Handcrafted Features Extraction-Based Epileptic Seizure Classification

Epilepsy diagnosis relies heavily on electroencephalogram (EEG) signals. Having epilepsy means frequent seizures, impaired motor and sensory functions, and a hampered normal lifestyle. Neurologists diagnose epileptic seizures by observing EEG signals visually or manually. In this paper, a deep learning-based model is proposed to classify epileptic seizures using EEG signals. Wavelet-based four time-frequency features are extracted named energy, amount of fluctuation, coefficient of variation, and recursive energy efficiency. Before feature extraction, a Savitzky-Golay filter was applied to denoise and smoothen the EEG signal. The proposed model yields accuracy, sensitivity, specificity, and F score of 99.333%, 99.35%, 99.666%, and 0.993, respectively, in seizure detection.