A combination of statistical parameters for the detection of epilepsy and EEG classification using ANN and KNN classifie
- PDF / 724,492 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 113 Downloads / 187 Views
ORIGINAL PAPER
A combination of statistical parameters for the detection of epilepsy and EEG classification using ANN and KNN classifier Hemant Choubey1 · Alpana Pandey1 Received: 28 August 2019 / Revised: 13 June 2020 / Accepted: 17 August 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Electrical activity of the brain reads through the technique called as electroencephalography for brain disorder like epilepsy. Epileptical signal is extracted from EEG signal through characteristics defined by statistical parameter like expected activity measurement, sample entropy and Higuchi fractal dimension as an input to a classifier. This paper works on the classification approach of EEG signal into healthy, inter-ictal and ictal signal using k-nearest neighbor and artificial neural network classifier according to the statistical parameter. Accuracy, sensitivity, selectivity, specificity and average detection rate are the performance parameter derived from both the classifier for comparison between k-NN and ANN classifier and also for detection of epilepsy with reduced sets of parameter. Keywords Electroencephalogram (EEG) signal · Levenberg Marquardt (LM) classifier · Epileptic seizure detection (ESD) · K-Nearest neighbor (k-NN) · Artificial neural network (ANN) · Variance, discrete wavelet transform (DWT)
1 Introduction Epilepsy is a neuronal chronic disease in which about 1% of the world population suffered from it .Current seizure prediction algorithm divides into three parts (a) preprocessing of EEG signal and removal of artifacts and other noise (b) extracting parameter from EEG signal (c) classification based on above parameter using k-NN and ANN classifier. Discrete wavelet transform is the prominent technique for detection of epilepsy and classification of EEG signal at different frequency band. The main purpose of using EEG recordings for epileptic seizure detection is, it discovers the inter-ictal and ictal activities, and localizes the region of these activities. Also, it is a kind of biomarker that measures the voltage fluctuations in the brain, which is highly useful for analyzing the epileptic syndrome. This paper organizes statistical parameter like expected activity measurement coefficient of EEG signal using image processing aspects described in paper of Imen et al. [15] in which four statistical EAM coefficients
B
1
Hemant Choubey [email protected]
of 64 × 64 image derived from 4096 samples of EEG signal, three parameters of Higuchi fractal dimension computed in Satapathy et al. [16] paper, and at the last three parameters of sample entropy of the EEG signals based on Song et al. [19] paper for the classification of EEG signals using k-NN and ANN classifier. Classification based on k-NN classifier is superior for predicting seizure signal from the EEG signals consisting of three set of data like healthy, inter-ictal and ictal (seizure) signal in terms of performance parameters like accuracy, sensitivity and specificity.
1.1 Objectives The research objectives of
Data Loading...
