Fig. 4.

Adolescent development of structure–function coupling. (A) We derived a functional connectivity (FC) matrix, or functional connectome, for each scan by estimating the pairwise correlations between resting-state fMRI time series averaged over all voxels in each of all possible pairs of two regions defined by the parcellation template. (B) We estimated global structure–function coupling by correlating the ranked edgewise structural and functional connectivity vectors derived from each participant’s FC matrix and MSN, respectively; and regional structure–function coupling as the correlation between the ranked vector of a region’s edges derived from the FC matrix and the MSN, respectively. (C) We estimated the linear effect of age on global structure–function coupling and found that there was a significant decline in coupling over the course of adolescence ($t=-5.04,P<0.001$). We next estimated the linear effect of age on regional structure–function coupling using linear mixed effects models. From this analysis, we derived (D) a map of baseline structure–function coupling as the predicted coupling at age 14, and (E) a map of the rate of change in coupling, or the $t$-value of the effect of age on structure–function coupling. We found that 10 regions showed significant changes in structure–function coupling during adolescence, after correction for multiple comparisons ($P_{\text{FDR}}<0.05$; SI Appendix, Fig. S12A). (E) We found that the age effect on morphometric similarity was significantly positively correlated with the rate of change in coupling ($r=0.36,P_{\text{spin}}<0.01$). Thus, isocortical regions that become more morphometrically dissimilar (structurally differentiated) tended to have decreased strength of structure–function coupling over the course of adolescence.