Combination of structural and functional connectivity explains unique variation in specific domains of cognitive function.

Abstract

The relationship between structural and functional brain networks has been characterised as complex: the two networks mirror each other and show mutual influence but they also diverge in their organisation. This work explored whether a combination of structural and functional connectivity can improve models of cognitive performance, and whether this differs by cognitive domain. Principal Component Analysis (PCA) was applied to cognitive data from the Human Connectome Project. Four components were obtained, reflecting Retention and Retrieval, Processing Speed, Self-regulation, and Encoding. The PCA-Regression approach was applied to predict cognitive performance using structural, functional and joint structural-functional components. Model quality was evaluated using model evidence, model fit and generalisability. Functional connectivity components produced the most effective models of Retention and Retrieval and Encoding, whereas joint structural-functional components produced most effective models of Processing Speed, and Self-regulation. The present study demonstrates that multimodal data fusion using structural and functional connectivity can help predict cognitive performance, but that the additional explanatory value (relative to overfitting) may depend on the specific selection of cognitive domain. We discuss the implications of these results for studies of the brain basis of cognition in health and disease.