Epileptic EEG Signal Classification Using Machine Learning based Model
Garima Chandel, Sandeep Kumar Saini, Ashish Sharma
Abstract
The incidence of spontaneous seizure attacks caused by neurological illnesses like epilepsy greatly impacts patients’ ability to lead regular lives. EEG has been shown to be the best method for identifying epileptic seizures. From past few years, different computer based epileptic seizure detection methods have been developed using EEG signals. A comparison of current machine learning-based seizure detection techniques is presented in this paper. These methods used various techniques depending on EEG analysis which are the signals captured from the human scalp employing temporal, frequential, and time-frequency based features. Also, scalp and intracranial EEG datasets are also described. The study shows the requirement of practical method for effective diagnosis of epileptic seizures and future possibilities of improvement are also discussed. This paper also proposes a methodology to detect epileptic EEG signals using machine learning approach. The dataset used for this study is benchmark dataset and it has been used so that a comparative analysis can be carried out. The classification between normal EEG and epileptic EEG has been carried out using three classification models; Random Forest (RF), Decision Tree (DT) and Extra Tree (ET). The algorithm performance is measured by three parameters; these are sensitivity, specificity and accuracy. Among all classifier, ET gave best performance, these parameters are 99.85, 99.42 and 99.63 respectively for the proposed method.