PRIM: An Efficient Preconditioning Iterative Reweighted Least Squares Method for Parallel Brain MRI Reconstruction

Area of Science:

• Medical Imaging
• Computational Science

Background:

• Parallel Magnetic Resonance Imaging (pMRI) aims to reduce scan times.
• Joint Total Variation (JTV) regularization accelerates pMRI but suffers from inefficient optimization.

Purpose of the Study:

• To develop a more efficient optimization method for JTV-regularized pMRI.
• To achieve linear convergence for JTV model optimization, overcoming current sublinear rates.

Main Methods:

• Proposed a novel linear-convergent optimization method based on the Iterative Reweighted Least Squares algorithm.
• Introduced a new preconditioner to accelerate the optimization process for the complex JTV objective.

Main Results:

• The proposed method achieves linear convergence, outperforming existing sublinear methods.
• Demonstrated superior accuracy and efficiency in pMRI compared to state-of-the-art techniques through extensive experiments.

Conclusions:

• The novel optimization approach significantly enhances the performance of JTV-regularized pMRI.
• This advancement offers a more accurate and efficient solution for accelerating MRI acquisition.