Work

Leveraging deep learning for enhanced diagnosis of autism spectrum disorder using resting-state functional magnetic resonance imaging and clinical data

Year: 2025

Type: article

Abstract: Autism Spectrum Disorder (ASD) presents a multifaceted challenge, char- acterized by intricate difficulties in social interaction, communication, and behavior. Timely and accurate diagnosis is crucial for effective intervention and support. However, existing methods often struggle to capture the in- tricate patterns indicative of ASD traits, leading to misdiagnosis or delayed intervention. In response, this research introduces an innovative approach leveraging Deep Attention Bidirectional Long Short-Term Memory (LSTM) networks alongside Convolutional Neural Networks (CNN) with Bidirectional LSTM with Attention for ASD classification. By analyzing resting-state func- tional Magnetic Resonance Imaging (rs-fMRI) data and phenotypic informa- tion from the Autism Brain Imaging Data Exchange (ABIDE) dataset, our model aims to dissect the complex temporal and spatial signatures associated with ASD. Our methodology focuses on integrating attention mechanisms to identify and prioritize key features in the input data, improving classifica- tion accuracy. Extensive training and validation are conducted to establish a strong classification framework. Evaluation metrics such as accuracy, preci- sion, sensitivity, and AUC-ROC are used to rigorously assess the effectiveness of our model. Preliminary results showcase promising outcomes: an accuracy of 93%, a precision of 0.90, and an AUC-ROC of 0.93. These findings not only advance the realm of ASD diagnosis research but also hold the potential to augment clinical decision-making with nuanced and reliable classification tools, ultimately improving outcomes for individuals affected by ASD. • A novel approach using Deep Attention CNN with Bidirectional LSTM. • ASD classification using rs-fMRI and phenotypic information. • Attention mechanisms enhance model accuracy. • Comparative analysis between models for clinical decision making.

Source: Results in Engineering

DOI: 10.1016/j.rineng.2025.104444

Authors: A. Sheik Abdullah, Keerthana V, S. Geetha, Utkarsh Mishra

Institution: Vellore Institute of Technology University

Language: English

Topic: Autism Spectrum Disorder Research

Subfield: Cognitive Neuroscience

Field: Neuroscience

Domain: Life Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $1,940