International Journal of Pharma and Bio Sciences
ijpbs.net
editorijpbs@rediffmail.com (or) editorofijpbs@yahoo.com (or) prasmol@rediffmail.com
10.22376/ijpbs.2019.10.1.p1-12
Volume 8 Issue 2
2017 (April - June)
Neural network based classification of eeg signals for diagnosis of autism spectrum disorder
Autism Spectrum Disorder(ASD) and its diagnosis is one of the streams where extensive researches are being carried out. Early diagnosis of autism can lead to better analysis and treatment of the person who suffers from autism. The idea that the same regions of brain are activated at or near the same time when the processing and/or execution of a specific task occurs is the basis for research using several forms of neuro-imaging and recording technology. As a step to develop an autism diagnosis system, 4 AR feature extraction algorithms are implemented on Electroencephalogram (EEG) signals of normal and autistic children. Moreover, 2 neural networks namely Cascade forward back propagation neural network model (CFBPNN) and Elman neural network (ENN) are used to identify the combination with highest classification accuracy. Network and Subject based classification is performed on the dataset. AR Burg and ENN combination were found to have the highest classification accuracy rate of 95.63%. Based on this research, a Human Machine Interface can be designed for the diagnosis of ASD.
LAXMI RAJA, M. MOHANA PRIYA
Autism Spectrum Disorder, Electroencephalogram, Auto Regressive Features, Elman Neural Network, Cascade Forward Back Propagation Neural Network
1020-1026