Title: Artificial neural network model for detection and classification of alcoholic patterns in EEG
Authors: Bollampally Anupama; Somayajulu Laxmi Narayana; K.S. Rao
Addresses: Department of ECE, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Vijayawada, Andhra Pradesh, 522502, India; B.V. Raju Institute of Technology, Narsapur, Medak, Telangana, 502313, India ' Department of ECE, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Vijayawada, Andhra Pradesh, 522502, India ' Department of ECE, Anurag Group of Institutions, Hyderabad, 500088, India
Abstract: Alcoholism influences the brain function and is one of the major causes for cognitive, emotional impairments. This work investigates alcohol and normal states by analysing electroencephalogram (EEG) signals recorded from Frontal lobes of brain. A second order finite impulse response (FIR) filter is designed to segregate Delta and Theta waves from EEG data. Empirical mode decomposition (EMD) technique is adopted to extract distinct features like kurtosis, median absolute deviation (MAD) and inter quartile range (IQR), which are given as input to artificial neural network (ANN) for the classification. After consumption of Alcohol, the amplitude and frequency corresponding to delta and theta waves, responsible for inactive and sleep states of the brain are found to be very low in comparison to a normal state. The results indicate that they are potential discriminators of alcoholics and normal with an accuracy of 92%, sensitivity 100% and specificity 83%.
Keywords: EEG; electroencephalogram; empirical mode decomposition; IMF; intrinsic mode functions; ANN; artificial neural network; back propagation algorithm; Hilbert transform.
International Journal of Bioinformatics Research and Applications, 2022 Vol.18 No.1/2, pp.84 - 100
Received: 07 Aug 2019
Accepted: 16 Apr 2020
Published online: 07 Apr 2022 *