Abstracts

Rationale: Electrical Source Imaging (ESI) and simultaneous electroencephalogram and functional Magnetic Resonance Imaging (EEG/fMRI) are two neuroimaging techniques clinically relevant for presurgical evaluation of patients with drug resistant focal epilepsy. A good level of spatial correspondence between EEG sources and fMRI hemodynamic responses to epileptic discharges has been demonstrated in our previous works, but the results of this study was based on EEG with 56 electrodes recorded outside the Magnetic Resonance (MR) scanner. The goal of this study is to evaluate the feasibility of ESI of epileptic discharges using high density EEG (hdEEG) recorded in the MR scanner. Performing ESI with EEG/fMRI data is a challenging task considering decreased EEG signal quality due to high magnetic fields inside the scanner (gradient artefact, ballistocardiogram). In this study, we compared the spatial concordance of ESI results of epileptic discharges recorded inside and outside the scanner with fMRI responses to those discharges. Methods: Three patients with focal epilepsy underwent two hdEEG acquisitions (256 electrodes, EGI, Eugene, OR): a) outside the MRI scanner and b) simultaneously with fMRI. Similar interictal discharges were marked for both sessions. EEG preprocessing and fMRI statistical analysis were done according to our standard procedure, described in [Bagshaw et al., NeuroImage 2004]. ESI analyses were performed according to [Heers et al., HumanBrainMapping 2014], using the coherent Maximum of Entropy on the Mean (cMEM), a distributed source localization technique sensitive to the spatial extent of the cortical sources. The reproducibility of ESI outside and inside the scanner was tested, and ESI results were compared to the fMRI responses. For patient 1, since the hdEEG/fMRI analysis failed because of too few discharges recorded in the scanner, ESI results were compared to the results of a previous EEG/fMRI session. Results: ESI maps obtained from hdEEG inside and outside the scanner showed good concordance (Figure 1), suggesting that cMEM localization is robust even in noisy environment of the MR scanner. ESI results were spatially concordant with the most significant fMRI positive response in the three patients. In patient 1, the ESI found a maximum in the right frontopolar region, concordant with the peak of fMRI positive response. In patient 2, ESI detected a source in the left temporopolar region which was discordant with the peak of negative response (highest absolute t-value) but close to the peak of positive response in the fMRI. In patient 3, ESI indicated a source in the left superior temporal gyrus, concordant with the peak of positive response in the fMRI. Conclusions: Our results suggest the feasibility of hdEEG to provide accurate source localization even when acquired in the scanner during the hdEEG/fMRI session. This opens a new prospect for combining the high temporal resolution of ESI and the higher spatial resolution of EEG/fMRI analysis and developing fMRI informed by ESI results. Funding: NSERC Discovery grant, FRQNT Research team grant and Savoy foundation