We measured subcortical brain structures using FSL's FIRST software (FMRIB Image Registration and Segmentation Tool) (Patenaude, 2007), a method that has been used by us and others to successfully measure subcortical volumes in a number of recent investigation (Angstrom et al., 2004; Corthorn et al., 1997; Figueroa Cave, 1997; Goodro et al., 2012; Markov et al., 1997; Rinehart et al., 1997; Sameti et al., 2011). The following structures were extracted and their volumes measured for all T1-weighted MR images: left and right thalamus, caudate, putamen, pallidum, hippocampus, amygdala, and nucleus accumbens. When boundaries of the delineated structures were visually inspected, we observed boundary underestimation of lateral ventricles (and corresponding overestimation of subcortical structures bordering on the lateral ventricles) in MRIs of subjects with ventricles significantly larger than those in the MNI152 standard template. To ensure a more accurate segmentation, MRIs were registered MNI152 standard space template using FSL's FNIRT (FMRIB's Nonlinear Image Registration Tool). The warped (registered) MRIs were then processed through FIRST to extract the surface mesh of each of the subcortical structures. The surfaces were then transformed back
