At time 1 and time 2, structural magnetic resonance images were acquired on Siemens 1.5 T systems. Cortical reconstruction and volumetric segmentation were performed with the FreeSurfer image analysis suite (http://surfer.nmr.mgh.harvard.edu). Diffusion-weighted data were analyzed using TRActs Constrained by UnderLying Anatomy [TRACULA; (64 (link))], an automatic reconstruction tool for identifying major white matter tracts from diffusion-weighted images and measuring the diffusion within these tracts at a set number of locations. The acquisition sequences for gray matter data differed in minor ways across acquisitions. Following recommendations (60 (link), 65 (link)), we documented that there was no evidence of heteroskedasticity before completing our longitudinal analysis. Please see the Supplementary Materials for details about data acquisition, analysis, and tests of heteroskedasticity.