A comparative study of gradient nonlinearity correction strategies for processing diffusion data obtained with...

Area of Science:

• Medical Imaging
• Neuroscience
• Biophysics

Background:

• Diffusion MRI data analysis requires corrections for gradient nonlinearities.
• Existing preprocessing pipelines do not account for motion-induced spatio-temporal B-matrix variations.
• These variations can introduce systematic bias in parameter estimates.

Purpose of the Study:

• To compare the performance of established diffusion MRI preprocessing pipelines.
• To introduce and evaluate a novel spatio-temporal B-matrix tracking (STB) framework.
• To assess the impact of subject motion on diffusion MRI data under gradient nonlinearities.

Main Methods:

• Diffusion Tensor MRI (DT-MRI) data acquired with a 300 mT/m gradient system.
• Data collected from volunteers in regions with significant gradient nonlinearities.
• Comparison of six processing pipelines, including the proposed STB approach.

Main Results:

• Neglecting gradient nonlinearities led to up to 30% errors in DT-MRI parameter estimates.
• The order of applying corrections (e.g., B0 inhomogeneity, eddy currents, gradient nonlinearities) significantly affected parameter estimate consistency.
• The STB approach demonstrated the most consistent parameter estimates under large gradient nonlinearities.

Conclusions:

• Preprocessing pipeline choice critically impacts diffusion parameter estimation under gradient nonlinearities.
• Motion-induced spatio-temporal B-matrix variations cause systematic bias, which STB can mitigate.
• The STB framework offers improved accuracy and consistency for diffusion MRI analysis.