Motion correction of multi-b-value diffusion-weighted imaging in the liver

Area of Science:

• Medical Imaging
• Biophysics

Background:

• Motion artifacts are a significant challenge in diffusion-weighted imaging (DWI), particularly during free-breathing acquisitions.
• Low signal levels at higher b-values exacerbate motion-related errors, complicating accurate parameter quantification.

Purpose of the Study:

• To present a reliable method for reducing motion artifacts in multi-b-value DWI during free-breathing.
• To enhance the robustness of diffusion parameter quantification in the liver, even with low signal-to-noise ratios.

Main Methods:

• Employed time-resolved, multi-b-value spin-echo echo planar imaging acquisitions of the liver during free breathing.
• Utilized image registration with a normalized mutual information similarity measure to correct for motion-induced spatial misalignment.
• Estimated registration accuracy by comparing normalized root-mean-square error (NRMSE) values before and after motion correction using the biexponential intra-voxel incoherent motion model.

Main Results:

• Motion correction significantly reduced the mean NRMSE values in regions of interest (ROIs) from 0.41 ± 0.13 to 0.38 ± 0.16 (P < .05).
• Visual inspection confirmed marked improvement in the alignment of liver contours between frames in datasets with substantial respiratory motion.
• The method demonstrated increased robustness in multi-b-value acquisitions.

Conclusions:

• The proposed image registration method effectively addresses motion-related artifacts in free-breathing multi-b-value DWI.
• This technique enhances the reliability of diffusion parameter quantification, particularly in challenging low-signal conditions.
• The findings support the clinical utility of this method for abdominal DWI.