An efficient algorithm for dynamic MRI using low-rank and total variation regularizations.

Area of Science:

• Medical Imaging
• Image Reconstruction
• Magnetic Resonance Imaging

Background:

• Dynamic MR images possess spatial and temporal redundancy.
• Existing models like low-rank or sparse methods may not fully exploit this redundancy.
• Minimizing energy functions with Total Variation (TV) and Nuclear Norm (NN) regularization is computationally challenging due to non-smoothness and non-separability.

Purpose of the Study:

• To propose an efficient algorithm for dynamic MR image reconstruction.
• To leverage both spatial and temporal redundancy using TV and NN regularization.
• To overcome the computational challenges of minimizing complex energy functions.

Main Methods:

• Developed a novel algorithm based on a primal-dual form of the TVNNR model.
• The algorithm efficiently minimizes the energy function incorporating TV and NN regularization.
• Theoretical convergence rate of O(1/N) for N iterations is proven.

Main Results:

• The proposed algorithm demonstrates superior performance compared to state-of-the-art methods.
• Experiments on single-coil and multi-coil dynamic MR data validate the method.
• Significant improvements in reconstruction accuracy and reduced time complexity were observed.

Conclusions:

• The proposed primal-dual algorithm offers an efficient solution for dynamic MR image reconstruction.
• The TVNNR model effectively utilizes spatio-temporal redundancy for better data modeling.
• This method presents a promising advancement in dynamic MRI reconstruction.