Morphological Versus Functional Network Organization: A Comparison Between Structural Covariance Networks and Probabilistic Functional Modes

Petra Lenzini, Thomas Earnest, Sung Min Ha, Abdalla Bani, Aristeidis Sotiras, Janine Bijsterbosch

Abstract: The degree to which gray matter morphology constrains brain function remains an elusive target of investigation due to the lack of a gold-standard against which to argue for a better or worse metric of neurobiological significance. Therefore, we sought to compare the output of state-of-the-art morphological and functional covariance decomposition methods directly to one another. Specifically, we compared the spatial network organization produced by non-negative matrix factorization of T1-weighted images and probabilistic functional modes of resting state functional MRI scans from 1297 UK Biobank subjects. We measured the cosine similarity of matched networks across 2 to 140 rank decompositions. Our findings revealed strong commonality between morphological and functional networks at the lowest rank (2). Morphology-function network commonality was retained across all ranks in the visual cortex, but broader network organization diverged between morphology and function at higher ranks.