Neural orientation distribution fields for estimation and uncertainty quantification in diffusion MRI.

Area of Science:

• Neuroimaging
• Diffusion MRI Analysis
• Computational Neuroscience

Background:

• Accurate estimation of the orientation distribution function (ODF) is crucial for in-vivo brain connectivity and structure inference from diffusion MRI (dMRI).
• Estimating ODFs presents challenges due to noise, high-dimensional parameter spaces, and sparse angular measurements in dMRI signals.

Purpose of the Study:

• To develop a novel deep learning methodology for continuous estimation and uncertainty quantification of the spatially varying ODF field.
• To improve the efficiency and accuracy of ODF estimation, particularly in sparse and noisy dMRI data regimes.

Main Methods:

• Utilized a neural field (NF) to parameterize a random series representation of latent ODFs, incorporating spatial correlations.
• Derived an analytic approximation for posterior predictive distribution to quantify ODF estimation uncertainty.
• Evaluated the method on synthetic and real in-vivo diffusion MRI data.

Main Results:

• The proposed deep learning method demonstrates improved efficiency and accuracy in estimating ODFs compared to existing approaches.
• Successfully quantified uncertainty in ODF estimates without resorting to computationally expensive resampling methods.
• Validated the approach on both simulated and real-world in-vivo dMRI datasets.

Conclusions:

• The novel deep learning approach offers a robust and efficient solution for ODF estimation and uncertainty quantification in dMRI.
• This method has the potential to advance in-vivo brain connectivity and structure analysis by overcoming key estimation challenges.