Convergence and divergence of thickness correlations with diffusion connections across the human cerebral cortex

Abstract

Cortical thickness correlation across individuals has been observed. So far, it remains unclear to what extent such a correlation in thickness is a reflection of underlying fiber connection. Here we explicitly compared the patterns of cortical thickness correlation and diffusion-based fiber connection across the entire cerebral cortex, in 95 normal adults. Interregional thickness correlations were extracted by using computational neuroanatomy algorithms based on structural MRI, and diffusion connections were detected by using diffusion probabilistic tractography. Approximately 35-40% of thickness correlations showed convergent diffusion connections across the cerebral cortex. Intriguingly, the observed convergences between thickness correlation and diffusion connection are mostly focused on the positive thickness correlations, while almost all of the negative correlations (>90%) did not have a matched diffusion connection, suggesting different mechanisms behind the positive and negative thickness correlations, the latter not being mediated by a direct fiber pathway. Furthermore, graph theoretic analysis reveals that the thickness correlation network has a more randomized overall topology, whereas the nodal characteristics of cortical regions in these two networks are statistically correlated. These findings indicate that thickness correlations partly reflect underlying fiber connections but they contains exclusive information, and therefore should not be simply taken as a proxy measure for fiber connections.