Correlation-Weighted Sparse Representation for Robust Liver DCE-MRI Decomposition Registration.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Motion artifacts significantly degrade the diagnostic quality of liver dynamic contrast-enhanced magnetic resonance (DCE-MR) imaging.
• Traditional intensity-based registration methods fail due to significant intensity variations introduced by contrast agents.

Purpose of the Study:

• To develop a robust framework for motion correction in liver DCE-MR imaging.
• To address the challenge of intensity variations caused by contrast agents during motion correction.

Main Methods:

• A correlation-weighted sparse representation framework was proposed to separate contrast agents from original liver DCE-MR images.
• The method recovers ideal contrast agent curves from a transpose dictionary (t-3D mode) using a "correlation-weighted" constraint.
• De-enhanced images were registered, and deformation fields were applied to original DCE-MR images for motion correction.

Main Results:

• The proposed method successfully separates contrast enhancement components, enabling robust registration without intensity variances.
• Experimental results on simulated and real data demonstrate superior registration performance compared to state-of-the-art methods.
• The framework achieves improved accuracy with reduced computational cost.

Conclusions:

• The correlation-weighted sparse representation framework offers an effective solution for motion correction in liver DCE-MR imaging.
• This approach mitigates the impact of contrast agent-induced intensity variations, enhancing image registration accuracy.
• The method provides a more efficient and accurate alternative for DCE-MR motion correction.