Epileptic EEG Identification Based on Deep Bi-LSTM Network

Automatic seizure detection has significance for the diagnosis of epilepsy and the reduction of massive workload for neuroscientists. In the analysis of electroencephalogram (EEG) to achieving the goal of detecting seizure activity automatically, in this paper, we proposed a novel algorithm. First, the original signal is filtered into a band-pass filter and the linear features are calculated; Then, deep neural network - bidirectional long-short term memory network (Bi-LSTM) is used to train and classify features. Ultimately, obtained the specificity is 97.14%, the sensitivity is 100%, the G-mean is 98.56% and the average accuracy is 98.56%. The experimental results show that the proposed method has good performance and potential for clinical application.