Fig. 1.

Establishing the backbone connection. The whole-brain streamlines (A) generated from tractography were classified according to their end-points (i.e., 84 cortical/subcortical ROIs (B)) to produce a structural connectivity matrix, where each element indicates the number of streamlines connecting any pair of ROIs. To eliminate false-positive connections arising from imaging noise, the mean connectivity matrix of all subjects (C) was binarized with a threshold of 10% (D). For each subject, tract-specific diffusion metrics were derived along all binarized connections, and tau-PET signals were summarized in all ROIs (H). Thus, “tau-WM” could be analyzed on “node-edge” pairs. Circle size in the brain illustration indicates the ROI size with a matching color code in (B, F, and G).