Epilepsy detection in EEG using firefly based feature selection and random forest classifier

Abstract

            An epileptic seizure causes by the abnormal synchronization of patients’ cerebral cortex which is identified by using the electroencephalograph (EEG) signals. The detection of epilepsy is necessary because around 1% of the people in the world are suffering from epilepsy. The detection of epilepsy from the EEG signals characteristics is difficult because of the chaotic and non-stationary nature of the EEG signals. In this paper, the detection of the epilepsy is based on the extracted features of the Variational Mode Decomposition (VMD). Here, the Firefly Optimization (FFO) based feature selection is used to improve the classification of the epileptic seizure pattern. The fitness function considered in the firefly algorithm is the error rate of EEG signals. The Random Forest Classifier (RFC) is used for classifying the epileptic and non-epileptic signals. The proposed methodology is named as VMD-FFO-RFC. The performance of VMD-FFO-RFC methodology is analyzed in terms of sensitivity, specificity and classification accuracy. The performance of VMD-FFO-RFC methodology is compared with an existing method EMD-DWT-KNNThe accuracy of the VMD-FFO-RFC methodology is 95%, which is more compared to the EMD-DWT-KNN.