Detection of Epileptic Seizure on EEG Signals using EMD and ICEEMDAN

Abstract

Elliptic seizure is identified and analyzed on Electroencephalogram (EEG) signals using different methods. In this research, EMD and ICEEMDAN are used to detect seizure and non-seizure signals by decomposing the EEG signals into modes called intrinsic mode function (IMF). EMD and ICEENDAN methods are applied to analyze EEG signals and extract its features, thus classification of these features is performed. In the process of classification, Linear Support Vector Machine (L-SVM) used which works on analyzing and identifying these signals. First of all, the ability of both (RMSE) Root Mean Squire Error and (PCC) Pearson Correlation Coefficient methods are compared via evaluating IMFs. Secondly, all extracted features based on EMD and ICEEMDAN were classified to obtain acceptable results in detecting seizures and nonseizure signals. The results achieved here are 100% for all accuracy and sensitivity using 14 trial for both EMD and ICEEMDAN. Its 99.97% and %99.95 in accuracy and sensitivity using ICEEMDAN while it is 98.975% and 97.95% in EMD respectively.