Abstracts

Rationale: We compared distributed source analysis to dipole analysis of MEG in patients with epilepsy and MRI diagnosed FCD for evaluation of the irritative zone. Methods: We recorded simultaneous EEG/MEG in 5 patients with FCD, with a Neuromag VectorView (306 MEG sensors, 70 EEG electrodes) or a VSM-CTF (271 MEG sensors, 64 EEG electrodes) MEG system. Data was acquired for 40 to 60 minutes and for each patient we selected one run (4-6min) for visual identification of spikes, which were marked, classified according to spatial distribution/morphology and averaged. Only MEG signals were analyzed. The forward model consisted of a single layer boundary element model generated from the outer-brain surface segmented from each patient’s anatomical MRI and co-registered with MEG data using fiducial points. Dipoles (ECDs) were fitted to the largest peak of the MEG signal (Xfit/MRIlab Neuromag soft) using a Goodness of Fit>70% and a dipolar moment<400nA.m. Three distributed sources models were used to calculate current estimates at the same averaged spikes: Minimum Norm Estimate (MNE), its normalized extension, dynamic Statistical Parametric Mapping (dSPM) and Maximum Entropy on the Mean (MEM). Distribution of activity in both hemispheres and relationship of its maximum to the lesion, was evaluated using a cortically constrained surface derived from FreeSurfer.Results: Two (n=3) or 3 (n=2) spike types/patient were analyzed, giving a total of 12 studies (spikes): number of spikes/average ranged from 4 to 22 (mean=10, median=11). In 4/5 patients at least one study’s ECDs were located within the lesion: in 2 patients, ECDs for the 2 studies were located within the lesion/perilesional area; in 1 patient 2/3 studies showed lesional/perilesional ECDs; in 1 patient, ECDs for 1 study was lesional, whereas for the other 2 studies were contralateral homologous to the lesion. Only 1 patient, with 2 studies, had ECDs in other distant areas. MNE, dSPM and MEM showed localization in the lesion topography in 8/12 studies. For only 1 study related to spikes contralateral to the lesion none of the estimates showed involvement of the lesion/perilesional areas. In the majority of studies, the maximum activity was indeed located in an area involving the lesion (although not always in the same part of it), with extension to other areas. Overall concordance between the 3 estimates was found within the lesion in 10/12 studies, within the perilesional areas in 7/12 studies and extra-lesional areas in 5/12 studies. In 4/5 studies with ECDs outside the lesion/perilesional areas, involvement of the lesion was demonstrated with the distributed methods.Conclusions: We successfully used distributed source analyses of epileptiform discharges on MEG to disclose epileptic networks. The different estimates complement each other in the identification of the irritative zone and provides converging evidence of source localization. Whereas ECD, MNE-dSPM and MEM rely on different assumptions, concordant localization increases the confidence of the location of the epileptic discharges.