Seizure onset zone localization from many invasive EEG channels using directed functional connectivity

Abstract

In this study we investigated how directed functional connectivity can be used to localize the seizure onset zone (SOZ) from ictal intracranial EEG (iEEG) recordings. First, simulations were conducted to investigate the performance of two directed functional connectivity measures, the Adaptive Directed Transfer Function (ADTF) and the Adaptive Partial Directed Coherence (APDC), in combinations with two graph measures, the out-degree and the shortest path, to localize the SOZ. Afterwards the method was applied to the seizure of an epileptic patient, recorded with 113-channel iEEG and localization was compared with the subsequent resection that rendered the patient seizure free. We found both in simulations and in the patient data that the ADTF combined with the out-degree and shortest path resulted in correct SOZ localization. We can conclude that ADTF combined with out-degree or shortest path are most optimal to localize the SOZ from a high number of iEEG channels.