Automated classification of epileptic electroencephalogram signals using wavelet entropies and energies

Epilepsy is a chronic electrophysiological disorder characterized by repetitive seizures. Electroencephalogram (EEG) signals represent the dynamic condition of the brain. Variations in the EEG signals cannot be predicted by ocular examination. Thus, signal processing methods are used to extract the hidden information from the EEG signals. In this work, three different types of EEG signals (normal, epileptic, background) are used and features are extracted using wavelet entropies and energies. ANOVA is performed followed by classification. A graphical user interface (GUI) is introduced that can help in the automatic classification of the EEG signals. Various classifiers such as decision tree, k-nearest neighbour, support vector machine and fuzzy classifiers are used. Classification accuracy of 97% is achieved using decision tree classifier with GUI.