CATERPillar: a flexible framework for generating white matter numerical substrates with incorporated glial cells

Area of Science:

• Neuroscience
• Biophysics
• Medical Imaging

Background:

• Diffusion MRI (dMRI) signal sensitivity to microstructure is key for understanding brain tissue.
• Accurate numerical phantoms representing cerebral white matter (WM) are essential for dMRI simulations.

Purpose of the Study:

• Introduce CATERPillar (Computational Axonal Threading Engine for Realistic Proliferation), a novel method for generating realistic numerical substrates for dMRI simulations.
• Enhance biological fidelity in simulations by generating realistic axonal and glial cell structures.

Main Methods:

• Simulate axonal growth using overlapping spheres as elementary units with collision prevention.
• Control key structural parameters: cellular density, undulation, beading, and myelination.
• Generate realistic glial cells alongside axonal structures.

Main Results:

• Generated substrates exhibit morphological parameter distributions consistent with histological studies.
• Quantitative validation of astrocytic component realism using Sholl analysis.
• Simulated diffusion accurately reflects theoretical models of short-range disorder in extra- and intra-axonal compartments.

Conclusions:

• CATERPillar provides realistic numerical phantoms for dMRI simulations of cerebral white matter.
• The open-source tool aids in developing new acquisition schemes, testing analytical models, and training machine learning models.