Affordable Access

Access to the full text

Epilepsy Detection Using DWT Based Hurst Exponent and SVM, K-NN Classifiers

Authors
  • Sharmila, Ashok
  • Madan, Saiby
  • Srivastava, Kajri
Type
Published Article
Journal
Serbian Journal of Experimental and Clinical Research
Publisher
Sciendo
Publication Date
Dec 01, 2018
Volume
19
Issue
4
Pages
311–319
Identifiers
DOI: 10.1515/sjecr-2017-0043
Source
De Gruyter
Keywords
License
Green

Abstract

Epilepsy is a typical neurological issue which influence the focal sensory system and can make individuals have seizure. It can be surveyed by electroencephalogram (EEG). A wavelet based HURST EXPONENT strategy is displayed for the analysis of epilepsy. This strategy deals with the nonlinear analysis of EEG signals. Discrete wavelet transform is used to disintegrate the original EEG signal into specific subbands. The hurst exponent of different sub-bands is employed and then fed into two classifiers, namely SVM and KNN. The highest classification accuracy obtained in the presented work is 99% for healthy subject data versus epileptic data is obtained by SVM. However, the corresponding accuracy between normal subject data and epileptic data using SVM is obtained as 99% and 93% for the eyes open and eyes shut conditions, respectively. The detailed analysis of the methodology and results has been discussed in the paper.

Report this publication

Statistics

Seen <100 times