*2.2. Image Acquisition*

Image acquisition was described previously [33]. Briefly, all scans were acquired on the same GE DiscoveryTM MR750 3T scanner (GE Medical Systems, Milwaukee, WI, USA). The following sequences were acquired for each participant: Pre-contrast 3D highresolution T1-weighted image (0.7 mm isotropic) using fast spoiled gradient echo (SPGR) with magnetisation-prepared inversion recovery pulse (TE/TI/TR=2.8/450/7.1 ms, flip angle = 12); and axial diffusion-weighted imaging (2 mm isotropic, TE/TR = 85/8325 ms) with a uniform gradient loading (*b* = 1000 s/mm2) in 64 directions and 2 *b*0 s. An additional *b*0 image with reversed phase-encoding was also acquired for distortion correction [34].

## *2.3. Image Processing to Obtain Structural Connectomes*

The T1 images were processed using a modified version of Freesurfer's recon-all (v6.0) [35], alongside an in-house skull-stripping tool (Sydney Neuroimaging Analysis Centre). Each subject was inspected, and minor segmentation errors were manually corrected. A 5 tissue-type (5TT) image [36] was generated using MRtrix3 [37]. The T1 image was registered to the mean *b*0 image; the warp was used to register the 5TT image, and the DK [32] parcellation image to the diffusion image.

Diffusion image processing was conducted using MRtrix3 [37]. The diffusion preprocessing included motion and distortion correction [34,38], bias correction using ANTs [39]. The dhollander algorithm [40] was used to estimate the response functions of the white matter, grey matter, and cerebral spinal fluid, from which constrained spherical deconvolution was used to estimate the fibre orientation distributions using MRtrix3Tissue [37]. The intensity of the white matter fibre orientation distributions was normalised [37], and used for anatomically constrained whole-brain tractography [41] (along with the registered 5TT image). The tractography specifications were as follows: 15 million tracks were generated, iFOD2 probabilistic fibre tracking [42], dynamic seeding [43], maximum length 300 mm, backtrack selected and crop at grey-matter-white-matter interface selected. For quantitative analysis, the corresponding weight for each streamline in the tractogram was derived using SIFT2 [43]. The streamlines and corresponding SIFT2 weights were used to create a weighted, undirected structural connectome ("SC") using the registered parcellation image. All image processing steps are shown in Figure 1 (Step 1, a).
