Practical estimate of gradient nonlinearity for implementation of apparent diffusion coefficient bias correction.

Area of Science:

• Medical Imaging
• Physics
• Biophysics

Background:

• Gradient nonlinearity in MRI scanners introduces bias in apparent diffusion coefficient (ADC) measurements.
• Accurate ADC values are crucial for quantitative analysis in diffusion-weighted imaging (DWI).

Purpose of the Study:

• To develop and validate an efficient empirical procedure for characterizing and correcting gradient nonlinearity bias on clinical MRI scanners.
• To reduce apparent diffusion coefficient (ADC) bias in diffusion-weighted imaging (DWI).

Main Methods:

• Spatial nonlinearity scalars were estimated using an e-water phantom and diffusion measurements.
• A literature-based digital nonlinearity model was rescaled with system-specific scalars to create 3D bias correction maps.
• Correction efficacy was validated by comparing corrected ADC values to unbiased measurements at the isocenter.

Main Results:

• Empirical nonlinearity scalars correlated with geometric distortion measurements.
• The correction reduced ADC bias from 20% to 2% at clinically relevant offsets for both isotropic and anisotropic media.
• Correction using either adjusted DWI intensities or b-values yielded similar results in brain and ice-water imaging.

Conclusions:

• Empirical adjustment of a gradient nonlinearity model effectively corrects DWI bias on clinical MRI scanners.
• The observed ADC bias correction efficiency aligns with theoretical predictions and simulations.
• This procedure offers a benchmark for evaluating nonlinearity bias correction methods in clinical MRI.