Abstracts

Rationale: For surgical candidates with medically intractable epilepsy, it is critical to detect and localize the seizure focus for surgical treatment. SAM(g2) analysis is a novel MEG spatial filtering technique based on a constrained minimum variance beamformer, synthetic aperture magnetometry (SAM), and the excess kurtosis (g2) statistic. This technique evaluates for locations of increased 'spike-like' activity compared to local background activity, and gives source waveforms for SAM virtual sensors (SAM-VSs) at those locations. Although source waveforms estimated by SAM have higher signal-to-noise ratio (SNR) than raw MEG, it is possible that abnormal cortical irritability and/or continuous spiking would be hidden from SAM(g2) by mimicking a 'noisy' background or poor SNR. Since SAM estimates the time-course of source activities at specific locations, SAM-VSs can be arbitrarily added to evaluate visually for 'spikiness'. The purpose of this study was to evaluate the clinical utility of SAM-VSs added as virtual '3D-grids' in the brain to detect regions of higher 'spikiness' than detectable by SAM(g2) alone. Methods: Seven patients with intractable epilepsy, with or without lesion on MRI, were studied presurgically with 275-channels MEG (VSM MedTech, Port Coquitlam, Canada) at 300Hz and/or 2 kHz decimated from 4 kHz sampling rate. At least 30 minutes recording for each patient were analyzed using SAM(g2). SAM-VSs 3D-grid positions were carefully selected based on the locations of the original peak SAM(g2) detections and MRI lesion, if present, and computed at 5 mm voxel-spacing between 420 (7×10×6) to 4056 (12×13×26) virtual sensors. SAM-VSs source waveforms were inspected carefully visually. Timing and frequency of VS spike onsets were compared to spike onset in the raw EEG and MEG signals and to locations of peak 'spikiness' by SAM(g2).Results: The SAM-VSs 3D-grids showed various findings (e.g., earlier onset spike in SAM-VSs than raw EEG or MEG spike onset, drop in spike amplitude between adjacent sulci). The SAM virtual sensor grids revealed regions of very high cortical irritability, not shown on raw MEG recording, adjacent to SAM (g2) locations of high 'spikiness'.Conclusions: The additional visual examination of 3D-grids of closely spaced virtual sensors in regions of increased 'spikiness', previously detected by SAM(g2), may be helpful in better defining the epileptogenic zone.