Efficient approximate signal reconstruction for correction of gradient nonlinearities in diffusion-weighted imaging.

Area of Science:

• Medical Imaging
• Neuroimaging
• Diffusion MRI

Background:

• Hardware nonlinearities in diffusion weighted MRI (DW-MRI) introduce spatial variations in diffusion weighting.
• Existing voxel-wise empirical corrections for gradient nonlinearities are complex and require reimplementation.
• These distortions affect the accuracy of diffusion MRI analyses.

Purpose of the Study:

• To propose a novel two-step signal approximation technique for correcting gradient nonlinearity effects in DW-MRI.
• To enable seamless integration of gradient nonlinearity correction into existing diffusion MRI workflows.
• To provide an efficient alternative to current voxel-wise correction methods.

Main Methods:

• A two-step signal approximation involving signal scaling and gradient orientation resampling was developed.
• The proposed technique was validated by fitting a neurite orientation dispersion and density imaging (NODDI) model.
• The study used data from five subjects in the MASiVar pediatric dataset.

Main Results:

• The proposed technique demonstrated uniform gradients across the corrected image.
• Analysis of intra-cellular volume fraction (iVF), CSF volume fraction (cVF), and orientation dispersion index (ODI) showed Cohen's d < 0.2.
• This indicates no significant differences between the proposed approximation and voxel-wise correction techniques.

Conclusions:

• The two-step signal approximation offers an efficient method for voxel-wise gradient table correction in DW-MRI.
• This technique can be easily incorporated into existing diffusion preprocessing pipelines.
• The method is implemented in the 'PreQual' automated pipeline for DW-MRI preprocessing and quality assurance.