Seeing the EpileptogenIc Zone through machine learning on strUctuRal, functional and clinical nEurological data – SEIZURE

Epilepsy is resistant to antiepileptic drugs for 30% of patients. For those patients, surgical resection of the brain epileptogenic zone (EZ) is the best option. No biomarker allows to perfectly detect the EZ and surgical outcome is not a trivial task. Recent progress in statistical image processing of magnetic resonance (MRI) and positron emission (PET) imaging has allowed the detection and phenotyping of subtle lesions, however, some remain hard or impossible to detect. Magnetoencephalography (MEG) offers the possibility to record brief electrophysiological events which propagate within dynamic neural networks including the epileptic focus. Promising advances in these signals analysis open the prospect to characterize brain dysfunctions induced by epilepsy. The anatomical approach with MRI and PET and the functional approach with MEG have largely evolved in parallel. The clinical integration of the diagnostic information conveyed by these different modalities is still empirically performed by epileptologists. SEIZURE aims to extract the most informative biomarkers from mp-MRI, PET and MEG associated to clinical data and combine them into a single statistical model that will directly provide the EZ localization to the clinicians and predict surgical outcome. The proposed analysis framework will leverage the most advanced statistical machine learning methods for image and signal graph processing as well as fusion of heterogeneous data. SEIZURE is a highly transdisciplinary project combining expertise in epilepsy neuroimaging, clinical epileptology as well as core and applied machine learning for signal and image analysis. It will be based on the multimodal EPIL-IA database (>200 patients) already collected by the clinician partner as well as open datasets (e.g. MELD). SEIZURE will produce innovative methodological tools in ML for neurological data analysis and bring these tools into clinical epileptology for objective identification of the EZ and robust prediction of surgery outcome.