“A CONVOLUTIONAL NEURAL NETWORK MODEL FOR EARLY STAGE DETECTION OF AUTISM SPECTRUM DISORDER USING DEEP LEARNING”

Abstract: Autism spectrum disorder (ASD) is a neurological disorder that begins in childhood and lasts the rest of person’s life. It has an influence on how a person communicates and learns, as well as how they act and connect with others. There are number of techniques that can be used to help the child to grow and acquire new abilities. Behavioural and communication therapy, skill training, and symptom-controlling medications are all options available for treatment which are time consuming and subjective. Therefore, early and accurate detection of ASD is required which will help in treatment planning. With the patient’s history and different medical tests, the brain MR scans can proceed towards the distinguish between the Typical controls (TC) and ASD controls. The work is towards the development of Computer Aided Diagnosis for ASD detection and its classification into Typical Control (TC) and ASD. This project is about the selection of CNN deep learning techniques for accuracy improvement. In this project we have used total 10878 image belongings to typical control and autism. The collected datasets are pre-processed and applied convolution neural network with four layers which is giving 99% accuracy for training and validating data.

Keywords: Autism Spectrum Disorder (ASD), Deep Learning, Classification, Convolutional Neural Network (CNN).