A new framework for MR diffusion tensor distribution.

Area of Science:

• Biomedical Imaging
• Neuroscience
• Diffusion MRI

Background:

• Characterizing intravoxel water diffusion heterogeneity and anisotropy non-invasively is crucial for biological, neurological, and medical research.
• Current MRI techniques face limitations in resolving diffusion processes at the picoliter scale within a microliter voxel.

Purpose of the Study:

• To develop a new theoretical framework and experimental design for describing and measuring intravoxel structural heterogeneity and anisotropy.
• To extend the diffusion tensor distribution (DTD) paradigm to a wider range of b-values while maintaining descriptive richness.

Main Methods:

• Proposed a constrained normal tensor-variate distribution (CNTVD) to model diffusion tensor variability within voxels.
• Introduced a Monte Carlo (MC) scheme for synthesizing 6D DTD numerical phantoms and inverting MR signals.
• Developed a robust estimation pipeline assessing parameter accuracy with added noise.

Main Results:

• Demonstrated that the MR signal inversion is well-posed.
• Successfully estimated CNTVD parameters parsimoniously by leveraging tensor symmetries.
• Presented invariant parameters and glyphs for characterizing microscopic heterogeneity within voxels.

Conclusions:

• The proposed CNTVD framework and MC scheme offer a powerful approach for detailed intravoxel diffusion analysis.
• This method enhances the characterization of microscopic tissue structure and diffusion properties non-invasively.
• The findings have significant implications for advancing biological, neuroscience, and medical imaging applications.