Latent biophysical diffusion properties underlying cerebral microstructure revealed through multimodal MRI covariation in the squirrel monkey

Abstract

Diffusion MRI provides multiple quantitative descriptors of neural tissue derived from distinct physical and biophysical models, each probing different aspects of cerebral microstructure. While these metrics are often interpreted independently, their joint statistical structure may encode information about latent biophysical properties that constrain and organize brain microarchitecture. Here, we investigate whether stable patterns of multimodal diffusion MRI covariation reveal underlying biophysical diffusion properties of cerebral microstructure in the squirrel monkey (Saimiri sciureus). Using a high-resolution multimodal diffusion MRI dataset acquired at 11.7T (400 m isotropic resolution) from 15 adult subjects, we analyzed tensor-based metrics alongside advanced multicompartment and axonal models, including NODDI and ActiveAx. Voxelwise and region-level correlation analyses were performed across multiple spatial scales to examine the structure, redundancy, and stability of relationships among diffusion-derived maps. We show that metrics originating from conceptually distinct models consistently organize into a low-dimensional structure characterized by stable clusters of covariation, robust to changes in parcellation and spatial aggregation. These patterns cannot be explained by trivial metric redundancy alone, but instead suggest convergence toward a reduced set of effective biophysical degrees of freedom governing diffusion behavior in neural tissue, including axonal organization, neurite density, orientation dispersion, and isotropic diffusion components. We interpret these covariation structures as manifestations of latent biophysical diffusion properties--emergent tissue states that are not directly observable through any single metric but become apparent through their structured relationships. From a systems neuroscience perspective, the stability of these latent dimensions supports the view that adult cerebral microstructure reflects the steady-state outcome of neurodevelopmental dynamics operating under biophysical constraints. Rather than providing a regional atlas, this work proposes a conceptual framework for interpreting multimodal diffusion MRI as a projection of an underlying low-dimensional biophysical state space organizing cerebral microstructure in primate brains. Key pointsO_LIMultimodal diffusion MRI metrics derived from distinct biophysical models exhibit stable and non-trivial patterns of covariation, revealing a low-dimensional organization of cerebral microstructure. C_LIO_LIThese covariation structures suggest the existence of latent biophysical diffusion properties that are not directly observable through individual metrics but emerge from their structured relationships across spatial scales. C_LIO_LIThe stability of these latent dimensions supports the interpretation of adult cerebral microstructure as a structurally stable outcome of neurodevelopmental dynamics governed by biophysical constraints. C_LI