Abstract

Human neuroimaging methods have provided a number of means by which the connectivity structure of the human brain can be inferred. For instance, correlations in blood-oxygen-level-dependent (BOLD) signal time series are commonly used to make inferences about “functional connectivity.” Correlations across samples in structural morphometric measures, such as voxel-based morphometry (VBM) or cortical thickness (CT), have also been used to estimate connectivity, putatively through mutually trophic effects on connected brain areas. In this study, we have compared seed-based connectivity estimates obtained from four common correlational approaches: resting-state functional connectivity (RS-fMRI), meta-analytic connectivity modeling (MACM), VBM correlations, and CT correlations. We found that the two functional approaches (RS-fMRI and MACM) had the best agreement. While the two structural approaches (CT and VBM) had better-than-random convergence, they were no more similar to each other than to the functional approaches. The degree of correspondence between modalities varied considerably across seed regions, and also depended on the threshold applied to the connectivity distribution. These results demonstrate some degrees of similarity between connectivity inferred from structural and functional covariances, particularly for the most robust functionally connected regions (e.g., the default mode network). However, they also caution that these measures likely capture very different aspects of brain structure and function.