Abstract
PURPOSE: Cortical thickness (CTh) is a valuable imaging biomarker of brain development and aging. The assessment of CTh using the two most widely utilized software packages, Computational Anatomy Toolbox (CAT12) and FreeSurfer, reveals systematic differences between the two tools. Nonetheless, longitudinal results are supposed to be less affected by such differences. To date, there is no comparison of longitudinal CTh data obtained with these preprocessing pipelines. METHODS: We evaluated cross-sectional and longitudinal measurements of CTh using CAT12 and FreeSurfer in healthy older individuals with structural brain MRI. First, we compared cross-sectional CTh values obtained with these two methods using paired t-tests and correlation analyses. Second, we compared longitudinal CAT12 and FreeSurfer-based data using annualized percent change (APC) of CTh between two timepoints. RESULTS: Cross-sectional CTh values were higher in FreeSurfer compared to CAT12 in most regions, albeit highly correlated and similarly distributed. In contrast, longitudinal analyses revealed significant differences in APC values with only weak to moderate correlation between the two methods. CONCLUSION: Significant differences in longitudinal results between CAT12 and FreeSurfer emphasize the need to consider the preprocessing methods used when interpreting MRI data in brain aging research. Further technical developments are warranted for reliable longitudinal CTh assessment in elderly cohorts.