Agreement between FreeSurfer and CAT12 Cortical Thickness Measurements: A Discovery and a Replication Study in...

Cortical thickness can be derived from magnetic resonance imaging scans using the software tools FreeSurfer or the Computational Anatomy Toolbox (CAT12). Two studies in children/adolescents have compared cortical thickness estimates generated by these two programs and reported positive correlations. However, these studies, as well as similar studies in adults, found that the magnitude of cortical thickness generated by these two programs for the most part was not comparable, with one program or the other generating relatively lower values, and the direction of this effect varying across and within studies. Here, we tackled this issue by using, for the first time, discovery and replication samples of children/adolescents. For both samples (age range 6-16 years), we observed strong positive correlations between cortical thickness values generated with FreeSurfer and with CAT12 at both the whole-cortex and regional levels. We also found that FreeSurfer yielded relatively lower cortical thickness values than CAT12, both at the level of the whole cortex and in ≥ 94% of the regions investigated. In supplementary analyses, we further evaluated several factors that could influence inter-software convergence and found that age, image quality, and scanner manufacturer were associated with the magnitude of agreement between the two programs. Given this unequivocal difference in magnitude across both samples, we conclude that cortical thickness values can only be compared amongst studies on children/adolescents if the studies implemented the same software program.