Autism Detection Using Xception-based CNN with SVM on Resized and Normalized Neuroimaging Data

A complicated neurological disorder known as Autism Spectrum Disorder (ASD) causes a variety of signs and behaviors. For prompt assistance and treatment, a prompt and correct diagnosis of ASD is essential. The deep learning model built on the Xception platform paired with Support Vector Machines (SVM) using MRI data is a unique method for diagnosing and identifying ASD presented in this paper. The first step of the suggested technique is preparing neurological data, including magnetic resonance imaging (MRI) and functional magnetic resonance imaging (fMRI), through downsizing and normalisation to improve extracting features and lower computing costs. Following this, the Xception-based Convolutional Neural Network (CNN) automatically extracts basic features from brain imaging information. The CNN model performs exceptionally well at identifying elaborate trends and differentiating between those with ASD and those who are usually developing. SVM is applied to the retrieved features to improve accurate classification and the ability even more. This novel technique uses SVM's ability to discriminate and deep learning's capability in feature extraction. The suggested technique achieves outstanding accuracy, sensitivity, and specificity in ASD identification and identification, as demonstrated by research results on a large dataset. On scaled and normalized brain data, the combination of Xception-based CNN and SVM shows promise for accelerating early detection of ASD and enhancing knowledge of the neurological basis of this complex condition. This study lays the door for more accurate and obtainable tools for medical practitioners to help with the early detection and treatment of ASD, thereby improving the standard of life for those who are impacted by this disorder. The proposed Xception-CNN with SVM-Lagrangian Optimizer obtains a remarkable accuracy measurement of 95.13 %.