Journal of Psychology & Psychotherapy
Open Access

Abstract

MVUT_GCN: Multi-View United Transformer Block-Graph Convolution Network for Diagnosis of Autism Spectrum Disorder

D. Darling Jemima*, Grace Selvarani and J. Daphy Louis Lovenia

Identifying Autism Spectrum Disorder (ASD) is challenging due to its complex and varied nature, making early detection important for effective intervention. Recently, there has been considerable discussion about using deep learning algorithms to improve ASD diagnosis through neuroimaging data analysis. To address the limitations of current techniques, this research introduces an innovative approach called the Multi-View United Transformer Block of Graph Convolution Network (MVUT_GCN). MVUT_GCN leverages the benefits of multi-view learning and convolution processes to extract subtle patterns from structural and functional Magnetic Resonance Imaging (MRI) data. A comprehensive analysis using the Autism Brain Imaging Data Exchange (ABIDE) dataset demonstrates that MVUT_GCN outperforms the existing Multi View Site Graph Convolution Network (MVS_GCN), achieving a +3.44% improvement in accuracy. This enhancement highlights the effectiveness of our proposed model in identifying ASD. Improved accuracy and consistency in ASD diagnosis through MVUT_GCN can facilitate early intervention and support for ASD patients. Additionally, MVUT_GCN's interpretability bridges the gap between deep learning models and clinical insights by aiding in the identification of biomarkers associated with ASD. Ultimately, this work advances our understanding of ASD and its practical management, with the potential to improve outcomes and quality of life for those affected. Understanding of ASD and its practical management, with the potential to improve outcomes and quality of life for those affected.

Published Date: 2024-08-21; Received Date: 2024-07-22