary sclerosis at the fracture site (see Fig. 25.43). CT findings follow closely those observed on conventional radiographs. Hypertrophic nonunion is characterized by an exuberant bone reaction, sclerosis, and flaring of bone ends. CT is able to demonstrate partial bridging within the hypertrophic reaction. Atrophic nonunion is characterized by round and smooth edges of the fractured bone with a residual gap. There may be sclerosis of the bone ends. CT rarely yields additional information over conventional radiographs. The presentation of infected nonunion depends on the activity of infection. Inactive osteomyelitis is characterized by a well-organized periosteal reaction, irregular thickening of
a
b
Nonunion and Arthrodesis
Fig. 25.67 Acetabularfracture after osteosynthetic repair. The streak artifact between two parallel screws (a) mimics a fracture line on the coronal MPR (b).
25 Musculoskeletal System Table 25.21
a
b
c
d
Fig. 25.68 Despite massive metal implants, artifacts on sagittal reformations are minor in this patient with an internal fixation of L2 (a), even in the immediate vicinity of the metal rods (b). Artifact reduction software can almost completely eliminate streak artifacts from metal objects (c. d) but requires substantially increased time for image reconstruction.
the cortex, and reactive sclerosis of cancellous bone. Active osteomyelitis may show soft-tissue swelling, destruction of cortex, and cancellous bone associated with new bone formation and sequestration. Osteonecrosis
Osteonecrosis, e.g., avascular necrosis of the femoral head, is diagnosed earlier (stage I) and more accurately with MRI than with CT (Table 25.21). CT should be used only in patients with contraindications or troublesome metallic implants.
•
Grading of Osteonecrosis
o
Ischemia, no necrosis
I
Necrosis, edema
II
Necrosis
ill
Fracture
IV
Impression fracture, osteoarthritis
CT Morphology
In Osteonecrosis, CT demonstrates heterogeneous sclerosis of the affected bone that reflects trabecular (micro) fractures with callus formation, and subchondral impression fractures within thejoint area (stage III) at an earlier stage than projection radiographs. Trabecular microfractures and subchondral impression fractures indicate mechanical insufficiency. MRI and CT are both suitable methods to demonstrate the location and extent of necrosis for treatment planning. Late sequelae of Osteonecrosis are cyst formation and articular surface fractures that lead to deformity of the affected bone In femoral head necrosis, axial scans with 1-2 mm slice thickness through the femoral head will reveal typical heterogeneities and areas of increased attenuation in the asterisk trabecular pattern indicating callus formation following microfractures. A subchrondral crescent sign is indicative of an impending impression fracture due to Osteonecrosis. Impression fractures lead to a contour depression, usually affecting a long segment of the subchondral portion of the femoral head. Sagittal and coronal MPRs demonstrate the abnormalities in multiple planes and are often superior for detection of subchondral involvement. The affected area of the femoral head surface can be estimated from the angle of the affected portions on images in the sagittal and coronal plane through the central line of the femoral head: - Necrotic area = sagittal angle/180x coronal angle/" 180xlOO%. Other sites where Osteonecrosis may occur include the humeral head, the wrist (scaphoid, lunatum), the proximal tibia, the medial femur condyle, the talus, and metatarsal bones (Fig. 25.69).
Osteonecrosis
a
b
Fig, 25.69 Osteonecrosis of the humeral head (4 x 2/ 11) with a hyperlucent crescent sign (a) and sclerosis associated
with a subchondral
fracture (b), On an exar-
c ticulated volume-rendered view of the humeral head, the extent of the necrosis can be directly visualized (c),
997
Index Page references in bold type refer to illustrations
D
aviator's astragalus fracture-dislocation,
D
spontaneous,
895
1060
Index
D
anatomy,
544,545
anatomy,
935
hematoma,
anatomy.
943
1066
Index
B.;
1068
Index
a Kaposi'ssarcoma,203,327,327,562-563,
anatomy.
743
E3
n
D
1080
Index
Ft
o
1084
Index
table feed (IF), 13.14,117
thrombus
ulcer
Index
E3
wedge ratto. thoracic spine, 952
1089
1090
Index
l!!l
Versinia enteritis, 574,579
119
369
Q
