Truncated enhanced constraint for low-rank plus sparse in cardiac dynamic MRI reconstruction

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Cardiovascular Imaging

Background:

• Dynamic MRI is crucial for cardiac cine and perfusion imaging.
• Low-rank and sparse priors are essential for reconstructing high-resolution dynamic MRI from undersampled data.
• Inaccurate singular value estimation can compromise low-rank regularization and image reconstruction quality.

Purpose of the Study:

• To introduce a novel reconstruction method, TRIONES, for dynamic MRI (dMRI).
• To address the challenge of accurate rank function estimation in cardiac dMRI.
• To enhance the precision of low-rank regularization in cardiac dynamic MRI.

Main Methods:

• TRIONES utilizes a truncated nuclear norm for improved singular vector weighting.
• This approach assigns differential shrinkage values to singular vectors, preserving key image data while reducing noise.
• The method was evaluated using both simulated and in vivo cardiac dynamic MRI datasets.

Main Results:

• TRIONES demonstrated superior performance compared to existing reconstruction methods.
• The proposed method achieved higher Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM).
• It also resulted in a smaller Normalized Root Mean Squared Error (NRMSE) in both simulated and in vivo experiments.

Conclusions:

• TRIONES offers accurate rank estimation, significantly improving low-rank based dynamic MRI reconstruction.
• The method enhances image quality, providing more accurate and efficient information for cardiac applications.
• This advancement holds potential benefits for the clinical diagnosis of heart-related diseases.