Autoencoding of long-term scalp electroencephalogram to detect epileptic seizure for diagnosis support system

Ali Emami, Naoto Kunii, Takeshi Matsuo, Takashi Shinozaki, Kensuke Kawai, Hirokazu Takahashi

Abstract: Highlights•Autoencoder (AE) was used to detect seizures as anomalous events from interictal EEG.•Long-term EEG data from subjects with various backgrounds were tested.•For these wide range of data, the AE error increased during detect seizures.•The AE error is a feasible candidate for a universal seizure detection feature.•An unsupervised form of learning in the AE offers practical advantages.