Diffusion-Weighted Imaging of the Abdomen: Correction for Gradient Nonlinearity Bias in Apparent Diffusion...

Area of Science:

• Medical Imaging
• Biophysics
• Radiology

Background:

• Gradient nonlinearity (GNL) introduces spatial nonuniformity bias in apparent diffusion coefficient (ADC) measurements, particularly at greater distances from the magnet isocenter.
• This bias can affect the accuracy of quantitative diffusion-weighted imaging (DWI) analysis.

Purpose of the Study:

• To investigate the impact of GNL on abdominal ADC measurements.
• To develop and validate a procedure for correcting GNL-induced ADC bias in abdominal imaging.

Main Methods:

• Retrospective analysis using two homemade water phantoms and 25 patients with pancreatic ductal adenocarcinoma (PDAC).
• Diffusion-weighted imaging (DWI) acquired on a 3.0T MRI scanner using a single-shot echo-planar imaging sequence.
• Development of spatial-dependent ADC bias correctors derived from phantom data; application and validation in phantom and abdominal tissues (liver, PDAC, spleen, kidneys, paraspinal muscles).

Main Results:

• Significant ADC bias was observed in phantoms, varying by direction and offset distance from the isocenter (e.g., 9.7% and 12.6% higher in right-left and anterior-posterior directions at 150 mm).
• The proposed correction method and a theoretical GNL correction method effectively reduced ADC bias and relative ADC errors in phantom studies.
• Post-correction, abdominal tissue ADCs were generally lower, except for the left liver lobe, indicating successful bias mitigation.

Conclusions:

• GNL-induced bias in abdominal ADC measurements can be accurately quantified using a DWI phantom.
• The developed correction procedure effectively mitigates GNL bias in abdominal ADC measurements, improving quantitative accuracy.