Abstracts

Rationale: Insular epilepsies are difficult to recognize clinically and by standard localization techniques, especially in nonlesional cases. The insula being located deep in the Sylvian fissure, scalp EEG is unable to detect insular spikes, unless they project to a broader area on the surface. Functional magnetic resonance imaging (fMRI) is an emerging non-invasive imaging technique that has the potential to help localize the epileptic focus by assessing hemodynamic changes associated with spikes. Objective: To assess the value of EEG-fMRI in identifying operculo-insular epilepsies. Methods: EEG-fMRIs of patients with operculo-insular epilepsies (confirmed by intracerebral EEG, the presence of an epileptogenic lesion or a MEG cluster of source localizations in the operculo-insular region) were analyzed retrospectively. EEG-fMRIs were acquired in 8 sessions of 6 minutes. EEGs were recorded on 64 electrodes according to the 10-20 system at 5 kHz with BrainCap and BrainRecorder. T2*-weighted fMRI BOLD were acquired with a 3T Philips scanner. EEGs were filtered with BrainVision Analyzer 2 software and then marked. A pipeline of general linear model (GLM) analyses was carried out, each with different assumptions regarding the hemodynamic response function and inclusion of confound variables. Results: 16 patients with operculo-insular epilepsies underwent a presurgical EEG-fMRI study. GLM analysis could not be carried out on 7/16 (44%) patients as no spikes were seen on EEG traces after removal of the MR gradient and cardioballistic artifacts. Of these 7 patients with no spikes, 3 had frequent spikes during prior long-term video-EEG monitoring but they were of low amplitude and very focal (limited in spatial distribution) and 4 had no or scarce spikes during long-term EEG monitoring. Five of these 7 patients with no spikes on the EEG-fMRI study eventually had an intracerebral EEG study, all of which revealed very active insular epileptiform discharges. Out of the remaining 9/16 (56%) patients with spikes which could be marked, BOLD activation was noted over the insula in 1/9 (11%), the insula + overlying perisylvian cortex in 2/9 (22%) and perisylvian cortex in 4/9 (44%). Two patients had no insular or perisylvian BOLD activations. Distant activations were noted in 7/9 (77%) patients in various regions (post-central gyrus, superior parietal lobule, middle or superior frontal gyri, anterior cingulate or medial frontal gyri). Conclusions: The value of EEG-fMRI in operculo-insular epilepsies is limited by the absence of spikes to be marked in several patients, either because of the inability of scalp EEG to detect spikes from the deeply-seated insula or because of their low amplitude rendering them no longer identifiable after filtering of EEG traces. For patients with visible spikes, insular but more frequently perisylvian BOLD activations were observed. The latter findings suggest that one should consider sampling operculo-insular areas in the presence of insular or perisylvian BOLD activations.