Sequential Segmentation of EEG Signals for Epileptic Seizure Detection using Machine Learning

Abstract

The problem of epilepsy has grown exponentially and is now considered as one of the most prevailing neurological disorders affecting around 50 million people around the globe. Epilepsy is identified by analyzing the interictal activity present in the EEG signal. Visual analysis of EEG is a tedious process and subject to human error. This work proposes a robust method to ease the burden of intractable seizures by automatic recognition of ictal epileptiform activity in the EEG of epileptic patients. The classification between EEG having an epileptic seizure and non-seizure is done using various machine learning algorithms. The classifiers used are Simple Decision tree, Quadratic Discriminant, Medium Gaussian SVM, Bagged Trees, and Subspace k-NN. The performance is assessed using 10-fold cross-validation.