Abstracts

Rationale: Magnetoencephalography (MEG) has been typically using modeling interictal activity as equivalent current dipoles (ECDs) to localize epileptic activity. Synthetic aperture magnetometry (SAM) is an adaptive spatial filtering algorithm for MEG. SAM kurtosis (SAM(g2)) provides source locations with excess kurtosis value (steepness) of spikes. To evaluate SAM(g2) in analyzing single/multiple epileptic foci, we applied SAM(g2) in children with intractable neocortical epilepsy. Methods: We analyzed SAM(g2) and equivalent current dipoles (ECDs) in 44 children (mean 9.9 years). We used whole head gradiometer Omega system (151 channels). SAM(g2) calculated kurtosis value (band pass filter, 20-70 Hz) at each voxel (5 mm distance). We selected epileptic voxels of SAM(g2) (evSAM(g2)) with local peak kurtosis higher than half of maximum value and more than 1.0 (background activity). A cluster of ECDs contains ?6 ECDs and a group of evSAM(g2) contains ?3 evSAM(g2), within 1cm distance of each other. We evaluated the concordance of evSAM(g2) with both single and multiple clustered ECDs, with and without lesions. We defined a case as concordant when ? 50 % of grouped evSAM(g2) overlapping with ECD clusters; partially concordant when < 50 %; discordant when there was no overlap. Results: The rate of grouped evSAM(g2) overlapping with ECD clusters ranged from 0 to 100 % with a mean of 73.1%. Thirty-four patients (77.3%) showed concordant with 20 single ECD clusters (11 lesions; 9 non-lesion) and 14 multiple ECD clusters (12 lesions, 2 non-lesion). Seven patients (15.9%) showed partially concordant with 2 single ECD clusters, (2 non-lesions) and 5 multiple ECD clusters (3 lesions, 2 non-lesions). Three patients (6.8%) showed discordant . Eleven patients with single ECD cluster with lesion showed higher concordant rate (mean 88%) than that of 12 patients with a single cluster without lesion (67.4%). Sixteen patients with multiple clusters with lesion also reached higher concordant rate (74.6%) than that of 5 patients without lesion (49.4%). In thirteen patients who underwent surgery, nine patients became seizure free and four had reduction in seizure frequency. Conclusions: SAM(g2) analysis succeeded in localizing epileptic sources correlating with both single and multiple clustered ECDs in patients with neocortical epilepsy. SAM(g2) showed highest concordant evSAM(g2) with the single clustered ECDs secondary to lesion. In multiple clustered ECDs, lesional epilepsy presented highly concordant evSAM(g2) comparing with those without lesion. SAM(g2) can assist ECD especially for the cases with multiple foci to determine epileptogenic hemisphere and surgical candidates.