Abstracts

Rationale: The objective of our study is to compare the localization of epileptic spikes using dipolar and distributed source imaging methods against interictal spiking regions and ictal onset zones identified on intracranial EEG studies in the same patients. Methods: : We retrospectively identified 20 patients with medically refractory epilepsy who underwent MEG followed by invasive intracranial EEG recordings for presurgial evaluation. Interictal epileptic spikes observed in the MEG recordings of these patients were localized using both dipolar and distributed source methods. Based on visual inspection of the resulting localizations of the interictal spikes, the studies were grouped into 3 categories: 1)those where the methods agree2) those where one or the other method failed to provide satisfactory models3) those where the results from the two methods are substantially different. Cortical regions identified as generators of both interictal spikes and seizures on intracranial EEG were compared to regions identified by the two source modeling techniquesResults: Our patient group had a mean age of 34.2 years (+/-14.27) with a female/male gender ratio of 13/7. Category 1: In 15/20(75%)of patients dipolar and distributed sources were concordant.In 11/15 (73.3%) patients with concordance between dipoles and distributed source models, localization was consistent with intracranial interictal spikes and ictal onset zones.In 1/15, patient both source models agreed but were discordant with the ictal onset zone and intracranial interictal data. 2 other patients with dipoles and distributed concordance had multifocal findings, and intracranial EEG identified additional areas of interictal spiking and seizure generation.Category 2: In 3 patients dipolar modeling was deemed unsatisfactory due to poor goodness of fits and large confidence volumes. Distributed modeling in all three yielded results consistent with both intracranial interictal and ictal findings. Category 3: In 1 patient with very high amplitude spikes distributed modeling identified broad temporal and temporo-parietal regions of spike generation. Dipolar sources consistently localized to a small volume in the medial mid temporal region, but with low goodness-of-fit despite large signals.iEEG confirmed a large temporo-parietal spike and seizure generator. In the final patient, dipolar sources were broadly distributed in the fronto-parietal deep white matter, while distributed sources identified a generator in right posterior frontal lobe related to deep sulcal dysplasia.Conclusions: : In majority of patients, dipolar and distributed source models of spikes recorded in MEG were concordant and predicted the areas of spike origin as well as seizure onset zones identified subsequently with intracranial EEG. In instances where dipolar modeling of spike events is found to be unsatisfactory, distributed source modeling of the events may provide useful localizing information relevant to identifying epileptogenic networks.