Prisma

Imaging protocols for the ABCD dataset are described elsewhere (Casey et al. 2018 (link)), and were harmonized for three 3T scanner platforms (Siemens Prisma, General Electric 750 and Philips) used across the 21 data acquisition sites.

Minimally processed data for baseline and 2-year follow-up from the ABCD repository were used. Participants were scanned at 21 sites using 3T Siemens Prisma, General Electric 750 or Phillips scanner. Data acquisition and image processing methods were harmonized between sites and scanners [26 (link), 27 (link)]. T1 scans were processed using FreeSurfer 5.3.0 and the Desikan–Killiany atlas was used for subcortical segmentation and cortical parcellation. Major white matter tracts were labeled using AtlasTrack [28 (link)]. Quality control was conducted following recommendations from the ABCD data team [26 (link)], and full details are in the Supplementary Materials.

MRI and DTI scans were collected across the 21 research sites using Siemens Prisma, GE 750 and Philips 3T scanners. Scanning protocols were harmonized across sites. Full details of all the imaging acquisition protocols and the processing methods used in ABCD are outlined elsewhere (Casey et al., 2018 (link); Hagler et al., 2019 (link)).
To determine the importance of grey and white matter we chose six structural metrics available in the ABCD study: cortical thickness (CT), surface area (SA) and volume (GMV) for grey matter and fractional anisotropy (FA), mean diffusivity (MD) and volume (WMV) for white matter.

The MRI data collection was harmonized across the 21 sites using 3 T scanners (Siemens Prisma, General Electric 750, Philips).^{39 (link),45 (link)} The diffusion-weighted acquisition was conducted as previously described.^{45 (link)} After preprocessing of DWIs (eMethods in the Supplement), RSI was used to fit fiber orientation density functions to model rND, rN0, and total hindered diffusion (hD) (eg, primarily extracellular space around neurites).^{45 (link)} The DTI outcomes included FA and MD. Major white matter tracts were labeled with AtlasTrack using prior probabilities and orientation of long-range projection fibers.^{46 (link)} Probability estimates for each white matter tract were used to calculate weighted means of the RSI and DTI measures for all white matter fibers as well as key association, commissural, and projection fiber tracts,^{45 (link)} including the anterior thalamic radiations (ATR), cingulum in the cingulate gyrus (CGC), cingulum adjoining the hippocampus (CGH), corpus callosum (CC), corticospinal tract (CST), fornix (FX), uncinate fasciculus (UNC), inferior frontal occipital (IFO), inferior longitudinal fasciculus (ILF), and the superior longitudinal fasciculus (SLF).

Neuroimaging data were acquired at 21 different sites and processed by the ABCD team. A 3-T Siemens Prisma, General Electric 750 or Phillips scanner was used for data acquisition. Protocols used for data acquisition and processing are described elsewhere (Casey et al., 2018 (link), Hagler et al., 2019 (link)). In brief, T1-weighted data was acquired by magnetisation-prepared rapid acquisition gradient echo scans with a resolution of 1 × 1 × 1 mm³, which was used for generating cortical structural measures, and diffusion-weighted data was obtained by high angular resolution diffusion imaging scans, used for generating white matter microstructural measures.
Two modalities of brain structural measures were used in the present study: grey matter cortical measures and white matter microstructural measures (Hagler et al., 2019 (link)). Cortical reconstruction and volumetric segmentation was performed with FreeSurfer 6.0 (Dale et al., 1999 , Fischl et al., 2002 (link)). White matter microstructural measures (DTI) were generated using AtlasTrack, a probabilistic atlas-based method for automated segmentation of white matter fiber tracts (Hagler et al., 2009 (link)). For cortical measures, global measures were generated for cortical surface area and thickness. For DTI, measures of FA and MD were generated over the whole brain. For full pipeline, see SI Section 3.