WAVELET TRANSFORM ANALYSIS (HAAR AND SYM8) FOR EPILEPSY CLASSIFICATION WITH SOFT DISCRIMINANT CLASSIFIER

Epilepsy is one of the prominent and disturbing neurological disorder and many people across the world are victims of this problem. The sudden motor disturbances in the brain cause and trigger these seizures. Due to the hypersynchronous discharges happening on the cortical regions of the brain, the activities of the motor becomes abnormal and so seizures are triggered. The seizures caused due to epilepsy are quite heterogeneous in nature and so diagnosing it is quite challenging. Electroencephalography (EEG) is the most widely used instrument for the detection of epileptic seizures. In this work, Haar and Sym8 wavelets are employed to extract the wavelet features at level 4 from EEG signals. The extracted features like alpha, beta, theta, gamma and delta are classified through the Soft Discriminant Classifier (SDC) to obtain the epilepsy risk level from EEG signals. The final results show that when Haar wavelet is employed and classified with SDC, an average classification accuracy of 95.20% is obtained and when Sym8 wavelet is utilized and classified with SDC, an average classification accuracy of 94.68% is obtained.