The results of this study provide evidence that longitudinal MRI assessment using robust registration is sufficiently sensitive to identify significant regional brain changes over about a half-year in boys and girls in early adolescence. The independent registration method used two inter-subject registration steps, whereas the sequential registration method relied on one inter-subject and one intra-subject registration. Reducing the number of inter-subject registrations minimized measurement error, as revealed by the lower CV for the sequential than independent methods, and likely enhanced detection of small volume changes occurring over short intervals. Consistent with current longitudinal studies based on longer follow-up intervals, the patterns of change differed by brain region. Specifically, developmental changes were greater in anterior than posterior cortex, suggesting remodeling of brain structure towards maturity (Gogtay et al., 2004; Sowell et al., 2004a). As with all in vivo imaging protocols, precision and registration are limited by the spatial resolution of acquired images, which in our case was about 1 mm3.
