A Unified Approach for Inverse Problem in EEG and Brain Connectivity with Application to Epilepsy. A Proof of Concept Study

The analysis of brain source connectivity using scalp EEG is conventionally performed in two successive steps: solving the EEG inverse problem then estimating brain connectivity. Such sequential manner is sub-optimal since the results of the latter step are highly subject to the quality of the source localization one. To address this limitation, a proof of concept study on solving the EEG inverse problem and inferring brain functional connectivity in only one single step leading to a new All-in-One approach is considered in this paper in the context of drug- resistant epilepsy. Both minimum norm and smoothness on graph assumptions on the target extended epileptic sources are used to restore the identifiability of the ill-posed EEG inverse problem. Preliminary results on realistic simulated surface epileptic EEG signals confirm the potential of the proposed approach.