An Explainable Connectome Convolutional Transformer for Multimodal Autism Spectrum Disorder Classification.

Researchers created a computer program that analyzes brain scans to help diagnose autism. The system looks at how different brain regions connect and work together using two types of brain imaging. When tested on a large database of brain scans, it was very accurate at identifying autism patterns. The brain areas it identified match what scientists already know about autism. This could potentially help make autism diagnosis faster and more objective than current methods.

This study developed the Connectome Convolutional Transformer (CCTF), a deep learning framework that combines brain imaging data from fMRI and sMRI to classify autism spectrum disorder. The system integrates functional and structural brain connectivity patterns using advanced machine learning techniques. Testing on the multi-site ABIDE dataset showed the combined fMRI+sMRI model achieved high accuracy in both within-site and across-site validation scenarios. The identified brain regions aligned with established autism neurobiology.

The framework demonstrated superior performance compared to existing state-of-the-art methods and maintained robustness across different research sites, suggesting potential for clinical translation.

This technology could potentially complement traditional autism diagnosis by providing objective neuroimaging-based assessment tools. However, clinical validation studies comparing against gold-standard diagnostic methods are needed before clinical implementation. The multimodal approach may enhance diagnostic accuracy but requires specialized neuroimaging infrastructure.

Specific accuracy values are not clearly reported in the abstract. Sample size and participant demographics are not provided. Clinical validation and comparison to standard diagnostic methods are not described. Real-world implementation feasibility remains unclear.

The diagnosis of autism spectrum disorder (ASD) is often hampered by its heterogeneity and reliance on time-consuming behavioral assessments. Automated neuroimaging-based diagnostic tools offer a promising alternative, but multi-site data integration often introduces variability, hindering the achievement of accurate and interpretable results. This study presents the Connectome Convolutional Transformer (CCTF), a multimodal deep learning framework that integrates functional and structural brain connectivity information from fMRI and sMRI modalities. The CCTF enriches feature representation by incorporating diverse functional connectivity metrics and structural covariance networks based on multiple morphological properties.

It employs a connectome convolutional embedding module and transformer encoder to capture and refine brain connectivity patterns. In addition, a node-to-graph pooling layer facilitates the identification of potential ASD biomarkers. Evaluation on the multi-site ABIDE dataset demonstrated that CCTF outperformed state-of-the-art methods, achieving accuracies of [Formula: see text] for fMRI, [Formula: see text] for sMRI, and [Formula: see text] for the ensemble fMRI+sMRI model in intra-site cross-validation. In the inter-site leave-one-site-out cross-validation, the CCTF maintained its superiority, with the ensemble model reaching [Formula: see text] accuracy, underscoring its robustness and generalizability across different sites.

The identified brain regions are consistent with established ASD neurobiology, underscoring CCTF's potential to advance the understanding of the neural mechanisms underlying this complex disorder.