Integration of preoperative and intraoperative functional brain mapping in a frameless stereotactic environment for lesions near eloquent cortex

Transcription

1 J Neurosurg 90:591Ð598, 1999 Integration of preoperative and intraoperative functional brain mapping in a frameless stereotactic environment for lesions near eloquent cortex Technical note JEFFREY D. MCDONALD, M.D., PH.D., BRIAN W. CHONG, M.D., JEFFREY D. LEWINE, PH.D., GREG JONES, PH.D., ROBERT B. BURR, PH.D., PAUL R. MCDONALD, B.SC., SPENCER B. KOEHLER, PH.D., JAY TSURUDA, M.D., WILLIAM W. ORRISON, M.D., AND M. PETER HEILBRUN, M.D. Departments of Neurosurgery and Radiology, University of Utah School of Medicine, Salt Lake City, Utah The authors present a method of incorporating preoperative noninvasive functional brain mapping data into the frameless stereotactic magnetic resonance (MR) imaging dataset used for image-guided resection of brain lesions located near eloquent cortex. They report the use of functional (f)mr imaging and magnetic source (MS) imaging for preoperative mapping of eloquent cortex in difficult cases of brain tumor resection such as those in which there are large expansive masses or in which reoperations are required and the anatomy is distorted from prior treatments. To correlate methods of preoperative and intraoperative mapping localization directly, the authors have developed techniques of importing preoperative MS and fmr imaging data into an image-guided frameless stereotactic computer workstation. The data appear as a seamless overlay on the same preoperative volumetric MR imaging dataset used for stereotactic guidance during the operation. Intraoperatively identified functional locations mapped by cortical stimulation are recorded as digitally registered points. This approach should prove useful in assessing the accuracy and reliability of various preoperative functional brain mapping techniques. KEY WORDS brain neoplasm electrocorticography functional brain mapping functional magnetic resonance imaging image-guided surgery magnetoencephalography I N patients with intracranial neoplasms or arteriovenous malformations (AVMs) that require surgical resection, these anomalies often are found to be adjacent to, or within, a region of the brain that is critical for motor, sensory, or cognitive functioning. In these cases, it is important to know as much as possible about the local functional organization of the brain before surgery because this can aid in planning the surgical approach and in defining realistic goals and limits of lesion resectability. This information can also be of assistance in advising patients regarding treatment risks and therapeutic options, such as the choice of surgical resection or biopsy. Available data such as standard magnetic resonance (MR) imaging anatomical landmarks 3 and standard brain atlases provide some general information about brain functional anatomical localization; however, differences between individual patients can be significant. 13,18 Also, in patients with large intracranial masses, the local anatomy may be distorted so J. Neurosurg. / Volume 90 / March, 1999 that it can be difficult to identify important neuroanatomical landmarks accurately, based on preoperative MR imaging alone. 8,17,19 We have instituted a prospective multimodality regimen of preoperative noninvasive functional brain mapping for patients with brain tumors or AVMs near eloquent cortex. The techniques used include magnetoencephalography (MEG) and functional (f)mr imaging, as well as other modalities. Magnetoencephalography detects minute regional alterations in magnetic flux associated with neuronal electrical activity related to a particular task. 7 Using mathematical models, the location of the neuronal generators of the specific signal components can be inferred. In magnetic source (MS) imaging, MEG source locations are plotted on spatially aligned MR images to produce MS localization images. Magnetic source imaging analysis appears to be very promising for the accurate noninvasive mapping of motor and sensory areas, seizure focus 591

2 J. D. McDonald, et al. identification, and, possibly, language localization. 6,10,11 Functional MR imaging can be used to predict brain functional areas based on small regional alterations in blood flow associated with neuronal activation and may be especially useful for identifying areas of motor and language function. 5,12 Direct cortical stimulation and recording techniques 4,16 allow the surgeon to map specific areas of motor cortex activation and, in the awake patient, specific cortical areas associated with reading or picture naming, expressive speech, and somatosensory cortex activation. To correlate methods of preoperative and intraoperative mapping localization directly, we have developed techniques of importing preoperative MS and fmr imaging data into an image-guided frameless stereotactic computer workstation. The data appear as a seamless overlay on the same preoperative volumetric MR imaging dataset used for stereotactic guidance during the operation. Intraoperatively identified functional locations mapped by cortical stimulation are recorded as digitally registered points. Direct cross-modality correlation and verification of preoperative and intraoperative functional maps are thus achievable within the frameless stereotactic working environment. In this report we demonstrate the technical feasibility and usefulness of this adjunct to functional brain mapping through illustrative cases. Description of Technique Patients undergo MEG recording for which five to eight surgical fiducial markers are placed on the scalp. A largearray whole-head 122-channel biomagnetometer is positioned over the head and a visual, auditory, somatesthetic, motor, or dichotic listening task is presented to the patient as appropriate for the location of the brain lesion. A multiple dipole spatial/temporal model is used to characterize MS locations for MS imaging. 9 Neuromagnetic data are recorded with a precise relationship to the anatomical landmarks on the patientõs head, which are defined via a three-dimensional digitizer and fiducial markers. Data analysis for MEG is performed using a software package with overlay of MEG sources as slice data points onto an anatomical volumetric T 1 -weighted MR image obtained immediately following the MEG examination. The choice of MR imaging parameters is determined by the prerequisites of the image integration software. Image integration of the preoperative functional mapping with the anatomical MR imaging dataset is performed within the frameless stereotactic workstation. It is a requirement of the current software (version 2.6) that the volume datasets be generated from isotropic voxels (that is, data points or voxels are the same dimension in the x, y, and z planes, specifically 1 mm). A three-dimensional volume gradient-echo sequence is used to acquire the anatomical MR image (stereotactic study) on one of two 1.5-tesla MR imaging systems (Picker or GE). Using the Picker MR imager, images are obtained using the following parameters: a 15-msec repetition time (TR), a 4.47-msec echo time (TE), and one excitation; a 20û flip angle (FA); 200 contiguous slices measuring 1 mm each; a matrix; and a 25.6-cm field of view (FOV). Using the GE unit, the parameters included: a 22-msec TR, minimum TE, and one excitation; a 45û FA; 160 to 180 contiguous slices, each measuring 1 mm; a matrix; and a 25.6-cm FOV. The fmr image is obtained on the day after the MEG examination. Data are acquired using the GE MR imager with a multiphase, single-shot, gradient-echo evoked potential index image. Imaging parameters include: a msec TR, 60-msec TE, and one excitation; a 90û FA; 5- mm-thick slices with a 2-mm gap; six slice locations with 80 phases per location for a total of 480 sections; a matrix; and a 25.6-cm FOV. For fmr imaging, the patient is presented with a visual, auditory, somatesthetic, motor, language, or mnemonic stimulus task. Stimuli or tasks are presented in alternating 30-second time periods (for example, finger flexion alternating with a rest period) time-locked to image acquisition. The functional data are initially analyzed off line by using a Sun workstation equipped with an image processing software package. Data processing consists of pixel-by-pixel statistical analysis using either a temporal correlation between a reference waveform and the task-generated function, or an unpaired Student t-test in which images acquired during performance of the task (for example, finger flexion) are compared with images acquired during the alternating condition (for example, rest). 1,2 The reference waveform is in the shape of a ÒboxcarÓ of alternating task conditions with an initial 6-second delay, approximating the delay between the task performance and the physiological response. Analysis is performed on the 64 64Ðpixel echoplanar images and a statistical threshold is set by using either cross-correlation values or z-scores generated from the unpaired Student t-test analysis. Statistical maps are either overlaid directly onto the corresponding Ðpixel anatomical image or rendered via spatial transformation into the anatomical volume. The spatial transformation is calculated using the Automated Image Registration package within the Medx program. Integration of preoperative functional mapping is performed within the StealthStation frameless stereotactic workstation using the ImMerge software module. Briefly, the MS imaging dataset output from NeuroMag in DICOM version 3.0 format is transferred via the computer network directly to the hard disk of the StealthStation computer by using a conversion utility called ÒstorecpÓ written within our department and a modification of the DICOM Toolkit version 3.11 public domain software to set the DICOM-formatted images on the hard drive. This program also outputs the information manually to create the required ÒattributesÓ file, containing pixel size, slice thickness, FOV, and other volumetric information. The DICOM images on the hard drive are converted into files compatible with StealthStation software using a utility written within our department entitled Òdicom2stax.Ó This creates a stereotactic dataset that can be opened as the ÒreferenceÓ dataset in ImMerge. We load our fmr imaging dataset, containing mapped functional points generated within Medx, as the ÒworkingÓ dataset in ImMerge. The Medx output is currently not available in DICOM format and thus must be transferred by file transfer protocol in isotropic Analyze format to the StealthStation. The Im- Merge software scales and transforms the MS and fmr imaging datasets into two-dimensionally matched volumes loadable into the cranial program of the Stealth- 592 J. Neurosurg. / Volume 90 / March, 1999

3 Integrated functional brain mapping Station as ÒVolume IÓ and ÒVolume II,Ó respectively. Merging of these two datasets is performed by following standard ImMerge guidelines using anatomical landmarks, after which the merged functional dataset (containing both MS imagingð and fmr imagingðmapped points of function) is used in the operating room as the stereotactic dataset. The patient undergoes surgery within 1 to 4 days after the preoperative functional brain mapping analysis has been completed. Point-merge and surface-merge registration of the merged preoperative stereotactic MR imaging datasets with the fiducial markers on the patientõs scalp is performed in the usual manner. Before biopsy or resection of the lesion, the patient undergoes a craniotomy followed by standard cortical stimulation for intraoperative mapping of language, sensory, and/or motor functional areas. Cortical stimulation is performed using an Ojemann-type bipolar current electrode (60 Hz, 1-msec pulse, 0.5Ð10- ma current) according to methods previously described. 16 During awake procedures, patients receive a continuous infusion of propofol/remifentanyl. Continuous electroencephalography recording is performed during mapping to monitor after-discharge potentials. When generalization of after-discharge potentials is identified, immediate termination of these potentials is achieved by intravenous administration of 50 to 100 mg of thiopental and/or cortical administration of ice-cold lactated RingerÕs solution, as previously described. 15 Areas of function that are intraoperatively mapped are stored digitally within the StealthStation Surgical Plan mode as Òtargets.Ó The direct correlation of these intraoperatively identified functional targets with the preoperative functional map can be performed quantitatively by measuring the distance in millimeters between preoperatively and intraoperatively identified areas of identical function (such as motor cortex localization of a finger contraction task identified both preoperatively on fmr imaging analysis and intraoperatively by motor cortex stimulation). If similar, but not identical, functional tasks are mapped preoperatively and intraoperatively (such as leg somatosensory cortex identified preoperatively on MS imaging and a distal leg motor contraction function identified intraoperatively by cortical stimulation, or hand motor cortex identified with a finger flexion task on preoperative fmr imaging and an area of motor cortex identified by intraoperative stimulation with production of complex handðwrist flexion), then correlation across mapping modalities can be performed qualitatively by localization and mapping of the homunculus of the patientõs primary motor and somatosensory cortex. J. Neurosurg. / Volume 90 / March, 1999 Sources of Supplies and Equipment The whole-head 122-channel biomagnetometer was obtained from NeuroMag Ltd. (Helsinki, Finland), as was the NeuroMag software package. Two 1.5-tesla MR imaging systems were used in this study: the Picker Eclipse Power Drive 250, available from Picker International (Cleveland, OH), and the GE Signa Echo Speed, available from General Electric Medical Systems (Milwaukee, WI). The StealthStation frameless stereotactic (Silicon Graphics O2) workstation and StealthStation software (version 2.6) were purchased from Sofamor Danek USA (Memphis, TN) as was the ImMerge software module, which was used with the workstation to integrate preoperative functional mapping. Initial analysis of functional data was performed using a Sun workstation obtained from Sun Microsystems (Palo Alto, CA) equipped with the Medx software package available from Sensor Systems Inc. (Sterling, VA). Analyze software was obtained from CN- Software (Wilmington, DE). An Ojemann-type bipolar current electrode, manufactured by Radionics, Inc. (Burlington, MA), was used for the cortical stimulation. Illustrative Cases Our experience to date with these techniques is summarized in Table 1. For the purposes of this technical report, two illustrative cases are presented in more detail. Case 1 This 25-year-old right-handed man experienced a single generalized tonicðclonic seizure. Examination. The patient was neurologically intact on referral to the neurosurgery service for consultation. Neuroimaging demonstrated a large lesion within, or just anterior to, the primary motor cortex. The anatomical MR images were ambiguous with regard to the location of the central sulcus, based on standard criteria (Fig. 1aÐd) and, thus, preoperative functional analysis was performed to assist in the assessment of the surgical risk of attempted resection of the lesion. Functional MR imaging (Fig. 1e) and MS imaging analysis (Fig. 2) produced discordant results with respect to motor cortex identification. Operation. We performed an awake craniotomy with intraoperative cortical stimulation for definitive localization of primary motor and somatosensory cortex, as well as areas associated with direct expressive speech arrest. For this case, MS imaging data were imported preoperatively into the frameless stereotactic workstation dataset and agreement was verified during intraoperative mapping with the identified locations for motor and somatosensory function (Fig. 2). The lesion was confirmed to be located nearly wholly within a markedly expanded primary motor gyrus. Attempted resection of the lesion was judged inappropriate given the high risk of significant neurological deficit, and only biopsy of the lesion was performed. Pathological Findings and Postoperative Course. The pathological diagnosis was benign oligodendroglioma. Given the large volume of the lesion, the patient underwent external-beam radiotherapy to 54 Gy and he remains neurologically intact 6 months postoperatively. Case 2 This 42-year-old right-handed woman had undergone two prior craniotomies in 1986 and 1990 for resection of a right frontal oligoastrocytoma located anterior to the primary motor cortex. She received external-beam radiotherapy to 54 Gy in 1987 and multiple cycles of procarbazine-lomustine-vincristine chemotherapy, which were completed in

4 J. D. McDonald, et al. TABLE 1 Summary of patient characteristics and preoperative and intraoperative mapping techniques used* Age Preop Additional Case (yrs), Lesion Mapping Preop Intraop No. Sex Pathological Diagnosis Type Lesion Location Modalities Analysis Anesthesia Functional Mapping 1 25, M oligodendroglioma new lt parietal MSI fmri awake speech, motor, somatosensory 2 42, F mixed oligoastrocytoma recurrent rt posterior frontal fmri, MSI Ñ general motor 3 45, F oligodendroglioma recurrent rt posterior frontal fmri, MSI Ñ general motor 4 42, F anaplastic astrocytoma recurrent rt posterior frontal fmri, MSI Ñ general motor 5 37, F mixed oligoastrocytoma recurrent rt parietal fmri, MSI Ñ general motor 6 45, F anaplastic astrocytoma recurrent lt frontal fmri, MSI Wada awake speech, motor 7 37, F glioblastoma recurrent rt parietotemporal fmri, MSI Wada awake speech, motor, somatosensory 8 56, M oligodendroglioma new lt frontal fmri, MSI Ñ awake speech, motor, somatosensory 9 32, F oligodendroglioma recurrent lt frontal fmri, MSI Wada awake speech, motor, somatosensory * All patients except the one in Case 1 underwent resection of the lesion. Abbreviations: fmri = fmr imaging; MSI = MS imaging; Wada = intracarotid amobarbital test; Ñ = not applicable. Data imported into StealthStation. Examination. Because the patient had experienced progressive leg weakness in the year preceding the present admission, an MR image was obtained that documented progression of her residual/recurrent tumor (Fig. 3). The patient underwent preoperative noninvasive functional analysis with fmr and MS imaging and the data were imported into the frameless stereotactic workstation before surgery. This allowed direct correlation with functional points mapped during surgery by cortical stimulation. This analysis confirmed highly concordant localization of function across the testing modalities (Fig. 4). The mapping data, combined with standard use of the stereotactic surgical anatomical information, facilitated a near-total resection of the lesion. FIG. 1. Case 1. Anatomical MR images (aðd) and fmr images (e) obtained in a patient who presented with a generalized seizure and a large left frontoparietal mass consistent with a low-grade glioma on MR imaging. Localization of the central sulcus by MR imaging anatomical criteria on high T 2 -weighted images (aðc) suggested posterior displacement of motor cortex (asterisk) by the large tumor mass. The parasagittal anatomy (d) revealed the cingulate sulcus to be terminated in two symmetrical sulci (small arrows), making unambiguous identification of the marginal sulcus and rolandic cortex difficult. Functional MR imaging analysis for hand motor activity (e) supported the identification of primary motor cortex as seen in panels a through c. 594 J. Neurosurg. / Volume 90 / March, 1999

5 Integrated functional brain mapping FIG. 2. Case 1. a: Intraoperative photograph obtained in the same patient, who underwent an awake craniotomy with intraoperative functional mapping by cortical stimulation. Anatomical margins of the tumor at the cortical surface are outlined (black arrows). Points 1 and 2 = expressive speech interruption; point 3 = face motor; point 4 = hand motor with finger and wrist contraction; point 5 = finger somatosensory; point 6 = thumb somatosensory; and points 7 and 8 = tongue somatosensory. The large, expanded gyrus involved with tumor is seen between points 3 and 4. Close correlation between functional locations mapped preoperatively and intraoperatively is shown (bðd). b: Magnetic source imaging revealing the hand motor point is 2 cm deep within the same gyrus and within the same coronal plane as the intraoperatively mapped location for hand motor activation (point 4). c: Magnetic source imaging revealing the hand somatosensory point is 5 mm deep within the same gyrus and within the same coronal plane as the intraoperatively mapped location for hand somatosensory function (point 5). d: Magnetic source imaging revealing that the tongue somatosensory point is located within the same axial plane, 1.5 cm deep and slightly posterior to the intraoperatively mapped location of identical function (point 8). The unlabeled MS imaging points in the sagittal views of panels b and c are that of tongue somatosensory function, seen in the axial view in panel d. MSI = magnetic source imaging; somato = somatosensory. J. Neurosurg. / Volume 90 / March,

6 J. D. McDonald, et al. Postoperative Course. A transient postoperative supplemental motor area deficit, 14 which produced moderate contralateral arm and leg weakness, resolved by 4 weeks postoperatively. FIG. 3. Case 2. Preoperative axial (left) and sagittal (right) gadolinium-enhanced T 1 -weighted MR images. Discussion An increasingly common technique of modern neurosurgical practice is the determination of functional brain organization during surgery with cortical stimulation mapping techniques. The goal of the use of this technique is to delineate critical areas of brain function so that they can be preserved during lesion resection, thus minimizing the direct risk of neurological deficit from the surgery. Although intraoperative mapping is still considered by most FIG. 4. Case 2. Close correlation between preoperatively mapped functional points determined by MS and fmr imaging with intraoperatively mapped functional points determined by cortical stimulation in the patient. The correlation illustrated here is for hand somatomotor function. a: Image saved from the StealthStation monitor, with the red pointer dot on the location of hand motor activation (finger contraction) mapped by cortical stimulation. In the trajectory orientation looking down the central sulcus, slightly off-coronal (upper left), off-sagittal (lower left), and off-axial (upper right) views are presented. b: Enlarged view of the central sulcus. With a setting of 0.5 on the alpha sliding scale of the StealthStation (allowing a continuous blending between ÒVolume IÓ of the cranial program, containing mapped MS imaging information, and ÒVolume II,Ó containing mapped fmr imaging information), both the preoperatively mapped point for hand somatosensory function determined by MS imaging (white line on primary somatosensory gyrus, inferior to the red dot) and the preoperatively mapped location for hand somatomotor function determined by fmr imaging (large white dot located at the depth of the sulcus, spanning from motor to somatosensory cortex) are visualized. c: Pure MS imagingðmapped function, with the alpha scale of the StealthStation set to 0.0. d: Pure fmr imagingðmapped function, with the alpha scale set to 1. Median n. somato = median nerve somatosensory; MSI = magnetic source imaging. 596 J. Neurosurg. / Volume 90 / Msarch, 1999

7 Integrated functional brain mapping neurosurgeons to be the Ògold standardó of functional mapping, there are several reasons why reliable preoperative noninvasive functional analysis is often desirable. First, with intraoperative mapping, relevant data become available only during the surgical procedure, after the patient has already committed to an open procedure for attempted lesion resection (as opposed to a needle biopsy). Second, intraoperative mapping procedures usually require a larger craniotomy opening to expose not only the lesion to be resected, but also the adjacent functional cortex to be mapped. Third, intraoperative mapping often places greater demands on the patient. It lengthens the duration of the operation, requiring the patient to be in the operating room longer, and intraoperative speech mapping, of course, requires the patient to undergo an awake craniotomy assisted by local anesthesia. If preoperative noninvasive functional brain mapping techniques can be developed and demonstrated to be highly accurate and reliable in both normal and pathological settings, then these may eventually reduce the need for the performance of intraoperative functional cortical mapping. The cost effectiveness of this approach should be demonstrable through the guidance of presurgical decision making, saving individual patients from unneeded craniotomies when needle biopsy alone might be appropriate. To assess the accuracy and reliability of preoperative functional brain mapping techniques by direct correlation with intraoperative localization of function determined by cortical stimulation, we believe the techniques reported here for importing preoperative MS and fmr imaging data into an image-guided frameless stereotactic computer workstation will prove useful. Use of the frameless stereotactic workstation for recording mapping data facilitates direct correlation of points of function mapped preoperatively with those mapped intraoperatively. This occurs not only because mapping results are directly recorded into digital computer file format, but also because of the ability to rotate the digital volume representing the patientõs brain so that it can be viewed at any point in any plane, rather than only in traditional true axial, coronal, or sagittal planes (see Fig. 4). This approach should also prove useful in identifying possible susceptibilities of the specific preoperative mapping techniques themselves to the occurrence of signal artifact or alteration due to local pathological processes, such as anatomical expansion of functional cortex by infiltrating low-grade neoplasms (Case 1), anatomical distortion due to tumor recurrence after previous treatment (Case 2), possible signal alteration due to peritumoral interstitial edema, or altered regional blood flow adjacent to AVMs. An example of this includes the apparently false localization of the motor cortex by fmr imaging analysis in Case 1. With this methodology, the data exist in a digital format amenable to archiving and compiling into future standard digital functional neuroanatomical atlases. Although additional experience will be required for the quantitative analysis of the accuracy and concordance of the functional maps produced by each testing modality, we have found this approach to be extremely useful in the care of patients with brain tumors or AVMs near eloquent cortex. Conclusions We have outlined methods to incorporate preoperative J. Neurosurg. / Volume 90 / March, 1999 functional brain mapping data into the operative frameless stereotactic dataset used to guide resection of brain tumors or AVMs located near functional cortex. This approach should allow the direct correlation within the digital frameless stereotactic environment of various noninvasive preoperative testing modalities (such as fmr imaging and MS imaging discussed in this report) with intraoperative brain mapping by cortical stimulation, as well as permit direct intermodality crosscorrelation. The standardized application of these techniques in the functional mapping of patients with lesions near eloquent cortex should facilitate our understanding of the accuracy of the various testing modalities, as well as the relative susceptibility of each modality to pathological states such as anatomical distortion caused by expansive masses, prior treatment, interstitial edema, and regional alterations in blood flow. Financial Disclosure Dr. Heilbrun is a consultant to Sofamor Danek USA. The MEG facility at the University of Utah School of Medicine (the Center for Advanced Medical Technologies) has received an institutional research support grant from Picker International, Inc. Acknowledgments The authors wish to acknowledge the expert clinical care of these patients provided by Shelly OÕMeara, R.N. We also acknowledge the contribution of Edward Kinghorn, Ph.D., in the performance of the preoperative neurocognitive assessment of the patients reported. References 1. Bandettini PA, Jesmanowicz A, Wong EC, et al: Processing strategies for time-course data sets in functional MRI of the human brain. Magn Reson Med 30:161Ð173, Baudendistel K, Schad LR, Friedlinger M, et al: Postprocessing of functional MRI data of motor cortex stimulation measured with a standard 1.5 T imager. Magn Reson Imaging 13: 701Ð707, Berger MS, Cohen WA, Ojemann GA: Correlation of motor cortex brain mapping data with magnetic resonance imaging. J Neurosurg 72:383Ð387, Berger MS, Ojemann GA: Intraoperative brain mapping techniques in neuro-oncology. Stereotact Funct Neurosurg 58: 153Ð161, Binder JR, Swanson SJ, Hammeke TA, et al: Determination of language dominance using functional MRI: a comparison with the Wada test. Neurology 46:978Ð984, Gallen CC, Schwartz BJ, Bucholz RD, et al: Presurgical localization of functional cortex using magnetic source imaging. J Neurosurg 82:988Ð994, Gallen CC, Sobel DF, Lewine JD, et al: Neuromagnetic mapping of brain function. Radiology 187:863Ð867, Lewine JD, Davis JT, Orrison WW Jr: Clinical magnetic source imaging: integration of MEG, EEG and MRI in a hospital setting, in Witte H, Zweiner U, Schack B, et al (eds): Quantitative and Topological EEG and MEG Analysis. Third International Hans Berger Congress. Erlangen: Druchhaus Mayer Verlag, 1997, pp 72Ð79 9. Lewine JD, Morrell F, Orrison WW Jr: Neuromagnetic characterization of normal and abnormal brain functions in epilepsy. Int J Neuroradiol 1:182Ð198, Martin NA, Beatty J, Johnson RA, et al: Magnetoencephalographic localization of a language processing cortical area adjacent to a cerebral arteriovenous malformation. Case report. J Neurosurg 79:584Ð588,

8 J. D. McDonald, et al. 11. Morioka T, Yamamoto T, Mizushima A, et al: Comparison of magnetoencephalography, functional MRI, and motor evoked potentials in the localization of the sensory-motor cortex. Neurol Res 17:361Ð367, Mueller WM, Yetkin FZ, Hammeke TA, et al: Functional magnetic resonance imaging mapping of the motor cortex in patients with cerebral tumors. Neurosurgery 39:515Ð521, Ojemann GA: Cortical organization of language. J Neurosci 11:2281Ð2287, Rostomily RC, Berger MS, Ojemann GA, et al: Postoperative deficits and functional recovery following removal of tumors involving the dominant hemisphere supplementary motor area. J Neurosurg 75:62Ð68, Sartorius CJ, Berger MS: Rapid termination of intraoperative stimulation-evoked seizures with application of cold RingerÕs lactate to the cortex. Technical note. J Neurosurg 88:349Ð351, Skirboll SS, Ojemann GA, Berger MS, et al: Functional cortex and subcortical white matter located within gliomas. Neurosurgery 38:678Ð684, Sobel DF, Gallen CC, Schwartz BJ, et al: Locating the central sulcus: comparison of MR anatomic and magnetoencephalographic functional methods. AJNR 14:915Ð927, Uematsu S, Lesser R, Fisher RS, et al: Motor and sensory cortex in humans: topography studied with chronic subdural stimulation. Neurosurgery 31:59Ð72, Yetkin FZ, Papke RA, Mark LP, et al: Location of the sensorimotor cortex: functional and conventional MR compared. AJNR 16:2109Ð2113, 1995 Manuscript received June 24, Accepted in final form October 9, Address reprint requests to: Jeffrey D. McDonald, M.D., Ph.D., Department of Neurosurgery, University of Utah School of Medicine, 50 North Medical Drive, Salt Lake City, Utah jdm@suzy.med.utah.edu. 598 J. Neurosurg. / Volume 90 / March, 1999