CCURATE identification of zones of seizure origin in patients with localization-related neocortical epilep- sy remains one of the greatest challenges to success- ful epilepsy surgery. In the absence of a clear-cut anatom- ical lesion such as a vascular malformation, tumor, or cor- tical migration defect, discrete focus localization typically requires the use of a wide variety of noninvasive tech- niques, as well as invasive EEG monitoring. Various imag- ing modalities have been developed to refine this localiza- tion problem. These methods include MR imaging, 8,11,18 MR spectroscopy, 4,11,13,14,16 PET scanning, 5,17 ictal and interictal SPECT scanning, 3,24,25 and a variety of sophisticated EEG dipole models. 2,10,20,21,23,28,29 Although each of these meth- ods has particular strengths, no single noninvasive tool has J. Neurosurg. / Volume 97 / October, 2002 J Neurosurg 97:865–873, 2002 Magnetoencephalography-directed surgery in patients with neocortical epilepsy ADAM N. MAMELAK, M.D., NANCY LOPEZ, R.E.E.G.T., MASSOUD AKHTARI, PH.D., AND W. WILLIAM SUTHERLING, M.D. Huntington Medical Research Institutes, Pasadena; Epilepsy and Brain Mapping Program, Huntington Memorial Hospital, Pasadena; and City of Hope National Medical Center, Duarte, California Object. Magnetoencephalography (MEG) and magnetic source (MS) imaging are techniques that have been increas- ingly used for preoperative localization of epileptic foci and areas of eloquent cortex. The use of MEG examinations must be carefully balanced against the high cost and technological investments required to perform these studies, par- ticularly when less expensive alternative localization methods are available. To help elucidate the value of MEG, the authors have critically reviewed their experience with whole-head MEG in the case management of patients undergo- ing epilepsy surgery. Methods. The authors identified 23 patients with suspected focal epilepsy who underwent whole-head MEG and MS imaging at Huntington Memorial Hospital and, subsequently, underwent invasive intracranial electrode monitor- ing and electrocorticography (ECoG) to localize the zone of seizure origin for surgical resection. The results of the MS imaging were retrospectively stratified into three groups by the number of interictal spikes recorded during a 4-hour recording session: Class I (no spikes), Class II ( five spikes), and Class III ( six spikes). Class III was further subdi- vided according to the clustering density of the interictal spikes: Class IIIA represents a mean distance between interic- tal spikes of 4 mm or greater (that is, diffusely clustered) and Class IIIB represents a mean distance between interictal spikes of less than 4 mm (that is, densely clustered). The authors analyzed these groups to determine to what extent the results of MS imaging correlated with the ECoG-determined zone of seizure origin. In addition, they assessed whether the MS imaging study provided critical localization data and correlated with surgical outcome following resec- tion. A statistical analysis of these correlations was also performed. Of the 40 patients studied, 23 underwent invasive monitoring, including 13 with neocortical epilepsy, four with mesial temporal lobe epilepsy, and six with suspected neocortical epilepsy that could not be clearly localized by ECoG. Depth electrodes were used in nine cases, subdural grids in nine cases, depth electrodes followed by subdural grids and strips in four cases, and intraoperative ECoG in one case. Electrocorticography was able to localize the zone of seizure origin in 16 (70%) of 23 cases. In 11 (69%) of the 16 cases in which ECoG was able to localize the zone of seizure origin, the interictal spikes on the MS images were classified as Class IIIB (densely clustered) and regionally corre- lated to the MS imaging–determined localization in all cases (that is, the same lobe). In contrast, no Class IIIB cases were identified when ECoG was unable to localize the zone of seizure origin. This difference showed a trend toward, but did not achieve, statistical significance (p  0.23), presumably because of the relatively small number of cases available for analysis. In three cases (all Class IIIB), MS imaging was used to guide invasive electrodes to locations that otherwise would not have been targeted and provided unique localization data, not evident from other imaging modalities, that strongly influenced the surgical management of the patient. The classification of findings on MS im- ages into subgroups and subsequent statistical analysis generated a model that predicted that Class IIIB MS imaging data are likely to provide reliable information to guide surgical placement of electrodes, but all other data groups do not provide localization information that is reliable enough to guide surgical decision making. Conclusions. Magnetic source imaging can provide unique localization information that is not available when other noninvasive methods are used. Magnetic source imaging appears most useful for cases of neocortical epilepsy. In par- ticular, when an MS imaging study revealed six or more interictal spikes that were densely clustered in a single ana- tomical location, the MS image was highly correlated with the zone of seizure origin identified by ECoG. In these cases the MS imaging data may be useful to guide placement of intracranial electrodes. KEY WORDS • magnetoencephalography • magnetic source imaging • electrocorticography • epilepsy • cortical mapping A 865 Abbreviations used in this paper: ECoG = electrocorticography; EEG = electroencephalography; MEG = magnetoencephalography; MR = magnetic resonance; MS = magnetic source; PET = positron emission tomography; SPECT = single-photon emission computer- ized tomography.