A Deep Learning Method for Autism Spectrum Disorder Classification Based on Multimodal Neuroimaging Data.

Researchers created a computer program that can help identify autism by analyzing two types of brain scans together. The program correctly identified autism in about 8 out of 10 people tested. By combining different brain imaging information, it performed better than using just one type of scan. This technology could potentially help doctors diagnose autism earlier and more accurately.

This study developed a deep learning model to classify autism spectrum disorder (ASD) using combined brain imaging data from both functional MRI (fMRI) and structural MRI (sMRI). The multimodal approach aimed to capture more comprehensive brain signatures than single imaging types alone. Testing on data from the ABIDE NYU site using five-fold cross-validation, the model achieved 82.63% accuracy, 89.31% AUC, 81.45% sensitivity, and 82.86% specificity for distinguishing individuals with ASD from typically developing controls. The researchers suggest this multimodal feature fusion strategy significantly enhances ASD identification and could support early clinical decision-making and personalized treatment strategies.

While promising for future diagnostic support, this technology requires further validation across diverse populations and clinical settings before implementation. The multimodal approach may inform development of objective diagnostic tools to complement clinical assessment, potentially supporting earlier and more accurate ASD identification.

Sample size not reported, limiting assessment of generalizability. Testing limited to single site (ABIDE NYU), which may not represent broader ASD population diversity. No comparison with current diagnostic methods or external validation reported.

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by impairments in social interaction and communication skills. Accurate, early-stage differentiation of individuals with ASD from typically developing controls (TC) is essential for timely intervention and treatment. In this paper, we propose a predictive model based on multimodal feature fusion, using both functional magnetic resonance imaging (fMRI) and structural magnetic resonance imaging (sMRI) data to improve the classification of ASD. By integrating complementary information from these two modalities, our method constructs a more comprehensive feature space, capturing complex neuropathological signatures that a single modality cannot provide.

We evaluated the proposed approach using imaging data from the ABIDE NYU site under a five-fold cross-validation scheme. The experimental results show that the proposed method achieved an average accuracy of 82.63%, an area under the receiver operating characteristic curve (AUC) of 89.31%, a sensitivity of 81.45%, and a specificity of 82.86%. These findings suggest that the proposed multimodal feature fusion strategy significantly enhances ASD identification, offering a promising approach to the precise diagnosis of brain disorders.Clinical Relevance- We proposed a learning framework that integrates multi-modality neuroimaging data, addressing the heterogeneity of ASD-related brain features and the challenges posed by limited training data. This framework contributes to improving diagnostic accuracy and supports early clinical decision-making for ASD, thereby facilitating timely intervention and the development of personalized treatment strategies in clinical practice.