Abstracts

Rationale: Determining the site of ictal onset can be difficult in some patients with refractory partial epilepsy. MEG is a diagnostic tool with excellent spatial and temporal resolution that can provide non-invasive interictal localizing information not available from other testing methods. Several MEG analytical methods are used to localize epileptiform activity. We wish to determine if techniques, beyond the standard ECD technique, can provide more accurate localizing information prior to surgical resection. We investigated the MEG data source localization techniques (ECD, Current Distribution, Beamformer, and Coherence) for evaluating a potential epilepsy surgery patient s focus. Methods: Sixty patients with refractory epilepsy were monitored with MEG. Visually identified epileptic spike activity was analyzed with four separate MEG techniques: Single Equivalent Current Dipole (ECD), Current Distribution (MR-FOCUSS), Beamformer (Borgiotti-Kaplan (BKbeam)), and Coherence (ICA-minimum norm). The ECD was performed with 4D Neuroimaging software and the others were all performed in MEG-TOOLS software. Results: Forty patients had 5-10 spikes each identified during MEG analysis. Of the patients who had detectable spikes, 39 had ECD source localizations, 40 had current distribution localizations, 40 had beamformer localizations, only one patient had a poor ECD localization that could not be used. Results from 31 patients using both advanced techniques (MR-FOCUSS and BKbeam) included the ECD localization and depicted a larger region of activity than the ECD symbol. Agreement among these three techniques was determined if the localization results all fell with in a region encompassing a 1 cm sphere. Coherence imaging displayed focal areas in 28 patients, of these concordance was found in 24 with ECD, 27 with current distribution and 25 with beamformer localizations. Conclusions: MEG is a safe and non-invasive technique which provides localization of epileptic areas for presurgical evaluation. Advanced MEG techniques (Current distribution, Beamformer, and Coherence) provide additional interictal epileptiform data than ECD alone, which may be helpful in the decision making process for surgical intervention. Research supported by NIH/NINDS Grant R01-NS30914.