Angular Upsampling in Infant Diffusion MRI Using Neighborhood Matching in x-q Space.

Area of Science:

• Medical Imaging
• Biophysics
• Computational Neuroscience

Background:

• Diffusion MRI (dMRI) requires extensive q-space coverage for accurate water diffusion mapping.
• Long acquisition times in dMRI pose challenges for pediatric and immobile patients.
• Current interpolation methods may not fully capture complex diffusion patterns.

Purpose of the Study:

• To develop an advanced q-space upsampling technique for diffusion MRI.
• To improve the quality and resolution of dMRI data, especially for challenging populations.
• To address the limitations of existing interpolation methods in dMRI.

Main Methods:

• Harnessing non-local self-similar information within the x-q space of dMRI data.
• Employing neighborhood matching to identify signal relationships in x-q space.
• Utilizing these relationships to regularize an ill-posed inverse problem for high angular resolution dMRI estimation.

Main Results:

• The proposed method effectively upsamples q-space in diffusion MRI data.
• Demonstrated superior performance compared to spherical radial basis functions and spherical harmonics interpolation.
• Achieved higher qualitative and quantitative quality in high angular resolution dMRI reconstruction.

Conclusions:

• The novel x-q space self-similarity approach significantly enhances dMRI data quality.
• This method offers a promising solution for reducing scan times and improving dMRI analysis in vulnerable populations.
• The framework effectively utilizes white matter structure information for robust dMRI reconstruction.