Early diagnosis of epileptic seizures over EEG signals using deeplearning approach

In this study it is proposed a novel automatic epileptic detection algorithm, that consists of different procedures that are EEG signal preprocessing including idle tone removal, clarifying cutoff time points of the preictal period, detecting the HFOs, choosing efficient channels, feature extraction and selection approaches, and a novel training method for deep nets-based information criteria, to provide a decision support system for the early diagnosis and interpretation of epilepsy. With the proposed algorithms, highly performance models were obtained and the model complexity was significantly reduced.