Fig. 1: Study overview.

A: MSN construction. The MSN was computed using a 308 × 308 matrix incorporating five features (surface area, cortical thickness, gray matter volume, Gaussian curvature, and mean curvature).The MSN strength was derived by calculating the average weighted correlation coefficients between each brain region and all other regions. B: Classification and Genes analysis: The HYDRA method was applied to identify distinct AMDD subtypes based on MSN strength profiles. PLS regression was employed to determine imaging-transcriptomic associations. The relationship between whole-brain gene expression patterns and morphometric alterations across different AMDD subtypes was elucidated through PLS weight mapping, functional enrichment analysis, and cellular transcriptional signature assessment.