Fig. 1.

Hand-tracing of the caudate and putamen. (A) MRI showing an axial MRI slice with caudate and putamen visible, (B) using our edge contrast-enhancing technique (Sobel-gradient filter), gray/white boundaries of the striatal regions (caudate and putamen) are enhanced, (C) caudate and putamen are outlined, and (D) the anatomical regions-of-interest are shown. As described by Buchsbaumet al. (2003), the most ventral and dorsal axial slices for which both the caudate and putamen were present were determined for each participant. An automated boundary-finding method based on the Sobel-gradient filter provides a reproducible structure edge, with little operator variability. The caudate and putamen were outlined on the MRI by depositing points by mouse on the magnified and enhanced white structure edge using a semi-automated 3 × 3 local pixel maximum search. This placed the point at the center of the edge, enhancing inter-operator consistency. A spline curve was fit to the points and the ROI edge was stored. We determined the top of the caudate and putamen as the most dorsal axial slice showing a visible gray patch and the bottomas the slice in which the caudate and putamen entirely merged. This distance was divided by six to yield five equally spaced slices for the caudate and putamen separately for tracing (e.g., most dorsal slice = 20, most ventral slice = 38, difference = 18. 18/6 = 3, slices 23, 26, 29, 32, and 35 are traced). For each region of interest, area was calculated in voxels, divided by whole brain volume, and then multiplied by 1000 to correct for variation in intracranial volume. Whole brain volume was expressed as the sum of the absolute gray and white matter volumes of 39 Brodmann areas within 33 coronal brain slices (see Mitelman et al., 2005). Although there are multiple ways to account for variation in brain size (O'Brien et al., 2006), our primary dependent variables are striatal size relative to whole brain because relative values are more widely reported than absolute size. However, we additionally conducted parallel analyses on absolute size and report these findings. To establish inter-rater reliability, two tracers from our group independently traced striatal ROIs at various ventral/dorsal slice levels in each hemisphere for a subset of the individuals (n = 10). The intra-class correlation coefficient (ICC) for the putamen was 0.98 and for the caudate, ICC = 0.92 (Buchsbaum et al., 2003) which is in line with kappa scores from other groups examining the striatum, e.g., Levitt et al. (2002).