Computer-Assisted Neurosurgery

Computer-Assisted Neurosurgery Computer-Assisted Neurosurgery Edited by Gene H. Barnett Cleveland Clinic Foundation...

Author: Gene H. Barnett | Robert Maciunas | David Roberts

54 downloads 1051 Views 18MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Computer-Assisted Neurosurgery

Computer-Assisted Neurosurgery

Edited by

Gene H. Barnett Cleveland Clinic Foundation Cleveland, Ohio, U.S.A.

Robert J. Maciunas Case Western Reserve University Cleveland, Ohio, U.S.A.

David W. Roberts Dartmouth Medical School Hanover, New Hampshire, U.S.A.

Published in 2006 by Taylor & Francis Group 270 Madison Avenue New York, NY 10016 © 2006 by Taylor & Francis Group, LLC No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International Standard Book Number-10: 0-8247-2837-8 (Hardcover) International Standard Book Number-13: 978-0-8247-2837-3 (Hardcover) Library of Congress Card Number 2005052998 This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Computer assisted neurosurgery / edited by Gene Barnett, Robert Maciunas, David Roberts. p. ; cm. Includes bibliographical references and index. ISBN-13: 978-0-8247-2837-3 (alk. paper) ISBN-10: 0-8247-2837-8 (alk. paper) 1. Computer-assisted neurosurgery. I. Barnett, Gene H. II. Maciunas, Robert J. III. Roberts, David W., 1950[DNLM: 1. Neuronavigation--methods. 2. Central Nervous System Diseases--surgery. 3. Neuronavigation--instrumentation. 4. Surgery, Computer-Assisted--methods. WL 368 C7379 2005] RD593.5.C66 2005 617.4'8059--dc22

2005052998

Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com Taylor & Francis Group is the Academic Division of Informa plc.

0

PREFACE

The practice of neurosurgery has been fundamentally changed by the evolution of the electronic computer. Digital imaging techniques such as computer-assisted tomography, magnetic resonance imaging, and digital angiography revolutionized diagnostic neurosurgery, supplanting pneumoencephalography, myelography, and many other invasive procedures. Although powerful diagnostic tools, these technologies have not replaced the diagnostician; rather they complement and confirm thoughtful deductions based on a careful neurologic history and examination. In the late 1970s, several neurosurgical visionaries recognized that these computerized imaging systems were capable of more than just aiding diagnosis. They devised equipment and techniques to harness the spatial information contained within these images for surgical guidance. Image-guided frame stereotaxy used a stereotactic reference frame and imaging modality-specific encoders to allow safe, accurate biopsy of imaged brain lesions. This technique was adapted to numerous neurosurgical procedures including depth electrodes, functional neurosurgery, radiation implants, and craniotomy. Virtually any intracranial target could be accessed along predefined approaches or trajectories. Although the calculations to decode spatial information from neuroimaging could be done manually or with calculators, the availability of reasonably priced computers and workstations simplified the process for the surgeon. As the ratio of computing power to cost increased, not only could points be defined stereotactically but volumes as well. Stereotactic radiosurgery and volumetric craniotomy benefited from these technological advances, but stereotactic craniotomy failed to be widely adopted in the neurosurgical community, perhaps as the benefit of guidance was not felt to outweigh the cumbersome logistics of the procedure. In the early 1980s, three technological advances occurred that allowed for the development of stereotactic systems that were not reliant on applied reference frames: (1) neuroimaging had progressed to the point where it was not only spatially accurate within a given image (i.e., slice) but also accurate throughout the entire volume of data; (2) the availability of accurate, inexpensive three-dimensional digitizers; and (3) another incremental increase in low-cost computing power such that three-dimensional transformations and other calculations could be performed and displayed using image data sets several megabytes in size and done so in near real time. Image-guided frameless stereotaxy could not only provide the guidance information of its frame counterpart, but provided truly iii

iv

&

PREFACE

interactive navigational information such as location and orientation during surgical procedures. These devices could provide instantaneous intraoperative answers to the questions, ‘‘Where am I in this head?’’ and ‘‘What’s on the other side of that structure?’’ Since then, these systems have become commonly known as surgical navigation systems or neuronavigation systems. Although surgical navigational systems have become mainstream devices, equally at home in the community as well as the academic medical center, far too many of these devices rest idle or are misused because the neurosurgeon has not been adequately exposed to the principles and nuances of computer-assisted neurosurgery beyond the information provided by a manufacturer’s device or sales representative. The thrust of this book is not to dwell on the specifics of any given surgical navigational system (although a brief primer on surgical navigational system technologies is provided for the newcomer), but rather it is to provide a resource on these computerized surgical tools with emphasis on how they work (as well as their limitations) and how best to use them in a variety of clinical settings. This follow-up to our previous book Image-Guided Neurosurgery: Clinical Applications of Surgical Navigation is divided into three sections: Basic Principles, Technologic Applications, and Clinical Applications. In Part I: Basic Principles, we review how these devices work starting with the process by which the computer relates image space to the three-dimensional space of the operating room—the so-called ‘‘registration’’ process. We then examine the technologies that allow these systems to work, with an emphasis on how the location of surgical tools is relayed to the computer via various three-dimensional digitizers. Optimal use of navigation for many procedures requires preplanning a target and the course or trajectory to get there; various means of presenting this information to the surgeon in a useful fashion are presented. Intrinsic to knowing how to use a tool effectively is to understand its limitations and how it can malfunction; these are reviewed Chapter 4: Pitfalls. Surgical navigations systems often work in concert with other technologies, and the same principles used for their use in the operating room apply to certain noninvasive radiosurgical tools. As intracranial surgery becomes progressively more complex and focused on maximal safe resection, surgical navigational systems are increasingly employing intraoperative imaging such as intraoperative magnetic resonance imaging, often in conjunction with neurophysiologic monitoring. The marriage of endoscopy with navigation has broadened the utility of these devices, not just for intraventricular procedures but also for some skull base and trans-sphenoidal operations and, perhaps, some day intraparenchymal tumors. This is explored in Part II: Technologic Applications. In the end, however, the reasons for using surgical navigational systems are to augment or guide certain surgical procedures. General overviews of how surgical navigational systems can be used for brain biopsy, related procedures, and minimal access craniotomies are reviewed in Part III: Clinical Applications (including a narrated video on the CD-ROM accompanying this book). We then delve into the specifics of maximizing the value of surgical navigation in several clinical disorders such as gliomas, meningiomas, pituitary tumors, vascular malformations, spinal, seizure, and skull base neurosurgery.

PREFACE

&

v

One final note: Just as computerized imaging has not replaced the diagnostician, these tools will not transform a bad surgeon into a great one, let alone replace the surgeon entirely. They remain an adjunct to the surgeon’s experience, judgment, and technical skills—one more instrument at the surgeon’s disposal. We believe that this technology is the next computer-generated revolution in neurosurgery, and we hope that you find the information and experiences of the authors of this book a useful guide on its utilization in common neurosurgical procedures and disorders. Gene H. Barnett Robert J. Maciunas David W. Roberts

0

CONTENTS

Preface . . . . iii Contributors . . . . xiii PART I: BASIC PRINCIPLES 1. Fundamentals of Registration . . . . . . . . . . . . . . . . . . . . . . . . . . 3 David W. Roberts Coordinate Systems . . . . 4 Registration Concepts . . . . 7 Transformation Matrices . . . . 8 Fiducial Point Strategies . . . . 11 Surface Matching Strategies . . . . 12 Conclusions . . . . 14 References . . . . 15 2. Surgical Navigation System Technologies . . . . . . . . . . . . . . . . . . 19 Narendra Nathoo and Gene H. Barnett Introduction . . . . 19 Current Surgical Navigation Systems . . . . 20 Surgical Planning and Simulation . . . . 20 Intraoperative Digitization . . . . 21 Registration of Images with the Patient . . . . 30 Intraoperative Navigation . . . . 30 Spine Navigational (Fluoroscopy) Technologies . . . . 32 Computer-Directed Neurosurgery/Robotic Surgery . . . . 33 Conclusion . . . . 33 References . . . . 34 3. Target and Trajectory Guidance . . . . . . . . . . . . . . . . . . . . . . . . 37 Charles P. Steiner System Description and Requirements . . . . 37 Localization and Orientation . . . . 38 Target Guidance . . . . 39 Trajectory Guidance . . . . 41 vii

viii

&

CONTENTS

Discussion . . . . 44 References . . . . 48 4. Pitfalls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Robert J. Maciunas Definition of Accuracy . . . . 50 Image Acquisition . . . . 52 Registration Techniques . . . . 55 Computers and Software Interfaces . . . . 57 Interactive Localization Devices and Intraoperative Use . . . . 58 Integration of Real-Time Data . . . . 62 Tissue Displacement . . . . 63 Robotics . . . . 64 Judgment and Clinical Experience . . . . 64 References . . . . 66 PART II: TECHNOLOGIC APPLICATIONS 5. Brain Biopsy and Related Procedures . . . . . . . . . . . . . . . . . . . . . 71 Vitaly Siomin and Gene H. Barnett Introduction . . . . 71 Suitability for Brain Biopsy . . . . 72 Image-Guided Frame-Based Stereotactic Brain Biopsy . . . . 72 Image-Guided Frameless Stereotactic Brain Biopsy . . . . 73 SBB-Related Procedures . . . . 81 Comparison of SBB Techniques . . . . 82 Complications . . . . 83 Postoperative Management . . . . 85 Conclusions . . . . 85 References . . . . 86 6. Intraoperative MRI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Andrew A. Kanner and Michael A. Vogelbaum Development of Intraoperative MR (iMRI) Imaging in Neurosurgery . . . . 91 The Concept of iMRI . . . . 92 Development of iMRI Systems . . . . 93 Instruments . . . . 99 Navigation . . . . 100 Interpretation of Intraoperatively Acquired Images . . . . 100 Artifacts . . . . 101 The Impact of Extent of Resection of Brain Tumors on Outcome . . . . 102 Conclusion . . . . 103 References . . . . 104

CONTENTS

&

ix

7. Awake Craniotomy and Intraoperative Imaging . . . . . . . . . . . . 109 Lilyana Angelov and Gene H. Barnett Introduction . . . . 109 Anesthetic Techniques . . . . 111 Intraoperative Functional Mapping . . . . 113 Static Image-Guided Resection . . . . 116 Dynamic Intraoperative Image-Guided Resection . . . . 117 Intraoperative Use of Magnetic Resonance Imaging . . . . 118 The Future . . . . 122 References . . . . 123

8. Computer-Assisted Neuroendoscopy: The Navigated Neuroendoscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Mark G. Luciano, Mohamed Ammar, and Stephen M. Dombrowski The Development of Modern Neuroendoscopy . . . . 127 Current Neuroendoscopic Technology . . . . 128 Problems in Neuroendoscopy . . . . 129 Computer-Assisted Visualization . . . . 133 Computer-Assisted Tools for Manipulation . . . . 133 Conclusion . . . . 141 References . . . . 142

PART III: CLINICAL APPLICATIONS 9. Minimal Access Craniotomy Using Surgical Navigation Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Gene H. Barnett Introduction . . . . 147 Surgical Technique . . . . 147 Summary . . . . 158 References . . . . 159

10. Image-Guided Treatment of Metastatic Brain Tumors . . . . . . . . 161 Vitaly Siomin and Michael A. Vogelbaum Introduction . . . . 161 Clinical Features . . . . 162 Imaging . . . . 162 Treatment . . . . 163 Image-Guided Surgery . . . . 164 Stereotactic Radiosurgery . . . . 168 Surgery or Radiosurgery? . . . . 171 Summary . . . . 173 References . . . . 174

x

&

CONTENTS

11. Surgical Navigation Systems for the Resection of Intracranial Gliomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 Michael W. McDermott, Devin K. Binder, Sandeep Kunwar, Andrew T. Parsa, and Mitchel S. Berger Preoperative Preparation . . . . 180 Considerations for the Use of Surgical Navigation Systems Based on Anatomic Site . . . . 182 University of California at San Francisco Experience . . . . 185 A Prospective Study on SNS and EOR . . . . 186 Technical Considerations . . . . 190 References . . . . 192 12. Intracranial Meningiomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 Joung H. Lee, Ajit A. Krishnaney, Michael P. Steinmetz, and Dae Kyu Lee Introduction . . . . 195 SNS Application in Meningioma Surgery . . . . 196 Summary . . . . 206 References . . . . 207 13. Transphenoidal Hypohysectomy . . . . . . . . . . . . . . . . . . . . . . . 209 Edwin Cunningham and Marc R. Mayberg Introduction . . . . 209 Direct Microscopic Approach . . . . 210 Conclusion . . . . 219 References . . . . 220 14. Vascular Malformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 William E. Thorell and Peter A. Rasmussen Introduction . . . . 221 Vascular Malformations . . . . 222 Review of Imaging Modalities . . . . 226 Stereotactic Radiosurgery . . . . 230 Frameless Stereotactic Image Guidance . . . . 231 Conclusions . . . . 232 References . . . . 233 15. Image-Guided Spinal Navigation . . . . . . . . . . . . . . . . . . . . . . . 235 Iain H. Kalfas Principles of Image-Guided Spinal Navigation . . . . 236 Clinical Applications . . . . 241 Pitfalls of Image-Guided Spinal Navigation . . . . 252 Fluoroscopic Navigation . . . . 253 Conclusion . . . . 254 References . . . . 256

CONTENTS

&

xi

16. Applications in Epilepsy Surgery . . . . . . . . . . . . . . . . . . . . . . . 259 David W. Roberts and Terrance M. Darcey Preoperative Investigation . . . . 260 Therapeutic Procedures . . . . 263 Future Development . . . . 271 References . . . . 272

17. Skull Base Neurosurgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Troy D. Payner and Armond L. Levy Utilization of Surgical Navigation . . . . 278 Types of Lesions Targeted . . . . 281 Technical Considerations . . . . 282 Skull Base Approaches . . . . 285 Complications and Limitations . . . . 291 Intraoperative Imaging . . . . 292 Learning Curve . . . . 293 Summary and Future . . . . 294 References . . . . 295

Index . . . . 297

0

CONTRIBUTORS

Mohamed Ammar Section of Pediatric and Congenital Neurosurgery, Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Lilyana Angelov Department of Neurological Surgery and Brain Tumor Institute, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Gene H. Barnett Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Mitchel S. Berger Department of Neurological Surgery, University of California, San Francisco, California, U.S.A. Devin K. Binder Department of Neurological Surgery, University of California, San Francisco, California, U.S.A. Edwin Cunningham Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Terrance M. Darcey U.S.A.

Dartmouth Medical School, Hanover, New Hampshire,

Stephen M. Dombrowski Section of Pediatric and Congenital Neurosurgery, Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Iain H. Kalfas Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Andrew A. Kanner Brain Tumor Institute and Department of Neurological Surgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Ajit A. Krishnaney Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. xiii

xiv

&

CONTRIBUTORS

Sandeep Kunwar Department of Neurological Surgery, University of California, San Francisco, California, U.S.A. Dae Kyu Lee Brain Tumor Institute and Department of Neurological Surgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Joung H. Lee Cleveland Clinic Brain Tumor Institute and Department of Neurosurgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Armond L. Levy Indiana, U.S.A.

Indianapolis Neurosurgical Groups, Indianapolis,

Mark G. Luciano Section of Pediatric and Congenital Neurosurgery, Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Robert J. Maciunas Department of Neurological Surgery, Case Western Reserve University, University Hospitals of Cleveland, Cleveland, Ohio, U.S.A. Marc R. Mayberg Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Michael W. McDermott Department of Neurological Surgery, University of California, San Francisco, California, U.S.A. Narendra Nathoo Brain Tumor Institute and Department of Neurological Surgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Andrew T. Parsa Department of Neurological Surgery, University of California, San Francisco, California, U.S.A. Troy D. Payner Indiana, U.S.A.

Indianapolis Neurosurgical Groups, Indianapolis,

Peter A. Rasmussen Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. David W. Roberts Dartmouth Medical School, Hanover and Section of Neurosurgery, Dartmouth–Hitchcock Medical Center, Lebanon, New Hampshire, U.S.A. Vitaly Siomin Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

CONTRIBUTORS

&

xv

Charles P. Steiner Division of Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Michael P. Steinmetz Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. William E. Thorell Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A. Michael A. Vogelbaum Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

0

PART I

Basic Principles

1

FUNDAMENTALS OF REGISTRATION David W. Roberts Dartmouth Medical School, Hanover and Section of Neurosurgery, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, U.S.A.

Mathematical quantification of spatial information and subsequent manipulation of that information enabling one to relate one set of data to another represent fundamental principles of stereotaxy. Numerous benefits are derived from the establishment of such spatial registration, including the identification in an imaging study of anatomic features from an atlas, the integration of multiple imaging modalities, and the tracking or delivery of a surgical instrument to a given location seen in an atlas or on a radiographic study. This chapter presents an overview of stereotactic coregistration, including the basic mathematical principles of coordinate systems and their manipulation. Prior to the current, relatively inexpensive availability of computing resources in the operating room, stereotaxy relied upon frame-based instrumentation and registration strategies that could be readily implemented by the neurosurgeon in the clinical environment. These strategies necessarily constrained the registration problem, and usually achieved their ends through simple arithmetic or graphic, geometric steps. More recently, computer resources have altered this in at least as significant a way as the technology of computed tomography (CT) and magnetic resonance imaging (MRI) scanning has broadened stereotaxy’s indications and facilitated its application. Inexpensive computational power has provided more efficient, algebraic solutions to many of the registration tasks required by stereotactic frame systems. In addition, it has made possible the sophisticated representation of reformatted atlas, individualized neuroimaging, and operative field information for two- and three-dimensional visual display during surgery. Despite increasingly intuitive interfaces, which facilitate the use of such technology without knowledge of the underlying mathematics, there are reasons why continuing to understand the fundamentals of stereotaxy may be advantageous to neurosurgeons. Foremost among these is the safeguard such an understanding provides during clinical application; errors that might arise may be better appreciated or prevented altogether. Equally compelling is the recognition that such 3

4

&

PART I: BASIC PRINCIPLES

understanding on the part of the surgeon can provide the impetus and direction to further development of the field.

COORDINATE SYSTEMS Pierre de Fermat and Rene´ Descartes, two early 17th century Frenchmen, independently recognized that a system comprised of two perpendicular lines could be used to identify any point within a plane (1). The distances along each of these lines (what may be called the x- and y-axes) from the origin (the axes’ intersection, where each is usually assigned a value of 0) to a given point provide an ordered pair of numbers, or coordinates, unique to that point. Such a rectangular coordinate system is commonly called Cartesian, in honor of Descartes, and may be extended to three-dimensions, with x-, y-, and z-axes. In this manner, each location in a given space may be uniquely and quantitatively defined. There are other ways, of course, to describe a point’s location in space as well. For example, the lines representing the coordinate system’s axes need not be perpendicular to one another (although for each to provide independent information about each point they must be nonparallel). Alternatively, a point in a plane may be described by specifying its distance and direction from an origin, i.e., a radius (r) and an angle (theta), using a polar coordinate system. This may be extended into three dimensions using a cylindrical coordinate system by specification of a distance along an additional (z) axis perpendicular to that original plane, or one may use a spherical coordinate system, specifying a radius and two angles (perpendicular with respect to each other) describing the direction from the origin. All of these systems are mathematically equivalent and achieve the same end in defining uniquely the location of a point in space. The coordinates of one system may be readily converted to those of another. It is an elementary exercise to show, for example, that the coordinates of a point (r, theta) in two-dimensional polar coordinate space may be converted to x, y coordinates in two-dimensional Cartesian coordinate space by the formulas ‘‘x ¼ r cos theta’’ and ‘‘y ¼ r sin theta,’’ and that similar conversion from three-dimensional spherical coordinates is represented by the formulas ‘‘x ¼ r sin theta cos phi,’’ ‘‘y ¼ r sin theta sin phi,’’ and ‘‘z ¼ r cos theta’’ (2). Various two- and three-dimensional coordinate spaces may be employed in a stereotactic procedure. As noted earlier, spatial information from imaging studies, atlases, stereotactic frames, or other intraoperative digitizers may all be represented in such coordinate spaces. Considering imaging studies first, the coordinates inherent in a single CT slice are most intuitive. The computer graphic representation of such a slice is generated using coordinates inherent to the scanner, and these machine-specific or gantry coordinates are usually readily accessible at the scanner console. By convention, an x coordinate is assigned along the right–left axis of the image and a y coordinate along the posterior–anterior axis. Thus, on any given two-dimensional CT, MR, positron emission tomography, or single photon emission tomography image, one may specify a given pixel’s location in terms of an x and y coordinate; whether this is in gantry coordinate space or some other based

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

5

upon part of a stereotactic frame that may also be in the scan is immaterial at this point in the discussion and will be revisited later in the discussion about transformations. Neurosurgical procedures must account for the three-dimensional space of the intracranial volume, and imaging information must include this third dimension. The location of a slice with respect to a frame or to other slices in the study may be determined in several ways, the most common being to monitor the position of the scanner table relative to the gantry or to use the geometry of frame components visible in the scanner that enable unique determination of the slice plane relative to the frame (i.e., the various N-shaped fiducial plates). A coordinate incorporating this third dimension may be added to the coordinates already derived in two dimensions to provide a three-dimensional address for any location within the imaging study. Keeping in mind such traps as gantry tilt that will introduce skew into the stack of slices (which can be accounted for), the Cartesian coordinate system of CT and MR imaging is intuitive, tractable, and convenient. The projection geometry of conventional radiographs and biplanar angiography is not as simple—accounting in part for early stereotaxy’s esoteric mystique— but remains directly solvable as well. In this instance, spatial information is obtained from two-dimensional images, and in order to derive three-dimensional information one must use data from at least two images obtained from different perspectives, identify the same object(s) in each image, and account for magnification and parallax. All of these issues have been well worked out mathematically in general and by multiple stereotactic systems in particular (3). Geometric methods using plotting graphs or the spiral diagram provided non-algebraic working solutions prior to computer availability. Analytic solutions for conventional angiography, however, have now been developed for multiple stereotactic frame systems using fiducial-embedded plates attached to the frame (4,5). Scalp-based fiducial markers in a frameless adaptation have also been described (6). Stereotactic atlases derive their utility from similar organization of their anatomic information into coordinate space. Nearly all stereotactic atlases are Cartesian, and from a mathematical perspective differ primarily in the selection of the coordinate system’s origin, the orientation of its axes (i.e., what other natural reference points are used to align one or more axes), the methodology of scaling, the intervals of data sampling, and the occasional incorporation of statistical methodology. Independent of these issues, each atlas assigns a unique three-dimensional address (the stack of anatomic slices compiled into a three-dimensional volume in a manner entirely analogous to that of stacking CT slices) to each anatomic point of interest, enabling subsequent manipulation and incorporation of that information into a stereotactic procedure. Multiple functions are made possible by stereotactic frame systems: the definition of a coordinate space that includes the relevant portion of a particular patient’s intracranial volume, the means by which coregistration of that coordinate space with information from imaging studies or anatomic atlases may be accomplished, and the delivery of an operative instrument to a selected coordinate address (stabilization of that instrument by the frame is an advantageous and convenient, but non-stereotactic, additional benefit). The first of these tasks may be achieved using any type of coordinate system, and indeed the various stereotactic

6

&

PART I: BASIC PRINCIPLES

frame systems utilize a variety. Leksell, Talairach, and Hitchcock frames, to cite a few, are all Cartesian-based; target coordinates on these frames are specified as ordered triplets (x, y, z). Target-centered Arc systems (like the Leksell system) have the attractive and convenient feature of enabling free adjustment of the angles of the arc and probe carrier (and thus the trajectory) with the target point always remaining centered. For applications of point targeting, these additional parameters need not be stereotactically defined. If a specific trajectory is to be stereotactically defined (e.g., one passing through more than one designated point), these angles, of course, must be specified; there are more than three parameters to such a localization, and information in addition to Cartesian coordinates of the target is necessary. Non-Cartesian coordinate systems are employed by still other stereotactic frame systems. The Reichert–Mundinger frame uses two angles on an arc, an additional two angles on the probe carrier, and a dimension of depth (7). Determination of this combination of spherical coordinates was originally performed noncomputationally, by mechanically adjusting the trajectory probe until it reached the simulated target on a phantom. (The positioning of this phantom target would already have been set using radiographically determined rectangular coordinates, in turn derived through projection geometry from multiple two-dimensional image rectangular coordinates.) The inconvenience and potential contamination of this step with a phantom was eventually replaced by the mathematical calculation of the frame angles and length, exemplifying the integration of increasingly accessible computational power. The coordinate space of the Brown–Roberts–Wells stereotactic system, in which targets and trajectories are defined by four angles and a length, is similarly non-Cartesian (8,9). Its development at a time when computational resources were available, initially using the CT scanner’s computer and later a programmable calculator, allowed computational conversion from Cartesian scanner coordinates to spherical coordinates, and its phantom base is used for mechanical confirmation only. The operative coordinate space is defined by the coordinate system of a particular stereotactic frame. More recently developed stereotactic systems that do not use a mechanical frame, nevertheless, still require an operative coordinate space, and this requirement is fulfilled through the use of a three-dimensional digitizer within the operating room. Such a device can locate a point within its working volume, assigning it a coordinate address (usually within whatever kind of coordinate space is desired, the interconversion being straightforward as noted earlier). A variety of digitizing technologies may be employed to accomplish this localization task, including those based upon ultrasound emitters and microphone arrays, articulated arms with potentiometers or optical encoders at the joints between links, light-emitting diodes and linear CCD camera arrays, stereo video or electronic cameras, and electromagnetic field transmitters and detecting coils. The software running these various digitizers calculates from the one or more sensors of the device, the location of a selected point, and this information is usually passed directly to further software determining coregistration, driving appropriate graphic displays, or interfacing with robotic instrumentation. The actual coordinate system and numeric coordinates for a selected point in operating room space in these implementations remain transparent to the procedure. A good illustration

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

7

of how such a digitizer can define an operating room stereotactic space is found in Friets’ description of a sonic digitizer used to track an operating microscope (10).

REGISTRATION CONCEPTS The establishment of the spatial relationship between the multiple coordinate spaces defined by imaging studies and by the intraoperative stereotactic system is the sine qua non of a stereotactic procedure. With the goal of minimizing calculation, particularly when computational aids were less available, most framebased methodologies were developed using some degree of mechanical alignment of frame and imaging studies. The problem of determining the relationship between coordinate spaces is in this manner at least partially constrained and in turn made more tractable. Using an orthogonal central beam a given distance from a frame’s side is an example, in the instance of working with projection geometry; placing a frame in a scanner perpendicular to the gantry and without rotation is another. In the latter example, elimination of rotation along any of three coordinate axes reduces the transformation from Cartesian scanner space to Cartesian frame space to a calculation of only three translations (Fig. 1). Continuing the imposition of constraints to an extreme, if two coordinate spaces were concordant, with corresponding points in the imaging study and the intracranial working space having identical coordinates, coregistration would in effect already be accomplished and no further manipulation or computation would be required. This is the situation in the recent development of operating within an open magnet MRI. With increasingly fast and inexpensive computation, more and more of the task of determining the relationship between different coordinate spaces can be shifted from the mechanical to the software realm. An example of this is the registration process for the Brown–Roberts–Wells frame in which a full threedimensional transformation is derived during the registration step; positioning of the frame within a CT scanner is thus less constrained at the expense of more elaborate, but still reasonably efficient calculation. Conceptually, this shift toward computational derivation of necessary transformations has led to elimination of the stereotactic frame altogether. The correspondence between one Cartesian coordinate space and another at its most fundamental level consists of the definition of six parameters. These are three angles of rotation by which the x, y, and z coordinate axes can be made parallel to one another, respectively, and three distances (along each of these axes) by which the origins of the two coordinate systems can be superimposed. An additional parameter of scale must, of course, also be explicitly or implicitly accounted for as well. It is not difficult to intuitively appreciate these rotations, translations, and scaling (Fig. 2). In registering a patient’s imaging study with the patient’s head in the operating room, the commonest method of registration is that of determining a rigid-body transformation, with the assumption being made that the morphology of the brain is constant (i.e., without deformation) in both coordinate systems. In instances where this assumption cannot be made (e.g., registering a

8

&

PART I: BASIC PRINCIPLES

Figure 1 The calculation of stereotactic frame coordinates at the CT or MRI console, in this instance of a stereotactic biopsy using the Leksell stereotactic frame, is facilitated by constraining the problem to one of straightforward translations. The left–right (x) and posterior–anterior ( y) distances between the center of the frame (as determined by the intersection of diagonal lines passing through symmetrical landmarks on the frame’s imaged superstructure) and the target are measured using the scanner console’s measure–distance function. The superior–inferior (z) coordinate is calculated from the distance between a vertical fiducial rod (posteriorly) and a diagonal fiducial rod whose geometrical relationship to the frame’s coordinate system is known.

stereotactic atlas to an individual patient), an elastic transformation may be necessary, but such algorithms are not yet as well developed.

TRANSFORMATION MATRICES This process of moving between different coordinate spaces is often represented mathematically by transformation matrices (3,11). Such a matrix for rotation about the x-axis has the form: 2

1 60 Rx ðfÞ ¼ 6 40 0

0 cos f sin f 0

0 sin f cos f 0

3 0 07 7 05 1

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

9

Figure 2 A rigid-body transformation from one coordinate space to another requires the designation of three angles of rotation and three translations.

and multiplication of the coordinates of a point by this matrix generates the new coordinates for that point: P 0 ¼ P Rx ðfÞ

¼ ½x

y

z

2

1

60 6 1 6 40 0

¼ ½x

0

0

0

3

cos f sin f 0 7 7 7 sin f cos f 0 5 0 0 1

y cos f z sin f y sin f þ z cos f 1

The new coordinates may be considered conceptually as either movement of the point to a new location within a single coordinate space or representation of an analogous point in a new coordinate system; the two are functionally equivalent. A matrix may be derived for each of the other rotations, translations, and scaling as follows (11): Rotation about the y-axis: 2 3 cos f 0 sin f 0 6 0 1 0 07 7 Ry ðfÞ ¼ 6 4 sin f 0 cos f 0 5 0 0 0 1

10

&

PART I: BASIC PRINCIPLES

Rotation about the z-axis: 2 cos f sin f 6 sin f cos f Rz ðfÞ ¼ 6 4 0 0 0 0 Translation along all three axes: 2 1 0 6 0 1 T ðDx ; Dy ; Dz Þ ¼ 6 4 0 0 Dx Dy

3 0 07 7 05 1

0 0 1 0

0 0 1 Dz

3 0 07 7 05 1

Scaling: 2

Sx 60 S ðSx ; Sy ; Sz Þ ¼ 6 40 0

0 Sy 0 0

0 0 Sz 0

3 0 07 7 05 1

Such multiple processes will be required in converting from one coordinate space to another, and the various matrix transformations may be performed sequentially. It will be observed that translation is an additive process and that rotation and scaling are multiplicative. By the use of what are called homogeneous coordinates (as given above), it is possible to treat all three operations as multiplications, and in this way one can combine sequential transformations in a process referred to as concatenation or composition. A rigid-body transformation combining all of the above operations is thus represented: T ¼ Rx Rx Rz Txyz Sxyz Registration across multiple coordinate spaces may be similarly performed: TAC ¼ TAB TBC Given a point whose coordinates are known in coordinate space A, coordinates for that point in coordinate space C may be derived by matrix multiplication: PC ¼ PA TAC Though such formulation has the appeal of formulaic simplicity, in the actual computation of transformational processes, the mathematical operations are actually sometimes broken back down into addition and multiplication for the sake of greater efficiency in terms of the number of operations required. It should also be pointed out in this regard that there are alternative methodologies to matrices for such operations, such as the use of direction cosines employed in early software for the Dartmouth microscope (10), but these are mathematically equivalent. The ability to move in the opposite direction between coordinate spaces—a desirable process in any truly interactive stereotaxy—is facilitated by the inverse

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

11

transformation, represented symbolically by T1: P2 ¼ P1 T P1 ¼ P2 T 1 Derivation of the transformation matrix required to move between patientspecific imaging coordinate spaces and stereotactic operating space can be achieved in a number of ways. The known geometry of specific stereotactic frame superstructures visible in imaging space allows convenient determination of the necessary transformation. As previously alluded to, the simple hand calculation of Leksell x and y coordinates from a CT or MR console represents such a calculation, albeit a fairly constrained one; similarly, identification in image space of the so-called fiducial rods (or, in the two-dimensional image, points) of a Brown– Roberts–Wells frame enables formulaic calculation of selected targets in stereotactic operating space. FIDUCIAL POINT STRATEGIES The simplest and most commonly used method at the present time, of achieving coregistration of coordinate spaces with frameless stereotactic systems, is that of matching a set of ordered points visible to both the imaging study and the intraoperative digitizer (10,12–20). A minimum of three noncolinear such pairs of points is required for determination of the transformation, although many systems employ or allow additional pairs of points to improve registration accuracy. Such points may be markers that have been attached to the scalp prior to imaging (e.g., glass beads, staples, Vitamin E capsules, skull-embedded screws, and now most commonly radiopaque, adhesive, foam ‘‘lifesavers’’). Alternatively, such algorithms may be used with natural anatomic features, such as the tragus, lateral canthi, or nasion. The less discrete nature of an anatomic feature may compromise accuracy, although such an approach has the advantage of not requiring a special imaging study. In either instance, the algorithms used are the same. Given a set of ordered points whose coordinates are known in two different coordinate spaces, there are a number of approaches to the problem of determining the parameters of the rigid body transformation relating those spaces. These include closed-form solutions, such as singular value decomposition of a matrix, eigenvalue–eigenvector decomposition of a matrix, or unit quaternions, as well as iterative solution techniques (21). Given a set of reference features as described above, an iterative registration process requires two conceptual steps. The first is that of defining a disparity function, by which a measure of the correspondence between two sets of corresponding features can be determined. The most common function is a least-squares criterion of the type: D ¼ iwi fdistance½FAi ; TðFBi ; tg 2 In this function, D is equal to the sum of squares of the distances between each of i corresponding points (w is a weighting factor related to the noise of the measurements) (22). The second step is that of optimization of the disparity function, by

12

&

PART I: BASIC PRINCIPLES

which is meant the adjustment of various transformation parameters such that a usable value of the function is achieved. In the above example, minimization of the function is desired, and when this is achieved, the registration process is complete. There are numerous optimization strategies, and when there is not a direct solution, nonlinear iterative methods are required. A frequently used method is that of gradient descent (22).

SURFACE MATCHING STRATEGIES In the registration strategy described using three or more reference points, the ordering of those corresponding points in both coordinate spaces is known. There are alternative strategies, however, in which unordered sets of points may be used, as in matching the surface of the head as imaged by CT, MRI, or positron emission tomography with the surface of the scalp as digitized in the operating room. This surface-to-surface matching has the advantage of using a natural feature of the head (so that additional, prospective imaging is not required) whose segmentation can be automated. A number of such techniques have been developed, the best known of which is the ‘‘hat-and-head’’ matching algorithm of Pelizarri and Chen (Fig. 3) (23–26). In their method, the distance between intraoperatively digitized surface points and the preoperatively imaged surface along rays extending from the centroid of that surface to each point is computed, and the sum of these squared distances then minimized. One may also derive a transformation using other features, such as the intrinsic curvatures of the scalp surface, to achieve alignment. By calculating from both imaging studies and from an intraoperative digitizer the mathematically defined curvatures over a sufficient portion of the scalp, it is then possible to align either

Figure 3 Two surfaces, represented by discrete points and contours lines may be matched, much like a hat and a head, using algorithms such as that of Pelizarri et al. Source: From Ref. 26.

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

13

Figure 4 Feature maps of the scalp’s surface, in this instance showing the inherent maximum and minimum curvatures at multiple, random points, have unique topography, which can be used for matching. These maps can be derived from information obtained from an intraoperative sonic digitizer or from CT and MR imaging. The resulting feature maps, whose unique topography can be readily appreciated, can then be matched. Source: From Ref. 27.

points of maximum curvature or related features (such as ridge lines, saddle points, etc.) in a similar manner (27–29) (Fig. 4). It will be appreciated that an iterative optimization process may inopportunely appear to have become optimized when it actually is not (the problem of local minima), and either initial guidance or later user intervention may be required to reduce this possibility. Numerous other surface-to-surface, surface-to-volume,

Figure 5 Cuchet and colleagues show in this graph the shapes of various cost functions using simulated data; only variation in rotation about the z-axis is shown. Source: From Ref. 42.

14

&

PART I: BASIC PRINCIPLES

and volume-to-volume matching methodologies have been reported (30–40). Each has respective advantages and disadvantages in terms of ease of automation and computational efficiency, and though they vary in the details of their algorithms, all conceptually share the component steps of definition of a disparity function and optimization of that function (22). The accuracy, efficiency, and speed of various strategies employing different functions and optimizations are understandably areas of considerable, current attention (41,42) (Fig. 5). The further development of these more sophisticated, either rigid or elastic, registration strategies based upon inherent anatomic features (i.e., not requiring placement of artificial fiducial markers on special scans) and fully automated segmentation and matching algorithms will make coregistration of multimodality and sequential imaging studies of a given patient a standard practice (43–45). The extension of these methods into the operating room using threedimensional digitizers will render nearly all procedures stereotactic in a manner transparent to the operating surgeon. CONCLUSIONS Coregistration, relating two or more coordinate spaces, is the process by which stereotactic operating systems achieve accurate integration of one or more databases into the operative field. Established mathematical methods for accomplishing this linkage have long been available, but newer algorithms adapted to the needs of the clinician are being developed and better characterized. The availability of powerful computational resources within the operating room allows the use of more sophisticated algorithms that have eliminated the need for a stereotactic frame, that have become increasingly more efficient and more accurate, and that have the potential to automate coregistration. Although the mathematical and computer skills enabling these latter developments have become more sophisticated, the fundamental mathematical principles underlying stereotaxy remain accessible to all. Reduction of potential error in clinical practice and facilitation of further development of the field will reward an understanding of these principles.

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

15

REFERENCES 1. West BH, Griesbach EN, Taylor JD, Taylor LT. The Prentice-Hall Encyclopedia of Mathematics. Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1982:119–126. 2. Clapham C. A Concise Dictionary of Mathematics. Oxford: Oxford University Press, 1990. 3. Lemieux L, Henri CJ, Wootton R, Collins DL, Peters TM. The mathematics of stereotactic localization. In: Thomas DGT, ed. Stereotactic and Image Directed Surgery of Brain Tumors. Chap. 12. Edinburgh: Churchill Livingstone, 1993:193–216. 4. Kelly PJ, Goerss SJ, Kall BA. Modification of Todd–Wells system for imaging data acquisition. In: Lunsford LD, ed. Modern Stereotactic Neurosurgery. Chap. 6. Boston: Martinus Nijhoff, 1988:79–97. 5. Lunsford LD, Kondiziolka D, Flickinge˙r JC, Bissonette DJ, Jungreis CA, Maitz AH, Horton JA, Coffey RJ. Stereotactic radiosurgery for arteriovenous malformations of the brain. J Neurosurg 1991; 75:512–524. 6. Grzeszczuk R, Tan KK, Levin DN, Pelizzari CA, Hu X, Chen GTY, Beck RN, Chen C-T, Cooper M, Milton J, Spire J-P, Towle VL, Dohrmann GJ, Erickson RK. Retrospective fusion of radiographic and MR data for localization of subdural electrodes. J Comput Assist Tomogr 1992; 16:764–773. 7. Mundinger F, Birg W. The imaging-compatible Riechert–Mundinger system. In: Lunsford LD, ed. Modern Stereotactic Neurosurgery. Chap. 2. Boston: Martinus Nijhoff, 1988:13–25. 8. Brown RA. A computerized tomography–computer graphics approach to stereotactic localization. J Neurosurg 1979; 50:715–720. 9. Apuzzo MLJ, Fredericks CA. The Brown–Roberts–Wells system. In: Lunsford LD, ed. Modern Stereotactic Neurosurgery. Chap. 5. Boston: Martinus Nijhoff, 1988: 63–77. 10. Friets EM, Strohbehn JW, Hatch JF, Roberts DW. A frameless stereotaxic operating microscope for neurosurgery. IEEE Trans Biomed Eng 1989; 36:608–617. 11. Foley JD, Van Dam A. Fundamentals of Interactive Computer Graphics. Reading, Massachusetts: Addison-Wesley Publishing Co, 1984:245–266. 12. Roberts DW, Strohbehn JW, Hatch JF, Murray W, Kettenberger H. A frameless stereotaxic integration of computerized tomographic imaging and the operating microscope. J Neurosurg 1986; 65:545–549. 13. Galloway RL Jr, Maciunas RJ, Edwards CA. Interactive image-guided neurosurgery. IEEE Trans Biomed Eng 1992; 39:1226–1231. 14. Maciunas RJ, Fitzpatrick JM, Galloway RL, Allen GS. Beyond stereotaxy: extreme levels of application accuracy are provided by implantable fiducial markers for interactive image-guided neurosurgery. In: Maciunas RJ, ed. Interactive Image-Guided Surgery. Chap. 21. AANS, 1993:259–270. 15. Day R, Heilbrun MP, Koehler S, McDonald P, Peters W, Siemionow V. Three-point transformation for integration of multiple coordinate systems: applications to tumor, functional, and fractionated radiosurgery stereotactic planning. Stereotact Funct Neurosurg 1994; 63:69–76.

16

&

PART I: BASIC PRINCIPLES

16. Heilbrun MP, Koehler S, MacDonald P, Siemionow V, Peters W. Preliminary experience using an optimized three-point transformation algorithm for spatial registration of coordinate systems: a method of noninvasive localization using frame-based stereotactic guidance systems. J Neurosurg 1994; 81:676–682. 17. Villalobos H, Germano IM. Clinical evaluation of multimodality registration in frameless stereotaxy. Comput Aided Surg 1999; 4:45–49. 18. Kozak J, Nesper M, Fischer M, et al. Semiautomated registration using new markers for assessing the accuracy of a navigation system. Comput Aided Surg 2002; 7:11–24. 19. Sadowsky O, Yaniv Z, Joskowicz L. Comparative in vitro study of contact- and image-based rigid registration for computer-aided surgery. Comput Aided Surg 2002; 7:223–236. 20. Krishnan R, Hermann E, Wolff R, et al. Automated fiducial marker detection for patient registration in image-guided neurosurgery. Comput Aided Surg 2003; 8:17–23. 21. Maurer CR, Fitzpatrick JM. A review of medical image registration. In: Maciunas RJ, ed. Interactive Image-Guided Surgery. Chap. 3. AANS, 17–44. 22. Lavalle´e S. Registration for computer-integrated surgery: methodology, state of the art. In: Taylor RH, Lavalle´e S, Burdea GC, Mo¨sges R, eds. Computer-Integrated Surgery. Chap. 5. Cambridge, Massachusetts: The MIT Press, 1996:77–97. 23. Levin DN, Pelizzari CA, Chen GTY, Chen CT, Cooper MD. Retrospective geometric correlation of MR, CT, and PET images. Radiology 1988; 169:817–823. 24. Pelizzari CA, Chen GTY, Spelbring DR, Weichselbaum RR, Chen C. Accurate threedimensional registration of CT, PET, and/or MR images of the brain. J Comput Assist Tomogr 1989; 13:20–26. 25. Tan KK, Grzeszczuk R, Levin DN, Pelizzari CA, Chen GTY, Erickson RK, Johnson D, Dohrmann GJ. A frameless stereotactic approach to neurosurgical planning based on retrospective patient-image registration. Technical note. J Neurosurg 1993; 79:296–303. 26. Pelizzari CA, Levin DN, Chen GTY, Chen C-T. Image registration based on anatomic surface matching. In: Maciunas RJ, ed. Interactive Image-Guided Surgery. Chap. 4. AANS, 1993:47–62. 27. Friets EM, Strohbehn JW, Roberts DW. Curvature-based nonfiducial registration for the frameless stereotactic operating microscope. IEEE Trans Biomed Eng 1995; 42:867–878. 28. Gueziec A, Ayache N. Smoothing and matching of 3-D space curves. Visualization in Biomedical Computing 1992. Proc SPIE 1992; 1808:259–273. 29. Balter JM, Pelizzari CA, Chen GTY. Correlation of projection radiographs in radiation therapy using open curve segments and points. Med Phys 1992; 19:329–334. 30. Jiang H, Robb RA, Holton KS. A new approach to 3-D registration of multimodality medical images by surface matching. Proceedings of the Second Conference on Visualization in Biomedical Computing, Chapel Hill, North Carolina, 1992:196–213. 31. Mangin JF, Frouin V, Bloch I, Lopez-Krahe J, Bendriem B. Fast Nonsupervised 3D Registration of PET and MR Images of the Brain. Paris: Telecom Paris, 1993.

CHAPTER 1: FUNDAMENTALS OF REGISTRATION

&

17

32. Hemler PF, Sumanaweera TS, van den Elsen P, Napel S, Adler J. A versatile system for multimodality image fusion. J Image Guided Surg 1995; 1:35–45. 33. Bachler R, Bunke H, Nolte LP. Restricted surface matching—numerical optimization and technical evaluation. Comput Aided Surg 2001; 6:143–152. 34. Brendel B, Winter S, Rick A, et al. Registration of 3D CT and ultrasound datasets of the spine using bone structures. Comput Aided Surg 2002; 7:146–155. 35. Glozman D, Shoham M, Fischer A. A surface-matching technique for robot-assisted registration. Comput Aided Surg 2001; 6:259–269. 36. Moore CJ, Graham PA. 3D dynamic body surface sensing and CT-body matching: a tool for patient set-up and monitoring in radiotherapy. Comput Aided Surg 2000; 5:234–245. 37. Muratore DM, Russ JH, Dawant BM, et al. Three-dimensional image registration of phantom vertebrae for image-guided surgery: a preliminary study. Comput Aided Surg 2002; 7:342–352. 38. Penney GP, Little JA, Weese J, et al. Deforming a preoperative volume to represent the intraoperative scene. Comput Aided Surg 2002; 7:63–73. 39. Fei B, Duerk JL, Wilson DL. Automatic 3D registration for interventional MRIguided treatment of prostate cancer. Comput Aided Surg 2002; 7:257–267. 40. Amin DV, Kanade T, DiGioia AM III, et al. Ultrasound registration of the bone surface for surgical navigation. Comput Aided Surg 2003; 8:1–16. 41. Simon DA, Hebert M, Kanada T. Techniques for fast and accurate intrasurgical registration. J Image Guided Surg 1995; 1:17–29. 42. Cuchet E, Knoplioch J, Dormont D, Marsault C. Registration in neurosurgery and neuroradiotherapy applications. J Image Guided Surg 1995; 1:198–207. 43. Bansal R, Staib L, Chen Z, et al. A minimax entropy registration framework for patient setup verification in radiotherapy. Comput Aided Surg 1999; 4:287–304. 44. Westermann B, Hauser R. Online head motion tracking applied to the patient registration problem. Comput Aided Surg 2000; 5:137–147. 45. Dawant BM, Hartmann SL, Pan S, et al. Brain atlas deformation in the presence of small and large space-occupying tumors. Comput Aided Surg 2002; 7:1–10.

2

SURGICAL NAVIGATION SYSTEM TECHNOLOGIES Narendra Nathoo Brain Tumor Institute and Department of Neurological Surgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Gene H. Barnett Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION The practice of neurosurgery is one of continuous progression, striving to improve accuracy and reduce the morbidity of intracranial surgical procedures by using the latest technological advances in computational power, imaging, materials, electronics, and robotics. The field of neurosurgical navigation has been marked by explosive progress, with advances that have generated more user-friendly systems that are, typical of computer and industrial technology in this era less expensive and more broadly available. The increasing importance and reliability of contemporary interactive image-guided surgery has led this technology to become the standard of care for many intracranial procedures. Interactive image guidance or neuronavigation implies synchronization or ‘‘registration’’ between surgical and image space. When properly achieved, this process results in accurate and reliable surgical navigation. Three features are necessary to achieve this objective: spatially accurate image acquisition and registration, sufficiently powerful and user-friendly planning and navigation computers, and accurate three-dimensional (3-D) digitizers that link surgical and image space. In this chapter, we review the technologies that provide critical human– machine interface. Related technologies such as intraoperative ultrasound and magnetic resonance imaging are discussed in later chapters.

19

20

&

PART I: BASIC PRINCIPLES

CURRENT SURGICAL NAVIGATION SYSTEMS The successful integration of surgical navigation systems (SNSs) into mainstream neurosurgery parallels the rapid advances in general computer technology. Driven primarily by innovative software applications, a myriad of these devices is presently available. Ongoing improvements include efforts to make systems more compact, with more intuitive control (e.g., touch screen), and display interfaces. Besides offering color 3-D graphics and guidance, other features include calculation and display of probe depth, autosegmentation (3-D visualization of critical structures and pathology), image fusion, laser scanning technology for surface registration, and means to facilitate camera adjustments. The most popular systems employ passive wireless digitizing technology with most of them having the capability of universal instrument registration. Integration with (sometimes robotic) microscopes can provide a virtual image overlay in the microscopes’ oculars, thereby allowing the surgeon to simultaneously view the surgical field and the virtual location of the lesion. Applications for spinal surgery such as intraoperative fluoroscopy and with some systems only utilizing an image intensifier are also available. To better understand the surgical navigation technology at its present level of development, we take a surgeon’s perspective examining the major technologies behind each step of the surgical process.

SURGICAL PLANNING AND SIMULATION The presentation of 3-D image data is intrinsic to the successful use of SNS for planning and during surgery. Manipulation of the spatial image data sets enables the neurosurgeon to simulate or plan surgical procedures (1,2). Viewing these anatomic relationships may help the surgeon by providing additional information regarding the optimal trajectory (craniotomy site, access routes to the lesion and corticotomy), and excision margins. In addition, a 3-D model for interactive surgical simulation such as trajectory optimization, access route selection, or avoidance of eloquent areas of the brain is possible. Most SNSs present information as several two-dimensional (2-D) planes with 3-D renderings of surfaces. This approach makes minimal demands on computational power and results in the fast generation of 3-D images. However, recent advances in computer hardware and software have made volume rendering a practical, interactive technique that allows processing and display of images to occur in real time (minimum, 5–10 frames/sec) using relatively inexpensive workstations (3). Unlike surface rendering, volume rendering uses the entire volume of data, sums the contributions of each voxel along a line from the viewer’s eye through the data set, and displays the resulting composite for each pixel of the display. Utilization of the entire 3-D volume data set for image reconstruction may result in greater fidelity of data with improved accuracy of autosegmentation, and unambiguous 3-D images with good depth cues. To further understand how this technique works, it may be helpful to think of the volume of data arranged as a 3-D matrix of volume

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

21

elements (voxels) floating within a 2-D computer screen composed of discrete picture elements (pixels). Utilizing defined algorithms, the 3-D volume-rendering technique determines for each pixel what portion of the data in memory (i.e., floating within the monitor) should be displayed on the screen and how that portion should be weighted to best represent spatial relationships (3). Therefore, volume rendering represents a flexible and accurate 3-D imaging technique that produces images of higher quality with depth perception. Image Fusion With the availability of advanced anatomical and functional imaging, the surgeon may wish to have a comprehensive presentation of all available digital information at his/her disposal. Most SNSs allow for integration of multimodality imaging (i.e., ‘‘fusion’’). Examples of data sets that are often useful for fusion include various sequences of MR imaging (e.g., volume acquisition images, FLAIR, etc.), functional MRI, computed tomography (CT), single photon emission computed tomography, and positron emission tomography (4). Fusion of CT and MR data sets combines the advantage of CT spatial accuracy with the superior tissue definition of MR. Such multimode image integration requires either a manual or automated fusion process. Even the most robust algorithm may, on occasion, produce an erroneous match, so it is important to be able to manually override an automated fusion. Control Interface To optimally harness the power of today’s surgical navigation, many manufacturers have directed considerable effort to develop appropriate ease of use through development of an uncomplicated interface, such as providing guided instructions in a step-by-step fashion. Some present-day SNSs allow direct intraoperative surgeon control via touch screen technology or innovative methods such as the ‘‘pointer-as-a-mouse’’ option, both of which remove the need for an operator at the workstation. Digital images are rendered onto liquid crystal display monitors integrated into dedicated workstations. Future advanced SNS may incorporate voice recognition software, biosensors, and a virtual keyboard, while information output devices may include head-mounted displays and holography. Digital information interfaced from multiple sources, such as the operating microscope and endoscope, may be easily observed by stereoscopic vision while allowing the surgeon increased freedom of movement. INTRAOPERATIVE DIGITIZATION For surgical navigation, the location of the pointing device, be it a wand, instrument, or microscope, must be accurately conveyed to the computer, which will then relate its position to imaging data. Digitizers or localizing systems are the means by which

22

&

PART I: BASIC PRINCIPLES

Table 1 Active vs. Passive Technologies Principles Active

Passive

The pointing device either emits or receives some type of signal, often requiring a wire to connect it to the surgical navigation system Either reflect a signal or are ‘‘seen’’ by some external detector

Examples Infrared emitting diodes, mechanical arms, electromagnetic Reflective infrared technologies and machine vision

information about location in surgery space is sensed by the navigation computer. The terminology associated with these devices is often daunting, and in this section we aim to review the principles by which 3-D digitizers operate, their relative benefits, and their intrinsic limitations. Today, most localizing systems use active or passive infrared or magnetic technologies, but we will endeavor to provide a balanced overview of the most common and promising systems (Table 1).

Multiarticulated Arms Among the first 3-D digitizers adapted for use with an SNS are articulated arms that provide the SNS with the orientation of the distal segment and whereabouts of its tip (Fig. 1) (5–10). The base of the arm is rigidly fixed, typically to the head immobilizer, thereby allowing head tracking (see below). Several arm segments of known length are attached at joints where sensors are fitted that determine the angle of the joint. When all the angles are known, the calculation of the location and orientation of the tip is determined by a computer. Sensors are usually potentiometers (analog) or digitizers (digital). The former are small and inexpensive, but are subject to electrical ‘‘drift’’ that may result in erroneous localization information and require recalibration during the case. Digital sensors are not subject to drift, but better accuracy requires larger, more expensive digitizers. The principal advantages of these arms are that they are simple, they do not require line of sight (see below), which is advantageous when using the surgical microscope, and they are an established technology. Unfortunately, they tend to be awkward to use as they drape over the surgeon’s arm, and are conceptually foreign since there is no comparable instrument mechanically tethered to a point that is familiar to surgeon. In order for use in a sterile field, they must be covered by a sterile drape. Furthermore, they are not readily adaptable for use with surgical instruments (e.g., bipolar, forceps, etc.) as the pointing device. Also, since these devices have limited length and orientations, it may not be possible to get the tip of the wand (especially along a desired trajectory) to a desired point in the surgical field during an intracranial or spine case. For these reasons, most SNSs use devices that do not rely on mechanical linkage.

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

23

Figure 1 Mechanical stereotactic arm. These linked technologies use analog or digital sensors at each joint to deduce the location of the ‘‘wand’’ at is tip. Low cost and lack of line-ofsight constraints are among its benefits. These mechanically linked digitizers, however, tend to be cumbersome, analog ones may drift, and they may have too-limited a reach.

Sonic Systems A discussion of sonic SNS is primarily of historic value, as it is among the earliest reported technology for use with surgical navigation (11–13). The principle here is that one or more emitters applied to a probe (14–16), endoscope (17), or microscope (11–13) produce an ultrasonic pulse that is detected by an array of microphones fixed in a known configuration (Fig. 2). If the speed of sound is known, the time delay between production of the pulse and its detection at each microphone allows the distances between emitter and microphones to be calculated. The first application of this system used emitters fixed to a surgical microscope and a detector array fixed near the ceiling of the operating room (11–13). Development of a hand-held wand required several modifications to the stereotactic microscope schema (Fig. 3) (14–16). Over several years, enhancements to determine local speed of sound (compensating for air temperature and humidity), timing to filter false echoes, and table-mounted arrays allowed for the development of a commercial system using sonic technology. Sonic systems required ‘‘line of site’’ between the emitters and all detectors. The merits and problems of this requirement follows. Although their low cost, large working volume, and accuracy were appealing, they were susceptible to

24

&

PART I: BASIC PRINCIPLES

Figure 2 Stereotactic surgical microscope used 3-D sonic digitizers to determine location of operating microscope in space. This system was not mechanically linked, accurate, and inexpensive, however was subject to error from drafts, echoes, and required line of sight between sources and receivers.

environmental noise and drafts. As such, all commercial sonic SNSs have converted to optical technologies (descriptions follow). Optical Systems Surgical navigation systems incorporating various forms of infrared digitizers are the mainstay of contemporary navigation. They are convenient, reliable, and accurate (18). Infrared Emitting Diode Systems The infrared emitting diodes (IREDs) used typically in SNS emit near-infrared light that is invisible to humans and is detected by solid-state cameras, usually composed of linear or 2-D charge-coupled devices (CCDs—similar in concept to what is used in commercial camcorders) (19–21). These line-of-site systems offer invisible, silent, almost magical operation compared to mechanical arms or sonic systems. To conserve emitter life, allow identification of which emitter is where, and

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

25

Figure 3 Adaptation of sonic technology to hand-held instruments required close placement of the receiving microphones and redesign of system electronics that minimized environmental factors except for drafts and ultrasonic noise. These systems still required line of sight and would fail if emitters got wet.

help differentiate the emitter from background light, the emitters are usually pulsed or strobed. Problems with IRED line of sight are generally readily overcome by strategic placement of the detecting camera. CCDs are quite sensitive in the infrared and ideally suited as detectors for such systems. As with any optical camera, direct and reflected light is refracted and focused to form an image within the CCD camera. Instead of film, however, a multiplicity of tiny CCD sensors detect the infrared energy, which is then converted to electrical impulses that may remain digital for further signal processing or for conversion to analog signals. The arrangement of CCD elements are either in a single line or a 2-D array. Contemporary SNS CCD cameras are very tolerant of environmental lighting, but if difficulty is encountered at surgery, sources of interference such as illuminated view boxes, sunlight, or the surgeon’s headlight should be considered. At times, intense OR lights illuminating the operative field may induce reflections that can also confuse the detection system.

Linear CCD One type of CCD camera for IRED SNS rely on CCD where tens of thousands of CCD elements are situated in a linear arrangement for each sensor. A lens

26

&

PART I: BASIC PRINCIPLES

Figure 4 Drawing of infrared optical system using three linear charge-coupled-device (CCD) cameras. These systems are silent and can readily provide continuous tracking of instruments equipped with infrared emitting diodes (IREDs). They require line of sight, are susceptible to interference from light sources, and linear cameras tend to be large and provide only the minimum information to localize an IRED in space.

focuses infrared light from the LED on a limited number of elements (Fig. 4). Localization may then be accomplished in several ways including picking the element with the strongest illumination, digital signal processing to minimize the effect of reflections, or conversion to an analog signal. Three or more of these linear CCD sensors are required to determine the 3-D coordinate of each emitter. Clearly, precise and stable alignment of cameras is a prerequisite of such systems. Advantages of one-dimensional CCD systems include high resolution and precision albeit at considerable expense due to their limited production compared to their 2-D counterparts.

Two-Dimensional CCD In this case, the CCD elements are positioned in a rectangular arrangement. Again, a lens distributes the LED emissions, but this time across the 2-D sensor. Similar detection techniques are used as described above (Fig. 5). Only two sensors are required for this arrangement, but critical alignment is again required.

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

27

Figure 5 Scheme for optical systems using a pair of 2-D CCDs (passive or active infrared, machine vision). Emitters or reference points are viewed stereoscopically. Principal advantages over linear systems are that they tend to be smaller, require a smaller line-of-sight corridor, and provide redundancy that may detect misalignments.

Today, even high-resolution, commercially available sensors are relatively inexpensive. Precise stable alignment remains critical and a blow to the camera may render it useless. Nonetheless, 2-D CCDs are, at present, the most common detector device in use for SNS.

Passive Infrared Here, similar cameras (linear or 2-D) can be used, but instead of active emitters, an infrared light source shines down on the operative field and reflective spheres are mounted on the pointing device(s). The principal advantage is that instruments tagged with these spheres are wireless and readily sterilized, whereas steam sterilization reduces the life of IRED. By virtue of this design, individual emitters cannot be strobed and passive reflectors all appear simultaneously to the cameras; thus, large numbers of these devices could confuse the system. However, in practice, they perform very well and allow for popular, accurate navigation.

Passive Ultraviolet Near-ultraviolet light can also be used to illuminate the surgical site. Fluorescent markers can then be used to absorb these frequencies of illumination and re-radiate energy at visible frequencies at levels above ambient reflected visible light. This situation is advantageous, as the markers are more easily distinguished from background data, the frequency (i.e., color) of re-irradiated light can be coded to define what marker is being imaged, and inexpensive color cameras can be used as detectors. Apart from alignment issues discussed above, work in our laboratory indicates

28

&

PART I: BASIC PRINCIPLES

that the major limitation to this approach is that ultraviolet radiation in this range is extremely fatiguing on the human eye and may be potentially dangerous. Machine Vision Another form of optical digitizing is to use a video camera (usually digital) in a known geometric confirmation aimed at the surgical field. This monitors the movements of instruments or the microscope to determine the position on 2-D video images viewed from two separate angles and calculates the 3-D positions in space. Accuracy for machine-vision-based systems has been reported to be as good as 1.5 mm for each vector calculated using this form of digitization (22,23). Surgical tools, and even the surgical microscope, are compared to a reference set of surgical tools of known geometries. Special passive markers may be applied in strategic locations to assist this recognition. Once identified, best-fit calculations can be performed to determine the location and orientation of the instrument in space, even if partially obscured by the surgeon’s hand. Although it is a promising technology and perhaps the ideal line-of-sight system, development has languished for SNS, probably due to the virtues of today’s active and passive IRED systems. In theory, any instrument could be used with minimal, if any, adaptation, once information is entered into the computer’s database. The hardware required for this technology is inexpensive, unless extremely high-resolution color systems are used. Software development, however, is the most complex of any SNS front end and is typical of those used in sophisticated military applications.

Magnetic Systems Electromagnetic digitizers use magnetic gradients or waves to provide localization to appropriate sensors. The principal advantage of these systems is that they do not require line of sight between the transmitter and receiver. This allows bulky surgical equipment such as the microscope to be used through the entire procedure and for continuous surgical navigation deep into the brain (24,25) (Fig. 6). In theory, these systems can work through acoustic and optical obstructions such as drapes and even human limbs. Metal objects common to the surgical field, however, can prove formidable obstacles to optimal use of these systems, although optimization of frequencies may minimize this problem. Unlike other digitizing technologies, which employ proximal sensor location, recent advances have led to the development of an SNS, which provides real-time tracking of the distal tips of flexible catheters, steerable endoscopes, and other instruments using ultra-low electromagnetic fields and a novel miniature distal position sensor for surgical navigation. Location determination is achieved with the use of active electromagnetic generators, which emit the ultra-low magnetic fields, and a passive sensor located within the probe/instrument. The principal advantage of this system is its small size, the ability to sense in six degrees of freedom, and accuracy.

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

29

Figure 6 Various surgical instruments for the Instatrak2 electromagnetic image-guided system requires ‘‘no line of sight.’’ Top left: biopsy probe; top right: shunt catheter passer; bottom left: suction tip; bottom right: pointing device.

Direct Current A linear relationship between what is detected and location is assumed. Unfortunately, metallic objects in or near the line of sight may result in spurious localization of the emitter. Electrical drift is also a consideration in this system. This system is not commercialized at this time in the United States. Alternating Current By producing an alternating magnetic field, interference from magnetic objects and drift can be minimized, but not eliminated. Surgery with exclusively nonmetallic instruments would substantially improve the feasibility of such systems. Use in a magnetic resonance imager would require complex integration of systems unless used only when the magnet was off, such as is feasible with resistive magnet systems. Future Localization Technologies A host of future front ends exist that may result in new SNS localization technologies. Accelerometers, similar to those used in automobile air bag systems, could provide location information without line of site or magnetic constraints if users

30

&

PART I: BASIC PRINCIPLES

are willing to place it back in a reference holder after a few minutes of use. Other techniques using radio waves, and even the global positioning system, are being considered.

REGISTRATION OF IMAGES WITH THE PATIENT Successful surgical navigation requires coherent correlation of spatial image data (image space) with the referenced surgical field (physical space) data. This registration process is discussed in the preceding chapter. Briefly, in frameless application registration, a 3-D coordinate system is established from fiducials (internal or external) on the patient and on the imaging study. Extrinsic or noninvasive temporary markers are most commonly used today; however, they have the disadvantage of potential inaccuracy if the fiducials or the scalp moved. Implantable fiducials may provide accuracy that equals or exceeds that achievable by a frame-based system, but they are more invasive and should therefore be used only when this extra accuracy is necessary. Other frameless techniques used to achieve registration include tracing surface curves or touching multiple points to generate surfaces that are used to match those of the image data. Recently, a laser-based scanning technique of surface registration has been shown to be an accurate and easy-to-use method for patient registration in interactive image guidance surgery in certain situations. A camera detects the skin reflections of the laser, and a virtual 3-D surface of the head is generated. A surface-matching algorithm then mates the 3-D matrix to the 3-D image data set. This registration method does not require skin fiducials and eliminates additional coordinate transformation to an external device (digitizer), which results in increased surgical accuracy (26). It is most useful, however, where there is substantial surface detail, such as the face and forehead, while accuracy may be substantially reduced over the vertex and posterior regions of the scalp (and also requires shaving hair in the region of registration).

INTRAOPERATIVE NAVIGATION Tracking Head Movement Virtually, all SNSs allow the surgeon to change the position of the patient after initial registration (i.e., raise-lower, rotate, pitch the head of the bed). To do this without necessitating re-registration, which may be tedious or impossible, some means of keeping track of the position of the head with respect to the detection mechanism is required. Common strategies include a reference directly affixed to the head or to the head holder, or by using a mechanical linkage of the primary sensor to the head holder or the operating table. Unfortunately, none of these systems are without fault, either in terms of convenience or accuracy. The ideal head tracking system accurately and continuously follows all subtle movements of the head, transparently adjusting the initial registration throughout

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

31

the entire surgical procedure while doing so in a noninvasive fashion, requiring no special head fixation device and without health risk to personnel in the room. Currently, such a system does not exist. Table-Mounted Detectors By mounting the detector (irrespective of technology) to the head portion of the operating table, SNS can take advantage of the mechanical linkage of the head to the table through the head fixation device. This coupling, however, is imperfect as the detector may move under its own weight and drapes or other forces may move the patient’s head with respect to the detector. Superior technologies described below have largely superceded this approach. Table-Mounted References Here a lightweight reference can be mounted to the table with a remote sensor. The disadvantages of head-to-table-to-reference coupling remain and there is little advantage apart from some increased flexibility of reference placement over fixation-mounted reference systems.

Head-Holder Mounts By directly attaching to the head holder, most (but not all) potential movement between the head and digitizer is eliminated so that head tracking is generally more accurate. Common three-point fixation devices, however, still allow head movement with respect to the main structural frame. Head-holder mounts are most commonly used for some arms and electromagnetic systems.

Head-Holder References—Dynamic Reference Frames This approach provide the advantages of head-holder mounting for systems that require the detector to be placed remotely and is employed by virtually all optical systems for intracranial navigation. When using the popular Mayfield head fixation device, most systems mount to the body or attachment portions of the device, which may allow for some head movement (principally rotational) with respect to the dynamic reference frames (DRFs) (Fig. 7). One commercial system attaches the head-tracking device directly to ‘‘C’’ portion of a Mayfield fixation device. This approach virtually eliminates issues of head movement (Fig. 7). The principal disadvantages of DRF are that not only line of site must be maintained between detector and the pointing device, but also the detector and the reference. Also, it is essential after registration that the DRF not be displaced

32

&

PART I: BASIC PRINCIPLES

Figure 7 Dynamic reference frame (DRF) using active infrared emitting diodes. When secured to the head holder, allows the navigation to track, and therefore compensate for head or camera movement.

from its initial relationship with the head, as it will then result in spurious intraoperative navigation.

SPINE NAVIGATIONAL (FLUOROSCOPY) TECHNOLOGIES Traditional means of intraoperative localization of spinal anatomy are generally considered to be inaccurate (27,28). The localization of critical bony landmarks during spinal surgery has largely depended on the experience and skill of the surgeon, knowledge of spinal anatomy, and the ability to think in a 3-D fashion. The anatomy of independent vertebral bodies, which, individually, remain anatomically constant despite varying intervertebral relationships, makes the spine ideally suitable for image-guided surgery. Intersegmental movement between the time of image acquisition and that of surgery is generally not a problem, as separate registration of each spinal segment is performed, thereby enhancing the accuracy of intraoperative navigation. To maintain accuracy between the image space and surgical space, the motion of the spine during surgery may be detected in a real-time fashion (dynamic referencing), although some investigators have not found this to be necessary. One approach has the ability to quickly and automatically register the patient without the need for a CT scan. Combining intraoperative C-arm fluoroscopy with free-hand surgical navigation techniques, modules were developed to automate digital X-ray image registration. Following attachment of a noninvasive localizer

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

33

to the image intensifier, the first X-ray image is acquired. Those are automatically detected and this information is used to correct for deformation. The patient is then registered to the intraoperatively acquired images. Therefore, quick and accurate registration is possible without the need for fiducials with cross-referencing of the image intensifier with the surgical object allowing real-time image-interactive navigation of surgical tools based on a single X-ray image, with no further updates (29). Other features offered with fluoroscopic navigation include fluoroscopic tracking of multiple surgical instruments and a reduction in radiation exposure to the surgical team by allowing surgical navigation in real time on the initial fluoroscopic shot. In addition, with universal adaptors, potentially any image intensifier may be used. Therefore, this approach to spinal SNS provides the surgeon with a multidimensional view of anatomic relationships in the operative field, which may increase surgical accuracy and safety. Undoubtedly, future spinal SNS will be incorporated into percutaneous and minimally invasive procedures. COMPUTER-DIRECTED NEUROSURGERY/ROBOTIC SURGERY Robots may be considered instruments of stereotaxy by virtue of their precise, spatially defined, deliberate movements. Computer-assisted neurosurgery using SNS may therefore be viewed as an intermediate technology to computer-directed robotic neurosurgery in the near future. The use of robots in surgery has three main advantages. First, they have superior 3-D spatial accuracy, especially when linked to digitized image information. Second, a significant improvement in manual dexterity is possible when using a robotic interface. This will allow the surgeon to operate through smaller corridors of access and choose longer working distances without tiring easily. Lastly, robots can be designed to be extremely precise, accurate, and as such, can potentially produce more reproducible outcomes with smaller margins of error. Conventional image-guided surgical systems, as we know them today, will likely incorporate robotic components in the near future.

CONCLUSION Technology for SNS continues to evolve. Commonplace systems of today will likely evolve into totally different technologies in the near future. Each present and future system has limitations and advantages that potential users should understand when considering purchase and to optimize their surgical benefit.

34

&

PART I: BASIC PRINCIPLES

REFERENCES 1. Jolesz F, Nabavi A, Kikinis R. Integration of interventional MRI with computer assisted surgery. Magn Reson Imaging 2001; 13:69–77. 2. Nakajima S, Atsumi H, Bhalerao AH, Jolesz FA, Kikinis R, Yoshimine T, Moriarty TM, Stieg PE. Computer assisted surgical planning for cerebrovascular neurosurgery. Neurosurgery 1997; 41:403–409. 3. Calhoun PS, Kuszyk BS, Heath DG, Carley JC, Fishman EK. Three-dimensional volume rendering of spiral CT data: theory and method. Radiographics 1999; 19:745–764. 4. Gering DT, Nabavi A, Kikinis RT, Eric W, Grimson L, Hata N, Everett P, Jolesz FA, Wells WM III. An integrated system for surgical planning and guidance using image fusion and interventional imaging. In: Proceedings of the Second International Conference on Medical Image Computing and Computer Assisted Intervention. Berlin: Springer-Verlag, 1999:809–819. 5. Guthrie BL, Adler JR. Frameless stereotaxy: computer interactive neurosurgery. In: Barrow DL, ed. Perspectives in Neurological Surgery. Vol. 2. No. 1, St. Louis: Quality Medical Publishing, 1991:1–19. 6. Watanabe E, Watanabe T, Manaka S, Mayanagi Y, Takakura K. Three-dimensional digitizer (neuronavigator): new equipment for computed tomography-guided stereotaxic surgery. Surg Neurol 1987; 27:543–547. 7. Golfinos JG, Fitzpatrick BC, Smith LR, Spetzler RF. Clinical use of a frameless stereotactic arm: results in 325 cases. J Neurosurg 1995; 83:197–205. 8. Olivier A, Germano IM, Cukiert A, Peters T. Frameless stereotaxy for surgery of the epilepsies: preliminary experience. J Neurosurg 1994; 81:629–633. 9. Doshi PK, Lemmieux L, Fish DR, Shorvon SD, Harkness WH, Thomas DG. Frameless stereotaxy and interactive neurosurgery with the ISG viewing wand. Acta Neuroschir Suppl (Wien) 1995; 64:49–53. 10. Guthrie BL, Adler JR Jr. Computer-assisted preoperative planning, interactive surgery, and frameless stereotaxy. Clin Neurosurg 1992; 38:112–131. 11. Friets EM, Strohbehn JW, Hatch JF, Roberts DW. A frameless stereotaxic operating microscope for neurosurgery. IEEE Trans Biomed Eng 1989; 36:608–617. 12. Roberts DW, Strohbehn JW, Friets EM, Kettenberger J, Hartov A. The stereotactic operating microscope: accuracy refinement and clinical experience. Acta Neurochir Suppl (Wien) 1989; 46:112–114. 13. Roberts DW, Strohbehn JW, Hatch JF, Murray W, Kettenberger H. A frameless stereotaxic integration of computerized tomographic imaging and the operating microscope. J Neurosurg 1986; 65:545–549. 14. Barnett GH, Kormos DW, Steiner CP, Piraino D, Weisenberger J, Hajjar F, Wood C, McNally J. Frameless stereotaxy using a sonic digitizing wand: development and adaptation to the Picker ViStar medical imaging system. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. American Association of Neurological Surgeons, 1993:113–119. 15. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Intraoperative localization using an armless, frameless stereotactic wand. J Neurosurg 1993; 78:510–514.

CHAPTER 2: SURGICAL NAVIGATION SYSTEM TECHNOLOGIES

&

35

16. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Use of a frameless, armless stereotactic wand for brain tumor localization with 2-D and 3-D neuroimaging. Neurosurgery 1993; 33:674–678. 17. Rhoten RP, Luciano M, Barnett GH. Computer-assisted endoscopy for neurosurgical procedures: technical note. Neurosurgery 1997; 40(3):632–637. 18. Li Q, Zamorano L, Jiang Z, Gong JX, Pandaya A, Perez R, Diaz F. Effect of optical digitizer selection on the application accuracy of a surgical localization system—a quantitative comparison between the OPTOTRAK and Flashpoint tracking systems. Comput Aided Surg 1999; 4(6):314–321. 19. Maciunas RJ, Galloway RL Jr, Fitzpatrick JM, Mandava VR, Edwards CA, Allen GS. A universal system for interactive image-directed neurosurgery. Stereotact Funct Neurosurg 1992; 58:108–113. 20. Ryan MJ, Erickson RK, Levin DN, Pelizzari CA, Macdonald RL, Dohrmann GJ. Frameless stereotaxy with real-time tracking of patient head movement and retrospective patient-image registration. J Neurosurg 1996; 85:287–292. 21. Smith KR, Frank KJ, Bucholz RD. The NeuroStation—a highly accurate, minimally invasive solution to frameless streotactic neurosurgery. Comput Med Imaging Graph 1994; 18:247–256. 22. Heilbrun MP, Koehler S, McDonald P, Peters W, Sieminov V, Wiker C. Implementation of a machine vision method for stereotactic localization and guidance. In: AANS Pulication Committee, Maciunas RJ, eds. Interactive Image-Guided Neurosurgery. Chicago: American Association of Neurological Surgeons, 1993. 23. Heilbrun MP, McDonald P, Wiker C, Koehler S, Peters W. Stereotactic localization and guidance using a machine vision technique. Stereotact Funct Neurosurg 1992; 58:94–98. 24. Manwaring KH. Magnetic field guided endoscopic dissection through a burr hole may avoid more invasive craniotomies—a preliminary report. Acta Neurochir Suppl 1993; 61:34–39. 25. Kato A, Yoshimine T, Hauakawa T, Tomita Y, Ikeda T, Mitomo M, Harada K, Mogami H. A frameless, armless navigational system for computer-assisted neurosurgery. Technical note. J Neurosurg 1991; 74:845–849. 26. Raabe A, Krishnan R, Wolff R, Herman E, Zimmerman M, Seifert V. Laser surface scanning for patient registration in intracranial image-guided surgery. Neurosurgery 2002; 50(4):797–801. 27. Foley KT, Smith MM. Image-guided spine surgery. Neurosurg Clin North Am 1996; 7(2):171–186. 28. Zamorano L, Vinas FC, Jiang Z, Diaz FG. Use of surgical wands in neurosurgery. Adv Tech Stand Neurosurg 1998; 24:77–128. 29. Nolte LP, Slomczykowski MA, Berlemann U, Strauss MJ, Hofstetter R, Schlenza D, Laine T, Lund T. A new approach to computer-aided spine surgery: fluoroscopybased surgical navigation. Eur Spine J 2000; 9(suppl 1):S78–S88.

3

TARGET AND TRAJECTORY GUIDANCE Charles P. Steiner Division of Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Image-guided frame stereotaxy permits accurate guidance to image-derived targets and can readily do so along predefined surgical trajectories (1–3). Frame stereotaxy systems have had limited application in procedures such as craniotomy (4–9) because many neurosurgeons find these devices cumbersome and logistically difficult to use for surgical guidance. Additionally, the ability of frame stereotaxy systems alone to provide localization and orientation information in real time is, at best, limited. In contrast, interactive surgical navigational systems (SNSs) are adept at showing the location and orientation of an instrument during surgery using image data obtained preoperatively (10–13). Unlike image-guided frame stereotaxy, however, the ability of SNSs to guide the surgeon to predefined targets along prescribed trajectories has been typically limited and awkward. Although most SNSs readily identify tip position and orientation of a pointing device (e.g., wand), they do not readily provide intuitive access to a predefined target or replicate a desired trajectory. The development of a target and trajectory guidance system has helped to extend the application of SNSs beyond craniotomy since it facilitates guidance to small, deep intracranial structures (14). A method of target and trajectory guidance that has proved to be simple, reliable, and accurate will be presented.

SYSTEM DESCRIPTION AND REQUIREMENTS Target and trajectory guidance routines were developed as enhancements for the prototype computer-assisted minimally invasive surgery (CAMIS) system at The Center for Computer-Assisted Surgery in The Cleveland Clinic Foundation’s Department of Neurosurgery. This SNS has a modular design and uses an infrared optical three-dimensional digitizer with dual two-dimensional charge-coupled device

Icons indicate materials available on the accompanying CD.

37

38

&

PART I: BASIC PRINCIPLES

cameras and DEC Alpha Station 5000 workstation. The image is displayed on 20-in. RGB monitors and a prototype head-mounted display. The software architecture of CAMIS incorporates an open toolkit methodology that allows for rapid prototyping of new features, typically on the order of hours as opposed to weeks. The fine-tuning of new techniques can be performed on a large collection of phantoms or in the operating room while providing adequate backup mechanisms. The target and trajectory guidance system design concept was to create a methodology that fulfilled the following criteria: & & & & & & &

Ability to select one or more targets Option to predefine one or more trajectory(ies) to reach the target Ability to select the entry point from image data or wand location Provide depth information from wand tip to target slice Visually provide reorientation information Automatic operation capabilities at surgery Portable to any SNS

Our approach to target guidance was to logically extend a previously designed method (10,11) to ‘‘look beyond’’ the physical limits of a navigational probe to the view(s) at a defined depth. This would require a mechanism to select targets and entries in three-dimensional space as well as a display interface to correctly describe one’s current orientation relative to the preselected target. This methodology has subsequently been published and incorporated into at least one commercial surgical navigation system (14).

LOCALIZATION AND ORIENTATION The CAMIS system provides two display motifs. A triplanar display simultaneously shows three orthogonal images with a common vertex (i.e., images parallel to coronal, sagittal, and transverse planes) and is typically used in surgery for purposes or real-time localization (Fig. 1). The manual selection of this vertex can be accomplished through the use of a mouse cursor, the localizing wand, or slider controls on screen. Placing the cursor over a point on any of the three orthogonal images and pressing a mouse button causes the system to extract and display the two other images that are perpendicular to the current image and parallel to the principal axis of the data set. This process can also be used to browse through the image data and to select voxel coordinates for storage, such as target and/or entry point selection. The second display mode of the CAMIS system is where images are extracted based on the orientation of the pointing device (typically a hand-held wand). The primary image obtained is perpendicular to the axis of the pointing device with the position of the pointer displayed in the center of this image (surgeon’s view). The rotation corresponds to the view of the operator down the wand axis, and the default depth is at the tip of the localizing wand. The distance

CHAPTER 3: TARGET AND TRAJECTORY GUIDANCE

&

39

Figure 1 Triplanar display of coronal, sagittal, and transverse planes with a common vertex.

or depth from the tip of the want to the surgeon’s view can be ‘‘pushed’’ deeper or ‘‘pulled’’ more superficial by the surgeon using either a mouse or foot pedal. Two other images, both containing and displaying the localizing axis of the pointing device and perpendicular to each other, are generated in real time, thereby creating a display with three mutually orthogonal images (‘‘multiplanar obliques’’ or MPR) although usually not parallel to the principal axis of the data set (Fig. 2).

TARGET GUIDANCE Target Push Target guidance on the CAMIS system uses this MPO display to show the relationships between predefined targets and/or trajectories with the actual wand trajectory (AWT). Once the AWT is determined by performing localization, one of the indefinite number of parallel images must be reformatted to provide the surgeon’s view image. The target push feature of target guidance selects the one image

40

&

PART I: BASIC PRINCIPLES

Figure 2 MPO display with mutually orthogonal images oriented based on the pointing device.

that is perpendicular to the viewing axis and contains the predefined target (Fig. 3). The distance from the wand tip to the plane along the present wand trajectory is also calculated and presented.

Concentric Circles Conceptually, the target can be considered a 4 mm sphere centered at the predefined location. The intersection of any target-push plane with this sphere results in a circle with a 4 mm diameter. Similarly, the wand can be considered a cylinder of larger diameter, and its projection along the AWT necessarily results in another circle where it intersects the auto-push plane (Fig. 4). On this image, the projected position of the pointing device is fixed as the center of the frame and is shown as a small circle in a user-selectable color. The position of the target is also shown on this image as a smaller circle in a different color. Should the boundaries of these circles overlap the smaller circle is presented on top of the larger one. Reorientation of the wand simply requires rotating or translating the wand to move the larger circle in a direction that will result in two circles being concentric.

CHAPTER 3: TARGET AND TRAJECTORY GUIDANCE

&

41

Figure 3 Surgeon’s view oblique image perpendicular to the viewing axis containing the predefined target depicted as a solid circle.

Together, target-push and concentric circles are sufficient to provide guidance to a target when it is not necessary to follow a predefined trajectory.

TRAJECTORY GUIDANCE Collinear Spheres and Concentric Circles The addition of a trajectory guidance function provides information about the position and orientation of the pointing device with respect to a predefined surgical trajectory. Creation of the surgical trajectory requires the same tools discussed earlier for target selection for the definition of an entry point. Together, target and entry points define a linear surgical trajectory. In theory, nonlinear trajectories could be approximated by a series of linear surgical trajectories. The target is once again represented by a small sphere, and the intersection of the target sphere with the target-push plane generates a small circle. Although the previous paradigm for generating the AWT circle remains valid, an alternative mechanism is to create a virtual sphere at the tip of the wand. Its projection

42

&

PART I: BASIC PRINCIPLES

Figure 4 Surgeon’s view oblique image perpendicular to the viewing axis containing the predefined target and probe position as two open circles.

down the AWT such that its center is in the target-push plane will also result in a circle where sphere surface and plane intersect. Finally, a virtual sphere of larger diameter is envisioned at the entry point and is projected onto the target-push plane along a line parallel to the AWT, the intersection of which results in a third circle. Appropriate wand rotations and translations of the wand are performed to make the three circles concentric. This is most easily achieved by first aligning the target circle with the entry circle using a series of rotations followed by wand translations to achieve concentricity. When the wand is appropriately oriented to access the target along this predefined trajectory, not only are the three auto-push circles concentric but the three virtual spheres are also collinear.

Trajectory Projection Because some users find it difficult to align the circles to replicate a desired trajectory, an additional method of trajectory guidance called trajectory projection was devised. The MPO images contain the AWT, and as a result the surgical trajectory can be projected onto these images to provide information as to the necessary

CHAPTER 3: TARGET AND TRAJECTORY GUIDANCE

&

43

Figure 5 (A) MPO display showing three concentric circles and alignment vectors, light gray indicating AWT and dark gray the intended trajectory. (B) The initial alignment step is to make vectors parallel by wand rotations. (C) The wand is then translated while maintaining the same rotational orientation. Collinear vectors result in concentricity of all circles, indicating correct surgical trajectory.

44

&

PART I: BASIC PRINCIPLES

Figure 5 (Continued )

translations and rotations required to orient the pointing device along the surgical trajectory. When the pointing device is properly oriented, the images perpendicular to the pointing device will show three concentric circles, and the other two MPO images will show vectors aligned parallel and collinear with the pointing device’s orientation (Fig. 5). The alignment method used by most surgeons is to first rotate (tilt) the wand such that the AWT is parallel to the predefined surgical trajectory. Final alignment is achieved by wand translations until the AWT and surgical trajectory are collinear. The amount and direction of translation required are depicted by the separation of the parallel vectors displayed on the alternate images.

DISCUSSION It became clear that virtually all of the design goals for this target and trajectory target system were easily met except for devising a simple, intuitive form of presentation to reorient the surgeon. Certainly, the most direct approach would be to have robotic manipulation of the wand (15–17). Although promising work on bringing robotic technology to the surgical arena has been reported, such a solution was beyond the scope of this project. Another approach considered

CHAPTER 3: TARGET AND TRAJECTORY GUIDANCE

&

45

was to create volume cutaways oriented such that one is looking down the AWT. This was excessively computer intensive and to some confusing. An alternative was to represent the target as a solid, seen through a semitransparent three-dimensional surface (Fig. 6A). Similarly, an idealized trajectory could be projected and compared to the AWT. Unless the images were presented stereoscopically or from orthogonal views, however, it was difficult to visually interpret the depth of the object or even which side of the brain it was on (Fig. 6B). Finally, this technique also proved too computationally intensive to provide useful information in a near real-time fashion. Instead of projecting the entry point onto the target-push plane, we attempted to show the entry point in a semitransparent plane parallel to the target-push plane. Although less computer intensive, this method proved too confusing to be reliable and helpful. After seeing the MPO display of a CAMIS-derived system, Dr M. Peter Heilbrun (personal communication, 1993) suggested the use of vectors to help define the direction and magnitude of reorientation for target trajectory guidance. Initial results appeared promising, since vectors could visually represent the necessary modifications to the AWT for target and/or trajectory guidance under most circumstances (Fig. 7). However, as the AWT approached the target or surgical trajectory, the vectors became so small that magnitude and direction of correction became unclear.

Figure 6 (A) Three-dimensional representation of a patient with the target depicted as a solid object viewable through a semitransparent surface (lateral view). (B) Three-dimensional representation of patient with the target depicted as a solid object viewable through a semitransparent surface (anteroposterior view).

46

&

PART I: BASIC PRINCIPLES

Figure 7 Surgeon’s view oblique image perpendicular to the viewing axis containing direction vectors indicating the direction and magnitude of correction necessary to position the AWT along the idealized surgical trajectory.

It became apparent that a similar visually intuitive display for target guidance could be achieved by representing the target and AWT as circles, the space and orientation between them providing the same information as the vector display. However, unlike vectors, as the AWT approaches the target, the reorientation information remains robust, aiding final reorientation of the wand resulting in concentric circles that denote being ‘‘on target.’’ Although other shapes (triangles, squares, etc.) could be have been used, rotational alignment of the vertices adds further computations but conveys no further information than that derived from circles. A similar paradigm developed to replicate trajectories with concentric circles and a triad of virtual collinear spheres proved less intuitive to some users while adopted easily by others. The use of trajectory projection onto the MPO planes serves as an additional or alternative guide when attempting to duplicate a predefined trajectory.

CHAPTER 3: TARGET AND TRAJECTORY GUIDANCE

&

47

The purpose of any surgical tool is to provide function with ease of use. Computer interfaces suitable for operation by a programmer may be far too cumbersome to be used in the operating room (16). This implementation of a surgical guidance tool incorporates the sophisticated features of a flight simulator while limiting the controls and configuration to those needed for surgery (18,19). Target guidance is the simplest function of this system, providing the information necessary to orient a pointing device in space such that it will intersect a predetermined target. The distance from the tip of the pointing device to the intersection of this target is displayed, information that is useful when approaching a subcortical lesion during craniotomy and also allowing appropriate depth setting of instruments used for biopsy, cyst aspiration, depth electrodes, and similar applications. The use of trajectory guidance may refine the surgical approach such as when it is desired to avoid eloquent areas or it is necessary to pass through an identifiable structure en route to the target (e.g., foramen of Munroe for third ventriculostomy). This implementation of target and trajectory guidance has been used in over 500 neurosurgical procedures and has proved a simple and accurate solution to the problem of guidance in surgery.

48

&

PART I: BASIC PRINCIPLES

REFERENCES 1. Abernathy CD, Camacho A, Kelly PJ. Stereotactic suboccipital transcerebellar biopsy of pontine mass lesion. J Neurosurg 1989; 70:195. 2. Brown RA. A computerized tomography-computer graphics approach to stereotaxis localization. J Neurosurg 1979; 50:715–720. 3. Galloway RL, Maciunas RJ, Latimer JW. The accuracies of four stereotactic frame systems: an independent assessment. Biomed Instrum Technol 1991; 25:457–460. 4. Barnett GH, McKenzie RL, Ramos L, Palmer J. Nonvolumetric stereotaxy assisted craniotomy. Results in 50 consecutive cases. Stereotac Funct Neurosurg 1993; 61:80–95. 5. Gomez H, Barnett GH, Estes ML, Palmer J, Magdinec M. Stereotactic and computer-assisted neurosurgery at the Cleveland Clinic. Review of 501 consecutive cases. Cleve Clin J Med 1993; 60:399–410. 6. Hassenbusch SJ, Anderson JS, Pillay PK. Brain tumor resection aided with markers placed using stereotaxis guided by magnetic resonance imaging and computed tomography. Neurosurgery 1991; 28:801–806. 7. Kelly PJ. Volumetric stereotactic surgical resection of intra-axial brain mass lesions. May Clin Proc 1988; 63:1186. 8. Kelly PJ, Kall BA, Goerss S, Earnest F IV. Computer-assisted stereotaxic laser resection of intra-axial brain neoplasms. J Neurosurg 1986; 64:427. 9. Moore MR, Black PM, Ellenbogen R, Gall CM, Eldredge E. Stereotactic craniotomy: methods and results using the Brown–Roberts–Wells stereotactic frame. Neurosurgery 1989; 25:572–577. 10. Barnett GH, Kormos DW, Steiner CP, Piraino D, Weisenberger J, Hajjar F, Wood C, McNally J. Frameless stereotaxy using a sonic digitizing wand: development and adaptation to the Picker ViStar Medical Imaging System. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:113–119. 11. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Intraoperative localization using an armless, frameless stereotactic wand [technical note]. J Neurosurg 1993; 78:510–514. 12. Guthrie BL, Adler JR. Frameless stereotaxy: computer interactive neurosurgery. Perspect Neurol Surg 1991; 2(1):1–19. 13. Maciunas RJ. Yesterday’s tomorrow: the clinical relevance of interactive imageguided surgery. Neurosurg Top 1993; 1:3–8. 14. Barnett GH, Steiner CP, Weinseberger J. Target and trajectory guidance for interactive surgical navigation systems. Stereotact Funct Neurosurg 1996; 66:91–95. 15. Buckingham R. Robotics in surgery. IEEE Rev 1994; 40:193–196. 16. Cohn MB. Surgical applications of milli-robots. J Robot Syst 1995; 12:401–416. 17. Glauser D, Fankhauser H, Epitaux M, Hefti JL, Jaccottet A. Neurosurgical robot Minerva. First results and current developments. J Image Guided Surg 1995; 1:266–272. 18. Beardsley T. A new view for surgeons. Sci Am 1994; 271:100. 19. Swenson HN, Jones RD. Combat automation for airborne weapon systems: man/ machine interface trends and technologies. AGARD N93–28850, 1993:11–54.

4

PITFALLS Robert J. Maciunas Department of Neurological Surgery, Case Western Reserve University, University Hospitals of Cleveland, Cleveland, Ohio, U.S.A.

Recent technologic advances have initiated an ongoing revolution in surgical navigation (1–3). Digital tomographic imaging modalities produce three-dimensional mosaics of patient data that are exquisite in sensitivity and detail and can be used as spatially accurate maps of patients’ anatomy. Novel methods of registering, or mapping, these image data sets onto patient anatomy as it exists in physical space have made stereotaxy frames obsolete for this purpose. Interactive localization devices have provided a means of three-dimensionally digitizing the surgical workspace, relating the imaging data sets and the patients’ anatomy to surgeons and their instruments. Computers enable us to manipulate enormous volumes of information and display this information in a visually compelling and clinically meaningful manner. The incorporation of real-time intraoperative data into our calculations is rapidly expanding. Further advances in visualization and robotics promise to further transform the neurosurgical process in our lifetimes. These technologic developments are sufficiently profound so that clinical optimism about what is possible to accomplish via neurosurgical intervention has reached new heights. All is for the best in this best of all possible worlds. —Voltaire, Candide Before we are overcome by the hubris of technologic advance and submit ourselves to the uncritical optimism of Dr. Pangloss, we should consider the inherent risks and limitations—whether resulting from imperfect manifestations of what can be accomplished with this technology, inherent limitations of the technologic solutions themselves, or limitations of our own abilities and judgment. It behooves us to become cognizant of the pitfalls of using surgical navigation technologies so that we might more safely, efficiently, and effectively make use of them to the benefit of those under our care. There lives more faith in honest doubt, believe me, than in half the creeds —Tennyson, In Memorium 49

50

&

PART I: BASIC PRINCIPLES

These new technologies for surgical navigation and image analysis have been termed interactive image-guided neurosurgery. A system is composed of five fundamental elements: a method for registration of images and physical space, an interactive localization device, a computer with its requisite software interface and video display system, the integration of real-time feedback, and robotics. The pitfalls of interactive image-guided neurosurgery may be analyzed by sequentially discussing the following nine topics: (1) definition of accuracy, (2) image acquisition, (3) registration techniques, (4) computers and software interfaces, (5) interactive localization devices and intraoperative use, (6) integration of real-time data, (7) tissue displacement, (8) robotics, and (9) judgment and clinical experience.

DEFINITION OF ACCURACY In an article on three-dimensional digitizers, T. Wohlers wrote a sidebar commentary entitled ‘‘How Accurate Is Accurate?’’ In his comments, he states: Comparing the accuracy of 3-D digitizers can be confusing. The terms ‘‘accuracy,’’ ‘‘precision,’’ ‘‘resolution’’ and ‘‘repeatability’’ are often misused, misunderstood, misleading or just plain omitted. Buyer confusion also occurs when digitizer suppliers publish numbers related to the specific parts . . . that make up the system. This can be misleading. The precision of the system’s mechanics does not reflect the accuracy of the process. In fact, the precision of the mechanics is . . . always better than the accuracy of the process (4). Because a stereotactic surgical guidance system is essentially a three-dimensional spatial digitizer coupled with mapping image space onto physical space, Wohlers’ caveats are appropriate. Although the terms ‘‘stereotactic’’ and ‘‘submillimetric accuracy’’ are usually closely associated in the literature, the neurosurgeon must exercise considerable caution when dealing with such assumptions. A considered approach to the subject of stereotactic accuracy requires attention to the definition of terms. First, a clear distinction must be drawn between three commonly misused terms: ‘‘unbiasedness’’ (or lack of skew), ‘‘precision,’’ and ‘‘accuracy.’’ A series of observations that tend to the true value are unbiased, or without skew. If these observations have considerable spread, however, they lack precision. A series of observations with little spread among them indicate precision, although they are biased, or skewed, if the observations tend to center at a value displaced from the true value. Accuracy encompasses both unbiasedness and precision: accurate measurements are both unbiased and precise, although inaccurate measurements may be biased, imprecise, or both. Because the mechanical precision of a stereotaxy frame system determines how well it can return to a point, it may be thought of as being, in effect, a measure of how finely incremented a relative ruler the apparatus is. Finer increments on the verniers or better machining tolerances can be expected to improve mechanical precision.

CHAPTER 4: PITFALLS

&

51

Incidentally, it is often this number that is advertised for stereotactic devices. When a system of interactive image-guided surgery demonstrates a tendency to consistently arrive at a wrong answer, this skew results from mechanical bias in a precise system. The mechanical accuracy of a surgical navigation apparatus is the accuracy with which it can bring the tip of a straight probe to a given coordinate within the stereotactic coordinate system. This requires both mechanical unbiasedness and mechanical precision, but does not include the effects of imaging technologies on localization. The standard performance specifications for cerebral stereotactic instruments issued by the American Society for Testing and Materials state that the mechanical accuracy of a stereotaxy frame-based system shall be submillimetric. In addition to the mechanical limitations of the stereotaxy frame, the error associated with the many steps of stereotaxy (including imaging techniques, point selection, vector calculation, vernier settings, mechanical couplings, and adjustments) contributes to the final clinically relevant error. Human operator error lies outside the realm of this discussion, although tedious calculations or taxing calibrations imposed by a system will make these errors more likely. It is in the interest of the neurosurgeon to go beyond assessment of the mechanical accuracy of various surgical navigation systems (SNSs) to assess what may be termed the application accuracy of these devices. This is a measure of the accuracy of these devices when they are used in their real-world setting. It is separate and distinct from any mere measurement of the unbiasedness, precision, or accuracy associated with the machinery of the interactive image-guided system or with the stereotaxy frame construction itself. When the application accuracy of various stereotaxy frames was analyzed, it was demonstrated that these devices had a millimetric to centimetric level of accuracy, which was strongly dependent upon scan slice thickness (5). Furthermore, the method of reporting this number was itself reassessed. What is reported is the mean of the errors and not the mean for all erroneous coordinates, which may end up being closer to the correct value if two oppositely incorrect values cancel each other out. Most previous reports had offered a mean value as the only numeric statement, and some reported a mean with the standard error of the mean. The standard error of the mean, however, is not relevant to this discussion and is smaller than the standard deviation (SD), which is the value that should be reported. Therefore, the mean SD is the standard for reporting the application accuracy, or target localization error, or values. Finally, a 99% confidence interval may be reported, which is equal to the mean plus three times the SD. This allows the surgeon to understand how the system will perform 99 times out of 100, a situation more analogous to the calculation of surgical clinical risk. When assessing accuracy, three separate, distinct, and not directly related numbers must be assessed. The first is fiducial localization error. This is the accuracy of the system in finding the fiducial in an image or in physical space. The next is fiducial registration error. This is the residual error after registration of two images or of an image with physical space as judged by the sum of the root mean square error for the mapping of fiducial markers in one image set onto fiducial markers in another image set. This is mathematically related to the fiducial localization error, and is a larger number. The most clinically

52

&

PART I: BASIC PRINCIPLES

important value is the target localization error, which is not directly correlated with the fiducial registration error, although it does track along with it and is often a larger number. The clinician wants to know the target localization error; instead, the SNS usually reports the fiducial registration error as a number for his or her approval. It is also important to realize that eliminating fiducials to decrease the fiducial registration error may create a seemingly good registration, but a very poor target localization error. This occurs when the fiducials eliminated are those farthest from the surgical field, leaving only a group clustered near one another; if the target is distant from these fiducials, then the target localization error may worsen when the distant fiducial is eliminated, and the registration will be inadvertently skewed. Ultimately, we are left with an operational definition of accuracy for imageassisted and image-guided surgery: an SNS is accurate (precise and unbiased) when the surgeon observes that the anatomic structure localized in physical space always corresponds to an appropriate address in image space. It’s hard to define, but I know it when I see it. —To paraphrase Justice Potter Stewart in Jacobelis v. Ohio, 378 U.S. 184, 197 (1964) This aphorism holds true so long as the surgeon is independently secure with an alternative way of defining a target; once the system is guiding rather than assisting the surgeon, full reliance is placed on its accuracy without a realistic capability for cross-checking with an anatomic landmark. This is especially true for pedicle screw placement, where cross-checking is possible for surface anatomy but not for deep pedicular anatomy and intraoperative fluoroscopy is used as an independent, intraoperative, real-time imaging modality to cross-check the system and perhaps to update the registration.

IMAGE ACQUISITION The ability of neurodiagnostic images to form a satisfactory basis for surgical navigation is dependent on the images’ spatial fidelity and precision and on their gray scale depth of discrimination. The capacity to provide true, point-to-point maps relies upon the images having a consistent, predictable, and finely resolved matrix of spatial addresses for each pixel. The surgeon must be certain that the presumed location of a specific anatomic detail corresponds to its displayed position in the image. Without a clearly defined relationship that can be relied upon in all the parts of an image, misregistration or mismatching of individual points between images and physical space will occur. The images will not serve as true maps but will, in effect, be cartoons that give the qualitative appearance of an anatomic structure without quantitatively providing a reliable address for each point. An example of this is digital angiography. Here, the photomultiplier tube generating the image will introduce a series of spatial distortions on the image that are variable, depending on the status

CHAPTER 4: PITFALLS

&

53

of the tube, patient, and radiographic settings. These distortions are not consistent, predictable, or reproducible, thus preventing engineers from producing an algorithm to correct spatial misrepresentations a priori. For this reason, digital angiography is used with circumspection in stereotactic radiosurgery and image-guided surgery. Maciunas et al. (4) definitively established that slice thickness is an over-riding factor in determining the maximal level of localization accuracy with a given imaging data set. This understanding has resulted in increasing clinical reliance on thinner slice thickness for navigation scans. Although in the early pioneering work of Kelly (6) and Kelly et al. (7) a slice thickness of 8 to 10 mm was used, routine clinical use now presumes slices of 1 to 4 mm in thickness. The overlapping of slices has been abandoned because thinner slices are a more effective means of improving resolution and the overlapping itself presents considerable complexities for the computerized reconstruction of a three-dimensional database from the slices. Similarly, spacing between slices has been eliminated in CT with the advent of rapid spiral acquisitions with parallel-slice reconstruction. Thus, no imaging information is lost in the gaps between slices. In MR images, spacing between slices is favored by radiologists and technicians because it prevents interslice interference. When used, the slices should be handled as if they were simply that much thicker than the thickness imaged. With the advent of three-dimensional fast acquisitions, this is becoming less of an issue. Some MR image scanners and pulse sequences are particularly prone to geometric distortions and to ‘‘wrapping’’ artifacts, thus limiting their use for surgical navigation. This has to be carefully and individually assessed by the clinician for each machine used. The soft tissue discrimination of MR imaging is superior to that of CT, but on the other hand, MR images manifest geometric distortions caused by many factors. Geometrically distorted MR images can lead to errors in fiducial localization, fiducial registration, and intracranial target localization. The causes of geometric distortion in MR images are multifactorial (8). Those resulting from the machine itself include inhomogeneity of the main magnetic field (Bo), nonlinear magnetic field gradients, imperfect slice or volume selection pulses, eddy currents caused by switching of the gradients, and magnetic debris caught up in the equipment. Those resulting from the patient include magnetic field susceptibility variations, chemical shifts, flow, induced distortions, and eddy currents cause by metallic objects, including stereotaxy frames or fiducial markers (9). The correction of distortions in MR images requires attention to manufacturer design and construction, site preparation, shimming of the magnet, selection of appropriate compromises for field and gradient strength, incorporation of flowcorrected MR angiographic pulse sequences, and avoidance of chemical shifts when selecting fiducial markers. A joint task force composed of neurosurgery and radiology department representatives as well as other interested parties is necessary to properly deal with these concerns. The rectification of distortions in MR images by universally applicable computer algorithms has been described by two groups of investigators from Vanderbilt and Stanford Universities (8,10–13). Both groups advocate postprocessing of additional scans. The technique used by the Stanford University group is elegant, but limited because it requires considerable postprocessing time and is unable to

54

&

PART I: BASIC PRINCIPLES

rectify MR images of patients when standard stereotaxy frames are attached. The technique used by the Vanderbilt University group produces a significant improvement in localization accuracy by using 4 mm thick spin echo MR images ranging from 3.833 0.992 to 1.986 0.605 mm. Less than one minute of postprocessing time is required. Image quality is not adversely affected. Chemical shifts in MR images can be induced by the chemicals used in fiducial markers. This is especially significant in fats or oils, causing vitamin E capsules to be unacceptable for use as fiducials, despite their bright signal characteristics. Additional spatial distortions are created by the flow of blood in MR angiographic images, and these distortions are not corrected for by any currently employed algorithm. The degree of shift is up to 5 mm, which represents a significant caution in the routine use of these images. For CT, the greatest source of spatial inaccuracy is error in calibration of table movement between slices or of patient movement during a scan set. If the patient moves substantially between the time, it takes to obtain two slices of a scan set but does not move during scanning, no movement artifact will be present, but the anatomic volumes imaged above and below the slice of movement will not be correlated with one another as a single volume. These issues of table movement and patient movement have been dramatically ameliorated by rapid scanning with spiral acquisitions and slice reconstructions. In my experience and that of my colleagues at Vanderbilt University, the average amount of error in CT has dropped approximately from 1.0–2.0 to 0.3 mm once these new scanners are used. Very pernicious indeed is the acquisition of a volumetric CT set with the gantry angled away from 0 . When this data set is reconstructed in the computer as if the gantry angle was 0 (the baseline assumption), the volume is dramatically and irreversibly skewed with the address for each successive slice offset in relation to the gantry angle. Equally subtle and devastating is the effect of ‘‘recentering’’ the slice image during slice acquisition, something that is done merely to improve the aesthetics of the picture. This has the similar effect of offsetting slices and destroying the spatial integrity of the image volume. Every neurosurgeon who intends to employ neurodiagnostic images for surgical navigation must develop a close collaborative working relationship with colleagues in his or her institution’s radiology department to effectively communicate the concerns and requirements to everyone involved. Because the accuracy of the system is based upon the images used, these must be appropriate for the intended purpose. This never proves to be a simple, let alone an automatic transfer of data. Instead, an ongoing and close personal relationship is critical to ensure that all parties maintain quality in the use of images in a manner entirely different from that anticipated by previous clinical requirements. Furthermore, once understanding of goals and requirements and a consensus of methods have been achieved, continual careful maintenance and servicing schedules must be observed to produce a consistently reliable source of clinical information. Often factors outside the awareness and control of one party will have a radical impact on the ability of the other parties involved. For example, routine installation of a minor upgrade in one function of a PACS network software interface will suddenly render image transfer to an SNS highly problematic if steps have not been taken beforehand to rewrite the appropriate code to deal with this potential pitfall. Most MR imaging units need to be reshimmed and

CHAPTER 4: PITFALLS

&

55

retuned on a weekly or monthly basis; when used for surgical navigation data acquisition, the system requirements may be sufficiently changed so as to require that weekly or even daily adjustments be made.

REGISTRATION TECHNIQUES Registration is a method for the mapping of individual patient-specific image spaces to one another and to physical space by means of identifying corresponding points in each image and in physical space. Registration is a fundamental process underlying the ability of all SNSs to perform the task at hand, and it belongs to a different category than reformatting or rendering of images. Reformatting is the reslicing of a volumetric image data set along a new angle to approximate the orientation of another image set. Rendering is the segmentation and display of image information in such a way as to present the data set as if it were a photograph of a three-dimensional object seen from a given perspective. These latter two procedures are ways of manipulating images for display purposes and do not fundamentally affect the relationships between different images or physical space. Therefore, improvements in techniques for the pseudo-threedimensional display of rendered image data cannot in any way affect the determination of corresponding points in two different views of the same object because this registration of those images logically precedes any manipulation of their display. Alternative techniques of registration have consisted of point methods (including stereotaxy frames, fiducial markers, and anatomic landmarks), curve and surface methods, moment and principal axes methods, correlation methods, and atlas methods (14). Extrinsic point methods using implantable fiducial markers provide the fastest and most accurate form of registration and can define the level of accuracy for each patient encounter. This has made them popular among investigators, despite their minimally invasive nature and their inability to retrospectively register image data sets obtained without markers already in place. Surface-based registration has the attraction of being the only method currently available that potentially requires no special imaging and can instead use ‘‘retrospectively acquired’’ images, that is, images not intended for navigation but only for initial diagnosis (15). This economic consideration may be significant, but it does not alter the highly variable ways in which these retrospectively acquired images might have been performed and collected. Their indiscriminate use is an invitation to introduce a variety of unpredictable sources of localization error about which nothing can be done post hoc. To say that bad images are better than expensive images or no images at all is an economic consideration, but not an engineering or medical consideration. On the other hand, most practitioners who attempt registration using surface-based algorithms do use carefully obtained prospectively acquired images to suit the needs of navigation accuracy. In these settings, the limitations of the registration technique become the most significant concern. Surface-based registration techniques are less accurate and

56

&

PART I: BASIC PRINCIPLES

less reliable than are fiducial marker-based registration techniques, and it takes considerably more technical support and operator time to make them work at all. Most typically, the level of application accuracy achieved by these techniques under optimal circumstances is of the order of 3 to 10 mm (16). Some investigators have preferred to use anatomic landmarks such as the bridge of the nose, tragi, or lateral canthus for registration (17,18). These intrinsic landmarks are also readily available and allow retrospectively acquired images to be used for registration and navigation. Their successful use does, however, entail a considerable component of individual operator skill and experience. Thus, their use has been limited to those centers with skilled practitioners who prefer them to surface registration techniques or to skin-mounted fiducial markers. The ideal intrinsic anatomic fiducial markers are as small as possible to minimize the inherent definitional challenge of finding them on CT or MR images and in physical space. Very few structures satisfy these criteria; instead, different landmarks may be chosen for different scans and vice versa. Bony foramina, joints, or spicules have been used to great advantage by some for CT-based registration during skull base surgery and pedicle screw placement surgery. Extrinsic fiducial markers may be attached to soft tissue (such as the skin or scalp) or to bony tissue (such as the skull or spinous process) (19). The movement of skin, especially of the scalp during prone positioning for suboccipital craniectomies, is a variable source of error that degrades both fiducial registration error and target localization error. Also, fiducials can fall off the skin if adhesives do not stick. To overcome the inadequacies of skin-mounted fiducials, most practitioners place many markers, often five to 10, for a given case. A minimum of three noncollinear points defines a volume, any more are redundant but serve to improve the chances of adequate registration. Imaging limitations of fiducial markers must be kept in mind. Some markers such as titanium pellets or screws are useful only with CT. Some skin-mounted fiducials using hydrogel compounds do not show up well in T2-weighted spin echo or FLAIR (fluid attenuation inversion recovery) MR images, limiting their use for low-grade glioma resectional surgery. The distribution of markers is also an important consideration. Clustering of the markers is suboptimal. The markers should be spread out as far over the cranial surface as possible, especially on the contralateral side. Clustering may produce less fiducial registration error, but this is misleading because target registration error rapidly degrades farther away from the fiducials one gets. On the other hand, widely spaced fiducials are less convenient for the patient preoperatively and for the surgeon intraoperatively during registration. Obviously, markers placed in the field of the craniotomy itself will need to be removed during surgery, preventing reregistration in the event of a registration failure. Bone-anchored fiducial markers provide the most reliable and accurate method of surgical registration and navigation (19–21). Under ideal circumstances, their application accuracy equals or exceeds that of stereotaxy frame systems. Their implantation involves an invasive surgical procedure that carries potential risks, and this fact must be balanced against their ease of implantation and their benefit to the surgery being planned. Four bone-implanted fiducial markers are typically

CHAPTER 4: PITFALLS

&

57

used in a given patient to provide some redundancy to the system. The larger the imaging fiducial marker is the more accurate is the registration of images because the centroid can be better calculated. This fact must be balanced against the surgeon’s desire for a minimally invasive marker. The closer to a sphere the fiducial marker is the easier are the calculations for identifying the centroid of the marker. Any shape known a priori can be used satisfactorily. Infection or local trauma has not been a significant observation to date. Perhaps, the greatest risk from fiducial markers is their potential for contributing to an uncritical attitude regarding the results of a given registration value intraoperatively should that value prove misleading. For the most part, however, fiducial markers appear to increase the level of reliability of SNSs.

COMPUTERS AND SOFTWARE INTERFACES Trifles make perfection, and perfection is no trifle. —Michelangelo, Lacon The clinically relevant display of registered imaging data to the surgeon requires appropriate imaging data input, computer hardware and software integration, and video display capabilities. Rapid advances in this field have produced a kaleidoscopic pattern of rapidly evolving attempted solutions to these system requirements. Because medicine is a ‘‘niche’’ market for the manufacturers involved, this field is largely driven by external forces of technology development and transfer. The anatomic detail provided by thin-slice gradient echo-planar MR images mapped onto the information derived from functional MR imaging or positron emission tomography studies, perhaps in combination with digitized stereotactic atlases, shows promise for optimal definition of these target volumes. The greatest pitfalls in the use of SNSs have involved the inability to translate a given scanner’s file format or header information into the form understood by the navigation system’s computer. Although DICOM3 compatibility promised by the manufacturers of scanners and PACS systems has the potential to clear this unnecessarily painful interface incompatibility thicket, it will not make these problems disappear altogether in the near future. Working in collaboration with radiologists and their technicians and corporate representatives, the surgeon must be prepared to expend a significant amount of time and capital to provide a solution for clinical use and maintenance. Once images have been successfully entered into the system, the surgeon must familiarize himself with the software architecture and menu structure of the navigation software. This must be done systematically and ‘‘off-line,’’ before any patient is operated on. The operating room during a surgical procedure is an unforgiving environment in which to learn nuances of menudriven program structure and function. If at all possible, the presence of a dedicated computer-literate technician in the operating room will reinforce the utility and benefit of these systems for the surgeon.

58

&

PART I: BASIC PRINCIPLES

The software used may have varying levels of robustness, transparency, or flexibility for the clinical situation at hand. Preoperative treatment planning and simulations can often speed the actual surgery along by offering those involved the opportunity to rehearse the desired procedures and anticipate common distractions or complications. Every system has its own dedicated computer hardware with its own peculiarities, requirements, and limitations. These must be appreciated to maintain functionality in the operating room. An uninterruptible power supply should be standard equipment in each system to prevent the loss of registration data in the event of disconnection and power loss. Although most hardware is well shielded against radio frequency currents, the simple physical assault of being bumped, dropped, or turned upside down must be anticipated. The violation of operating field sterility is always a potential hazard with large computer monitors situated nearby. The two-dimensional video display of three-dimensional information introduces an inevitable layer of graphics interface distortion and misrepresentation of reality. Each surgeon must determine which type of image display is the most meaningful and least confusing for a given operation. Triplanar orthogonal views of anatomy are extremely helpful in orienting the surgeon during complex surgery. Oblique views have limited benefit for craniotomies to resect tumors, although they can be very helpful during stereotactic biopsies and pedicle screw placements. Rendered information is of some benefit during cortical localization and superficial resective surgery, but is less helpful when dealing with subcortical or intraventricular anatomy. The solid tumor tissue component of a malignant brain tumor is often best appreciated on contrast-enhanced CT, although it is clearly seen but occasionally over-reported on contrasted T1-weighted MR images. Rendered outlines of lesions seen through a microscopic navigation system are subject to misregistration and are confusing to some surgeons. Ultimately, when a disagreement arises between the graphics display and what the surgeon visualizes directly, the surgical procedure should be stopped for a moment of reflection and analysis to resolve the conflict. Although the navigation system must be treated skeptically at times, the surgeon must also reassess how much has actually been accomplished at any given point during the surgery in light of conflicting information. This represents a critical moment during the operation, demanding the highest levels of concentration and using the surgeon’s knowledge base and judgment for correlation, skepticism, and resolution.

INTERACTIVE LOCALIZATION DEVICES AND INTRAOPERATIVE USE A fool with a tool is a fool. —Aphorism attributed to Lars Leksell An interactive localization device is necessary to define the spatial coordinates of physical space in the operating room. Alternative technologies used to date may be categorized as either tethered or untethered systems. Tethered systems are

CHAPTER 4: PITFALLS

&

59

conceptually and technically simpler from an engineering point of view, but have the disadvantages of drag and tethering of the device in the surgeon’s hand. Untethered devices are more acceptable as surgical instruments because they are less intrusive and do not have physical links or tethers, but they introduce a number of clinical and engineering concerns such as line-of-sight issues. The individual surgeon must evaluate which system is best suited to the operative procedure planned and the clinical setting available. Tethered interactive localization systems include stereotaxy frame systems (6,22), active robotic arms (23), passive localization arms with joint-position sensors using potentiometers or optical encoders (24), and intraoperative CT or MR imaging scanners. Non-tethered systems include spark-gap emitters with microphone triangulation as sonic localizers, infrared light emitting diode (LED), emitters with optical linear charge-coupled device (LCCD) camera, triangulation arrays as optical localizers, passive optical motion detectors using shaped light patterns, magnetic field distortion sensors, and gravitometers and inertial movement sensors. The use of a passive localization arm for neurosurgical guidance was first described by Watanabe (25). The neuronavigator consisted of a personal computer programmed in C, an in1age display monitor, and an articulated arm with six joints, each of which was equipped with a 50 kO potentiometer. The arm endpoint position was determined by digitizing the angles at each of the arm joints and inserting these into the forward kinematic equation. This elegant system had two major disadvantages. First, the potentiometers used for angle-position detectors in each of the arm joints were not finely incremented and allowed for only a coarse measurement of angular displacement. The linearity of potentiometers was also inconstant, introducing fluctuations into the measurement of the angles. The second disadvantage of the system was its limited computational abilities in supporting the video display of medical images. The ISG Viewing Wand system (ISG Technologies, Toronto, Canada) has been employed by a number of surgeons. The interactive localization device for the ISG Viewing Wand system is a passive localization arm (FARO Medical Technologies, Miami, Florida, U.S.A.). The ergonomics of this arm are praiseworthy, although it has a tendency toward backlash and forward drag at times. This sixjointed arm employs electrogoniometers for angular detectors, thus limiting its accuracy and precision. Guthrie and Adler (26) reported on an evolving series of passive localization arms for measuring angular joint displacement based on optical encoder technology. Earlier versions of localization arms from this group possessed five articulating joints with low-resolution optical encoders. To achieve higher levels of localization accuracy for more recent versions of the arm, higher-resolution optical encoders have been encased in six articulated joints. Elegant counterbalancing of the joints of these arms has been the hallmark of the mechanical construction displayed by this group; the arms ‘‘float’’ much like a Calder mobile, with occasional backlash. In 1986, Roberts et al. (27) first described microscope-based frameless stereotaxy. Spark-gap ultrasonic emitters were attached to an operating microscope in a known geometric configuration. Sound detectors for these ultrasonic noises were arrayed throughout the operating room. The combination of these ultrasonic

60

&

PART I: BASIC PRINCIPLES

emitters and their detectors created a three-dimensionally registered space through which the microscope moved. The focal length of the optics of the microscope was known and, accordingly, the microscope’s focal point became a virtual end effector for the microscope/spark-gap emitter assembly. In this way, the microscope could be used as an optical interactive localization device for patient landmarks. A fundamental limitation of this system is the use of spark-gap emitters. Because the speed of sound in air is highly temperature dependent and the distances between ultrasonic emitters and their detectors are long, the temperature fluctuations commonly encountered in surgical suites cause variations in the speed of sound sufficient to produce potentially large errors in target position location. The pathways between detectors and emitters must be kept clear to prevent signal displacement by intervening structures or materials. Echoes from the multitude of surfaces in the operating suite produce significant interference with surgical localization. Considerable work has been carried out by this Dartmouth University group to correct for temperature fluctuations, echoes, and other sources of error in the system. The error in locating the focal point of the operating microscope in a recent clinical series of 17 patients was 6.33 3.37 mm. Interestingly, this system has been employed in a series of seven patients undergoing conventional neurosurgical procedures on the lumbosacral spine for herniated discs or spondylolisthesis. The mean accuracy of locating the operating microscope’s focal point was 6.04 mm. At the level of the disc space, however, the combination of all sources of error coupled with mobility of the soft tissue resulted in an average error of 28.81 mm. Barnett and colleagues (28–32) incorporated spark-gap ultrasonic emitter technology with a hand-held interactive localization device (Picker International, Highland Heights, Ohio, U.S.A.). This pointing device carries the spark-gap emitters rather than coupling the emitters to the operating microscope. The repetitive (50 kHz) ultrasonic pulses produced by the interactive localization device are detected by three or more microphone/emitter units comprising a portable detector/calibration array that is mounted to the side rails of a standard operating table. The positioning of the detector/calibration array was believed to be superior to ceiling-mounted or wall-mounted placement; the shorter triangulation distances and optimal slant ranges measurably improved localization accuracy. This system demonstrates a fair degree of sensitivity to angulation of the interactive localization device with respect to the detector/calibration array. Careful positioning of the interactive localization device is necessary for proper intraoperative localization. Software algorithms for temperature and humidity fluctuation effects are critical to the functionality of the system. Several groups have investigated the use of infrared LEDs as an alternative form of optical localization (19,33,34). By having two or more infrared LEDs attached to a surgical instrument, the position in space of that instrument can be tracked over time by a triangulation technique employing three or more ‘‘cameras.’’ These cameras are specialized 2000-to-4000-element LCCDs with cylindrical lenses that focus light from the infrared LEDs onto the individual elements of the array. Three of these one-dimensional arrays are precisely oriented on an aluminum cross beam so that their overlapping visual fields define any point in threedimensional space.

CHAPTER 4: PITFALLS

&

61

These optical triangulation techniques are highly accurate and precise, surprisingly robust within the noisy operating theater environment, and extremely flexible in their use. Line-of-sight considerations apply in this situation much as they do with sonic digitizer triangulation methods. Unlike sonic digitizers, infrared LED arrays are insensitive to temperature or humidity fluctuations. Interference from surgical light sources has not been proved to be a problem. The infrared wavelength of their emissions is not disruptive to surgery (5). Bucholz and Smith (33) employed an array of two infrared LEDs attached to a surgical bipolar cautery forceps. Maciunas et al. (19) developed an alternative manifestation of this optical triangulation technique. The position of the interactive localization device is tracked by a precision-milled, three-camera LCCD array and embedded computer program with knowledge-based software using centerfinding algorithms to better detect the infrared LEDs. The interactive localization device is studded with an array of multiple infrared LEDs to increase the robustness and accuracy of localization. In this way, ‘‘blind spots’’ are more easily avoided. Heilbrun et al. (36) have proposed a technique of surgical guidance employing machine vision in which the three-dimensional position of an object in space is determined from its differential position on two-dimensional video images viewed from two different angles. Two video cameras situated approximately 1 m apart are aimed at the surgical workspace. A ‘‘video localizer,’’ consisting of eight fiducial objects set in a box-like configuration of known dimensions, is placed within the field of view of the two video cameras. A ‘‘photogrammetric’’ projection algorithm defines the three-dimensional coordinates of the eight fiducial objects. Subsequently, the position and field of view of the video cameras remain unchanged. Thus, the three-dimensional position of any object in the surgical workspace of their field of view can be determined and tracked. Surgical instruments may be tracked within the surgical space of the video cameras by affixing fiducial markers to the surgical instruments in a known geometric configuration. The advantages of this method are its flexibility, its potential to track standard surgical instruments as stereotactic points with minimal modification, and its reliance on readily available video technology. It remains sensitive to line-of-sight considerations, the need for the video cameras to maintain their relationship with the preoperatively defined surgical space, limitations to the resolution of currently available video technology, and the computationally intensive nature of intraoperative registration. Many investigators have employed a magnetic field guidance system manufactured by Polhemus Navigation Sciences (Colchester, Vermont, U.S.A.) for directing endoscopic visualization and dissection. A magnetic field is established with the use of a transmitter antenna near the patient’s head, and its strength is measured by a receiving antenna mounted on the surgical instrument. Because magnetic field strength diminishes as a function of the distance between the transmitter and the receiving antenna, the tip of the receiving instrument is measured in x, y, and z planes of space in relation to the repetitive direct, current pulse magnetic field in x, y, and z hemispheres. This process is carried out at 100 Hz. Magnetic field guidance has the attraction of being mechanically simple and inexpensive. Unfortunately, it has been proved to be less accurate than other

62

&

PART I: BASIC PRINCIPLES

technologies in clinical settings and is susceptible to both static and active field distortion in the surgical suite environment because of the presence of common metals as well as electromagnetic field interference. The Regulus navigation system (Stereotactic Medical Systems, Rochester, Minnesota, U.S.A.) employs software with algorithms that attempt to compensate for the static magnetic field inhomogeneities to minimize the drift and inaccuracies encountered in the clinical use of magnetic localization systems. Regardless of the interactive localization system used, attention must be paid to the degradation of registration information through system drift, head movement, and other sources of noise. Other than optical encoder-based joint-position sensors in linked passive arms and active, infrared-emitting diode-based optical triangulation systems, all other technologies to date have demonstrated some degree of drift of positional information over time. Head movement in the pinion head holder has been identified as a very real source of positional drift. Optical systems have incorporated a reference emitter to track head movement and readjust the registration accordingly; for this to function properly, the reference emitter must be attached to the head properly, something that is not always done. In spinal surgery, the movement of anatomic structures during surgical manipulation is so significant that it is virtually imperative to consider a reference emitter for each mobile element to achieve maximal accuracy. Whenever any interactive localization device is used, some error with its use should be anticipated. As a result, there is inevitable error accumulating during fiducial localization and registration, which is compounded during target localization. The surgeon can therefore assume that the target localization accuracy of any system will always be worse than the displayed fiducial registration accuracy due to this ‘‘double sampling error’’ and can apply the requisite caution in relying on these systems.

INTEGRATION OF REAL-TIME DATA It’s like de´ja` vu all over again. —Yogi Berra Despite the registration of image to physical space, digital scan information remains historical data and is subject to becoming outdated during the course of the surgical manipulation of tissues. The feedback loop of navigation can be closed by incorporating real-time intraoperative imaging and monitoring; these data are used to refine and modify the patient-specific maps being used for surgical guidance. The use of historically obtained (preoperative) imaging sets is further augmented by intraoperative updating and reregistration with real-time microendoscopy, spatially resolved electrophysiology and cortical mapping techniques, ultrasonography, and interventional CT and MR imaging. By digitizing the video output from intraoperative visualization techniques (ultrasonography, endoscopy, tomography, and electrophysiologic recordings), these images may be treated as simply another source of spatially registered medical information (5,37,38).

CHAPTER 4: PITFALLS

&

63

Preliminary steps are being taken to incorporate this information into surgical guidance databases. Microscopes pose a considerable challenge to engineers attempting to produce a surgically accurate and useful device. Despite claims to the contrary, every surgeon must realize that these devices involve a real learning curve if used properly. The depth of field of microscopes, misalignments of their optical axis, miscalibrations of image injection systems, and visually confusing or deceiving perspective distortions must be taken into consideration with every clinical use. Because visually augmented reality can be most compelling, even when it is misregistered or incorrect, interpretation of this information as it is presented dictates extreme caution. Individual surgeons have varying degrees of stereoscopic vision and three-dimensional visualization skills, and this may affect how much can be gained with the use of these systems. Some microscopes assume right-eye dominance in their displays, limiting their use to those of us who are left-eye dominant. Video microscopy has been widely discussed as a means of integrating digital tomographic information and real-time video. Nevertheless, this system awaits the introduction of higher-resolution LCCD arrays before it can replace optical microscopy for surgery; at present, it represents nothing more than a technical curiosity. In the meantime, spatially registered endoscopy is becoming incorporated into routine clinical practice. The use of angled eyepieces can be visually confusing at times, and no good solution for registering flexible endoscopes is currently available. Slippage of the infrared LED attachments or miscalibration of the device can lead to erroneous positional data. Intraoperative tomography offers the promise of high-resolution digital volumetric imaging data for continuous real-time updating of the surgical field. Intraoperative CT has been employed extensively, and intraoperative MR image scanning has been introduced into neurosurgery. The potential benefits of these devices include reregistration of preoperative databases when structures shift during surgery, visual assessment of the adequacy of surgical resection, and minimally invasive heating of selected tissues under imaging control. To date, the restrictive physical environment of these scanners has somewhat impeded their surgical utility. The considerable expenditures for the hardware and facilities of such systems have raised concerns about their cost effectiveness in any but high-volume clinical settings (9).

TISSUE DISPLACEMENT All current medical registration techniques involve rigid body assumptions. This allows the computational load to be manageable and leads to therapeutically appropriate interfaces. With any surgical intervention, however, some movement of soft tissue is inevitable. For some time, this has been a recognized limitation in the indiscriminate reliance on these navigation systems. Recent quantitation of this effect during craniotomies has indicated superficial brain shifts approaching 1 cm intraoperatively (40). Deeper structures, especially those along the cranial base, along the falx, or near the deep gray nuclei, in fact, shift less than do the

64

&

PART I: BASIC PRINCIPLES

superficial overlying structures. This improves the reliability of navigation the deeper one goes, permitting these systems to be used safely and effectively. Nonetheless, this remains an issue of concern not only academically but also for every clinical application. First and foremost, the presence of tissue distortions mandates that the highest level of accuracy and freedom from drift be maintained for any SNS. Otherwise, the accumulation of errors threatens to engulf any effort at localization in positional noise. Second, certain precautions are necessary. It appears that the greatest shift is in the direction of gravity. Thus, in a given proposed surgical trajectory, the critical structures of interest may prove to shift deeper along the trajectory rather than away in an unpredictable manner. Minimizing osmotic diuretics does not, at present, seem to be of any significant benefit, and standard surgical considerations should not be jeopardized. The venting of ventricular cerebrospinal fluid can produce unpredictable shifts in superficial structures, and should be avoided until after the primary approach has been accomplished. Debulking of tumors may be modified by minimizing the ‘‘inside to outside’’ approach to debulking, making an effort to define the outer margins of intrinsic tumors with subsequent debulking carried out at the earliest opportunity (2,34,41,42). Some authors have described ‘‘refreshing’’ the registration by reregistering to shifted cortical or subcortical anatomic fiducials to keep up with these shifts, but this has the potential for transmitting extremely confusing information to the surgeon. Until more systematic protocols and software for accomplishing this procedure in a safe manner have been developed, it remains of limited benefit. Research is under way to integrate intraoperative imaging studies such as ultrasonography, CT, and MR imaging to warp preoperative images to a new registration based on tissue displacement. Although highly promising for the future, no reliable method is yet available for clinical use.

ROBOTICS Currently, there is considerable speculation as to the applicability of robotic control technologies and the employment of virtual reality computer interfaces to integrate the various imaging databases obtained for a given patient. This automation of the surgical process will likely increase rather than decrease the involvement of the neurosurgeon in ensuring oversight, calibration, and knowledge-based feedback. JUDGMENT AND CLINICAL EXPERIENCE First, do no harm. —Hippocrates As in all other aspects of neurosurgery, the clinician is called on to definitively assess the conflicting or confusing sets of information about a patient’s status and

CHAPTER 4: PITFALLS

&

65

diagnosis. Rather than uncritically accepting whatever spatial and diagnostic information is provided, the surgeon must continue to exercise surgical judgment based on training and clinical experience to achieve optimal clinical results. Although a computer can effectively handle vast quantities of quantitative imaging data, it does not at present have the knowledge-based intelligence to appropriately weigh the evidence, as it is presented and to distinguish between disease and healthy tissue or between surgical danger and safe passage. This weighing of factual information will remain the greatest responsibility (and generate the greatest excitement) in the clinical decision-making process for the surgeon. Numerous limitations and pitfalls become apparent in the use of surgical navigation devices. Nonetheless, these are exceedingly useful devices and benefit neurosurgery on a daily basis. Given their deficiencies, how can they be so useful? There are a number of ameliorating factors that mitigate the detrimental aspects of this technology. First and foremost, most of us have the pleasure of spending our lives practicing in the median range of the universal Gaussian distribution of error rather than at the 99th percentile limits of confidence. Accordingly, by definition, most of the time these systems will perform better than they do in the worst-case scenario. In any event, most of the lesions that we approach are larger than the limits of resolution for these systems, and the overconfigured safety cushion provided by tight tolerances for accuracy pays off in reliable performance in most clinical situations. When a primary brain tumor is resected, the tumor architecture and biology provide an indistinct border to the target lesion, causing most surgeons to bracket the lesion coordinates for safety’s sake. During a craniotomy, direct visualization of the lesion provides real-time feedback to update the registration intraoperatively. During functional stereotactic procedures such as thalamotomies or pallidotomies, electrophysiologic feedback is ultimately the technique for reregistering the appropriate anatomy and successfully completing the operation. The neurosurgeon must bring to bear all the available tools for localization of cerebral function and lesional anatomy to accurately distinguish them (2,7,32,34,40–43). And, in spite of pride, in erring Reason’s spite, one truth is clear, whatever is, is right. —Pope, Essay on Man, I

66

&

PART I: BASIC PRINCIPLES

REFERENCES 1. Galloway RL, Maciunas RJ. Stereotactic neurosurgery. Crit Rev Biomed Eng 1990; 18:207–233. 2. Maciunas RJ, ed. Clinical Frontiers of Interactive Image-Guided Neurosurgery. Neurosurg Clin N Am 1996. 3. Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993. 4. Maciunas RJ, Galloway RL, Latimer JW. The application accuracy of stereotactic frames. Neurosurgery 1994; 35:682–695. 5. Galloway RL Jr, Berger MS, Bass WA, et al. Registered intraoperative information: Electrophysiology, ultrasound, and endoscopy. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:247–258. 6. Kelly PI. Computer-assisted stereotaxis: new approaches for the management of intracranial intra-axial tumors. Neurology 1986; 36:535–541. 7. Kelly PI, Daumas-Duport C, Scheithauer B, et al. Stereotactic histologic correlation of computer tomography and magnetic resonance imaging-defined abnormalities in patients with glial neoplasms. Mayo Clin Proc 1987; 62:450–459. 8. Maciunas RJ, Fitzpatrick JM, Gadamsetty S, et al. A universal method for geometrical correction of magnetic resonance images for stereotactic neurosurgery. Stereotact Funct Neurosurg 1996; 66:137–140. 9. Plante E, Turkstra L. Sources of error in quantitative analysis of MRI scans. Magn Reson Imaging 1991; 9:589–595. 10. Chang H, Fitzpatrick JM. A technique for accurate magnetic resonance imaging in the presence of field in homogeneities. IEEE Trans Med Imaging 1992; 11: 319–320. 11. Dong S, Fitzpatrick JM, Maciunas RJ. Rectification of distortion in MRI for stereotaxy. In: Proceedings of the Fifth Annual IEEE Symposium on Computer-Based Medical Systems, 1992:181–189. 12. Maurer CM Jr, Aboutanos GB, Dawant BM, et al. Effect of geometrical distortion correction in MR on image registration accuracy. J Comput Assist Tomogr 1996; 20:666–679. 13. Sumanaweera TS, Glover GH, Binford TO, et al. MR susceptibility misregistration correction. IEEE Trans Med Imaging 1993; 12:251–259. 14. Maurer CM, Fitzpatrick JM. A review of medical image registration. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:17–44. 15. Pelizzari CA, Chen GTY, Spelbring DR, et al. Accurate three-dimensional registration of CT, PET, and/or MR images of the brain. J Comput Assist Tomogr 1989; 13:20–26. 16. West J, Fitzpatrick JM, Wang MY, et al. Comparison and evaluation of retrospective intermodality image registration techniques. J Comput Assist Tomogr 1997; 21: 554–566.

CHAPTER 4: PITFALLS

&

67

17. Adler JR Jr. Image-based frameless stereotactic radiosurgery. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:81–89. 18. Ge Y, Fitzpatrick JM, Votaw JR, Gadamsetty S, et al. Retrospective registration of PET and MR brain images: an algorithm and its stereotactic validation. J Comput Assist Tomogr 1994; 18:800–810. 19. Maciunas RJ, Fitzpatrick JM, Galloway RL, et al. Beyond stereotaxy: extreme levels of application accuracy are provided by implantable fiducial markers for interactive image-guided neurosurgery. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:259–270. 20. Mandava VR, Fitzpatrick JM Jr, Maurer CR, et al. Registration of multimodal volume head images via attached markers. In: Proceedings of the SPIE Medical Imaging Conference, 1992:271–282. 21. Wang MY, Maurer CR, Fitzpatrick JM, et al. A knowledge-based technique for localizing externally attached markers in MR and CT volume images of the head. Proc Ann Int Conference IEEE Eng Med Biol Soc 1993; 15:120–121. 22. Shelden C, McGann G, Jacques S, et al. Development of a computerized microstereotaxic method for localization and removal of minute CHS lesions under direct 3-D vision [technical report]. J Neurosurg 1980; 52:21–27. 23. Drake JM, Joy M, Goldenberg A, et al. Computer- and robot-assisted resection of thalamic astrocytomas in children. Neurosurgery 1991; 29:27–33. 24. Maciunas RJ, Galloway RL, Fitzpatrick JM, et al. A universal system for interactive image-directed neurosurgery. Stereotact Funct Neurosurg 1992; 58:108–113. 25. Watanabe E. The neuronavigator: a potentiometer-based localization arm system. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:135–148. 26. Guthrie BL, Adler JR Jr. Computer-assisted preoperative planning, interactive surgery, and frameless stereotaxy. Clin Neurosurg 1992; 38:112–131. 27. Roberts DW, Strohbehn JW, Hatch JF, et al. A frameless stereotaxic integration of computerized tomographic imaging and the operating microscope. J Neurosurg 1986; 65:545–549. 28. Barnett GH. Surgical management of convexity and falcine meningiomas using interactive image-guided surgery systems. Neurosurg Clin North Am 1996; 7:274–284. 29. Barnett GR, Kormos DW, Steiner CP, et al. Frameless stereotaxy using a sonic digitizing wand: development and adaptation to the Picker Vistar medical imaging system. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:113–120. 30. Barnett GH, Kormos DW, Steiner CP, et al. Intraoperative localization using an armless, frameless stereotactic wand. J Neurosurg 1993; 78:510–514. 31. Barnett GH, Kormos DW, Steiner CP, et al. Use of a frameless, armless wand for brain tumor localization with two-dimensional and three-dimensional neuroimaging. Neurosurgery 1993; 33:674–678.

68

&

PART I: BASIC PRINCIPLES

32. Gomez H, Barnett GR, Palmer J. Contemporary stereotactic neurosurgery at the Cleveland Clinic. Review of 501 consecutive cases. Cleve Clin Med 1993; 60: 399–410. 33. Bucholz RD, Smith KR. A comparison of sonic digitizers versus light-emitting diode-based localization. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:179–200. 34. Maciunas RJ, Berger MS, Copeland B, et al. A technique for interactive image-guided neurosurgical intervention in primary brain tumors. Neurosurg Clin N Am 1997; 7:245–266. 35. Horstmann GA, Reinhardt HF. Micro-stereometry: a frame less computerized navigating system for open microsurgery. Comput Med Imaging Graph 1994; 18: 229–233. 36. Heilbrun MP, McDonald P, Wiker C, et al. Stereotactic localization and guidance using a machine vision technique. Stereotact Funct Neurosurg 1992; 58:94–98. 37. Berger MS. Ultrasound-guided stereotaxic biopsy using a new apparatus. J Neurosurg 1986; 65:550–554. 38. Berger MS, Kinkaid J, Ojemann GA, et al. Brain mapping techniques to maximize resection, safety, and seizure control in children with brain tumors. Neurosurgery 1989; 25:786–792. 39. Moriarty TM, Kikinis R, Jolesz FA, et al. Magnetic resonance imaging therapy: Intraoperative MR imaging. Neurosurg Clin N Am 1996; 7:323–331. 40. Hill DLG, Maurer CL Jr, Wang MY, et al. Estimation of intraoperative brain surface movement. In: Troccaz J, Grimson E, Mosges R, eds. CYR Med-MRCAS ’97. Berlin: Springer-Verlag, 1997:449–458. 41. Berger MS, Deliganis A V, Dobbins J, et al. The effect of extent of resection on recurrence in patients with low-grade cerebral hemisphere gliomas. Cancer 1994; 74:1784–1791. 42. Galloway RL, Maciunas RJ, Failinger AL. Factors affecting perceived tumor volumes in magnetic resonance imaging. Ann Biomed Eng 1993; 21:367–375. 43. Olivier A, Germano I, Cukiert A, et al. Frameless stereotaxy for surgery of the epilepsies: Preliminary experience. J Neurosurg 1994; 81:629–633.

Part Two PART II

Technologic Applications

5

BRAIN BIOPSY AND RELATED PROCEDURES Vitaly Siomin Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Gene H. Barnett Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION Stereotactic brain biopsy (SBB) is a technique of undisputed value in today’s neurosurgery. It provides a straightforward and highly precise means of obtaining diagnostic tissue within the skull or brain with minimal damage to the surrounding structures. Although there has been remarkable technical progress in this area, the basic principle of advancing an instrument under geometric and anatomic guidance is based on the principles established by Horsley and Clarke (1) and their first stereotactic frame for animal experiments in 1908. These achievements led the way to the development of various devices for human stereotaxy, referenced to X rays of anatomic structures as visualized using pneumoencephalography (2,3). The advent of computerized tomography (CT) in the early 1970s, however, marked the beginning of the modern era in brain biopsy. SBB using reference frames guided by CT scans (4,5), and later magnetic resonance imaging (MRI) (6), became commonplace. Another major accomplishment in the last two decades was the development of surgical navigation systems (SNSs) (7,8). A growing number of reports demonstrate that image-guided frameless brain biopsy is gaining popularity among neurosurgeons and may, eventually, displace frame-based technologies. This chapter reviews the principles of different techniques of SBB with an emphasis on ‘‘frameless stereotaxy’’ using SNSs.

71

72

&

PART II: TECHNOLOGIC APPLICATIONS

SUITABILITY FOR BRAIN BIOPSY As stated by Kulkarni and Bernstein (9), the ideal lesion for SBB would be the one that is ‘‘too deep, too small, too diffuse, multiple, and located in the eloquent brain’’ for open surgery. Lesions where surgical debulking is not necessary or beneficial may be appropriately accessed using SBB. Also, significant medical illness may render a patient, who would otherwise benefit from excisional biopsy, more suitable for SBB. In some situations where the clinical diagnosis of a brain lesion is ‘‘obvious’’ based on the imaging and/or clinical presentation, confirmation of the diagnosis by SBB may be less appealing. This approach may be particularly justified in lesions such as brainstem gliomas, where modern imaging is considered sufficient for diagnosis, or in patients with multiple ring-enhancing lesions and a history of cancer that tends to metastasize to the brain. It is important to recognize, however, that in up to a quarter of cases, SBB yields totally unexpected diagnoses (e.g., demyelination, inflammatory lesions, vascular events, etc.) that dramatically changes treatment strategies (10,11). Indications for SBB as they relate to specific pathologies are beyond the scope of this chapter but can be accessed in the literature.

IMAGE-GUIDED FRAME-BASED STEREOTACTIC BRAIN BIOPSY Frame-based stereotactic biopsy utilizes spatial information encoded into imaging studies, then mathematically extracted and applied to stereotactic frames. While certain systems allow simple derivation of the target in stereotactic space (e.g., Leksell stereotactic frame, Elekta Instrument AB, Stockholm, Sweden) (Fig. 1), others require special computer programs allowing surgeons to match the imaging with the frame, and all may benefit from software that allows planning and visualization of the target and trajectory on neuroimaging studies (12–14). The following is a general description of the procedure, although there are many variations of technique. Frame-based biopsy requires application of a rigid frame directly to a patients’ skull, generally using screw pins, under local anesthesia. The position of the frame remains unchanged throughout imaging (be it CT, MR, PET, or any other imaging) and the procedure itself. It, thus, literally serves as a ‘‘frame of reference.’’ Prior to scanning, some type of adapter is secured to the frame to spatially encode the images. This adapter then produces reference marks, or ‘‘fiducials’’ on each image slice. The pattern of fiducials can be decoded and the coordinates (often Cartesian x, y, and z) of the target (vis-a`-vis the rigid frame) are calculated. To direct the procedure, the guidance jig (target centered, multi-arc, ball and socket) attached to the frame is used. Either a twist drill or a burr hole may be used to penetrate the skull. After perforating the dura with the monopolar cautery, the biopsy instrument is inserted to the previously defined depth. For side-cutting biopsy probes, the sample is obtained by applying gentle suction (1–2 cm3) on a syringe affixed to the end of the biopsy needle, while it is rotated within an outer cannula.

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

73

Figure 1 Leksell stereotactic frame (Elekta Instrument AB, Stockholm, Sweden).

The sample is then sent to pathology for examination. If the tissue is deemed suitable and sufficient for diagnosis, the procedure is over, and the instrument withdrawn.

IMAGE-GUIDED FRAMELESS STEREOTACTIC BRAIN BIOPSY Various techniques for frameless stereotactic biopsy have been described. They may be divided into two main categories: SBB with the use of SNS (15–18) and SBB using intraoperative MRI (19–21). The authors have used both techniques at the Cleveland Clinic Foundation. Brain biopsy with the use of SNS is based on preoperatively obtained images that are then transferred to the planning station. Most of the modern navigation

74

&

PART II: TECHNOLOGIC APPLICATIONS

systems have image fusion capabilities. In addition to the regular MRI and CT scans, MR spectroscopic images, PET scan, MR venography, functional MRI, etc. may be incorporated for planning or navigation in the operating room as necessary. Our approach to SBB begins with application of special skin markers applied to the patient’s head prior to imaging. They serve as reference points for subsequent registration (analogous to the ‘‘fiducials’’ in frame-based stereotaxy). Skull-mounted fiducials may be used to enhance accuracy (22,23). The procedure may be performed under local or general anesthesia, the latter preferred by the senior author. The head is usually secured in a three-point fixation device to which a dynamic reference frame (DRF) is attached to allow the computer to track and compensate for head or camera movements (Fig. 2). The operating principles and registration processes of SNS are addressed elsewhere in this book (chap. 2) and surgeons should be familiar with strategies to maximize accuracy (Table 1). A target and, often, an entry point are selected to optimize diagnostic yield and minimize risk of adverse outcomes such as hemorrhage. Using the principles of target and trajectory guidance (chap. 3), the probe is aligned with the preoperative plan and the scalp marked, shaved, and prepared with a povodone iodine solution. A sterile eye drape is placed with the opening incorporating the planned trajectory. The authors use the Voyager surgical navigation system (Z-KAT, Inc., Hollywood, Florida, U.S.A.), which incorporates target and trajectory guidance (chap. 3). Its biopsy hardware was devised in such a way that it could be readily adapted to virtually any navigation system (15). Orientation and fixation of the probe and biopsy needle are provided by an extra-large Greenberg instrument holder (Codman and Schurtleiff, Randolph, Massachusetts, U.S.A.) fixed to the head clamp. A metal guide block machined to hold common stereotactic reducing tubes is secured in the jaws of the instrument holder (Fig. 3). Following the usual process of positioning and registration, the navigation wand is positioned in the guide block in such a way that the three colored circles on the screen of the planning station, representing the tip of the wand, the entry point, and the target, are concentric, indicating that the wand is aligned with the planned trajectory. The instrument holder is then tightened securely and is subsequently used to hold the twist drill guide. Drilling is performed through a short skin incision using a 4.5 mm twist drill (Radionics, Burlington, Massachusetts, U.S.A.) (Fig. 4). Although these procedures may be performed through a larger bur hole, we have not found this necessary and had to only convert a twist drill hole to bur hole (to secure bleeding) in a few cases out of several hundreds. The wand tip is then inserted through the guide block again and into the twist drill hole. This maneuver is used to check accuracy and, as necessary, correct the trajectory after drilling, and to record the SNS-generated distance from the probe tip to the target (depth). The probe tip length (PTL—tip to shank) is a constant for any given probe and should be recorded (100 mm for these biopsy probes). The wand offset (WO) between the shank and the guide block is recorded (Fig. 5). The wand is removed and the reducing tube for the biopsy probe is placed. The distance from the top of the guide block to the top of the reducing tube is recorded (reducing tube offset—RTO) (Fig. 6). The formula used to

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

75

Figure 2 The head of the patient is secured in a three-point fixation device. The skin markers (fiducials) are seen on the forehead. The dynamic reference frame (DRF) is attached on the right side of the head fixation device. The registration is carried out by touching the fiducials with the tip of the wand held by the surgeon.

determine the working length of the biopsy needle (i.e., distance from the top of the reducing tube to the target) is Working length ¼ Depth þ PTL þ RTO WO Perforation of the dura, insertion of the biopsy needle, and aspiration of the specimen are performed as in frame-based procedures. Similar methodology can

76

&

PART II: TECHNOLOGIC APPLICATIONS

Table 1 Tips for More Accurate Brain Biopsy Using SNS Standard placement of fiducials for all intracranial procedures (e.g., four markers on the forehead, two in the anterior parietal areas, two over the mastoid processes) and one at the vertex ensure adequate coverage for most procedures and eliminate unnecessary improvisations Avoid placement of fiducials on the easily movable areas (e.g., cheeks, back of the head, neck, etc.) Affix fiducials directly to the skin, and shave, if necessary. Secure with cyanoacrylate. Mark the fiducial sites with ink for accurate replacement, should they fall off Make sure that the fiducials are located far apart from each other Place the dynamic reference frame as close to the patient as possible, but avoid crowding the operative field Patient positioning should be carried out prior to rigid fixation to insure a good ‘‘line-ofsight’’ between the emitter on the head and the camera Avoid placement of the head clamp pins close to the fiducials. This can lead to scalp distortion and fiducial movement Hold the wand with both hands at registration to avoid excessive pressure and displacement of fiducials The head, once it is registered can be moved during surgery (e.g., positioned on the pillow as opposed to the rigid holder), provided that the head clamp and the dynamic reference frame remain a rigid unit Source: From Ref. 24.

Figure 3 A metal guide block machined to hold common stereotactic reducing tubes or wand is secured in the jaws of the instrument holder.

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

77

Figure 4 Drilling is performed through the instrument holder secured in the guide block.

be used when placing ventricular or tumor cyst catheters, electrodes for recording, or stimulation and radiation implant catheters or seeds (15,25–27). Other Methodologies Alternative approaches to SBB using surgical navigation include methods of tracking the biopsy probe and the use of other types of guidance/fixation apparatus. An array of infrared emitting diodes or passive reflective markers is secured to the biopsy instrument (Fig. 7), calibrated, and the tip and trajectory of the probe

Figure 5 The wand tip is inserted through the guide block and into the twist drill hole. The probe tip length (PTL—tip to shank) is a constant for any given probe. The wand offset (WO) is the distance between the shank and the guide block.

78

&

PART II: TECHNOLOGIC APPLICATIONS

Figure 6 The reducing tube is inserted into the guide block. The reducing tube offset (RTO) is a distance from the topof the guide block to the top of the reducing tube.

are tracked as it is advanced intracranially. The principal advantages of this method are that it obviates the need for the calculation shown above, and allows for freehand biopsy. This approach does assume, however, that the probe/array structure is rigid and the probe is linear. If the weight of the array bends the probe, the navigation will present a spurious location of the probe tip. This problem may be amplified when performing related procedures such as placement of ventricular or tumor cyst catheters as they are typically less rigid than biopsy instruments.

Figure 7 An array of passive reflecting markers is secured to the biopsy instrument. Reducing tubes of various sizes are also shown on this figure.

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

79

Nonetheless, this technique has been successfully used in such applications (28). Although the notion of real-time tool tracking has clear advantages, its popularity among neurosurgeons remains unclear. There are multiple devices available to guide and hold the biopsy probe. In addition to the method outlined above, one may use the arc of a conventional stereotactic frame (without using it for localization or planning) aligned using the navigation probe. Another popular approach is to use a ball-and-socket device (e.g., Navigus, Image Guided Neurologics) secured to the skull, where the guidance sleeve is ‘‘steered’’ into place using a navigational wand (Fig. 8). The sleeve is then locked in place for subsequent drilling and biopsy. The intraoperative MRI (iMRI) -based technique does not require preoperative navigation scans because they may be obtained in near real time intraoperatively. More importantly, the positioning of an MR-compatible biopsy needle within the targeted lesion can be directly visualized prior to obtaining tissue. The authors use a scalp-mounted guidance device (Navigus, Image-Guided Neurologics) (Fig. 8). This apparatus allows biopsy to be performed percutaneously via twist-drill hole in conjunction with a 0.12 T iMRI (Odin Medical Systems, Inc., Newton, Massachusetts, U.S.A.) (Fig. 9). In our experience, accuracy of this system’s navigation component is comparable to conventional frame or frameless navigation (29). Bernstein et al. (30) reported on the use of a 0.2 T iMRI (GE Medical Systems, Milwaukee, Wisconsin, U.S.A.). This system also combines intraoperative

Figure 8 The scalp-mounted ball-and-socket device (Navigus, Image-Guided Neurologics) allows biopsy to be performed percutaneously via twist drill hole in conjunction with intraoperative MRI.

80

&

PART II: TECHNOLOGIC APPLICATIONS

Figure 9 The patient is positioned in the 0.12 T intraoperative MRI (Odin Medical Systems, Inc., Newton, Massachusetts, U.S.A.). The head is fixed in a three-point MR-compatible

imaging with image guidance. It incorporates a Polaris position tracker (Northern Digital, Waterloo, Canada), which, on the basis of the positional information, calculates and displays the position of the registered tools over the scanned images. Use of this system has resulted in real-time imaging to facilitate surgical planning, confirm entry into lesions, accuracy of catheter placement, and to limit postoperative complications (30).

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

81

Liu et al. (21) reported on a method to remotely steer a biopsy probe within a 1.5 T iMRI (NT-ACS, Philips Medical Systems) using intraoperative imaging for guidance and confirmation of appropriate placement. They found improved targeting accuracy, ‘‘even in the presence of brain shift.’’ Hall et al. (31) combined intraoperative MR guidance with intraoperative turbo spectroscopy and more sensitive single-voxel spectroscopy in their series of 42 biopsies. Intraoperative turbo spectroscopy, although associated with higher ‘‘contamination’’ and somewhat less precision than single-voxel studies, proved to be helpful in target selection (32). Areas of elevated choline relative to creatinine were targeted during the biopsy. This approach resulted in a 100% diagnostic success rate (33). It remains to be determined whether the benefits of the iMRI-guided frameless biopsy and related procedure outweigh the disadvantages (such as inconvenience of the iMRI suite setup, logistics, and cost).

SBB-RELATED PROCEDURES A number of procedures employ the principles of SBB. Placement of an Ommaya reservoir facilitates percutaneous access to intracranial fluid compartments. Most frequently, it is used for delivery of chemotherapy for patients with hematopoietic and solid malignances, drainage of cystic tumors, and drainage or delivery of radioisotopes in cystic craniopharyngioma (27,34–36). Insertion of this device into a relatively large nondisplaced ventricle may not be difficult; however, it can be challenging when ventricles are small or displaced, or when the targeted cyst is small or deep. We have found that use of SNS for this application is straightforward and can be carried out in an unencumbered sterile field. Brachytherapy refers to placement of radioactive substances (i.e., isotopes) within or near tumors. Modern brachytherapy involves stereotactic targeting, cross-sectional imaging, and computerized dosimetry. The principles of stereotactic frameless navigation have been applied to brachytherapy (15,26). We used an SNS to direct permanent iodine implants into meningiomas in poor surgical/radiosurgical candidates with excellent tumor control. Julow et al. described the use of the BrainLAB SNS (BrainLAB AG, Heimstatten, Germany) in their procedures. Fusion of the CT, MRI, PET, and SPECT images provided accurate and precise target volumes, more exact localization of catheters and isotope seeds (verification fusion), differentiation between the localization and amount of the necrotic and proliferating parts of the tumors, and showed volume changes resulting from interstitial irradiation (37). Frameless stereotactic techniques are not widely employed in functional neurosurgery, as compared to some previously described areas of neurosurgery. The principal reasons for this probably relate to the (spurious) perception that frameless systems are less accurate than frames and a lack of appropriate guidance platform that allows for microelectrode recording and repositioning. Nevertheless, the evidence that image-guided techniques may have a promising future in functional neurosurgery is growing. Roux et al. (38) described frameless stereotactically

82

&

PART II: TECHNOLOGIC APPLICATIONS

guided placement of electrodes using fMRI data to control phantom limb pain. Chu et al. (20) reported that intraoperative MR imaging techniques have the potential to greatly improve the stereotactic methods used for functional neurosurgery. They claim that evolving intraoperative imaging may have a positive impact on accuracy and complication avoidance even with deep brain stimulating electrodes and depth electrodes for epilepsy.

COMPARISON OF SBB TECHNIQUES Frameless vs. Frame Although image-guided frame-based SBB was a major technological breakthrough and a gold standard of brain biopsy due to its low morbidity and high diagnostic yield, it was used by a relatively small fraction of neurosurgeons. Some neurosurgeons still claim its superiority, particularly with regards to accuracy over the frameless techniques (9). Others, however, favor frameless SBB and use skull fiducials when accuracy comparable to frame SBB is required (15,16,18,19,29). The principal advantages of frameless SBB over frame are: 1. Logistics. The timing of frame placement and imaging, prior to the biopsy later that day in frame-based procedures, can be eliminated, as imaging for frameless procedures can be done days before surgery. 2. Patient comfort. In the authors’ experience, patients who have had both frame and frameless procedures universally prefer the frameless methodology. 3. Interactive capabilities. SNSs allow the surgeon to try various trajectories during surgery by reorienting the navigation wand with instantaneous visualization of the results. This dynamic feature allows for visualization of the true (as opposed to planned) trajectory during surgery, allowing for changes as deemed necessary by the surgeon (as compared to ‘‘static’’ frame-based biopsies). 4. SNSs usually offer more complex image processing than frame systems such as incorporating additional imaging information (e.g., functional MRI, MR spectroscopy, or PET scan). 5. Frameless systems tend to be more user friendly and eliminate the need for setting complex frames. Dorward et al. (16) compared the results of 76 frame-based and 79 frameless procedures and concluded that frameless techniques were associated with an overall decrease in the operating room occupancy, duration of anesthesia, and reduction of cost. It is our opinion, however, that their assertion that SBB using frameless techniques has less risk than frame procedures should be interpreted with caution, as there is little reason to believe this to be true for surgeons experienced in frame techniques. With regard to accuracy, it has been shown that frameless procedures can be as good, if not better, than frame-based biopsies. Moriarty et al. (39) made phantom experiments that demonstrated a 1:1 correlation between the MR image of a stereotactically guided probe and its relationship to a target and the actual

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

83

relationship of the probe and target. Maciunas et al. (40) showed that the application accuracy of frame stereotactic procedures (when done with slice-by-slice registration) is substantially worse than achievable with a frameless system using skull-implanted fiducials. Our own clinical experience demonstrated that even with scalp-applied fiducials, our rate of diagnosis was comparable to those obtained in an era when we used frames (8,15). Role of Intraoperative MRI-Guided Biopsy Brain biopsy using iMRI-guided technique has a number of potential advantages. The main benefits are to avoid preoperative imaging with fiducials, and to allow visualization of the location of the biopsy instrument within the target. In general, SBB is usually not associated with significant intraoperative brain movement. Shift, however, may become a problem when the lesion is located close to the ventricular system, or has a considerable cystic component. In such instances, an egress of CSF or cystic fluid can lead to changes in the configuration and size of the lesion and brain shift. Intraoperative MRI may be helpful in detecting these changes. It is not rare for the first biopsy specimen to be nondiagnostic (41). If such tissue is ‘‘reactive’’ and a more definitive diagnosis is expected based on diagnostic imaging, the surgeon must guess where to biopsy next when relying on preoperative imaging. Biopsy using iMRI allows the surgeon to rationally redirect the instrument to ensure diagnosis (especially with small lesions) and, potentially, obviates the need for frozen section. Also, iMRI may allow visualization of the relationship of the biopsy tool to nearby blood vessels prior to taking a specimen, potentially reducing the risk of clinically significant hemorrhage. Some high-resolution iMRI machines (e.g., 1.5 T NT-ACS iMRI, Philips Medical Systems) can perform diffusion weighted imaging, MR spectroscopy, MR angiography, and MR venography in addition to T1-weighted, T2-weighted, and turbo FLAIR (fluid-attenuated inversion recovery) imaging. These advanced imaging features may result in improved diagnostic yield of these procedures. Most criticism of iMRI-guided SBB is related to its high cost, need for special MR-compatible biopsy instruments, and increased operative time. Although the authors are not aware of any study directly comparing the duration of procedures and cost efficiency of MR-guided biopsy to other biopsy techniques, such costbenefit analyses should also consider that intraoperative MRI SBB makes preoperative as well as postoperative imaging unnecessary, and may obviate the need for intraoperative frozen section or smear. COMPLICATIONS Hemorrhage Hemorrhage is by far the most dreaded and, thus, clinically significant complication of SBB and surgeons should be familiar with management strategies (Table 2). Fortunately, with the use of modern image guidance or frames, the likelihood of

84

&

PART II: TECHNOLOGIC APPLICATIONS

Table 2 Management of Intraoperative Bleeding During Stereotactic Biopsy (1) Leave the outer sheath of the biopsy needle in place to allow free exit of blood (2) Elevate the head of the bed. If the bleeding is venous, this (and steps 3 and 4) measure may stop it (3) Irrigate the needle sheath using a narrow gauge spinal needle, to prevent formation of the blood clot within the needle or immediately around it. Such a clot may obstruct the exit of blood and make a false impression that the bleeding has stopped. (4) Mechanically obturate the sheath with its stylette to clear deep clot (5) If the bleeding is arterial and does not stop with irrigation, flush the biopsy needle with 0.5 to 2 cm3 of thrombin (5000 U/cm3) Source: From Ref. 49.

injury to the vessels is relatively low. In most series, the incidence of SBB-related hemorrhage ranges from 1.2% to 8% (42) and clinically significant hemorrhage (leading to neurological morbidity and mortality) in 1% to 2% of cases. Patients with vascular malignant lesions or with human immunodeficiency virus (HIV) infection are believed to be more prone to hemorrhage (7,43). Contemporary frameless stereotaxy (and some frame systems) allow the surgeon to visualize the path of the biopsy needle from the cortex down to the target. Such tracking may help avoid superficial sulcal blood vessels and deeper vessels as well. Nondiagnostic Biopsy Nondiagnostic biopsy is defined as failure to achieve definitive microbiological or histological diagnosis based on the tissue obtained (44) and may occur either due to technical reasons or unfavorable lesion characteristics. Technical failures may be associated with various problems in SNS biopsies such as miscalculations, inaccurate registration (perhaps due to movement of scalp fiducials), or movement of the guidance apparatus, deflection of the biopsy probe against the walls of a twist drill skull hole, displacement of the DRF after registration, or brain movement. These technical problems can often be solved by appreciating the potential causes and methodically eliminating them, one by one. On occasion, reregistration, or even reimaging may be required. With regard to unfavorable lesion characteristics, it was reported that biopsy of nonenhancing or non-neoplastic lesions as well as lesions in immunocompromised patients may be associated with higher incidence of nondiagnostic biopsy (45). Brainard et al. (41) showed that the rate of misdiagnosis after the ‘‘first pass’’ may be as high as 50% in non-neoplastic lesions, compared to 27% in neoplastic lesions. Soo et al. (44) reported an 8.7% failed biopsy rate in their series of 518 procedures. HIV-positive patients had higher incidence of nondiagnostic biopsies. In a situation of nondiagnostic biopsy, Brainard et al. (41) suggested to take more samples (up to four) at adjacent sites, as this may increase the diagnostic yield to 89%. Our practice is to either advance or withdraw the biopsy needle a 5 to 10 mm, or rotate it within the lesion, based on the surgeon’s judgment.

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

85

Highly vascular lesions may cause sufficient bleeding on acquisition of the first specimen such that the procedure may need to be aborted without diagnosis. As in any type of SBB, biopsy of malignant lesions may also be associated with seeding of the SBB tract (46). This is an extremely rare but serious complication. POSTOPERATIVE MANAGEMENT Although there is evidence that most complications (i.e., hemorrhage) occur within the first six hours of the procedure, the issue of timing of the patient’s discharge after SBB remains controversial. Opinions range from discharging patients home after four hours of observation ‘‘a safe and well-tolerated practice’’ (47,48) to a more conservative overnight stay, justified by a reported 0.4% risk of symptomatic delayed neurological deficit despite normal postbiopsy imaging (42). At present, we obtain a CT scan without contrast two hours following surgery and if no gross bleeding occurred at surgery, the patient has not changed neurologically, and this delayed scan shows no evidence of bleeding greater than 1 cm in dimension, the patient is transferred to the regular nursing floor with the expectation of discharge the next day. When any of these factors are not true, patients are subject to more intensive observation. In addition to screening for hemorrhage, postbiopsy CT is helpful to verify the location of the tip of the biopsy needle, because it often demonstrates a small air bubble or a spot of blood at the biopsy site. CONCLUSIONS Image-guided SBB and related techniques continue to evolve. The future of this area of neurosurgery probably resides in further development of multimodality imaging (e.g., MR spectroscopy, PET), intraoperative MR imaging (for better target definition and improved safety) and its use for molecular diagnosis and delivery of new treatments.

86

&

PART II: TECHNOLOGIC APPLICATIONS

REFERENCES 1. Horsley V, Clarke RH. The structure and function of the cerebellum examined by a new method. Brain 1908; 31:45–124. 2. Conway LW. Stereotaxic diagnosis and treatment of intracranial tumors including an initial experience with cryosurgery for pinealomas. J Neurosurg 1973; 38(4):453–460. 3. Waltz JM, Cooper IS. A comparative study of ventriculography and pneumoencephalography for roentgenographic landmarks during cryothalamectomy. Am J Roentgenol Radium Ther Nucl Med 1966; 97(3):583–587. 4. Brown RA. A computerized tomography-computer graphics approach to stereotaxic localization. J Neurosurg 1979; 50(6):715–720. 5. Kelly PJ, Earnest F 4th, Kall BA, Goerss SJ, Scheithauer B. Surgical options for patients with deep-seated brain tumors: computer-assisted stereotactic biopsy. Mayo Clin Proc 1985; 60(4):223–229. 6. Lufkin R, Duckwiler G, Spickler E, Teresi L, Chang M, Onik G. MR body stereotaxis: an aid for MR-guided biopsies. J Comput Assist Tomogr 1988; 12(6): 1088–1089. 7. Apuzzo ML, Chandrasoma PT, Cohen D, Zee CS, Zelman V. Computed imaging stereotaxy: experience and perspective related to 500 procedures applied to brain masses. Neurosurgery 1987; 20(6):930–937. 8. Gomez H, Barnett GH, Estes ML, Palmer J, Magdinec M. Stereotactic and computer-assisted neurosurgery at the Cleveland Clinic: review of 501 consecutive cases. Cleve Clin J Med 1993; 60(5):399–410. 9. Kulkarni A, Bernstein M. Stereotactic biopsy. In: Bernstein M, Berger M, eds. Neurooncology: the Essentials. New York: Thieme Medical Publishers, Inc., 2000: 122–129. 10. Friedman WA, Sceats DJ Jr, Nestok BR, Ballinger WE Jr. The incidence of unexpected pathological findings in an image-guided biopsy series: a review of 100 consecutive cases. Neurosurgery 1989; 25(2):180–184. 11. Lunsford LD, Martinez AJ. Stereotactic exploration of the brain in the era of computed tomography. Surg Neurol 1984; 22(3):222–230. 12. Couldwell WT, Apuzzo ML. Initial experience related to the use of the Cosman– Roberts–Wells stereotactic instrument. Technical note. J Neurosurg 1990; 72(1):145–148. 13. Heilbrun MP, Roberts TS, Apuzzo ML, Wells TH Jr, Sabshin JK. Preliminary experience with Brown–Roberts–Wells (BRW) computerized tomography stereotaxic guidance system. J Neurosurg 1983; 59(2):217–222. 14. Leksell L. A stereotactic apparatus for intracerebral surgery. Acta Chir Scand 1949; 99:229–233. 15. Barnett GH, Miller DW, Weisenberger J. Frameless stereotaxy with scalp-applied fiducial markers for brain biopsy procedures: experience in 218 cases. J Neurosurg 1999; 91(4):569–576. 16. Dorward NL, Paleologos TS, Alberti O, Thomas DG. The advantages of frameless stereotactic biopsy over frame-based biopsy. Br J Neurosurg 2002; 16(2):110–118.

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

87

17. Grunert P, Espinosa J, Busert C, Gunthner M, Filippi R, Farag S, Hopf N. Stereotactic biopsies guided by an optical navigation system: technique and clinical experience. Minim Invasive Neurosurg 2002; 45(1):11–15. 18. Paleologos TS, Dorward NL, Wadley JP, Thomas DG. Clinical validation of true frameless stereotactic biopsy: analysis of the first 125 consecutive cases. Neurosurgery 2001; 49(4):830–835; discussion 835–837. 19. Bernays RL, Kollias SS, Khan N, Brandner S, Meier S, Yonekawa Y. Histological yield, complications, and technological considerations in 114 consecutive frameless stereotactic biopsy procedures aided by open intraoperative magnetic resonance imaging. J Neurosurg 2002; 97(2):354–362. 20. Chu RM, Tummala RP, Kucharczyk J, Truwit CL, Maxwell RE. Minimally invasive procedures. Interventional MR image-guided functional neurosurgery [review]. Neuroimaging Clin North Am 2001; 11(4):715–725. 21. Liu H, Hall WA, Truwit CL. Neuronavigation in interventional MR imaging [review]. Prospective stereotaxy. Neuroimaging Clin North Am 2001; 11(4):695–704. 22. Maurer CR Jr, Fitzpatrick JM, Wang MY, Galloway RL Jr, Maciunas RJ, Allen GS. Registration of head volume images using implantable fiducial markers. IEEE Trans Med Imaging 1997; 16(4):447–462. 23. Wang MY, Maurer CR Jr, Fitzpatrick JM, Maciunas RJ. An automatic technique for finding and localizing externally attached markers in CT and MR volume images of the head. IEEE Trans Biomed Eng 1996; 43(6):627–637. 24. West JB, Fitzpatrick JM, Toms SA, Maurer CR, Maciunas RJ. Fiducial point placement and the accuracy of point-based, rigid body registration. Neurosurgery 2001; 48(4):810–816; discussion 816–817. 25. Dorward NL, Alberti O, Dijkstra A, Buurman J, Kitchen ND, Thomas DG. Clinical introduction of an adjustable rigid instrument holder for frameless stereotactic interventions. Comput Aided Surg 1997; 2(3–4):180–185. 26. Obasi PC, Barnett GH, Suh JH. Brachytherapy for intracranial meningioma using a permanently implanted iodine-125 seed. Stereotact Funct Neurosurgery 2002; 79(1):33–43. 27. Voges J, Sturm V, Lehrke R, Treuer H, Gauss C, Berthold F. Cystic craniopharyngioma: long-term results after intracavitary irradiation with stereotactically applied colloidal beta-emitting radioactive sources. Neurosurgery 1997; 40(2):263–269; discussion 269–270. 28. Gil Z, Siomin V, Beni-Adani L, Sira B, Constantini S. Ventricular catheter placement in children with hydrocephalus and small ventricles: the use of a frameless neuronavigation system. Childs Nerv Syst 2002; 18(1–2):26–29. 29. Kanner AA, Vogelbaum MA, Mayberg MR, Weisenberger JP, Barnett GH. Intracranial navigation by using low-field intraoperative magnetic resonance imaging: preliminary experience. J Neurosurg 2002; 97(5):1115–1124. 30. Bernstein M, Al-Anazi AR, Kucharczyk W, Manninen P, Bronskill M, Henkelman M. Brain tumor surgery with the Toronto open magnetic resonance imaging system: preliminary results for 36 patients and analysis of advantages, disadvantages, and future prospects. Neurosurgery 2000; 46(4):900–907; discussion 907–909. 31. Hall WA, Liu H, Maxwell RE, Truwit CL. Influence of 1.5-Tesla intraoperative MR imaging on surgical decision making. Acta Neurochir Suppl 2003; 85:29–37.

88

&

PART II: TECHNOLOGIC APPLICATIONS

32. Martin AJ, Liu H, Hall WA, Truwit CL. Preliminary assessment of turbo spectroscopic imaging for targeting in brain biopsy. AJNR Am J Neuroradiol 2001; 22(5):959–968. 33. Hall WA, Martin A, Liu H, Truwit CL. Improving diagnostic yield in brain biopsy: coupling spectroscopic targeting with real-time needle placement. J Magn Reson Imaging 2001; 13(1):12–15. 34. Al-Anazi A, Bernstein M. Modified stereotactic insertion of the Ommaya reservoir. Technical note. J Neurosurg 2000; 92(6):1050–1052. 35. Mottolese C, Stan H, Hermier M, Berlier P, Convert J, Frappaz D, Lapras C. Intracystic chemotherapy with bleomycin in the treatment of craniopharyngiomas. Childs Nerv Syst 2001; 17(12):724–730. 36. Rogers LR, Barnett G. Percutaneous aspiration of brain tumor cysts via the Ommaya reservoir system. Neurology 1991; 41(2 Pt 1):279–282. 37. Julow J, Major T, Emri M, Valalik I, Sagi S, Mangel L, Nemeth G, Tron L, Varallyay G, Solymosi D, Havel J, Kiss T. The application of image fusion in stereotactic brachytherapy of brain tumours. Acta Neurochir (Wien) 2000; 142(11):1253–1258. 38. Roux FE, Ibarrola D, Lazorthes Y, Berry I. Chronic motor cortex stimulation for phantom limb pain: a functional magnetic resonance imaging study: technical case report. Neurosurgery 2001; 48(3):681–687. 39. Moriarty TM, Quinones-Hinojosa A, Larson PS, Alexander E III, Gleason PL, Schwartz RB, Jolesz FA, Black PM. Frameless stereotactic neurosurgery using intraoperative magnetic resonance imaging: stereotactic brain biopsy. Neurosurgery 2000; 47(5):1138–1145; discussion 1145–1146. 40. Maciunas RJ, Galloway RL Jr, Fitzpatrick JM, et al. A universal system for interactive image-directed neurosurgery. Stereotact Funct Neurosurg 1992; 58:108–113. 41. Brainard JA, Prayson RA, Barnett GH. Frozen section evaluation of stereotactic brain biopsies: diagnostic yield at the stereotactic target position in 188 cases. Arch Pathol Lab Med 1997; 121:481–484. 42. Field M, Witham TF, Flickinger JC, Kondziolka D, Lunsford LD. Comprehensive assessment of hemorrhage risks and outcomes after stereotactic brain biopsy. J Neurosurg 2001; 94(4):545–551. 43. Luzzati R, Ferrari S, Nicolato A, Piovan E, Malena M, Merighi M, Morbin M, Gerosa M, Rizzuto N, Concia E. Stereotactic brain biopsy in human immunodeficiency virus-infected patients. Arch Intern Med 1996; 156(5):565–568. 44. Soo TM, Bernstein M, Provias J, Tasker R, Lozano A, Guha A. Failed stereotactic biopsy in a series of 518 cases. Stereotact Funct Neurosurg 1995; 64(4):183–196. 45. Whiting DM, Barnett GH, Estes ML, Sila CA, Rudick RA, Hassenbusch SJ, Lanzieri CF. Stereotactic biopsy of non-neoplastic lesions in adults. Cleve Clin J Med 1992; 59(1):48–55. 46. Steinmetz MP, Barnett GH, Kim BS, Chidel MA, Suh JH. Metastatic seeding of the stereotactic biopsy tract in glioblastoma multiforme: case report and review of the literature. J Neurooncol 2001; 55(3):167–171. 47. Kaakaji W, Barnett GH, Bernhard D, Warbel A, Valaitis K, Stamp S. Clinical and economic consequences of early discharge of patients following supratentorial stereotactic brain biopsy. J Neurosurg 2001; 94(6):892–898.

CHAPTER 5: BRAIN BIOPSY AND RELATED PROCEDURES

&

89

48. Bhardwaj RD, Bernstein M. Prospective feasibility study of outpatient stereotactic brain lesion biopsy. Neurosurgery 2002; 51(2):358–361. 49. Chimowitz MI, Barnett GH, Palmer J. Treatment of intractable arterial hemorrhage during stereotactic brain biopsy with thrombin. Report of three patients. J Neurosurg 1991; 74(2):301–303.

6

INTRAOPERATIVE MRI Andrew A. Kanner Brain Tumor Institute and Department of Neurological Surgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Michael A. Vogelbaum Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

DEVELOPMENT OF INTRAOPERATIVE MR (iMRI) IMAGING IN NEUROSURGERY The recent introduction of intraoperative magnetic resonance imaging (iMRI) into the operating room is the culmination of decades of progress in the neurosurgeon’s ability to integrate three-dimensional imaging information into pre- and intraoperative surgical planning (1–19). Progress in this arena dates back to the early 20th century, when innovators of what we refer to as modern neurosurgical care recognized the need to perform intraoperative imaging assessments of intracranial pathology. At the early part of the last century plain skull X-ray images, pneumoencephalography, ventriculography, and soon after cerebral angiography were the first imaging modalities that became available (20,21). Soon these imaging modalities were used in conjunction with stereotactic frames, enabled stereotactic procedures, and helped to guide instruments (precisely) into the depth of the human brain (22,23). The invention of computerized tomography (CT) imaging allowed neurosurgeons to obtain more detailed images of a larger area of the brain than could be performed with any prior imaging technology. The application of this imaging modality to the intraoperative environment was an obvious step to take and this technology was adopted by a number of hospitals (24–26). Initial efforts led to the integration of mobile CT units into the operating room in the early 1980s. However, there were significant limitations associated with intraoperative CT including exposure to radiation for both patient and staff, fixed slice orientation, and poor soft-tissue contrast (27,28). The introduction of MRI into the clinical arena in the late 1970s and early 1980s has revolutionized the neurosurgeon’s ability to assess both neuroanatomy

Icons indicate materials available on the accompanying CD.

91

92

&

PART II: TECHNOLOGIC APPLICATIONS

and clinical neuropathology noninvasively. Its superiority over other imaging modalities for this purpose has been increasingly well documented. This superiority, in particular over CT, is due to the sensitivity of MRI to clearly image detailed neuroanatomy in all three dimensions, its ability to detect even subtle pathological changes in the brain parenchyma, and its additional application for functional and metabolic studies (29). Shortly following the widespread use of CT and MRI in neuroimaging, efforts were undertaken to develop intraoperative computerized frameless stereotactic guidance systems, which would allow the integration of detailed imaging data sets into intraoperative navigation. These image-guided navigation systems have truly revolutionized intracranial procedures (9,30–33). Current conventional framebased and frameless systems used for image guidance use a set of preoperative acquired and archived CT or MR images, which are loaded into a planning workstation. For the frame-based systems, the frame fixed to the patient’s head helps to establish the reference for further navigation. In the case of a frameless system, patients typically have scalp fiducials that were placed at the time of imaging and that can be used in the operating room. In both cases, these fiducial markers are necessary to define the position of working volume in space. The utility of conventional navigational systems may be limited by reliance on preoperative imaging (34–37). Gross movement of the brain (brain shift) from loss of cerebrospinal fluid and local tissue distortions or swelling by the surgical resection itself can change the relative position of the brain and the lesion(s) being targeted (38,39). In addition, the relationship of biopsy instruments or other instruments to critical structures or small lesions cannot be confirmed without a means to update the image data in the operating suite. While ultrasound (40–42) and CT scans have been used as intraoperative imaging tools, the introduction of MRI into the operating room has made it possible to produce high-quality, near real-time images throughout neurosurgical procedures (13,14,19,43–46). iMRI provides the surgeon with information about progress throughout a procedure, which may have an impact on the extent of tumor resection—a factor that is believed by many to be an important, favorable prognostic factor for glial tumors (12,47,48). The use of intraoperatively updated neuronavigation enhances the surgeon’s ability to perform a radical resection and morbidity may be lower as well (49).

THE CONCEPT OF iMRI The concept of intraoperative image-guided MRI use has been described by one of the leading pioneers in the field, Peter Black. He defined the concept of iMRI with its three main characteristics. First, the use of iMRI allows the surgeon to ‘‘overcome spatial restrictions’’ imposed by the surgical approach and makes it possible for the surgeon to view ‘‘beyond the exposed’’ surgical field. It helps in precisely localizing a lesion, in planning each step of the procedure via a multiplanar model on the computer display, and in calculating the ideal access to the tumor before the operation. This first principle helps to virtually conduct the surgical procedure and minimize the surgical trauma and prevent unnecessary exposure of normal brain

CHAPTER 6: INTRAOPERATIVE MRI

&

93

parenchyma. Secondly, iMRI allows for the recognition and identification of ‘‘changes in anatomy,’’ both normal and tumor, which occur ‘‘during (the course of) the procedure.’’ The dynamic alterations that take place can be documented and the surgeon can modify or adjust the progress of the surgery using this real-time data, and finally the use of this intraoperatively acquired data sets for ‘‘improved intraoperative neuronavigation’’ (50). DEVELOPMENT OF iMRI SYSTEMS Groundbreaking iMRI Systems The first documented efforts to integrate MRI systems for interventional procedures date back in the late 1980s and early 1990s (51–53). In 1986, the first MRI-guided brain biopsies were reported (54). Subsequent development of MRcompatible instruments and other interventional material was followed by the first reports for the use of MRI to aid biopsies of liver, prostate, and head and neck lesions (55–57). These efforts were primarily undertaken to perform controlled tissue biopsies or cyst aspirations. Procedures were performed in low-field magnets that allowed some percutanous and minimally invasive procedures. However, they were limited by the restricted access to the patient. The first attempt to incorporate MRI into the neurosurgical field began with the development of an MRI scanner dedicated for intraoperative use. This unit, the Signa SP Intraoperative Imaging System (the ‘‘Double Doughnut’’; General Electric (GE) Medical Systems, Milwaukee, Wisconsin, U.S.A.), was proposed by GE and a group of investigators at the Brigham and Women’s Hospital (1,58) (Fig. 1). Initially installed in 1994, the

Figure 1 One of the two large tori is seen, with the patient couch entering the axis (standard for most cases). The fully MRI-compatible anesthesia machine and surgical instrument cabinet are seen on the right.

94

&

PART II: TECHNOLOGIC APPLICATIONS

system underwent a test phase with a number of modifications until it went into practical clinical use two years later. Surgery was performed within the 56 cm vertical gap space within the magnet. Image monitors are integrated into the magnet and allow the surgeon to see near-real imaging without moving the patient and displaying the diameter of the spherical imaging volume at 30 cm. This mid-field scanner of 0.5 T incorporates a number of unique features. A tracking device is integrated into the scanner, which is LED based (optical tracking system) and provides precise navigational accuracy within 1 mm. Specially designed receive and transmit surface coils were produced in two sizes, which allow direct surgical access when placed on a patient’s head. The acquisition time for intraoperative T1weighted images is 5 to 66 seconds, T2-weighted images between 26 and 351 seconds, and proton density images between 19 and 248 seconds. More important was the demonstration that with defined sequence motion, ultimately, robust imaging is possible (2). Subsequently, at the University of Minnesota, a conceptionally similar system was modified for direct neurosurgical interventional in an operating room (59–61) (Fig. 2). The University of Minnesota system incorporates a confined highfield MRI scanner and portable C-arm angiography system. The MRI consists of a 1.5 T high-field strength interventional magnet (ACS-NT; Philips Medical Systems, Best, The Netherlands) with a 180 cm bore length having 100 cm flared openings at the front and rear entries, which offers adequate patient access. The high-field strength system provides advanced MRI features such as real-time interactive imaging, magnetic resonance angiography, magnetic resonance thermometry, MRS, functional assays such as brain activation and perfusion assays, and diffusion.

Figure 2 View through the tori shows the patient couch docked perpendicular to standard, giving room directly over the patient’s head for the surgeon only. The couch occupies the assistant’s position. This is useful for very large patients (coat size > 50). The two large tori of the GE MRT system, standing on edge, are seen. The liquid crystal display screen is visible between them, above the operative field.

CHAPTER 6: INTRAOPERATIVE MRI

&

95

These additional sequences represent a major benefit compared with low- and mid-field systems.

Second-Generation iMRI Other institutions have also developed a variety of low- and high-field magnets within or adjacent to the operating room. Two groups of investigators in cooperation with Siemens took this slightly different approach to the idea of using MRI for neurosurgical interventional procedures, which they refer to as the second-generation iMRI (4,45,62). The ‘‘twin-room’’ concept incorporated a 0.2 T C-shaped resistive magnet (Magnetom Open; Siemens AG, Erlangen, Germany) with a lateral patient access of 240 within a 44 cm gap (Fig. 3). However, there was no vertical access during scanning and no lateral patient positioning was possible. A magnetically shielded cabin was established adjacent to one of the neurosurgical operating theaters, which harbors the magnet. Patients are moved to the cabin whenever the need for an additional imaging arises. The conditions within the cabin are comparable to the sterile OR standards and enable the surgeon to conduct some of the procedure directly in the magnet. A number of modifications and

Figure 3 Patient being positioned down the axis of the two tori. The surgeon has access to the patient from the right, and an assistant has similar access to the patient from the opposite side. The surgeon holds Pixsys pointer (LED linkend to overhead video cameras to precisely localize the pointer and biopsy vector). The respirator is visible to the right, behind the patient couch.

96

&

PART II: TECHNOLOGIC APPLICATIONS

adaptations of instruments were necessary to allow undisturbed use in the magnetic field. Using the fast imaging with steady-state precession (FISP) two-dimensional sequence, one image can be obtained every two seconds, producing almost real-time imaging conditions. This enables the placement of catheter or biopsy instrument ‘‘under vision.’’ However, only minor procedures can be performed within the scanner, in contrast to craniotomies, which have to be performed in the adjacent OR. Typically, the craniotomy is performed with preoperative acquired MRI images for navigation and throughout the procedure the need for updated images is assessed. The most important disadvantage of the ‘‘twin-room’’ concept is the need to transfer patient and data between OR and scanner room. The image transfer after obtaining the data set, back to the navigational computer in the OR, may take 20 to 40 minutes (62), as could the patient transfer. Phantom testing revealed field homogeneity for 30 cm phantoms, with only minor distortions, which have been further reduced by a correction program (63). Third-Generation iMRI: Current State of the Art The latest models of iMRI, which we refer to as third-generation iMRI, are mobile and have the capability to move in and out of the operating area. There have been two different systems created to date. One is a ceiling-mounted mobile unit with a superconductive 1.5 T magnet (Magnex Scientific, Abingdon, Oxon, U.K.) installed at the University of Calgary (5,64,65) (Fig. 4). A major modification to the standard OR required for this system is the installation of ceiling support track, to support the 4 ton magnet, enabling the movement in and out of the operating

Figure 4 Superconductive 1.5 T magnet. (Magnex Scientific intraoperative MRI, Calcary. Courtesy of Dr Sutherland.)

CHAPTER 6: INTRAOPERATIVE MRI

&

97

field. In contrast to other iMRI systems that require extensive room radio frequency (RF) shielding, this system has a local RF shielding method, which consists of a silver-impregnated tent attached to the magnet and placed around the patient and OR table. At the time of imaging, any non-MR-compatible equipment has to be moved out of the 5 G line (radial dimension 2.9 m). Navigation is accomplished by using Vitamin E capsules placed for orientation. The other mobile system is a 0.12 T permanent scanner (Odin PoleStar N-10, Medtronic Surgical Navigation Technologie, Louisville, Colorado, U.S.A.) mounted on a mobile gantry (6,17,66) (Fig. 5). The PoleStar N-10 magnet can move under a conventional OR table and be elevated in a programmed fashion whenever imaging is needed. This system has an optical image-guidance system for intraoperative navigation. The movable magnet can be placed in a magnetically shielded cage. At times, it is not needed and conventional surgery without any restriction can be performed. The design of this system obviates or mitigates most of the major problems associated with previous generations of iMRI, although it is not without its own limitations. The field of view is limited, due to the small size magnet producing a 14 l5 l2 cm3 spheroid field only. Larger lesions might not be visualized on one image set. There are also physical limitations, which limit the size of patients (in particular, their shoulders) that will fit into the gap between the magnets. On the other hand, this design may allow for use of more conventional instrumentation and better utilization of the operating suite when iMRI is not used. The individual characteristics of the three different generations of iMRI systems have been summarized in Table 1.

Figure 5 0.12 T permanent magnet (Odin PoleStar N-10, Medtronic Surgical Navigation Technologies, Louisville, Colo.) Draped and set up iMRI guided brain biopsy. The infrared camera is seen in the background.

None, patient inside MRI at all times

30 (spherical) Flexible Good Inside Brain, spine, body Available MRI compatible MRI compatible Adjacent room Additional

Open, vertical superconductive Dedicated OR >70 db 56 0.5 (mid-field)

Source: From Ref. 6.

Abbreviations: MRI, magnet resonance imaging; OR, operating room.

Image (cm) Coil Image quality MR imaging in OR Procedure Neuronavigation Instruments Anesthesia machine Work station Personnel (MRI-technician, radiologist) Portability

5-Gauss line (m)

System Location RF shielding Gap (cm) Field strength (tesla)

14 15 12 (0.12 T) Flexible Good (appropriate) Inside Brain, all positions Integrated Both MRI compatible and regular MRI compatible Inside OR No additional Portable, MRI moves to Patient

Patient moves to MRI

Open, vertical permanent RF-shielded regular OR <70 db 25 0.12 (low-field) 1.5 (high-field) 1.5 (0.12 T)

Third Generation

2.1 2.3 (0.2 T) 7.8 5 (1.5 T) 30 (spherical) Integrated Diagnostic Outside No lateral position Limited, some systems (MRI compatible) (MRI compatible) Adjacent room Additional

Open, horizontal ‘‘Twin-room’’ >70 db 40–66 0.2–1.5 (mid-high field)

Second Generation

&

Table 1 Comparison of the Three Generations of iMRI First Generation

98 PART II: TECHNOLOGIC APPLICATIONS

CHAPTER 6: INTRAOPERATIVE MRI

&

99

INSTRUMENTS For persons not familiar with the MRI environment, it is of vital importance to be aware of the following definitions, as their recognition is a matter of safety (67). MR environment: This term is used to describe the general environment present in the vicinity of an MR scanner, which includes all places within the MR procedure and operating room, including the center of the bore of the MRI scanner. In particular, this refers to the area within the 5 G line (defined below) around the scanner. Characteristics of the environment include the following: (i) the static magnetic field [the range of 0.12–1.5 T (1 T ¼ 10,000 G)], which is most common for iMRI and (ii) RP magnetic field pulses (of the order of tens to hundreds of MHz). MR safe: This term indicates that a device, when used in the MR environment, has been demonstrated to present no additional risk to the patient, but may affect the quality of the diagnostic information. MR compatible: This term indicates that a device, when used in the MR environment, is MR safe and has been demonstrated to neither significantly affect the quality of the diagnostic information nor have its operations affected by the MR device (68). Surgical tools and anesthesia equipment impose some special difficulties in this newly created electromagnetic environment (69). Much time and attention were spent on the development of MR safe and compatible instruments and equipment that are normally present in the operating room (55–57). There are two principal challenges involved in using any instrument or machinery in close vicinity of an MR scanner. First, there is a safety issue: ferromagnetic material will be pulled into the magnet (‘‘missile effect’’). This can cause significant injuries and damage to patient, staff, and equipment within the line of the ‘‘projectile.’’ With the initial high-field iMRI designs, any known ferromagnetic equipment had to be eliminated from the OR and any ‘‘nonferrous’’ instrument needed to be tested and labeled prior to its use in the operating field. This problem has been mitigated to a significant degree by the development of low-field systems, which have a much smaller 5 G line radius. This line specifies the perimeter around an MR scanner within which the static magnetic fields are higher than 5 G. Five gauss and below are considered ‘‘safe’’ levels of static magnetic field exposure for the general public. In the case of movable iMRI systems, moving the magnet in its ‘‘resting’’ place allows for the safe use of conventional instrumentation during the majority of surgical procedures. The other problem that needed to be solved was that devices containing any chrome, steel, ferromagnetic materials, or alloys produce magnetic field inhomogeneities, which produce considerable artifacts and distortions during the imaging process (see artifacts below). Instruments that passed the first stage of evaluation and were found to be MR safe had to be placed in the field to evaluate MR compatibility. Of note, electronics in many of the basic monitoring and ventilatory equipment used by anesthesia (70) and other electronic instrumentation produce sufficient RF noise to result in poor image quality. To summarize, one has to be aware of the potential dangers involved in working with the MR environment, namely, tissue injury caused by rotation or acceleration of ferromagnetic objects, burning injuries due to heating caused by RF-induced currents, and potential

100

&

PART II: TECHNOLOGIC APPLICATIONS

equipment malfunction caused by electromagnetic interference. This makes testing and evaluation as described above of any equipment and instrument used within the MR environment necessary.

NAVIGATION There are two navigational strategies that we use for iMRI-guided procedures (6): One is to perform an initial scan followed by navigation alone throughout the remainder of the procedure, with a final scan at the end of the procedure. This approach, used most commonly for stereotactic biopsies, is not very different from conventional neuronavigation with the exception that the final scan verifies that the biopsy was performed in the area targeted, and it provides the surgeon with immediate information regarding a possible procedure-related complication. The second approach is to perform repetitive scanning throughout a procedure to assess the progress of the resection. This approach permits the surgeon to ‘‘look’’ beyond the exposed field and aids in decision-making when brain and tumor tissue appear visibly undistinguishable. This is particularly so in the case when brain shift and local tissue deformation makes navigation using preoperative imaging data sets inaccurate. Updated imaging may be most useful when resections are performed close to anatomically defined eloquent cortex. The integrated software allows interactive neuronavigation. The display offers all basic navigational functions including target and trajectory guidance and various three-planar views that enable better orientation of anatomical relationships of tumor to surrounding tissue (6,16,17,37,50,66). Although very useful, the quality of reconstructed images in the three-planar view is strongly dependent on the selected scanning sequence and on the field strength of the system. The use of short image sequences that have thick slicing (4–10 mm) results in low-quality reconstruction. INTERPRETATION OF INTRAOPERATIVELY ACQUIRED IMAGES As with the use of iMRI units, there is a learning curve involved in the process of interpretation of intraoperatively acquired images (43). Several pitfalls have been reported (6,62). Gadolinium leakage may occur, particularly late in the procedure, and tissue manipulation may produce high signal on FLAIR and T2 imaging at the boundary of resection that could be potentially confused with residual tumor (71). Also, the disruption of the blood–brain barrier at the advanced stage of the procedure could lead to extravasation of contrast media and result in the assumption that residual tumor is present. To circumvent this problem, one is advised to carefully compare previous pre- and intraoperative images and the localization and pattern of enhancement. Some systems mitigate these issues with a compare function, which enables viewing of several image sets simultaneously, thereby helping the surgeon to differentiate tumor tissue from surgical artifact. When intraoperative imaging is inconclusive, navigation-guided serial biopsies could be obtained and sent for intraoperative pathological analysis. The pooling

CHAPTER 6: INTRAOPERATIVE MRI

&

101

of blood during image acquisition may also be problematic, especially for transsphenoidal procedures (6). Finally, the ‘‘unnatural’’ appearance of the air–brain interface poses some interpretation challenges for both radiologists and neurosurgeons. It is clear that monitoring the degree of resection, one of the main advantages of iMRI, requires a good interpretation of the defined resection margins of a lesion in the context of the above-described changes. ARTIFACTS Image artifact: This is a general term that refers to an inappropriate image signal at a specified spatial location. It is generally characterized as increased signal intensity in an area, which is known to contain no signal producing material or decreased signal intensity (voids) where signal should be produced (67). The most recognizable image disturbances are motion artifacts. Even a small amount of patient motion can result in severe artifacts. The nature of the artifact depends on the timing of the motion with respect to the acquisition and the extent of movement. However, for systems where the patients’ head is fixed in a head holder (with pins) this type of image artifact might be less common. There are a number of other elements that contribute to an unsatisfactory MR image in the OR environment. Although a lot of these interferences are specific for certain iMRI systems, some factors are of general importance. There are external factors that interfere with the homogeneity of the field and with stability of the magnet. Instability of the magnet is referred to as changes of the magnetic field over time, whereas inhomogeneity refers to spatial changes in the field. These changes can be caused by endogenous (magnet) or by external (environmental) factors. Such effects are strongly dependent on the nature of the magnet, e.g., being permanent or superconductive. For permanent magnets, like the PoleStar N-10, one factor could be a change in temperature. Any magnet is sensitive to configuration changes that can occur as a result of vibration, motion, or absorption of physical impact, which could cause movement of substantial magnet part and result in field inhomogeneity. This is of importance especially in the OR environment where objects could collide with the magnet. In addition, the presence of ferromagnetic materials, inaccurate shimming, and effects of equipment and instruments that become magnetized can influence the magnetic field and cause artifacts. These types of interference in particular will cause shape distortions. Another important source of image artifacts is electronic equipment and machines that emit RF noise causing artifact appearance or completely disturbed images. Such equipment includes anesthesia equipment, bipolar or other electronic surgical tools, and also electric cables, overhead lights, and a motorized OR table. The RF noise signal is picked up by the receive antenna along with the real MRI signal and, depending on the nature of the noise, appears on the images in the form of streaks, lines, or white noise within the displayed image. This type of interference makes it necessary that the room be shielded to prevent RF noise from outside the room, which is one of the structural measures that has to be completed in order to transform the OR suite into an MR area. Depending on the iMRI system, any equipment potentially causing RF noise needs to be removed from the room, simply switched off, or unplugged during the time of scanning. Any

102

&

PART II: TECHNOLOGIC APPLICATIONS

device that is essential or vital (e.g., life support equipment) and cannot be switched off has to be MRI compatible (67). General Benefits of Using iMRI Although both iMRI-based and conventional navigation provide precise localization, which allows for minimal access craniotomy and reduced tissue trauma, the former obviates the need for fiducial placement and utilization of a diagnostic MRI for preoperative imaging (31). However, loss of cerebrospinal fluid, retraction, edema, and resection of the tumor itself are responsible for intraoperative brain shift and tissue distortion, which adversely impact on the accuracy of conventional neuronavigation (35). This problem only increases as the case proceeds (34,38). It is the ability of iMRI to produce subsequent intraoperative images that provides the neurosurgeon with an unsurpassed ability to monitor resection progress and compensate for brain movement during surgery. The ability to visualize brain shift when coupled with navigation may enhance the surgeon’s ability to perform a maximal, safe resection (34,39). THE IMPACT OF EXTENT OF RESECTION OF BRAIN TUMORS ON OUTCOME The potential of iMRI to provide the surgeon an increased ability to achieve a gross total or maximal safe resection of a primary brain tumor needs to be considered in the context of the impact of extent of tumor resection on outcome. There is evidence that the extent of resection (EOR) of low-grade gliomas has a favorable impact on progression-free and overall survival, but this issue remains controversial (49,72,73). In patients with high-grade gliomas, there are also conflicting results regarding the impact of EOR (47,48,74), although the preponderance of evidence, to date, favors a beneficial effect (12,49,72). Shortcomings of most historical studies include inconsistencies in reporting the degree of resection, as well as differences in the modality and timing of neuroimaging. It is beyond the scope of this chapter to review this subject in detail. However, it is clear that if EOR, particularly radiographic gross total resection, does result in better outcome, then methods to intraoperatively monitor resection such as iMRI should aid in this effort (31). For pituitary tumors, iMRI may aid in facilitating complete resection, especially for tumors not invading the cavernous sinus. Whether such benefit will produce measurable improvement in outcome remains uncertain. The potential of iMRI to help produce improved resections may allow us to obtain a better understanding of the true relationship between EOR and outcome. Impact on Brain Biopsies Another potential benefit of iMRI navigation is enhancing the yield and safety of stereotactic brain biopsies (75). Although these represent only a small percentage

CHAPTER 6: INTRAOPERATIVE MRI

&

103

of the cases performed using iMRI, all of the brain biopsies reported in one series produced diagnostic tissue (6); a result that is favorable when compared to larger series of frame-based and frameless stereotactic biopsies with nondiagnostic biopsy rates of 3.7% to 8.1% (11,74). Intraoperative MRI allows for confirmation of biopsy instrument placement in the desired part of a lesion, visualization of what action needs to be performed to correct for a near miss of a lesion, and confirmation that the instrument is not immediately adjacent to nearby vascular structures. Such information may ultimately obviate the need for intraoperative histopathology review or enhance the safety of the procedure (75). Especially suitable for iMRI-guided biopsies are very small lesions and lesions with small or enhancing region, which might represent more aggressive tumor areas. CONCLUSION As experience has grown with iMRI, a number of clinical studies on the feasibility of the different systems have been published (1,18,36,45,46,50,61,65,76). These studies address the use of iMRI in various neurosurgical procedures and, in particular, they focus on the safety and accuracy of these systems. However, to date no studies addressing statistical and comparative evaluation of outcome or the effects of more radical tumor resection, biopsy accuracy, and yield have been published. A carefully designed assessment of intraoperative MRI compared to conventional navigational systems, which rely only on preoperative image data, will be necessary to justify this expensive technique. A clear characterization of indication defining patients who would benefit the most is necessary and consequently thereafter the appropriate cost-effectiveness studies. This type of study would help define the situation in which iMRI provides the greatest cost efficacy. In terms of improving current systems, optimization of image quality, shorter acquisition times, and software upgrades to allow fusion of images from external/ other modalities would add to the benefit of iMRI. The ability to perform fusion with MR sequences not available on a number of iMRI scanners (FLAIR, diffusion/perfusion studies, MR venography or angiography), single-photon computed tomography, positron emission tomography, or digital subtraction angiography image data sets would be of particular benefit. Integrated navigational guidance systems will be necessary for all systems. Novel applications for local RF shielding of the patient within the scanner are necessary to eliminate the high cost involved in shielding of the whole room. Ultimately, for iMRI to become the intraoperative imaging modality of choice in brain tumor surgery, high-quality imaging associated with the first two generation iMRI systems will need to be combined with the small footprint and flexibility of the third-generation iMRI system.

ACKNOWLEDGMENT The authors wish to thank Mrs. Martha Tobin for her assistance with the preparation of this manuscript.

104

&

PART II: TECHNOLOGIC APPLICATIONS

REFERENCES 1. Black PM, Alexander E III, Martin C, Moriarty T, Nabavi A, Wong TZ, et al. Craniotomy for tumor treatment in an intraoperative magnetic resonance imaging unit. Neurosurgery 1999; 45:423–431; discussion 431–433. 2. Kacher DF, Maier SE, Mamata H, Mamata Y, Nabavi A, Jolesz FA. Motion robust imaging for continuous intraoperative MRI. J Magn Reson Imaging 2001; 13: 158–161. 3. Maurer CR Jr, Hill DL, Martin AJ, Liu H, McCue M, Rueckert D, et al. Investigation of intraoperative brain deformation using a 1.5-T interventional MR system: preliminary results. IEEE Trans Med Imaging 1998; 17:817–825. 4. Steinmeier R, Fahlbusch R, Ganslandt O, Nimsky C, Buchfelder M, Kaus M, et al. Intraoperative magnetic resonance imaging with the magnetom open scanner: concepts, neurosurgical indications, and procedures: a preliminary report. Neurosurgery 1998; 43:739–747; discussion 747–748. 5. Kaibara T, Saunders JK, Sutherland GR. Utility of a moveable 1.5 Tesla intraoperative MR imaging system. Can J Neurol Sci 1999; 26:313–316. 6. Kanner AA, Vogelbaum MA, Mayberg MR, Weisenberger JP, Barnett GH. Intracranial navigation by using low-field intraoperative magnetic resonance imaging: preliminary experience. J Neurosurg 2002; 97:1115–1124. 7. Alexander E III, Kooy HM, van Herk M, Schwartz M, Barnes PD, Tarbell N, et al. Magnetic resonance image-directed stereotactic neurosurgery: use of image fusion with computerized tomography to enhance spatial accuracy. J Neurosurg 1995; 83:271–276. 8. Black PM, Moriarty T, Alexander E III, Stieg P, Woodard EJ, Gleason PL, et al. Development and implementation of intraoperative magnetic resonance imaging and its neurosurgical applications. Neurosurgery 1997; 41:831–842; discussion 842–845. 9. Alexander E III, Moriarty TM, Kikinis R, Black P, Jolesz FM. The present and future role of intraoperative MRI in neurosurgical procedures. Stereotact Funct Neurosurg 1997; 68:10–17. 10. Barnett GH, Steiner CP, Weisenberger J. Target and trajectory guidance for interactive surgical navigation systems. Stereotact Funct Neurosurg 1996; 66:91–95. 11. Barnett GH. The role of image-guided technology in the surgical planning and resection of gliomas. J Neurooncol 1999; 42:247–258. 12. Berger MS. Intraoperative MR imaging: making an impact on outcomes for patients with brain tumors. Am J Neuroradiol 2001; 22:2. 13. Bernstein M, Al-Anazi AR, Kucharczyk W, Manninen P, Bronskill M, Henkelman M. Brain tumor surgery with the Toronto open magnetic resonance imaging system: preliminary results for 36 patients and analysis of advantages, disadvantages, and future prospects. Neurosurgery 2000; 46:900–907; discussion 907–909. 14. Buchfelder M, Ganslandt O, Fahlbusch R, Nimsky C. Intraoperative magnetic resonance imaging in epilepsy surgery. J Magn Reson Imaging 2000; 12:547–555.

CHAPTER 6: INTRAOPERATIVE MRI

&

105

15. Fahlbusch R, Ganslandt O, Buchfelder M, Schott W, Nimsky C. Intraoperative magnetic resonance imaging during transsphenoidal surgery. J Neurosurg 2001; 95: 381–390. 16. Fahlbusch R, Ganslandt O, Nimsky C. Intraoperative imaging with open magnetic resonance imaging and neuronavigation. Childs Nerv Syst 2000; 16:829–831. 17. Hadani M, Spiegelman R, Feldman Z, Berkenstadt H, Ram Z. Novel, compact, intraoperative magnetic resonance imaging-guided system for conventional neurosurgical operating rooms. Neurosurgery 2001; 48:799–807; discussion 807–809. 18. Liu H, Hall WA, Martin AJ, Maxwell RE, Truwit CL. MR-guided and MR-monitored neurosurgical procedures at 1.5 T. J Comput Assist Tomogr 2000; 24:909–918. 19. Zimmermann M, Seifert V, Trantakis C, Kuhnel K, Raabe A, Schneider JP, et al. Open MRI-guided microsurgery of intracranial tumours. Preliminary experience using a vertical open MRI-scanner. Acta Neurochir (Wien) 2000; 142:177–186. 20. Clark RA, Obenchain TG, Hanafee WN, Wilson GH. Pneumoencephalography. Comparison of complications in 100 pediatric and 100 adult cases. Radiology 1970; 95:675–678. 21. Funch RB, MacMoran JW, Parker JA, Bernhard RA, D’Orazio EA. Radiology of the skull and central nervous system. Prog Neurol Psychiatry 1970; 25:238–251. 22. Yoshida T, Miyazaki H, Moriyama T, Nakashima H. Stereotactic operations with the help of a skull x-ray apparatus. Confin Neurol 1975; 37:291–295. 23. Hodges PC, Garcia-Bengochea F. Precise alignment of x-ray beams for stereotactic surgery. Am J Roentgenol Radium Ther Nucl Med 1969; 105:260–269. 24. Okudera H, Kyoshima K, Kobayashi S, Sugita K. Intraoperative CT scan findings during resection of glial tumours. Neurol Res 1994; 16:265–267. 25. Uematsu S, Rosenbaum AE, Erozan YS, Gupta PK, Moses H, Nauta HJ, et al. Intraoperative CT monitoring during stereotactic brain surgery. Acta Neurochir Suppl (Wien) 1987; 39:18–20. 26. Shalit MN, Israeli Y, Matz S, Cohen ML. Experience with intraoperative CT scanning in brain tumors. Surg Neurol 1982; 17:376–382. 27. Lunsford LD, Parrish R, Albright L. Intraoperative imaging with a therapeutic computed tomographic scanner. Neurosurgery 1984; 15:559–561. 28. Lunsford LD, Coffey RJ, Cojocaru T, Leksell D. Image-guided stereotactic surgery: a 10-year evolutionary experience. Stereotact Funct Neurosurg 1990; 54–55:375–387. 29. Albert FK, Forsting M, Sartor K, Adams HP, Kunze S. Early postoperative magnetic resonance imaging after resection of malignant glioma: objective evaluation of residual tumor and its influence on regrowth and prognosis. Neurosurgery 1994; 34:45–60; discussion 60–61. 30. Barnett GH, Miller DW, Weisenberger J. Frameless stereotaxy with scalp-applied fiducial markers for brain biopsy procedures: experience in 218 cases. J Neurosurg 1999; 91:569–576. 31. Kikinis R, Gleason PL, Moriarty TM, Moore MR, Alexander E III, Stieg PE, et al. Computer-assisted interactive three-dimensional planning for neurosurgical procedures. Neurosurgery 1996; 38:640–649; discussion 649–651.

106

&

PART II: TECHNOLOGIC APPLICATIONS

32. Maciunas RJ, Berger MS, Copeland B, Mayberg MR, Selker R, Allen GS. A technique for interactive image-guided neurosurgical intervention in primary brain tumors. Neurosurg Clin N Am 1996; 7:245–266. 33. Perry JH, Rosenbaum AE, Lunsford LD, Swink CA, Zorub DS. Computed tomography/ guided stereotactic surgery: conception and development of a new stereotactic methodology. Neurosurgery 1980; 7:376–381. 34. Hill DL, Maurer CR Jr, Maciunas RJ, Barwise JA, Fitzpatrick JM, Wang MY. Measurement of intraoperative brain surface deformation under a craniotomy. Neurosurgery 1998; 43:514–526; discussion 527–528. 35. Roberts DW, Hartov A, Kennedy FE, Miga MI, Paulsen KD. Intraoperative brain shift and deformation: a quantitative analysis of cortical displacement in 28 cases. Neurosurgery 1998; 43:749–758; discussion 758–760. 36. Roberts TS, Brown R. Technical and clinical aspects of CT-directed stereotaxis. Appl Neurophysiol 1980; 43:170–171. 37. Rubino GJ, Farahani K, McGill D, Van De Wiele B, Villablanca JP, Wang-Mathieson A. Magnetic resonance imaging-guided neurosurgery in the magnetic fringe fields: the next step in neuronavigation. Neurosurgery 2000; 46:643–653; discussion 653–654. 38. Nabavi A, Black PM, Gering DT, Westin CF, Mehta V, Pergolizzi RS Jr, et al. Serial intraoperative magnetic resonance imaging of brain shift. Neurosurgery 2001; 48:787–797; discussion 797–798. 39. Nimsky C, Ganslandt O, Cerny S, Hastreiter P, Greiner G, Fahlbusch R. Quantification of, visualization of, and compensation for brain shift using intraoperative magnetic resonance imaging. Neurosurgery 2000; 47:1070–1079; discussion 1079–1080. 40. Dohrmann GJ, Rubin JM. Dynamic intraoperative imaging and instrumentation of brain and spinal cord using ultrasound. Neurol Clin 1985; 3:425–437. 41. White DN. The early development of neurosonology: I. Echoencephalography in adults. Ultrasound Med Biol 1992; 18:115–165. 42. Chandler WF, Knake JE, McGillicuddy JE, Lillehei KO, Silver TM. Intraoperative use of real-time ultrasonography in neurosurgery. J Neurosurg 1982; 57:157–163. 43. Bohinski RJ, Kokkino AK, Warnick RE, Gaskill-Shipley MF, Kormos DW, Lukin RR, et al. Glioma resection in a shared-resource magnetic resonance operating room after optimal image-guided frameless stereotactic resection. Neurosurgery 2001; 48:731–742; discussion 742–744. 44. Kaibara T, Saunders JK, Sutherland GR. Advances in mobile intraoperative magnetic resonance imaging. Neurosurgery 2000; 47:131–137; discussion 137–138. 45. Wirtz CR, Bonsanto MM, Knauth M, Tronnier VM, Albert FK, Staubert A, et al. Intraoperative magnetic resonance imaging to update interactive navigation in neurosurgery: method and preliminary experience. Comput Aided Surg 1997; 2:172–179. 46. Wirtz CR, Knauth M, Staubert A, Bonsanto MM, Sartor K, Kunze S, et al. Clinical evaluation and follow-up results for intraoperative magnetic resonance imaging in neurosurgery. Neurosurgery 2000; 46:1112–1120; discussion 1120–1122. 47. Curran WJ Jr, Scott CB, Horton J, Nelson JS, Weinstein AS, Nelson DF, et al. Does extent of surgery influence outcome for astrocytoma with atypical or anaplastic foci (AAF)? A report from three Radiation Therapy Oncology Group (RTOG) trials. J Neurooncol 1992; 12:219–227.

CHAPTER 6: INTRAOPERATIVE MRI

&

107

48. Simpson JR, Horton J, Scott C, Curran WJ, Rubin P, Fischbach J, et al. Influence of location and extent of surgical resection on survival of patients with glioblastoma multiforme: results of three consecutive Radiation Therapy Oncology Group (RTOG) clinical trials. Int J Radiat Oncol Biol Phys 1993; 26:239–244. 49. Berger MS, Deliganis AV, Dobbins J, Keles GE. The effect of extent of resection on recurrence in patients with low grade cerebral hemisphere gliomas. Cancer 1994; 74:1784–1791. 50. Schwartz RB, Hsu L, Wong TZ, Kacher DF, Zamani AA, Black PM, et al. Intraoperative MR imaging guidance for intracranial neurosurgery: experience with the first 200 cases. Radiology 1999; 211:477–488. 51. Gronemeyer DH, Kaufman L, Rothschild P, Seibel RM. New possibilities and aspects of low-field magnetic resonance tomography. Radiol Diagen (Berl) 1989; 30:519–527. 52. Duckwiler G, Lufkin RB, Teresi L, Spickler E, Dion J, Vinuela F, et al. Head and neck lesions: MR-guided aspiration biopsy. Radiology 1989; 170:519–522. 53. Lufkin R, Duckwiler G, Spickler E, Teresi L, Chang M, Onik G. MR body stereotaxis: an aid for MR-guided biopsies. J Comput Assist Tomogr 1988; 12:1088–1089. 54. Thomas DG, Davis CH, Ingram S, Olney JS, Bydder GM, Young IR. Stereotaxic biopsy of the brain under MR imaging control. Am J Neuroradiol 1986; 7:161–163. 55. Mueller PR, Stark DD, Simeone JF, Saini S, Butch RJ, Edelman RR, et al. MR-guided aspiration biopsy: needle design and clinical trials. Radiology 1986; 161:605–609. 56. Lufkin R, Teresi L, Hanafee W. New needle for MR-guided aspiration cytology of the head and neck. Am J Roentgenol 1987; 149:380–382. 57. Herman SD, Friedman AC, Radecki PD, Caroline DF. Incidental prostatic carcinoma detected by MRI and diagnosed by MRI/CT-guided biopsy. Am J Roentgenol 1986; 146:351–352. 58. Alexander E III, Moriarty TM, Kikinis R, Jolesz FA. Innovations in minimalism: intraoperative MRI. Clin Neurosurg 1996; 43:338–352. 59. Hall WA, Martin AJ, Liu H, Pozza CH, Casey SO, Michel E, et al. High-field strength interventional magnetic resonance imaging for pediatric neurosurgery. Pediatr Neurosurg 1998; 29:253–259. 60. Hall WA, Martin AJ, Liu H, Nussbaum ES, Maxwell RE, Truwit CL. Brain biopsy using high-field strength interventional magnetic resonance imaging. Neurosurgery 1999; 44:807–813; discussion 813–814. 61. Hall WA, Liu H, Martin AJ, Pozza CH, Maxwell RE, Truwit CL. Safety, efficacy, and functionality of high-field strength interventional magnetic resonance imaging for neurosurgery. Neurosurgery 2000; 46:632–641; discussion 641–642. 62. Tronnier VM, Wirtz CR, Knauth M, Lenz G, Pastyr O, Bonsanto MM, et al. Intraoperative diagnostic and interventional magnetic resonance imaging in neurosurgery. Neurosurgery 1997; 40:891–900; discussion 900–902. 63. Schad LR, Ehricke HH, Wowra B, Layer G, Engenhart R, Kauczor HU, et al. Correction of spatial distortion in magnetic resonance angiography for radiosurgical treatment planning of cerebral arteriovenous malformations. Magn Reson Imaging 1992; 10:609–621.

108

&

PART II: TECHNOLOGIC APPLICATIONS

64. Sutherland GR, Kaibara T, Louw DF. Intraoperative MR at 1.5 Tesla–experience and future directions. Acta Neurochir Suppl 2003; 85:21–28. 65. Sutherland GR, Louw DF. Intraoperative MRI: a moving magnet. CMAJ 1999; 161: 1293. 66. Schulder M, Liang D, Carmel PW. Cranial surgery navigation aided by a compact intraoperative magnetic resonance imager. J Neurosurg 2001; 94:936–945. 67. CDRH Magnetic Resonance Working Group. A Primer on Medical Device Interactions with Magnetic Resonance Imaging Systems. 97. 2003. 68. Anonymous. Thermal Injuries and Patient Monitoring during MRI Studies. Health Devices 1991; 20:362–363. 69. Schenck JF. The role of magnetic susceptibility in magnetic resonance imaging: MRI magnetic compatibility of the first and second kinds. Med Phys 1996; 23:815–850. 70. Berkenstadt H, Perel A, Ram Z, Feldman Z, Nahtomi-Shick O, Hadani M. Anesthesia for magnetic resonance guided neurosurgery: initial experience with a new open magnetic resonance imaging system. J Neurosurg Anesthesiol 2001; 13:158–162. 71. Knauth M, Aras N, Wirtz CR, Dorfler A, Engelhorn T, Sartor K. Surgically induced intracranial contrast enhancement: potential source of diagnostic error in intraoperative MR imaging. Am J Neuroradiol 1999; 20:1547–1553. 72. Keles GE, Anderson B, Berger MS. The effect of extent of resection on time to tumor progression and survival in patients with glioblastoma multiforme of the cerebral hemisphere. Surg Neurol 1999; 52:371–379. 73. Keles GE, Lamborn KR, Berger MS. Low-grade hemispheric gliomas in adults: a critical review of extent of resection as a factor influencing outcome. J Neurosurg 2001; 95:735–745. 74. Soo TM, Bernstein M, Provias J, Tasker R, Lozano A, Guha A. Failed stereotactic biopsy in a series of 518 cases. Stereotact Funct Neurosurg 1995; 64:183–196. 75. Brainard JA, Prayson RA, Barnett GH. Frozen section evaluation of stereotactic brain biopsies: diagnostic yield at the stereotactic target position in 188 cases. Arch Pathol Lab Med 1997; 121:481–484. 76. Moriarty TM, Kikinis R, Jolesz FA, Black PM, Alexander E III. Magnetic resonance imaging therapy. Intraoperative MR imaging. Neurosurg Clin N Am 1996; 7: 323–331.

7

AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING Lilyana Angelov Department of Neurological Surgery and Brain Tumor Institute, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Gene H. Barnett Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION Methods to localize and, ultimately, protect functional cortex through direct cortical and subcortical stimulation as well as electrocorticography have long been recognized in neurosurgical practice. Awake craniotomy with intraoperative brain mapping has, however, evolved over the past few decades to become the gold standard of physiologic monitoring when attempting resection of lesions in or near eloquent brain. Even more dramatic are recent advances in neuroimaging. Originally, neurosurgeons relied on indirect and imprecise information about intracranial lesion location as inferred through perturbations of normal anatomy as seen on pneumoencephalograms or angiograms. Today’s imaging afford us exquisitely detailed information about the anatomical features of lesions, their surrounding structures, and some insights into the likely pathology they represent. Surgical navigation exploits this anatomic information to provide localization, orientation, and guidance, typically using preoperatively acquired image data. Navigation using intraoperatively acquired imaging, particularly using intraoperative magnetic resonance imaging (iMRI), potentially provides critical updates to this anatomic data, but presents unique logistical challenges. This chapter will outline many aspects of these two modalities, presenting indications for when and how they can be applied and ultimately integrated as we strive to improve the success and safety of neurosurgical procedures. While much of this chapter is generally applicable, the discussion will be focused on these strategies as they directly apply to tumor resections.

Icons indicate materials available on the accompanying CD.

109

110

&

PART II: TECHNOLOGIC APPLICATION

Indications Awake craniotomy is most commonly indicated for patients with a lesion arising in or near critical areas of the brain. It allows for brain mapping and hence preservation of critical areas of eloquent cortex and its associated neurological function during neurosurgical procedures. The technique of cortical mapping of speech and motor areas was originally pioneered in epilepsy surgery. However, its relative ease and significant advantage in decreasing morbidity has allowed it to be applied in the resection of tumors or vascular lesions within or bordering language or sensorimotor regions. In the area of the primary motor cortex, anatomic identification alone is often inadequate for safe resections in this location. This is due to the fact that the central sulcus cannot be consistently visually identified intraoperatively, even with extensive surgical exposure (1,2). Further, the primary motor area may extend, more than 20 mm anterior, to the rolandic fissure in patients with intra-axial tumors (3). However, these problems may be mitigated by use of surgical navigation and the fact that the central sulcus can be reliably located using somatosensory evoked potentials where the sulcus is marked by a phase reversal. As such, awake craniotomy is used rarely, solely for identification of central cortex. In dominant hemisphere perisylvian lesions, the identification of the cortical sites for language is more problematic. There are significant variations in language representation among individuals making reliable anatomic localization impossible (4,5). Moreover, there is no proven method for identifying essential language centers in a patient under general anesthesia either by visual inspection alone or electrophysiologically. Thus, awake craniotomy is the approach of choice for resection of these lesions. Mapping requires a fully awake patient where the cerebral cortex can be electrically stimulated directly in order to disrupt the performance of various tasks and thereby better identify areas essential to normal functioning such as sensory, motor, and language pathways. In this way, postoperative deficits can be minimized or avoided through the preservation of functional neural tissue. In the context of tumor surgery, maximizing resection while preserving normal brain function is an important objective of the surgery. The impact of maximal tumor resection of gliomas on the interval to tumor progression and overall outcome remains controversial. Several studies, however, demonstrate the benefit of radical tumor resection on outcome in both high- and low-grade gliomas (6–12). Further, factors such as small lesions, location near eloquent cortex and proximity to vital structures make surgical resections additionally challenging. Other features that may make tumor resection more difficult include large expansive masses or cases where reoperation is required and the anatomy is distorted from prior treatments. Another important issue in tumor surgery is that new neurological deficits cannot be reliably avoided simply by debulking infiltrating tumors from within, as gliomas (particularly low-grade tumors) have been reported to contain functional tissue; hence, ancillary techniques are required to minimize morbidity while trying to maximize tumor resection (13,14). Consequently, awake craniotomy plays a major role in the safe resection of lesions that are otherwise considered inoperable using conventional techniques due to their high risk of attendant neurological morbidity.

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

111

While the predominant advantage of awake craniotomy is that it allows for brain mapping, another benefit is that it avoids general anesthesia and the potential complications associated with it. Intraoperative endotracheal intubation and general anesthetic patients monitoring devices such as arterial lines, central venous lines, and indwelling urethral catheters and their associated complications, which are used during general anesthesia, can often be avoided in awake patients. Also, in the absence of general anesthesia, patients generally regain an alert conscious state shortly after the termination of the procedure often resulting in a shorter postoperative hospital stay, which is advantageous for both the patient and the healthcare system. Awake craniotomy is also advantageous in the setting of patients with significant medical comorbidities, decreasing the risk of surgery under general anesthesia in these individuals. These benefits, however, may be offset by the relative loss of control over the patient during the operation. Agitated patients may be particularly troublesome during awake procedures, and it may be very difficult to control brain swelling during these procedures. Also, not all patients are appropriate candidates for an awake craniotomy procedure. Patients are not suitable for awake craniotomy if they are unable to cooperate due to profound dysphasia, significant language barrier, confusion, decreased level of consciousness, mental retardation, or emotional instability (15). Relative contraindications include low occipital lesions that may require the placement of the patient in the prone position, which is uncomfortable for the patient during an awake craniotomy, patients who present acutely with increase in intracranial pressure as they are typically unable to cooperate completely during an awake procedure, and pediatric patients younger than 11 years of age (15,16).

ANESTHETIC TECHNIQUES Many anesthetic techniques have been used for awake neurosurgical procedures, as the challenge is to provide analgesia and sedation for periods of intense surgical stimulation needed for cranial fixation, scalp, and muscle incision and dissection, bone flap removal and dural handling while having an alert and cooperative patient during functional testing. A combination of local anesthesia and conscious sedation with a variety of agents including droperidol, short-acting narcotics, or propofol infusion is now most commonly used for these procedures (17–22). Generally, these agents provide appropriate analgesia and sedation, allow for rapid recovery of consciousness when required, and interfere minimally with electrocorticography. At its inception, one major concern with the technique was the potential for airway problems occurring during the craniotomy, resulting in brain swelling and difficult airway access. During these cases, the airway is typically unsecured or maintained with a nasal airway. This is an issue since attempts to better control any surgical pain with increased intravenous sedation may result in apnea or airway obstruction. Hence, careful titration of the intravenous agent and appropriate administration of local anesthetic as required are important considerations.

112

&

PART II: TECHNOLOGIC APPLICATION

Another issue is that while such combination techniques are commonly used at most centers when performing awake craniotomies, the procedure may occasionally have to be aborted due to pain, poor patient cooperation such as with postictal confusion, brain swelling requiring hyperventilation, or airway obstruction (17–19,23). Early reports in the literature suggested that unexpected intubation may be required in 2% to 16% of the cases, although this rate may be lower as experience with the technique grows (17–19,24). In this context, emergent intubation can be rather difficult in these cases at times. The anesthetist is required to navigate an unfamiliar airway, while the patient is often in a non-neutral position, immobilized in pin fixation with their face covered by surgical drapes. Several strategies to deal with this issue have been proposed in the literature including blind nasal intubation (25), awake fiberoptic intubation followed by general anesthesia (23), or a laryngeal mask airway (26). However, for the most part, recent studies suggest that normally no significant anesthetic complications arise during the procedure as both neurosurgeons and neuroanesthesiologists become more experienced with the techniques and surgical times decrease (15,27). From the patient’s perspective, while the awake craniotomy procedure is more challenging than having the same procedure performed under general anesthesia, these procedures appear to be generally well tolerated (27). At our institution, patients are typically brought to the operating room and placed in the supine or lateral position while fully awake to ensure patient comfort, as they must maintain this position for the duration of the procedure (Fig. 1). The lateral position is preferred as it thwarts the tendency of supine patients to try to ‘‘sit up’’ when awakening. Careful attention is given to padding pressure areas and avoiding an uncomfortable position. Significant lateral rotation of the neck is avoided. Intravenous access is established and a nasal cannula is placed for oxygen administration. Noninvasive blood pressure, cardiac and oxygen saturation monitors are also applied. Patients are then sedated with intravenous sedation by the anesthetist. Local anesthetic is used prior to head-frame placement and a dynamic reference frame (intraoperative neuronavigation tracking device) is affixed to the Mayfield head frame. The head frame itself is not secured to the OR table, rather the head frame and the head rest on a pillow, in order to allow some patient movement aiding in patient comfort over the long term. Neuronavigation with head position registration is performed in a standard fashion and a minimal-access incision is planned. We then awaken the patient so that we can gauge their reaction to emergence and the patient can experience it before doing so during surgery. The patient is resedated and infiltrated with local anesthetic as guided by the preoperative imaging. A standard craniotomy is then performed with the patient sedated. Sterile drapes are placed such that the anesthesiologist and neurophysiologist have appropriate access to the patient. The patient is kept deeply sedated during the incision and craniotomy and then allowed to lighten after the dura is opened. Hemodynamic and respiratory parameters as well as patient discomfort are used to adjust the depth of conscious sedation. Sedation is decreased and the patient is completely alert at the critical portion of the surgery where patient participation is required for motor, sensory, or language testing and evaluation. Complete patient cooperation is essential for the success of the surgery. When the relevant surgical resection is complete, the patient’s sedation is once again increased to allow the procedure to be completed with minimal patient discomfort. At the end of the case, emergence from conscious is typically rapid allowing

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

113

Figure 1 In many cases, the patient is positioned in the lateral position with the head and neck in a neutral position relative to the body allowing for anesthetic monitoring, surgeon access, and patient comfort during the procedure. Note that the neuronavigation system dynamic reference frame is attached directly to the head frame; however, the entire unit is not fixed to the bed, which allows for some patient movement during the procedure increasing the patient’s overall comfort without compromising the ability to use image guidance during the procedure.

prompt and thorough evaluation of the patient’s neurological status. Conversely, if the patient is not immediately fully alert, the surgeon is more quickly alerted to a perioperative complication and appropriate measures can be undertaken.

INTRAOPERATIVE FUNCTIONAL MAPPING Undoubtedly, the main reason for performing an awake craniotomy is when resection of a lesion in or near eloquent cortex is intended. While anatomical landmarks identified on MRI and related to standard brain atlases may provide some general information about functional localization, differences between individuals can be significant (3,28). Furthermore, the local anatomy may be distorted in patients with large intracranial masses, making it difficult to identify important neuroanatomic landmarks accurately based on preoperative MR images alone. In these cases, functional cortical and subcortical mapping is performed in an awake patient to identify anatomic areas related to sensory, motor, or speech function. This remains the gold standard for the accurate assessment of functional cerebral cortex. In this way, intraoperative functional mapping prior to and during resection assists in decreasing

114

&

PART II: TECHNOLOGIC APPLICATION

the surgical morbidity and postoperative neurological deficits, while allowing for more extensive resection of lesions that were previously considered inoperable. Functional mapping may include sensory cortex identification, definition of sensorimotor transition zone, delimitation of the motor cortex, and cortical surface stimulation (Fig. 2) (29–31). The sensorimotor cortex, can often be identified using somatosensory evoked potentials. This can be performed using a subdural grid array, which is also used for initial electrocorticography. The central sulcus can often be identified using somatosensory evoked potentials and a subdural strip electrode. In this setting, a phase reversal wave can frequently be seen at the sensorimotor transition across the central sulcus and the precentral motor cortex. (This methodology does not require the patient to be awake.) In addition, direct cortical stimulation with a monopolar or bipolar stimulator can be performed. For the duration of stimulation, the patient is asked to report any evoked sensations. Motor responses can be evaluated using either clinical observation during cortical stimulation or monitoring for electromyography changes. The patient is observed for spontaneous movements of the face, arm, or leg and formal motor testing is also performed. Speech monitoring is performed for resections in the dominant hemisphere. Any speech arrest or dysphasia suggests that critical speech is represented in that area of the cortex. Mapping in the dominant parietal areas monitors the reception of auditory stimuli. While difficult and at times apparently incongruous, awake speech mapping is indispensable as the distance of the resection area from language localization (as determined by

Figure 2 Cortical mapping can be performed in an awake patient through a small craniotomy. A subdural strip electrode is seen in the center of the field as well as a bipolar stimulator, which are used to evaluate eloquent cortex and guide resection during surgery.

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

115

intraoperative cortical stimulation) is the most important variable in predicting absence or recovery from postoperative aphasia (4). Subcortical stimulation can also be used in addition to cortical mapping in order to map eloquent subcortical white matter such as descending motor pathways (32). Electrocorticography can also be used to detect and resect perilesional epileptiform foci potentially improving seizure control postoperatively (33). During the stimulation procedure, electrocorticography is maintained to monitor for afterdischarges. Intraoperative simple partial seizures may arise during brain mapping in 5% to 20% of cases (31). Traditionally, these seizures had been treated with intravenous short-acting barbiturates with antiepileptic properties. The drawback with these agents was that their anticonvulsant effect was somewhat delayed, temporary postictal cortical inactivation occurred, and the patient often became more lethargic interfering with language mapping. Today, topical cortical irrigation with cold Ringer’s solution (4 C) is used to stop the seizure within seconds and also allows the cortex to be immediately restimulated as required in order to continue with intraoperative mapping (34). The information obtained through functional mapping permits surgeons to plan a safe avenue to the tumor allowing for maximum resection while decreasing the risk of neurological morbidity. A corticotomy is carried out avoiding all sites that are identified as eloquent cortex during mapping. The lesion is resected using standard neurosurgical techniques with functional mapping continued throughout the resection process. Although optimal use of surgical navigation suggests an en bloc technique of lesion resection, awake procedures are best performed using gradual resection starting at the least eloquent brain and advancing toward the most hazardous regions. Areas that are identified as functional are not resected even when they are grossly involved with tumor. If neurological changes occur, the resection is stopped and any retractors are released or repositioned. In the majority of cases, negative mapping reassures the surgeon that the chosen transcortical approach and the degree of resection is safe. However, the absence of detectible brain mapping does not always guarantee the absence of functioning cortex resulting in neurological deficits post-op despite negative intraoperative mapping, nor does it address the resectability of subcortical brain (15). Moreover, the presence of a preoperative neurological deficit is known to increase the difficulty and decrease the reliability of brain mapping (35). Overall, however, intraoperative mapping of eloquent cortex during craniotomy with the patient undergoing monitored conscious sedation and local anesthesia is a wellestablished and safe technique allowing for maximum tumor excision involving eloquent cortex while minimizing neurological deficits (4,33,36–39). Nonoperative brain mapping can also be performed using a variety of imaging techniques and while it does not represent the gold standard in localizing functional brain, it can compliment information obtained through preoperative neuronavigational imaging and awake intraoperative cortical mapping. Positron emission tomography (PET) can demonstrate regions of metabolic activity in functional cortex in response to activity. The disadvantage, however, is that PET is not available in most centers, is costly, requires injection of radioactive material, and is associated with poor image resolution. A more common imaging modality used to localize cortical activity is functional magnetic resonance imaging (fMRI). fMRI is

116

&

PART II: TECHNOLOGIC APPLICATION

a noninvasive technique that detects an increase in T2-weighted MR signal related to a decrease in local deoxyhemoglobin concentration. This arises due to a mismatch between regional cerebral blood flow and oxygen extraction when a specific task is being performed (40,41). An advantage of fMRI is that it noninvasively provides the surgeon with additional information about the local functional organization of the brain before surgery and may aid in deciding whether to biopsy or resect a lesion, assist in planning the surgical approach, and help to define realistic goals and limits in the resection of the lesion. Several reports have demonstrated the value of fMRI in the localization of functionally eloquent cerebral cortex including motor, somatosensory, language, and visual areas, as confirmed by comparing the activation maps generated with fMRI and anatomic localization intraoperatively with cortical stimulation (42–46). Nevertheless, some controversies exist especially in language mapping due to uncertainty regarding the exact techniques and interpretation of language-related fMRI. We have had several cases where fMRI inaccurately localized motor or lateralized speech function. Also, visualization of fMRI activity does not necessarily indicate that the region is critical to that function. Hence, it is currently felt to lack sensitivity and specificity in language mapping and is somewhat unreliable. Further, fMRI also suffers from the limitations of other static neuronavigation approaches in that it is unable to give real-time data and also gives no information about subcortical eloquent areas. Thus, at present, we perceive fMRI’s main role to be that of a guide, complimenting and enhancing information obtained through other modalities.

STATIC IMAGE-GUIDED RESECTION In an attempt to obtain accurate image-guided information during neurosurgical procedures, a variety of different systems have been developed to compliment and supplement information obtained during awake craniotomy and functional mapping. While image-guided resections were initially pioneered in the setting of craniotomy under general anesthesia, there has been a seamless transition of this technology for procedures involving awake craniotomy using conscious sedation. Although this subject is discussed extensively in other chapters, a brief review is warranted. Computer-assisted image-guided neurosurgery was first introduced in the 1980s and has now ultimately become an integral component of many neurosurgical procedures (47–50). Frameless stereotaxy enables the neurosurgeon to accurately localize areas of the brain and lesions in this area by relating fixed points on the skull and face to regions of interest on an MR or computed tomography (CT) image without applying a conventional steriotactic head frame. Frameless steriotactic devices have a spatial accuracy of several millimeters or better and are hence appropriate for guiding craniotomies and intradural surgery (51–53). Image-guided resections involve three major components: (i) image data acquisition with coordinate space definition from one or several imaging modalities; (ii) preoperative surgical planning and simulation of the surgical approach; and (iii) the surgical procedure itself, which involves integration of the image and the surgical coordinate space, as well as localization and resection of the lesion.

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

117

At our center, image-guided resections are performed in a fairly standardized fashion. The day prior to surgery, 8 to 10 scalp fiducials are attached to the patient’s scalp in a scattered pattern for referencing. The patient then undergoes a volumetric T1-weighted post-Gd-DTPA MRI with 1 mm isotropic voxels. These 3-D MRI data are then transferred via Ethernet to the operating room navigation system. On the day of surgery, the patient’s head is positioned in the Mayfield head frame and the dynamic reference frame is attached to the Mayfield head holder, which is left free of the base, providing rigid fixation to the head while still allowing the patient to maintain a wide range of freedom of movement and flexibility. The fiducial markers are then registered relative to the MRI images according to the requirement of the specific neuronavigation system. Other available modalities such as CT, further MRI sequences, fMRI, PET, and SPECT are also sent to the intraoperative computer workstation as required, where they can be coregistered with the new fiducial containing MRI and used for intraoperative planning. The computer program then creates a virtual patient environment, which includes a geometric analysis of patient data, stereotactic position measurements, distance measurements, and area measurements in order to plan the surgical approach. Various 3-D and 2-D displays such as oblique and trajectory view are analyzed and utilized to approach the lesion as well as to maximize resection. The size and shape of the craniotomy is determined at this stage such that only as much brain as necessary is exposed in order to allow resection and permit for placement of monitoring devices for intraoperative mapping. The incision is marked out, the area shaved, prepped and draped and a standard craniotomy, either under general anesthesia or conscious sedation, is performed as required.

DYNAMIC INTRAOPERATIVE IMAGE-GUIDED RESECTION While conventional neuronavigation systems have clearly revolutionized the planning and performance of a variety of neurosurgical procedures, they also have the obvious limitation of being unable to provide any information about dynamic changes that occur during surgery. New approaches have thus explored ways to incorporate intraoperative imaging modalities into surgical navigation that would better represent up-to-the-minute image information throughout the surgical procedure. Strategies were explored that would be able to update neuronavigation images acquired preoperatively while demonstrating the current extent of tumor resection and compensating for any intraoperative brain shift or local tissue deformations. The first reported application of dynamic intraoperative imaging involved (CT) and was reported in the early 1980s (54,55). Using this technique, the biopsy target, extent of resection, or intraoperative hemorrhage in the tumor bed could be detected. However, while the technique was rapid and relatively inexpensive, CT did not develop into the intraoperative imaging modality of choice due to the associated ionizing radiation, scan times were long, and the fact that its resolution was less than adequate for interactive image-guided surgery. It has, thus, for the most part been largely abandoned, although there is some renewed interest in the technique with a new generation of portable intraoperative CT imagers.

118

&

PART II: TECHNOLOGIC APPLICATION

By contrast, intraoperative ultrasound is an excellent, readily affordable, and straightforward technique that can be applied at various stages of the neurosurgical procedure (56–58). It provides real-time guidance to the neurosurgeon with respect to the location and trajectory of the lesion as well as extent of resection. Further, intraoperative ultrasonic images can be obtained in stereotactic coordinates and linked directly to the preoperatively acquired MRI or CT images on the neuronavigation system (59–61). A substantial drawback of intraoperative ultrasound is that image quality is often poor, although greatly improved as the technology has developed. Further, while in the majority of cases it has been found to accurately represent the extent of tumor resection, this does not apply when used in lesions that demonstrate radiation effects from previous treatments (56). As well, a thin hyperechoic rim surrounding the resection cavity (less than 3 mm in diameter) can occasionally occur as a nonspecific finding potentially masking a thin layer of residual tumor (58). Finally, surgeons tend to be more familiar with images obtained on CT or MRI scans, consequently experiencing some difficulty in interpreting the ultrasound images.

INTRAOPERATIVE USE OF MAGNETIC RESONANCE IMAGING As noted in the previous section, some common concerns with navigational systems using preoperative images are the brain deformations produced by partial resection of space occupying lesions, and shifts due to the drainage of cerebrospinal fluid or swelling related to resection. In an attempt to improve the degree of surgical accuracy and minimize the errors caused by this brain shift, intraoperative MRI was developed and first reported in 1995 (62–64). This state-of-the-art modality is able to provide high-quality intraoperative near-real-time neuronavigation images throughout the procedure allowing for up-to-the-minute evaluation of surgical progress including redefining the location of the lesion and the current extent of resection (Fig. 3). Further, recent modifications in magnetic resonance imaging scanners and their software allows for high-resolution multiplanar scanning and has enhanced our ability to utilize instruments under continuous guidance defining their position in relation to the lesion or critical structures. The subject of intraoperative MRI is addressed in detail in the previous chapter. There are several ways to categorize these systems, one of which is by field strength: high (0.5 T or higher) and low (less than 0.5 T). Examples of high-field strength systems include the superconductive magnets used by General Electric BrainLAB. Lower-field systems include modified open MRIs using resistive magnets (Hitachi, Philips, Picker, Siemens, Fonar, Odin). These systems have variable capabilities to be combined with image-guided neuronavigation. High-field strength systems produce rapid high-quality full head images and some may be used to perform intraoperative angiography and spectroscopy. However, drawbacks of this type of intraoperative magnet include a high investment and running cost, limited surgical working space, the need to redesign and shield the operating room, increased risk of injury due to attraction of ferrous objects,

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

119

Figure 3 Intraoperative MRI being performed. In the foreground, the PoleStar N10 magnet can be seen draped and positioned below the field allowing for surgical access during the procedure with standard instruments. The magnet is then raised into the field and images are acquired during surgery as required.

and the development of specially modified operating room equipment, which includes nonmagnetic anesthetic machines, laryngoscopes, head holders, and MR-compatible surgical and microsurgical instruments with appropriate tracking systems. Some low-field strength MRI systems have many of the same liabilities as their high-field cousins. One ultra-low-field (0.12 T) system (PoleStar N10, Odin Medical Technologies, Newton, Massachusetts, U.S.A.) is significantly smaller, more light weight, portable, and more cost effective than the higher-field strength magnets. Further, it allows the surgeon better access to the patient during the procedure. Also, the combination of low magnetic field strength with the ability to move the magnet away from the operative field allows for the use of standard instruments throughout much of the procedure and to perform cortical mapping and stimulation. Using conventional instruments is often beneficial in terms of cost, mechanical ‘‘feel’’ and the diversity of choice. Some drawbacks in using the low-field strength MRI units include sacrifice in field of view, decreased image quality, and increased acquisition time, although advocates of these systems suggest that the image provided is sufficient for their intended intraoperative use (65). However, while there are many obvious and significant advancements related to the use of intraoperative MRI systems, as in any novel and developing technology certain limitations do exist. Paramount among these is the significant

120

&

PART II: TECHNOLOGIC APPLICATION

expense of purchasing, installing, and running one of these systems, making their general application prohibitive in many centers. A further drawback in the use of intraoperative MR imaging is that additional surgical time is required with an average intraoperative imaging time of 35.5 minutes for a standard craniotomy with a range of 7.5 to 60.5 minutes (65). In general, the fastest sequence that can be obtained is a T2-weighted sequence with a one-minute acquisition time, but resolution is often sacrificed with rapid studies. This may become particularly relevant when used in the awake patient setting, where at the end of a case the patient may be somewhat more restless and uncomfortable and the longer motionless time to perform a T1-Gd study can become an issue if the patient is required to be awake for further residual tumor resection. To this end, resection margins may themselves be difficult to interpret. Surgically induced artifacts can give rise to contrast enhancement in areas where no tumor was seen on preoperative imaging. These areas may be misjudged as residual tumor, resulting in more extensive resections than required. Finally, there are some software limitations with the different systems such that neuronavigation or simultaneous fusion with information from other modalities such as digital angiography, spectroscopy, or PET cannot be performed.

Awake Procedures Using iMRI The combination of iMRI in the awake patient setting has significant and obvious merits and the smooth integration of these modalities in the operating room is the task at hand in the coming years. Intraoperative magnets, which can provide increased surgical working space while allowing for a greater variety of surgical positioning options, enhancing patient comfort without compromising field of view or image quality, will be invaluable in the merging of these approaches. While currently iMRI allows the surgeon to evaluate movement and distortion of the brain during surgery, not all iMRI systems have neuronavigation capability allowing for direct comparison to the preoperative image data. As the systems and software evolve, this integration of information will provide an important reality check on the extent of resection and enhance the overall safety of the resection. In the future, iMRI navigation systems will further be able to coregister macroscopic atlas data such as the visible human data set morphed to match the patients’ own images providing a new level of information about a given areas’ function, connections, and vascular supply, making this a very powerful adjunct to safe resection in the future (Fig. 4). A recent study has proceeded to further integrate iMRI and awake craniotomy by demonstrating that accurate fMRI can be performed with a low-field PoleStar N-10 iMRI during surgery (66). This approach complements the information obtained with awake intraoperative direct cortical stimulation and overcomes the limitations relating to intraoperative brain shift and its comparison to data obtained preoperatively. At the present time, while cortical electrodes for awake mapping and intraoperative MRI are employed during the same case, they are not used concurrently.

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

121

Figure 4 In the future, multiple modalities will be used in the intraoperative setting to optimize safe resections of lesions. Work is ongoing to coregister intraopertive MRI images with the preoperative studies that have also been coregistered with a macroscopic atlas (in this case from the Visible Human Dataset) data set as represented in this synthesized figure.

This is related to the theoretical safety concern related to potential heating and injury to the underlying brain secondary to induction of electrical currents during magnetic resonance imaging. While the risk for electrode heating has not been directly evaluated in this context, it should be noted that patients with implanted neurostimulation devices routinely have MRIs. A recent abstract specifically addressed this question specifically evaluating the resultant heating at the electrode tips in the most commonly utilized neurostimulation systems (67). The heating in these systems ranged from 0.1 C to 11.9 C with the authors noting that the use of various coils, strength of the magnet used, the electrode configurations as well as a variety of other factors may effect the heating profile. Hence, no specific generalization about the safety of performing awake mapping with cortical electrodes during iMRI can be inferred from this work and is an area requiring further evaluation. Ideally, if the grid could be left in place during iMRI image acquisition and this information overlayed and integrated onto the pre- and intraoperatively acquired

122

&

PART II: TECHNOLOGIC APPLICATION

images it improves the accuracy and safety of the tumor resection. This powerful integration of information however awaits further research in the future.

THE FUTURE Image-guided surgery has become a standard of practice in neurosurgery at most centers. This has been largely spearheaded by the advantages provided by sophisticated cerebral anatomy imaging. When combined with the advanced computer technologies that support complex image acquisition and the display of high-quality complicated graphics often fusing several image modalities, surgeons are able to minimize the degree of ‘‘surgical exploration’’ undertaken, thereby decreasing morbidity and improving surgical outcome overall. Sophisticated image acquisition and merging strategies combined with awake craniotomy and cortical mapping continue to be developed and should allow for more aggressive surgical resection of tumors while continuing to improve the safety of these procedures by minimizing neurological deficits. Strategies to merge volumetric 2-D and 3-D MRI and CT images and actual intraoperative subdural electroencephalogram electrodes location (even when a portion of the grid is located below the bone flap) have been described and continue to evolve (68). As we strive to gain evermore precise and revealing information about the brain, specifically as it is perturbed by tumors, information from a myriad of modalities will inevitably be combined, ultimately enhancing our ability to control optimize safe resection of these lesions.

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

123

REFERENCES 1. Berger MS, Ojemann GA, Lettich E. Neurophysiological monitoring during astrocytoma surgery. Neurosurg Clin North Am 1990; 1:65–80. 2. King RB, Schell GR. Cortical localization and monitoring during cerebral operations. J Neurosurgery 1987; 67:210–219. 3. Uematsu S, Lesser R, Fisher RS, Gordon B, Hara K, Krauss GL, Vining EP, Webber RW. Motor and sensory cortes in humans: topography studied with chronic subdural stimulation. Neurosurgery 1992; 31:59–72. 4. Haglund MM, Berger MS, Shamseldin M, Lettich E, Ojemann GA. Cortical localization of temporal lobe language sites in patients with gliomas. Neurosurgery 1994; 34:567–576. 5. Schwartz TH, Devinsky O, Doyle W, Perrine K. Preoperative predictors of anterior temporal language areas. J Neurosurg 1998; 89:962–970. 6. Ammirati M, Galicich JH, Arbit E, Liao Y. Reoperation in the treatment of recurrent intracranial malignant gliomas. Neurosurgery 1987; 21:607–614. 7. Ammirati M, Vick N, Liao Y, Ciric I, Mikhael M. Effect of the extent of surgical resection on survival and quality of life in patients with supratentorial glioblastomas and anaplastic astrocytomas. Neurosurgery 1987; 21:201–206. 8. Laws ER Jr, Taylor WF, Clifton MB, Okazaki H. Neurosurgical management of low grade astrocytomas of the cerebral hemisphere. J Neurosurgery 1984; 61:665–673. 9. Quigley MR, Maroon JC. The relationship between survival and the extent of resection in patients with supratentorial malignant gliomas. Neurosurgery 1991; 29:385– 389. 10. Shaw EG, Scheithauer BW, O’Fallon JR, Tazelaar HD, Davis DH. Oligodendroglioas: the Mayo Clinic experience. J Neurosurg 1992; 76:428–434. 11. Keles GE, Anderson B, Berger MS. The effect of extent of resection on time to tumor progression and survival in patients with glioblastoma multiforme of the cerebral hemisphere. Surg Neurol 1999; 52:371–379. 12. Keles GE, Lamborn KR, Berger MS. Low-grade hemispheric gliomas in adults: a critical review of extent of resection as a factor influencing outcome. J Neurosurg 2001; 95:735–745. 13. Ojemann GA, Miller JW, Silbergeld DL. Preserved function in brain invaded by tumor. Neurosurgery 1996; 39:253–259. 14. Skirboll SS, Ojemann GA, Berger MS, Lettich E, Winn HR. Functional cortex and subcortical white matter located within gliomas. Neurosurgery 1996; 38:678–685. 15. Taylor MD, Bernstein M. Awake craniotomy with brain mapping as the routine surgical approach to treating patients with supratentorial intraaxial tumors: a prospective trial of 200 cases. J Neurosurg 1999; 90:35–41. 16. Berger MS. The impact of technical adjuncts in the surgical management of cerebral hemispheric low-grade gliomas of childhood. J Neurooncol 1996; 28:129–155. 17. Archer DP, McKenna JMA, Morin L, Ravussin P. Conscious sedation analgesia during craniotomy for intractable epilepsy. A review of 354 consecutive cases. Can J Anaesth 1988; 35:338–344.

124

&

PART II: TECHNOLOGIC APPLICATION

18. Gignac E, Manninen PH, Gelb AW. Comparison of fentanyl, sufentanil,and alfentanil during awake craniotomy for epilepsy. Can J Anaesth 1993; 40:421–424. 19. Manninen PH, Contreras J. Anesthetic considerations for craniotomy in awake patients. Int Anesthesiol Clin 1986; 24:157–174. 20. Silbergeld DL, Mueller WM, Colley PS, Ojemann GA, Lettich E. Use of propofol (Diprivan) for awake craniotomies: technical note. Surg Neurol 1992; 38:271–272. 21. Weiss FR, Schwartz R. Anesthesia for awake craniotomy. Can J Anaesth 1993; 40:100–103. 22. Welling EC, Donegan J. Neuroleptanalgesia using alfentanil for awake craniotomy. Anesth Analg 1989; 68:57–60. 23. Huncke K, Van de Wiele B, Fried I, Rubinstein EH. The asleep–awake–asleep anesthetic technique for intraoperative language mapping. Neurosurgery 1998; 42(6):1312–1317. 24. Trop D. Conscious-sedation analgesia during neurosurgical treatment of epilepsies. Practice at the Montreal Neurological Institute. Int Anesthesiol Clin 1986; 24:175– 184. 25. Penfield W, Pasquet A. Combined regional and general anesthesia for craniotomy and cortical exploration. Anesth Analg 1954; 33:145–164. 26. Hagberg C, Berry J, Haque S. The laryngeal mask airway for awake craniotomy in pediatric patients [abstr]. Anesthesiology 1996; 83:A184. 27. Danks RA, Rogers M, Aglio LS, Gugino LD, Black PM. Patient tolerance of craniotomy performed with the patient under local anesthesia and monitored conscious sedation. Neurosurgery 1998; 42:28–36. 28. Ojemann GA. Cortical organization of language. J Neurosci 1991; 11:2281–2287. 29. Duffau H, Capelle L, Sichez J, Faillot T, Abdennour L, Law Koune JD, Dadoun S, Bitar A, Arthuis F, Van Effenterre R, Fohanno D. Intraoperative direct electrical stimulations of the central nervous system: the Salpetriere experience with 60 patients. Acta Neurochir (Wien) 1999; 141(11):1157–1167. 30. Krombach GA, Spetzger U, Rohde V, Gilsbach JM. Intraoperative localization of functional regions in the sensorimotor cortex by neuro-navigation and cortical mapping. Comput Aided Surg 1998; 3(2):64–73. 31. Sartorius CJ, Wright G. Intraoperative brain mapping in a community setting—technical considerations. Surg Neurol 1997; 47:381–388. 32. Berger MS, Ojemann GA. Intraoperative brain mapping techniques in neuro-oncology. Stereotact Funct Neurosurg 1992; 58:153–161. 33. Berger MS, Kincaid J, Ojemann GA, Lettich E. Brain mapping techniques to maximize resection, safety, and seizure control in children with brain tumors. Neurosurgery 1989; 25:786–792. 34. Sartorius CJ, Berger MS. Rapid termination of intraoperative stimulation-evoked seizures with application of cold Ringer’s lactate to the cortex. J Neurosurg 1998; 88:349–351. 35. Berger MS. Functional mapping-guided resection of low-grade gliomas. Clin Neurosurg 1994; 42:437–452.

CHAPTER 7: AWAKE CRANIOTOMY AND INTRAOPERATIVE IMAGING

&

125

36. Black PM, Ronner SF. Cortical mapping for defining the limits of tumor resection. Neurosurgery 1987; 20:914–919. 37. Ebeling U, Schmid UD, Ying H, Reulen HJ. Safe surgery of lesions near the motor cortex using intra-operative mapping techniques: a report of 50 patients. Acta Neurochir (Wien) 1992; 119:23–28. 38. Gregorie EM, Goldring S. Localization of function in the excision of lesions from the sensorimotor region. J Neurosurg 1984; 61:1047–1054. 39. Walsh AR, Schmidt RH, Marsh HT. Cortical mapping and local anesthetic resection as an aid to surgery of low and intermediate grade gliomas. Br J Neurosurg 1992; 6:119–124. 40. Schulder M, Maldjian JA, Liu WC, Holodny AI, Kalnin AT, Mun IK, Carmel PW. Functional image-guided surgery of intracranial tumors located in or near the sensorimotor cortex. Neurosurg 1998; 89:412–418. 41. Puce A, Constable RT, Luby ML, McCarthy G, Nobre AC, Spencer DD, Gore JC, Allison T. Functional magnetic resonance imaging of sensory and motor cortex: comparison with electrophysiological localization. J Neurosurg 1995; 83:262–270. 42. Atlas SW, Howard RS, Maldjian J, Alsop D, Detre JA, Listerud J, D’Esposito M, Judy KD, Zager E, Stecker M, Howard RS 2d. Functional magnetic resonance imaging of regional brain activity in patients with intracerebral gliomas: findings and implications for clinical management. Neurosurgery 1996; 38:329–338. 43. Jack CR, Thompson RM, Butts RK, Sharbrough FW, Kelly PJ, Hanson DP, Riederer SJ, Ehman RL, Hangiandreou NJ, Cascino GD, Jack CR Jr. Sensory motor cortex: correlation of presurgical mapping with functional MR imaging and invasive cortical mapping. Radiology 1994; 190:85–92. 44. Mueller WM, Yetkin FZ, Hammeke TA, Morris GL, Swanson SJ, Reichert K, et al. Functional magnetic resonance imaging mapping of the motor cortex in patients with cerebral tumors. Neurosurgery 1996; 39:515–521. 45. Ogawa S, DW T, Menon R. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 1992; 89:5951–5955. 46. Turner R, Jezzard P, Wen H. Functional mapping of the human visual cortex at 4 and 1.5 tesla using deoxygenation contrast EPI. Magn Reson Med 1993; 29:277–279. 47. Barnett GH, Steiner CP, Weisenberger JP. Target and trajectory guidance for interactive surgical navigation systems. Stereotact Funct Neurosurg 1996; 66:91–95. 48. Brown RA, Roberts TS, Osborn AG. Stereotaxic frame and computer software for CT-directed neurosurgical localization. Invest Radiol 1980; 15:308–312. 49. Leksell L, Jernberg B. Stereotaxis and tomography. A technical note. Acta Neurochir (Wien) 1980; 52:1–7. 50. Roberts TS, Brown R. Technical and clinical aspects of CT-directed stereotaxis. Appl Neurophysiol 1980; 43:170–171. 51. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Intraoperative localization using an armless, frameless stereotactic wand. Technical note. J Neurosurg 1993; 78:510–514. 52. Germano IM. The NeuroStation system for image-guided frameless stereotaxy. Neurosurgery 1995; 37:438–450.

126

&

PART II: TECHNOLOGIC APPLICATION

53. Golfinos JG, Fitzpatrick BC, Smith LR, Spetzler RF. Clinical use of a frameless stereotactic arm: results of 325 cases. J Neurosurg 1995; 83:197–205. 54. Lunsford LD, Rosenbaum AE, Perry J. Stereotactic surgery using the ‘‘therapeutic’’ CT scanner. Surg Neurol 1982; 18:116–122. 55. Lunsford LD, Parrish R, Albright L. Intraoperative imaging with a therapeutic computed tomography scanner. Neurosurgery 1984; 15:559–561. 56. Hammoud MA, Ligon BL, elSouki R, Shi WM, Schomer DF, Sawaya R. Use of intraoperative ultrasound for localizing tumors and determining the extent of resection: a comparative study with magnetic resonance imaging. J Neurosurg 1996; 84:737–741. 57. LeRoux PD, Winter TC, Berger MS, Mack LA, Wang K, Elliott JP. A comparison between preoperative magnetic resonance and intraoperative ultrasound tumor volumes and margins. J Clin Ultrasound 1994; 22:29–36. 58. Woydt M, Krone A, Becker G, Schmidt K, Roggendorf W, Roosen K. Correlation of intraoperative ultrasound with histopathologic findings after tumor resection in supratentorial gliomas: a method to improve gross total resection. Acta Neurochir (Wien) 1996; 138:1391–1398. 59. Trobaugh JW, Richard WD, Smith KR, Bucholz RD. Frameless stereotactic ultrasonography: methods and applications. Comput Med Imaging Graph 1994; 18:235– 246. 60. Giorgi C, Casolino DS. Preliminary clinical experience with intraoperative stereotactic ultrasound imaging. Stereotact Funct Neurosurg 1997; 68:54–58. 61. Unsgaard G, Ommedal S, Muller T, Gronningsaeter A, Nagelhus Hernes TA. Neuronavigation by intraoperative three-dimensional ultrasound: initial experience during brain tumor resection. Neurosurgery 2002; 50:804–812. 62. Black PM, Moriarty T, Alexander EI, Stieg P, Woodard EJ, Gleason PL, et al. Development and implementation of intraoperative magnetic resonance imaging and its neurosurgical applications. Neurosurgery 1997; 41:831–845. 63. Schenck JF, Jolesz FA, Roemer PB, Cline HE, Lorensen WE, Kikinis R, Silverman SG, Hardy CJ, Barber WD, Laskaris ET. Superconducting open-configuration MR imaging system for image guided therapy. Radiology 1995; 195:805–814. 64. Tronnier VM, Wirtz CR, Knauth M, Lenz G, Pastyr O, Bonsanto MM, Albert FK, Kuth R, Staubert A, Schlegel W, Sartor K, Kunze S. Intraoperative diagnostic and interventional MRI in neurosurgery. Neurosurgery 1997; 40:891–902. 65. Kanner AA, Vogelbaum MA, Mayberg MR, Weisenberger JP, Barnett GH. Intracranial navigation by using low-field intraoperative magnetic resonance imaging: preliminary experience. J Neurosurg 2002; 97:1115–1124. 66. Schulder M, Azmi Hm, Biswal B. Functional magnetic resonance imaging in a lowfield intraoperative scanner. Stereotact Funct Neurosurg 2003; 80(1–4):125–131. 67. Sharan AD, Rezai AR, Hrdlicka G, Nyhenuis J, Tkach J, Ruggieri P, Baker K, Phillips M. MR Safety in Patients with Implanted Neurostimulators: Update and Future Implications. 68. Barnett GH, Kormos DW, Steiner CP, Morris H. Registration of EEG electrodes with three-dimensional neuroimaging using a frameless, armless stereotactic wand. Stereotact Funct Neurosurg 1993; 61:32–38.

8

COMPUTER-ASSISTED NEUROENDOSCOPY: THE NAVIGATED NEUROENDOSCOPE Mark G. Luciano, Mohamed Ammar, and Stephen M. Dombrowski Section of Pediatric and Congenital Neurosurgery, Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

‘‘Minimally invasive neurosurgery’’ has grown through the evolution of neuroendoscopy and neuronavigation from cumbersome technologies used in a minority of specialized centers, to tools, which are an essential part of any neurosurgical operating room. The inevitable combination of these complementary technologies has provided a safer and less disruptive approach to many deep targets. Over the last decade, the coupling has become smoother, faster, and more ‘‘user friendly.’’ Since the ventricular system is a natural cavity lending itself to visual exploration, many initial neuroendoscopic procedures were related to hydrocephalus and paraventricular lesions. While this is still the case today, neuroendoscopy is also used in the brain cisterns, pathological cysts, spine and spinal canal, skull base, and cranium. Importantly, the endoscope is used increasingly to assist in open operative neurosurgical procedures to see closer, at a different angle, or through a smaller incision. Just as the development of the operative stereomicroscope ushered in the era of ‘‘microsurgery,’’ advances in the ‘‘navigated neuroendoscope’’ may be the next step in the evolution of minimally invasive neurosurgery, putting the operative microscope inside the brain. THE DEVELOPMENT OF MODERN NEUROENDOSCOPY Endoscopy itself is not new. The earliest endoscopy in natural body orifices was limited by availability of light to allow exploration and navigation. In 1805, Phillip Bozzini of Vienna used a series of concave mirrors to channel light into the rectum, only to be censured by his society for inappropriate curiosity (1). Max Nitze created the first successful endoscope in 1879 by using light from a glowing wire and magnifying the image with lenses (2). Modern endoscopy began with the more efficient delivery

Icons indicate materials available on the accompanying CD.

127

128

&

PART II: TECHNOLOGIC APPLICATIONS

of light and transmission of images using glass rod scopes developed by Hopkins in the 1960s. His work with fiber optics was also instrumental in the development of flexible fiberscopes (3). Neuroendoscopy began in 1910 with L’Espinasse who is credited with the first neuroendoscopic procedure (4). In 1922, Walter Dandy (5,6) published on choroid plexus removal and in 1923 Mixter (7) performed the first endoscopic third ventriculostomy. In the 1990s, interest in neuroendoscpy grew with fiber optic technology and a general interest in minimally invasive procedures in all surgery. The addition of navigation to endoscopy has roots not only in hydrocephalus but also in tumor surgery, since it was in this area that stereo tactic techniques were widely developed and employed. While the spectrum of current neuroendoscopic procedures can be seen in Table 1, the most frequent procedures remain the third ventriculostomy and ventricular tumor biopsy. Ventricular procedures are often performed by pediatric neurosurgeons; however, skull base, spinal, and tumor, and vascular neurosurgeons have increasingly found uses outside the ventricular system. In all these locations and purposes, neuronavigation has been helpful where anatomy is abnormal, complex, deep, and near eloquent regions. CURRENT NEUROENDOSCOPIC TECHNOLOGY Endoscopes employ either glass rods or glass fiber optics for light transmission and may be rigid, semirigid, flexible, or steerable. The glass rod scopes have been preferred when a straight trajectory is possible since the optics are superior. However, with increases in the number of fibers in fiberscopes (to 30,000), the image has improved greatly. Fiberscopes can be made small enough to fit inside ventricular catheters and a variety of ‘‘seeing’’ dissection tools for open surgical procedures. The flexible scope can also be ‘‘steerable,’’ allowing a limited flexion. For example, steerable endoscopes are useful in visualizing both the floor and the posterior third ventricle through one burr hole and trajectory. Importantly, however, flexible scopes do not allow the use of stereotaxic navigation, which is limited to rigid systems where the probe tip is fixed and can be calculated.

Table 1 Current Applications of Neuroendoscopy Third ventriculostomy Fenestrations: septum, arachnoid cysts Tumor biopsy/excision Pituitary and skull base surgery Brain abscess and hematoma drainage Catheter placement and foreign body removal Craniectomy in craniosynostosis and chiari decompression Spinal surgery—discs, intrathecal masses, syrinx Endoscopic-assisted surgery—posterior fossa angle, aneurysms, parenchymal tumors

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

129

The light and camera of the endoscopic system generally consist of a Xenon light source and a charge-coupled device, which has been improving in recent years to the current ‘‘three chip’’ systems with high resolution (8–10). Endoscopic tools for manipulation of tissue consist of 1 to 2 mm rigid, semirigid, or flexible instruments for dissection, cutting, biopsy, cautery (mono- and bipolar), and suctioning. Specialized tools, which may cut and coagulate or suction simultaneously have also been described, as well as other lesioning devices such as the laser and ultrasonic destruction (11,12). Except in the case of endoscopicassisted surgery, these tools are designed to fit down the endoscope-working sheath in a coaxial fashion.

PROBLEMS IN NEUROENDOSCOPY The neuroendoscope is limited in three major ways: (i) movement, (ii) visualization, and (iii) difficulty in manipulation (Table 2). 1. Endoscope movement: For many procedures in dilated ventricles, freehand insertion and movement of the endoscope may be safe and effective. In these cases, inaccuracy of trajectory is tolerated and the endoscope may be moved laterally while in-based on orientation to internal visual. However, when ventricles are small, or when the target is a cistern or small tumor, precise targeting is needed. After endoscope insertion, the lateral movements necessary for visual exploration or for instrument manipulation may damage adjacent structures unseen behind or lateral to the limited field of view. In abnormal anatomy, exploratory endoscope movement may result in disorientation and brain injury. 2. Visualization: While CSF is an excellent medium for visualization, a minimal amount of blood can greatly reduce clarity. Circulation of irrigation is currently the primary tool for maintaining clarity. In addition, the endoscopic view is Table 2 Limitations of Neuroendoscopy Endoscope movement Planning and executing safe trajectories to targets Damage with scope movement when at target Orientation—knowing where you are with scope movement Visibility Limited and distorted view Two-dimensional images Clear fluid (or air) required Tissue manipulation Small size of tools and exit port Poor mobility of tools—depends on scope movement Limited articulation of tools Limited ports for tools

130

&

PART II: TECHNOLOGIC APPLICATIONS

distorted (‘‘fish eye’’), two dimensional, and usually fixed to a particular angle (such as 0 or 30 ). Some of these constraints may be compensated by operator experience and some may be amenable to improvement through mechanical or computer advances. 3. Tissue manipulation: The endoscope’s working sheath and operative space places constraints on the size, strength, and articulation of instruments. Manipulation is largely limited to lesioning (cutting, coagulation, or laser) and tissue removal through a small port (suctioning or biopsy forceps). Though a great variety of tools have been employed, their articulation is often limited. In the case of coaxial instruments, movement depends on movement of the entire endoscope. While endoscope movement has little impact in a large dilated ventricle, in a parenchymal tract and in small target spaces, such as cistern, these movements can cause injury. Computer guidance can play a role in efforts to minimize all three types of neuroendoscope limitations, especially in endoscope navigation and robotics.

Improving Neuroendoscopy with Computer Technology Intraoperative ‘‘Real-Time’’ Neuroendoscopic Navigation Ideally, neuroendoscopes should be guided through the brain using detailed intraoperative imaging and real-time feedback. Theoretically, intraoperative imaging is an important advantage in ventricular neuroendoscopy where fluid shifts make navigation based on preoperative images less reliable. These brain shifts are most dramatic in more superficially around drained ventricles and less dramatic in more central brain structures. Although attempts to minimize the effect of intraoperative shifts are helpful (13,14), this limitation remains and therefore updated imaging during a case is optimal for continuous accuracy. The three major modalities for intraoperative imaging are ultrasound, computerized tomography (CT), and MRI. Ultrasound Guidance. Ultrasound has the advantage of a being a current, familiar, low-cost, and truly ‘‘real-time’’ intraoperative imaging modality, which can be easily adapted for neuroendoscopic guidance. One disadvantage of ultrasound guidance is the need for a more extended craniotomy to allow an ultrasound window of view. This window may be adjacent or remote from the neuroendoscope. Strowitzki et al. (15) describe the physical coupling of an ultrasound probe to the neuroendoscope to maintain visual guidance at the working tip. While the development of smaller ultrasound probes allows smaller windows, recently, microprobes small enough to be placed through the endoscope-working channel have been developed and provide a limited image. Resch (16) reports using a sono catheter in 52 patients for transendoscopic imaging to facilitate navigation and tumor biopsy. Micro-Doppler probes used in vascular medicine are also small enough to fit down the endoscope (17,18) and can detect blood flow towards

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

131

or away from the probe. While they are useful in detecting the basilar artery below the floor of the third ventricle, the range of view is limited and other blood flow, like those from the basilar perforators, is not reliably detected. Another disadvantage of ultrasound imaging is that low resolution and difficulty in interpretation may limit guidance to targets, especially small tumors or blood. For this reason, more advanced systems such as ‘‘3D-ultrasound’’ navigation have been utilized clinically with some success (19). In the future, computer image fusion of the more detailed anatomy available from a pre-op MRI, with the changes seen intraoperatively through ultrasound, may allow the needed detail for guidance but with the ease and low cost of ultrasound. Intraoperative MRI. Intraoperative imaging, especially MRI, obviously provides the greatest opportunity for a high-resolution real-time or ‘‘near-real-time’’ guidance. The expense of intraoperative imaging and the need to develop MRIcompatible devices, however, limits the general use of these systems beyond a few centers. In addition, many current intraoperative MRI systems significantly increase the procedure time, restrict the operative working space, and provide limited image resolution. To improve resolution in neuroendoscopy, gadolinium contrast dye has been injected into the endoscope-perforated cavity (20) and has been successfully used in children for fenestration of complex cysts. Computed Tomography. While CT scanning maybe somewhat less cumbersome and less expensive, this modality suffers from some of the same disadvantages as MRI and significantly increases exposure to ionizing radiation.

Stereotactic Guidance: Use of Preoperative Imaging While the advantage of real-time intraoperative image guidance is clear, practical and financial considerations have limited its use. On the other hand, despite the potential problem of brain shift, the widespread use and documented efficacy of stereotactic procedures demonstrates the utility and practicality of preoperative images for guidance. Stereotactic guidance is readily available, familiar, relatively inexpensive, and accurate. As with stereotactic biopsy (21), this approach adds minimal time to the endoscopic procedure. Further, use of preoperative surgical planning and trajectory setup can even reduce operative time (19). The evolution of stereotactic procedures generally is detailed elsewhere in this book, but has generally proceeded from awkward and inflexible framed systems to frameless systems. Historically, stereotactic frames were first adapted for neuroendoscopic guidance most often for visually enhanced sterotactic biopsy (12,22). Framed systems were especially helpful in targeting trajectory planning and holding the endoscope in a fixed position. An open guide tube was developed to allow coaxial introduction of instruments (23). However, when multiple targets were needed or when movement of the scope under visual guidance was also desired, these systems became quite cumbersome. Thus, the system is useful for the initial

132

&

PART II: TECHNOLOGIC APPLICATIONS

approach but less for use after endoscope insertion when free hand movements are needed for instrument use and multiple tasks. The development of frameless stereotaxy offers the precision of guidance while allowing flexibility in freehand use important in many forms of neuroendoscopy. In 1995, we described the adaptation of the Picker sonic frameless stereotactic system to rigid endoscope guidance called ‘‘computer-assisted neuroendoscopy’’ or CANE (24). Patients underwent a pre-op MRI (or CT) with feducials and the head was fixed intraoperatively in a pin headrest. A custom obturator for insertion of a peal-away sheath was fitted with ultrasonic sound emitters and registered to allow ventricular access with guidance. A separate custom probe was designed to lock into the working channel of the endoscope to allow localization of the tip during endoscopy (Fig. 1A). In frameless CANE cases, care is taken to keep the entry point at the top to avoid fluid drainage and irrigation is continuously circulated for visual clarity and to avoid ventricular or cyst collapse. As with other sterotactic procedures, anatomical accuracy is greater in deep and midline structures. In the last seven years, the many technical improvements in navigation have made its use with neuroendoscopy simpler. Accuracy has improved with the switch to infrared emitters instead of ultrasonic pulses and to the use of a reference array (Fig. 1A, B). Passive reflectors have eliminated the need for wired probes, which is especially important given the multiple cords and lines already needed for endoscopy alone. Since a single endoscopic case may require the use of a standard probe, introducer sheath, and multiple endoscopes, the ability to register any tool in the operating utilizing clip on reflectors allows navigation from the beginning of the case and throughout, quickly and without custom instruments (Fig. 1C).

Figure 1 The CANE system originally described in 1997 (A) was cumbersome with additional probes and wires. Today’s system (B) has a passive wireless system with a three-ball reference array to correct for table movement and clip on instrument array. A new endoscope can be registered intraoperatively with a registration block array (C).

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

133

Recent advances in head immobilization and registration may further simplify use and make navigated neuroendoscopic cases easier in children. For example, surface registration obviates the need for a pre-op fiducial placement and head fixation, through oral occlusion splints, which can eliminate skull pin fixation (19). COMPUTER-ASSISTED VISUALIZATION While most efforts to maintain good visualization have depended on irrigation and instruments allowing less vascular injury (25) or better coagulation, computer imaging and navigation systems may also play a role in the future. A guidance system (Cybion, Inc) can, for example, superimpose a computer-generated virtual image on the actual endoscopic image to allow better intraoperative recognition and targeting. A recently described experimental system automatically generates a virtual endoscope view based on previous endoscope images, which automatically is displayed when the ‘‘red out’’ of an acute bleed is detected. This may allow some continued effort at vessel coagulation without direct vision (26). The combination of images such as the endoscopic view, operative microscope view, virtual image has been proposed through microscope oculars or video monitors (27). Recently, the combination of endoscope and virtual views may be facilitated further with heads up screens (28). Other significant improvement in visualization through the endoscope will depend on the development of 3D imaging (difficult in very small spaces) and with variable angle rotating optics facilitating a ‘‘look around’’ without scope movement. COMPUTER-ASSISTED TOOLS FOR MANIPULATION Manipulation, lesioning, and removal of tissue through the neuroendoscope are still limited by the small space and the coaxial approach. Systems of robotic and computer-navigated endoscope movement have been described (29) and used clinically (30). While this system adds accuracy and minimizes unnecessary and potential injurious movement, the system is slow, inflexible, and expensive. In addition, this approach does not improve the effectiveness of instruments. While remote robotics have been employed clinically (the Da Vinci system), in other areas of surgery for the remote manipulation of tools, these systems are too big for neurosurgery. A small coaxial robotic system, which allows multiarticulated forceps to be controlled remotely and which employs stereotactic navigation and 3-D imaging has been described experimentally but is not yet clinically tested (31). Systems like this may eventually allow safer navigation to small targets, better visualization, and better ability to manipulate what is seen. Clinical Use of the Navigated Endoscope Over the last 10 years, about 356 CSF-based neuroendoscopic procedures were performed at the Cleveland Clinic Foundation, 119 in children and 237 in adults.

134

&

PART II: TECHNOLOGIC APPLICATIONS

The case mix can be seen in Table 3, which shows that the majority were third ventriculostomies. In addition to the cases listed, the endoscope has been used outside the CSF spaces in pituitary tumor resection, parenchymal tumor resection, and chiari malformation decompression. Overall, computer navigation guidance was used approximately in 15% of the cases. While it was used in the minority of third ventriculostomies (reserved for cases with small or abnormal third ventricles), it was used more frequently in neuroendoscopic tumor biopsy, multiloculated cyst fenestration, and colloid cyst resection. This is in agreement with others where tumor biopsy may make up more than 50% of their navigated endoscopic cases and third ventriculostomy 7% (32,33). Schroeder et al. (33) reviewing 44 navigated endoscope cases found it most useful in accessing small ventricles and in targeting posterior third ventricular masses and parenchymal or multiloculated cysts. Of the non-CSF or ‘‘open air’’ endoscopic uses, navigation was utilized in endoscopic-assisted parenchymal tumor resection and CSF leak repair. In each kind of case navigation, it can be used at two levels. First, computer navigation may be used for entry point and working channel (often a peel-away sheath) trajectory planning. After this, an un-navigated neuroendoscope is placed in and used under direct visual guidance. Second, the neuroendoscope itself may be navigated to orient and move inside the brain during the procedure. More recently, computer navigation has been utilized clinically in a third way, i.e., preoperative rehearsal through virtual images. Using this method, a surgeon may see an approach simulation based on individual preoperative MRI data and rehearse a more complex case (34,35). Schroeder et al. (33) used navigation in 44 of 135 cases, but in only 10 cases did they use it during the endoscopic procedure itself. In agreement with this, Tirakotai et al. (36), in reviewing their experience in 450 cases over 20 years, found that the most frequent uses of navigation was in initial trajectory planning. Obstructive Hydrocephalus The endoscopic third ventriculostomy has become an accepted primary treatment for obstructive hydrocephalus. Obstruction may occur in the aqueduct due to congenital malformation, inflammation, or tumor, or in the posterior fossa most often secondary to tumor. In our own series, overall success rate for third ventriculostomy is approximately 70%, and over 80% in aqueductal stenosis. Since the fenestration reclosure rate is only 7.5%, failure is most often due to a residual communicating hydrocephalus. Basilar artery system injury is the most concerning complication; Table 3 Endoscopic Case Mix at Cleveland Clinic Third ventriculostomy Tumor biopsy Endoscopic catheter placement/movement Cyst fenestration Colloid cyst excision

261 30 30 24 11

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

135

however, injury to the hypothalamus or fornix is also possible. Although navigation may reduce the frequency of these complications, their incidence in third ventriculostomy is already very low, generally less than 2%. While third ventriculostomy is one of the most common neuroendoscopic procedures, it is the least likely to require computer navigation (33). Because of obstructive hydrocephalus the lateral and third ventricles are usually large and the ventricular anatomy is intact, obviating the need for navigation in approach or orientation. The floor of the third ventricle is often stretched and translucent allowing visual determination of the safest point for fenestration, usually in the midline just posterior to the clivus. Earlier, the development of neuroendoscopic navigation guidance was used more frequently even in simple third ventriculostomies. In 13 third ventriculostomies, Schroeder et al. (33) used navigation to plan the optimal entry point and compared the location of the stereotactic determination to a point selected before use of navigation. In no case was the point significantly different. Grunert et al. (37) compared their success rate with stereotactic and freehand techniques and found no significant difference, 74.2% and 74.7%, respectively. Kanner et al. (38) analyzed the best site for third ventriculostomy based on their series of 27 patients where the entry point was planned stereotacitically. They found that the optimal site was 3 cm lateral to midline and 1 cm anterior to the coronal suture. They concluded that individual planning was still needed since there was variability in the optimal site, but that this site could be identified on a coronal and sagittal MRI, and did not require computer navigation. Even considering the above, there are cases where navigation is helpful in third ventriculostomy. When the third ventricle is small, a more medial entry point may be planned for access to the ventricular floor. In myelomeningocele, the anatomy of the floor may be abnormal and a large massa intermedia may restrict access to the ventricle, making the anterior–posterior entry position more critical. If the basilar artery or ventricular floor anatomy is unusual, the fenestration target may be more safely determined with guidance. In entry point and trajectory planning for third ventriculostomy, the target is chosen in the midline of the floor anterior to the basilar artery. The foramen of Monroe is the entry point and the trajectory is extrapolated superiorly to its intersection with the skull. The standard pointer is used to identify and mark the entry point on the skull before skin preparation. The insertion itself can also be guided by registering the peel-away sheath after clipping the three reflecting sphere array to the shaft. Although rarely needed, the endoscope itself can be registered for guidance during ventricular floor exploration and perforation. Pineal region and posterior third ventricle tumors, which obstruct the aqueduct, are ideally treated neuroendoscopically with simultaneous biopsy and third ventriculotomy (39,40). This procedure has been performed through two burr holes with rigid endoscopy, approach for the posterior ventricular access and another for the floor (Fig. 2). Navigation is especially useful in planning the more anterior burr hole targeting the tumor. Use of the flexible scope, however, allows the biopsy and the fenestration through a single burr hole. In this case, navigation may be helpful for planning the skull entry point to allow closest access to both targets, but cannot be used during the flexible scope procedure.

136

&

PART II: TECHNOLOGIC APPLICATIONS

Figure 2 One pineal region or tectal mass is an ideal use of neuroendoscopy since there is both obstructive hydrocephalus and a paraventricular mass (A). The combined procedure can be performed through two separate approaches (dotted and dashed) for tumor biopsy and ventricular fenestratation, respectively, or through a single approach with a stearable neuroendoscope solid (B).

Multiloculated Hydrocephalus Standard treatment for multiloculated hydrocephalus has involved placement of multiple catheters for even drainage. However, shunting systems have a higher probability of complications and attempts at trying to communicate all compartments to allow drainage through a single catheter is appropriate (41). While this may be performed through an open craniotomy, these fenestrations may be created endoscopically (42,43). Multicompartment hydrocephalus can be one of the most challenging endoscopic procedures with potential disorientation, hemorrhage, missed cysts, and inadequate fenestration reclosure. Because of the abnormal anatomy and the complexity of some loculations and membranes, the precise entry points, trajectories, and orientation allowed by navigation are helpful. Using the navigation system, the best site of fenestration in a cyst can be identified based on avoiding unseen structures behind and trajectory for best entry into other cysts (Fig. 3). While multiloculated hydrocephalus is relatively unusual, it is a very good indication for navigated neuroendoscopy (24,32).

Cyst Fenestration The endoscope is widely used for cyst fenestration (44,45).

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

137

Figure 3 Multiloculated cysts demonstrate complex abnormal anatomy risking disorientation (A). Navigation allows pre- and intraoperative planning of trajectories to more safely fenestrate multiple cysts (B).

Arachnoid Cysts. Cerebral arachnoid cysts such as those in the middle or posterior fossa may be communicated with the subarachnoid space using laser, coagulation, cutting, or balloon dilatation. In middle fossa cysts, where medial fenestration into the suprasellar cistern is usually attempted, navigation allows identification and avoidance of critical structures. In the posterior fossa, the entry point, trajectory, and fenestration of a qudrageminal cistern cyst may be performed with navigated endoscopy. Intraventricular Ependymal Cysts. When intraventricular cysts are communicated with the ventricle, the angle of approach and fenestration site can be critical. Third ventricular cysts are treated dual fenestration at the dome (which may extend into the lateral ventricle) and then at the base, which can extend to the prepontine cistern. Colloid Cysts. Colloid cysts have previously been treated with open craniotomy and a ventricular or interhemispheric approach. These lesions can also be resected via neuroendoscopy using coagulation, cutting, and suction. The authors have removed 12 colloid cysts by this approach without permanent complication or recurrence. The approach is facilitated by a lateral approach to the ventricle, which allows visualization of the third ventricular roof. Cystic Abscess and Hematoma. Cystic abscesses and hematomes in the ventricle or brain parenchyma are good indications for navigated endoscopy (32,46). In

138

&

PART II: TECHNOLOGIC APPLICATIONS

these cases, navigation is used for initial approach to the cyst and once inside, the endoscope itself is computer navigated since initial visual guidance is not possible through the opaque fluid. After drainage and irrigation clears the cyst fluid sufficiently, the endoscopic view can be used to insure a complete evacuation. Tumors Stereotactic endoscopy was developed, first framed then frameless, in large part for treatment of tumors. The Marburg, Germany group (32) that began with framed stereotactic neuroendoscopy performed 213 tumor biopsies of 412 total navigated endoscope procedures. Zamorano et al. (12) report its uses in 152 cases where laser was also used for resection and fenestration. Ventricular endoscopy is well suited to biopsy of paraventricular tumors given the difficulties of blind stereotactic biopsy in these areas (Fig. 4). Endoscopic drainage of craniopharyngiomas with catheter placement was performed to two patients. Navigation is also useful in the approach to the dome of the cystic tumor in the third ventricle and with endoscope navigation in the cyst for drainage and catheter placement (Fig. 5). The advantage of the combination of stereotaxy and endoscopy in tumor biopsy is the combination of ease of approach afforded by navigation, followed by the ability to take the biopsy under direct visualization, differentiating abnormal

Figure 4 Paraventricular tumors like this germinoma may be hard to biopsy neuroendoscopically due to small ventricles and difficulty in visualizing the tumor. Navigation allows trajectory planning.

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

139

Figure 5 A cystic craniopharyngioma (A and B) was approached, fenestrated, evaluated, and catheterized (C) via the navigated neuroendoscope.

Figure 6 Multiple hemangioblastomas were resected via three separate planned trajectories from an incision less than 1 in. The patient was discharged without deficit within 48 hours.

140

&

PART II: TECHNOLOGIC APPLICATIONS

tissue or tumor cyst wall. While endoscopic biopsies may be small, since they are performed under direct visualization, the diagnostic yield is high. In our series of pineal region tumors, endoscopic biopsy yielded the diagnosis 88% of the time.

Endoscope-Assisted Surgery With fiber optic ‘‘seeing’’ dissections tools, neuroendoscopes can be used to see into and around corners during open microsurgical procedures, viewing angles not possible with the operative microscope (27). In this way, neuroendoscopy can be used in air, outside CSF spaces, to assist a variety of operations such as pituitary and aneurysm surgery. Experimental studies suggest that this approach may enhance safe access to eloquent areas such as the rostral brainstem (11). The endoscope can replace the operative microscope entirely for the resection of pituitary and skull base tumors (39), parenchymal tumors (47,48), vascular malformations (49), and CP angle epidermoids (50). For example, in a Von Hippel Landau patient with multiple posterior fossa cysts and hemangioblastomas, three

Figure 7 The navigated neuroendoscope can be used in place of the operative microscope during minimal retraction to resect deep-seated paramchymal tumors.

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

141

growing lesion were resected from a single small (less than 1 in) using navigated endoscope trajectories and minimal retraction (Fig. 6). A small deep frontal brain tumor was resected with assistance from a navigated endoscope, using minimal retraction (Fig. 7).

CONCLUSION The navigated endoscope is a tool used within and outside the ventricular system, in fluid or in air, alone or in conjunction with the operative microscope. The two technologies complement each other to extend the safe use of minimally invasive endoscopic approaches. The coupling of the endoscope and navigation has become simpler, faster, and less cumbersome. While use in common ventricular cases such as the third ventriculostomy is usually not needed, series have suggested that the addition of navigation to endoscopy adds little operative time and is advantageous in more complex cases such as those with abnormal anatomy, poor visualization, and small targets. The navigated endoscope, especially with future improvements in visualization and tissue manipulation, may increasingly replace the standard operative microscope, placing the operating eye within the brain itself. ACKNOWLEDGMENTS We would like to thank Mahmoud Ragab, M.D., for his help in compiling the clinical data presented in Tables 1 to 3.

142

&

PART II: TECHNOLOGIC APPLICATIONS

REFERENCES 1. Bush RB, Leonhardt H, Bush IV, Landes RR. Dr. Bozzini’s lichtleiter. A translation of his original article (1806). Urology 1974; 3(1):119–123. 2. Schultheiss D, Truss M, Jonas U. History of direct vision internal uretherotomy. Urology 1987; 52(4):729–734. 3. Hopkins H, Kapany. A flexible light transmitting tube. Nature 1954; 173:39–41. 4. Abbott R. History of neuroendoscopy. Neurosurg Clin North Am 2004; 15(1):1–7. 5. Dandy WE. An operative procedure for hydrocephalus. John Hopkins Hosp Bull 1922; 33:188–190. 6. Dandy WE. Surgery of the brain. In: Lewis D, ed. Practice of Surgery. Vol. 12. Hagerstown, MD: WF Prior, 1932 (chapter 1). 7. Mixter WJ. Ventriculoscopy and puncture of the floor of the third ventricle. Boston Med Surg J 1923; 188:277–278. 8. Yanagisawa E, Godley F, Muta H. Selection of video cameras for stroboscopic videolaryngoscopy. Ann Otol Rhinol Laryngol 1987; 96(5):578–585. 9. Surgical video systems. Health Devices 2004; 33(4):109–147. 10. Berber E, Pearl JM, Siperstein AE. A simple device for measuring the resolution of videoscopic cameras and laparoscopes in the operating room. Surg Endosc 2002; 16(7):1111–1113. 11. Cohen AR, Perneczky A, Rodziewicz GS, Gingold SI. Endoscope-assisted craniotomy: approach to the rostral brain stem. Neurosurgery 1995; 36(6):1128–1129. 12. Zamorano L, Chavantes C, Moure F. Endoscopic stereotactic interventions in the treatment of brain lesions. Acta Neurochir Suppl 1994; 61:92–97. 13. Kelly PJ. Volumetric stereotactic surgical resection of intra-axial brain mass lesions. Mayo Clin Proc 1988; 63(12):1186–1198. 14. Kelly PJ, Kall BA, Goerss SJ. Results of computed tomography-based computerassisted stereotactic resection of metastatic intracranial tumors. Neurosurgery 1988; 22(1 Pt 1):7–17. 15. Strowitzki M, Kiefer M, Steudel WI. A new method of ultrasonic guidance of neuroendoscopic procedures. Technical note. J Neurosurg 2002; 96(3):628–632. 16. Resch KD. Endo-neuro-sonography: first clinical series (52 cases). Childs Nerv Syst 2003; 19(3):137–144. 17. Schmidt RH. Use of a microvascular Doppler probe to avoid basilar artery injury during endoscopic third ventriculostomy. Technical note. J Neurosurg 1999; 90(1):156–159. 18. Yamasaki T, Moritake K, Akiyama Y, Oi S. Microvascular Doppler ultrasoundassisted neuroendoscopic surgery. Br J Neurosurg 2000; 14(5):464–467. 19. Vougioukas VI, Hubbe U, Hochmuth A, Gellrich NC, van V V. Perspectives and limitations of image-guided neurosurgery in pediatric patients. Childs Nerv Syst 2003; 19(12):783–791. 20. Balmer B, Bernays RL, Kollias SS, Yonekawa Y. Interventional MR-guided neuroendoscopy: a new therapeutic option for children. J Pediatr Surg 2002; 37(4):668–672.

CHAPTER 8: COMPUTER-ASSISTED NEUROENDOSCOPY

&

143

21. Alberti O, Dorward NL, Kitchen ND, Thomas DG. Neuronavigation—impact on operating time. Stereotact Funct Neurosurg 1997; 68(1–4 Pt 1):44–48. 22. Bauer BL, Hellwig D. Minimally invasive endoscopic neurosurgery—a survey. Acta Neurochir Suppl 1994; 61:1–12. 23. Otsuki T, Jokura H, Yoshimoto T. Stereotactic guiding tube for open-system endoscopy: a new approach for the stereotactic endoscopic resection of intra-axial brain tumors. Neurosurgery 1990; 27(2):326–330. 24. Rhoten RL, Luciano MG, Barnett GH. Computer-assisted endoscopy for neurosurgical procedures: technical note. Neurosurgery 1997; 40(3):632–637. 25. Paladino J, Rotim K, Stimac D, Pirker N, Stimac A. Endoscopic third ventriculostomy with ultrasonic contact microprobe. Minim Invasive Neurosurg 2000; 43(3): 132–134. 26. Scholz M, Fricke B, Tombrock S, Hardenack M, Schmieder K, von During M, et al. Virtual image navigation: a new method to control intraoperative bleeding in neuroendoscopic surgery. Technical note. J Neurosurg 2000; 93(2):342–350. 27. Perneczky A, Fries G. Endoscope-assisted brain surgery: part 1—evolution, basic concept, and current technique. Neurosurgery 1998; 42(2):219–224. 28. Levy ML, Day JD, Albuquerque F, Schumaker G, Giannotta SL, McComb JG. Heads-up intraoperative endoscopic imaging: a prospective evaluation of techniques and limitations. Neurosurgery 1997; 40(3):526–530. 29. Zimmermann M, Krishnan R, Raabe A, Seifert V. Robot-assisted navigated endoscopic ventriculostomy: implementation of a new technology and first clinical results. Acta Neurochir (Wien) 2004; 146(7):697–704. 30. Zimmermann M, Krishnan R, Raabe A, Seifert V. Robot-assisted navigated neuroendoscopy. Neurosurgery 2002; 51(6):1446–1451. 31. Hongo K, Kobayashi S, Kakizawa Y, Koyama J, Goto T, Okudera H, et al. NeuRobot: telecontrolled micromanipulator system for minimally invasive microneurosurgery—preliminary results. Neurosurgery 2002; 51(4):985–988. 32. Hellwig D, Benes L, Bertalanffy H, Bauer BL. Endoscopic stereotaxy—an eight year’s experience. Stereotact Funct Neurosurg 1997; 68(1–4 Pt 1):90–97. 33. Schroeder HW, Wagner W, Tschiltschke W, Gaab MR. Frameless neuronavigation in intracranial endoscopic neurosurgery. J Neurosurg 2001; 94(1):72–79. 34. Rohde V, Krombach GA, Struffert T, Gilsbach JM. Virtual MRI endoscopy: detection of anomalies of the ventricular anatomy and its possible role as a presurgical planning tool for endoscopic third ventriculostomy. Acta Neurochir (Wien) 2001; 143(11):1085–1091. 35. Auer LM, Auer DP. Virtual endoscopy for planning and simulation of minimally invasive neurosurgery. Neurosurgery 1998; 43(3):529–537. 36. Tirakotai W, Bozinov O, Sure U, Riegel T, Bertalanffy H, Hellwig D. The evolution of stereotactic guidance in neuroendoscopy. Childs Nerv Syst 2004; 20(11–12): 790–795. 37. Grunert P, Charalampaki P, Hopf N, Filippi R. The role of third ventriculostomy in the management of obstructive hydrocephalus. Minim Invasive Neurosurg 2003; 46(1):16–21.

144

&

PART II: TECHNOLOGIC APPLICATIONS

38. Kanner A, Hopf NJ, Grunert P. The ‘‘optimal’’ burr hole position for endoscopic third ventriculostomy: results from 31 stereotactically guided procedures. Minim Invasive Neurosurg 2000; 43(4):187–189. 39. Pillay PK, Leo CW, Sethi DS. Computer-aided/image-guided and video-endoscopic resection of pituitary tumors. Stereotact Funct Neurosurg 2000; 74(3–4):203–209. 40. Oi S, Shibata M, Tominaga J, Honda Y, Shinoda M, Takei F, et al. Efficacy of neuroendoscopic procedures in minimally invasive preferential management of pineal region tumors: a prospective study. J Neurosurg 2000; 93(2):245–253. 41. Rhoton AL Jr, Gomez MR. Conversion of multilocular hydrocephalus to unilocular. Case report. J Neurosurg 1972; 36(3):348–350. 42. Nowoslawska E, Polis L, Kaniewska D, Mikolajczyk W, Krawczyk J, Szymanski W, et al. Effectiveness of neuroendoscopic procedures in the treatment of complex compartmentalized hydrocephalus in children. Childs Nerv Syst 2003; 19(9):659–665. 43. Nowoslawska E, Polis L, Kaniewska D, Mikolajczyk W, Krawczyk J, Szymanski W, et al. Neuro-endoscopic techniques in the treatment of complex hydrocephalus in children. Neurol Neurochir Pol 2003; 37(1):99–111. 44. Hellwig D, Bauer BL, Schulte M, Gatscher S, Riegel T, Bertalanffy H. Neuroendoscopic treatment for colloid cysts of the third ventricle: the experience of a decade. Neurosurgery 2003; 52(3):525–533. 45. Bognar L, Orbay P. Neuroendoscopic removal of a colloid cyst of the third ventricle. Orv Hetil 2000; 141(3):125–127. 46. Nakano T, Ohkuma H, Ebina K, Suzuki S. Neuroendoscopic surgery for intracerebral haemorrhage—comparison with traditional therapies. Minim Invasive Neurosurg 2003; 46(5):278–283. 47. Hor F, Desgeorges M, Rosseau GL. Tumour resection by stereotactic laser endoscopy. Acta Neurochir Suppl (Wien) 1992; 54:77–82. 48. Teo C, Nakaji P. Neuro-oncologic applications of endoscopy. Neurosurg Clin North Am 2004; 15(1):89–103. 49. Otsuki T, Jokura H, Nakasato N, Yoshimoto T. Stereotactic endoscopic resection of angiographically occult vascular malformations. Acta Neurochir Suppl 1994; 61: 98–101. 50. Schroeder HW, Oertel J, Gaab MR. Endoscope-assisted microsurgical resection of epidermoid tumors of the cerebellopontine angle. J Neurosurg 2004; 101(2):227–232.

Part Three PART III

Clinical Applications

9

MINIMAL ACCESS CRANIOTOMY USING SURGICAL NAVIGATION SYSTEMS Gene H. Barnett Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION The primary application of surgical navigation systems (SNSs) for intracranial use by most practitioners is likely to be fashioning minimal access craniotomies (MACs) for resection of primary and metastatic tumors (1–13). Knowing the precise location of an intracranial lesion along with vascular and cortical relationships may allow for very small cranial openings and minimize the need for intracranial exploration. Computer-assisted minimally invasive craniotomies appear to reduce neurological and wound complications as well as operative time and patient discomfort compared to traditional techniques (1,2,6,7,14). Proper use of SNSs to select a surgical trajectory (ST) (15) requires an understanding of the interpretation of SNS displays and strategies to manage brain movement after dural opening and local tissue deformations from the resection itself. This chapter reviews these issues as well as techniques used by the author in several hundred minimal access craniotomies. An MPEG movie of the author performing a MAC is included on the compact disc that accompanies this book.

SURGICAL TECHNIQUE Surgical Planning Although most SNSs allow for ‘‘on the fly’’ visualization of the projections of the navigation wand after registration, thereby allowing for use without preoperative planning, the author believes that in most cases it is worth the extra time (usually minimal) to plan the case and to more thoughtfully interpret all the imaging data

Icons indicate materials available on the accompanying CD.

147

148

&

PART III: CLINICAL APPLICATIONS

available to the person. Most navigation systems allow for the coregistration (fusion) of two or more image data sets. In addition to the high-resolution volume MRI typically used for navigation, other MRI sequences or functional data may prove useful for planning and during navigation at surgery. Examples of such data include magnetic resonance angiography or venography (Fig. 1) for display of relevant vessels, fluid attenuation inversion recovery (FLAIR—Fig. 2) to highlight low-grade tumor, and functional MRI to provide some information on cortical function. MRI sequences that show directionality of diffusion (e.g., anisotropic diffusion-weighted imaging or diffusion tensor imaging) are emerging as potentially important sources of information on subcortical white matter tract anatomy.

Figure 1 Magnetic resonance venogram (MRV) in blue, superimposed on high-resolution contrastenhanced MRI (gray). Abbreviation: MRI, magnetic resonance imaging.

CHAPTER 9: MINIMAL ACCESS CRANIOTOMY

&

149

Figure 2 Fluid attenuation inversion recovery (FLAIR) MRI in color, superimposed on high resolution contrast-enhanced MRI (gray). This display is useful for planning and intraoperative navigation of low-grade gliomas. Abbreviation: MRI, magnetic resonance imaging.

Display Interpretation A surgical trajectory avoiding functionally eloquent cortex and tracts is defined. Although the ultimate target is generally the center of the tumor or lesion, the best approach may not always be via the shortest distance. We have found that simple triplanar (i.e., axial, sagittal, and coronal) displays are inadequate for optimal planning of a surgical trajectory for several reasons: (i) They may be misleading, particularly with respect to where the margins of a tumor project to the scalp surface (Fig. 3). For instance, novice users of SNSs will often move the pointing device (wand) on the scalp surface to where the tumor margin disappears on a given plane and mark that spot as the projected tumor boundary. To do so is only valid if the

150

&

PART III: CLINICAL APPLICATIONS

Figure 3 (A) Proper placement of minimal access craniotomy requires proper use of SNS displays. Approaches must account for the curvature of the skull. If a standard triplanar display is used (i.e., sagittal, coronal, transverse), the craniotomy margin must be placed on a line perpendicular to the skull at the most extreme portion of the tumor (A) and not on a line either horizontal (B) or vertical on the display as the actual inner margin of the cut will be misdirected (C). This effect is minimized at the 12, 3, 6 and 9 o’clock positions of each display (D and D0 ). (B) Oblique displays are perhaps better suited to defining placement of the bony opening as long as the axis is perpendicular to the skull. Abbreviation: SNS, surgical navigation system.

final surgical trajectory (ST) is truly parallel to the line of projection. This situation only exists at the vertex, immediately above the ears, and above the inion and nasion. Elsewhere, physical constraints usually demand that the scalp and bony openings will be perpendicular to the surface of the skull in that region and this may depart as much as 45 from the line of projection when the above technique is used. (ii) It is difficult to interpret the ST with respect to vascular structures, especially major cortical and parasagittal veins, as these structures typically appear as nondescript spots of high intensity on most triplanar displays. (iii) Often, the relationship of the ST to cortical anatomy and its inferred function is obscure. For these reasons, the presentations that are steered by the pointing device are quite valuable. A pair of perpendicular views that both contain the wand axis can be used to insure that the ST is truly perpendicular to the skull, and a view that is perpendicular to the wand axis at a user-determined depth provides accurate projection of the lesion margins, as well as visualization of vascular and cortical anatomy that may suggest redirection of the planned ST (Fig. 4).

CHAPTER 9: MINIMAL ACCESS CRANIOTOMY

&

151

Figure 4 Oblique displays showing (A) superior sagittal sinus and draining vein, (B) motor strip (M), sensory strip (S), and astrocytoma (A).

Patient Positioning, Preparation, and Opening Ideally, the patient should be positioned so that this surgical trajectory is nearly vertical to help the surgeon manage intraoperative movements of the brain and tumor (‘‘brain shift’’) (1,6,7). This sag of the brain due to drainage of the cerebrospinal fluid (CSF) occurs principally downward (16), so using a vertical ST allows the surgeon to contend with shift largely confined to the ST (i.e., straight down) instead of some obscure three-dimensional angle to the ST. The MAC is located directly over a superficial lesion using the method described above to project the tumor boundary. Placement of the MAC for subcortical lesions is generally directed toward avoiding functionally eloquent cortex, major veins and, often, entering the brain through a sulcus to minimize parenchymal dissection (1,6,7). Superficial lesions require that the craniotomy be at least as large as the cortical presentation of the tumor, while deeper lesions allow for MACs smaller than the lesion, as the surgeon may work from the apex of a conical surgical corridor to the lesion. Lesions requiring frontal or temporal lobectomy are best accessed using MACs located at the posterior boundary of the lesion, typically at the coronal suture or a few centimeters posterior to the temporal tip, respectively. The head is secured in a fixation device so as to avoid displacement of reference fiducials (if used), to allow line-of-sight between the SNS wand, dynamic reference frame (i.e., head tracker) and detector (as required by the SNS), and to allow a near vertical surgical trajectory. A minimal hair shave is performed encompassing about 1 cm to either side of the prospective incision and the region prepared with an antiseptic solution such as povodone-iodine (Fig. 5) (17). Draping is tacked in place with skin staples to cover the remaining hair and a lazy S-shaped incision centered on the surgical trajectory is then fashioned (Fig. 6). Skin clips are usually not required if hemostasis is secured early. The periosteum is elevated and self-retaining retractors are placed (Fig. 7) in lieu of fish-hooks or other retractors that may move

152

&

PART III: CLINICAL APPLICATIONS

Figure 5 Minimal access craniotomy for malignant glioma begins with minimal hair shave over proposed incision which is centered on the SNS-defined surgical trajectory. Abbreviation: SNS, surgical navigation system.

the head in the positioning device and risk spatial misregistration. Some recommend drilling small holes for use as secondary registration points at this stage of the procedure. A small bone flap is elevated using either a trephine or craniotome (Fig. 8). The use of a trephine (Codman, Randolph, Massachusetts, U.S.A.) may provide a spatial reference when used with some SNSs (6,7), but may be more likely to result in cortical laceration than a craniotome in some hands. Dural tacking sutures are then placed through small holes drilled around the periphery of the craniotomy. For most lesions, the dura is opened in a cruciate fashion with attention to underlying venous structures viewed on the SNS (Fig. 9). It is advantageous to tent the dural flaps with sutures placed at the base of the leaflets rather than the apices, as this approach maximizes exposure. Dural-based lesions are approached by opening the dura circumferentially (Fig. 10).

Figure 6 Drapes are stapled over the remaining hair and a shallow S-shaped incision made. Hemostasis can usually be obtained by cautery without the need for clips.

CHAPTER 9: MINIMAL ACCESS CRANIOTOMY

&

153

Figure 7 The periosteum is elevated and self-retaining retractors placed to provide adequate opening for craniotomy.

Figure 8 A small (1–2 in) bone flap is then fashioned using a trephine or craniotome.

Figure 9 The dura is opened in four or more leaflets. It is best to retract the leaflets with sutures at their bases as it provides better exposure than when the apices are used. The superficial tumor is situated in the center of the minimal access craniotomy and is bounded by sulci laterally. The accuracy of the SNS can be confirmed by localizing in one of the sulci. Abbreviation: SNS, surgical navigation system.

154

&

PART III: CLINICAL APPLICATIONS

Figure 10 Dural-based lesions require opening the dura peripherally (dotted line).

Figure 11 Use of triplanar display showing location at edge of brain–tumor interface (cross hairs).

CHAPTER 9: MINIMAL ACCESS CRANIOTOMY

&

155

Figure 12 The sulcus is then opened and the plane of dissection carried into the gyrus.

The exposed cortical anatomy and veins provide an ideal opportunity to confirm the accuracy of the SNS prior to cerebrotomy. Using the view perpendicular to the wand, veins, gyri, and sulci should be clearly visible, especially when MRI is used for guidance, and correlate to what the surgeon sees. It is important to bear in mind that at this point the brain is often displaced a few millimeters through the dural plane so that there is often a predictable discrepancy of depth when using this technique. Triplanar displays can also be used to track visible sulci to ensure that there is concordance. Superficial lesions, particularly gliomas, often conform to sulci and when visually different from surrounding brain (Fig. 9) provide a reference to confirm the accuracy of the SNS. Lesion Resection and Tissue Deformation The localization function of a surgical navigation system can provide valuable information on the boundary between solid tumor and brain, particularly when

Figure 13 The localization function is used to assist defining the plane around the entire solid portion of the tumor. If the lesion is away from functionally eloquent brain, resection can be extended in an effort to obtain better margins.

156

&

PART III: CLINICAL APPLICATIONS

contrast-enhanced MRI or FLAIR is used for navigation (Fig. 11). Optimal use of these devices, however, requires some modification of traditional microsurgical techniques. Traditional gutting of the tumor followed by centripetal resection to surrounding brain (14) often renders preoperative imaging inaccurate as a navigational guide due to local tissue deformations. As the tumor shrinks, use of the SNS is still referenced to the original tumor size thereby suggesting that more tumor remains than in reality. Similarly, it is unwise to decompress cystic portions of the tumor early in the resection, as they may unpredictably displace the entire region of interest. For these reasons, tumor resection where the SNS is used to assist tumor– brain boundary definition is best accomplished by an en bloc resection (1,2,6,7). The tumor–brain interface, defined using visual, tactile, and SNS information, is developed (Fig. 12) working first around the superficial portion of the tumor and then proceeding to its base (Fig. 13). If a complete radiographic resection is secured then primary hemostasis is usually obtained easily without the need for lining the cavity with hemostatic agents (Fig. 14). Not all tumors can truly be removed in one piece without sacrificing safety. In such cases, attention is first directed to regions where information provided by the SNS would be most useful such as the portion nearest eloquent brain or deep structures. In situations where brain distortion or shift is suspected, it is better to leave tumor behind than risk intrusion into sensitive areas. Updated image data sets with ultrasound (18) computerized tomography, or intraoperative MRI may allow for better, safer resections than are feasible with preoperative imaging alone.

Figure 14 The tumor is then removed in essentially one piece and hemostasis obtained, usually without the need for topical agents.

CHAPTER 9: MINIMAL ACCESS CRANIOTOMY

&

157

Figure 15 Dural approximation is usually sufficient unless the ventricles or cisterns have been entered. Layers of fibrin glue, absorbable gelatin sponge, and fibrin glue are applied followed by the bone flap secured with mini-plates and screws.

Closure Wound closure is accomplished by reapproximation of dural leaflets (Fig. 15). It is often impractical or impossible to achieve a watertight dural closure if the dura is opened in a cruciate fashion. In this situation, a layer of fibrin glue is applied to the dura, followed by a piece of Gelfoam (Pharmacia & Upjohn Company, Kalamazoo, Michigan, U.S.A.) cut to the size of the bony defect, and then another layer of fibrin glue applied. The bone flap is then pressed into place and secured with small titanium plates and self-drilling, self-tapping screws. This approach results in a high-pressure zone in the extradural space that helps prevent the extracranial flow of CSF. A subgaleal drain may be placed, if desired. The scalp is closed with inverted, interrupted absorbable suture and the skin with a running vertical mattress suture, providing an additional barrier to CSF leakage and an excellent cosmetic result. Antibiotic ointment is placed on the incision and a small dressing is stapled to the skin (Fig. 16).

Figure 16 Antibiotic ointment is applied and a small dressing stapled to the scalp.

158

&

PART III: CLINICAL APPLICATIONS

SUMMARY Minimal access craniotomy may be facilitated by use of SNSs, but optimal results require knowledge of the limitations of SNSs based on preoperative imaging and some modifications of traditional neurosurgical techniques.

CHAPTER 9: MINIMAL ACCESS CRANIOTOMY

&

159

REFERENCES 1. Barnett GH. Stereotactic techniques in the management of brain tumors. Contemp Neurosurg 1997; 19(10):1–9. 2. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Use of a frameless, armless stereotactic wand for brain tumor localization with two-dimensional and threedimensional neuroirmaging. Neurosurgery 1993; 33:674–678. 3. Doshi PK, Lemmieux L, Fish DR, Shorvon SD, Harkness WH, Thomas DG. Frameless stereotaxy and interactive neurosurgery with the ISG viewing wand. Acta Neuroschir Suppl (Wien) 1995; 64:49–53. 4. Golfinos JG, Fitzpatrick BC, Smith LR, Spetzler RF. Clinical use of a frameless stereotactic arm: results in 325 cases. J Neurosurg 1995; 83:197–205. 5. Guthrie BL, Adler JR Jr. Computer-assisted preoperative planning, interactive surgery, and frameless stereotaxy. Clin Neurosurg 1992; 38:112–131. 6. Kelly PJ. Volumetric stereotactic surgical resection of intra-axial brain mass lesions. Mayo Clin Proc 1988; 63:1186–1198. 7. Kelly PJ, Kall BA, Goerss S, Earnest F. 4th computer-assisted stereotaxic laser resection of intra-axial brain neoplasms. J Neurosurg 1986; 64:427–439. 8. Maciunas RJ, Galloway RL Jr, Fitzpatrick JM, et al. A universal system for interactive image-directed neurosurgery. Stereotact Funct Neurosurg 1992; 58:108–113. 9. Manwaring KH. Magnetic field guided endoscopic dissection through a burr hole may avoid more invasive craniotomies—a preliminary report. ACTA Neurochir 1994; 61(suppl):34–39. 10. Murphy MA, Barnett GH, Koimos DW, Weisenberger J. Astrocytoma resection using an interactive frameless stereotactic wand. An early experience. J Clin Neurosci 1994; 1:33–37. 11. Rhoten RL, Luciano MG, Barnett GH. Computer-assisted endoscopy for neurosurgical procedures: technical note. Neurosurgery 1997; 40(3):632–637. 12. Roberts DW, Strohbehn JW, Friets EM, Kettenberger J, Hartov A. The stereotactic operating microscope: accuracy refinement and clinical experience. Acta Neurochir Suppl (Wien) 1989; 46:112–114. 13. Smith KR, Frank KJ, Bucholz RD. The NeuroStation—a highly accurate, minimally invasive solution to frameless streotactic neurosurgery. Comput Med Imaging Graph 1994; 18:247–256. 14. Walters CL, Schmidek HH. Surgical management of intracranial gliomas. In: Schmidek HH, Sweet WH, eds. Operative Neurosurgical Techniques: Indications, Methods and Results. Philadelphia: W.B. Saunders, 1988:431–450. 15. Barnett GH, Steiner CP, Weisenberger J. Target and trajectory guidance for interactive surgical navigation systems. Stereotact Funct Neurosurg 1996; 66:91–95. 16. Roberts DW, Hartov A, Kennedy FE, Miga MI, Paulsen KD. Intraoperative brain shift and deformation: a quantitative analysis of cortical displacement in 28 cases. Neurosurgery 1998; 43(4):749–758. 17. Winston KR. Hair and neurosurgery. Neurosurgery 1992; 31:320–329. 18. Trobaugh JW, Richard WD, Smith KR, Bucholz RD. Frameless stereotactic ultrasonography: method and applications. Comput Med Imaging Graph 1994; 18:235–246.

10

IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS Vitaly Siomin Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Michael A. Vogelbaum Brain Tumor Institute and Department of Neurological Surgery, Taussig and Case Comprehensive Cancer Centers, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION Epidemiology of Metastatic Brain Tumors Cerebral metastases (CM) are the most common brain tumors seen in clinical practice, comprising over half of all brain tumors. The annual incidence of CM in the United States is more than 150,000 cases, compared to only 17,000 for primary brain tumors (1). About one-third of patients with systemic cancer eventually develop CM, and in patients lacking any cancer history about 15% present initially with CM (2–4). Various types of primary cancer metastasize to the brain with different frequencies. Table 1 shows the sources of CM in patients who underwent an autopsy at Sloan-Kettering Cancer Center (5,6). These data reinforce two major points about CM: & &

Over half of all CM come from lung and breast primary sources. Primary cancers have varied propensities to metastasize to the brain, with highest prevalence of CM observed in melanoma patients.

Computed tomography (CT) data suggest that CM from solid tumors tend to be solitary in 50% of patients at diagnosis (7). However, magnetic resonance imaging (MRI)-based studies suggest that only less than one-third of patients have a solitary CM at the time of diagnosis of CM (2,8). Such a considerable difference in sensitivity is an indicator of the inadequacy of CT in determining the number of CM. This issue is further discussed in the Imaging section of this chapter. 161

162

&

PART III: CLINICAL APPLICATIONS

Table 1 Most Common Primary Cancers in Patients with Cerebral Metastases (CM) % of given primary out Autopsy incidence of CM for Primary cancer of all autopsy cases given primary cancer Lung Breast Melanoma Gastrointestinal Renal cell

40.5 18.6 9.5 6.7 6.2

34 30 72 7 23

Source: From Refs. 5, 6.

The majority of CM are located in the cerebral hemispheres (80–85%). Up to 15% of metastases develop in the cerebellum and only 1% to 5% of patients have metastases in the brain stem (9). Most parenchymal CM arise near the junction of parietal, temporal, and occipital lobes, behind the Sylvian fissure, possibly due to an embolic spread of tumor cells to the distal MCA branches (10). Although, most CM look well demarcated from surrounding brain on imaging and even seem to have a ‘‘capsule’’ that some surgeons try to preserve during operation to ensure ‘‘complete’’ removal, there is evidence that these lesions may actually have an infiltrative growth pattern (11). CLINICAL FEATURES Clinical features of CM generally are similar to those arising from other expanding brain mass lesions (12,13). Signs and symptoms can be broadly categorized into four groups: (i) as a consequence of increased intracranial pressure (either due to mass effect or blocked CSF flow); (ii) location-dependent focal neurological deficits; (iii) seizures; and (iv) depressed mental status. Patients with single CM tend to present with focal signs, whereas presentation of patients with multiple metastases may be more diffuse and nonlocalized. In patients with no known primary cancer, it may be difficult to differentiate metastatic disease from a primary brain tumor based upon clinical grounds alone. Indeed, even in patients with known primary cancer a newly diagnosed brain mass can turn out to be a primary brain tumor, or other non-metastatic disease in about 5% of cases (3). IMAGING MRI with gadolinium enhancement is by far the most sensitive imaging modality for both diagnosis and follow-up of patients with CM (8,14,15). T2-weighted images are used to estimate the extent of peritumoral edema. A metastasis itself may have various signal intensities depending on the types of tissue present within the lesion (e.g., blood products, necrosis, melanin). The appearance of blood products on T1-weighted images depends on the ‘‘age’’ of bleeding, and ranges from isointense in the first 24 hours to hyperintense after 24 to 72 hours. Melanin is

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

163

paramagnetic and hyperintense on T1 and hypointense on T2-weighted images. Metastatic melanomas often have both components (blood and melanin), which appear bright on noncontrast T1-weighted scans. Carcinomatous meningitis is usually seen as an irregular brightly enhancing pial surface on T1-weighted imaging with contrast. The arachnoidal surfaces, ventricular ependyma, or dura may enhance pathologically as well. Although MRI is the imaging modality of choice in diagnosis of CM, in ‘‘real life’’ most physicians first perform CT to confirm the diagnosis suggested by history and clinical picture. Bell et al. (16) reported an 82% accuracy of CT in CM diagnosis. The authors did not specifically address its accuracy in evaluating the posterior fossa, where CT scans are particularly unreliable. This raises a question whether CT is an adequate screening tool for CM. The answer is probably ‘‘no,’’ and it is the opinion of these authors that suspected CM cannot be ruled out without contrast-enhanced MR imaging. The radiological diagnosis of CM in a patient with multiple lesions and suggestive history is usually straightforward. The diagnosis of CM in a patient with a single lesion even with progressive systemic cancer is more challenging. Primary brain tumors, abscess, meningioma, or lymphoma may resemble CM and should be ruled out. The issue becomes even more complicated in patients without a history of systemic disease. The probability that an otherwise ‘‘typically’’ looking lesion is metastasis in such cases is less than 15% (17). The interpretation of radiological follow-up studies in patients who underwent a complete surgical resection of their tumor(s) is more straightforward, as any new enhancement in the lesion site can be considered recurrence. This is not always the case after radiosurgery, where recurrent or residual tumor needs to be differentiated from radiation necrosis. Radiation necrosis and recurrent brain tumor can present with similar symptoms and are usually indistinguishable on MRI. 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) has been proposed as a diagnostic alternative, particularly when coregistered with MRI. For brain metastasis with MRI coregistration, FDG PET had a sensitivity of 86% and specificity of 80% (18). There are few preliminary reports on utilization of MR spectroscopy to distinguish between radiation necrosis and recurrent brain metastasis (19,20).

TREATMENT General Considerations Since the purpose of this chapter is to describe the application of image guidance in treatment of metastatic brain tumors, we will focus primarily on the two areas where this technology is utilized most: radiosurgery and image-guided surgery. When treating patients with CM, therapeutic decisionmaking is not always straightforward. One must consider the status of the systemic disease, patient’s neurological status, number and location of lesions, and presence of leptomeningeal disease. It has been shown in multiple studies that the status of the primary

164

&

PART III: CLINICAL APPLICATIONS

Table 2 Considerations in Therapeutic Decisionmaking Status of systemic disease Neurological status Age Number of metastases Location of lesions Presence of leptomeningeal disease Other medical morbidities

cancer remains the most important factor influencing survival in patients undergoing surgical resection (21,22). In addition, even in highly selected and aggressively treated cases of CM, up to 70% of such patients die of progression of the primary cancer, rather than neurological causes (3). Therefore, the results of treatment of patients with poorly controlled systemic disease are often disappointing. A thorough preoperative work-up to properly stage the systemic cancer is appropriate, if done within a reasonable time frame. A summary of considerations in therapeutic decision-making is provided in Table 2. The patient’s preoperative neurological status also correlates well with the outcome; patients with severe neurological deficits tend to have shorter survival (21,22). Most current studies use recursive partitioning analysis (RPA) of prognostic factors to assign patients with brain metastases to one of three prognostic classes. Factors evaluated to determine the RPA class of a patient include Karnofsky performance status (KPS), status of primary disease, presence of extracranial metastases and age. RPA Class 1 includes patients with KPS 70, younger than 65 years, with controlled primary disease and no extracranial metastases and it has the best prognosis. Patients in RPA Class 3 include those with KPS 60, and this class has the worst prognosis. All other patients fall into Class 2 (23). Other factors, negatively influencing survival, are presence of multiple metastases, infratentorial location, and leptomeningeal spread (24). Based on these considerations, the clinician must initially choose between either palliative treatment [steroids, whole brain radiotherapy (WBRT)], or aggressive treatment (surgery, radiosurgery with or without WBRT, chemotherapy).

IMAGE-GUIDED SURGERY Surgical resection of CM was considered for many years a form of palliative therapy only, having most benefit for patients with radioresistant tumors, such as colon carcinoma, melanoma, or renal-cell carcinoma (24–27). More recently, prospective studies have demonstrated that, in appropriately selected cases, surgical treatment can effectively prolong survival in patients with CM. The main goals of surgery are listed in Table 3. Surgical removal of a CM leads to immediate eradication of the tumor, elimination of mass compressing surrounding brain structures or causing blockage of

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

165

Table 3 Goals of Surgery for Brain Metastases Immediate relieve of symptoms due to mass effect Obtain a tissue diagnosis Obtain local control Avoid producing new neurological deficits

the CSF flow, and removal of the source of perifocal edema. Surgery is particularly of use in patients with large (over 3 cm in size) lesions, especially those in the posterior fossa. Surgical patients may also benefit from a more rapid taper of steroids in the postoperative period, thereby reducing the risk of potential complications associated with their use. These complications include hyperglycemia, GI ulceration, weight gain, immunosuppression, and peripheral edema (28). Surgical resection of metastases is also of value to confirm, or define the diagnosis. Patchell et al. (3) noted that 6 of 54 patients thought to have ‘‘classical’’ metastases before surgery turned to harbor other lesions, of which three were non-neoplastic. Even in an era of widespread use of novel imaging techniques (e.g., MRI, MR Spectroscopy, PET, etc.), surgery remains the only treatment modality that provides actual tissue diagnosis. The benefit of surgery in the treatment of a single CM has been demonstrated in a number of prospective studies. Patchell et al. (3) showed a significant increase in median survival of surgical patients who underwent gross total resection of a single CM followed by WBRT (median survival of 40 weeks vs. 15 weeks in the WBRT-only group). Median survivals following gross total resection with WBRT reported in other studies ranged from 10 to 14 months, depending on pathology (25,29). One randomized prospective study, however, disputed the benefit of surgery. The authors of this study did not find any significant difference in survival among patients treated with surgery and WBRT versus WBRT alone (5.6 vs. 6.3 months, respectively) (2). It should be noted, however, that 73% of patients in this study had poorly controlled extracranial disease, the distribution of primary pathologies was not equal between the groups (i.e., greater proportion of radioresistant colorectal cancer in surgery group and radiosensitive breast cancer in WBRT group), and survival times were not calculated in uniform manner. Van der Ree et al. reported that surgical resection of CM, particularly in the posterior fossa, may cause leptomeningeal dissemination of the tumor. In their series, 33% of patients developed leptomeningeal metastasis 2 to 13 months after surgery, which included six of the nine patients operated on for posterior fossa metastasis (30). Some authors, therefore, suggest an en bloc removal of metastatic lesions (30,31). In our opinion, image guidance becomes essential if such approach is utilized. Not all patients will benefit from surgical resection. The location and accessibility of the tumor(s) is a crucial factor in surgical decision-making. Moser and Johnson (32) defined accessibility as ‘‘the risk and extent of neurologic injury the patient is willing to accept’’. Two essential parameters taken into consideration in assessing the accessibility are: (i) the depth of the lesion, and (ii) proximity to the eloquent cortex (e.g., speech or language areas, motor strip). Due to advances in

166

&

PART III: CLINICAL APPLICATIONS

Figure 1 Metastatic adenocarcinoma in the left frontal lobe of the right-handed patient. (A) axial view, (B) coronal view, (C) sagittal view. Note the location of the mass within the Broca speech area. Postoperative axial, (D) and coronal, (E) views demonstrating complete removal of the lesion. Small hyperdense foci in the tumor bed represent blood products.

image guidance, intraoperative imaging, and functional mapping, many CM, previously thought to be associated with prohibitive surgical morbidity, are now accessible and can be resected with acceptable levels of morbidity (Fig. 1). From a technical standpoint, computer-assisted image-guided surgery has become the most advanced standard method of treatment of CM (27,33). For image-guided surgery, preoperative images (including various sequences of ‘‘regular’’ MR, functional MR, or CT) are transferred to a computer workstation (navigation system) where they are coregistered with radiographic markers on the patient’s head and a reference device, which is fixed to the head holder and which can be tracked throughout the procedure by the navigation system. The surgeon then can easily identify the tumor and normal structures on the navigation system planning station. The high degree of precision provided by the stereotactic navigation system allows the surgeon to select the safest possible route to the tumor, make smaller craniotomies, minimize exposure of surrounding normal brain, and make the surgery more ‘‘elegant.’’ Addition of intraoperative MRI or

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

167

ultrasound adds a ‘‘real-time’’ imaging capability to surgery and may further increase the accuracy, particularly after the brain shift has occurred. When operating on metastases located in the eloquent areas, many surgeons use intraoperative functional mapping of motor or sensory cortex, or perform surgeries under awake conditions with monitoring of speech or motor function throughout resection. The details of these techniques are beyond the scope of this chapter and discussed elsewhere in the book. The employment of these adjuncts along with refined microsurgical techniques can bring surgical mortality and morbidity to the acceptable levels of 1.6% and 7.7%, respectively (34). The justification for the use of surgical resection for the treatment of multiple CM is less clear than it is for single or solitary CM. Uncontrolled, retrospective studies (35,36) suggested that patients with multiple CM did not receive any benefit from surgery, compared to the historical results with WBRT alone. These studies included patients with advanced systemic disease, or cases in which only a fraction of the total number of metastases were treated with surgery. A significantly longer survival was found in patients who underwent removal of all metastases than in patients in whom at least one lesion was not resected (14 vs. 6 months) (26). The conclusion of this study was that removal of all tumors in patients with multiple metastases may have survival rates comparable to what is observed following resection of a single metastasis.

Is WBRT Required in Combination with Surgery? In all of the prospective studies examining the role of surgery for the management of patients with CM, all patients received WBRT following surgery. Several investigators subsequently examined whether WBRT is necessary following surgical resection of a CM. A retrospective review from the Mayo Clinic examined patients who had undergone surgical resection of a single CM followed by observation or WBRT. The authors found that 85% of the patients in the observation group subsequently had brain relapse (defined as local or distant tumor growth) whereas relapse occurred in only 21% of patients in those treated with WBRT. Median survival was longer for the WBRT group as well (21 vs. 11.5 months) (37). A randomized trial of observation versus postoperative WBRT in patients with gross total resection of a single CM showed that recurrence of tumor in the brain was less frequent in the WBRT group than in observation group (18% vs. 70%). Furthermore, patients in the WBRT group were less likely to die of neurologic causes than patients in the observation group (14% vs. 37%). However, there was no difference in the overall survival or the length of time that patients remained independent as the majority of patients died of their systemic disease (38). The failure of most studies to demonstrate any advantage in survival from adding WBRT to surgery can be explained by the fact that most patients with CM die from systemic progression too early to definitively determine the duration of brain control. Consequently, the rationale for withholding WBRT after surgery may need to be re-evaluated in the future, as the results of systemic cancer improve and survival time increase.

168

&

PART III: CLINICAL APPLICATIONS

Potential alternatives to WBRT for maintenance of local control following surgery are brachytherapy or intracavitary chemotherapy. While there are multiple delivery systems available for these modalities, there is not a sufficient clinical experience available at this time to evaluate their efficacy. A phase II multicenter study evaluating the local control efficacy of brachytherapy following gross total resection of a single brain metastasis recently closed to accrual and data should be available to the public early in 2004.

STEREOTACTIC RADIOSURGERY Stereotactic radiosurgery (SRS) is a technique introduced in the early 1950s by the Swedish neurosurgeon Lars Leksell, M.D. His original idea was to destroy the lesion with a high dose of radiation precisely focused from dozens of small radiation beams. The high precision is achieved due to computer-guided stereotactic principles employed in these machines. There are two methods available to deliver SRS. The first is the Leksell gamma knife (Elekta), which operates using the original principle developed by Dr. Leksell. This device consists of 201 concentrically aligned 60Co sources, which emit gamma irradiation. Four different sizes of collimators are available and are placed into spherical ‘‘helmets,’’ which, when in the treatment position, align with the radiation sources. The center of delivery of each dose (or ‘‘shot’’) of radiation is always in the center of the sphere defined by the helmet; the head position within the helmet is determined by the treatment planning system and set using a stereotactic frame. The second method for delivering SRS is by X radiation produced by a linear accelerator (LinAc). There are several systems designed to utilize this latter method. Unlike the gamma knife, where the patient’s head is moved from position to position, in LinAc radiosurgery the patient’s head is maintained in a fixed position and the linear accelerator is moved along arcs of travel, or from point to point. Gamma knife SRS can be performed only on lesions above the foramen magnum; LinAc radiosurgery can be used to treat lesions anywhere in the body as long as stereotactic localization can be performed accurately. Radiosurgery is an outpatient procedure. Patients undergo stereotactic frame placement under mild intravenous sedation and local anesthetic infiltration of the pin sites. A contrast-enhanced MRI with 1 mm axial slices and CT studies are obtained. In our opinion, CT fused with MRI may both enhance accuracy of planning and serve as an internal control for spatial accuracy, to offset potential warping produced by an inhomogeneous MRI field. The images are exported to a workstation where the contours of the lesion(s) are defined and the treatment planning is performed. Conformal SRS allows delivery of radiation to only a small volume of surrounding normal brain with a rapid fall-off of the dose. Thus, brain tissue as close as 2 to 3 mm away from the lesion may receive less than 20% of the peak dose (Fig. 2). Unlike conventional fractionated radiosurgery, which requires activation of cell death processes, SRS is believed to cause a direct cytotoxic effect on the tumor tissue. Traditionally, radioresistant histologies tend to be more sensitive

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

169

Figure 2 Recurrent metastatic left frontal lobe renal cell carcinoma treated with the Gamma Knife (Elekta, Stockholm, Sweden) radiosurgery. Axial, coronal, and sagittal views are seen on the screenshot. Inner line, 50% isodose; outer line, 30% isodose.

to SRS than to conventional fractionated radiation treatment. In addition, SRS causes indirect vascular injury and subsequent sclerosis of the blood vessels and eventual compromise of blood supply and circulation within the tumor (39). Various doses have been used to treat CM in many retrospective studies. A prospective Radiation Therapy Oncology Group 90–05 study examined the maximum tolerated dose (MTD) of single fraction RS in patients with tumors (both primary and metastatic) previously treated with conventional fractionated radiation therapy (40). MTD was defined as the dose of SRS that produced greater than a 5% incidence of radiation-induced swelling or necrosis. They found that MTD for tumors having a maximum dimension from 3.1 to 4.0 cm was 15 Gy and for tumors 2.0 to 3.0 cm in size it was 18 Gy. The MTD for tumors less than 2 cm in size was not identified, as the investigators did not go above a dose of 24 Gy. The potential advantages of RS are listed in Table 4. The majority of studies evaluating the efficacy of SRS have been retrospective. Flickinger et al. reported a multicenter series of 116 patients treated with SRS alone, SRS combined with WBRT up front, or SRS combined with WBRT following recurrence. The median survival for the entire group of patients was

170

&

PART III: CLINICAL APPLICATIONS

Table 4 Advantages of Radiosurgery Noninvasive, outpatient procedure Can treat surgically inaccessible lesions Useful for multiple metastases Does not require general anesthesia Lower risk of neurological complications Lower cost

11 months with a local tumor control rate of 85%. The authors noted particular benefit for patients with ‘‘radioresistant’’ histologies (i.e., melanoma and renal cell carcinoma). These results are very favorable compared to WBRT alone, in which case the median survival is approximately 3 to 4 months (41). Similar results have been reported in five more recent large studies with median survivals ranging from 7 to 15 months and the local control rates of 84% to 100% (25,42–46). These initial studies were followed by studies, which included various combinations of SRS and WBRT and which produced mixed results. In one study, no significant survival difference was observed between SRS alone and SRS plus WBRT. However, in a subgroup analysis of patients who had no evidence of extracranial disease, an increased survival was seen in patients who received WBRT in addition to SRS (15.4 vs. 8.3 months in SRS only group) (47). In contrast, Goyal et al. found that in patients with renal cell carcinoma the addition of WBRT to SRS did not affect survival. Nevertheless, the authors concluded that patients who present with multiple brain lesions may be more likely to benefit from the addition of WBRT because they appeared to be considerably more prone to develop distal brain failure than patients with single lesion (48). Kondziolka et al. performed a prospective randomized study to compare SRS plus WBRT versus WBRT alone in patients with multiple CM. The rate of local failure was 100% in WBRT-alone group and only 8% in patients who had boost SRS, a finding that lead to premature termination of the study. The median time to local failure was 6 months after WBRT alone in comparison to 36 months after WBRT plus SRS. Patients who received WBRT alone lived a median of 7.5 months, while those who received WBRT and SRS lived 11 months (49). A recently completed multicenter prospective, randomized trial examined the use of a radiosurgical boost following WBRT for up to three brain metastases (40). This study, RTOG 9508, was stratified for patients with a single brain metastasis and those with two to three brain metastases. All patients received WBRT. The results of this study showed that the addition of SRS to WBRT improves survival for solitary CM (6.5 vs. 4.9 months in WBRT-alone group), but had no survival benefit in patients with two to three metastases. Survival benefit was also seen in RPA Class 1 and patients younger than 50 years old, and in cases of small cell and non-small cell lung cancer. This study also demonstrated improved local control with addition of RS in both single and multiple CM groups and reduced steroid usage. Of note, more than two to three patients died as a result of their systemic disease progression, which undoubtedly had an impact upon the ability of the study to demonstrate a survival advantage associated with SRS. Although disappointing in terms of the impact, SRS had on overall survival in this

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

171

study; it did have a clear impact on quality of life. Further prospective studies are being designed to try to more carefully select for patients who have the best possible prognosis, from the standpoint of their systemic disease. An ongoing randomized prospective trial of SRS alone versus SRS plus WBRT for the treatment of patients with one to three CM will help further define the respective roles of SRS and WBRT in the management of CM (ACoSOG Z0300). This study, which opened in 2002, will accrue over a period of five years, so results will not be available until 2008 at the earliest.

SURGERY OR RADIOSURGERY? Although almost every retrospective study carried out in the last decade calls for a multicenter, prospective, randomized trial of SRS versus surgery for surgically resectable lesions, no such study has been completed. Largely because of the relative ‘‘ease’’ of performing SRS and the widespread belief that it costs less than surgery, one may think that SRS should eventually replace open surgery, at least for lesions smaller than 3 cm. The lack of unequivocal data to support the superiority of surgery versus SRS only adds to this controversy. We are aware of three published papers more or less directly comparing the results of surgical resection and SRS in the treatment of CM. Bindal et al. reported on a retrospectively matched series of 75 patients treated by surgery and SRS. The median survival was 7.5 months in the RS and 16.4 months in the surgical groups. The authors concluded that surgery was superior to SRS in the treatment of CM (25). In contrast, Auchter et al. retrospectively analyzed a multi-institutional outcome of SRS in 122 highly selected patients with a resectable single CM. WBRT was performed on all patients except five. The median survival was 56 weeks, comparable to the results of most surgical series (50). More recently, O’Neil et al. retrospectively compared these two modalities in treatment of solitary CM and found no significant difference in patient survival. Their difference in local tumor control rate, however, was significantly different (100% following SRS and 58% after surgery) (51). One of the arguments of the proponents of surgery is rapid improvement of neurological symptoms after removal of the mass. However, there have been reports of reduced mass effect and neurological symptoms following SRS as well. Hoshi et al. (42) reported a series of patients with renal cell carcinoma in which 80% of patients had rapid neurological improvement after SRS and complete disappearance of the lesion was seen in 9% of cases. In addition, Sheehan et al. (45) showed that in patients with non-small cell lung cancer 60% of tumors decreased in size, whereas 24% remained stable and 16% enlarged after SRS. It has been suggested that SRS may be more cost effective than surgery for the treatment of CM. Two studies have shown SRS to be associated with a lower cost than surgical resection (based upon U.S. reimbursement patterns). Rutigliano et al. estimated costs from the societal viewpoint using the 1992 Medicare Provider Analysis and Review database. The analysis revealed that SRS had a lower uncomplicated procedure cost ($20,209 vs. $27,587), a lower average complication cost

172

&

PART III: CLINICAL APPLICATIONS

per case ($2534 vs. $2874), and a lower total cost per procedure ($22,743 vs. $30,461), was more cost effective ($24,811 vs. $32,149 per life year), and had a better incremental cost effectiveness ($40,648 vs. $52,384 per life year) than surgical resection. The authors also found that elimination of all surgical resection morbidity cost would still result in superior incremental cost effectiveness for SRS (52). Mehta et al. reviewed a total of 46 surgical resections and 135 RS procedures during the period of time from 1989 to 1994. These data were compared to the analysis of 454 WBRT procedures performed within the same time frame. The results indicated that both resection and RS yielded superior survival and functional independence compared to WBRT alone. Surgical resection, however, resulted in 1.8-fold increase in cost compared to RS. The average cost per week of survival was $310 for radiotherapy, $524 for resection plus radiation, and $270 for RS plus radiation (53). It should be pointed out, however, that (1) both of these reports may not be currently relevant, as analysis was based on cost data which are a decade old, and (2) although the differences were statistically significant, they should not create an impression that SRS is ‘‘cheaper’’ solution for all patients. The results of these studies examining both cost efficacy and clinical efficacy of SRS and surgery demonstrate that both of these modalities are associated with comparable survival rates. We have summarized the advantages and disadvantages of each treatment modality in Table 5. In our opinion, the concept that these treatment modalities are competitors is artificial and without merit. Instead, they should Table 5 Comparison of Stereotactic Radiosurgery and Image-Guided Surgery Characteristic Stereotactic radiosurgery Image-guided surgery Precision Tumor control rate Treatment of small deep-seated lesions Tolerability

Excellent 84–95% (40,44,45,51) Excellent

Excellent 46–58% (38,51) Poor

Excellent

Need for general anesthesia Postoperative recovery

Anxiolytic and local anesthetic for frame placement Not an issue

Need for postoperative steroids Mass effect

Prolonged use of steroids may be required Rarely eliminated

Tissue diagnosis Follow-up

Not available More difficult, as the lesions rarely disappear, need to distinguish from radionecrosis Lower

May be poor due to age/ medical conditions Usually general anesthesia, except awake craniotomies Usually takes a few weeks, more after posterior fossa surgery Usually can be tapered off quickly Immediate elimination after resection Available Easier, any new enhancement after complete removal is probably recurrence Higher (52,53)

Cost

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

Table 6 Indications for SRS and Open Surgery Surgery Surgically accessible tumor >35 mm Mass effect (especially in posterior fossa) with or without a neurological deficit Seizure control Uncertain diagnosis

&

173

Stereotactic radiosurgery Multiple lesions Tumor <35 mm No mass effect In or near eloquent cortex Deep lesion Poor anesthesia risk Radioresistant histology

be considered complimentary. A complimentary, multifaceted approach is particularly suitable for patients with multiple CM, who may often have the largest accessible lesions resected surgically and small inaccessible lesions treated with SRS. Pollock et al. demonstrated a good illustration of this ‘‘aggressively eclectic’’ approach. They concluded that properly selected patients with multiple CM may benefit from aggressive local therapy that would employ both treatment modalities (54). Although we attempted to summarize the indications for these two modalities in Table 6, it is our belief that care should be individualized as not all patients would fit precisely into this treatment paradigm.

SUMMARY CM is very common in patients with systemic cancer. Employment of various treatment modalities, particularly SRS and image-guided surgery, allows clinicians who are focused upon the treatment of CM to achieve superior levels of control of brain disease and the result is that patients’ prognoses are much more dependent upon the status of their systemic disease. Although the existing published data allows us to define indications for each treatment approach, the needs of every individual patient should guide the therapy. In modern days, surgery and SRS should no longer be considered rivals, but rather complimentary tools in aggressive treatment of metastatic brain disease.

174

&

PART III: CLINICAL APPLICATIONS

REFERENCES 1. Posner JB. Management of brain metastases. Rev Neurol (Paris) 1992; 148: 477–487. 2. Mintz AP. Treatment of a single brain metastasis. JAMA 1998; 280:1527–1529. 3. Patchell RA, Tibbs PA, Walsh JW. A randomized trial of surgery in the treatment of single metastasis to the brain. N Engl J Med 1990; 322:494–500. 4. Voorhies RM. The single supratentorial lesion. J Neurosurg 1980; 53:364–368. 5. Posner JB. Neurologic Complications of Cancer. Contemporary Neurology Series. Philadelphia: Davis1995; Vol. 45:3–14, 77–110. 6. Posner JB, Chernik NL. Intracranial metastases from systemic cancer. Adv Neurol 1978; 19:579–592. 7. Zimm S. Intracerebral metastases in solid-tumor patients: natural history and results of treatment. Cancer 1981; 48:384–394. 8. Davis PC, Hudgins PA. Diagnosis of cerebral metastasis: double-dose delayed CT vs. contrast-enhanced MR imaging. Am J Neuroradiol 1991; 12:293–300. 9. Delattre JY. Distribution of brain metastases. Arch Neurol 1988; 45:741–744. 10. Kindt GW. The pattern of location of cerebral metastatic tumors. J Neurosurg 1964; 21:54–57. 11. Henson RA, Urich H. Cancer and the Nervous System. London: Blackwell Scientific, 1982:657. 12. Cairncross JG, Kim JH, Posner JB. Radiation therapy for brain metastases. Ann Neurol 1980; 7(6):529–541. 13. Hirsch FR, Paulson OB, Hansen HH, Larsen SO. Intracranial metastases in small cell carcinoma of the lung. Prognostic aspects. Cancer 1983; 51(3):529–533. 14. Healy ME, Hesselink JR. Increased detection of intracranial metastases with intravenous Gd-DTPA. Radiology 1987; 165:619–624. 15. Sze G. Detection of brain metastases: comparison of contrast-enhanced MRI with unenhanced MRI and enhanced CT. Am J Neuroradiol 1990; 11:785–791. 16. Bell D, Grant R, Collie D, Walker M, Whittle IR. How well do radiologists diagnose intracerebral tumour histology on CT? Findings from a prospective multicentre study. Br J Neurosurg 2002; 16(6):573–577. 17. Lang FF, Wildrick DM, Sawaya R. Metastatic brain tumors. In: Bernstein M, Berger M, eds. Neurooncology. The Essentials. New York: Thieme Medical Publishers, 2000:329–337. 18. Chao ST, Suh JH, Raja S, Lee SY, Barnett G. The sensitivity and specificity of FDG PET in distinguishing recurrent brain tumor from radionecrosis in patients treated with stereotactic radiosurgery. Int J Cancer 2001; 96(3):191–197. 19. Galanaud D, Nicoli F, Le Fur Y, Roche P, Confort-Gouny S, Dufour H, Ranjeva JP, Peragut JC, Viout P, Cozzone PJ. Contribution of magnetic resonance spectrometry to the diagnosis of intracranial tumors [in French]. Ann Med Interne (Paris) 2002; 153(8):491–498.

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

175

20. Sabatier J, Ibarrola D, Malet-Martino M, Berry I. Brain tumors: interest of magnetic resonance spectroscopy for the diagnosis and the prognosis [in French]. Rev Neurol (Paris) 2001; 157(8–9 Pt 1):858–862. 21. Galicich JH, Sundaresan N. Surgical treatment of single brain metastasis: factors associated with survival. Cancer 1980; 45:381–386. 22. Sundaresan N, Galicich JH. Surgical treatment of brain metastases. Clinical and computerized tomography evaluation of the results of treatment. Cancer 1985; 55: 1382–1388. 23. Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997; 37(4):745–751. 24. Lang FF, Sawaya R. Surgical management of cerebral metastases. Neurosurg Clin North Am 1996; 7:459–484. 25. Bindal AK, Bindal RK, Hess KR, Shiu A, Hassenbusch SJ, Shi WM, Sawaya R. Surgery versus radiosurgery in the treatment of brain metastasis. J Neurosurg 1996; 84(5):748–754. 26. Bindal RK, Sawaya R, Leavens ME. Surgical treatment of multiple brain metastases. J Neurosurg 1993; 79:210–216. 27. Lang FF, Sawaya R. Surgical treatment of metastatic brain tumors. Semin Surg Oncol 1998; 14:53–63. 28. Hempen C, Weiss E, Hess CF. Dexamethasone treatment in patients with brain metastases and primary brain tumors: do the benefits outweigh the side-effects? Support Care Cancer 2002; 10(4):322–328. 29. Vecht CJ, Haaxma-Reiche H, Noordijk EM, Padberg GW, Voormolen JH, Hoekstra FH, Tans JT, Lambooij N, Metsaars JA, Wattendorff AR, et al. Treatment of single brain metastasis: radiotherapy alone or combined with neurosurgery? Ann Neurol 1993; 33(6):583–590. 30. van der Ree TC, Dippel DW, Avezaat CJ, Sillevis Smitt PA, Vecht CJ, van den Bent MJ. Leptomeningeal metastasis after surgical resection of brain metastases. J Neurol Neurosurg Psychiatry 1999; 66(2):225–227. 31. Salvati M, Cervoni L, Caruso R, Gagliardi FM. Solitary cerebral metastasis from melanoma: value of the ‘en bloc’ resection. Clin Neurol Neurosurg 1996; 98(1):12–14. 32. Moser RP, Johnson ML. Surgical management of brain metastases: how aggressive should we be? Oncology 1989; 3:123–128. 33. Lang FF, Wildrick DM, Sawaya R. Management of cerebral metastases: the role of surgery. Cancer Control 1998; 5(2):124–129. 34. Muacevic A, Kreth FW, Horstmann GA, Schmid-Elsaesser R, Wowra B, Steiger HJ, Reulen HJ. Surgery and radiotherapy compared with gamma knife radiosurgery in the treatment of solitary cerebral metastases of small diameter. J Neurosurg 1999; 91(1):35–43. 35. Haar F, Patterson RH Jr. Surgery for metastatic intracranial neoplasm. Cancer 1972; 30:1241–1245. 36. Ransohoff J. Surgical management of metastatic tumors. Semin Oncol 1975; 2:21–27.

176

&

PART III: CLINICAL APPLICATIONS

37. Smalley SR, Schray MF, Laws ER Jr, O’Fallon JR. Adjuvant radiation therapy after surgical resection of solitary brain metastasis: association with pattern of failure and survival. Int J Radiat Oncol Biol Phys 1987; 13(11):1611–1616. 38. Patchell RA, Tibbs PA, Regine WF. Postoperative radiotherapy in the treatment of single metastases to the brain: a randomized trial. JAMA 1998; 280(17):1485–1489. 39. Szeifert GT, Massager N, DeVriendt D. Observations of intracranial neoplasms treated with gamma knife radiosurgery. J Neurosurg 2002; 97(5 suppl):623–626. 40. Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loeffler J, Farnan N. Single dose radiosurgical treatment of recurrent previously irradiated primary brain tumors and brain metastases: final report of RTOG protocol 90–05. Int J Radiat Oncol Biol Phys 2000; 47(2):291–298. 41. Flickinger JC, Kondziolka D, Lunsford LD, Coffey RJ, Goodman ML, Shaw EG, Hudgins WR, Weiner R, Harsh GR 4th, Sneed PK, et al. A multi-institutional experience with stereotactic radiosurgery for solitary brain metastasis. Int J Radiat Oncol Biol Phys 1994; 28(4):797–802. 42. Hoshi S, Jokura H, Nakamura H. Gamma-knife radiosurgery for brain metastasis of renal cell carcinoma: results in 42 patients. Int J Urol 2002; 9(11):618–625. 43. Petrovich Z, Yu C, Gianotta SL. Survival and pattern of failure in brain metastasis treated with stereotactic gamma knife radiosurgery. J Neurosurg 2002; 97(5 suppl):499–506. 44. Sheehan JP, Sun MH, Kondziolka D. Radiosurgery in patients with renal cell carcinoma metastasis to the brain. J Neurosurg 2003; 98(2):342–349. 45. Sheehan JP, Sun MH, Kondziolka D. Radiosurgery for non-small cell lung carcinoma metastatic to the brain: long-term outcomes and prognostic factors. J Neurosurg 2002; 97(6):1276–1281. 46. Sneed PK, Suh JH, Goetsch SJ. A multi-institutional review of radiosurgery alone vs. radiosurgery with whole brain radiotherapy as the initial management of brain metastases. Int J Radiat Oncol Biol Phys 2002; 53(3):519–526. 47. Pirzkall A, Debus J, Lohr F, Fuss M, Rhein B, Engenhart-Cabillic R, Wannenmacher M. Radiosurgery alone or in combination with whole-brain radiotherapy for brain metastases. J Clin Oncol 1998; 16(11):3563–3569. 48. Goyal LK, Suh JH, Reddy CA, Barnett GH. The role of whole brain radiotherapy and stereotactic radiosurgery on brain metastases from renal cell carcinoma. Int J Radiat Oncol Biol Phys 2000; 47(4):1007–1012. 49. Kondziolka D, Patel A, Lunsford LD, Kassam A, Flickinger JC. Stereotactic radiosurgery plus whole brain radiotherapy versus radiotherapy alone for patients with multiple brain metastases. Int J Radiat Oncol Biol Phys 1999; 45(2):427–434. 50. Auchter RM, Lamond JP, Alexander E, Buatti JM, Chappell R, Friedman WA, Kinsella TJ, Levin AB, Noyes WR, Schultz CJ et al. A multiinstitutional outcome and prognostic factor analysis of radiosurgery for resectable single brain metastasis. Int J Radiat Oncol Biol Phys 1996; 35(1):27–35. 51. O’Neil BP, Iturria NJ. A comparison of surgical resection and stereotactic radiosurgery in the treatment of solitary brain metastases. Int J Radiat Oncol Biol Phys 2003; 55(5):1169–1176.

CHAPTER 10: IMAGE-GUIDED TREATMENT OF METASTATIC BRAIN TUMORS

&

177

52. Rutigliano MJ, Lunsford LD, Kondziolka D, Strauss MJ, Khanna V, Green M. The cost effectiveness of stereotactic radiosurgery versus surgical resection in the treatment of solitary metastatic brain tumors. Neurosurgery 1995; 37(3):445–453. 53. Mehta M, Noyes W, Craig B, Lamond J, Auchter R, French M, Johnson M, Levin A, Badie B, Robbins I et al. A cost-effectiveness and cost-utility analysis of radiosurgery vs. resection for single-brain metastases. Int J Radiat Oncol Biol Phys 1997; 39(2):445–454. 54. Pollock BE, Brown PD, Foote RL. Properly selected patients with multiple brain metastases may benefit from aggressive treatment of their intracranial disease. J Neurooncol 2003; 61(1):73–80.

11

SURGICAL NAVIGATION SYSTEMS FOR THE RESECTION OF INTRACRANIAL GLIOMAS Michael W. McDermott, Devin K. Binder, Sandeep Kunwar, Andrew T. Parsa, and Mitchel S. Berger Department of Neurological Surgery, University of California, San Francisco, California, U.S.A.

Primary glial neoplasms, especially astrocytomas in adults, behave clinically in a malignant fashion due to their infiltrating nature, making them microscopically truly ‘‘unresectable.’’ In spite of this, local recurrence is still the major mode of failure for patients even after intensive local therapies. Most commonly, the recurrence is within 2 cm of the original site in 80% to 90% of cases (1–4). Recurrence distally within the central nervous system or systemically is uncommon. Local therapies, in which open operation for tumor resection plays a major role, have been shown to improve patient survival. Although there is some controversy on the role of debulking surgery for elderly patients with malignant gliomas, a number of retrospective and prospective randomized trials, conducted by the Radiation Therapy Oncology Group, have proven the benefit of surgical resection improving survival for patients with supratentorial malignant gliomas (5–8). Laws et al. (9) in an evaluation of the Glioma Outcomes Project data found that resection instead of biopsy was associated with prolonged survival in patients with grade III and IV gliomas. For low-grade gliomas as well, there is a greater body of literature indicating the survival benefit of more extensive surgical resection (10). In addition, Barker et al. (11) found that age and extent of resection were significantly independent predictors of radiation response in patients with glioblastoma. For recurrent gliomas, the volume of residual disease prior to chemotherapy also seems to have an influence on outcome. Keles et al. (12) found that patients in whom the volume of residual disease was less than 10 cm3 had a six-month progression free survival of 32% compared with 8% and 3% for those with volumes of 10 to 15 cm3 and greater than 15 cm3, respectively. For the pediatric population with malignant gliomas, previous studies have shown that surgical resection has been a very important factor in determining outcome (13,14). Regardless of the measure of resection as either the percentage of the original volume or as the volume of postoperative 179

180

&

PART III: CLINICAL APPLICATIONS

residual disease, it is frequently the case that many patients with gliomas live ‘‘better, if not longer’’ after debulking surgery. Given that surgical resection is an important component in the management of patients with intracranial gliomas, any technique that can improve our orientation within brain tissue at the time of surgery, and assist us with determining the margins of solid tumor tissue with surrounding infiltrated brain, should improve the safety and efficacy of these procedures. In the past, we have relied on a conceptualized three-dimensional volume of an intra-axial tumor based on our knowledge of neuroanatomy, imaging studies, intraoperative two-dimensional imaging information from ultrasound, and visual and tactile feedback to orient ourselves with respect to margins of solid tumor tissue and surrounding brain. During a surgical procedure, with a confined operative and visual field, once surface sulcal and gyral landmarks have been excluded and we proceed using the operating room microscope in a subcortical dissection, orientation for the surgeon becomes a problem. Surgical navigation systems (SNSs) help maintain surgical orientation and provide us additional information from preoperative imaging sets. SNSs are not designed to replace standard surgical techniques or surgeon judgment in deciding on the completeness or safety of surgical resection. There exist problems with these systems in that the preoperative registration accuracy is affected by intraoperative brain shift, distortion, and removal of cerebrospinal fluid. However, SNSs constitute a significant technical advance as they enable more efficient, safe, and complete surgical resection compared to technologies available in the past. At our institution, SNSs have become standard equipment in the surgical resection of primary intracranial gliomas.

PREOPERATIVE PREPARATION We have used skin-based fiducials as markers to assist with the registration of physical to imaging space. Because intracranial anatomic detail is better defined with magnetic resonance (MR) imaging, we have used this as the imaging study of choice for intra-axial tumors. The standard approach is to use a volume acquisition with 124 slices, 1.5 mm thick in the axial plane, no skip, field of view 26 using an SPGR sequence with intravenous contrast. Radiology technicians are careful to avoid compression of the tragus, straps over the forehead, or sponges in the headrest that would distort the skin surface, which is used for part of the surface merge process when the fiducial-based registration is less than optimal. For discrete nonenhancing presumed low-grade intra-axial neoplasms, standard T1 imaging may not define the boundaries well. If we are interested in a T2 volume for surgery, we use a fast spin echo image without fat suppression, which allows the skin surface to be seen for reconstruction and registration and allows the margins of the nonenhancing lesion to be seen more clearly. In the future, revisions to the image sequence coding and software for the system should allow a variety of image sequences to be presented for interactive display during surgery. Already, superimposition of somatosensory dipoles is routine and now MR tractography can be used to superimpose the descending corticospinal pathways

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

181

(15,16). Combined with intraoperative physiologic recordings at both the cortical and subcortical levels, surgical navigation systems have increased the speed, precision, and safety of the resection of intra-axial lesions (17). Because most systems rely on skin-based fiducials for registration of imaging to physical space, a potential source of error is scalp movement due to patient positioning or to the effects of pin fixation. We now use a routine set of 10 skin fiducials for every case (Table 1). Fiducials are placed over the junction over the frontozygomatic process with the body of zygoma bilaterally, the midline supraorbital margin; two on the vertex; one on each mastoid process and one or two distributed over the parietal eminence on each side, anterior to where the skin will be compressed by the pad in the MR head holder. Anatomic landmarks are used to check the localization accuracy of the registration set and the operator not relying on the root-mean square error number displayed by most systems, which is a mathematical estimation of the minimization function performed to map the physical surface to that derived from the three-dimensional image set reconstruction. Even with these fiducial placements, difficulties with image registration can be ameliorated by using a combination of skin-based surface fiducials and random surface registration points. In some cases, the process of acquiring the skin surface points has been automated using a laser-scanning device over anterior face and skull features. In the operating room, we use pin fixation for all patients, including those who require local anesthesia for intraoperative speech mapping. Ultra-short-acting intravenous anesthetic agents in combination with local anesthetic field blocks can allow one to apply the Mayfield pin apparatus in a patient who will be later recovered for speech/language assessment. We have not had problems with this technique. Once the registration process has been completed, and the accuracy of the registration confirmed, the surgeon can use the SNS in manual mode to review the anatomy of the tumor in three planes and ‘‘scroll’’ through the tumor volume. Having confirmed the operative approach and plan, the SNS is used in active mode to localize the tumor under the scalp and skull, and help plan skin incisions. Once the scalp flap has been turned back, the probe can be brought back into the field to confirm the margins for the craniotomy/craniectomy. After the dura has been turned back if swollen gyri are not present to indicate the tumor’s presence, the probe can be used and viewed in a number of orientations. We find it most Table 1 Standard Fiducial Sets for Registration Fiducial location Superior orbital margin Glabella Bregma Lambda Mastoid tips bilaterally Parietal eminence bilaterally Total

No. 2 1 1 1 2 3 10

182

&

PART III: CLINICAL APPLICATIONS

accurate to use the trajectory views, which are images, reconstructed in the plane of the probe, not the plane defined by the preoperative imaging, such as axial or coronal.

CONSIDERATIONS FOR THE USE OF SURGICAL NAVIGATION SYSTEMS BASED ON ANATOMIC SITE Supratentorial Frontal Lobe Tumors For intra-axial tumors of the frontal lobe, use of SNS is straightforward. Fiducials that are placed on the opposite side of the head, behind the ear, may not be accessible with head rotation. When a frontal lobectomy is planned, one of the issues that the surgeon should be aware of is the inaccuracy of the registration of position for brain structures, as brain shift occurs with drainage of CSF or coagulation and cutting of parasagittal draining veins, both of which allow the frontal lobe to move posteriorly and inferiorly. In an attempt to allow the surgeon to assess this degree of shift, in the past we placed small black dot tickets on the surface of the brain, similar to the numbers or letters used during functional mapping, at the start of the procedure. A second registration was required at the beginning of the procedure if the surgeon planned to use these ‘‘brain landmarks.’’ Using this second registration set, these black dot tickets on the surface of the brain were used as landmarks and the surgeon could then check the minimum tip-to-landmark distance to get some assessment of the degree of shift at the brain surface. Currently, an ultrasound probe has been coupled to imaging space by using a four light emitting diode tracking device. The SNS can display the MR images in the plane of the ultrasound probe similar to the plane of imaging seen on the two-dimensional ultrasound image. Relatively immobile structures such as the falx and choroid plexus can be marked on either image set to assess the coregistration accuracy between the MR and ultrasound images. Then the preresection tumor margin can be compared with the preoperative imaging to determine any shift. The postoperative resection brain shift can be measured by comparing the preresection position of known structures with the same point post resection. Keles et al. (18) found that the brain shift after resection of glioma varied from 0 to 6.4 mm and the direction of the shift was always towards the resection cavity, except for one case. They also found that the degree of postresection shift was greater in those who received mannitol. Tumor volume was not related to shift. Tumor position was with a greater amount of shift seen for tumors within 3 cm of the cortical surface. Temporal Lobe Tumors Tumors of the anterior temporal lobe are approached in similar fashion to those of the frontal pole. For mid-temporal and post-temporal tumors where the head will be rotated beyond 60 from the upright position, frequently fiducials on the

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

183

opposite side of the head towards the floor cannot be accessed for registration. Even if only five of the fiducials are used from the anterior, midline, and the same side of the scalp, the registration can still be accurate enough (less than 2.5 mm) to proceed. Again the registration set can be supplemented with 40 skin surface points. Artifact may be created by using fiducials on the dependent portions of the skin surface that sag away from the bone with extremes of positioning. Lesions that are in the medial portion of the temporal lobe are also subject to lack of anatomic correlation with registration when retraction is required for elevation of the temporal lobe to gain access to the medial structures. The surgeon can reorient him/herself relative to the dura of the middle cranial fossa, the lateral wall of the cavernous sinus, and the free edge of the tentorium in order to attempt to estimate the degree of shift of temporal lobe structures caused by retraction. Tumors in a posterior medial temporal lobe location can be exposed also via a supratentorial medial suboccipital route minimizing temporal lobe elevation for access (19). Parietal Lobe Tumors For tumors in the parietal lobe, the main concern is whether the operation will be done awake or asleep. Awake craniotomy allows for low-voltage stimulation of the cortex to identify the sensory/motor cortex. For those patients who will be asleep, superimposition of somatosensory dipoles from magnetic source imaging can be done. The surgeon should consider the effect that patient positioning will have on brain shift and inaccurate anatomic localization after the dura is opened and further as debulking proceeds. The slouching position with the neck flexed on the chest and the head flexed on the neck allows for a greater contribution of the effect of gravity and brain shift than would a lateral position. Issues of brain shift can be resolved with the addition of ultrasound as above (20). Occipital Lobe Tumors For tumors in this location, it is predominantly the posterior fiducial set that will be used for registration. Those on the forehead are usually difficult to access. The operator should be warned not to supplement the fiducial registration with skin surface points on the back of the head as this site is always distorted by supine positioning in a head holder for imaging. Other Brain Locations SNSs are particularly helpful using a trajectory view for accessing tumors within the ventricular cavity. This trajectory view can be used to guide the surgeon transcortically to the ventricle, and also allow for determination of where along the corpus callosum to begin an incision. While the intraventricular landmarks assist with orientation during the procedure, for large tumors, surgical navigation devices

184

&

PART III: CLINICAL APPLICATIONS

are particularly helpful in determining the posterior margin of such lesions. All too often, we have operated assuming a greater extent of surgical resection for intraventricular lesions than was actually the case, based on postoperative imaging studies. This seems to be less of a problem since using this device. SNSs have also been particularly helpful to pediatric surgeons for the surgical debulking of optic chiasm hypothalamic gliomas. Significant resections have been accomplished safely, and there is anecdotal evidence to suggest that for these tumors the extent of resection may, as well, influence outcome, although this remains yet to be analyzed. Orientation with respect to the lateral, posterior, and superior margins of these lesions is, of course, crucial, and is assisted by the SNS. Focal low-grade contrast-enhancing neoplasms within the brainstem are now being approached surgically by our pediatric colleagues and there is again some evidence that this may improve tumor control. Lesions of the midbrain, diencephalon, pons, and medulla can be approached accurately, and selective incisions made in the pial surface. Because of their brainstem location, errors in registration related to brain shift are minimized. The closer one is to the central axis of the brain. Surgeons should be mindful of the focal length objective of the microscope and the length of the probe to be used. The shorter the focal length for the microscope lens, the more difficulty the surgeon will have with placing the probe deep into the wound towards the brainstem at the time of surgery. Lenses of 350 to 400 mm are most appropriate, and the long probe should be used for deep trans-Sylvian approaches or subtemporal transtentorial approaches to the brainstem. Current SNS systems also allow for the focal point within the microscope field to act as the virtual probe and LEDs are mounted on the microscopes so that they may be tracked as the pointing device (21).

Infratentorial Cerebellar Tumors The posterior fiducial set is most often used for accurate registration for these patients, whether the patient is positioned prone or in the concorde position. These fiducials include the mastoid (2), parietal eminence (2,3), lambda (1), and bregma (1). Tumors within the cerebellar hemisphere or vermis can be more accurately localized, and again, the deeper the location of lesion in the hemisphere, the less important are issues related to brain shift. However, supracerebellar infratentorial approaches to anterior vermian lesions require consideration on the surgeon’s part of the effects of inferior displacement of the cerebellum during tumor exposure.

Fourth Ventricular Tumors Exophytic brain stem tumors or tumors filling the fourth ventricle are ideally imaged in the sagittal plane for maintaining the surgeon’s orientation with respect

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

185

to the floor of the fourth ventricle. When the rostral or caudal portion of the normal ventricular floor cannot be seen because of the mass of tumor, SNSs assist with maintaining orientation of position relative to the position of the floor of the fourth ventricle above and below the margin of tumor. In all cases, image guidance is supplemented by intraoperative neurophysiologic monitoring of somatosensory and lower cranial nerves. Tumors that do not arise from the floor of the fourth ventricle primarily, such as medulloblastomas, can also benefit from the use of the device by allowing the surgeon to reorient himself as the resection proceeds, and presumably increasing the efficiency and safety of tumor resection. Posterior Fossa Biopsies In the past, frame-based stereotactic biopsies have been used in selected cases prior to instituting therapy via a transcerebellar route through the middle cerebellar peduncle. After selecting a target point and entry point to define the surgical plan, a small burr hole can be placed on the occipital bone below the transverse sinus and then either a skull mounted or external fixator biopsy guide can be used to bring the guide to the correct trajectory. The biopsy needle itself can be tracked, as it is delivered to the target position. Snap shots of the computer screen with the needle at the target position are excellent documentation for later reference should the biopsy results be unclear. Lesions down to the junction of the dorsal pons and middle cerebellar peduncle have been approached in this fashion.

UNIVERSITY OF CALIFORNIA AT SAN FRANCISCO EXPERIENCE During a prospective clinical trial of an SNS at University of California at San Francisco (UCSF) between 1993 and 1994, in 61% of the cases the device was used for operation on an intra-axial primary brain neoplasm. In this group of 51, 66% of the patients had high-grade gliomas, and 25% were being operated for recurrent disease. The average time for registration was 13 minutes during this study, and the additional time for setup of the equipment in the operating room was approximately 18 minutes. The majority of the surgeons (85%) felt that the device was easy to use, and most (76%) felt that the device was helpful in assisting them with tumor resection. There were no cases of a break in the sterile field or an adverse patient event in any of these cases. Our experience with the device has continued since that time and is now over 300 cases with a number of different commercial SNS products. The SNS is now considered standard equipment, at least for supratentorial neoplasms, for first-time operations and reoperations, on intra-axial neoplasms. Because of time constraints and use of the device twice in 1 day, patients to have surgery starting at 0800 hours are scanned the afternoon before surgery, and patients scheduled for later in the day have their imaging scans performed on the morning of surgery. Operating room technicians and the surgeon work together to perform the registration process, and fiducial locations for registration can be selected ahead of time on the computer,

186

&

PART III: CLINICAL APPLICATIONS

which speeds up the process. Intraoperatively, particularly when working under the microscope, the surgeon requires the assistance of operating room personnel to freeze the image position so that he or she can safely turn away from the microscope to look at the computer screen. Newer systems allow the surgeon to view the outline of the tumor volume as a line superimposed on the microscopic field.

A PROSPECTIVE STUDY ON SNS AND EOR Since SNS are used to improve extent of resection (EOR) of intra-axial brain neoplasms, we decided to assess the efficacy of SNS at improving EOR. A prospective study was undertaken involving patients undergoing resection of gliomas. Surgeons were enlisted to mark out the estimated EOR postoperatively on navigational scans. This estimated EOR was compared with pre- and postoperative tumor volumes to allow comparison of surgeons’ estimated EOR with actual EOR. The primary goal of this investigation was to assess the efficacy of SNS at improving actual EOR.

Patient Enrollment Subjects (n ¼ 28) undergoing surgical treatment of primary CNS gliomas by one of three UCSF neurosurgeons were prospectively enrolled in the study. Patient records and radiologic studies were identified in the Departments of Neurological Surgery/ Division of Neuro-oncology and Department of Radiology at UCSF. Subsequent to enrollment, one subject was excluded for alternative diagnosis (schwannoma) and two subjects were excluded because of no postoperative imaging study.

Imaging Studies All patients received preoperative navigational MRI imaging studies (Stealth protocol). Surgeons marked out the estimated EOR immediately postoperatively by drawing on individual navigational scans. Following surgery, each patient underwent postoperative diagnostic MRI imaging. Three volumes were obtained for each patient: (i) preoperative tumor volume; (ii) surgeons’ estimated resection volume; and (iii) postoperative tumor residual volume.

Image Analysis Preoperative and postoperative tumor volumes were taken to be entire T2 volume for nonenhancing tumors and contrast-enhancing T1 volume for enhancing tumors. Area measurements were digitized on a GE workstation with digitizing

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

187

software and then integrated over slice thickness to obtain tumor volumes. For surgeons’ estimated resection volume, the extent of tumor estimated to have been resected intraoperatively was marked, slice by slice, on navigational MRI image sets. Marked areas were digitized using a light table and digitizing tablet, and then scaled to the areas measured pre- and postoperatively on the workstation with a calibrated scaling ratio to ensure consistency between area measurements derived from hard copy films and from workstation images. Preoperative, postoperative, and surgeons’ resection volumes were measured by three separate investigators blinded to each other’s measurements. In cases in which residual tumor was difficult to distinguish from postoperative injury, hemorrhage, edema, or infarction, other studies (such as T1-weighted precontrast or diffusion imaging) were consulted.

Estimated vs. Actual EOR Estimated EORs were calculated as the ratio of surgeons’ predicted resection volume to total preoperative volume 100%. There were a few cases in which the surgeon’s predicted resection volume was slightly higher than total preoperative volume because of nonenhancing tissue marked as part of the resection. Actual EORs were calculated as preoperative volume postoperative residual volume/preoperative volume 100%. Difference between estimated and actual EOR was simply calculated as the difference in percentages. A greater than 10% difference between estimated and actual EOR was judged to be potentially clinically significant.

Patient Characteristics Twenty-eight patients were enrolled prospectively in the study. Three patients were excluded (one due to alternative diagnosis—schwannoma, and two due to lack of postoperative imaging). The following information applies to the remaining 25 patients. Mean patient age was 47 (range 28–74). Diagnoses included both highgrade gliomas (grade III and IV) (n ¼ 21) and low-grade gliomas (grade II) (n ¼ 4). These included glioblastoma (n ¼ 15, nine with first resection and six recurrent), anaplastic astrocytoma (n ¼ 1), oligodendroglioma (n ¼ 4, one grade II and three grade III), oligoastrocytoma (n ¼ 3, one grade II and two grade III), and ganglioglioma (n ¼ 2) (Table 2). Tumor locations included frontal (n ¼ 12), temporal (n ¼ 7), and parietal (n ¼ 6) lobes. Three patients also received radioactive seeds as part of the resection procedure.

Tumor Volumes Preoperative tumor volumes were digitized on 24 patients (Fig. 1). Mean preoperative tumor volume was 50.6 cm3 (median 40.7, range 4.1–161.1). Postoperative

188

&

PART III: CLINICAL APPLICATIONS

Table 2 Pathologic Diagnoses for Intra-axial Neoplasms Evaluated in EOR Study Newly diagnosed Glioblastoma multiforme 9 Anaplastic astrocytoma 1 Oligodendroglioma 4 Mixed oligoastrocytoma 3 Ganglioglioma 2 Recurrent Recurrent GM 6

tumor residual volumes were digitized in 25 patients (Fig. 2). Mean postoperative residual volume was 2.0 cm3 (median 0.48, range 0–26.4). Of these 25 patients, 16 had 10 cm3 or less residual volume (gross-total or near-gross-total resection) (64%). By this criterion, gross-total or near-gross-total resection was achieved in 71% of high-grade gliomas (15/21) and 25% of low-grade gliomas (1/4). Surgeons’ estimated resection volume was digitized on 17 patients (Fig. 3). Mean estimated resection volume was 35.0 cm3 (median 35.0, range 5.0–81.9).

Estimated and Actual EORs Estimated and actual EORs were calculated for each tumor. Estimated EOR was defined as (surgeons’ estimated resection volume/preoperative volume) 100%. In cases in which surgeon’s estimated resection volume was larger than preoperative volume (see ‘‘Methods’’), estimated EOR was set at 100%. Estimated EORs were available on 16 patients. Mean estimated EOR was 83.6% (median 88.5%, range 38–100%).

Figure 1 Bar histogram showing distribution of tumor volumes in EOR study.

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

189

Figure 2 Postoperative residual tumor volume. For nonenhancing tumors entire T2 volume was used for pre- and postoperative comparison. For high-grade tumors, only contrastenhancing volume was measured.

Actual EOR was defined as (preoperative volume postoperative residual volume/preoperative volume) 100%. Actual EORs were available on 24 patients. Mean actual EOR was 96.7% (median 98.6, range 80.8–100%). Differences between estimated and actual EOR were expressed as percentages. Mean difference was 12.0% in favor of underestimation of actual EOR. In the five cases in which surgeons’ estimated EOR was an overestimate of actual EOR, the mean overestimate of tumor resection was 6.3% (median 3.3, range 1.1–18.9). The individual percentages in these five cases were 1.1%, 1.8%, 3.3%,

Figure 3 Distribution of surgeons EOR. Mean surgeon-estimated EOR was 83.6% and actual was 96.7%.

190

&

PART III: CLINICAL APPLICATIONS

6.5%, and 18.9%. It is important to note that the two largest discrepancies (6.5% and 18.9%) were both low-grade gliomas (ganglioglioma and grade II oligodendroglioma, respectively). Thus, at least for high-grade gliomas, differences between actual EOR and estimated EOR appear to be clinically negligible.

Study Conclusions In cases in which surgeons’ EOR overestimated actual EOR, the difference was 6.3% by volume, and this discrepancy was contributed disproportionately by low-grade gliomas. The estimated and actual EOR in experienced hands using navigational devices appear to be acceptable, at least for high-grade gliomas. In contrast to the findings in our study, Knauth et al. (22) evaluated the use of intraoperative MR as an adjunct to the use of SNSs for the resection of high-grade gliomas. Surgeons in this study ceased operating when they felt that all the enhancing tumors had been removed. Using a 0.2 T iMR machine, imaging was then performed. In 53.7% residual enhancing tissue was seen and in 77% of these cases further surgical resection was carried out. Early postoperative imaging on a 1.5 T MR showed residual tumor in 19.5% of patients. The authors concluded that the results of SNS-based surgical resection could be improved significantly by the use of intraoperative imaging. Further results on the use of this modality and its impact on patient outcome are pending.

TECHNICAL CONSIDERATIONS The choice of the operative approach to a primary intra-axial neoplasm may influence how accurately preoperative imaging reflects intraoperative localization during the resection. Generally, there are two approaches for removing malignant gliomas in the supratentorial compartment in a noneloquent location. The first is by limited corticectomy and internal debulking; the second is a larger area of corticectomy immediately superficial to the area of underlying brain abnormality. In the first approach, considerable retraction along the corticectomy is required in order to expose peripheral margins of the tumor, and this retraction may add to the displacement of brain tissue, thus affecting the accuracy of the intraoperative location of the probe tip. With a more generous superficial corticectomy, such retraction is usually not required, and one would suppose that artifacts from retraction in this situation would be minimized. We have avoided the ‘‘fence-posting’’ technique employed by surgeons using frame-based systems, where catheters are placed stereotactically at the margins of the tumor prior to any resection and then the operation carried out following these catheters. Instead, once we have outlined the area for corticectomy, we proceed initially with a resection in a narrow trough at the margins of lesion down to the inferior extent of the tumor, as visualized through the microscope and as represented on the SNS. Once troughs around the margin of the tumor have been created, then presumably the margins have

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

191

been more accurately defined, as opposed to proceeding first to internal debulking, where brain shift may interfere with accurate localization based on preoperative images. We have also noted that there is a significant difference between first and second operations for a patient with a glioma, relating to the effect of prior therapies on brain elasticity. For initial operations, the brain is relatively more elastic, and tumor removal allows normal tissue to move into the space created by the resection cavity. This movement may make subsequent localization inaccurate, and visual and tactile clues to the margins of solid tumor tissue with surrounding more normal edematous or infiltrated brain thus become more important. With recurrent tumors, the brain is less elastic, and does not rebound back into the resection cavity after removal of tissue. On the other hand, with recurrent tumors, visual and tactile clues to the junction between solid tumor tissue and surrounding brain are more poorly defined. In spite of these limitations, we have found SNS to be particularly helpful with achieving a more complete resection in these recurrent cases. The surgeon should also be mindful of the fact that SNSs are adjuncts to standard surgical techniques, and do not replace them. Anatomic localization does not guarantee functional localization, and in areas of eloquent tissue, intraoperative electrophysiologic monitoring and mapping is essential in order to reduce the likelihood of postoperative deficits (17). The combination of the use of surgical navigation systems and electrophysiologic monitoring has allowed for more complete resection of lesions previously thought to be unapproachable or amenable only to ‘‘subtotal’’ removal of solid tumor tissue. In summary, our experience with one SNS has been gratifying. Surgeons have found the device easy to use, reliable, and informative. At our institution, it has become standard equipment for resection of supratentorial intra-axial neoplasms, and will likely become a specialty standard within the next 5 years. Future applications of the technology include combining functional MR or magnetic source imaging data with the SNS, possibly reducing the need for intraoperative functional localization, at least at the cortical surface. Furthermore, since intraoperative volumetric localization is possible at multiple sites within a tumor, there are basic science research questions comparing pathology, proliferative indices, special stains, molecular biologic analyses, with the results of diffusion-weighted imaging, cerebral blood volume mapping MR spectroscopy.

192

&

PART III: CLINICAL APPLICATIONS

REFERENCES 1. Binder DK, Keles GE, Berger MS. Aggressive glial neoplasms. In: Batjer HH, Loftus CM, eds. Textbook of Neurological Surgery. Philadelphia, Pennsylvania: Lippincott Williams & Wilkins, 2003:1270–1280. 2. Choucair AK, Levin VA, Gutin PH, Davis RL, Silver P, Edwards MS, Wilson CB. Development of multiple lesions during radiation therapy and chemotherapy in patients with gliomas. J Neurosurg 1986; 65:654–658. 3. Liang BC, Thornton AF Jr, Sandler HM, Greenberg HS. Malignant astrocytomas: focal tumor recurrence after focal external beam radiation therapy. J Neurosurg 1991; 75:559–563. 4. Sneed PK, Gutin PH, Larson DA, Malec MK, Phillips TL, Prados MD, Scharfen CO, Weaver KA, Wara WM. Patterns of recurrence of glioblastoma multiforme after external irradiation followed by implant boost. Int J Radiat Oncol Biol Phys 1994; 29:719–727. 5. Curran WJ Jr, Scott CB, Horton J, Nelson JS, Weinstein AS, Fischbach AJ, Chang CH, Rotman M, Asbell SO, Krisch RE, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst 1993; 85:704–710. 6. Devaux BC, O’Fallon JR, Kelly PJ. Resection, biopsy, and survival in malignant glial neoplasms. A retrospective study of clinical parameters, therapy, and outcome. J Neurosurg 1993; 78:767–775. 7. Rostomily RC, Spence AM, Duong D, McCormick K, Bland M, Berger MS. Multimodality management of recurrent adult malignant gliomas: results of a phase II multiagent chemotherapy study and analysis of cytoreductive surgery. Neurosurgery 1994; 35:378–388. 8. Simpson JR, Horton J, Scott C, Curran WJ, Rubin P, Fischbach J, Isaacson S, Rotman M, Asbell SO, Nelson JS, et al. Influence of location and extent of surgical resection on survival of patients with glioblastoma multiforme: results of three consecutive Radiation Therapy Oncology Group (RTOG) clinical trials. Int J Radiat Oncol Biol Phys 1993; 26:239–244. 9. Laws ER, Parney IF, Huang W, Anderson F, Morris AM, Asher A, Lillehei KO, Bernstein M, Brem H, Sloan A, Berger MS, Chang S. Survival following surgery and prognostic factors for recently diagnosed malignant glioma: data from the Glioma Outcomes Project. J Neurosurg 2003; 99:467–473. 10. Keles GE, Lamborn KR, Berger MS. Low-grade hemispheric gliomas in adults: a critical review of extent of resection as a factor influencing outcome. J Neurosurg 2001; 95:735–745. 11. Barker FG II, Chang SM, Larson DA, Sneed PK, Wara WM, Wilson CB, Prados MD. Age and radiation response in glioblastoma multiforme. Neurosurgery 2001; 49:1288–1297; discussion 1297–1288. 12. Keles GE, Lamborn KR, Chang SM, Prados MD, Berger MS. Volume of residual disease as a predictor of outcome in adult patients with recurrent supratentorial glioblastomas multiforme who are undergoing chemotherapy. J Neurosurg 2004; 100:41–46.

CHAPTER 11: SNS FOR RESECTION OF INTRACRANIAL GLIOMAS

&

193

13. Duffner PK, Horowitz ME, Krischer JP, Burger PC, Cohen ME, Sanford RA, Friedman HS, Kun LE. The treatment of malignant brain tumors in infants and very young children: an update of the Pediatric Oncology Group experience. Neurooncology 1999; 1:152–161. 14. Duffner PK, Horowitz ME, Krischer JP, Friedman HS, Burger PC, Cohen ME, Sanford RA, Mulhern RK, James HE, Freeman CR, et al. Postoperative chemotherapy and delayed radiation in children less than three years of age with malignant brain tumors. N Engl J Med 1993; 328:1725–1731. 15. Berman JI, Berger MS, Mukherjee P, Henry RG. Diffusion-tensor imaging-guided tracking of fibers of the pyramidal tract combined with intraoperative cortical stimulation mapping in patients with gliomas. J Neurosurg 2004; 101:66–72. 16. Roberts TP, Zusman E, McDermott M, Barbaro N, Rowley HA. Correlation of functional magnetic source imaging with intraoperative cortical stimulation in neurosurgical patients. J Image Guid Surg 1995; 1:339–347. 17. Henry RG, Berman JI, Nagarajan SS, Mukherjee P, Berger MS. Subcortical pathways serving cortical language sites: initial experience with diffusion tensor imaging fiber tracking combined with intraoperative language mapping. Neuroimage 2004; 21: 616–622. 18. Keles GE, Lamborn KR, Berger MS. Coregistration accuracy and detection of brain shift using intraoperative sononavigation during resection of hemispheric tumors. Neurosurgery 2003; 53:556–562; discussion 562–564. 19. Russell SM, Kelly PJ. Volumetric stereotaxy and the supratentorial occipitosubtemporal approach in the resection of posterior hippocampus and parahippocampal gyrus lesions. Neurosurgery 2002; 50:978–988. 20. Keles GE, Lamborn KR, Berger MS. Coregistration accuracy and detection of brain shift using intraoperative sononavigation during resection of hemispheric tumors. Neurosurgery 2003; 53:556–562. 21. Roessler K, Ungersboeck K, Czech T, Aichholzer M, Dietrich W, Goerzer H, Matula C, Koos WT. Contour-guided brain tumor surgery using a stereotactic navigating microscope. Stereotact Funct Neurosurg 1997; 68:33–38. 22. Knauth M, Wirtz CR, Tronnier VM, Aras N, Kunze S, Sartor K. Intraoperative MR imaging increases the extent of tumor resection in patients with high-grade gliomas. Am J Neuroradiol 1999; 20:1642–1646.

12

INTRACRANIAL MENINGIOMAS Joung H. Lee Cleveland Clinic Brain Tumor Institute and Department of Neurosurgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Ajit A. Krishnaney and Michael P. Steinmetz Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Dae Kyu Lee Brain Tumor Institute and Department of Neurological Surgery, Taussig Cancer Center, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION Surgical management of patients with meningiomas is arguably the most rewarding, challenging, and at times, daunting task for a neurosurgeon. It is rewarding because of the benign nature of most meningiomas, leading to a possibility of providing a cure following total removal, in addition to reversal or improvement of preoperative symptoms. It is definitely challenging due to the tumor’s common compression, adherence, or displacement of adjacent neurovascular structures, especially for those arising along the skull base (SB), making surgery difficult and highly risky (1–3). Simultaneously, it is daunting because of the associated risks of surgery, the tumor’s tendency to recur following incomplete (and at times complete) removal, and its frequent involvement of the dural sinuses, surrounding skull base bone, dura, and neurovasculature, making complete removal often not possible (4). Real-time computer-assisted surgical navigation systems (SNSs) may mitigate some of the problems and challenges inherent in meningioma surgery, which would ultimately improve patient outcome (5–7). Many of meningioma’s characteristic features, such as its common skull base origin, its extra-axial location, and its intimate relationship with surrounding neurovascular structures, and involvement of the adjacent bone/dura, make meningioma surgery an ideal venue for application of SNS. Main advantages of SNS application in meningioma surgery would include: (i) accurate tumor localization; (ii) optimal positioning, placement of incision, and craniotomy/tumor exposure; (iii) assessment of the extent of bone removal during exposure of SB meningiomas; (iv) complete delineation of tumor and the involved dura and bone, including the limit of ‘‘dural tail’’; (v) localization of skull base landmarks; (vi) localization of adjacent neurovascular structures; and (vii) estimation of tumor removal. Principles, technical considerations, and application of computer-assisted SNS in cranial surgery, in general, are discussed in detail in other chapters in this book, and these will not be repeated here. Rather, only the special considerations, 195

196

&

PART III: CLINICAL APPLICATIONS

indications, and nuances unique to the application of SNS in meningioma surgery will be delineated in this chapter. In particular, the authors’ insights and opinions on when and how to best use SNS in meningioma surgery are presented. SNS APPLICATION IN MENINGIOMA SURGERY Meningiomas require surgical approaches that are primarily dictated by their locations. Additionally, the tumor size, the patient’s body habitus and anatomic variations, and the surgeon’s personal experience and preference may further contribute to selecting a particular approach. However, the following basic principles and surgical steps hold for meningioma surgery of most locations: (i) optimal patient positioning and incision; (ii) craniotomy/tumor exposure; (iii) early tumor devascularization; (iv) internal decompression/extracapsular dissection; (v) early localization and preservation of adherent or adjacent neurovasculature; (vi) removal of bone and dura involved by the tumor; and (vii) closure (8). The authors find SNS quite helpful in all steps, except during closure. Sequential SNS application in meningioma surgery is outlined. Positioning and Incision Patient positioning, appropriate incision placement, and selection of optimal approach for tumor exposure are the critical elements of successful meningioma surgery. The patient is positioned in a way that maximizes patient safety. Moreover, the ideal position must allow for an approach that provides complete exposure of the tumor and the involved surrounding bone and dura. At the same time, maximal brain relaxation should be achieved by use of gravity and uncompromised venous drainage to minimize the need for retraction. The head should be not be lower than the level of the heart regardless of the position selected, and excessive neck rotation or flexion must be avoided, as this may obstruct jugular venous outflow. SNS accurately localizes the tumor and delineates its full extent prior to the start of surgery, in addition to providing a preview of the trajectory to the tumor. This information is helpful in optimizing patient positioning, especially in terms of the amount of head rotation, elevation, and flexion needed to achieve the best trajectory, while maximizing patient safety. The exact size and location of scalp incision are more critical for superficial than deep-seated meningiomas. For superficial tumors (e.g., convexity or parasagittal), the planned scalp flap or incision should contain the tumor in the center. Notably, the incision must be planned to avoid any visible cosmetic defect or significant compromise to the scalp vascular supply. If a horseshoe-shaped incision is planned, the depth must not exceed the width of the flap. Again, for superficial tumors, the size of the scalp and bone flaps must be large enough to allow for maximal exposure of the tumor, the involved bone and dura, as well as the limits of the dural tail, as noted on preoperative MRI scans (Fig. 1). SNS is especially helpful in minimizing the incision and craniotomy for small superficial meningiomas or meningiomas of all sizes with significant dural tail.

CHAPTER 12: INTRACRANIAL MENINGIOMAS

&

197

Figure 1 SNS allows full delineation of the tumor as well as its dural tail so as to include this involved dura in the craniotomy (upper panel). Complete removal may then be performed (lower panel).

Craniotomy/Tumor Exposure In addition to minimizing the size of bone opening, SNS helps avoid potential complications during craniotomy and subsequent skull base bone removal. For anterior frontal convexity or parasagittal meningiomas, SNS can delineate the

198

&

PART III: CLINICAL APPLICATIONS

Figure 2 SNS may be used during the removal of large clinoidal meningiomas (upper panel). SNS is helpful to delineate the ethmoid and sphenoid sinuses during drilling of the anterior clinoid process and optic canal roof to prevent inadvertent entry. SNS is also helpful in localization of optic canal and the ICA. Aggressive tumor resection is then possible without associated morbidity (lower panel).

margins of the frontal sinus in order to avoid sinus entry. Likewise, during skull base drilling to remove the optic canal roof and the anterior clinoid process for surgery of large clinoidal or tuberculum sella meningiomas, the ethmoid and sphenoid sinus entry can be prevented by the use of SNS (Fig. 2). When presigmoid suboccipital craniectomy is planned for cerebellopontine angle or petrous

CHAPTER 12: INTRACRANIAL MENINGIOMAS

&

199

meningiomas, SNS localization of the transverse and sigmoid sinuses, as well as mastoid air cells, may facilitate suboccipital craniectomy and reduce complications. In most cases, an optimal approach should provide the shortest and direct route to the tumor without ‘‘sacrificing’’ any normal brain tissue or undo brain damage created by retraction. The need for retraction is minimized by, when appropriate, judicious use of gravity, cerebrospinal fluid (CSF) drainage, and skull base bone removal. For example, for surgery of an olfactory groove meningioma, the head can be slightly hyperextended. For large, deep falcine tumors, the patient’s head may be placed with the side of tumor down and the direction of the sagittal sinus parallel to the operating room floor. By previewing the SNS trajectory to the tumor, optimal head positioning is selected. In all of these examples mentioned, the brain falls away from the tumor and its attachment. In deep-seated tumors, brain retraction may be minimized by use of CSF drainage, either via a ventriculostomy (in patients with large ventricles) or lumbar drain. Furthermore, many of the skull base approaches developed over the last two decades, which convert the deep basal meningiomas to more superficial ‘‘convexity’’ lesions by reducing the working distance to the tumor, may reduce the need for brain retraction. In performing transcondylar approach for foramen magnum meningiomas, for example, the extent of condyle removal, and therefore the adequacy of skull base exposure, can be assessed by the use of SNS (Fig. 3). Similarly, the extent of bone removal during transpetrosal approach for petroclival meningiomas can be assessed with SNS, again to provide adequate bone removal, which ensures optimal tumor exposure. Other helpful applications of SNS include reoperation for recurrent meningiomas (Fig. 4). With exact localization of the tumor and its margin, SNS helps the surgeon to differentiate the tumor from surrounding scar or gliosis from previous surgery. Additionally, for patients with multiple small meningiomas, with the help of SNS the surgeon can ensure complete removal of all tumors (Figs. 1 and 4).

Tumor Devascularization Many meningiomas can be quite vascular. In addition to utilization of preoperative embolization when appropriate, early intraoperative devascularization of the tumor reduces blood loss and facilitates the resection. In superficial tumors, upon dural exposure, attention should be first directed at coagulating and dividing all the dural feeding vessels, most commonly the branches or the main trunk of middle meningeal artery. For most meningiomas, SNS has no significant role in specifically devascularizing meningiomas. Yasargil (9) advocates initial transtumoral devascularization of basal meningiomas by working through a small ‘‘window’’ created in the tumor to reach the blood supply coming through the base. However, this technique may not be suitable for an inexperienced surgeon as there may be a significant risk of injury to unexposed neurovascular structures that may be located on the other side of the tumor, although this may be mitigated by judicious use of SNS.

200

&

PART III: CLINICAL APPLICATIONS

Figure 3 SNS may be used during the exposure of foramen magnum meningiomas (A), especially during a transcondylar approach. This ensures an adequate exposure. Moreover, it allows localization of the vertebral artery, which may be encased by tumor as seen in this patient, as the decompression progresses. Safe and aggressive resection can be achieved (B).

CHAPTER 12: INTRACRANIAL MENINGIOMAS

&

201

Figure 4 SNS significantly aids the surgeon in removing recurrent and/or multiple meningiomas (upper panel). Accurate tumor localization and delineation from the surrounding gliotic brain is facilitated by SNS to allow complete and safe tumor removal (lower panel).

202

&

PART III: CLINICAL APPLICATIONS

Internal Decompression and Extracapsular Dissection Although small meningiomas may be removed en bloc, internal decompression is often a key initial step in actual tumor removal for many meningiomas following adequate exposure and initial devascularization. Internal debulking is carried out until a thin rim of exposed portion of the tumor is remaining. This internal debulking minimizes brain retraction and facilitates extracapsular dissection. SNS is used to assess the extent of initial debulking and the thickness of the residual tumor capsule. This application may be particularly helpful in avoiding damage to any adherent neurovascular structures, which are often located ventral to the tumor in many skull base meningiomas. In superficial meningiomas, this application of SNS is somewhat limited by errors produced by brain and tumor shift from retraction, CSF drainage, and tumor removal. There errors are relatively insignificant in meningiomas attached to the skull base. However, after significant central tumor debulking, SNS may be no longer reliable for the portion of the tumor remaining on the brain side. In situations of planned subtotal resection, as in patients with large petroclival meningioma, SNS provides accurate real-time assessment of the residual tumor size attached to the site of origin along the skull base (Fig. 5). Localization and Preservation of Adjacent Neurovasculature In large meningiomas, which displace or encircle intradural neurovasculature, early localization of these structures, which are distorted from its normal anatomic position, can help prevent any critical intraoperative injury. For removal of a large olfactory groove, clinoidal, and tuberculum sella meningiomas displacing and obscuring the view of optic nerve(s), the optic canal, which is a fixed skull landmark unaffected by brain shift or retraction, can be easily localized with SNS as tumor removal progresses (Fig. 6). Once the optic nerve is identified, the displaced internal carotid artery, which is normally just lateral to the intradural optic nerve, can often be safely exposed. Similarly, in sizeable petroclival or petrous meningiomas covering and displacing the cranial nerve complex VII/VIII, SNS can be used to locate the internal acoustic canal (IAC) (Fig. 5). In meningiomas involving the middle fossa floor and invading into the temporal bone, critical structures localized with the assistance of SNS include the petrous segment of ICA, cochlea, air cells, semicircular canals, and IAC (Fig. 7). In large foramen magnum meningiomas completely encircling the vertebral artery, which has a relatively ‘‘fixed’’ point of dural entry into the posterior fossa, the artery can be localized as the tumor removal progresses (Fig. 3).

Removal of the Involved Bone and Dura Complete removal of the bone and dura involved by the tumor minimizes the risk of recurrence. In meningiomas with significant hyperostosis or dural involvement,

CHAPTER 12: INTRACRANIAL MENINGIOMAS

&

203

Figure 5 Planned subtotal removal may be performed in large skull base meningiomas such as those in the petroclival/cavernous sinus region (upper panel). Furthermore, SNS allows early identification of critical structures and landmarks, such as the IAC in this particular case, ensuring preservation of the CN VII/VIII complex. In those patients with planned subtotal removal of the tumor, SNS provides accurate real-time assessment of the residual tumor size attached to the site of origin along the skull base (lower panel). The residual tumor then can be treated with gamma knife radiosurgery at a later date when there is a radiographic or clinical progression.

204

&

PART III: CLINICAL APPLICATIONS

Figure 6 A large olfactory groove meningioma obscuring the bilateral optic nerves (A). As tumor removal progresses, SNS can provide the real-time information regarding the exact location of the optic canals in relation to where the surgeon is working. SNS, therefore, aids in complete resection while preventing any intraoperative injury to the optic nerves (B).

CHAPTER 12: INTRACRANIAL MENINGIOMAS

&

205

Figure 7 (A) This large middle fossa meningioma has invaded the middle fossa floor into the temporal bone. SNS was used to localize the IAC, cochlea, air cells, semicircular cannals, and the petrous segment of the carotid artery. (B) This permitted complete removal of this large tumor.

206

&

PART III: CLINICAL APPLICATIONS

SNS accurately delineates the extent of such involvement. This is especially helpful in orbitosphenoid meningiomas, which can often produce extensive hyperostosis or convexity meningiomas with a significant ‘‘dural tail’’ (Fig. 1). As previously mentioned, while drilling the involved bone, SNS helps prevent untoward entry into paranasal sinuses, temporal air cells, and the inner ear structures.

SUMMARY SNS application in meningioma surgery provides the following important advantages: (i) accurate tumor localization; (ii) optimal positioning, placement of incision, and craniotomy/tumor exposure; (iii) assessment of the extent of bone removal during exposure of SB meningiomas; (iv) complete delineation of tumor and the involved dura and bone, including the limit of ‘‘dural tail’’; (v) localization of skull base landmarks; (vi) localization of adjacent neurovascular structures; and (vii) estimation of tumor removal. However, SNS is certainly not necessary in removal of many meningiomas. Those particular tumors that would benefit from the SNS application include: (i) small superficial tumors; (ii) tumors with dural tail; (iii) tumors with significant hyperostosis; (iv) recurrent tumors with significant scarring; and (v) large skull base tumors displacing or encircling critical neurovasculature. Despite its helpfulness in meningioma surgery, SNS should be utilized to complement rather than to replace the surgeon’s knowledge of surgical anatomy required for successful meningioma removal.

CHAPTER 12: INTRACRANIAL MENINGIOMAS

&

207

REFERENCES 1. Lee JH, Jeun SS, Evans JJ, Kosmorsky G. Surgical management of clinoidal meningiomas. Neurosurgery 2001; 48:1012–1021. 2. Tobias S, Kim CH, Kosmorsky G, Lee JH. Management of clinoidal meningiomas. Neurosurg Focus 2003; 14(6):1–7; article 5. 3. Sekhar LN, Babu RP, Wright DC. Surgical resection of cranial base meningiomas. Neurosurg Clin North Am 1994; 5(2):299–330. 4. Simpson D. The recurrence of intracranial meningiomas after surgical treatment. J Neurol Psychiatry Neurosurg 1957; 20:22–39. 5. Barnett GH. Surgical management of convexity and falcine meningiomas using interactive image-guided surgery systems. Neurosurg Clin North Am 1996; 7(2):279–284. 6. Barnett GH, Kaakaji W. Intracranial meningiomas. In: Barnett GH, Roberts DW, Maciunas RJ, eds. Image-Guided Neurosurgery. Clinical Applications of Surgical Navigation. St. Louis, Missouri: Quality Medical Publishing, Inc., 1998. 7. Barnett GH, Steiner CP, Kormos DW, et al. Intracranial meningioma resection using interactive frameless stereotaxy-assistance. J Image Guided Surg 1995; 1:46–52. 8. Evans JJ, Lee JH, Suh J, Golubic M. Meningiomas. In: Newell DW, Moore AJ, eds. Neurosurgery: Principles and Practice. London: Springer-Verlag, 2005. 9. Yasargil MG. Meningiomas. Microneurosurgery of CNS Tumors. Vol. IVB. Stuttgart: George Thieme Verlag, 1996.

13

TRANSPHENOIDAL HYPOHYSECTOMY Edwin Cunningham and Marc R. Mayberg Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION In the 1960s, the advent of operative microscopy and intraoperative fluoroscopy significantly diminished the morbidity of pituitary surgery and made selective microadenomectomy possible by transphenoidal approaches (1–3). Transphenoidal surgery remains the primary approach for the majority of pituitary fossa and parasellar lesions because of its safety, low morbidity, and direct access to the pathology. For experienced surgeons, intraoperative fluoroscopy provides adequate trajectory guidance to the sella in the majority of primary (first operation) cases. However, single plane fluoroscopic images provide only two-dimensional information to the surgeon (i.e., anterior/posterior and rostral/caudal), and do not directly reflect right/left orientation with respect to the midline. The orientation to midline in transphenoidal surgery is especially important because of the potentially devastating consequence of injury to carotid artery, cranial nerves in the skull base, or violation of the middle fossa. In addition, when tumor has destroyed bony landmarks, fluoroscopy may not delineate critical boundaries for the surgeon. In reoperation or in cases where the lesion has extended beyond the sella, anatomic landmarks may be distorted or absent. In these cases, image-guided surgery (IGS) confirms correct trajectory and midline orientation, minimizing the risk of neurovascular complications. IGS provides highly accurate, real-time information to the surgeon regarding localization, trajectory, and depth, and frequently enables less invasive approaches to lesions of the brain, skull base, and spine. Endoscopy has also been applied to neurosurgical procedures, including intraventricular lesions, sellar and parasellar lesions, intraspinal and paraspinal lesions, extra-axial intracranial lesions, and vascular lesions. Pituitary tumors and other lesions of the anterior skull base and clivus are particularly well-suited to transphenoidal endoscopic surgery due to the ease of access through the nostril, working space provided by the sphenoid sinus, and fields of view potentially wider and nonlinear compared to direct microscopy. However, transphenoidal endoscopy sacrifices the stereoscopic view provided by microscopy, and orientation (especially midline) may be more difficult without exposure of

Icons indicate materials available on the accompanying CD.

209

210

&

PART III: CLINICAL APPLICATIONS

anatomical landmarks. For these reasons, many surgeons prefer a transnasal, extramucosal approach using direct microscopy (4) since the morbidity of this approach is comparable to that of endoscopy. This chapter summarizes the two common transphenoidal approaches (direct microscopic and endoscopic) to the sella and the role of IGS in these techniques.

DIRECT MICROSCOPIC APPROACH Currently, most neurosurgeons use a microscope for transphenoidal surgery. The sphenoid sinus may be accessed through a sublabial trans-septal, transnasal transseptal, or endonasal septal pushover approach (5–7). The last technique is the fastest, most direct route, and is steadily gaining favor among surgeons (Fig. 1). For any of the approaches to the pituitary fossa, the two surgical imperatives are (i) maintaining the appropriate trajectory toward the sphenoid ostia during the intranasal dissection and (ii) defining the midline of the sphenoid bone (rostrum). Failure to adhere to these principles may result in misdirection superiorly into the ethmoid sinus/anterior fossa, or lateral entry into the cavernous sinus, carotid artery, middle fossa, or infratemporal fossa.

Figure 1 Schematic of transnasal septal pushover approach to sella and parasellar region.

CHAPTER 13: TRANSPHENOIDAL HYPOHYSECTOMY

&

211

Primary (First Operation) Transphenoidal Surgery The majority of neurosurgeons use intraoperative videofluoroscopy to confirm correct trajectory toward the sphenoid sinus and hypophyseal fossa. Videofluoroscopy provides real-time two-dimensional data in the sagittal plane, but does not provide any medial/lateral orientation (i.e., midline) for the surgeon. In this approach, midline orientation relies upon recognition of anatomic landmarks. After the cartilaginous and bony septa are displaced laterally, the rostrum of the sphenoid bone is located. This structure is the key midline landmark of the exposure. Once the rostrum is identified, the anterior wall of the sphenoid sinus may be safely removed and the sella visualized. The direct microscopic approach provides excellent stereoscopic visualization of the sphenoid sinus and pituitary fossa. Smith et al. summarized the complications associated with primary transphenoidal operations performed using fluoroscopic guidance. For surgeons who had performed more than 200 procedures, the risk of neurovascular complication (e.g., carotid artery injury) was minimal (Table 1). With experience, additional right/left orienting landmarks can be ascertained from comparison of preoperative imaging to septae in the sphenoid sinus and identification of the carotid canals at the inferior aspect of the cavernous sinus. If preoperative imaging demonstrates intact anterior sphenoid bone anatomy and pathology that is primarily in the pituitary fossa and adjacent sphenoid sinus, intraoperative fluoroscopy provides sufficient guidance for the experienced surgeon to safely perform primary transphenoidal surgery without the aid of IGS. A less experienced surgeon may prefer the additional data provided by IGS to confirm the fluoroscopic and visual feedback until adequate comfort with the latter techniques is obtained. In cases with disruption of normal sphenoid bone anatomy, the risk of deviating from the midline is significantly higher. Surgeon comfort and experience must also be considered.

Table 1 Percentage of Operations, in Three Experience Groups, Resulting in Each Complication of Transphenoidal Pituitary Surgery in the National Survey Percentage of operations resulting in complicationa Complication Carotid artery injury Central nervous system injury Loss of vision Ophthalmoplegia Cerebrospinal fluid leak Meningitis Postoperative sinusitis a

Estimation by participating neurosurgeons. Number of previous operations.

b

Source: From Ref. 19.

<200b 1.4 1.6 2.4 1.9 4.2 1.9 9.6

200–500 0.6 0.9 0.8 0.8 2.8 0.8 6.0

>500 0.4 0.6 0.5 0.4 1.5 0.5 3.6

212

&

PART III: CLINICAL APPLICATIONS

Some authors have advocated using frameless image guidance instead of videofluoroscopy for all transphenoidal approaches (4,8,9). The major advantage of computed tomography (CT) image guidance over videofluoroscopy is that it provides data in all three planes, allowing the surgeon to continuously orient to the midline and to continuously identify bony anatomy. Critics of transphenoidal IGS have cited the following problems: (i) additional preoperative imaging that is otherwise unnecessary; (ii) increased intraoperative time spent on data transfer and registration; (iii) lack of real-time imaging to account for tissue shift during operation; and (iv) additional cost for marginal benefit in a procedure that is already safely and effectively performed (10–12). One analysis demonstrated that the cost in dollars and operating time associated with transphenoidal IGS is not substantially greater than videofluoroscopic-guided transphenoidal surgery (9). These authors noted a net cost difference to the patient of $318 with a six-minute increase in mean duration of surgery versus videofluoroscopy. The authors concluded that the ‘‘use of frameless stereotaxy . . . in transphenoidal surgery is justified by the additional axial and coronal guidance it provides while requiring minimal additional operating time and costs.’’ In our experience (unpublished data), CT-guided frameless stereotactic assistance for primary direct transphenoidal surgery is at least $700 more expensive per case than videofluoroscopy. More importantly, we do not believe that the additional information it provides is necessary in the majority of first-time approaches in patients who have normal sphenoid bone/sinus anatomy.

Transphenoidal Reoperations and Invasive Parasellar Tumors In the patient who has undergone a previous transphenoidal approach, a preoperative MRI and/or CT scan assists in determining the need for IGS in any subsequent operation. If the sphenoid sinus was packed with fat in a previous procedure and current imaging shows loss of sphenoid bone landmarks combined with extensive neoplasm or reactive tissue within the sinus, then IGS is probably indicated. On the other hand, if much of the sphenoid bone remains, the sphenoid sinus is well pneumatized and there is no significant soft tissue obscuring the anterior wall and floor of the sella, which may be safe to perform a reoperation with videofluoroscopy alone. Additionally, individual surgeons experience and comfort level must be considered.

IGS for Transphenoidal Surgery CT-guided stereotaxy is the mainstay of IGS for transphenoidal surgery because CT accurately depicts the bony anatomy that must be defined to safely approach the sella. In some cases, T1-weighted MRI data may also be of value and the CT/MRI data can be fused. Fiducials are placed on the patient’s scalp and a preoperative stereotactic

CHAPTER 13: TRANSPHENOIDAL HYPOHYSECTOMY

&

213

CT scan is obtained. After intubation, the patient’s head is secured in the Mayfield head holder and the tracking device is attached to the head holder. For the transphenoidal approach, the head is laterally flexed to the left and slightly extended to facilitate the surgeon’s approach from the patient’s right side. The fiducials are then registered and the accuracy of stereotaxy is checked against common surface landmarks. After initial experience is gained with the navigation software, registration should not add more than 7 to 10 minutes to the operation. The surgeon may approach the sphenoid sinus through one of the three routes previously described. As the depth of the intranasal exposure increases, correct trajectory should be confirmed with the stereotactic probe while viewing the sagittal plane on the monitor. With intermittent confirmation from IGS, the surgeon more confidently proceeds with the exposure of the pituitary fossa. It must be reemphasized that the visual feedback provided on the computer monitor is based on the preoperative scan. There is no real-time picture of the anatomy in situ as the surgery progresses. This is especially important to remember during resection of the suprasellar portion of tumor, which frequently shifts down into the sella. Inappropriate reliance on IGS at this time of the operation may lead the surgeon to aggressively dissect in the suprasellar cistern, risking injury to the optic apparatus or hypothalamus. More recently, a fluoroscopic stereotactic system has been developed and described by Jane et al. (13). The FluoroNav Virtual Fluoroscopy SystemÕ (Medtronic Sofamor Danek, Inc., Memphis, Tennessee, U.S.) uses software that is part of the StealthStation Treatment Guidance PlatformÕ (Medtronic Sofamor Danek, Inc.). A dynamic reference arc is fixed to a radiolucent head holder that is firmly secured to the patient’s skull. Frontal and lateral videofluoroscopic images are obtained, calibrated, and stored in the software platform. The fluoroscope is then removed prior to the initiation of surgery. With a reference probe, the surgeon can refer to the stored images throughout the operation and visualize the instruments in real time. Using this technique, Jane et al. reviewed a consecutive series of 20 patients: 10 patients were operated on using FluoroNavÕ and 10 other patients underwent transphenoidal surgery with standard videofluoroscopy (non-FluoroNavÕ group). Tables 2 and 3 summarize the results of the comparison between the groups of surgical times and costs. FluoroNavÕ assisted surgery was approximately $740/ Table 2 Comparison of Surgical Times Between Groups Variable Mean setup time (min) (P ¼ 0.067) Mean operative time (min) (P ¼ 0.031) Mean no. of fluoroscopic X rays (P ¼ 0.03) a

FluoroNav surgeries

Non-FluoroNav surgeries

42.8 85.6 3.667

35.6 102.56 5.44

Setup time refers to the time from intubation to incision; operative time refers to the time from incision to closure. Source: From Ref. 13.

214

&

PART III: CLINICAL APPLICATIONS

Table 3 Cost Comparison Between FluoroNav and Traditional Transphenoidal Surgeries Non-FluoroNav Cost category FluoroNav costs ($) costs ($) Fluoroscopy Mean setup time ($20.14/min) Mean operative time ($20.14/min) Surgical fee for stereotaxy Total cost Cost with operative time excluded

179.09 861.99 1723.98 240.4 3005.46 1281.48

254.56 716.98 2065.56 0 3037.01 971.45

Source: From Ref. 13.

case less expensive than CT-guided frameless stereotaxis (previously published data from the authors’ institution) in the study. The results from this single study imply that fluoroscopic-guided stereotaxy may be an equivalent or superior alternative to videofluoroscopy for first-time transphenoidal approaches. As discussed, for redo approaches in patients with distorted sphenoid bone/sinus anatomy, we believe that CT-guided stereotaxy is more appropriate.

Endoscopic Approach Over the last decade, an increasing number of neurosurgeons have developed techniques for an endoscopic transphenoidal approach (14,15). Although endoscopic TS surgery has been promoted as ‘‘less invasive,’’ in fact the approach to the sphenoid sinus is similar to that of the endonasal septal pushover approach with the microscope. The middle turbinate is outfractured and the posterior nasal septum is displaced to the contralateral side to expose the rostrum of the sphenoid bone and both sphenoid ostia. Lateral videofluoroscopy is used to confirm correct trajectory. A 1.5 to 2 cm anterior sphenoidotomy is performed to provide adequate exposure of the sella. A combination of 0 -angled and 30–70 -angled rigid endoscopes are used to perform the procedure along with other endoscopic instruments that are passed through the same or opposite nostril. In this manner, endoscopic surgery provides the capacity to view structures beyond ‘‘line of sight’’ restrictions imposed by the operating microscope and speculum (Fig. 2). Similarly, the smaller aperture of the endoscope compared to speculum allows a wider excursion of the endoscope tip, which can be positioned at nearly any position in the ethmoid, sphenoid, and maxillary sinuses. Despite this flexibility, endoscopic TS has certain disadvantages. First, the endoscope image is monocular and depth must be inferred by the surgeon from visual clues. Second, the light intensity and image clarity are less in the endoscope compared to the operating microscope, and the endoscope tip may become obscured by blood. Surgeons experienced with traditional parasellar anatomy and the depth of field provided by binocular operating microscopes usually have little difficulty adapting to the endoscope. Less experienced surgeons, however, may have a more difficult time

CHAPTER 13: TRANSPHENOIDAL HYPOHYSECTOMY

&

215

Figure 2 Schematic representation of expanded field of view to anterior skull base by transnasal approaches.

interpreting the three-dimensional anatomy of the sphenoid sinus, and require more time for the procedure due to visual obscuration as above. In addition, midline disorientation can be magnified in endoscopic TS due to the wide excursion of the endoscope tip. The results of a retrospective analysis by Jho and Carrau (15) demonstrated that the endoscopic approach can be performed effectively and safely. We believe that the same principles regarding the application of IGS to microscope-assisted transphenoidal surgery apply to endoscopic approaches. For experienced surgeons, first-time endoscopic transphenoidal approaches may be safely performed without IGS if there is no significant disruption of sphenoid bony landmarks. If the pathology has invaded the sphenoid base or if preoperative imaging suggests distortion of residual anatomy in the reoperative case, then IGS may be of assistance. Lasio et al. (16) assessed the impact of IGS for endoscopic reoperation. During a 45-month period, 19 endoscopic endonasal transphenoidal reoperations were performed for recurrent pituitary adenomas. In 11 of 19 cases, the procedure was performed with the aid of IGS. There was no difference between the groups in morbidity or mortality (there was none for the 19 patients). For the IGS-assisted procedures, the mean setup time was 13 minutes longer, but the mean operative time was 36 minutes shorter. The results of this small series show that surgery is not prolonged with IGS and suggest that surgeon comfort was increased significantly by the presence of image guidance.

216

&

PART III: CLINICAL APPLICATIONS

Endoscopic Surgery Using Computer-Assisted Virtual Navigation To address the limitations of endoscopic transphenoidal surgery, we have used a technique that combines image guidance with computer-generated ‘‘virtual’’ threedimensional images. The ability to project accurately registered, three-dimensional rendered images of objects along the trajectory of pointing devices has been available for several years, and is included in commercially available navigation platforms. However, hardware and software to update and reconstruct these images in real time have only recently become available. The CBYONÕ IGS system includes software that enables coregistration of a perspective-based three-dimensional virtual image along the trajectory of a pointing device. By attaching localizing spheres to an endoscope, the tip can be tracked conventionally, and the ‘‘line of site’’ can be accurately coregistered to the endoscopic image using a visual fiducial field (Fig. 3). In this manner, the IGS monitor can be configured to show triplanar representations of endoscope tip position, the ‘‘real video’’ view of the endoscope camera video output, and a ‘‘virtual’’ view of the computed endoscope field of view (Fig. 4). In the virtual image, bone and soft tissue can be semiopaque or transparent, and critical structures (tumor, carotid artery, optic nerves) can be color-highlighted and opaque. Using the virtual image, the surgeon can quickly and accurately visualize the important structures in the operative field, navigate the endoscope tip to the appropriate location, and use three-dimensional cues to maintain orientation. A 5 mm Storz DCIÕ rigid endoscope was used for all procedures. This endoscope has a digital video output; an automatic rotational orientation feature was disabled to accommodate the coregistration of rendered IGS images. Scope angles of 0 , 30 , and 70 were used interchangeably. For lesions of the sella and parasellar region, a transnasal, transphenoidal approach was used as above. The osteotomy at the anterior sphenoid was enlarged compared with the usual endo-

Figure 3 Registration device for CBYONÕ image-guided endoscopy. The endoscope image (right) is registered to an array of circles of various dimensions held at a fixed distance from the endoscope tip. The computer-generated ‘‘virtual’’ image (left) shows good concordance.

CHAPTER 13: TRANSPHENOIDAL HYPOHYSECTOMY

&

217

Figure 4 Typical screen display using CBYONÕ image-guided endoscopy. The real-time image from a digital endoscope is at upper right; computer-generated ‘‘virtual’’ representation of previously designated anatomic structures (tumor; carotid arteries) oriented in line-of-sight trajectory of the endoscope at upper left. Standard triplanar views of endoscope tip position and trajectory are aligned across the bottom.

scopic technique to accommodate direct microscopy for verification of anatomic localization. In all cases, an operating microscope was used to confirm the accuracy of surface landmarks identified by image-guided endoscopy (IGE). No complications occurred attributable to IGE or otherwise. The initial registration of the IGS system and coregistration of the endoscope added approximately 15 to 30 minutes to the case length; an additional 15 to 30 minutes was spent prior to surgery for image transfer and processing. Complete gross resections were obtained for all pituitary tumors. Excellent accuracy was obtained for IGE images. Endoscope tip position corresponded to surface landmark anatomy with accuracy comparable to standard IGS pointing devices. Similarly, there was an excellent concordance between direct endoscopic view and rendered virtual images, as assessed by surface landmark topography and lesion margins; the accuracy of representation for nonvisualized structures (carotid artery, optic nerves) could not be assessed. Although time required for final endoscope positioning was not compared to non-IGE approaches, it was our impression that IGE significantly facilitated accurate placement of the endoscope within seconds of insertion.

218

&

PART III: CLINICAL APPLICATIONS

Advantages of Image-Guided Endoscopy For several reasons, the potential for central nervous system endoscopy has not been fully realized. First, abdominal, thoracic, joint, and cranial sinus endoscopy utilize distinct cavities, which enable a working space with generally resilient margins and relatively noninvasive access. Thus, in general, neuroendoscopic approaches developed for natural cavities (ventricle, paranasal sinuses, disc space, thoracic cavity) are applicable for a limited subset of lesions within or immediately adjacent to those cavities. Second, neurosurgeons are nearly uniformly comfortable and adept at microsurgery using the operating microscope, which affords high resolution, excellent illumination, and stereoscopic view, often through exposures that are not significantly more invasive than endoscopy. Third, neuroendoscopic instrumentation has not been developed to accommodate specific tissue-handling issues (e.g., brain retraction) or neurosurgical tasks (e.g., tumor removal or aneurysm clipping). Nevertheless, the central nervous system and spine may be highly suited to endoscopy, and current limitations should be surmountable through technology development and physician training. In contrast to neuroendoscopy, neuronavigation using IGS has substantially preceded similar applications in other venues outside of the nervous system. In a relatively short interval, IGS has become an integral part of neurosurgical technique. In theory, IGS enables less invasive and safer approaches to brain and spine, although these advantages have not been definitively demonstrated. A major disadvantage of IGS has been the necessity to display navigation data on a workstation screen, which is outside the operative field, requiring the surgeon to reorient between the two views. This has been addressed by overlay of navigation data on the operating microscope field of view, including trajectories and target contours. These overlays, however, provide only two-dimensional data, and target contours derived from a single perpendicular plane are less helpful in the typical microscope depth of field. Strategies to inject threedimensional data using binocular overlays are currently in development (17). In addition, computer-generated virtual reality (VR) representations can be combined with real-world views to produce augmented reality (18). IGE provides several unique features, which may be of potential benefit to the surgeon. First, standard navigation coordinates of the endoscope tip and trajectory provide three-dimensional orientation. Second, the navigation data and VR image provide cues for depth perception. Third, the VR image can display both pathologic and normal structures hidden from direct view by soft tissue or bone, and maintain a visual orientation during intermittent loss of endoscope image fidelity (e.g., obscuration by blood). Fourth, visualization of the three-dimensional VR lesion target during placement facilitates rapid and accurate transit and positioning of the endoscope. Finally, the accommodation of angled endoscopes enables both direct and virtual inspection of structures tangential to the direct line of vision. On the other hand, there are several limitations to current IGE. Preoperative imaging, image processing, and registration add time and potential cost to the procedure. As in any IGS, inaccurate registration may result in complications related to a false sense of surgical security. In our experience, endoscope and VR registration

CHAPTER 13: TRANSPHENOIDAL HYPOHYSECTOMY

&

219

were accurate and IGE did not add significant time to the procedure. Despite the wider field of view afforded by angled scopes, the system at present is restricted to rigid endoscopes, and thus linear approach trajectories. Future development of IGE may enable several new applications of neuroendoscopy. For example, 120 or greater IGE devices might enable direct and virtual views of the opposite side of lesions (e.g., aneurysm, acoustic neuroma) during microsurgical or endoscopic treatment. Also, flexible IGS-tracked endoscopes may facilitate endoscopic procedures through nonlinear approaches, such as the subarachnoid or epidural spaces. Finally, high-resolution virtual endoscopy may augment diagnostic image interpretation and surgical or endovascular planning. Ultimately, IGE and head-mounted displays may eliminate the operating microscope and the need for brain retraction or wide exposures to provide linear trajectories to lesions of the brain or spine.

CONCLUSION The transphenoidal approach remains the surgical procedure of choice for lesions located in the pituitary fossa and parasellar region. Most first-time approaches to the sella may be safely performed with videofluoroscopy alone or equivalently priced fluoroscopic-guided computer-assisted navigation. Computer-assisted navigation is of primary importance in transphenoidal reoperations and tumors invading the skull base, in which sphenoid bone landmarks are disrupted or absent. New techniques may combine incorporation of three-dimensional virtual navigation images with endoscopic techniques to provide safer and more effective surgery.

220

&

PART III: CLINICAL APPLICATIONS

REFERENCES 1. Bateman G. Trans-sphenoidal hypophysectomy: a review of 70 cases treated in the past two years. Trans Am Acad Ophthalmol Otolaryngol 1962; 66:103–110. 2. Bouche J, Guiot G, Freche. Evaluation of hypophysectomy by low transseptal approach based on 120 operations. Ann Otolaryngol Chir Cervicofac 1966; 83(6):466–471. 3. Hardy J, Wigser SM. Trans-sphenoidal surgery of pituitary fossa tumors with televised radiofluoroscopic control. J Neurosurg 1965; 23(6):612–619. 4. Walker DG, Ohaegbulam C, Black PM. Frameless stereotaxy as an alternative to fluoroscopy for transphenoidal surgery: use of the InstaTrak-3000 and a novel headset. J Clin Neurosci 2002; 9(3):294–297. 5. Wilson WR, Khan A, Laws ER Jr. Transseptal approaches for pituitary surgery. Laryngoscope 1990; 100(8):817–819. 6. Roux FX, Page P, Nataf F, Devaux B, Djian MC, Joly LM. The endonasal approach to pituitary adenomas: experience in 105 procedures. Ann Endocrinol (Paris) 2002; 63(3):187–192. 7. Griffith HB, Veerapen R. A direct transnasal approach to the sphenoid sinus. Technical note. J Neurosurg 1987; 66(1):140–142. 8. Kacker A, Komisar A, Huo J, Mangiardi J. Transphenoidal surgery utilizing computerassisted stereotactic guidance. Rhinology 2001; 39(4):207–210. 9. Elias WJ, Chadduck JB, Alden TD, Laws ER Jr. Frameless stereotaxy for transphenoidal surgery. Neurosurgery 1999; 45(2):271–275; discussion 275–277. 10. Hardy J. Neuronavigation in pituitary surgery. Surg Neurol 1999; 52(6):648–649. 11. Hardy J. Frameless stereotaxy for transphenoidal surgery. Neurosurgery 2000; 46(5): 1269–1270. 12. Warnke PC. Neuronavigation and surgical neurology: the beginning of a new age or the end of an old age? Surg Neurol 1999; 52(1):7–8; discussion 8–12. 13. Jane JA Jr, Thapar K, Alden TD, Laws ER Jr. Fluoroscopic frameless stereotaxy for transphenoidal surgery. Neurosurgery 2001; 48(6):1302–1307; discussion 1307–1308. 14. Jho HD, Alfieri A. Endoscopic endonasal pituitary surgery: evolution of surgical technique and equipment in 150 operations. Minim Invasive Neurosurg 2001; 2001(1):1–12. 15. Jho HD, Carrau RL. Endoscopic endonasal transphenoidal surgery: experience with 50 patients. J Neurosurg 1997; 87(1):44–51. 16. Lasio G, Ferroli P, Felisati G, Broggi G. Image-guided endoscopic transnasal removal of recurrent pituitary adenomas. Neurosurgery 2002; 51(1):132–136; discussion 136–137. 17. Edwards PJ, King AP, Maurer CR Jr, de Cunha DA, Hawkes DJ, Hill DLG, Gaston RP, et al. Design and evaluation of a system for microscope-assisted guided interventions (MAGI). IEEE Trans Med Imaging 2000; 19:1082–1093. 18. Bockholt U, Bisler A, Becker M, Muller-Witrig W, Voss G. Intra-operative support via augmented reality techniques in endoscopic surgery. Comput Aided Surg 2003; 8(6):310–315. 19. Ciric I, Ragin A, Baumgartner C, Prence D. Complications of transphenoidal surgery. Neurosurgery 1997; 40(2):225–237.

14

VASCULAR MALFORMATIONS William E. Thorell and Peter A. Rasmussen Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

INTRODUCTION During the last half of the 20th century, many improvements in neurosurgical care emanated directly from developments in neuroimaging. These developments have advanced the understanding of both normal and abnormal states of the central nervous system. The digital information collected with modern computer-dependent neuroimaging techniques, once utilized only for planning operations, can now be manipulated and updated during therapeutic procedures. Neurosurgery is approaching the point where the operating room will not only function as the place where treatment is administered based on previously gathered information but also the entire treatment will be modified based on computerized anatomic and physiologic data gathered continuously during procedures. This paradigm exists, in part, already and is used increasingly in all realms of neurosurgery, including the evaluation and treatment of vascular malformations and lesions of the central nervous system. Therefore, although computer technology has primarily impacted the diagnostic evaluation of vascular malformations through advancements in neuroimaging, it has also impacted their treatment. In an autopsy series published in 1966, McCormick (1) described a series of vascular lesions classified into four groups: arteriovenous malformations, cavernous malformations, capillary telangiectasias, and venous angiomas. The understanding of the pathophysiology, imaging characteristics, and clinical significance of these lesions has continuously evolved, especially with the advent of computed tomography (CT) in the 1970s and magnetic resonance imaging (MRI) in the 1980s. Subsequent refinements in the techniques along with enhanced computer speed, data storage, and networking capabilities constantly improve imaging and, therefore, the treatment of patients with these afflictions. Although substantial influence on the surgical treatment for vascular malformations comes from improvements in neuroimaging, computer technology has significantly changed the treatment strategies as well. Two broad areas of computer applications directly influencing the therapeutic realm of neurosurgery are stereotactic radiosurgery and frameless, stereotactic image guidance. Icons indicate materials available on the accompanying CD.

221

222

&

PART III: CLINICAL APPLICATIONS

VASCULAR MALFORMATIONS Arteriovenous Malformations Arteriovenous malformations (AVMs) of the brain are congenital malformations that form between the 3rd and 8th week of gestation. Within the AVM, blood is shunted directly from primitive arteries into veins without any intervening capillary network (1). Some brain tissue can be intermingled within the malformation but is comprised primarily of nonfunctional glial tissue (Fig. 1) (1). Although somewhat controversial, the risk of hemorrhage from an AVM is approximately 2% to 4% per year (2). Patients who harbor an AVM most frequently present with a headache, a hemorrhage, or a seizure. Because AVMs are high-flow,

Figure 1 2-D intra-arterial digital subtraction angiogram and MRI demonstrating AVM.

CHAPTER 14: VASCULAR MALFORMATIONS

&

223

high-pressure lesions, instances of hemorrhage have a relatively high probability of causing significant morbidity or death. A hemorrhagic episode carries approximately 20% risk of major neurological deficit and 10% risk of death (2). However, unlike cerebral aneurysms, rehemorrhage during the acute stages following a hemorrhage is infrequent. Therefore, in the absence of surgical indications for evacuation of large intracerebral hematomas (ICH) or intraventricular hemorrhage, treatment is often delayed to allow for convalescence. Treatment options include surgical resection and stereotactic radiosurgery. Neuroendovascular options are, for the most part, an adjunct and in most instances are not considered a curative treatment option.

Cavernous Malformations Cavernous malformations (a.k.a. cavernoma, cavernous angioma) are abnormal, dilated, sinusoidal collections of vessels lined with a single layer of endothelium with no interposed brain tissue whose etiology is unclear (1). Cavernous malformations have been described in young children but have also been clearly demonstrated as de novo lesions in adults (3,4). They are considered low-flow lesions, which are associated with a developmental venous anomaly (venous angioma) in approximately 15% of cases (5). These lesions are frequently not visible during angiography. A characteristic hypointense ring secondary to hemosiderin and ferritin deposition from microhemorrhages is frequently seen on MRI (Fig. 2). They can infrequently present with larger, more catastrophic ICH. Many times, however, they are found incidentally or present with seizures. The rate of hemorrhage ranges between 0.11% and 1.1% (4–6). Intractable seizures, significant ICH, enlargement of the lesion, or peripheral location are indications for resection. Stereotactic radiosurgery has been reported for treatment of symptomatic cavernous malformations, which are located in areas of the brain where surgical risks are high (7–9). Because surgical risks and observation both represent very high risk options in certain cases, stereotactic radiosurgery is an increasingly utilized modality. This is not without controversy. The natural history of cavernous malformations is not well defined and it is not clear that stereotactic radiosurgery significantly impacts this course. Szeifert et al. (10) argued that myofibroblasts are responsible for the response of an AVM to radiation. Since cavernous malformations are lined with a single layer of endothelium, this response cannot account for any change in the natural history of symptomatic cavernous malformations. Furthermore, because histopathology is not readily available with stereotactic radiosurgery, the tissue response remains elusive. Finally, the complication rate in treated cases has prompted caution and calls for further study from several authors (7,8). Venous Angiomas Venous angiomas (a.k.a. venous malformations, developmental venous anomalies) are an anomalous configuration of normal veins draining normal brain tissue (1). The veins are most frequently arrayed angiographically in a radial pattern and are

224

&

PART III: CLINICAL APPLICATIONS

Figure 2 MRI with T1- and T2-weighted images demonstrating midbrain cavernous malformation with characteristic hypointense ring secondary to deposition of hemosiderin and ferritin from microhemorrhages.

sometimes referred to as a caput medusae. Frequently, they resemble a chicken’s foot on T1-weighted MRI scans with contrast. They are easily discerned with contrast-enhanced brain imaging (Fig. 3). Once thought to represent a pathologic entity because of their association with intracerebral hemorrhage, improved imaging modalities, especially MRI, have clearly changed this viewpoint. It is now clear that 15% of cavernous malformations are associated with a developmental venous anomaly and were probably the culprit when ICH was reported. Venous angiomas represent an anomalous drainage pattern of normal brain and treatment (i.e., resection or radiotherapy) interrupts the normal venous drainage system and potentially leads to venous infarction and/or hemorrhage.

CHAPTER 14: VASCULAR MALFORMATIONS

&

225

Figure 3 CT without (A) and with intravenous contrast (B–D) demonstrating venous angioma of the left cerebellar hemisphere and left paraclinoid aneurysm.

Capillary Telangiectasias Capillary telangiectasias represent an abnormal, dilated collection of thin-walled capillaries in the brain (1). Normal brain tissue is interposed amongst the abnormal vessels. This malformation is most commonly found in the pons and is usually a small, incidental finding on contrast-enhanced MRI. Some have speculated that

226

&

PART III: CLINICAL APPLICATIONS

they may represent a precursor to cavernous malformations (11). They rarely hemorrhage and treatment is not indicated.

REVIEW OF IMAGING MODALITIES Magnetic Resonance Angiography Magnetic resonance angiography of the brain generates an image of the intracranial vasculature from image voids created in magnetic resonance by flowing blood. This provides a noninvasive intracranial vessel imaging modality. Aneurysms greater than 3 mm can be discerned with magnetic resonance angiography (12,13).

Intraoperative Magnetic Resonance Imaging Intraoperative MRI combines the detailed images delivered by MRI with real-time acquisition of anatomic information during an operation. The initial experience with this imaging modality centers on the resection of brain tumors. As experience increases, the use of this modality is expanding to all aspects of neurosurgery. With this tool, a neurosurgeon can assess the accuracy and extent of lesion resection, confirm placement of a biopsy needle, and identify complications without leaving the operating room (13). It also allows for procedural modifications to structural changes, such as brain shift. The inability to compensate for brain shift during an operation is perhaps the most obvious weakness of current image-guidance systems. With intraoperative MRI, the data/image acquisition phase of conducting an imageguided surgery is not limited to the preoperative stage. The application of intraoperative MRI to the surgical treatment of AVMs is not nearly as useful as with tumors. Angiography remains the gold standard for demonstrating arteriovenous shunting and is mandatory to demonstrate complete removal of an AVM after resection. Currently, MRI can be utilized intraoperatively to assess extent of resection, but still cannot effectively confirm complete resection. Intraoperative MRI can be helpful to confirm resection of cavernous malformations. Disadvantages of intraoperative MRI include large initial outlays for installation and alteration of the operating room and relatively poor image quality when judged against traditional MRI (14).

CT Angiography Using a timed intravenous contrast bolus, multirow detector spiral CT scanners can be used to generate a three-dimensional (3-D) representation of human vasculature. It is possible to image vessels as small as 1 mm in diameter and with

CHAPTER 14: VASCULAR MALFORMATIONS

&

227

high-speed scanners images from the aortic arch through the circle of Willis and beyond. These can be obtained in only 15 to 30 seconds. There are several distinct advantages of CT angiography (CTA). First, it is fast and relatively noninvasive. It does, however, require a fast multirow detector CT scanner. Those units with at least 16 rows of detectors offer the thinnest slices and highest levels of resolution. Complications are primarily those associated with contrast reactions and infiltration at the injection site. The ability of the viewer to readily manipulate the images in 3-D is very important (Fig. 4). The quality and speed of the 3-D reconstruction depends on the computer hardware used and the software algorithms employed. CT technicians can readily perform the study with only a minimum of additional training. Therefore, additional specialized personnel are not required. Patients with suspected cerebrovascular pathology who are undergoing CT to rule out ICH can have useful information concerning large intracranial vessels obtained quickly and noninvasively. The images are readily transmitted over the Internet so that an outside party can view and consult on the images. The sensitivity to cerebral aneurysms has been reported to be approximately 97% (15). The role of CTA is not limited to intracranial aneurysms. CTA can be used to visualize intra- and extracranial vascular malformations as well as cervical and intracranial stenosis (Figs. 5 and 6). Disadvantages of CTA include difficulty visualizing small, distal arteries. This complicates the analysis of infundibuli and perforator vessels related to aneurysms (15). In addition, vessel measurement can be inaccurate with CTA (15). With regards to AVMs, CTA lacks the ability to represent the rate of flow and hemodynamic properties, which the sequential imaging of catheter angiography provides.

Figure 4 CT angiogram with left paraclinoid aneurysm (seen on CT with contrast enhancement in Fig. 3 B–D).

228

&

PART III: CLINICAL APPLICATIONS

Figure 5 CT angiogram and 2-D angiogram demonstrating AVM of the scalp in the right supraorbital region.

CT Perfusion CT perfusion is a relatively new technique that, like CTA, is based on a timed, intravenous contrast bolus followed into the intracranial vessels. A bolus of contrast followed into a specified volume of brain allows calculation of cerebral blood volume, flow and time-to-peak values. This imaging modality, like that of CTA, is in its relative infancy and new applications are being described continuously. For example, CTP can be used along with CT and CTA at the time of evaluation of

CHAPTER 14: VASCULAR MALFORMATIONS

&

229

Figure 6 3-D rotational angiogram demonstrating aneurysm of the basilar summit.

acute stroke. It may also prove helpful in the management of vasospasm after subarachnoid hemorrhage. 3-D Rotational Angiography In neurosurgery, one of the most important aspects of all new imaging modalities is the ability to manipulate 3-D images. This significantly improves a surgeon’s understanding of the underlying anatomy of a given disease process. Angiography, like MRA and CTA, now has the ability to render in a 3-D format (Fig. 7). This improves the interpretation of the images, decreases the number of angiograms and the amount

230

&

PART III: CLINICAL APPLICATIONS

Figure 7 3-D rotational angiogram demonstrating small AVM of the right cerebellar hemisphere.

of contrast required. A significant advantage of 3-D rotational angiography, especially when evaluating a possible AVM, is the ability to collect flow information during the study. Of course, the major drawbacks to catheter angiography are the risk of thromboembolic complications, vessel injury, and hematoma formation at the access site.

STEREOTACTIC RADIOSURGERY Stereotactic radiosurgery (SRS) delivers focused radiation from multiple point sources to a focal point. The convergence of radiation at a focal point maximizes the delivered dose to the pathologic target, while minimizing dosage to the surrounding brain. Several systems exist which differ in their source of radiation but work on the same underlying principle of beam focusing. Each of these systems relies on computer software for target planning (Fig. 8). There are numerous applications for this technology outside the treatment of vascular malformations of the central nervous system including neuro-oncology, functional neurosurgery, and pain management. Nevertheless, the most common use in vascular neurosurgery is the treatment of AVMs especially small, deep lesions or those located in eloquent regions. In most instances, SRS is offered as the best treatment option to patients who harbor an AVM less than 3 cm in maximum diameter (Spetzler–Martin Grade 3 or less) and in whom the combined surgical endovascular embolization complication rate exceeds the permanent neurological deficit rate combined with the risk of hemorrhage over the approximate 2-year delay to complete involution. The disadvantage of SRS with the treatment of intracranial AVMs is delayed involution and risk of radiation-induced neurological deficits. Approximately two years may be required before lesions up to 3 cm in largest diameter will have

CHAPTER 14: VASCULAR MALFORMATIONS

&

231

Figure 8 Gamma knife stereotactic radiosurgery planning software for treatment of Spetzler–Martin Grade 3 AVM (3 cm) located in the left premotor area.

no arteriovenous shunting. The risk of ICH exists until shunting is eliminated entirely, but it does not appear to be increased after treatment in most patients (16,17). Approximately 70 to 80% of lesions will be obliterated at two years post-treatment (9,18). Transient neurological deficits related to radiation injury occur in 3 to 4% of cases and permanent injuries occur in 1% (17). There are descriptions of stereotactic radiosurgery for cavernous malformations, but this is not currently widely utilized (19) and is under investigation as a treatment option for patients with inoperable cavernous malformations. FRAMELESS STEREOTACTIC IMAGE GUIDANCE Stereotaxy allows for precise localization of intracranial structures and lesions. Several authors have described using traditional, frame-based stereotactic systems as adjuncts used for the resection of vascular malformations of the brain (20,21). Traditional stereotaxy has a limited breadth of use. It is used most frequently for biopsy, lesion making for a functional procedure or for ventricular access (22). Frameless stereotaxy, however, combines the accuracy of traditional framed-based stereotaxy with less obstructive equipment. In addition, frameless stereotactic

232

&

PART III: CLINICAL APPLICATIONS

systems utilize digital neuroimaging to provide real-time visual/stereotactic feedback to the surgeon. Frameless stereotaxy allows for nearly continuous display of the surgical location (22). The advantages of such systems include potentially smaller craniotomies and cortisectomies, better trajectory planning, less brain retraction, less exploration, and improved margins during resection of AVMs cavernous malformations, or other lesions (20,21,23,24). This lessens the probability of leaving residual vascular malformation. By registering information regarding arterial inflow from preoperative angiography, frameless stereotaxy can be used to demonstrate large feeding arterial pedicles for early division (25). Functional information gathered during intraoperative testing can also be registered along with the imaging data with the results visually displayed. Some potential weaknesses of frameless stereotactic image guidance include the lack of up-to-the-minute real-time imaging leading to inaccuracy. The most common example of this is brain shift after opening the dura. However, intraoperative ultrasound and/or MRI combined with the use of image guidance can overcome this problem (23). In addition, early brain shift can be minimized by limiting the use of diuretics, hyperventilation, and cerebrospinal fluid drainage. This maximizes the benefit of image guidance for planning a minimal incision, craniotomy, and accurate trajectory. Brain shift is usually not a significant issue during resection of vascular lesions after localization is complete because they are usually easily discernible from surrounding brain. This differs considerably from the resection of an infiltrating primary brain tumor for which continuous, accurate localization and spatial orientation are critical for safe, accurate resection. Image guidance is most helpful with small, deep AVMs or cavernous malformations with no cortical presentation and aneurysms in unusual locations (24). The major limitation of image-guided neuronavigation when utilized for the treatment of vascular malformations is the inability to ensure total resection of the lesion. Currently, intraoperative or early postoperative angiography is the gold standard for the determination of complete resection of an AVM by confirming no arteriovenous shunting. CONCLUSIONS Extensive technological advances have occurred in neurosurgery in large part due to advances in neuroimaging. Improvements in the imaging of vascular lesions of the central nervous system include MR and CT angiography, intraoperative MRI, as well as 3-D rotational angiography. These techniques have improved the understanding of the pathophysiology of vascular malformations of the brain. When combined with new computer-based techniques such as frameless stereotactic image guidance, craniotomies, cortisectomies, and brain retraction can be minimized and approach trajectories can be optimized. The amalgamation of stateof-the-art neuroimaging with computer-based surgical techniques has changed the preoperative assessment of patients with cerebrovascular disease, as well as altered the operating room from an isolated surgical suite to a combination surgical-imaging suite integrated with computer technology. All of these changes continue to improve the care of patients with neurosurgical diseases.

CHAPTER 14: VASCULAR MALFORMATIONS

&

233

REFERENCES 1. McCormick WF. The pathology of vascular (‘‘arteriovenous’’) malformations. J Neurosurg 1966; 24(4): 807–816. 2. Ondra SI, Troupp H, George ED, Schwab K. The natural history of symptomatic arteriovenous malformations of the brain: a 24-year follow-up assessment. J Neurosurg 1990; 73:387–391. 3. Narayan P, Barrow DL. Intramedullary spinal cavernous malformation following spinal irradiation: case report and review of the literature. J Neurosurg 2003; 98(Spine 1):68–72. 4. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural history of familial cavernous malformations: results of an ongoing study. J Neurosurg 1994; 80:422–432. 5. Robinson JR, Awad IA, Little JR. Natural history of cavernous angioma. J Neurosurg 1991; 75:709–714. 6. Curling OD, Kelly D, Elster AD, et al. An analysis of the natural history of cavernous angiomas. J Neurosurg 1991; 75:702–708 7. Kida Y, Kobayashi T, Mori Y. Radiosurgery of angiographically occult vascular malformations. Neurosurg Clin North Am 1999; 10:291–303. 8. Kim DG, Choe WJ, Paek HT, et al. Radiosurgery of intracranial cavernous malformations. Acta Neurochir 2002; 144:869–878. 9. Lunsford LD, Kondziolka D, Flickinger JC, et al. Stereotactic radiosurgery for arteriovenous malformations of the brain. J Neurosurg 1991; 75:512–524. 10. Szeifert GT, Kemeny AA, Timperley WR, et al. The potential role of myofibroblasts in the obliteration of arteriovenous malformations after radiosurgery. Neurosurgery 1997; 40:61–66. 11. Rigamonti D, Spetzler RF, Johnson PC, et al. Cavernous malformation and capillary telangiectasia: a spectrum within a single pathological entity. Neurosurgery 1991; 28:60–64. 12. Adams WM, Laitt RG, Jackson A. The role of MR angiography in the pretreatment assessment of intracranial aneurysms: a comparative study. AJNR 2000; 21: 1618–1628. 13. Ross JS, Masaryk TJ, Modic MT, et al. Incracranial aneurysms: evaluation by MR angiography. AJNR 1990; 11:449–456. 14. Martin C, Alexander E, Jolesz F, et al. Surgery in the MRI environment. In: Kaye AH, Black PM, eds. Operative Neurosurgery. 1st ed. London: Harcourt, 2000:185–198. 15. Kato Y, Katada K, Hayakawa M, et al. Can 3D-CTA surpass DSA in diagnosis of cerebral aneurysms. Acta Neurochir (Wien) 2001; 143:245–250. 16. Friedman WA, Blatt DL, Bova FJ, et al. The risk of hemorrhage after radiosurgery for arteriovenous malformations. J Neurosurg 1996; 84:912–919. 17. Friedman WA, Bova FJ, Bollampally, Sirisha, Bradshaw, Patrick. Analysis of factors predictive of success or complications in arteriovenous malformation radiosurgery. Neurosurgery 2003; 52:296–308.

234

&

PART III: CLINICAL APPLICATIONS

18. Steiner L, Lindquist C, Adler JR, et al. Clinical outcome of radiosurgery for cerebral arteriovenous malformations. J Neurosurg 1992; 77:1–8. 19. Kondziolka D, Lunsford LD, Flickinger JC, Kestle JRW. Reduction of hemorrhage risk after stereotactic radiosurgery for cavernous malformations. J Neurosurg 1995; 83(5):825–831. 20. Davis DH, Kelly PJ. Stereotactic resection of occult vascular malformations. J Neurosurg 1990; 72:698–702. 21. Sisti MB, Solomon RA, Stein BM. Stereotactic craniotomy in the resection of small arteriovenous malformations. J Neurosurg 1991; 75:40–44. 22. Bucholz RD, Levy AL, Marvowk S. Stereotactic and image-guided techniques. In: Kaye AH, Black PM, eds. Operative Neurosurgery. 1st ed. London: Harcourt, 2000:99–116. 23. Barnett G. Frameless stereotaxy. In: Kaye AH, Black PM, eds. Operative Neurosurgery. 1st ed. London: Harcourt, 2000:117–124. 24. Russell SM, Woo HH, Joseffer SS, Jafar JJ. Role of frameless sterotaxy in the surgical treatment of cerebral arteriovenous malformations: technique and outcomes in a controlled study of 44 consecutive patients. Neurosurgery 2002; 51(5):1108–1118. 25. Muacevic A, Steiger HJ. Computer-assisted resection of cerebral arteriovenous malformations. Neuro surgery 1999; 45:1164–1170.

15

IMAGE-GUIDED SPINAL NAVIGATION Iain H. Kalfas Department of Neurological Surgery, The Cleveland Clinic, Cleveland, Ohio, U.S.A.

Image-guided spinal navigation is a computer-based surgical technology that was developed to improve intraoperative orientation to the unexposed anatomy during complex spinal procedures (1,2). It evolved from the principles of stereotaxy, which have been used by neurosurgeons for several decades to help localize intracranial lesions. Stereotaxy is defined as the localization of a specific point in space using three-dimensional (3-D) coordinates. The application of stereotaxy to intracranial surgery initially involved the use of an external frame attached to the patient’s head. However, the evolution of computer-based technologies has eliminated the need for a frame and has allowed for the expansion of stereotactic technology into other surgical fields such as spinal surgery. The management of complex spinal disorders has been greatly influenced by the increased acceptance and use of spinal instrumentation devices as well as the development of more complex operative exposures. Many of these techniques place a greater demand on the spinal surgeon by requiring a precise orientation to that part of the anatomy that is not exposed in the surgical field. In particular, the various fixation techniques that require placing bone screws into the pedicles of the thoracic, lumbar, and sacral spine and into the lateral masses of the cervical spine and across joint spaces in the upper cervical spine require ‘‘visualization’’ of the unexposed spinal anatomy. Although conventional intraoperative imaging techniques such as fluoroscopy have proven useful, they are limited in that they provide only two-dimensional (2-D) imaging of a complex three-dimensional structure. Consequently, the surgeon is required to extrapolate the third dimension based on an interpretation of the images and a knowledge of the pertinent anatomy. This so-called ‘‘dead reckoning’’ of the anatomy can result in varying degrees of inaccuracy when placing screws into the unexposed spinal column. Several studies have shown the unreliability of routine radiography in assessing pedicle screw placement in the lumbosacral spine. The rate of penetration of the pedicle cortex by an inserted screw ranges from 21% to 31% in these studies (3–5). The disadvantage of these conventional radiographic techniques in orienting the spinal surgeon to the unexposed spinal anatomy is that they display, at most, only two planar images. While the lateral view can be relatively easy to assess, the anterioposterior or oblique view can be difficult to interpret. For most screw

Icons indicate materials available on the accompanying CD.

235

236

&

PART III: CLINICAL APPLICATIONS

fixation procedures, it is the position of the screw in the axial plane that is most important. This plane best demonstrates the position of the screw relative to the neural canal. Conventional intraoperative imaging cannot provide this view. To assess the potential advantage of axial imaging for screw placement, Steinmann et al. used an image-based technique for pedicle screw placement that combined computed tomography (CT) axial images of cadaver spine specimens with fluoroscopy. This study demonstrated an improvement in pedicle screw insertion accuracy with an error rate of only 5.5% (6). Image-guided spinal navigation minimizes much of the ‘‘guess work’’ associated with complex spinal surgery. It allows for the intraoperative manipulation of multiplanar computed tomographic (CT) images that can be oriented to any selected point in the surgical field. Although it is not an intraoperative imaging device, it provides the spinal surgeon with superior image data compared to conventional intraoperative imaging technology (i.e., fluoroscopy). It improves the speed, accuracy, and precision of complex spinal surgery, while in most cases it eliminates the need for cumbersome intraoperative fluoroscopy.

PRINCIPLES OF IMAGE-GUIDED SPINAL NAVIGATION The use of an image-guided navigational system for localizing intracranial lesions has been previously described (7,8). Image-guided navigation establishes a spatial relationship between a preoperative CT image data and its corresponding intraoperative anatomy. Both the CT image data and the anatomy can each be viewed as a 3-D coordinate system with each point in that system having a specific x, y, and z Cartesian coordinate. Using defined mathematical algorithms, a specific point in the image data set can be ‘‘matched’’ to its corresponding point in the surgical field. This process is called registration and represents the critical step of image-guided navigation. A minimum of three points needs to be matched, or registered, to allow for accurate navigation. A variety of navigational systems have evolved over the past decade. The common components of most of these systems include an image-processing computer workstation interfaced with a two-camera optical localizer (Fig. 1). When positioned during surgery, the optical localizer emits infrared light towards the operative field. A hand-held navigational probe mounted with a fixed array of passive reflective spheres serves as the link between the surgeon and the computer workstation (Fig. 2). Alternatively, passive reflectors may be attached to standard surgical instruments. The spacing and positioning of the passive reflectors on each navigational probe or customized trackable surgical instrument is known by the computer workstation. The infrared light that is transmitted towards the operative field is reflected back to the optical localizer by the passive reflectors. This information is then relayed to the computer workstation, which can then calculate the precise location of the instrument tip in the surgical field, as well as the location of the anatomic point on which the instrument tip is resting. The initial application of navigational principles to spinal surgery was not intuitive. Early navigational technology applied to intracranial surgery used an

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

Figure 1 Image-guided navigational workstation with infrared camera localizer system.

Figure 2 Navigation probe and drill guide for spinal surgery.

&

237

238

&

PART III: CLINICAL APPLICATIONS

external frame mounted to the patient’s head to provide a point of reference to link preoperative image data to intracranial anatomy. This was not practical for spinal surgery. The current generation of intracranial navigational technology uses reference markers or fiducials that are attached to the patient’s scalp prior to imaging. However, the use of these surface-mounted fiducials for spinal navigation is not practical because of accuracy issues related to a greater degree of skin movement over the spinal column (9,10). This is less of a problem with intracranial applications because of the relatively fixed position of the overlying scalp to the attached fiducials. The application of navigational technology to spinal surgery involves using the rigid spinal anatomy itself as a frame of reference. Bone landmarks on the exposed surface of the spinal column provide the points of reference necessary for image-guided navigation. Specifically, any anatomic landmark that can be identified intraoperatively, as well as in the preoperative image data set can be used as a reference point. The tip of a spinous or transverse process, a facet joint, or a prominent osteophyte can all serve as potential reference points (Fig. 3). Since each

Figure 3 Navigational workstation screen demonstrating a paired point registration plan for the insertion of T12 pedicle screws. Three discreet bony landmarks are selected at the T12 level. In this case the lateral margins of the two T12 transverse processes and the tip of the T12 spinous process have been selected.

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

239

vertebra is a fixed, rigid body, the spatial relationship of the selected registration points to the vertebral anatomy at a single spinal level is not affected by changes in body position. Two different registration techniques can be used for spinal navigation: paired point registration and surface matching. Paired point registration involves selecting a series of corresponding points in a CT or magnetic resonance imaging (MRI) data set and in the exposed spinal anatomy. The registration process is performed immediately after surgical exposure and prior to any planned decompressive procedure. This allows for the use of the spinous processes as registration points. A specific registration point in the CT image data set is selected by highlighting it with the computer cursor. The tip of the probe is then placed on the corresponding point in the surgical field and the reflective spheres on the probe handle are aimed towards the camera. Infrared light from the camera is reflected back allowing the spatial position of the probe’s tip to be identified. This initial step of the registration process effectively ‘‘links’’ the point selected in the image data with the point selected in the surgical field. When a minimum of three such points are registered, the probe can be placed on any other point in the surgical field and the corresponding point in the image data set will be identified on the computer workstation. Alternatively, a second registration technique called surface matching can be used. This technique involves selecting multiple nondiscreet points only on the exposed and debrided surface of the spine in the surgical field. This technique does not require the preselection of points in the image set, although several discreet points in both the image data set and in the surgical field are frequently required to improve the accuracy of surface mapping. The positional information of these points is transferred to the workstation and a topographic map of the selected anatomy is created and ‘‘matched’’ to the patient’s image set (11). Typically, paired point registration can be done more quickly than surface mapping. The average time needed for paired point registration is 10 to 15 seconds. The time needed for surface mapping is much longer with difficult cases requiring as much as 10 to 15 minutes. With the need to perform several registration processes during each surgery, this time difference can significantly impact the length of the navigational procedure and the surgery itself (12). The purpose of the registration process is to establish a precise spatial relationship between the image space of the data with the physical space of the patient’s corresponding surgical anatomy. If the patient is moved after registration, this spatial relationship is distorted making the navigational information inaccurate. This problem can be minimized by the optional use of a spinal tracking device, which consists of a separate set of passive reflectors mounted on an instrument that can be attached to the exposed spinal anatomy (Fig. 4). The position of the reference frame can be tracked by the camera system. Movement of the frame alerts the navigational system to any inadvertent movement of the spine. The system can then make correctional steps to keep the registration process accurate and eliminate the need to repeat the registration process. The disadvantage of using a tracking device is the added time needed for its attachment to the spine, the need to maintain a line of sight between the device and the camera, and the inconvenience of having to perform the procedure with the device placed in the

240

&

PART III: CLINICAL APPLICATIONS

Figure 4 Reference frame attached to a spinous process in the surgical field. The reference frame monitors inadvertent movement of the spinal anatomy that may affect navigational accuracy.

surgical field. It is particularly cumbersome when image-guided navigation is used during cervical procedures. Alternatively, image-guided spinal navigation can be performed without a tracking device (1,12). This involves acknowledging the effect of patient movement on the accuracy of image-guided navigation and maintaining reasonably stable patient position during the relatively short amount of time needed (i.e., 10–20 seconds) for the selection of each appropriate screw trajectory. Patient movement can potentially occur with respiration, from the surgical team leaning on the table, or from a change of table position. Movement associated with patient respiration is negligible and does not require any tracking even in the thoracic spine. Although movement associated with leaning on the table or repositioning the table or the patient will affect registration accuracy, it can be easily avoided during the short navigational procedure. If inadvertent patient movement does occur, the registration process can be repeated. Repeating the registration process is easiest when using the shorter paired point technique, as opposed to the more time-consuming surface mapping technique. When the registration process has been completed, the probe can be positioned on any surface point in the surgical field and three separate reformatted CT images centered on the corresponding point in the image data set are

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

241

immediately presented on the workstation monitor. Each reformatted image is referenced to the long axis of the probe. If the probe is placed on the spinal anatomy directly perpendicular to its long axis, the three images will be in the sagittal, coronal, and axial planes. A trajectory line representing the orientation of the long axis of the probe will overlay the sagittal and axial planes. A cursor representing a cross-section through the selected trajectory will overlay the coronal plane. The insertional ‘‘depth’’ of the trajectory can be adjusted to correspond to selected screw lengths. As the depth is adjusted, the specific coronal plane will also adjust accordingly with the position of the cursor demonstrating the final position of the tip of a screw placed at that depth along the selected trajectory. As the probe is moved to another point in the surgical field, the reformatted images, as well as the position of the cursor and trajectory line, will also change. The planar orientation of the three reformatted images will also change as the probe’s angle, relative to the spinal axis, changes. When the probe’s orientation is not perpendicular to the long axis of the spine, the images displayed will be in oblique, or orthogonal planes. Regardless of the probe’s orientation, the navigational workstation will provide the surgeon with a greater degree of anatomic information than can be provided by any intraoperative imaging technique. The application of image-guided navigation to spinal surgery is directed by the complexity of the procedure and, specifically, by the need to ‘‘visualize’’ the unexposed spinal anatomy. Image-guided navigation can be used with or without standard intraoperative imaging techniques (i.e., fluoroscopy). In either case, image-guided navigation provides the surgeon with an improved orientation to the pertinent spinal anatomy, which subsequently facilitates the accuracy and effectiveness of the procedure. CLINICAL APPLICATIONS Image-guided spinal navigation was initially evaluated for the insertion of pedicle screws in the thoracic and lumbosacral spines of cadaver specimens. The accuracy of screw insertion was documented by plain film radiography and thin section CT imaging of the instrumented levels. All inserted pedicle screws were satisfactorily placed (2). The clinical application of image-guided spinal navigation began with its use in lumbosacral pedicle fixation (1,13,14). Other spinal applications gradually evolved including transoral decompression, cervical screw fixation, thoracic pedicle fixation, anterior thoracolumbar decompression and fixation procedures, and application to spinal metastasis (12,15–20).

Pedicle Fixation Pedicle fixation has gained acceptance as an effective and reliable method of spinal stabilization. However, because of the variations of pedicle anatomy within each patient, the safe and precise placement of pedicle screws can be difficult. Suboptimal screw placement can result in varying degrees of neural injury and fixation

242

&

PART III: CLINICAL APPLICATIONS

failure. These complications can be minimized if the surgeon is provided with accurate spatial orientation to each pedicle to be instrumented prior to screw insertion. Image-guided spinal navigation can now be used routinely in place of fluoroscopy for the insertion of pedicle screws in both the thoracic and lumbosacral spine. Although fluoroscopy provides real-time imaging of the spinal anatomy, the views generated represent only 2-D images of a complex 3-D structure. Manipulation of the fluoroscopic unit can reduce this problem but these maneuvers can be cumbersome and time-consuming. Other disadvantages include the radiation exposure and the need to wear lead aprons during the procedure. Fluoroscopy cannot provide a view of the spinal anatomy in the axial plane. It is this axial view provided by image-guided navigation that makes it superior to fluoroscopy for spinal screw fixation procedures. The application of image-guided navigation to the spine involves obtaining a preoperative CT scan through the appropriate spinal segments to be instrumented. The images consist of a 3-D volume data set of contiguous axial CT images. Alternatively, MRI data may also be used. The image data are then transferred to the computer workstation via optical disc or a high-speed data link. If paired point registration is to be used, three to five reference points for each spinal segment to be instrumented are selected and stored in the image data set. Intraoperatively, a standard exposure of the spinal levels to be instrumented is performed. A lateral radiograph can be obtained to confirm the appropriate level. The computer workstation and camera localizer are then positioned. The infrared camera detector is mounted at the foot of the table and aimed rostrally for thoracic and lumbosacral procedures. Image-guided navigation is typically used prior to any planned decompression in order to utilize the intact posterior elements as registration points. The first spinal segment to be instrumented is registered using either the paired point or surface mapping technique. When the registration process has been completed, the navigational workstation will calculate and display a registration error (expressed in millimeters) that is directly dependent on the surgeon’s registration technique. The error presented does not represent a linear error but rather a volumetric calculation comparing the spacing of registration points in the surgical field to the spacing of the corresponding points in the image data set. This figure is, at best, a relative indicator of accuracy. A better method of assuring registration accuracy is the verification step. This step is typically performed immediately after completing either registration process. The surgeon places the navigational probe on a discreet landmark in the surgical field. With the navigational system now tracking the movement and position of the probe, the trajectory line and cursor on the workstation screen will, if accurate registration has been achieved, move to the corresponding point in the image data set. If registration accuracy has not been achieved, the cursor and trajectory line may rest on a point other than that selected in the surgical field. If this occurs to a significant degree, the registration process needs to be repeated. This step is more of an absolute indicator of registration accuracy and is important to perform prior to proceeding with navigation. When an accurate registration of the first spinal level to be instrumented has been verified, standard bony landmarks for pedicle localization are used to

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

243

approximate the screw entry point. A drill guide is placed on this entry point and the navigation probe passed through the guide. The navigational system is activated permitting tracking of the probe in the surgical field. Three separate reformatted views are displayed on the workstation screen. Each view represents a separate plane passing through the selected point in the surgical field. For most pedicle fixation cases, these views typically consist of a sagittal, an axial, and a coronal reconstruction. A trajectory line, referenced to the long axis of the probe, is superimposed on the sagittal and axial views. A round cursor, representing a cross section through the selected trajectory is superimposed on the coronal view. As the probe is moved through the surgical field, the position of the trajectory line and cursor will change accordingly. Both the width of the trajectory line and the diameter of the cursor can be adjusted to match the relative diameter of the pedicle screws to be used. The length of the trajectory line can also be adjusted (Fig. 5). As the probe is placed on each pedicle entry point, the images on the workstation screen are presented in real time. As the angle of the probe is adjusted in both the axial and sagittal planes, the images will update to show the corresponding trajectories. The depth of the coronal view can be adjusted to show the cross-sectional anatomy at any point along the selected trajectory. The orientation of each pedicle to be instrumented can be assessed rapidly and accurately. Any

Figure 5 Workstation screen demonstrating navigation for an L3 pedicle screw.

244

&

PART III: CLINICAL APPLICATIONS

errors in trajectory or entry point selection can be determined and corrected by adjusting the position of the probe and the drill guide through which it passes. When a satisfactory screw entry point and trajectory have been selected, the current images on the workstation screen are frozen, the probe is removed from the drill guide, and a drill (3 mm diameter) is positioned through the guide. The purpose of the drill guide is to preserve the physical trajectory and entry point information just acquired through the navigation of that pedicle. If a drill guide is not used, it may be difficult to precisely position a drill or pedicle probe on the same point and with the same trajectory as the removed navigational probe. The use of a drill guide also permits using a hand-held drill to place the small pilot hole as opposed to trying to recreate the navigational information with a blunt and awkward pedicle probe. When the pilot hole is placed, a sound can be passed down the hole to assure adequate positioning. Navigation is then performed for the contralateral pedicle. For each additional vertebra to be instrumented, a new set of registration points at that level is selected. This method, termed segmental registration, eliminates any potential discrepancy in anatomic orientation that may be related to a change in the patient position between the preoperative CT scan and surgery. Since each vertebra is a fixed, rigid body, the spatial relationship of the selected registration points to the vertebral anatomy at a single spinal level is not affected by changes in body position. After all pilot holes have been drilled, they are tapped and the appropriate size screws inserted. C-arm fluoroscopy or serial radiographs are not required. Typically, the average time for registration, navigation, and the insertion of four pedicle screws is approximately 8 to 10 minutes when using a paired point registration technique. This figure is considerably higher when using a surface mapping technique due to the greater time it takes to achieve adequate registration with surface mapping. In addition to screw placement in the large pedicles of the lumbosacral spine, image-guided navigation can also facilitate screw placement into the smaller pedicles of the thoracic spine (Fig. 6). The added precision for screw placement into thoracic pedicles greatly expands the fixation options for managing the unstable thoracic spine and cervicothoracic junction. Image-guided navigation can also be used in place of fluoroscopy for placement of interbody cages in the lumbosacral spine. During removal of the intervertebral disc, the navigational probe can be inserted into the evacuated disc space. With the trajectory length set at zero, the three reformatted images displayed provide optimal spatial orientation to the disc space allowing for precise placement of the cages (Fig 7A and B). C1–C2 Transarticular Screw Fixation This procedure involves the passage of a screw through the pars interarticularis of C2, across the facet joint, and into the lateral mass of C1. The risks of screw insertion include injury to the vertebral artery if the screw is placed too laterally or ventrally, injury to the spinal cord if the screw is placed too medially, and failure to engage the

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

245

Figure 6 Workstation screen demonstrating navigation for a T8 pedicle screw in a patient with mycotic aneurysm of the aorta.

lateral mass of C1 if the screw trajectory is too ventral. The insertion of a screw on either side may be contraindicated if the pars interarticularis of C2 is too narrow. The procedure is typically performed bilaterally using fluoroscopic guidance. The selection of the appropriate screw entry site and trajectory requires a thorough understanding of the atlanto-axial anatomy. Although fluoroscopy provides real-time imaging of the relevant spinal anatomy, the views generated represent only 2-D images of a complex 3-D anatomic region. Manipulation of the fluoroscopic unit can reduce this problem but these maneuvers can be cumbersome and time-consuming. Other disadvantages include the radiation exposure and the need to wear lead aprons during the procedure. Fluoroscopy cannot provide a view of the spinal anatomy in the axial plane. It is this axial view, provided by image-guided navigation, that makes it superior to fluoroscopy for spinal screw fixation procedures. The application of image-guided navigation to this procedure adds a significant layer of accuracy for proper screw placement. The technique for applying image-guided navigation to posterior C1–C2 screw fixation involves acquiring a preoperative CT scan that extends from the lower occipital region to C3. The image data are transferred to the computer workstation and can be used to create a preoperative screw trajectory plan. A

246

&

PART III: CLINICAL APPLICATIONS

Figure 7 (A) Workstation screen prior to L5-S1 disc excision for a posterior interbody fusion. (B) Workstation screen after L5-S1 disc excision. The depth within the disc space and the extent of disc removal can be determined prior to cage or bone graft insertion. Fluoroscopy is not needed.

proposed entry point and target can be selected at the C2 and C1 levels, respectively. The image data set can then be manipulated in multiple planes between these two points to demonstrate the position of a screw placed along the selected trajectory. In addition to a sagittal image that demonstrates the same information provided by lateral fluoroscopy, two other images are presented. One of the images lies perpendicular to the sagittal image along the selected trajectory. It represents an orthogonal view that lies approximately midway between the coronal and axial planes through the spine. It demonstrates a second view of the selected trajectory. An additional view demonstrates an image-oriented perpendicular to the long axis of the probe and, therefore, the selected trajectory. A cursor superimposed on this image can show the position of the screw tip along the selected trajectory at millimetric increments. By scrolling through this image, the proposed position of the screw along the selected trajectory can be assessed along its entire path. While this planning technique does not assure safe screw placement intraoperatively, it can preoperatively alert the surgeon to avoid screw placement in patients with insufficient anatomy and to select an alternate approach. Intraoperatively, the patient is positioned and the posterior C1–C2 complex is exposed. A wire (cable) and bone graft stabilization procedure at the C1–C2 level is performed prior to navigation and screw insertion. Performing this step first minimizes any independent motion between C1 and C2 during navigation and makes tap and screw insertion easier. If a reference frame is used, it is typically attached to the spinous process of C2. Following placement of the graft and cable, three to five registration points are selected at the C2 level. It is not necessary to include registration points at C1. Although the spatial relationship of C1 and C2 may change between the

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

247

preoperative scanned position and the intraoperative position, the ability of imageguided navigation to facilitate accurate screw placement is not significantly affected. The technical difficulty of this procedure is the accurate passage of the screw through the narrow pars interarticularis of C2. The lateral mass of C1 is a relatively large target that can be easily reached, provided there is a reasonably acceptable realignment of C1 and C2 as well as an optimal positioning of the screw within the appropriate C2 anatomy. While the relative position of C1 and C2 in both the preoperative image set and in the surgical field is important, it is not critical enough to interfere with the process of image-guided navigation. Two separate stab incisions are made on either side of the midline at the C7-T1 level. A drill guide is placed through one of the stab incisions, passed through the paravertebral musculature and into the operative field. A small divot is drilled at the proposed entry site in order to provide for secure placement of the drill guide. The registration process is performed at the C2 level and its accuracy confirmed using the verification step. The probe is passed through the drill guide and, as its position is adjusted in the surgical field; the images on the workstation screen will adjust accordingly to show the corresponding trajectory in two separate planes and the projected location of the screw tip in the third plane. Orientation to the correct screw position can be assessed rapidly and accurately (Fig. 8). Any errors in trajectory or entry point selection can be determined and corrected by adjusting the position of the probe and the drill guide through which it passes. When the correct screw insertion parameters have been selected, the probe is removed from the drill guide and a drill inserted. A hole is drilled along the selected trajectory, tapped, and the appropriate length screw inserted. The process is repeated on the opposite side. The purpose of the drill guide is to preserve the physical trajectory and entry point information just acquired through the navigation of that pedicle. If a drill guide is not used, it may be difficult to precisely position a drill or pedicle probe on the same point and with the same trajectory previously conveyed by the navigational probe after probe removal. While image-guided navigation does not guarantee accurate screw placement, it does provide the surgeon with a greater degree of anatomical information than fluoroscopy alone. The addition of fluoroscopy to this navigational technique provides the greatest degree of precision to the procedure. In this case, however, navigational technology significantly reduces the time of intraoperative fluoroscopic usage as it is typically used only to help position the patient preoperatively and as a final check of the selected trajectory in the sagittal plane immediately following the navigational step.

Segmental C1–C2 Screw Fixation As an alternative to transarticular screw fixation, segmental fixation of C1–C2 can be used for managing atlantoaxial instability (21). The procedure involves placing a screw into each of the two lateral masses of C1 and two screws down the pedicles of C2. The polyaxial screw heads on each side are then connected with rods.

248

&

PART III: CLINICAL APPLICATIONS

Figure 8 Workstation screen demonstrating a trajectory for insertion of a C1–C2 transarticular screw. The lower right screen shows the trajectory in the sagittal plane. The lower left screen represents an orthogonal plane lying between the axial and coronal planes. It conveys the medial–lateral trajectory. The upper left screen represents a plane that is perpendicular to the two other images. It demonstrates the location of the screw tip inserted along the selected trajectory at the indicated depth.

Although this approach potentially reduces the risk of injury to the vertebral artery during screw insertion, it does not eliminate the risk altogether. As with the transarticular technique, precise anatomic orientation is required to avoid arterial or neural injury. Image guidance can supplement intraoperative fluoroscopy in order to provide the necessary orientation for accurate screw insertion. As with the transarticular screw fixation technique, a preoperative CT is obtained. The posterior C1–C2 spine is exposed and a wire and cable fixation procedure is carried out. Registration is first performed at C1 for placement of the C1 lateral mass screws. The three registration points typically used at C1 include the midline posterior tubercle and the bilateral points marked by the junction of the small pedicle of C1 with its lateral mass (immediately above the two exiting C2 nerve roots). Once registered, the correct trajectory into the lateral mass can be displayed on the workstation screen and the screws inserted (Fig. 9A and B). To use image guidance for inserting C2 pedicle screws, the same registration points

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

249

Figure 9 Workstation screen demonstrating navigational information for placement of a screw into the (A) lateral mass of C1 and (B) pedicle of C2.

are used at C2 as those used for transarticular fixation (the C2 spinous process and the two lateral margins of the C2–C3 facet). The entry point for the screw is more lateral and the trajectory more medially oriented than for a transarticular screw. The navigation probe is placed through a drill guide onto this entry point and the selected trajectory is displayed on the workstation screen. When the correct entry point and trajectory have been selected, the probe is removed, a drill is inserted, and the pilot hole is drilled. The process is then repeated for the other side. The heads of the screws are then connected with two short rods. Transoral Surgery Transoral decompression of the upper cervical spine typically requires intraoperative fluoroscopy to help maintain proper anatomic orientation during the procedure. Although orientation in the sagittal plane is easy to obtain with fluoroscopy, depth and medial–lateral orientation are more difficult to assess. Image-guided technology can be used to orient the surgeon in multiple planes during transoral surgery (12,19). Unlike other spinal applications of image guidance, discreet registration points are not readily available during transoral surgery. In this setting, surfacemounted markers (fiducials) are applied to the patient prior to obtaining the preoperative CT. Typically, two fiducials are applied to the mastoid processes and two are applied to the lateral orbital margins or to both malar eminences. The nasal septum can also be used as an inherent registration point. The patient is positioned in a three-point head holder. The registration process is performed prior to draping the patient using the surface-mounted fiducials. Because the registration points will not be accessible during the procedure,

250

&

PART III: CLINICAL APPLICATIONS

a reference frame is used for transoral navigation. This allows for changes in patient positioning during surgery without the need to reregister. The reference frame can be attached to the three-point head holder. During the procedure, the probe can be placed into the site of the decompression. Reformatted sagittal, axial, and coronal CT images are immediately generated providing the surgeon with a precise orientation to the pertinent surgical anatomy. In particular, orientation in the axial plane minimizes the risk of lateral deviation towards the vertebral artery during the decompression (Fig. 10). If a posterior fixation is indicated following transoral decompression, the same CT image data set can be used for C1–C2 screw placement. Anterior Thoracolumbar Surgery Image-guided spinal navigation can be applied to anterior thoracolumbar surgery to help orient the surgeon to the extent of anterior decompression and to facilitate the precise placement of fixation screws. Although the selection of reference points for anterior spinal surgery is limited by the relative lack of prominent bony

Figure 10 Workstation screen demonstrating navigational information during transoral decompression.

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

251

landmarks on the anterior aspect of the spinal column, the degree of accuracy required is less than that needed for most posterior screw fixation procedures. This degree of accuracy, termed ‘‘clinically relevant accuracy,’’ will change according to the procedure being performed. It represents the degree of accuracy needed to achieve a particular surgical task. For example, insertion of a C1–C2 transarticular screw has a higher clinically relevant accuracy demand than placing an anterior fixation screw across a large thoracic or lumbar vertebral body. In both cases, image-guided navigation provides clinically relevant accuracy more consistently than fluoroscopy alone. Potential registration points for the use of image-guided navigation in anterior thoracolumbar surgery include selected landmarks on the vertebral endplates, pedicles, head of the rib, and prominent ventral osteophytes. In general, higher registration errors can be tolerated because of the lower accuracy requirements for most anterior thoracolumbar procedures compared to posterior screw fixation procedures. The accuracy verification step performed immediately after registration can further confirm the achievement of clinically relevant accuracy before proceeding with navigation. During anterior decompression, the probe can be placed into the partially decompressed site to orient the surgeon to the contralateral margin of the spinal column and, more importantly, to the location of the epidural space (Fig. 11A). Orientation to tumor margins can also be obtained by placing the probe into the partially decompressed tumor bed. Following decompression, image guidance can be used to guide anterior fixation screws across the vertebrae at either end of the corpectomy site (Fig. 11B).

Figure 11 (A) Workstation screen demonstrating navigation during removal of an L2 metastasis. Orientation to the contralateral side as well as the epidural space can be obtained. (B) Workstation screen demonstrating a selected trajectory for an anterior lumbar fixation screw.

252

&

PART III: CLINICAL APPLICATIONS

Other Spinal Applications Image-guided technology has several other applications in the management of complex spinal disorders. Any procedure in which intraoperative imaging is required to improve a surgeon’s orientation to nonexposed spinal anatomy can benefit from image guidance. Other procedures to which image guidance has been applied include anterior screw fixation for nondisplaced odontoid fractures, cervical lateral mass screw fixation, cervical corpectomy, and the removal of paraspinal neoplasms. The navigational workstation also provides superior image manipulation capabilities. This allows the surgeon to scroll through reformatted CT images in multiple planes providing for optimal preoperative planning, as well as improved intraoperative anatomic assessment.

PITFALLS OF IMAGE-GUIDED SPINAL NAVIGATION While image-guided spinal navigation has proven to be a versatile and effective tool for facilitating complex surgical procedures, it can be prone to several potential problems prior to and during its use. In general, these pitfalls and errors are related to issues of accuracy, technique, and overall ease of use of the technology during surgery. A thorough understanding of these potential problems is required to assure the efficient and effective use of image-guided navigation for spinal surgery. Like any other computer-based technology, image-guided navigation is highly dependent on the quality of the information imported into the system. While obtaining the properly formatted CT images and having them correctly transferred to the navigational workstation is important, the critical step of image guidance is the registration process. If the surgeon takes a casual approach to registration, inaccurate information will be displayed during intraoperative navigation. Another important principle of image guidance is the understanding that the navigational information provided needs to be correlated with the surgeon’s own knowledge of the surgical anatomy and the appropriate screw trajectories through that anatomy. Image-guided navigation is not a replacement for the surgeon knowing the pertinent spinal anatomy and surgical technique. It merely serves to help confirm a surgeon’s estimation of the nonexposed anatomy by providing image information that typically exceeds that of intraoperative fluoroscopy. Image-guided technology also has varying degrees of intraoperative functionality depending on the features of the navigational system used. This translates into an ease of use factor that can either simplify or complicate the overall procedure. Typically, the use of the surface mapping registration technique and a reference frame add time to the navigational procedure, frequently making it longer and more complicated than using fluoroscopy alone. The use of the paired point registration technique, without a reference frame, simplifies the spinal navigation process. Using this approach, the insertion of four pedicle screws typically takes not more than 8 to 10 minutes. By optimizing the ease of use of navigational technology, standard fluoroscopy becomes unnecessary for most spinal screw fixation procedures.

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

253

FLUOROSCOPIC NAVIGATION Fluoroscopic navigation is the combination of standard fluoroscopy with imageguided navigational technology. It was developed to counter the user difficulties of some earlier image-guided systems that typically took much longer to use than standard fluoroscopy (22). Its advantage is that it allows for a reduction in fluoroscopic time during the procedure. With the patient in position prior to surgery, an anteroposterior and lateral fluoroscopic view of the pertinent spinal anatomy is obtained. This is done with a customized reference frame attached to the C-arm or to the patient. This frame serves to superimpose a specific grid on the two images obtained. The navigational workstation can then take the two images and relate the spatial position of the imaged anatomy to a navigational probe. A navigational trajectory line and cursor can then be superimposed on the lateral and anteroposterior images, respectively. As the probe is moved over the exposed spinal anatomy during surgery, the trajectory line and cursor will adjust their position on the stationary fluoroscopic images (Fig. 12).

Figure 12 Workstation screen of a fluoroscopic navigational system. Only the standard anterioposterior and lateral views are provided. Unlike CT-based image-guided navigation, fluoroscopic navigation does not provide the critical axial plane view.

254

&

PART III: CLINICAL APPLICATIONS

The disadvantage of fluoroscopic navigation is that it is still only fluoroscopy. The same difficulties experienced with standard fluoroscopy are present with fluoroscopic navigation. Specifically, only an anteroposterior and lateral images are provided. The critical plane for most spinal screw fixation procedures is the axial plane. This is the only plane that can definitively demonstrate violation of the spinal canal by a medially displaced screw. Only CT-based image-guided navigation can demonstrate this view, although current developments with standard fluoroscopy will eventually allow for axial reconstructions. Furthermore, any region of the spinal column that is difficult to image with standard fluoroscopy (i.e., upper thoracic) will be difficult to image with fluoroscopic navigation. The early goals of image-guided spinal navigation were to improve the surgeon’s orientation to the intraoperative spinal anatomy in a time- and cost-efficient manner and to ultimately replace fluoroscopy. While earlier image-guided systems were difficult to use intraoperatively, several advances have made some systems much easier to use. Several years of clinical experience have helped modify and improve navigational techniques. The use of paired point registration and the optional use of a reference frame have both been found to significantly reduce the difficulties of using image-guided technology for spinal procedures. Advances in computer and localizer technology have also contributed to an improved functionality of these systems. This improved ease of use of the current advanced image-guided systems coupled with superior accuracy, image manipulation, and orientation capabilities provides image-guided technology with a clear advantage over any fluoroscopic-based technology.

CONCLUSION Image-guided navigational technology has been successfully applied to spinal surgery. By linking digitized image data to spinal surface anatomy, image-guided spinal navigation facilitates the surgeon’s orientation to unexposed spinal structures improving the precision and accuracy of the surgery. It is typically used to optimize the placement of spinal fixation screws and to monitor the extent of complex decompressive procedures. It can also be used as a preoperative planning tool. While image-guided spinal navigation is a versatile and effective technology, it is not a replacement for the surgeon having a thorough knowledge of the pertinent spinal anatomy as well as correct surgical techniques. It merely serves as an additional source of information used by the surgeon to make selected intraoperative decisions. In this way, it is similar to more conventional intraoperative imaging techniques (i.e., fluoroscopy) except that it provides a much greater degree of information to the surgeon. Despite the advantages of image guidance, the surgeon must ultimately assess the information provided by these systems and determine if it correlates with the surgeon’s estimation of the nonexposed anatomy and the proposed surgical plan. If good correlation exists between the two, the surgical step can be carried out. However, if sufficient correlation is not present, the surgeon needs to reassess both the spinal anatomy and the image-guided registration accuracy before proceeding.

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

255

Ideally, the clinical application of this technology to spinal surgery should facilitate a reduction in operative time, morbidity, and costs. It should be capable of minimizing or eliminating the need for conventional intraoperative imaging. It should be fast, easy to use, reliable, and capable of being used briefly to provide accurate intraoperative information while minimizing any disruption to the standard routine of each surgical procedure. Ultimately, beyond each individual surgical application, image-guided navigation technology needs to be clinically versatile. It is the routine use of this technology by multiple surgical specialties that will drive its continued evolution and development as well as establishing it as a cost-effective surgical tool.

256

&

PART III: CLINICAL APPLICATIONS

REFERENCES 1. Kalfas IH, Kormos DW, Murphy MA, McKenzie RL, Barnett GH, Bell GR, Steiner CP, Trimble MB, Weisenberger JP. Application of frameless stereotaxy to pedicle screw fixation of the spine. J Neurosurg 1995; 83:641–647. 2. Murphy MA, McKenzie RL, Kormos DW, Kalfas IH. Frameless stereotaxis for the insertion of lumbar pedicle screws: a technical note. J Clin Neurosci 1994; 1(4): 257–260. 3. George DC, Krag MH, Johnson CC, Van Hal ME, Haugh LD, Grobler LJ. Hole preparation technique for transpedicle screws: effect on pull-out strength from human cadaveric vertebrae. Spine 1991; 16:181–184. 4. Gertzbein SD, Robbins SE. Accuracy of pedicle screw placement in vivo. Spine 1990; 15:11–14. 5. Weinstein JN, Spratt KF, Spengler D, Brick C, Reid S. Spinal pedicle fixation: reliability and validity of roentgenogram-based assessment and surgical factors on successful screw placement. Spine 1988; 13:1012–1018. 6. Steinmann JC, Herkowitz HO, El-Kommos H, Wesolowski DP. Spinal pedicle fixation: confirmation of an image-based technique for screw placement. Spine 1993; 18:1856–1861. 7. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Use of a frameless, armless stereotactic wand for brain tumor localization with two-dimensional and threedimensional neuroimaging. Neurosurgery 1993; 33:674–678. 8. Barnett GH, Kormos DW, Steiner CP, Weisenberger J. Intraoperative localization using an armless, frameless stereotactic wand. Technical note. J Neurosurg 1993; 78:510–514. 9. Brodwater BK, Roberts DW, Nakajima T, Friets EM, Strohbehn JW. Extracranial application of the frameless stereotactic operating microscope: experience with lumbar spine. Neurosurgery 1993; 32:209–213. 10. Bryant JT, Reid JG, Smith BL, Stevenson JM. A method for determining vertebral body positions in the sagittal plane using skin markers. Spine 1989; 14:258–265. 11. Pellizzari CA, Levin DN, Chen GTY, Chen CT. Image registration based on anatomic surface matching. In: Maciunas RJ, ed. Interactive Image-Guided Neurosurgery. Park Ridge, Illinois: American Association of Neurological Surgeons, 1993:47–62. 12. Kalfas IH. Image-guided spinal navigation. Clin Neurosurg 1999; 46:70–88. 13. Foley KT, Smith MM. Image-guided spine surgery. Neurosurg Clin North Am 1996; 7(2):171–186. 14. Glossop ND, Hu RW, Randle JA. Computer-aided pedicle screw placement using frameless stereotaxis. Spine 1996; 21:2026–2034. 15. Assaker R, Reyns N, Vinchon M, Demondion X, Louis E. Transpedicular screw placement. Image-guided versus lateral-view fluoroscopy: in vitro simulation. Spine 2001; 26:2160–2164. 16. Kalfas IH. Image-guided spinal navigation: application to spinal metastasis. In: Maciunas RJ, ed. Advanced Techniques in Central Nervous System Metastasis. Lebanon: AANS Publications, 1998:245–254.

CHAPTER 15: IMAGE-GUIDED SPINAL NAVIGATION

&

257

17. Kalfas IH. Frameless stereotaxy assisted spinal surgery. In: Renganchary SS, ed. Neurosurgery Operative Color Atlas. Lebanon: AANS Publications, 2000:123–134. 18. Laine T, Lund T, Ylikoski M, Lohikoski J, Schlenzka D. Accuracy of pedicle screw insertion with and without computer assistance: a randomized controlled clinical study in 100 consecutive patients. Eur Spine J 2000; 9(3):235–240. 19. Welch WC, Subach BR, Pollack IF, Jacobs GB. Frameless stereotactic guidance for surgery of the upper cervical spine. Neurosurgery 1997; 40(5):958–964. 20. Youkilis AS, Quint DJ, McGillicuddy JE, Papadopoulos SM. Stereotactic navigation for placement of pedicle screws in the thoracic spine. Neurosurgery 2001; 48(4): 771–778. 21. Harms J, Melcher R. Posterior C1–C2 fusion with polyaxial screw and rod fixation. Spine 2001; 26:2467–2471. 22. Foley KT, Simon DA, Rampersaud YR. Virtual fluoroscopy: computer-assisted fluoroscopic navigation. Spine 2001; 26(4):347–351.

16

APPLICATIONS IN EPILEPSY SURGERY David W. Roberts Dartmouth Medical School, Hanover and Section of Neurosurgery, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, U.S.A.

Terrance M. Darcey Dartmouth Medical School, Hanover, New Hampshire, U.S.A.

The remarkable growth in awareness of the role surgery may play in the management of the patient with medically intractable epilepsy seen in the previous decade has steadily continued among both physicians and patients with epilepsy. A tremendous increase in the number of medical centers with surgical epilepsy programs, and in the number of patients being so evaluated and treated, was seen through the 1990s, in part stimulated by the first and second international symposia held on the ‘‘Surgical Treatment of the Epilepsies’’ in 1986 (1) and 1991 (2), respectively. This growth has also continued, with increased sophistication and maturity characterizing these many programs. Continued advances in the diagnostic and therapeutic technologies, which play prominent roles in the management of these patients, have driven this evolution, and image-guided surgical techniques are among the most important of these (3). The goal of the evaluation of the patient with intractable epilepsy is the identification and localization of the seizure focus; the complementary therapeutic intervention is the eradication of that same focus. These processes require accurate spatial localization in preoperative imaging and physiologic investigations as well as accurate coregistration of the preoperative database with the surgical field. It is an inherently stereotactic task. This overview will look at the role of emerging interactive image-guided techniques first in the preoperative seizure evaluation phase, including intracranial electrode investigation, and then in the planning and execution of the surgical intervention.

259

260

&

PART III: CLINICAL APPLICATIONS

PREOPERATIVE INVESTIGATION Noninvasive Studies The major advances in neuroradiology brought by CT and MR imaging, single photon emission tomography (SPECT), and positron emission tomography (PET) have had as great an impact on epilepsy management as on any other field of neurosurgery (3,4). With the advent of CT scanning, direct assessment of soft tissue volumes as well as potentially epileptogenic lesions became possible, and the improved soft tissue resolution of MRI has been even more revolutionary. Functional MRI’s ability to visualize functionally significant cortical areas in direct relation to anatomy is proving increasingly valuable in the determination of resection limits (5). Further, the imaging of physiologic substrates by PET, SPECT, and electromagnetic source imaging (6,7) is currently providing new, nonstructural information. At our own institution, ictal SPECT in combination with an interictal study has made feasible the workup of MRI-normal epilepsy patients and guided intracranial studies (8,9). The ability to coregister each of these modalities with the others underlies their powerful value in trying to understand what is often a complex pathophysiology. This coregistration exploits the complementary strengths of each study, be it spatial fidelity or bony detail in CT, sulcal delineation on MRI, or relative regional blood flow in SPECT, by combining them into a single multimodal fused image set. To be optimally useful to the neurosurgeon, this data set must also be coregistered with the patient in the operating room. Neuroradiologists and neurosurgeons have found common ground in establishing these spatial relationships, as coregistration algorithms such as those designed by Pelizzari and Chen (10) for fitting CT and MR scans are also applicable to matching imaging studies and scalp contours intraoperatively. Video-EEG recording of actual seizure events, a central piece of the investigation of the seizure patient, yields an enormous amount of potentially valuable information for the central database as well. High-resolution digital EEG and/ or MEG may be considered an imaging modality, especially when coregistered and visualized along with anatomical and functional images, and display of their information to the investigator is an area of continued development. There has been an evolution of techniques from schematic graphical presentations (e.g., BEAM) to spatially accurate representations and superimposed depictions of electrode or sensor locations and field maps of epileptiform spikes, seizures, and evoked potentials. With accurate electrode localization and anatomical detail, finite element modeling techniques may be applied to quantitatively estimate the sources of electrical activity in three dimensions. Intracranial Studies—Depth Electrodes In 1986, a survey of surgical epilepsy centers at the first Palm Desert Symposium showed that centers were fairly evenly divided between those inserting multicontact depth electrodes by freehand techniques and those using stereotactic

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

261

methodology (1). Five years later, a similar survey revealed the great majority of centers using stereotactic technique (2). Driven by both the demand for the additional information that accurately placed depth electrodes can provide and the facilitation of the stereotactic procedure by CT and MRI guidance, intracranial electrodes have enabled successful localization of more difficult seizure disorders and have a well-established role today (11–17). In early analyses, Spencer and colleagues (14,17) demonstrated that a significant number of additional patients could be offered surgical treatment from information made available only by such investigation, and conversely that many patients who otherwise would have been offered surgery could be spared what would ultimately be unsuccessful intervention (14,17). While centers vary in their utilization of invasive recording, most utilize such recording techniques selectively today. Depth electrode implantation by stereotactic technique offers the two advantages of increased accuracy in placement and increased safety. Direct visualization of the amygdala and hippocampus by MRI, coupled with an accuracy of several millimeters using frame-based stereotaxy, has enabled trajectories down the long axis of the temporal lobe with assured implantation of critical mesial structures. Preoperative planning at a graphics workstation allows renderings of proposed trajectories in triplanar display as well as in reformatted planes along the electrode course. In addition to the feasibility of achieving such trajectories, the methodology may be coupled with digital subtraction angiographic information or less invasive CT or MR angiography, to allow visualization of vascular structures to be avoided (18–20).

Intracranial Studies—Subdural Strip and Grid Electrodes As epilepsy is primarily a disease of the cerebral cortex in most patients, an additional method for its evaluation when the results of the noninvasive workup are equivocal is that of subdural electrode recording (12,15,16,21–30). Subdural strip electrodes consist of a linear array of stainless steel or platinum disc-like recording contacts, commonly eight in number, evenly spaced a centimeter apart, embedded in a flexible silastic strip. They are most commonly inserted through a burr hole and cruciate dural opening, with each entry accommodating placement of one or more electrodes. A tail containing the electrode leads is subcutaneously tunneled and brought out through a separate stab incision. These have most commonly been placed over the temporal and frontal convexity, but are often directed subtemporally, subfrontally, interhemispherically, and over the parietal and occipital lobes as well. Following placement, patients are monitored with continuous video-EEG recording in the epilepsy unit until a sufficient number of seizures have been recorded. Studies of the electroencephalographic and clinical value of subdural and depth electrodes have variably concluded that depth electrodes are more accurate (15,26) and, alternatively, that depth and subdural electrodes can be of equal utility (12,16,22–24). Subdural electrodes have the advantages of coverage of the cortical surface, easier placement operatively, and preservation of brain parenchyma

262

&

PART III: CLINICAL APPLICATIONS

integrity. On the negative side, they do not sample medial temporal structures as well as depth electrodes. The most critical factor in the reliability and usefulness of subdural strip electrodes is the accuracy of their placement. The strategy employed by Awad and colleagues (31) is to insert strips beneath the temporal lobe and to the undersurface of the medial temporal lobe structures; if adjacent to these structures as intended, such electrodes will be sensitive to seizure activity arising medially. In patients in whom invasive recording is desired to clarify the extent of the epileptogenic region associated with a subcortical structural lesion, or to help determine which of multiple abnormalities may be responsible for a given seizure disorder, such electrodes may be directed over the pathology. In both the medial temporal and the subcortical lesion instances, accurate placement is essential, and while insertion is in general relatively easy, accurate placement cannot be taken for granted. Towards that end, interactive image-guided techniques may be useful (32). Given the spatial limitations of intracranial recording contacts and the resultant insensitivity of contacts distant to seizure onset, there is a premium on accurate placement. Using a stereotactic frame to guide the insertion of subdural electrodes is not usually done, although it has on occasions been useful in assisting such electrode placement over lesions when it has already been placed on the patient for concurrent depth electrode placement. In such instances, a frame’s probe may serve simply as a pointer on the overlying scalp. Frameless stereotactic techniques, on the other hand, make such insertion guidance reasonable and accessible on a routine basis. At Dartmouth, the SurgiScope stereotactic operating microscope system (Elekta AB, Stockholm) has been regularly used for this purpose. With the patient awake but the head immobilized on a headrest and secured with tape, coregistration with preoperative MRI studies is accomplished using natural fiducials including the tragus, lateral canthi, and nasion. The burr hole entry site and orientation points along the desired trajectory of each subdural strip electrode are then indicated by a laser light directed through the robotically driven operating microscope; the operating microscope itself is not optically used by the surgeon during the procedure. Although this has also been accomplished using the LED hand-held probe that comes with the system, by moving the probe until it matches the planned entry and trajectory reference points, the efficiency of noniterative, direct robotic guidance is particularly apparent in this multitargeting procedure. In the manner described, preoperative electrode array plans have been executed with consistent placement of contacts over specific gyri, representing a clear improvement over our previous unguided technique. In all of our intracranial electrode placements, the actual location of recording contacts is reconstructed from CT imaging following implantation, and thus interpretation of recorded data is not directly dependent on preoperative intentions. Subdural grid electrodes are similar to strip electrodes but consist of a twodimensional array of contacts (rows and columns) embedded in a silastic or similar sheet. Although they are most commonly placed subdurally, they can be placed epidurally either routinely or in special instances of reoperation where subdural dissection may be difficult. They require a craniotomy for placement but sample a larger area than strip electrodes and can more accurately localize a seizure focus within that region. In addition, they provide the opportunity for more complete

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

263

mapping of motor, language, and visual cortex through cortical stimulation techniques (21,33). Subdural grid electrodes, like subdural strip electrodes, are placed without stereotactic guidance and are in most centers currently. In most instances, this approach may be fully adequate provided that very accurate placement is not required and that final electrode locations are determined by postimplant scanning. Ready access to frameless stereotactic guidance, however, has enabled its utilization in grid placement as well. Placement over possibly epileptogenic cortical or subcortical lesions, such as tumors, cavernous angiomas, heterotopias, and schizencephalies, has been unequivocally facilitated by image guidance in our patients. Placement of grids such as to facilitate functional mapping can also be accomplished in this manner. In SurgiScope-assisted subdural grid placement, coregistration is performed following the administration of general anesthesia and securing of the head by three-point pin fixation. The scalp flap and craniotomy are planned with the assistance of the SurgiScope’s LED hand-held probe or with the microscope’s directed laser light. Following elevation of the bone flap and opening of the dura, the robotic microscope is directed to show the surface locations of any pathologic areas identified in the images. Placement and orientation of the grid electrode may be accomplished in one of several ways. The most obvious and simplest manner of placement simply centers the grid over the already demonstrated area of suspected epileptogenicity. Alternatively, the grid’s location may be planned preoperatively, and the proposed location of at least two of the grid’s contacts entered at a workstation prior to surgery. Intraoperatively, with the cortex exposed, the grid is moved over the surface of the brain until the identified contacts are aligned with the laser light or LED probe. Once positioned, the electrode leads from the grid are tunneled subcutaneously and brought out through separate incisions, as with strip electrodes. It has been possible to place such arrays within several millimeters of a preoperative plan, optimizing the ability to achieve successful recordings and to obtain desired functional mapping. CT imaging prior to returning to the video-EEG monitoring unit following surgery is obtained in all implanted patients. The documentation of exact electrode locations is critical, and prior to recent development assessment entailed conventional review of CT (for optimal electrode visualization) or MRI (for optimal neuroanatomical detail) imaging. Automated electrode recognition algorithms and coregistration techniques to enable interactive display of CT-derived electrode locations on a prior high-resolution MRI have facilitated and improved this process and are now part of standard protocol at our institution.

THERAPEUTIC PROCEDURES Lesionectomy While interactive image-guided systems have been used the most worldwide in the resection of tumors, their adaptation to epilepsy surgery has been equally useful and similar in practice (Figs. 1 and 2). Surgical treatment of seizure disorders

264

&

PART III: CLINICAL APPLICATIONS

Figure 1 The position of the operating microscope’s focal point (represented by a large ‘‘X’’) relative to the preoperative MRI can be followed in this dissection down a large schizencephalic cleft.

associated with structural lesions has been highly successful in producing seizurefree outcomes. In the combined series survey generated from the Second Palm Desert International Conference on the Surgical Treatment of Epilepsies, lesionectomy achieved a seizure-free result in 66.6% of 293 cases, with improvement noted in an additional 21.5% (35). These are comparable to seizure-outcome results achieved with temporal lobe resections and considerably better than can be expected with non-lesional extratemporal resections. More recent surgical series have shown even better seizure outcomes (3). In the intractable epilepsy population, the most common structural pathologies have been developmental abnormalities (most commonly neuronal migration disorders), tumors (notably ganglioglioma, astrocytoma, dysembryoplastic neuroepithelial tumor, and oligodendroglioma), vascular malformations (cavernous hemangiomas and arteriovenous malformations), and post-traumatic gliotic scars. The tremendous advances that have taken place in neuroimaging over the past decade, particularly in MRI, have led to a much higher recognition of such pathologies in the seizure patient population. Although the relationship between these structural abnormalities and the actual seizure disorder remains incompletely understood, this relationship has significant implications for surgical intervention. In a retrospective analysis of 47 patients who underwent surgery for epilepsy associated with structural lesions,

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

265

Figure 2 The microscope’s focal point (represented by a large ‘‘X’’) has now been directed to a portion of the deep heterotopia (same patient as in Fig. 1). Source: From Ref. 18.

Awad and colleagues (31) explored this question. Seventeen of 18 patients who had complete lesion resection independent of extent of seizure focus excision became seizure-free; five of six patients who had complete focus resection but incomplete lesion resection had a similar outcome. Only 12 of 23 patients with incomplete resection of both the lesion and the focus achieved seizure control. Statistical analysis of the series showed the importance of completeness of lesion resection, but the study was not prospective (31). Piepgras et al. (36) reported a series of 102 patients with AVMs who had had at least one preoperative seizure and found 83% seizure-free at long-term follow-up (36). Montes et al. (37) recently reported 16 of 18 children with lesional epilepsy free of seizures following lesionectomy alone. Limitations to a structural approach alone have been suggested by other studies. Cascino and colleagues (38) reported their findings in a series of 23 patients with intractable partial epilepsy and foreign-tissue lesions treated by stereotactic lesionectomy, finding 74% of patients had a 90% reduction, but only 56% a Class I outcome. A nonrandomized comparative study from the same center looking at anterior temporal lobectomy versus stereotactic lesionectomy for medically intractable, lesional epilepsy showed less morbidity with the latter but considerably poorer success in achieving a seizure-free outcome (50% vs. 90%) (39). In a provocative retrospective review, Jooma et al. (40) reported three of 16 (18.8%)

266

&

PART III: CLINICAL APPLICATIONS

patients seizure-free following lesionectomy alone, compared with 13 of 14 (92.8%) patients seizure-free with electroencephalographic definition of the seizure focus and wider resection. These clinical studies do not definitively resolve the question of the optimal extent of resection or the means of achieving that resection, but collectively they do illustrate the complexity of epileptogenicity, as it relates to multiple factors including the anatomic lobe involved, the duration of seizures, and the number of preoperative seizures (31,36,38,41,42). One may conclude that there remains an important role for the multidisciplinary epilepsy team in the evaluation and management of these patients and that significantly better surgical outcomes may be achieved when intervention planning considers a broader database than imaging alone. The ability to seamlessly integrate multiple three-dimensional information data sets, including structural imaging, functional imaging, and electrophysiological data, is becoming a necessity for surgical epilepsy centers and will soon be a standard feature of intraoperative image-guiding systems (3,18,43–45).

Anterior Temporal Lobectomy Navigational guidance may be useful at several points during anterior temporal lobe resection. In determination of the lines of cortical resection at the beginning of the procedure and in orientation during removal of medial structures, correlation of the surgical field with imaging studies can improve the accuracy of the resection as well as shorten operative time. Established practice commonly resects 4.5 cm back from the temporal pole, along the middle temporal gyrus, on the language-dominant side, and approximately a centimeter further back on the nondominant side. Language function on the dominant side more anterior than 4.5 cm in some individuals has been demonstrated by Ojemann and others (46), who have consequently cautioned against the use of such arbitrary guidelines in favor of individualized functional mapping. Regardless of whether functional mapping is employed, however, the surgical team typically orients itself to the operative field following dural opening by measuring distances to the temporal pole and middle fossa floor. Given the curvature of both the temporal lobe and of the irregular middle fossa, these measurements can be imprecise and not necessarily concordant with the actual volumetric anatomy of the brain. In this determination, intraoperative guidance systems have enhanced surgical precision (18,19,32) by allowing convenient preoperative measurement of the imaged temporal lobe and identification of the proposed line of resection for subsequent display during surgery. When language or sensorimotor mapping has been performed preoperatively, guidance systems can easily incorporate that information at the time of surgical planning and do not require that the electrodes be left in place. Even if the electrode grid has been previously removed or demonstrated to have shifted during reexposure, the localized information from functional mapping is unaffected, provided that postimplantation scans with electrodes in place are used to plan and execute the surgery. In those instances in which the grid has remained in place (a useful navigational guide in conventional epilepsy surgery), the transfer of that

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

267

grid’s localizing information to the underlying cortex becomes transparent and automatic with the capability of memorizing and returning to selected locations in the surgical field. Removal of medial structures follows the above-described anterolateral resection step in the majority of temporal lobe resections for non lesional epilepsy. The approach to the anterior portion of the temporal horn may be straightforward for those with experience, but directional assistance to that structure may be appreciated by both younger and older surgeons. Whether a hand-held probe or microscope focal point is used as a guide, the temporal horn is visualized as much as any other structure or lesion (Fig. 3). Once the ventricular system has been entered and the hippocampus exposed, further application of a guidance system may continue to be useful. In our experience, this has been demonstrating the superior extent of desired amygdala resection, where natural reference guidelines are less available, and in assuring resection of the desired length of hippocampus. The latter can always be directly measured intraoperatively, but with a guidance method already functioning in the previous stages of the operation, its continued application at this point adds no additional time or effort.

Selective Medial Temporal Resection With recognition of the primary importance of medial temporal structures in the great majority of complex partial epilepsies and with increasing confidence in the ability to localize seizure disorders to those structures in selected patients, there is

Figure 3 This intraoperative display during a left anterior temporal resection demonstrates the position of the focal point (the lighter ‘‘X’’ in a circle) on the anterior hippocampus as well as the optical axis (the lighter oblique line) after the microscope has been robotically driven to this point.

268

&

PART III: CLINICAL APPLICATIONS

considerable interest in the procedure of selective amygdalo-hippocampectomy. Weiser and Yasargil’s (47) original description of the procedure entailed exposure of that anatomy through the anterior sylvian fissure. A number of centers have followed the lead of a more selective resection in certain patients, but have accomplished this through minimization of the anterolateral neocortical resection; a posterior line of resection 3.5 cm from the temporal pole, for example, provides sufficient access to the hippocampus. Alternatively, selective resection of medial temporal structures can be accomplished through a lateral corticotomy without any neocortical resection. Interactive, image-guided techniques can be of considerable assistance in this procedure (45,48,49). Selective amygdalo-hippocampectomy through a lateral corticotomy is performed at our institution in the following manner. The anesthetized patient is placed on the operating room table in the supine position, with the head secured in threepoint fixation, turned approximately 30 to the contralateral side. The microscope’s laser pointer is now directed along the preoperatively planned trajectory, which lies approximately 15 mm anterior to the pinna and is directed at the hippocampus approximately 10 mm posterior to its anterior most extent. The middle temporal gyrus is exposed through a 4 cm linear incision that begins just above the level of the zygoma and a free-bone flap measuring approximately 2.5 cm in greatest diameter. The draped microscope is now brought into the field, and following the laser guide, a 2 cm cortical incision can be made in either the middle temporal gyrus or immediately inferior sulcus. The dissection is directly carried down to the temporal horn, following either the laser guide or the progress of the microscope’s focal point as depicted on the adjacent monitor. A user-selected panel of images, including triplanar and 3-D rendered formats, is displayed on this monitor, and reference to a high magnification coronal image is sometimes especially helpful. After entering the temporal horn, self-retracting narrow brain ribbons are placed, and with the microscope in spherical mode (rotating around the microscope’s maintained focal point on the hippocampus), the exposure of the medial structures is examined. A subpial dissection is begun just lateral to the anterior hippocampus, down to the pia of the parahippocampal gyrus, although alternatively the dissection may be performed down the lateral occipitotemporal (fusiform) gyrus if desired. The dissection is now carried anteriorly and medially around the hippocampus to the region of the uncus, and the amygdala is then dissected. As previously described, guidance along the superior plane of the amygdala resection is available if needed. Directing attention posteriorly, the subpial dissection parallel to the hippocampus is carried back the length of planned hippocampal resection (Fig. 4) by measurement as well as by stereotactic guidance provided by both laser light and monitor. In those instances in which previous evaluation has included depth electrode recording along the length of the hippocampus, the extent of hippocampal resection to assure removal of epileptogenic tissue as well as the relationship between previous electrode contact positions and the exposed surgical field are clear to the surgeon during the procedure. The hippocampus is now transected, and opening of the exposed choroidal fissure begins posteriorly. The vascular pedicles to the hippocampus and parahippocampal gyrus are sequentially identified, coagulated, and divided. Anteriorly, the dissection is carried across the uncinate fasciculus and the fibers

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

269

Figure 4 During preoperative planning for a selective right amygdalo-hippocampectomy, a

of the anterior commissure, arriving at the previously dissected region of the amygdala. The hippocampus and adjacent parahippocampal tissue are delivered en bloc. Closure and postoperative management follow standard practice.

Neocortical Resection Among the most challenging of intractable seizure disorders are those involving non-lesional, extratemporal cortex. Such resections generally follow careful demarcation of the epileptogenic zone by subdural grid and strip electrode recording, and frequently incorporate functional mapping and/or evoked potential information as well. The distribution of these seizure foci and their subsequent resection is individualized to the patient to an even greater degree than in mesial temporal lobe epilepsy. On the parts of primarily ictal video-EEG findings and motor and language mapping information, a preoperative resection plan is developed jointly with the multidisciplinary surgical epilepsy team to achieve the goal of maximally resecting the responsible seizure focus while preserving critical cortex. The actual cortical resection is usually performed at the time of grid electrode removal. The grid itself may serve as a valuable, albeit sometimes inconvenient, guide during resection of underlying cortex, but with the availability of image

270

&

PART III: CLINICAL APPLICATIONS

guidance, the limits of the cortical resection can be visualized directly on the cortical surface. The cortical resections in which such guidance has been most helpful have been those in which the cortical surface is not widely exposed, as it would be on the convexity. In particular, resections of the supplementary motor area and of the subfrontal regions have been facilitated to the greatest degree. The corticectomy itself is carried out in standard fashion once the lines of cortical resection in the surgical field have been determined.

Corpus Callosum Section Callosal section has been performed less frequently since vagal nerve stimulation has been popularized, but it remains a valuable surgical procedure in selected patients (50). The reported clinical experience with navigational assistance during commissurotomy remains anecdotal but positive (51–53). Inadvertently incomplete or (in the instance of partial callosal section) insufficient division of the commissure was not uncommon in the early experience of corpus callosotomy for the treatment of intractable generalized seizure disorders without an identifiable, resectable seizure focus. As more recent analyses have shown a correlation between length of callosal section and seizure outcome, the surgeon’s ability to precisely execute a given length of callosal section has become more important (53). Methodologies to address this have evolved involving radiographic or stereotactic feedback regarding the length of section (53–55), but obtaining an intraoperative film or employing a stereotactic frame may encumber an otherwise straightforward procedure. Currently emerging frameless stereotactic systems, on the other hand, provide this feedback easily. As dissection along the corpus callosum proceeds, the ability to monitor the extent of section either by graphics on an adjacent screen or superposed through the optics of the operating microscope can be a useful adjunct.

Reoperation There is increasing evidence that reoperation can be worthwhile in many patients who respond suboptimally to initial epilepsy surgery (56–59). Re-evaluation and consideration for further surgery in selected individuals is very much appropriate. In this regard, perhaps no surgery for epilepsy is better assisted by navigational assistance than reoperation. All of the potential advantages that interactive image guidance can provide first time surgery apply at reoperation, but in the setting of postsurgical resection cavities, scarring, and adhesions that can obscure normal anatomical landmarks, such assistance can prove invaluable. In addition to the now obvious aid in orientation and location that can assist the surgeon intraoperatively, there is the additional advantage that information from previous evaluations can be integrated into the current workup. If intracranial electrode recording and/or functional mapping have been obtained previously, that information can be incorporated with the most recent

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

271

recordings or imaging. Although some plasticity and reorganization of functional cortex are possible, it is entirely reasonable in most instances not to repeat mapping of speech and motor cortex. Unlike first surgery where an orienting grid electrode may still be present at the time of cortical resection and provide an alternative to navigational assistance, second surgery may be greatly aided by the display of that previous grid on a recent coregistered study. Attention to any invalidating degree of subsequent shift of remaining brain tissue is in this instance clearly appropriate.

FUTURE DEVELOPMENT Few areas of neurosurgery are as complicated physiologically as epilepsy, and the role of image guidance in seizure surgery has not surprisingly become widely appreciated. Ever more sophisticated, complementary localizing techniques are enabling surgical intervention in a larger proportion of patients, with improved outcome prospects, but require computerized databases and coregistered surgery. It is likely that future surgical intervention will benefit from a number of further developments, including accurate noninvasive source localization based on realistic head models derived from individual scan data and electromagnetic source imaging. Stereotactic correlation will benefit from increasingly transparent and accurate noncontact coregistration methodologies, and modeling, prediction, and tracking of brain deformations and shifts during surgery will help maintain validity of that registration. Intraoperative MR imaging will likely be widely available. Highend visualization, including real-time electrophysiology (e.g., electrocorticography) and timely analysis in the operating room, will be commonplace. Multimodal heads-up displays will inform the operating surgeon, whose tools will include, among other advances, refined endoscopic techniques, robotically driven instruments at both the macro- and microlevel, and various ablative devices such as focused radiation and other high-energy sources. Computer-assisted, image-guided surgery already has found a welcome place in the treatment of medically intractable epilepsy, and in the future it will be essential.

272

&

PART III: CLINICAL APPLICATIONS

REFERENCES 1. Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1987. 2. Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993. 3. Shaefi S, Harkness W. Current status of surgery in the management of epilepsy. Epilepsia 2003; 44(suppl 1): 43–47. 4. Scha¨uble B, Cascino GD. Advances in neuroimaging: management of partial epileptic syndromes. Neurosurg Rev 2003; 26:233–246. 5. Nimsky C, Ganslandt O, Fahlbusch R. Functional neuronavigation and intraoperative MRI. Adv Tech Standards Neurosurg 2004; 29:229–263. 6. Assaf BA, Karkar KM, Laxer KD, et al. Ictal magnetoencephalography in temporal and extratemporal lobe epilepsy. Epilepsia 2003; 44:1320–1327. 7. Smith JR, King DW, Park YD, et al. A 10-year experience with magnetic source imaging in the guidance of epilepsy surgery. Stereotact Funct Neurosurg 2003; 80:14–17. 8. Lewis PJ, Siegel AH, Siegel AM, Studholme C, Sojkova J, Roberts DW, Thadani VM, Gilbert KL, Darcey TM, Williamson PD. Does performing image registration and subtraction in ictal brain SPECT help localize neocortical seizures? J Nucl Med 2000; 41:1619–1626. 9. Siegel AM, Jobst BC, Thadani VM, Rhodes CH, Lewis PJ, Roberts DW, Williamson PD. Medically intractable, localization-related epilepsy with normal MRI: presurgical evaluation and surgical outcome in 43 patients. Epilepsia 2001; 42:883–888. 10. Pelizzari CA, Chen GTY. Registration of multiple diagnostic imaging scans using surface fitting. Proceedings of the 9th International Conference on the Use of Computers in Radiation Therapy. North Holland: Elsevier Science Publishers, 1987: 437–440. 11. McCarthy G, Spencer DD, Riker RJ. The stereotaxic placement of depth electrodes in epilepsy. In: Lu¨ders H, ed. Epilepsy Surgery. New York: Raven Press Ltd, 1991:385–393. 12. Smith JR, Flanigin HF, King DW, et al. Analysis of a four-year experience with depth electrodes and a two-year experience with subdural electrodes in the evaluation of ablative seizure surgery candidates. Appl Neurophysiol 1987; 50:380–385. 13. So NK. Depth electrode studies in mesial temporal epilepsy. In: Lu¨ders H, ed. Epilepsy Surgery. New York: Raven Press Ltd, 1991:371–384. 14. Spencer SS. Depth electroencephalography in selection of refractory epilepsy for surgery. Ann Neurol 1981; 9:207–214. 15. Spencer SS. Depth versus subdural electrode studies for unlocalized epilepsy. J Epilepsy 1989; 2:123–127. 16. Spencer SS, Spencer DD, Williamson PD, et al. Combined depth and subdural electrode investigation in uncontrolled epilepsy. Neurology 1990; 40:74–79. 17. Spencer SS, Spencer DD, Williamson PD, et al. The localizing value of depth electroencephalography in 32 patients with refractory epilepsy. Ann Neurol 1982; 12:248–253.

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

273

18. Olivier A, Germano IM, Cukiert A, et al. Frameless stereotaxy for surgery of the epilepsies: preliminary experience: technical note. J Neurosurg 1994; 81:629–633. 19. Olivier A, Lacertee D, Cukiert A, et al. Frameless stereotactic craniotomies in the surgical treatment of epilepsy: preliminary experience in 70 patients. Scientific Program for the 62nd Annual Meeting of the American Association of Neurological Surgeons, San Diego, 1994:352. 20. Watson V, Xei J, Wilson C, et al. MR stereotactic procedure planning with vascular and cortical guidance: comparison and correlation of MRA, 3-D MRI, digital subtraction angiography (DSA), and intracerebral electrodes. Scientific Program for the 62nd Annual Meeting of the American Association of Neurological Surgeons, San Diego, 1994:351. 21. Arroyo S, Lesser RP, Awad IA, et al. Subdural and epidural grids and strips. In: Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993:377–386. 22. Blom S, Flink R, Hetta J, et al. Interictal and ictal activity recorded with subdural electrodes during preoperative evaluation for surgical treatment of epilepsy. J Epilepsy 1989; 2:9–20. 23. Devinsky O, Sato S, Kufta CV, et al. Electroencephalographic studies of simple partial seizures with subdural electrode recordings. Neurology 1989; 39:527–533. 24. Lu¨ders H, Han JF, Lesser RP, et al. Basal temporal subdural electrodes in the evaluation of patients with intractable epilepsy. Epilepsia 1989; 30:131–142. 25. Rosenbaum TJ, Laxer KD, Vessely M, et al. Subdural electrodes for seizure focus determination. Neurosurgery 1986; 19:73–81. 26. van Veelen CWM, Debets RMChr, van Huffelen AC, et al. Combined use of subdural and intracerebral electrodes in preoperative evaluation of epilepsy. Neurosurgery 1990; 26:93–101. 27. Wyler AR. Subdural strip electrodes in surgery of epilepsy. In: Lu¨ders H, ed. Epilepsy Surgery. New York: Raven Press Ltd., 1991:395–398. 28. Wyler AR, Ojemann GA, Lettich E, et al. Subdural strip electrodes for localizing epileptogenic foci. J Neurosurg 1984; 60:1195–1200. 29. Wyler AR, Walker G, Somes G. The morbidity of long-term seizure monitoring using subdural strip electrodes. J Neurosurg 1991; 74:734–737. 30. Wyler AR, Wilkus RJ, Blume WT. Strip electrodes. In: Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993:387–397. 31. Awad IA, Rosenfeld J, Ahl J, et al. Intractable epilepsy and structural lesions of the brain: mapping, resection strategies, and seizure outcome. Epilepsia 1991; 32: 179–186. 32. DeSalles AAF, Hoebel B, Bhenke E, et al. Frameless stereotaxy validation in general neurosurgery. Scientific Program for the 62nd Annual Meeting of the American Association of Neurological Surgeons, San Diego, 1994:357. 33. Lesser, RP, Gordon B, Fisher R, et al. Subdural grid electrodes in surgery of epilepsy. In: Lu¨ders H, ed. Epilepsy Surgery. New York: Raven Press Ltd., 1991:399–408. 34. Uematsu S, Lesser RP, Gordon B, et al. Total resection of perioccipital epileptogenic area with preservation of visual function: the advantage of preoperative subdural grid

274

&

PART III: CLINICAL APPLICATIONS

35.

36. 37. 38.

39.

40.

41. 42. 43. 44.

45.

46.

47. 48.

49.

50.

mapping. Scientific Program for the 62nd Annual Meeting of the American Association of Neurological Surgeons, San Diego, 1994:357. Engel J Jr, Van Ness PC, Rasmussen TB, et al. Outcome with respect to epileptic seizures. In: Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993:609–621. Piepgras DG, Sundt TM Jr, Ragoowansi AT, et al. Seizure outcome in patients with surgically treated cerebral arteriovenous malformations. J Neurosurg 1993; 78:5–11. Montes JL, Rosenblatt B, Farmer J-P, et al. Lesionectomy of MRI detected lesions in children with epilepsy. Pediatr Neurosurg 1995; 22:167–173. Cascino GD, Kelly PJ, Sharbrough FW, et al. Long-term follow-up of stereotactic lesionectomy in partial epilepsy: predictive factors and electroencephalographic results. Epilepsia 1992; 33:639–644. Moore JL, Cascino GD, Trenerry MR. A comparative study of lesion resection with corticectomy with stereotactic lesionectomy in patients with temporal lobe lesional epilepsy. Epilepsia 1992; 33(suppl 3):96. Jooma R, Yeh H-S, Privitera MD, et al. Lesionectomy versus electrophysiologically guided resection for temporal lobe tumors manifesting with complex partial seizures. J Neurosurg 1995; 83:231–236. Cohen DS, Zubay BA, Goodman RR. Seizure outcome after lesionectomy for cavernous malformations. J Neurosurg 1995; 83:237–242. Yeh HS, Tew JM, Gartner M. Seizure control after surgery on cerebral arteriovenous malformations. J Neurosurg 1993; 78:12–18. Harvey AS, Freeman JL, Berkovic SF, et al. Transcallosal resection of hypothalamic hamartomas in patients with intractable epilepsy. Epileptic Disord 2003; 5:257–265. Roberts DW, Darcey TM. The evaluation and image-guided surgical treatment of the patient with a medically intractable seizure disorder. Neurosurg Clin North Am 1996; 7:215–227. Wurm G, Wies W, Schnizer M, Trenker J, Holl K. Advanced surgical approach for selective amygdalohippocampectomy through neuronavigation. Neurosurgery 2000; 46:1377–1382. Ojemann GA, Sutherling WW, Lesser RP, et al. Cortical stimulation. In: Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993: 399–414. Weiser HG, Yasargil MG. Selective amygdalohippocampectomy as surgical treatment of mediobasal limbic epilepsy. Surg Neurol 1984; 17:445–457. Miyagi Y, Shima F, Ishido K, Araki T, Taniwaki Y, Okamoto I, Kamikaseda K. Inferior temporal sulcus approach for amygdalohippocampectomy guided by laser beam of stereotactic navigator. Neurosurgery 2003; 52:1117–1124. Van Roost D, Schaller C, Meyer B, Schramm J. Can neuronavigation contribute to standardization of selective amygdalohippocampectomy? Stereotact Funct Neurosurg 1997; 69:239–242. Roberts DW, Rayport M, Maxwell RE, et al. Corpus callosotomy. In: Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993:519–526.

CHAPTER 16: APPLICATIONS IN EPILEPSY SURGERY

&

275

51. Bucholz RD, Baumann CK. The use of image-guided frameless stereotaxy during epilepsy surgery. Scientific Program for the 62nd Annual Meeting of the American Association of Neurological Surgeons, San Diego, 1994:348. 52. Bucholz RD, Smith KR. A comparison of sonic digitizers versus light emitting diodebased localization. In: Maciunas R, ed. Interactive Image-Guided Neurosurgery. American Association of Neurological Surgeons, 1993:519–526. 53. Roberts DW, Reeves AG, Nordgren RE. The role of posterior callosotomy in patients with suboptimal response to anterior callosotomy. In: Reeves AG, Roberts DW, eds. Epilepsy and the Corpus Callosum. 2nd ed. New York: Plenum Press, 1995:183–190. 54. Awad IA, Wyllie E, Lu¨ders H, et al. Intraoperative determination of the extent of corpus callosotomy for epilepsy: two simple techniques. Neurosurgery 1990; 25:102–106. 55. Roberts DW, Strohbehn JW, Hatch JF, et al. A frameless stereotaxic integration of computerized tomographic imaging and the operating microscope. J Neurosurg 1986; 65:545–549. 56. Awad IA, Nayel MH, Lu¨ders H. Second operation after failure of previous resection for epilepsy. Neurosurgery 1991; 28:510–518. 57. Polkey CE, Awad IA, Tanaka T, et al. The place of reoperation. In: Engel J Jr, ed. Surgical Treatment of the Epilepsies. New York: Raven Press Ltd., 1993:663–667. 58. Siegel AM, Roberts DW, Thadani VM, McInerney J, Jobst BC, Williamson PD. The role of intracranial electrode re-evaluation in epilepsy patients following failed initial invasive monitoring. Epilepsia 2000; 41:571–580. 59. Wyler AR, Hermann B, Richey ET. Results of reoperation for failed epilepsy surgery. J Neurosurg 1989; 71:815–819.

17

SKULL BASE NEUROSURGERY Troy D. Payner and Armond L. Levy Indianapolis Neurosurgical Groups, Indianapolis, Indiana, U.S.A.

Although computer-assisted neurosurgery has much broader applications, skull base lesions remain an important indication. By their nature, skull base lesions require deep exposures that result in limited visibility within and around the target. The analogy of reading a road map through a dime-sized hole still applies. You may know which road you are on but are uncertain about which direction leads to your destination. Frameless surgical navigation provides a broad view of the map in the form of a CT or MRI slice with a marker that says, ‘‘You are here.’’ Experienced skull base surgeons developed their skills without the benefit of this technology. However, the learning curve for residents and other neurosurgeons to successfully operate in the skull base region can be significantly shortened by its use. The development of intraoperative MRI and CT scanning enables the surgeon to update the ‘‘map’’ in the middle of the operation to improve the accuracy of the navigation system. Surgical navigation in skull base surgery can reduce the incidence of complications, increase the extent of resections, and shorten the duration of procedures. The value and versatility of surgical navigation in skull base surgery is that it provides guidance in planning the incision, determining the size of the craniotomy flap, assessing the extent of bone resection at the skull base, and preserving vital neural and vascular structures (1,2). Additionally, surgical navigation orients the surgeon intraoperatively by assessing the extent of the target lesion. Various pathologic targets including bone lesions, tumors, and vascular anomalies can be localized. Lesions within and attached to the bone of the skull base are ideal for surgical navigation because the bone remains fixed in relation to the fiducial landmarks used to calibrate the equipment. This relative fixation maintains the accuracy of surgical navigation systems (SNSs) throughout the operation as compared to intra-axial lesions that are subject to error related to brain shift from retraction, hyperventilation, cerebrospinal fluid drainage, or mannitol administration. Intraoperative imaging allows the surgeon to update the navigation system to ensure accuracy despite these potential shifts. This chapter reviews the indications for and limitations of surgical navigation in skull base surgery. Its use for identifying landmarks, guiding approaches, localizing specific lesions, and estimating the extent of resection are discussed for several types of skull base approaches. The role of intraoperative imaging as it applies to 277

278

&

PART III: CLINICAL APPLICATIONS

skull base lesions and stereotactic navigation is also discussed. Case illustrations are presented from our experience. UTILIZATION OF SURGICAL NAVIGATION Patient Positioning One of the most important and often underemphasized aspects of neurosurgical procedures is positioning of the patient. This is especially true for skull base approaches. Having the head rotated too little or extended too much might be undesirable when approaching a superficial target; such a positioning error is amplified significantly for skull base lesions. Angling the microscope in an awkward position to resect a deep or large lesion can result in surgeon fatigue, incomplete resection, and increased complications. Stereotactic navigation can eliminate this problem by providing a preview of the trajectory to the target. This can be tested with the microscope to ensure that the positioning will allow a comfortable path to the lesion. In some systems, the navigation is integrated with the microscope itself allowing the trajectory to be tested at the same time as the navigation system is calibrated. An experienced surgeon may feel confident in positioning the patient without navigation but the brief additional time required to test the trajectory may be recouped many times over. Placement of the Incision and Craniotomy One of the fundamental functions of surgical navigation, not unique to skull base surgery, is the identification of underlying surface structures. The skin incision must be placed to allow access to the essential surface area of the skull for each approach. Surgical navigation eliminates insufficient exposure of the skull as a result of anatomic variation among patients. This same advantage applies to planning the craniotomy, thus avoiding injury to vital underlying structures while drilling. For example, the bone overlying the transverse sinus is intentionally traversed when performing combined middle and posterior fossa craniotomies. The bone covering the sigmoid sinus is then removed as part of the transpetrosal exposure. Advance knowledge of the precise location of the sinuses can reduce the potential complications associated with their inadvertent injury or sacrifice. Burr holes can be placed to straddle the transverse sinus, enabling unroofing with a punch rather than a drill or enabling separation of the sinus dura from the bone prior to drilling across it (3). The surgeon can adequately expose both the middle and posterior fossa as necessary for each case and improve the safety of this exposure. Approaches to the anterior skull base may require a bifrontal craniotomy with or without opening the frontal sinus. Since there are no reliable surface landmarks to identify the frontal sinus, surgical navigation provides the necessary intraoperative guidance to avoid unintentional opening of the frontal sinus, thus preventing cerebrospinal fluid leak and infection. Additional operative time, which would have been necessary to reconstruct or cranialize the sinus, had it been opened, is eliminated.

CHAPTER 17: SKULL BASE NEUROSURGERY

&

279

These two examples of the utility of surgical navigation in identifying superficial structures support its benefit in preventing potential complications and reducing the operative time. The availability of immediate localization of a surface landmark by radiographic images in three planes is an added assurance for experienced neurosurgeons. This same benefit is amplified for residents and less experienced neurosurgeons.

Identification of Skull Base Landmarks A thorough knowledge of the bony anatomy of the skull is prerequisite, especially for approaches involving the petrous bone and for those to the cavernous sinus, clivus, and anterior foramen magnum. Skull base surgeons who are very knowledgeable about this anatomy, at times, will experience uncertainty. For transpetrosal approaches, an SNS produces reliable images to localize the mastoid air cells, antrum, semicircular canals, petrous carotid artery, internal auditory canal, and petrous apex. It shortens the time needed to obtain adequate exposure by avoiding drill injury of vital bony landmarks and by reducing the amount of drilling necessary, once the desired trajectory to the target lesion is confirmed by the SNS. Improved outcome should result, especially hearing preservation because the vital structures for hearing can be anticipated and visualized before actual exposure. For anterior approaches in a normal skull, structures easily identified by the SNS include the sphenoid and ethmoid sinuses, superior orbital fissure, optic foramen, anterior clinoid, foramen ovale, foramen spinosum, and foramen rotundum. However, aggressive meningiomas and malignant tumors involving the skull base often erode, invade, or distort the normal bony anatomy. In these cases, even the most experienced surgeons will value surgical navigation for localizing vital structures such as the optic nerve and carotid artery. Other useful bony landmarks include the transverse foramen at C1 and the occipital condyle for far lateral approaches to the anterior foramen magnum region. An SNS can also identify the midline, which is an essential orientation in transfacial approaches to the clivus and upper cervical spine. Cadaver studies confirm that surgical navigation can detail specific landmarks that can be localized in various skull base approaches (4,5). A final feature of SNS particularly well-suited to skull base surgery is the ability to reregister intraoperatively utilizing fixed bony landmarks to obtain superior system accuracy (1). Localization of the Boundaries of the Lesion Circumferential visualization of skull base tumors is usually difficult; therefore, internal debulking of larger tumors is usually necessary. However, there is always uncertainty as to how much of the tumor remains and when the other side of the tumor is near. Surgical navigation makes this information immediately available, thus reducing the likelihood of injury to vital structures (e.g., brainstem) that result from debulking beyond the limits of the tumor (1). For example, while resecting

280

&

PART III: CLINICAL APPLICATIONS

Figure 1 An axial MR image after contrast shows the close relationship of the basilar and internal carotid arteries to the interstices of the tumor. Surgical navigation enabled the surgeon to anticipate when these arteries were near prior to direct visualization.

this aggressive pituitary adenoma, it was useful for the surgeon to anticipate the locations of the basilar and both internal carotid arteries, all intimately associated with tumor (Fig. 1). Because circumferential visualization of skull base tumors is usually difficult, surgical navigation can confirm the location of a certain border. For example, during resection of a large tumor at the petrous apex, it may appear that the most inferior limit of the tumor has been reached. However, surgical navigation may show that another lobule of tumor remains and that further dissection is necessary. Likewise, if the inferior extent of the tumor is reached, but is not recognized, additional dissection may risk inadvertent neurologic injury. SNS guidance could prevent excessive retraction and exposure beyond the limits of the tumor. Estimation of the Extent of Resection Most skull base tumors are removed in sections; one portion may be debulked and then its capsule dissected from surrounding structures. Surgical navigation allows identification of these poles of the tumor so that the extent of resection can be assessed. Usually, an inferior or superior pole of the tumor is resected and some normal anatomic structures identified. By informing the surgeon how large that pole of the tumor is, surgical navigation provides an estimate of the remaining lesion. Frequently, a resected pole of a large skull base tumor is far smaller than

CHAPTER 17: SKULL BASE NEUROSURGERY

&

281

anticipated. Some tumors such as epidermoids can have projections into crevices that may go unrecognized. Until recently, surgical navigation could demonstrate only the preoperative status of the tumor; this is usually sufficient to identify projections that may not be recognized intraoperatively. Image updating during surgery will be discussed later in this chapter. SNSs are especially valuable when subtotal resection is planned; for example, relief of brainstem compression in an elderly patient, or preservation of hearing in a patient with neurofibromatosis type 2. In these cases, it is often difficult to estimate how much tumor has been removed. However, an SNS pinpoints the location within the tumor and allows an accurate assessment of the remaining lesion. This accuracy improves the likelihood of sufficient decompression of the tumor and preservation of vital surrounding structures, while a portion of the tumor may be intentionally left unresected.

TYPES OF LESIONS TARGETED Surgical navigation has a variety of indications in neurosurgery, with initially reported cases involving applications for tumors (1,6–8). Since that time many new applications have been developed. Spinal procedures now compose a significant portion of its use. Among cranial applications in our practice, tumors remain the most common indication for SNS. However, meningiomas are no longer the predominant tumor. Astrocytomas are now the most common tumor, which is consistent with other published reports (7). Metastases and meningiomas are the next most common indication. Functional procedures for the treatment of Parkinson’s disease and other movement disorders are the fastest growing cranial use of SNS in our practice. A breakdown of our utilization of surgical navigation for the most recent 155 cranial cases is given in Table 1 . Table 1 Types of Lesions Among 155 Consecutive Cases Using Surgical Navigation Lesion Number Astrocytoma Meningioma Metastasis Functional Shunt Arteriovenous malformation Abscess Cavernous malformation Epilepsy Hemangioblastoma Intracerebral hemorrhage Total

52 37 34 21 3 3 1 1 1 1 1 155

282

&

PART III: CLINICAL APPLICATIONS

Skull base lesions are uniquely suited for SNS because most of them are intimately associated with bony structures that remain in a fixed position. The majority of these tumors are meningiomas; chordomas, acoustic and trigeminal neuromas, esthesioneuroblastomas, squamous cell carcinomas, and adenocystic carcinomas comprise most of the remainder. The strongest advantage of SNS in these lesions, as opposed to tumors with a glial origin, is their attachment to the immovable skull base (2,9). There is less error associated with the use of surgical navigation for lesions secured to bone than those within the brain substance that can ‘‘move’’ during resection, as well as with the use of hyperventilation, diuretics, and cerebrospinal fluid drainage. The accuracy is reliant on preoperative calibration using fiducial markers. As expected, the spatial orientation of these markers reflects the position of targets fixed to bone better than those attached to brain. The advent of intraoperative imaging to ‘‘update’’ the SNS after partial resection can overcome this problem but is less necessary for skull base lesions (2). TECHNICAL CONSIDERATIONS Preoperative Selection of an Imaging Modality Details of the procedure necessary for fiducial placement and preoperative imaging as well as image transfer to the computer are discussed elsewhere in this book. However, in skull base surgery, choosing the appropriate imaging modality (e.g., CT or MRI) is essential to obtain maximal benefit from the technology. During the initial phase of the operation in which the skull base is exposed, bony anatomy is most critical, and CT provides the most relevant images. However, during tumor resection, MRI is generally more useful. Careful review of the preoperative CT and MR images is necessary to determine which will be most useful intraoperatively. If the surgeon is confident in performing the exposure without optimal bony detail provided by SNSs, then MR images can be used to guide the resection of the tumor. If the tumor can be adequately seen on CT images, then this modality can be sufficient for exposure and resection. Current technology allows changes in the windowing of the images during the operation without recalibration. This allows use of a bone window to guide the exposure and a soft tissue window to guide tumor resection. CT and MR images can also be superimposed on each other using image fusion to provide the optimal detail for both exposure and resection of skull base lesions (2). Positioning the Equipment The use of SNSs in skull base surgery is certainly an asset; however, this must not be at the expense of sterile technique and an unobstructed operative field. Most current SNSs are camera-based systems requiring a reference frame near the targeted lesion as well as a clear line of sight. This can be problematic for skull base approaches in which multiple pieces of equipment and surgical personnel already surround the patient. Positioning the equipment in a functional manner is essential

CHAPTER 17: SKULL BASE NEUROSURGERY

&

283

for making effective use of it. The surgeon, surgeon’s assistant, scrub nurse, microscope, retractor system, and drapes must not interfere with the path between the camera, the reference frame, and the pointer. We have found that the cameras are best placed at the foot of the patient for procedures performed in the supine position. The computer monitor is placed to the right and slightly behind the surgeon. This enables proximity for ease of visualization but does not interfere with the scrub nurse or surgical assistant. The microscope should be based to the left of the surgeon (Fig. 2). Posterior fossa approaches done in the lateral or park bench position require more creative positioning and differ depending on whether it is a left-sided or right-sided approach. The cameras are now at the head of the patient, and the reference frame is placed toward the vertex of the head because the shoulder interferes with placing it caudally. The computer monitor is placed across the patient

Figure 2 Operative setup, supine position. The detector/camera array is at the foot of the bed; the reference frame is adjacent to the patient’s chin. The microscope is to the surgeon’s left, and the SNS monitor is to his right.

284

&

PART III: CLINICAL APPLICATIONS

for a right-sided approach and behind the surgeon for a left-sided approach. The microscope has a greater potential to interfere with the line of sight between the cameras and the reference frame (Fig. 3). There are two common types of dynamic references frames (DRFs): one is a semicircle generally mounted to surround the patient’s head and the other is a smaller ‘‘V’’ or ‘‘X’’ shaped frame, which can be mounted to the side of the patient’s head or head holder. The side-mounted DRF may be slightly less accurate but it is preferred in skull base surgery especially if the Budde Halo2 (OMI, Cincinnati, Ohio, U.S.A.) is used as a retractor. The surgeon must maintain awareness of this DRF because if it is bumped or leaned upon the accuracy of registration will decline. In our experience, semicircular DRF may be more accurate but it is more likely to interfere with retractors and instruments passing in and out of the field. Sterility of the DRFs can be handled in two ways. A sterile drape can be used to cover the reference frame or the frame itself can be sterilized and reattached after the patient is draped. To drape the reference frame, a clear drape must be used, but there is a small risk of decreasing the accuracy due to refraction of light

Figure 3 Operative setup, lateral position, right-sided approach. The camera array moves to the head of the bed. The reference frame is at the patient’s vertex, and the SNS monitor is across from the surgeon. For a left-sided approach, the surgeon, assistant, and scrub person switch sides; anesthesia, the microscope, and SNS equipment remain as is.

CHAPTER 17: SKULL BASE NEUROSURGERY

&

285

passing through the drape. Sterilizing the DRF eliminates this source of error; however, attaching the sterile frame requires penetrating the drape and may compromise sterility of the field. Being aware of these potential problems should enable effective use of either technique. SKULL BASE APPROACHES SNS guidance in our practice has been most useful for four basic skull base approaches. These include transpetrosal approaches to lesions of the middle fossa, posterior fossa, or both; transfacial and frontobasal approaches to lesions in the anterior midline involving the sella, clivus, and upper cervical spine; extradural resection of the anterior clinoid process for lesions along the proximal internal carotid artery, medial sphenoid wing, and cavernous sinus; and far lateral approaches to lesions eccentrically positioned. Although an anteroposterior radiographic view can be obtained to identify the midline, it may obstruct the working area, particularly if a microscope is used. Anteroposterior fluoroscopy may also increase the risk of contamination of the anterior foramen magnum and lower clivus. The most common skull base approach using SNS guidance in our experience is transpetrosal. Case examples are provided for each of these approaches.

Transpetrosal Approach Because of the complexity of the lesions approached at the petrous apex and along the clivus, transpetrosal approaches are among the most useful applications of SNS in skull base surgery. Surgical navigation is useful for positioning the patients, planning the incisions, placing burr holes, providing landmarks for the otologist during petrosectomy, and localizing landmarks within and around the lesions during neurosurgical resection. Case Illustration—Petrosal Approach A 49-year-old man underwent trans-sphenoidal partial resection of a clival chordoma 23 years previously, followed by conventional radiotherapy. Two episodes of double vision three years previously prompted MRI scanning, which revealed tumor recurrence. Subsequent MRIs at six-month intervals revealed steady tumor enlargement; the patient’s double vision became constant, and he began to have subjective balance difficulties. Neurologic exam revealed only a right sixth nerve palsy. MRI (Fig. 4A) demonstrated right petroclival tumor distorting the brainstem, extending from the cavernous sinus inferiorly to the seventh and eighth nerves. SNS-guided transpetrosal craniotomy for tumor resection was performed. Postoperative CT (Fig. 4B) and MRI (Fig. 4C) demonstrate SNS-guided bony and tumor resection, respectively. Postoperatively, the patient had a transient seventh nerve palsy which resolved by discharge.

286

&

PART III: CLINICAL APPLICATIONS

Figure 4 A 49-year-old man with residual and recurrent chordoma 23 years after initial transsphenoidal resection presented with double vision from a new sixth nerve palsy. (A) Preoperative MRI shows right petroclival mass distorting the brainstem. (B) Postoperative CT shows the bony resection of the petrosal approach. (C) Postoperative MRI demonstrates tumor resection.

Frontal Approach Frontal approaches have been significantly improved by the use of surgical navigation. Historically, these approaches required intraoperative fluoroscopy, usually oriented in a lateral position. Identification of the midline is essential during the exposure but can be misleading if the usual landmarks (e.g., nasal septum or vomer) are distorted. Surgical navigation can eliminate the use of fluoroscopy completely

CHAPTER 17: SKULL BASE NEUROSURGERY

&

287

by identifying the midline during the exposure and guiding the extent of resection in the axial, coronal, and sagittal planes. Case Illustration—Frontobasal Approach A 43-year-old man presented with blurred vision two years previously; examination had demonstrated bitemporal field cuts, and MRI revealed a large tumor in the sphenoid sinus extending upward through the sella tursica, and into the clivus. The patient underwent trans-sphenoidal resection and chordoma was confirmed histologically. Further symptomatology five months later included left orbital headache and deteriorating left vision; the patient then underwent pterional craniotomy with resection of the anterior clinoid process for further tumor debulking of the suprasellar component of tumor. Eight months later, the patient lost left optic nerve function and developed a left sixth nerve palsy; MRI was repeated (Fig. 5A) and radical resection was then undertaken. The SNS was utilized to guide a bifrontal craniotomy with bilateral orbital rim and orbitofrontal osteotomy, and resection of the cribriform plate. Image guidance-enabled confirmation of the clival anatomy, petroclival dura, and inferior and lateral margins of chordoma. Postoperative CT (Fig. 5B) and MRI (Fig. 5C), respectively, illustrate bony and soft tissue resections. Extradural Resection of the Anterior Clinoid Process Cavernous sinus and medial sphenoid wing approaches have been enhanced by the extradural bony resection of the anterior clinoid process as described by Dolenc (10). This approach aids exposure of tumors, vascular anomalies, and bony lesions in this region although we would not recommend this approach for paraclinoid aneurysms. Surgical navigation is especially useful when the normal anatomy is distorted such as that occurs from hyperostosis caused by a meningioma. Identifying the optic foramen is rendered more facile with stereotactic guidance. Case Illustration—Clinoid Resection A 52-year-old man developed paresthesias in his left jaw, tongue, and cheek, with occasional imbalance. Exam confirmed decreased sensation in the left V3 distribution, with a right-sided Hoffman’s sign. MRI demonstrated an enhancing mass extending from the posterior fossa through Meckel’s cave into the cavernous sinus in a dumbbell shape. The man underwent a transpetrosal approach for the posterior fossa component of the tumor, and pathology confirmed the diagnosis of chondroid chordoma. Postoperatively, the patient developed transient left fifth and seventh nerve palsies and expressive dysphasia; a sixth nerve palsy persisted. MRI (Fig. 6A) demonstrated the residual 3 cm cavernous tumor, which was deemed too large for proton beam radiotherapy, and he therefore underwent a second surgery for debulking the anterior portion of the tumor. Image-guided

288

&

PART III: CLINICAL APPLICATIONS

Figure 5 A 43-year-old man presented with progressive left-sided vision loss and sixth nerve palsy, status post redo trans-sphenoidal and single pterional approaches to the clival chordoma of the patient. (A) Preoperative MRI. Tumor fills the clivus, sella, spheno-ethmoids, and region of the left anterior clinoid. (B) Postoperative CT. The bony cuts are visualized. (C) Postoperative MRI. Tumor is largely absent.

CHAPTER 17: SKULL BASE NEUROSURGERY

&

289

Figure 6 Surgical navigation was used to localize the anterior clinoid process, identify the optic foramen extradurally, resect tumor, and prevent inadvertent entry into the ethmoid sinuses. (A) Preoperative MRI. (B) Postoperative MRI.

frontotemporal craniotomy was performed with extradural resection of the orbital roof, opening of the optic foramen, and removal of the anterior clinoid process, followed by dural opening, cavernous sinus dissection, and tumor debulking through Parkinson’s triangle. Since the location of the tumor, optic apparatus, and cavernous sinus were identified by surgical navigation, the risk of neural injury was reduced. A more precise assessment of the extent of resection was performed intraoperatively by placing the probe circumferentially around the tumor bed. In addition, surgical navigation helped ensure that the ethmoid sinuses were not entered unintentionally, exposing the patient to the risk of cerebrospinal fluid leak. His only new deficit from the second surgery was mild hypesthesia of the left V1 distribution. Postoperative MRI (Fig. 6B) demonstrates tumor resection. Far Lateral Approach The far lateral approach is the least utilized skull base approach in our practice. Perhaps the limited number of cases makes the use of surgical navigation even more valuable. Far lateral approaches are used for vascular lesions or tumors near the anterior foramen magnum up to the internal auditory canal. In each case, the exposure was modified to meet the specific trajectory necessary to reach the target lesion. Anatomic landmarks to determine the limit of resection are lacking in the occipital condyle, until the hypoglossal canal is encountered; however, this may risk injury of the nerve. Surgical navigation can provide an immediate assessment of the trajectory toward the target lesion as the condyle is drilled. Once the desired trajectory is reached, additional drilling becomes unnecessary and the risk of complications is reduced.

290

&

PART III: CLINICAL APPLICATIONS

Figure 7 Figure (Caption7 on facing page) (Caption on facing page)

CHAPTER 17: SKULL BASE NEUROSURGERY

&

291

Case Illustration—Far Lateral Approach A 53-year-old man presented with headaches and ataxia that became progressively worse over a three-month period. A cerebral aneurysm at the pontomedullary junction anteriorly was suspected based on an MR image and was confirmed by angiography (Fig. 7A). A far lateral approach was taken from the left side. Surgical navigation was useful in identifying the vertebral artery, particularly at the transverse foramen of the atlas. However, the most useful advantage of this technology was during the drilling of the lateral mass of the atlas and the occipital condyle. In this case, surgical navigation was valuable during the approach, but was not necessary during dissection and clipping of the aneurysm. Intraoperative angiography (Fig. 7B) and postoperative CT scan (Fig. 7C) show the angle of approach and extent of the bone resection, respectively, that enabled safe clipping of this aneurysm.

COMPLICATIONS AND LIMITATIONS In our experience, we have had no complications directly attributable to the use of SNS in skull base surgery. We have aborted its use in cases where the imaging data did not transfer to the workstation or the accuracy was recognized to be inaccurate. In all cases, the maximal accepted error at the time of calibration was less than 5 mm but was determined to be less than 3 mm in most. The accuracy of SNSs is reported elsewhere in this book. However, in no case in our series did the degree of error result in injury to a patient. It is imperative for neurosurgeons to have a thorough knowledge of the anatomy encountered during skull base approaches and not to rely on surgical navigation as a substitute. Although surgical navigation has been a tremendous asset to skull base surgeons, there are definite limitations. In skull base approaches, there are frequently deep exposures that become quite small at their depth, much like a cone. A probe or calibrated instrument must be directed into this exposure. However, the surgeon must be able to precisely visualize where the tip of the probe is to make use of the SNS computer image that pinpoints an anatomic landmark. Even at their tips, these probes may be several millimeters in diameter, reducing their precision. Combining the size of the tip of the calibrated instrument with the inherent degree of error in all SNS systems, it may be possible to identify the general vicinity of surgical activity but not an exact anatomic marker. Figure 7 (Facing page) A 53-year-old man presented with headaches and ataxia. (A) Preoperative left vertebral artery angiogram shows a large broad-necked aneurysm arising just beyond the vertebrobasilar junction. (B) Intraoperative right vertebral artery angiogram confirms the patency of this artery through the fenestration of two clips and obliteration of the aneurysm. (C) A postoperative CT scan shows the extent of the bone resection after a far lateral exposure using surgical navigation guidance.

292

&

PART III: CLINICAL APPLICATIONS

Another limitation of SNS not unique to skull base surgery but perhaps most manifest therein is the requirement of the surgeon to look first at the operative field where the probe is touching an anatomic landmark, and then at the computer monitor. To do this, the surgeon must stop operating and look away from the field to check the surgeon’s position on the SNS. This not only slows the operation but can be dangerous: should the tip of the probe ‘‘move’’ while the surgeon is looking away, it could conceivably injure a delicate structure. In our opinion, this lessens the advantage of calibrating other instruments to use as probes because the surgeon cannot continue to operate while looking away from the field regardless of which instrument is chosen as a probe. Some systems have a heads-up display in the microscope field of view that shows the preselected outline of the target and the relation of the probe tip to that target. Some microscopes now display the computer monitor images in a heads-up display, but this compromises the surgical field of view and has to be turned on and off. As the technology improves, this limitation will likely be solved.

INTRAOPERATIVE IMAGING The most recent advancement in SNS is the use of intraoperative imaging to update the accuracy of the stereotactic guidance. Skull base lesions are rarely affected by shift of neural structures because they are fixed to the skull, which is essentially immobile. However, updated imaging is very useful to assess partially resected lesions and ensure the continued accuracy of the navigation system. Further, the ability of intraoperative imaging to ‘‘see around’’ structures that are opaque in the surgical field may indeed be able to reduce the extent of soft tissue and bony dissection currently advocated for skull base approaches in deference to neural tissue (1,11). Although use of intraoperative CT and, especially, ultrasound has been well described, the imaging characteristics of these modalities are often suboptimal for visualization of intracranial pathology. MRI was brought into the operating room in the mid-1990s, and modified several times since then. The first-generation intraoperative MRI is an open design wherein the patient’s head or spine is placed at the center of a ‘‘double donut’’ configuration; the surgeon and assistant operate essentially within the scanner (11,12). Near-real-time image acquisition and thus surgical navigation with up-to-date images is possible with this system (11). However, this concept requires significant redesign of both operating room (RF shielding, structural support) and instrumentation (surgical tools, anesthesia equipment, microscopes). Second-generation intraoperative MRI involves moving the patient to the scanner, either in or near the operating room. This obviates much, but not all, of the need for specialized MRI-compatible instrumentation; however, major structural modifications remain generally necessary, and there are some obvious risks involved in moving the patient to the scanner (12). Third-generation scanners change the paradigm again. Two different concepts are a ceiling-mounted highfield device which can be tracked into and out of one or more operating rooms (13), and an ultra-low field device which can be stationed at the head of the operating table and raised or lowered as needed for imaging (12). These latter two

CHAPTER 17: SKULL BASE NEUROSURGERY

&

293

designs nearly eliminate instrument modification, excepting the skull fixation device. The former device can produce extraordinary images and can be used for routine diagnostic imaging when not being utilized in the operating room; there is, however, a modest increase in operative time due to transferring the magnet in and out of the field (13). The latter device interferes minimally with surgery, but has a fairly small aperture, which can force changes in preoperative positioning in order to accommodate the intraoperative imaging (12). Dort and Sutherland (13) described the use of intraoperative MRI (ceilingmounted) for a series of skull base procedures. They found the system useful for preoperative planning, including rechecking lesion size compared to occasionally temporally remote preoperative MRI, for intraoperative updating, and for immediate postoperative checking for resection completeness and absence of complications such as stroke or hematoma (13). The majority (9 of 11 cases) of MRI-related interventions (i.e., unsuspected residual tumor resulting in the need for additional resection) in their series, however, related to pituitary adenomas. Pitfalls of intraoperative MRI do exist. As mentioned above, there may be substantial and possibly prohibitively expensive modifications required in both the physical plant and the surgical instrumentation. Further, there are potential artifacts of intraoperative imaging, such as real-time formation of T2 signal changes and contrast enhancement, which can be frustrating as well as dangerously misleading if they become the basis for unnecessarily aggressive tumor resection and unintentional damage to tumor-free structures (12). Whether the potential benefits of intraoperative MRI outweigh its risks and expense remain to be seen, particularly in skull base surgery.

LEARNING CURVE As with most new instrumentation, there is a learning curve associated with the use of surgical navigation in general and for skull-based approaches in particular. This learning curve involves the proper placement of the fiducials to cover a sufficient area around the head while maintaining easy access to these fiducials during the calibration process. As discussed earlier in this chapter, experience also improves the surgeon’s ability to position the equipment to avoid interference during the operation. Perhaps, the greatest learning curve relates to knowing when the SNS is useful and when it is superfluous. Early in our experience, the probe was frequently used to confirm exposed anaomy that was already recognized. The more familiar we have become with skull base approaches, the fewer interruptions have been needed during the operation to confirm the exposed anatomy. Surgical navigation is most useful for invasive tumors and redo operations in which the anatomy may be distorted or when scarring obscures the normal structural landmarks. It is also more useful for large skull base tumors, particularly at the petrous apex and clivus with brainstem compression in which circumferential exposure is limited. In these cases, more frequent SNS localizations allow safer and faster resection of these tumors by preventing resection beyond the limits of the tumor into vital brain structures.

294

&

PART III: CLINICAL APPLICATIONS

SUMMARY AND FUTURE Surgical navigation serves many purposes in neurosurgery and has now become an essential component of modern practices. It has a special niche in skull base surgery. Although most of the skull base techniques were developed prior to the advent of this technology, parallel advances in both areas have enabled their integration. Future advancements will increase the accuracy and the resolution of SNSs, perhaps enabling individual cranial nerves and small arteries to be localized around tumors to further ensure their preservation. Image fusion with CT angiography may improve the safety and outcome of aneurysm surgery particularly in larger aneurysms, which cannot be treated by an endovascular approach. Accuracy is now enhanced by intraoperative imaging, which can provide updated data to reflect partial resection of tumors and account for brain shift related to hyperventilation, mannitol administration, and cerebrospinal fluid drainage. Further advancements in computer and robotic technology may enable computer-assisted exposure and resection of tumors to supplement the localization features that already exist.

CHAPTER 17: SKULL BASE NEUROSURGERY

&

295

REFERENCES 1. Robinson JR, Golfinos JG, Spetzler RF. Skull base tumors: a critical appraisal and clinical series employing image guidance. Neurosurg Clin North Am 1996; 7: 297–311. 2. Sure U, Alberti O, Petermeyer M, Becker R, Bertalanffy H. Advanced image-guided skull base surgery. Surg Neurol 2000; 53:563–572. 3. Tew JM, van Loveren HR. Atlas of Operative Microneurosurgery. Philadelphia: W. B. Saunders, 1994: 44–53. 4. Bumpous JM, Curtin HD, Prokopakis EP, Janecka IP. Applications of image-guided navigation in the middle cranial fossa: an anatomic study. Skull Base Surg 1996; 6:187–190. 5. Fernandez PM, Zamorano L, Nolte L, Jiang Z, Kadi AM, Diaz F. Interactive image guidance in skull base surgery using an opto-electronic device. Skull Base Surg 1997; 7:15–21. 6. Carrau RL, Curtin HD, Snyderman CH, Bumpous J, Stechison M. Practical applications of image-guided anterior craniofacial resection. Skull Base Surg 1995; 5:51–55. 7. Golfinos JG, Fitzpatrick BC, Smith LR, Spetzler RF. Clinical use of a frameless stereotactic arm: results of 325 cases. J Neurosurg 1995; 83:197–205. 8. McDermott MW, Gutin PH. Image-guided surgery for skull base neoplasms using the ISG viewing wand: anatomic and technical considerations. Neurosurg Clin North Am 1995; 7:285–295. 9. Mcdermott MW, Gutin PH. Image-guided surgery for skull base neoplasms using the ISG viewing wand: anatomic and technical considerations. Neurosurg Clin North Am 1996; 7:285–295. 10. Dolenc VV. Anatomy and Surgery of the Cavernous Sinus. Vienna: Springer-Verlag, 1989:233–256. 11. Moriarty TM, Kikinis R, Jolesz FA, Black PM, Alexander E. Magnetic resonance imaging therapy: intraoperative MR imaging. Neurosurg Clin North Am 1996; 7:323–331. 12. Barnett GH. Intraoperative magnetic resonance imaging. Contemp Neurosurg 2002; 24:1–6. 13. Dort JC, Sutherland GR. Intraoperative magnetic resonance imaging for skull base surgery. Laryngoscope 2001; 111:1570–1175.

INDEX

3-D digitizers, 6–7, 22 3-D rotational angiography, 227–228 Actual wand trajectory (AWT), 39 Autosegmentation, 20 Awake craniotomy and intraoperative imaging advantages, 104–105 anesthetic techniques, 105–107 awake procedures using iMRI, 114 context of tumor surgery, 104 disadvantages, 105 dynamic intraoperative image-guided resection, 111–112 future, 115 indications, 103–105 intraoperative functional mapping, 107–110 intraoperative use of MRI, 112–114 mapping, 104 positioning of patients, 106–107 static image-guided resection, 110–111 see also Surgical navigation systems (SNSs) Brachytherapy, 81 Brain biopsy complications hemorrhage, 83–84 nondiagnostic biopsy, 84 image-guided frame-based stereotactic, 72 image-guided frameless stereotactic, 72–78 iMRI-guided, 72 other technologies, 78–81 postoperative management, 85

[Brain biopsy] SBB-related procedures frameless vs. frame, 82–83 role of intraoperative MRI-guided biopsy, 83 using SNS, 75 see also Surgical navigation systems (SNSs) Brown–Roberts–Wells stereotactic system, 6, 11 registration process, 7 Cartesian coordinate systems CT and MR imaging, 5 stereotactic atlases, 5 see also Non-Cartesian coordinate systems CBYONÕ IGS system, 214 Cerebral metastases (CM), 155 clinical features, 156 image-guided surgery, 158–162 image-guided surgery vs. radiosurgery, 165–167 imaging, 156–157 stereotactic radiosurgery, 162–165 treatment, 157–158 Computed tomography (CT) angiography (CTA), 224–225 calculation of stereotactic frame coordinates, 6 invention of, 91 perfusion, 226 vs. videofluoroscopy, 210 vs. MR imaging, 53 Computer-assisted minimally invasive surgery (CAMIS), 37 description and requirements, 37–38

297

298

&

INDEX

[Computer-assisted minimally invasive surgery (CAMIS)] localization and orientation, 38–39 target guidance concentric circles, 40–41 target push, 39–40 trajectory guidance collinear sphere and concentric circles, 41–42 trajectory projection, 42–44 Computer-assisted neuroendoscopy (CANE), 126 Computer-directed neurosurgery, 32–33 Coordinate spaces operative, 6 relationship between, 7 transformation matrices, 8–11 Coordinate systems Cartesian, polar, spherical or cylindrical, 4 Pierre de Fermat and Rene´ Descartes, 4 Coregistration, 12 Cortical mapping of speech and motor, 104 Craniotomy see Awake craniotomy and intraoperative imaging; Minimal access craniotomy (MAC) using SNSs Cysts arachnoid, 131 colloid, 131 cystic abscess and hematoma, 131–132 intraventricular ependymal, 131 Digital angiography, 52–53 Digital tomographic imaging modalities, 49 Digitizers, 21–22 see also Electromagnetic digitizers; Intraoperative digitization; 3-D digitizers Dynamic reference frames (DRF), 74, 282 Electromagnetic digitizers, 28 Endoscopic third ventriculostomy, 128–129 Endoscopy, 121

[Endoscopy] see also Neuroendoscopy Epilepsy surgery future development, 269 preoperative investigations depth electrodes, 258–259 noninvasive studies, 258 subdural strip and grid electrodes, 259–261 therapeutic procedures anterior temporal lobectomy, 264–265 corpus callosum section, 268 lesionectomy, 261–264 neocortical resection, 267–268 reoperation, 268–267 selective medial temporal resection, 265–267 Extent of resection (EOR) estimated vs. actual EOR, 181, 182–184 impact on EOR of brain tumors on outcome, 101 prospective study on SNS and, 180–184 Fiducial localization error, 51–52 Fiducial markers bone-anchored, 56–57 extrinsic, 56 intrinsic, 56 Fiducial point strategies, 11–12 Fiducial registration error, 51–52 Fluoroscopic navigation, 251–252 Fluoroscopy, 31–32 Frame stereotaxy systems, 37 Frame-based stereotactic biopsy, 72 Frameless application registration, 29 see also Registration Frameless stereotactic biopsy SBB using intraoperative MRI, 79–81 SBB with the use of SNS, 73–79 Frameless stereotactic image guidance, 229–230 Frameless stereotaxy system, 110, 126 Functional MRI (fMRI), 109–110 Functional neurosurgery, 81–82 ‘‘Fusion’’ see Multimodality imaging Gantry coordinates, 4–5 Gliomas see Intracranial gliomas, SNSs for resection of

INDEX

‘‘Hat-and-head’’ matching algorithm (Pelizarri and Chen), 12 Hemorrhage, 83–84 Hydrocephalus multiloculated, 130 obstructive, 128–129 Image artifact, 99 Image-guided endoscopy (IGE), 215–217 Image-guided frame stereotaxy, 37 vs. interactive SNSs, 37 Image-guided resections, 110–111 Image-guided spinal navigation clinical applications anterior thoracolumbar surgery, 248–249 C1–C2 transarticular screw fixation, 242–245 other spinal applications, 250 pedicle fixation, 239–242 segmental C1–C2 screw fixation, 245–247 transoral surgery, 247–248 demerits, 250 fluoroscopic navigation, 251–252 principles, 234–239 registration techniques, 237–238 Image-guided surgery, 158–162 Interactive image guidance see Neuronavigation Interactive image-guided neurosurgery computers and software interfaces, 57–58 definition of accuracy, 50–52 image acquisition, 52–65 integration of real-time data, 62–63 interactive localization devices and intraoperative use, 58–62 judgment and clinical experience, 64–65 registration techniques, 55–57 robotics, 64 tissue displacement, 63–64 Interactive localization device, 58 Internal acoustic canal (IAC), 197, 198 Intracerebral hematomas (ICH), 221 Intracranial gliomas, SNSs for resection of considerations for use based on anatomic site, 176–179

&

299

[Intracranial gliomas] infratentorial, 178–179 supratentorial, 176–178 preoperative preparation, 174–176 technical considerations, 184 UCSF experience, 179–180 Intracranial meningiomas, SNS application in surgery craniotomy and tumor exposure, 191–194 internal decompression and extracapsular dissection, 195–196 localization and preservation of adjacent neurovasculature, 196–198 positioning and incision, 190 removal of involved bone and dura, 198 surgical management and characteristic features, 189 tumor devascularization, 194–195 Intraoperative digitization 2-D CCD passive infrared, 27 passive ultraviolet, 27–28 active vs. passive technologies, 22 future localization technologies, 29 machine vision, 28 magnetic systems, 28–29 alternating current, 29 direct current, 29 multiarticulated arms, 22 optical systems IREDs, 24–25 linear CCD, 25–26 sonic systems, 23–24 Intraoperative magnetic resonance imaging (iMRI), 224 artifacts, 99–100 benefits of using, 100–101 concept, 92 development in neurosurgery, 91–92 development of systems groundbreaking iMRI systems, 93–94 second-generation iMRI, 94–96 third-generation iMRI, 96–97 impact on brain biopsies, 101 impact on EOR of brain tumors on outcome, 101 instruments, 97–98

300

&

INDEX

[Intraoperative magnetic resonance imaging (iMRI)] interpretation of intraoperatively acquired images, 99 navigation, 98–99 Intraoperative navigation tracking head movement, 30 head-holder mounts, 31 head-holder references–dynamic reference frames, 31 table-mounted detectors, 30 table-mounted references, 30–31 see also Fluoroscopy Intraoperative tomography, 63 Intraoperative ultrasound, 112 Karnofsky performance status (KPS), 158 Localizing systems see Digitizers Magnetic resonance angiography, 224 Magnetic resonance imaging (MRI), 3 calculation of stereotactic frame coordinates, 6 introduction of, 91 open magnet, 7 see also Intraoperative magnetic resonance imaging (iMRI) Maximum tolerated dose (MTD), 163 Meningiomas see Intracranial meningiomas, SNS application in surgery Microscopes, 62–63 Minimal access craniotomy (MAC) using SNSs closure, 151 display interpretation, 143–144 lesion resection and tissue deformation, 149–150 patient positioning, preparation and opening, 145–149 surgical planning, 141–142 see also Awake craniotomy and intraoperative imaging MPO display, 38–39 MR compatible, 98 MR environment, 97

MR images correction of distortions, 53 rectification of distortions, 53–54 soft tissue discrimination, 53 spacing between slices, 53 MR safe, 97 Multimodality imaging, 21 Neuroendoscopy, 122 computer-assisted tools for manipulation cyst fenestration, 130–132 multiloculated hydrocephalus, 130 obstructive hydrocephalus, 128–129 tumors, 132–134 computer-assisted visualization, 127 current applications, 122–123 development of modern, 121–122 endoscope-assisted surgery, 134–135 improving with computer technology, 124 intraoperative ‘‘real time’’ neuroendoscopic navigation, 124–125 stereotactic guidance: use of preoperative imaging, 125–127 problems/limitations, 123–124 movement, 123 tissue manipulation, 124 visualization, 123–124 Neuronavigation, 19 see also Fluoroscopy Neurosurgical guidance infrared LEDs attached to surgical bipolar cautery forceps, 61 ISG Viewing Wand system, 59 machine vision, 61 magnetic field guidance system, 61 microscope-based frameless stereotaxy, 59–60 optical triangulation techniques, 60–61 passive localization arm, 59 spark-gap ultrasonic emitter technology, 60 Neurosurgical navigation, 19 see also Fluoroscopy Non-Cartesian coordinate systems, 6 Nondiagnostic biopsy, 84 Nonoperative brain mapping, 109–110

INDEX

Ommaya reservoir, 81 Paired point registration, 237 Pedicle fixation, 239–242 Point targeting, 6 Positron emission tomography (PET), 258 Precision, 50 Primary glial neoplasms, 173–174 see also Intracranial gliomas, SNSs for resection of Radiosurgery see Stereotactic radiosurgery (SRS) ‘‘Real time’’ intraoperative image guidance CT, 125 iMRI, 125 ultrasound guidance, 124–125 Recursive portioning analysis (RPA), 158 Reformatting, 55 Registration, 7–8, 55 alternative techniques, 55 standard fiducial sets for, 175 Regulus navigation system, 61 Reichert–Mundinger frame, 6 Rendering, 55 Robotic surgery, 32–33 Scalp-based fiducial markers, 5 Single photon emission tomography (SPECT), 258 Skull base neurosurgery complications and limitations, 289–290 future, 292 intraoperative imaging, 290–291 lateral setup, 282 learning curve, 291 skull base approaches extradural resection of the anterior clinoid process, 285–287 far lateral approach, 287–289 frontal approach, 284 transpetrosal approach, 283 SNS utilization estimation of EOR, 278–279 identification of skull base landmarks, 277 localization of boundaries of the lesion, 277–278

&

301

[Skull base neurosurgery] patient positioning, 276 placement of incision and craniotomy, 276–277 supine setup, 281 technical considerations positioning of the equipment, 280–283 preoperative selection of an imaging modality, 280 types of lesions targeted, 279–280 Slice thickness, 53 Spatially registered endoscopy, 63 Spine navigational technologies see Fluoroscopy Stereotactic atlases, 5 Stereotactic brain biopsy (SBB), 71 see also Brain biopsy; Frameless stereotactic biopsy Stereotactic frame systems, 5–6 Stereotactic procedures, evolution of, 125–126 Stereotactic radiosurgery (SRS), 162–165, 228–229 Leksell Gamma Knife, 162 LinAc, 162 vs. image-guided surgery, 165–167 Stereotaxy, 3–4, 233 image-guided frame, 33 Stereotaxy frame application accuracy, 51 mechanical limitations, 51 Surface matching registration, 237 Surface matching strategies, 12–14 Surface registration, 29–30 Surface-based registration, 55–56 Surface-rendering technique, 20 see also Volume-rendering technique Surgical navigation systems (SNSs), 19–20 prospective study on EOR and estimated and actual EORs, 182–184 estimated vs. actual EOR, 181 image analysis, 180–181 imaging studies, 180 patient characteristics, 181 patient enrollment, 180 study conclusions, 184 tumor volumes, 181–182 surgical planning and simulation, 20

302

&

INDEX

[Surgical navigation systems (SNSs)] control interface, 21 image fusion, 21 see also Brain biopsy; Fluoroscopy; Frameless stereotactic biopsy; Intraoperative navigation; Intracranial gliomas, SNSs for resection of; Intracranial meningiomas, SNS application in surgery; Skull base neurosurgery Target and trajectory guidance system, 37, 38 Target localization error, 52 Target-centered arc systems, 6, 8 Tethered interactive localization systems, 58–59 see also Untethered interactive localization systems Transformation matrices rotation by axes, 8–10 scaling, 10 translation, 10 Transphenoidal hypohysectomy, 207–208 advantages of image-guided endoscopy, 216–217 direct microscopic approach, 208 endoscopic approach, 212–213 endoscopic surgery using computerassisted virtual navigation, 214–215 IGS for transphenoidal surgery, 210–212 primary transphenoidal surgery, 209–210

[Transphenoidal hypohysectomy] transphenoidal reoperations and invasive parasellar tumors, 210 Triplanar display, 38 Tumor surgery, 104 Tumors, 132–134 cerebellar, 178 fourth ventricular, 178–179 frontal lobe, 176 occipital, 177 other brain locations, 177 pareital lobe, 177 posterior fossa biopsies, 179 temporal lobe, 176–177 Unbiasedness, 50 Untethered interactive localization systems, 58–59 Vascular malformations, 219 arteriovenous malformations (AVMs), 220–221 capillary telangiectasias, 223–224 cavernous malformations, 221 venous angiomas, 221–222 Ventricular endoscopy, 132 Video microscopy see Microscopes Videofluoroscopy, 209 vs. CT, 210 Volume-rendering technique, 20–21 Whole brain radiotherapy (WBRT), 161–162, 164