Electroencephalography-based Neural Network Approach for Epileptic Seizure Detection using Spatial Features

Epilepsy is a persistent dysfunction of the brain and nervous system that causes seizures and affects patients every day. Epilepsy affects more than 2% of people worldwide. Electroencephalogram (EEG) signals continue to play an essential role in diagnosing of patients with epileptic disorders. This article classifies EEG Signals for epilepsy through the Convolutional Deep Neural Network. The EEG signal is first decomposed and converted into a two-dimensional time-frequency image. Using the heuristic mode decomposition format, the EEG signal is divided into a specific category of band-limited signal defined as "intrinsic mode functions (IMFs)". According to this process, data from stable and epileptic patients are classified as IMFs. The time rate function is defined using the support vector machine (SVM), the k-nearest neighbor (k-NN), and ensemble classification using the AlexNet model. The results of this analysis indicate that the new EEG and deep neural network-based detection system is currently the most powerful way to identify epileptic seizures.