Abstracts

Rationale: Limbic networks are of highly importance in the field of epilepsy as they are often involved in the epileptic circuitry. Hippocampus and amygdala in particular are deep structures that are highly prone to epileptic discharges. A few studies have reported that these structures can be recorded in magnetoencephalography (MEG), despite their depth and complex architecture. Simultaneous MEG and intracebral EEG (stereotaxic EEG, SEEG) recordings constitute a powerful way for validating these results based on direct recordings within mesial structures. A key point to keep in mind is that amygdala and hippocampus are often involved in conjunction with other temporal lobe structures in the epileptic network in temporal lobe epilepsy, and often with small propagation delays: it is thus very difficult to ensure that the actual source of MEG signals is the hippocampus or amygdala and not nearby (neocortical) structures.  Methods: To investigate these issues, we analyzed simultaneous MEG (4D Neuroimaging, gradiometers) and SEEG recordings in 8 patients undergoing a SEEG recording for drug resistant epilepsy. Epileptic spikes were marked on electrodes within the hippocampus and the amygdala. Independant component analysis (ICA) was performed on the MEG traces segmented around the SEEG triggers, and also on the continuous signals for comparison. We computed the correlation of single trials ICA time courses with all SEEG traces, either among time (corrcoeff), or at each time point.  Independent component analysis (ICA) was performed on the MEG traces segmented around the SEEG triggers and on continuous recordings.  Results: ICA permitted to disentangle distinct brain networks involving many temporal lobe structures and extra temporal limbic regions (i.e. thalamus, orbito-frontal cortex) thus forming extended or more focal networks (only one to two structures) (Fig 1). ICA components presenting correlation with mesial structures were obtained in 6/8 patients for the analysis triggered on SEEG events, and in 4/8 patients for the analysis on continuous data (i.e., without information from SEEG). Within these components, we observed maximal correlation values (Table 1) with the hippocampus (4), amygdala (2), thalamus (1), insula (1), perirhinal cortex (1) collateral sulcus (1) and the temporal pole (1). Conclusions: Our study confirms that mesial structures from the limbic system, including thalamus, can produce significant signals on MEG sensors. This could help in the comprehension with non-invasive technics of limbic system involvement in the epileptic network.  Future implications are promising for both presurgical evaluation of epilepsy and for cognitive paradigms involving these structures.  Funding: Ligue française contre l'epilepsie (LFCE)