A Deep Learning Algorithm To Detect Children With Autism Spectrum Disorder (Asd) Using Electroencephalogram (Eeg) Signal

Rosman, Afrina Aida (2019) A Deep Learning Algorithm To Detect Children With Autism Spectrum Disorder (Asd) Using Electroencephalogram (Eeg) Signal. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (24) A Deep Learning Algorithm To Detect Children With Autism Spectrum Disorder (Asd) Using Electroencephalogram (Eeg) Signal.pdf - Submitted Version Download (694kB)

Abstract

Autism Spectrum Disorder (ASD), also known as autism is a condition where neurological disorder found in the brain development of the human being. Autistic patient will develop communication disorder and lack of social interaction. The number of children that have been diagnosed with autism increased each year. Therefore, it is significant to have early detection of presence of autism symptom in a child from an early age. This project is aimed to integrate signal from autistic child by using Electroencephalogram (EEG). This project successfully investigated the brain signal database using pattern recognition techniques from a deep learning method. The extracted features will undergo multilayer perceptron network for the classification process. The proposed method is able to give a high intermediate accuracy to detect autism presence in a child. The dataset obtained from University King Abdulaziz, Jeddah Saudi Arabia. The obtained dataset is three normal person and two autistic patients. By using 2D CNN, the obtained accuracy is 70%, while by using 1D CNN the obtained accuracy is 76%. In conclusion, the result achieved has proved that the deep learning method is viable for autism-normal EEG signal classification.

Actions (login required)

View Item