[Paper Review] Decoding Gray Matter: large-scale analysis of brain cell morphometry to inform microstructural modeling of diffusion MR signals

Charlie Aird-Rossiter, Hui Zhang+3 more|arXiv (Cornell University)|Jan 3, 2025

Advanced Neuroimaging Techniques and Applications5 citations

TL;DR

The paper analyzes ~3,500 3D brain cell reconstructions across mouse, rat, monkey, and human cortex to quantify structural, topological, and shape descriptors that inform diffusion MRI microstructure models of gray matter.

ABSTRACT

The structure of grey matter has long been a key focus in neuroscience, as cell morphology varies by type and can be affected by neurological conditions. Understanding these variations is essential for studying brain function and disease. Diffusion-weighted MRI (dMRI) is a powerful non-invasive tool for examining cellular microstructure in vivo. However, for dMRI to accurately reflect cellular features, it is crucial to determine which aspects of morphology influence its measurements. Proper interpretation of dMRI data depends on understanding its sensitivity to different cellular characteristics. Despite growing interest in cellular morphology, there has been no systematic report on the key features defining different neural cell types. To address this, we analyzed over 11,500 three-dimensional cellular reconstructions across three species and nine cell types, establishing reference values for critical morphological traits. These traits fall into three categories: structural features that define the cell's skeletal framework, shape features that describe spatial organization, and topological features that break down cellular structure to distinguish cell types. Beyond reporting these reference values, we examine their relevance for dMRI, identifying which neural features dMRI can detect and which cell types may be distinguishable. To complement the statistical analysis, we also provide high resolution 3D surface meshes representative of each cell type and species. This work provides essential benchmarks for grey matter research, offering new guidelines on linking neuroimaging measurements to neurobiology. These reference values will be a valuable resource for neuroscientists and neuroimaging researchers, aiding in the interpretation of imaging data and the refinement of brain tissue models.

Motivation & Objective

• Characterize gray matter cellular morphology relevant to diffusion MRI microstructure models.
• Quantify structural, topological, and shape descriptors from 3,598 SWC reconstructions across species.
• Provide distributions and reference values to inform biophysical GM models and dMRI acquisition design.
• Assess differences between neuronal and glial cells and across species to guide model parameterization.

Proposed method

• Download and preprocess 3,598 SWC cell reconstructions from Neuromorpho.org. Apply quality filtering to retain 3,598 healthy, complete 3D reconstructions and exclude 680 failures.
• Segment each cell into soma versus projections using a 1.5x soma radius threshold and compute soma metrics (R_soma, S/V_soma, eta_soma).
• Decompose projections into branches and compute branch-level metrics (L_branch, R_branch, S/V_branch, CV_branch, muOD_branch, Rc, tau_branch).
• Compute general cell metrics (R_domain, N_projection, BO) and soma/branch topology-derived measures (Topological Morphology Descriptor, TMD).
• Model diffusion-relevant shape descriptors by breaking segments into 10 μm cylinders to estimate FA and orientation dispersion (OD) via eigenvalue decomposition and appropriate distribution fitting (Watson/Bingham).
• Compare neuronal vs. glial morphologies and derive distributions of features (e.g., soma size, branch length) and topological distances across species.

Figure 1: An example of SWC file and how it relates to the cellular geometry. We highlight the structural elements used to estimate the morphological features. Note that the first node in the SWC file is the so called ‘root’. It often coincides with the soma’s centre and it is used to compute metric

Experimental results

Research questions

• RQ1 What are the statistical distributions of key morphological features of gray matter cells across species and cell types?
• RQ2 How do structural, topological, and shape descriptors relate to diffusion MRI signal modelling in gray matter?
• RQ3 What are the differences between neuronal and glial morphologies and across species in terms of features relevant to dMRI microstructure models?
• RQ4 How can the quantified morphometric features inform the design of biophysical GM models and diffusion acquisition strategies.

Key findings

• The study provides mean and SD for structural descriptors across eight cell types and species, highlighting both conserved and variable features.
• Neurons generally have larger soma and domain, longer branches, and fewer primary projections and lower soma-surface coverage by projections than glia.
• Shape descriptors show high FA and low orientation dispersion for polarized neuronal types, while glial projections are more dispersed with lower FA.
• Distributions of soma radius, branch length, and intra-cellular residence times are presented and can be used to parameterize GM diffusion models.
• Topological persistence analyses reveal clear distinctions between glial and neuronal organization, with cerebellar neurons exhibiting distinct topology compared to cortical neurons.
• The work provides reference values and distributional info to guide the development and validation of GM diffusion models (e.g., SANDI-like frameworks) and acquisition design.

Figure 2: Illustration of the structural descriptors investigated for an exemplar cell. We estimated general features of the whole structure and separated soma from projections, processing them individually to estimate a set of other relevant features. Additionally, we display the Gaussian curvature