HYpergraph-based Prediction of Hidden Epileptogenic Nodes

Epilepsy is one of the most common neurological disorders in the European Union (EU) and is characterized by unpredictable interruptions of brain functions, known as epileptic seizures. In drug-resistant epilepsy, the most effective path to seizure freedom is epilepsy surgery, which targets the removal of the Epileptogenic Zone (EZ), the brain area responsible for seizure generation. The localization of the EZ can be aided by Stereo-Electroencephalography (SEEG), a presurgical intervention that records brain activity in cortical and subcortical areas via intracranial electrodes. SEEG signals are highly complex to interpret, and the success rate of epilepsy surgery remains below 70% despite recent technological advancements. The identification of new biomarkers of the EZ is thus an essential priority for the medical community and the European Reference Network for Rare and Complex Epilepsies. With this aim, the HYPHEN project will develop a new framework for analysing SEEG data. The project will be among the first to represent SEEG signals as hypergraphs, in which each SEEG channel is modelled as a node and connections between multiple nodes are captured via high-order interactions. This approach will make it possible to identify epileptogenic channels and infer the presence of epileptogenic structures not sampled by the SEEG implant. In the latter cases, HYPHEN will design a graph-completion strategy to predict hidden epileptogenic nodes, enabling epilepsy surgery in scenarios currently considered too complex for treatment. These activities will be conducted at Aix-Marseille University (France), under the supervision of Prof. Bénar, in close collaboration with Prof. Bartolomei, two world leaders in the field of SEEG analysis. Complementary methodological expertise will be provided through a secondment at Northeastern University London (UK), in the laboratory of Prof. Petri, a leading authority in high-order networks and their application to neuroscience.