Abstracts

Rationale: Detection of epileptic discharges in MagnetoEncephaloGraphy (MEG) from background brain activity requires synchronized neuronal activity over a minimum of 4 cm2, as estimated when compared with ElectroCorticoGraphy data (Agirre-Arrizubieta Z. et al Brain 2009, Oishi et al Epilepsia 2002). We proposed and evaluated the principle of the Maximum Entropy on the Mean (MEM) (Grova et al. Neuroimage 2006, Lina et al. IEEE TBME 2012), as a distributed sources localization method able to accurately recover the generators of epileptic spikes together with their spatial extent along the cortical surface. The purpose of this study is to assess how the spatial extent recovered from MEG sources could be validated using intracranial EEG (iEEG) recordings. In order to take into account the limited spatial sampling of implanted depth electrodes, we propose to simulate iEEG potentials from MEG sources, to assess what part of MEG sources is seen by iEEG contacts. Methods: We evaluated 4 patients with drug-resistant focal epilepsy who had a pre-operative MEG acquisition and who underwent iEEG with MRI-compatible electrodes for precise registration of recording contacts. Three patients had right orbitofrontal (ROF) spikes and one patient had spikes in the right supplementary motor area (SMA). MEG was acquired in a CTF System equipped with 275 axial gradiometers. Similar spikes were marked and averaged in both MEG and iEEG recordings. Averaged MEG spikes were localized, using MEM, along the cortical surface segmented from an anatomical MRI. This MRI was subsequently co-registered with the MRI obtained with iEEG electrodes in place, for precise identification of every iEEG contact. An iEEG forward model estimating the influence of every dipolar source of the cortical surface on each iEEG contact was then estimated. Applying this iEEG forward model to MEG sources, we simulated the iEEG potentials that would have been generated by these MEG sources estimated along the cortical surface. MEG-simulated iEEG potentials were compared with real iEEG signals at the peak of the averaged spikes. Results: Excellent MEG/iEEG correlation was found over few lateral ROF iEEG contacts for Patient #1 (Figure 1) and #2, and over several contacts of three SMA electrodes for patient #4 (Figure 2). For patient #3, the main generator identified at a deep mesial ROF iEEG contact was not found in MEG sources. Moreover, MEG-simulated iEEG data showed that the main right frontal lateral MEG source could not have been recorded in iEEG. However, an early generator (half the peak of the spike) was identified in iEEG and MEG-simulated iEEG data over few lateral ROF contacts. Conclusions: Precise location of active iEEG contacts allows optimization of the anatomical correlation between MEG and iEEG findings. We demonstrated that simulating iEEG potentials from MEG sources allows quantifying (1) what percentage of the cortical surface was covered by each patient-specific iEEG implantation and (2) whether the spatial extent of spike generators estimated in MEG could be validated using iEEG.