Fig. 3: Associations of regional CT maturation with the WM network architecture considering age as a continuous variable or performing individual-level analysis.
(Panel a to d: Statistical Mode II; Panel e and f: Statistical Model III). a Brain map of regional age effects on CT maturation from childhood to adolescence using the GAM (Bonferroni corrected at p = 5 × 10−5, two-sided). b Two representative GAM fitting curves (central lines in deep red) of cortical thinning in the prefrontal cortex (PFC) and inferior parietal cortex (IPC), each with a 95% confidence interval (shadows in light red). c Brain maps of regional CT maturation rates (the first derivative of the GAM fitting curves) are shown at representative ages of 6, 10, and 14 years. d Significant Pearson correlations were observed between the nodal CT maturation rate and the mean of its directly connected neighbors at each age point. The observed correlations (red dots) were compared to the correlations obtained from 1000 rewired tests (light blue boxes) and 1000 spin tests (deep blue boxes). Asterisks denote statistical significance level at p < 0.001, one-sided. e Longitudinal WM network-based CT maturation analysis at the individual level. The definition of nodal CT maturation rates and their brain maps were shown in the left panel. Negative values indicate cortical thinning while positive values indicate cortical thickening with development. The Pearson correlation between nodal CT maturation rate and the mean of its SC-defined neighbors for a representative participant is given in the right panel (radj = 0.56, p = 3.75 × 10−82, two-sided, linear fit (central line in green) with a 95% confidence interval (shadows in gray)). f The distribution of these correlations across individuals (pspin < 0.05 in all longitudinal samples and prewired < 0.05 in 98.6% (204/207) of longitudinal samples, one-sided). In (d, f) boxes represent the IQR, with the median shown as an inside line, while the lower and upper boundaries of the box correspond to the 25th and 75th percentiles. The whiskers extend to the minimum and maximum values within 1.5×IQR, and data points beyond the whiskers are displayed as outliers.
