We performed statistical analysis on two types of surface based data maps, cortical thickness and white matter extent. FreeSurfer calculates cortical thickness as the closest distance from the grey/white boundary to the grey/CSF boundary at each vertex on the tessellated surface (47). White matter extent is measured by the jacobian of the spherical transform derived during atlas registration and reflects the amount of distortion needed to warp individual participant’s white matter maps into register with the common atlas. Difference maps for cortical thickness and white matter extent were created by subtracting each participant’s follow-up map from his/her baseline map, then smoothed at 10 mm FWHM prior to statistical analysis. FreeSurfer’s mri_glmfit program was used to perform general linear model (GLM) analysis of relations between cortical thickness or white matter extent difference maps and non-brain variables. To maintain proper statistical control over family-wise error in the mapping analysis, GLM effects were thresholded with reference to Monte Carlo–style simulations (10,000 iterations) using the FreeSurfer program mri_surfcluster under the assumption that the statistical output was z-distributed. These simulations provided estimates of the size
