Morphometric Similarity Networks (MSNs) estimate structural 'connectivity' as a biologically meaningful set of statistical similarities between cyto-architectural features derived in-vivo from multiple MRI sequences. These networks have shown to be clinically relevant, predicting 40% variance in IQ. However, the sequences required (T1w and T2w 3D anatomical, DWI) to produce these networks typically have long acquisition times, which are less feasible in some populations. Thus, estimating MSNs using features from only a T1w MRI is attractive to both clinical and developmental neuroscience. We aimed to determine whether reduced-feature approaches approximate the original MSN model as a potential tool to investigate brain structure. Using Human Connectome Project data, we extended previous investigations of reduced-feature MSNs by comparing not only T1w-derived networks but additional MSNs generated with fewer MR sequences to their full acquisition counterparts. We produce MSNs which are highly similar at the edge-level, to those generated with multi-modal imaging. We also find that, regardless of the number of features, these networks have limited predictive validity of generalised cognitive ability scores in contrast to previous research. Overall, settings in which multi-modal imaging is not available or clinically/developmentally appropriate, T1w-restricted MSN construction provides a valid estimate of the MSN.
Copyright the authors 2019. Creative Commons Attribution 4.0 International Public License
Published version of Record of this article was published in 2020 and can be found here: https://doi.org/10.1162/netn_a_00123