Seizure detection in EGG signal with novel optimization algorithm for selecting optimal thresholded offset Gaussian feature

摘要

EEG analysis is a responsible device in assessing the neurophysiological disorders associated to the postsynaptic movement happened in the neocortex. In detecting the epileptic seizure disorder, the EEG signals of the patient are diagnosed. In this analysis, the occurrence of epileptiform discharges in between the happening of seizure disorder should be identified. The introduction of automatic computerized process is required to analyze the EEG signals for detecting the seizure disorder. For obtaining the accurate classification result, the features should be carefully extracted. Though it is better to take many features in order to obtain clear and detail information of the signal, the classifier performance will be affected when large number of features is applied. Hence the optimal set of features must be obtained using the suitable optimization algorithm. In this work the Divergence State Estimation based Bio-geography based Optimization (DSEBBO) is presented for feature selection process whereas the for feature extraction method, Thresholded Offset Gaussian (TOG) method is employed. The selected optimal features are applied to the SVM classifier and the results are validated for the proposed method by comparing with the other optimization algorithms as well as other feature extraction methods. From the classification results, it is clearly elucidated that the TOG feature extraction and the utilization of DSEBBO based optimal feature selection strategy gives the better accuracy which is 4.64% more than the method that does not employ feature selection technique. (C) 2019 Elsevier Ltd. All rights reserved.