Classification of normal, seizure and seizure-free EEG signals using EMD and EWT

Objectives: Electroencephalogram (EEG) plays an important role in recording the activity of human brain. Identification of epileptic seizures can be done using EEG signals. Methods/ Statistical Analysis: In this work for classification of EEG signals a method known as Empirical mode decomposition (EMD) is used and compared with empirical wavelet transform (EWT) based method. Findings: In this paper the EMD has been considered for five classes of EEG signals. Intrinsic Mode functions obtained for these EEG signals have been shown. The amplitude modulation bandwidth BAM and frequency modulation bandwidth BFM have been calculated. Applications/ Improvements: The classification based on bandwidth features and least square support vector machine (LS-SVM) provided better categorization accuracy than earlier adopted methods. Results have been shown in this report.