A Convolutional Long Short-Term Memory Based Neural Network for Epilepsy Detection from EEG

Epilepsy is a severe neurological disorder characterized by recurrent seizures, which increases the risk of death three times more than normal. Currently Electroencephalography (EEG) has emerged as highly promising technique for diagnosis epilepsy. The majority current EEG-based epilepsy detection research have employed variety deep learning-based models, but most approaches suffer from poor generalizability, optimal design and performance rates. To address these issues, this study aims to develop an efficient framework based on spatio-temporal neural network called convolutional long short-term memory (ConvLSTM) EEG signals. In proposed model, firstly standard 19 channel data are selected resampled at 256Hz, then those signals segmented into three-second time frames. Afterward, fed input ConvLSTM model identifying epileptic patients normal subjects. generalize we tested it two different datasets with varying population sizes. We used five-fold cross validation leave-one-out schemes eliminate experiment’s biases. further validate framework, carried out various ablation studies. experimental results demonstrate that outperforms state-of-the-art studied datasets, making suitable use automated system