Automated Feature Extraction of Epileptic EEG Using Discrete Wavelet Transform and Approximate Entropy

Abstract

The disease epilepsy is characterized by a sudden and recurrent malfunction of the brain that is termed seizer. The electroencephalogram (EEG) signals play an important role in the diagnosis of epilepsy. Nonlinear analysis quantifies the EEG signal to address randomness and predictability of brain activity. In this study, the wavelet subband decomposition and Approximate Entropy (ApEn) is used for epilepsy detection from EEG signals. In first stage, EEG signals are decomposed into five EEG subbands viz. delta, theta, alpha beta and gamma, using Discrete wavelet transform (DWT). The second stage consists of the feature extraction of EEG using ApEn. The methodology is applied to two different EEG signals: 1) Normal 2) Epileptic. For each subband ApEn is calculated and it is observed that the each EEG subband value of ApEn drops during an epileptic seizures. Accuracy is calculated by using thresholding. Classification accuracy is determined by applying thresholding. The overall accuracy as high as 96% is achieved for EEG subbands as compared to the without wavelet decomposition accuracy value is 86%.