Abstract—This paper presents a study which was done in an attempt to discriminate between two motor actions; eyes-open task and eyes-closed task, for two classes; Autism Spectrums Disorders (ASD) and Typical or Normal (TP). Both of these groups were composed of school children with ages between 6 to 9 years. Utilizing the Time Different of Arrival (TDOA) approach applied with raw Electroencephalography (EEG) data for feature extracted in time domain. For each action, specific features were calculated and a Multilayer Perception (MLP) based Neural Network was used to classify the data into the two classes. The classification process was carried out for three scenarios for each group; first, all task for both group were combined together, second, eyes-open were classified for both groups separately, and third, eyes–closed was classified separately. The results show accuracy over 90 % and clearly discriminate for the features.

Index Terms—Multilayer perceptron (MLP) classifier, source location, ASD detection, EEG signals, angular displacements, motor actions.

Uvais Qidwai is with Computer Science and Engineering Department, Qatar University, Qatar (e-mail: uqidwai@qu.edu.qa).
Wafaa khazaal Shams is with Computer Science Department, International Islamic University, Malaysia (e-mail: wafaa_dth@yahoo.com, abdulwahab@iium.edu.my).

Cite:Uvais Qidwai and Wafaa khazaal Shams, "A Source-Discrimination Approach for Detection of ASD Using EEG Data," International Journal of Bioscience, Biochemistry and Bioinformatics vol. 3, no. 5, pp. 492-496, 2013.