Autism Spectrum Disorders Identification from MRI Using Deep Learning

Abstract

Autism Spectrum Disorder is a neurodevelopmental condition that affects social communication and behavior, characterized by diverse symptoms and varying severity levels. Accurate and early diagnosis remains crucial for effective intervention and support. This study focuses on developing a deep learning model with a 3D convolutional neural network (CNN)based approach for ASD identification using sMRI and fMRI data, integrating models such as 3D Resnet, DenseNet, and VGG16 to extract spatial patterns associated with ASD. To evaluate the performance of our proposed system, we conducted several experiments based on multiple parameters have been performed using the publicly challenged ABIDE dataset of unconstrained images. The obtained experimental results proved the effectiveness of the proposed system against deep CNN architectures, as well as with recent state-of-the-art methods.