Smoothness Priority Approach Based Epileptic Seizure Classification Using ANN

摘要

Seizures due to epilepsy are futile, normal activities of patients are hampered and motor along with sensorial function becomes enfeeble. The Neurologist diagnoses epileptic seizures by visual or manual inspection of Electroencephalogram (EEG) signals. This paper provides a simple classification of epileptic seizures through the smoothness priority approach (SPA) de-trending technique. The SPA technique removes very low frequency, time-varying trending interference from the EEG signal. DWT MATLAB toolbox is used for decomposing the EEG signal. Four important time-frequency domain features termed as relative energy, index of fluctuation, entropy, recursive energy efficiency is extracted. The final classification is performed by an artificial neural network (ANN) and 98% accuracy, 100% sensitivity, 96.2% specificity, 100% precision is obtained.