HFD and MCFET Based Feature Extraction Technique for Detection of Epilepsy Using ANN Classifier

Abstract

Listen

Translate article

A neurological disorder called Epilepsy which causes the sudden occurrence of epileptic seizures. The electroencephalogram (EEG) is the recorded electrical activities of the brain to examine the epileptic patient through EEG pattern for diagnosis. Epileptic seizure is one of the abnormality or brain disorder in which seizure patterns shows large spikes for specific time domain or area. This work mainly focused on detecting the Epileptic seizures or Epilepsy through the extracted feature like Higuchi Fractal Dimension (HFD) and Masking and Check-in based feature extraction technique (MCFET). Three scaling features of HFD viz. fractal dimension, the standard deviation of fractal dimension and scaling factor while twenty masking and check-in-based features of the upper and the lower envelope along with ten features of the Discrete Wavelet Transform (DWT) coefficients (Table 1) from raw EEG signals are required as input to the Artificial Neural Network (ANN) for classifications. The overall performance is improved in terms of Accuracy, Sensitivity, Specificity through both HFD and MCFET features. Further, the overall accuracy using HFD and MCFET based feature extraction technique around 98% with a bit of computational time of about 1 second by reducing the training percent from 80% to 60%.