Methods

Participant Analysis

The participant analysis comprises two levels: the first one computes the reproducibility/reliability metrics per participant using all participant’ ROIs, and the second one analyses the relationship between the computed metrics and age. In the end of the first level analysis, each participant has three reproducibility/reliability metrics: slope, intercept, and . A statistical test was performed, per reproducibility/reliability metric to compare the metric means of different groups. The statistical test performed depend upon the objective and were the following: objective (a) - to compare the reproducibility metric mean of paediatric and adults’ groups an ANCOVA was performed, the dependable variable was the metric whereas the factors where the age, the age-group and the acquisition setting; objective (b) – to test the acquisition setting effect on the reproducibility metric an ANCOVA in which the independent variables were the acquisition setting and age and the dependent variable the reproducibility metric was executed ; and objective (c) – to compare the test-retest reliability metric means of different frameworks a paired t-test was conducted. To assess the age impact on the reproducibility/reliability metrics, a second level analysis was carried out. The values from the first level analysis were grouped by metric and group. Then another linear fit was performed in which the independent and dependent variable was the age and the metric, respectively. It was considered that age influenced the metric if the p-value of the second level regression was lower than the significance value (p-value=0.05). The relation of age and a metric is given by the slope of the second level analysis. In this paper, the second level analysis slope of the and is given by , and , respectively. To further verify whether the relationship between age and reproducibility/reliability metrics was the same in every group, a pairwise independent t-test comparisons was performed, per metric, comparing the group’ slopes.

ROI analysis

This analysis aimed to understand the reproducibility/reliability across ROIs. In this case, a linear regression was fitted for each ROI in which the dependent and independent variables were the CAT12 and FreeSurfer estimations of cortical thickness for all individuals. To compare whether the reproducibility metrics across ROIs were identical in every group, a pairwise independent t-test comparison was performed, and corrected with the Bonferroni method. Concerning the test-retest reliability across ROIs, a paired-test was performed, per metric, to compare the ROI reliability metrics of the two frameworks. Moreover, for the all objectives a paired t-test comparing the cortical thickness estimations of CAT12 and FreeSurfer were performed per ROI and group/framework.