Rank-One and Transformed Sparse Decomposition for Dynamic Cardiac MRI

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Cardiovascular Research

Background:

• Dynamic cardiac MRI requires high spatiotemporal resolution for accurate diagnosis.
• Reconstructing cardiac MRI from under-sampled k-t space data is crucial for reducing scan times and motion artifacts.
• Classical low-rank and sparse models have limitations in fully exploiting data correlations.

Purpose of the Study:

• To introduce novel models and algorithms for reconstructing dynamic cardiac MRI data from under-sampled k-t space.
• To develop advanced reconstruction techniques that surpass the performance of traditional methods.
• To improve the quality and efficiency of dynamic cardiac MRI.

Main Methods:

• Proposed two novel reconstruction models: a rank-one model and a transformed sparse model.
• Developed two algorithms: Projected Alternative Direction Method (PADM) and Alternative Hard Thresholding Method (AHTM) to solve the proposed models.
• Applied these methods to reconstruct under-sampled k-t space cardiac MRI data.

Main Results:

• The proposed rank-one and transformed sparse models effectively exploit data correlations.
• PADM and AHTM algorithms demonstrate improved performance in reconstructing dynamic cardiac MRI.
• Numerical experiments showed enhanced results for cardiac perfusion and cardiac cine MRI.

Conclusions:

• The novel models and algorithms offer a promising approach for high-resolution dynamic cardiac MRI.
• These methods can significantly improve the reconstruction of cardiac MRI from under-sampled data.
• The findings suggest potential for enhanced diagnostic capabilities in cardiovascular imaging.