Multimodal Evidence and Applications of Neuroimaging in the Rehabilitation of Autism Spectrum Disorder

Currently, the clinical assessment of autism spectrum disorder (ASD) mainly relies on behavioral scales and structured interviews, lacking objective, repeatable, and easily trackable quantitative indicators. This review systematically integrates the main research on ASD neuroimaging in recent years. Based on summarizing the core findings, it carefully reviews the scope application and limitations of the existing evidence. In terms of structural imaging, structural magnetic resonance imaging (sMRI) studies generally reveal that individuals with ASD have nonlinear brain development trajectories and unique structural phenotypes in certain brain regions. In terms of functional networks, resting-state functional magnetic resonance imaging (rs-fMRI) shows significant abnormalities in functional connectivity (FC) between large-scale brain networks, and dynamic FC analysis suggests that these abnormalities may be related to changes in "network flexibility", but its reliability still requires sensitivity testing and external validation of independent cohorts. In terms of white matter pathways, diffusion tensor imaging (DTI) has identified differences in microstructural indicators in key white matter pathways such as the corpus callosum, providing a basis for understanding information transmission disorder, but the biological specificity of traditional diffusion indicators is limited. In other imaging techniques, magnetic resonance spectroscopy (MRS) provides metabolic-level support evidence for the brain excitation-inhibition imbalance hypothesis; functional near-infrared spectroscopy (fNIRS) has become a feasible tool for follow-up and intervention monitoring of young children due to its convenient operation and strong anti-interference ability; magnetoencephalography (MEG) can supplement the evidence of brain interconnection during the semantic-executive processing from the time and frequency domains. In terms of translational applications, multimodal fusion and interpretable artificial intelligence help to analyze heterogeneity and promote risk stratification and classification studies. In ASD rehabilitation practice, the application of imaging indicators can be summarized into three paths: assisting in early screening, biological subtype matching, and efficacy monitoring. Looking forward to the future, it is necessary to integrate the imaging results and functional indicators from multi-center longitudinal cohorts to establish a standardized, reproducible, and easily clinically interpretable assessment system, promoting the substantive transformation of ASD neuroimaging from a research tool to a rehabilitation auxiliary means.