The within-subject stability of cortical thickness, surface area, and brain volumes across one year

T1-weighted (T1w) imaging is widely used to examine brain structure based on image-derived phenotypes (IDPs) such as cortical thickness, surface area, and brain volumes. The reliability of these IDPs has been extensively explored, mainly focusing on the inter-subject variations, whereas the stability of the within-subject variations has often been overlooked. Additionally, how environmental factors such as time of day and daylight hours impact the structural brain is poorly understood. Therefore, we aimed to address the stability of T1w-derived phenotypes and explore the effecting factors including processing stream, brain region size, time of day, daylight hours, and head movement.

Three subjects in their late 20s, early 30s, and early 40s were scanned repeatedly on the same scanner over one year, from which a densely sampled dataset was acquired with 38, 40, and 25 sessions for subjects 1, 2, and 3, respectively. The temporal stability was evaluated using within-subject percentage change and coefficients of variation (CV). The effects of time of day and daylight hours were assessed by fitting general linear models, aggregating the effect size with meta-analysis, and equivalent analysis.

The longitudinal processing stream generates more stable results than the cross-sectional processing stream, hence the following results are all derived from the longitudinal processing stream. First and foremost, most IDPs demonstrated percentage changes within 5% and CVs within 2% for almost all brain regions, indicating high stability. Interestingly, we found that the brain region size negatively correlates with the CVs. Specifically, several small brain regions, including the temporal pole, frontal pole, pericalcarine, entorhinal cortex, and accumbens area, showed low stability. In addition, cortical thickness change was strongly and positively correlated with that of volume change while being negatively correlated with change in surface area, illustrating their distinct roles in brain anatomy. Moreover, the time of day could be ignored when evaluating the total surface area and total cortical brain volume but not the average cortical thickness and total subcortical brain volume. Furthermore, daylight hours could be left out when evaluating IDPs since there was no appreciable effect of daylight hours on the IDPs stability. Lastly, apparent head motion causes cortical thickness and volume underestimated and surface area overestimated.