Epileptic Electrophysiological Brain Networks Modeling with Partially Observable Brain Regions

Project Summary Epilepsy is one of the most common neurological disorders, affecting 50 million people worldwide of all ages. Seizures in up to one third of epilepsy patients are drug refractory, where surgery offers a necessary means to resect epileptogenic regions (EZs). However, the outcomes of epilepsy surgery may often be suboptimal as on average approximately 50% of patients are still not seizure free after surgery, mainly due to the EZs cannot be accurately localized. Traditionally, the EZ is clinically determined by using ictal intracranial electroencephalography (iEEG) recordings, which are invasive, and can only cover partial brain regions, and with a waiting period of days to weeks for seizure to occur. In more recent years, it has been recognized by researchers that seizures arise from a coordinated activity across large scale epileptic networks, thus characterizing the connections among the involved epilepsy regions is profound to understand the transmission pathways of epileptic brain network and epileptogenic foci. However, the current ESI frameworks are insufficient to tackle the changes. Firstly, iEEG electrodes can only record from partial brain regions, and it is infeasible to record the whole brain intracranial data due to its invasive nature, and secondly, the existing ESI frameworks suffer from very low accuracy when it comes to the estimation of whole brain electrophysiological networks. In this project, we aim to develop a unified paradigm using a simultaneous multimodal measurement of scalp EEG and iEEG signals to estimate the electrophysiological networks of the whole brain, thus enabling the characterization of seizure onset zones (SOZ) and its key transmission pathways from the level of whole brain source space. We aim to bridge the gap between the less accurate whole brain network reconstruction using source localization with scalp EEG and the more accurate but regional brain networks constructed from iEEG by using a principled machine learning framework and letting the partially observable brain signals to “semi- supervise” the whole brain networks estimation process. We term this new paradigm of source imaging as “electrophysiological networks imaging” (ENI) in contrast to the traditional electrophysiological source imaging (ESI). The proposed framework will enable delineation of brain regions involved in seizure generation and propagation at the whole brain scale and improve understanding of different epilepsy biomarkers from a network perspective and their values in the pre-surgery decision making and surgical outcomes.