Abstracts

Rationale: Given surgical success rates for treatment of localization related epilepsy, methods for localization of seizure onset remain of interest. While there remains debate on the critical role of the epileptic network, surgical successes and failures argue that the process of seizure localization is important, and thus information on defining candidate regions of seizure onset is constructive. In addition to semiology, clinical imaging (structural MRI, PET), ictal SPECT studies, MEG and MR spectroscopic imaging (MRSI) are noninvasive functional tests that are believed to evaluate different properties of the epileptic brain and thus are of interest to compare. In this report we describe n=15 surgical patients who underwent both MEG and MRSI and assess their overlap locations of abnormality. Methods: Adult with localization related epilepsy undergoing surgical evaluation at the UPMC Comprehensive Epilepsy Center were recruited for study. Consensus evaluation was based on semiology, scalp EEG monitoring, MRI, PET and ictal SPECT as available. 15 patients were identified to have completed both MEG and large volume MR spectroscopic imaging (MRSI). MEG: An Elekta Neuromag MEG system comprises of 102 triple sensors (102 magnetometers and 204 planar gradiometers) in a helmet-shaped array covering the entire scalp was used as previously described (Bagic 2011). Participants were recorded in supine position. The locations of the nasion, two preauricular points, and the four HPI coils were digitized prior to each MEG study using a 3D-digitizer (ISOTRAK; Polhemus, Inc., Colchester, VT) to define the subject-specific Cartesian head coordinate system. MEG data were acquired at a sampling rate of 1 kHz, with on-line filtering of 0.10 - 330 Hz, using a 20min acquisition period while the subject's eyes were closed. Analysis was performed using Graph and X-fit Programs (Elekta Neuromag Oy, Helsinki, Finland), according to the standard clinical routine (Bagic 2011). : MRSI: Siemens 3T Trio and 7T Magnetom systems were used. Large volume moderate echo (TE 40ms) MR spectroscopic imaging of NAA, creatine and choline was performed at 3T (fronto-parietal) and 7T (temporal) regions, methods as previously described (Pan 2013, Schirda 2015). The nominal voxel sampling was ~0.5cc in fronto-parietal-temporal regions (acquired as 2 slabs), and 1cc temporal (acquired as 2 slices). T1 weighted MPRAGE structural acquisitions were performed for anatomical localization and Freesurfer was used for tissue parcellation and segmentation. Semi-automated spectral analysis was performed using LCM, with significance testing performed on NAA/Cr after tissue correction for fraction gray matter content. Results: Out of the n=15 patients, consensus localization defined n=8 temporal and n=7 extra-temporal patients. From the n=8 temporal patients there was regional overlap in identified regions in 7 patients between the MEG and the MRSI, with one patient being inverted in severity. From the n=7 extra-temporal patients there was overlap in 2 patients. However, of these 7 patients, 3 patients did not display conclusive MEG abnormalities. Conclusions: The identification of dysfunctional regions in the temporal lobe patients appears to be more consistent between the MEG and MRSI than in the extra-temporal patients. More patients are needed for evaluation but thus far the nature of the overlap is consistent with a network distribution that is more clearly identified in the temporal lobe patients. Funding: Funding is provided from NIH NS090417 and EB011639. References: (1) Bagic AI et al. J Clin Neurophysiol. 2011 Aug;28(4):348-54; (2) Pan JW et al Epilepsia 2013 54(9):1668-78; (3) Schirda CS et al Magn Res Med 2015 Epub ahead of print