Diffusion MRI Microstructural Similarity Networks Map the Heritable and Hierarchical Organization of Gray Matter in the Human Brain

Motivation: In-vivo microstructural similarity between gray matter regions remains poorly understood in the human brain. Goal(s): We aim to map the regional microstructural covariation of human gray matter with diffusion MRI. Approach: We employ rotationally invariant spherical harmonic features to compute diffusion similarity networks (DSNs) for Human Connectome Project data. Results: Compared to previously reported measures of gray matter similarity, DSNs have greater biological fidelity, higher heritability, and better correlation to development and cognition. DSNs encode for functional and structural connectivity and gene expression; follow cytoarchitectural and laminar differentiation; align with the sensory-fugal and sensorimotor-association axes; and contribute to neural oscillatory dynamics. Impact: Diffusion similarity networks can be easily integrated into conventional diffusion MRI analysis, adding information complementary to structural connectivity, and could prove useful in investigating an array of neurological and psychiatric conditions.