Abstracts

Rationale: There is limited literature comparing magnetoencephalography (MEG) to functional MRI (fMRI). We compared magnetoencephalography (MEG) and functional magnetic resonance imaging in (fMRI) language mapping in localization-related epilepsy (LRE) using synthetic aperture magnetometry, a method that localizes changes in neural activity associated with higher cognitive function. Methods: Seven patients with LRE had MEG were scanned using a CTF 275 MEG system and a GE Sigma 3T MRI scanner with a gradient echo-planar sequence while performing an auditory description decision task previously proven to activate frontal and temporal eloquent cortex with fMRI. For the task, subjects hear a definition of a word followed by an answer: patients press buttons indicating the statement’s validity. The control stimuli were reverse waveforms of experimental stimuli; subjects responded indicating the presence/absence of a terminal noise burst. fMRI was a block design (5 cycles, total time 5 minutes), and MEG was an event related design. Task duration was approximately 20 min with 90 experimental and control stimuli, and a 7:3 true/noise:false/no-noises ratio. Synthetic aperture magnetometry (SAM), a minimum-variance beamformer, was used to localize event-related experimental versus control differences in oscillatory power. Estimated source activity spectrograms were visually inspected to identify time-frequency windows for computing SAM volumes. These volumes were used to compute middle temporal gyrus (MTG), middle frontal gyrus (MFG), inferior frontal gyrus (IFG) asymmetry indices (AI)= (|L|-|R|)/(|L|+|R|) using maximum and minimum SAM values. These were averaged to estimate anatomically-specific and overall hemispheric AIs. AI between +/- 0.1 were considered bilateral activation and |AI| > 0.1 were left laterlized. fMRI images were analyzed using SPM2: ROI analysis was performed with a bootstrap method, and an AI calculated, AI was lateralized if |AI| <0.2. Results: MTG and overall AIs showed 100% concordance between fMRI and MEG. The MEG IFG AI showed 89% and the MFG AI revealed a 75% concordance. One subject showed no frontal activation in MEG, 1 patient had right lateralization in IFG, 2 patients had partial discordance and one patient had bilateral activation. One-way ANOVA showed a significant correlation between fMRI and MEG AI for MTG (p<0.05) and a trend towards significance overall (p<0.06). Adjusted R-squared values for linear regressions (y-intercept set to 0) were 0.68 for MTG and 0.32 overall. Conclusions: Our preliminary data suggest that SAM and fMRI concur for assessment of language dominance and that MEG may be useful for predicting language dominance in localization related epilepsy particularly for receptive language cortex in the middle temporal gyrus.