2D affective space model for detecting autistic children motor imitation development disorder

Abstract

Autism Spectrum Disorder (ASD) is a complex developmental disorder that represents abnormality development of brain and its function. Few studies had seen the impact of early detection and intervention, and how it gives positive effect for the children to improve their life as good as other normal children. There are three major impairments of the autistic children namely: social interaction with others, communication and fine motor movements. A simple case of fine motor movements could be finger tapping, imitating other action and also managing their fingers towards objects. Many psychologists and psychiatrists diagnosed these autistic children as early as three years old. In Functional Magnetic Resonance Imaging (fMRI) studies, finding shows the brain region such as prefrontal cortex (especially in the pre-motor cortex), occipital and etc, did involve in motor impairment in autistic children. Thus, it is possible to differentiate between normal and autistic children using motor imitation skills but all these process require the psychologists, psychiatrist or expert to monitor and make decision on their observations. However, fMRI is an expensive process and could not be automated and be used for masses. In this research, it aims to look at motor imitation skills for autistic children using electroencephalogram (EEG) approach. Preliminary result has shown great potential in identifying motor imitation skills impairment in autistic children. EEG device will be used to capture the brain signals, and two methods for analysis such as Gaussian Mixture Model (GMM) and Kernel Density Estimation (KDE). The Mutilayer Perceptron (MLP) will be adopted as the classifier. Although the MLP was used as classifier the end results shows great potential of using the EEG model approach based on the 2D affective space model (ASM) for early detection of motor imitation skills development disorder among autistic children.