Fig. 1

Analysis flow chart. A Surface rendering of the 200-ROI brain atlas, based on which we estimated B cortical MWF and C group-level tau-PET scores, which were further vectorized (D) and spatially correlated. E The same brain atlas shown in (A) was applied to each participant’s tau-PET score change rates, individual values were vectorized and all possible pairs of ROIs across participants were correlated to obtain a covariance in tau-PET score change matrix (F). G Using the 200 ROI brain atlas shown in (A), resting-state fMRI functional connectivity was assessed on 100 participants of the human connectome project (HCP). H Diffusion MRI from the HCP was used to estimate the fiber-tracts between each pair of ROIs in the brain atlas shown in (A). I The MWF atlas was overlayed to extract regional MWF values of underlying fiber tracts. J Linear regression analysis was performed with the covariance in tau-PET score change as the dependent variable, and the interaction of functional connectivity by MWF in fiber-tract as the predictor. In a sensitivity analysis, we further controlled the above-mentioned analyses for regional amyloid-PET levels or the covariance in amyloid-PET change (not shown)