Q-space imaging based on Gaussian radial basis function with Laplace regularization.

Area of Science:

• Diffusion MRI
• Neuroimaging
• Biomedical Engineering

Background:

• Diffusion MRI (dMRI) is crucial for non-invasively probing neural microstructure.
• Reconstructing the ensemble average diffusion propagator (EAP) is key to understanding water diffusion in biological tissues.
• Current methods face challenges with sparse and noisy data, limiting accuracy.

Purpose of the Study:

• To enhance dMRI by incorporating spherical harmonics (SH) basis into Gaussian radial basis function (GRBF) q-space imaging.
• To achieve robust reconstruction of the EAP.
• To improve microstructure imaging and orientation distribution function (ODF) estimation.

Main Methods:

• Introduced Laplacian regularization into the GRBF-based dMRI method.
• Derived microstructure imaging indicators and ODFs from EAP.
• Combined the method with multi-compartment models for axon diameter calculation.
• Evaluated results through qualitative and quantitative signal fitting assessments.

Main Results:

• Achieved significant accuracy improvements in signal reconstruction.
• Estimated ODFs exhibited sharper profiles and fewer spurious peaks, even in sparse, noisy conditions.
• Reduced mean and standard deviation of axon diameter estimates compared to state-of-the-art methods in most experiments.

Conclusions:

• The proposed SH-GRBF method outperforms standard GRBF in signal fitting and EAP/ODF estimation with multi-shell sparse samples.
• Demonstrates potential for accurate microstructure feature recovery with reduced uncertainty when combined with multi-compartment models.