Structural Similarity as a Predictor of Resting State Functional Connectivity

Abstract

Background

Understanding the brain's patterns of connectivity is vital to understanding its mechanisms. An emerging avenue of research however, has emerged from the insight that there is more than one way to understand connectivity; brain regions with similar physical features more often have physical connections, and possibly functional ones as well. Recent studies have found structural connectivity to correlate with a variety of modalities, such as gene expression, neurotransmitter receptor density, cellular morphology, glucose metabolism, haemodynamic activity, and electrophysiology [1].

T1 structural MRI data is among the most accessible and reliable forms of brain imaging, but it lacks the resolution to detect microscopic differences. A clever workaround is the technique of morphometric similarity networks (MSNs): they aggregrate various structural properties at one point into a single vector, and then use this measurement to estimate the similarity between two brain regions of a given subject. Regions rated similar by MSNs have been found to have high similarity in microstructure, as well as higher axonal and functional connectivity (FC) [2].

Purpose

Morphometric INverse Divergence (MIND) is a relatively new MSN developed by Sebenius and colleagues [2] that aggregates five structural features that can be derived from T1 structural MRI images: cortical thickness, grey matter volume, large/small-scale curvature, and vertex surface area (a quantity derived from postprocessing transformations). Compared to many other MSNs, MIND does quite well at predicting various types of similarity. MIND similarity correlates well with both microstrucrure and axonal connectivity. In addition to having higher heritability, it also does well at capturing age-related change and several pathologies.

Preliminary results also point at a coupling of MIND and FC, but the technique is new and literature is sparse. Specifically there is a lack of research on MIND and resting state FC, which we aim to address.

Methods

Using a sample of 77 healthy older adults (min. age 65, mean age 68), we will test the resting-state FC between the same cortical regions that are used to define a MIND network. To compute MIND, we took T1 MRI images with the MPRAGE sequence on a 3 T MRI. Structural features were acquired using the recon-all function of FreeSurfer and the MIND pipeline developed by Sebenius et al. Resting state fMRI images were taken with a conventional gradient-echo-EPI sequence. Functional data processing and network-based analyses were done using FSL and FSLNETS.

References

[1] J. Y. Hansen et al., “Integrating multimodal and Multiscale Connectivity Blueprints of the human cerebral cortex in health and disease,” PLOS Biology, vol. 21, no. 9,  e3002314, 2023. doi: 10.1371/journal.pbio.3002314.

[2] I. Sebenius et al., “Structural MRI of brain similarity networks,” Nature Reviews Neuroscience, vol. 26, pp. 42–59, 2025. doi: 10.1038/s41583-024-00882-2.