Bayesian insights into exchange and restriction in gray matter diffusion MRI

Area of Science:

• Neuroimaging
• Biophysics
• Computational Neuroscience

Background:

• Diffusion MRI (dMRI) enables gray matter microstructure characterization.
• Reliability of biophysical models, especially those with water exchange, needs further study.

Purpose of the Study:

• Investigate accuracy, precision, and degeneracy of NEXI and SANDIX gray matter models.
• Assess parameter estimation under various conditions using simulated and in vivo data.
• Evaluate the utility of the µGUIDE Bayesian inference framework.

Main Methods:

• Utilized established dMRI acquisition protocols.
• Employed µGUIDE, a deep learning-based Bayesian inference framework.
• Analyzed simulated and in vivo human brain data.

Main Results:

• Some microstructural parameters (e.g., extra-cellular diffusivity, neurite signal fraction) were robustly estimated.
• Other parameters (e.g., exchange time, soma radius) exhibited high uncertainty and bias, particularly with noise and reduced protocols.
• µGUIDE effectively quantified uncertainty and detected degeneracies, flagging unreliable estimates.

Conclusions:

• Model-based dMRI parameter estimates require reporting of uncertainty and consideration of degeneracies.
• Uncertainty-aware methods offer critical advantages over traditional fitting techniques.
• Probabilistic fitting approaches are advocated for improved reproducibility and biological interpretability in neuroimaging pipelines.