Estimation of individual axon bundle properties by a Multi-Resolution Discrete-Search method

Area of Science:

• Neuroimaging and Biomedical Engineering
• Diffusion Magnetic Resonance Imaging (dMRI) Analysis

Background:

• Accurate estimation of intra-voxel diffusion properties in dMRI is crucial for understanding brain microstructure.
• Existing multi-fiber algorithms face limitations with complex microstructures, low signal-to-noise ratio (SNR), and small crossing angles.
• Robust and precise estimation of fiber bundle orientation and diffusion parameters remains a challenge in challenging configurations.

Purpose of the Study:

• To present a stable, accurate, and noise-robust method for estimating intra-voxel bundle-wise diffusion properties in dMRI.
• To overcome limitations of current multi-fiber algorithms, particularly in challenging microstructural configurations.
• To improve the estimation of intra-voxel geometry, diffusion profiles, and individual fiber bundle properties.

Main Methods:

• A Multi-Resolution Discrete-Search for accurate fiber bundle orientation and sparsity constraint.
• F-test combined with Rician bias correction for determining the number of fiber bundles.
• Simultaneous denoising and fitting exploiting spatial redundancy for noise robustness.
• Voxel-wise estimation framework for independent axial and radial diffusivity parameter determination.

Main Results:

• Significant improvements in estimating the number and orientation of fiber bundles compared to Spherical Deconvolution on multi-shell data.
• Robust estimation of axial and radial diffusivity parameters through voxel-wise diffusion profile estimation.
• Demonstrated superior performance on synthetic data, a challenging phantom, and the in-vivo MASSIVE dataset.

Conclusions:

• The proposed methodology offers a stable, accurate, and noise-robust approach for multi-fiber tractography and diffusion property estimation.
• The method enhances the understanding of individual bundle diffusion properties in complex brain microstructures.
• This work provides a novel evaluation metric and methodology for assessing dMRI analysis techniques.