Spatial image gradient estimation from the diffusion MRI profile.

Area of Science:

• Medical Imaging
• Neuroscience
• Biophysics

Background:

• Diffusion MRI (dMRI) models typically do not account for varying tissue relaxation-time properties experienced by water molecules.
• This limitation hinders the extraction of detailed information from dMRI signals.

Purpose of the Study:

• To develop a new mathematical relationship to mine dMRI signals for spatial variations in tissue relaxation-time properties.
• To enable the estimation of image spatial gradients directly from dMRI signals.

Main Methods:

• Derived a novel mathematical relationship between dMRI signal and image spatial gradient.
• Validated the method by comparing estimated gradients to gold-standard finite difference approximations.
• Assessed the impact of "fiber continuity" as a confounding factor.

Main Results:

• The image spatial gradient estimated from the new diffusion model showed significant correlation with finite difference approximations.
• The mean/median acute angle between estimated and gold-standard gradients was significantly smaller than chance (p < 10⁻¹⁰).
• Fiber continuity effects were observed but did not overlap with the primary hypothesized effect.

Conclusions:

• The study supports the hypothesized relationship between within-voxel dMRI signal and image gradient.
• The findings are not attributable to the confounding factor of fiber continuity.
• This new approach enhances the information obtainable from dMRI data.