Epileptic seizure classification using novel entropy features applied on maximal overlap discrete wavelet packet transform of EEG signals

Using electroencephalography for diagnosis of seizure attacks has been in a great attention as it records abnormal electrical activities of the brain. This paper proposes a novel technique for diagnosis of epileptic seizures based on non-linear entropy features extracted from maximal overlap discrete wavelet packet transform (MODWPT) of EEG signals. Discriminative features are selected by a t-test criterion and used for the classification with two different classifiers. The proposed method is evaluated and compared to the previous methods in EEG seizure classification by using a publically available EEG dataset with different healthy and seizure suffering subjects. The obtained results show the superiority of the proposed method over the previous techniques in classification performance.