ABSTRACT Background Structural white matter (WM) alterations are recognized in Autism Spectrum Disorder (ASD), yet the functional connectivity (FC) of WM networks and its clinical significance remain largely underexplored. Methods This study aimed to investigate aberrant FC patterns within intra‐WM (WM–WM) and WM–grey matter (WM–GM) networks in a large ASD cohort. Resting‐state fMRI data from 272 ASD individuals and 368 typical controls (TC) from the ABIDE‐II dataset were analysed. We constructed WM–WM and WM–GM FC networks using Pearson correlations between atlas‐defined regions, applied ComBat harmonization and employed Network‐Based Statistic (NBS) to identify group differences. Associations with clinical symptoms were assessed using the Social Responsiveness Scale (SRS) scores, and a CatBoost algorithm was used for diagnostic classification based on connectivity features. Results NBS analyses revealed significantly increased connectivity in ASD for 116 WM–WM pairs and 58 WM–GM pairs ( p < 0.05, FWER‐corrected). Critically, the strength of these aberrant WM–WM functional connections exhibited a significant negative correlation with SRS total scores ( r = −0.22, p < 0.001), whereas WM–GM connectivity showed no such significant association. The hybrid CatBoost classifier, integrating both WM–WM and WM–GM features, achieved moderate diagnostic discrimination (AUC = 0.669 ± 0.040). Conclusion These results offer novel insights into the aberrant functional architecture of WM‐related networks in ASD, particularly linking intra‐WM dysconnectivity to symptom severity, thereby enhancing our understanding of the neural substrates underlying social‐communicative deficits.
Wu et al. (Wed,) studied this question.