Efficient Compressed Sensing SENSE pMRI Reconstruction With Joint Sparsity Promotion

Area of Science:

• Magnetic Resonance Imaging
• Signal Processing
• Medical Imaging

Background:

• Compressed Sensing (CS) accelerates k-space data acquisition in parallel MRI (pMRI).
• Existing CS reconstruction models for pMRI have limitations in performance and recovery guarantees.
• There is a need for improved CS reconstruction accuracy and flexibility in pMRI.

Purpose of the Study:

• To propose a novel Joint Sparsity CS SENSE (JS CS SENSE) framework for pMRI.
• To enhance reconstruction accuracy from limited k-space data.
• To improve recovery guarantees and model flexibility in CS-based pMRI.

Main Methods:

• Developed a JS CS SENSE reconstruction framework.
• Utilized split Bregman, variable splitting, and combined-variable splitting for efficient algorithm development.
• Proposed a residual-JS regularized sensitivity estimation model, extended to calibration-less (CaL) JS CS SENSE.

Main Results:

• The JS CS SENSE model offers improved recovery guarantees compared to conventional CS SENSE and is comparable to coil-by-coil CS.
• JS CS SENSE demonstrates superior reconstruction accuracy over conventional CS SENSE.
• CaL JS CS SENSE outperforms other state-of-the-art calibration-less CS methods, especially for non-piecewise constant images.

Conclusions:

• The proposed JS CS SENSE framework significantly enhances pMRI reconstruction accuracy and offers improved theoretical guarantees.
• The developed efficient algorithms facilitate fast and accurate image reconstruction.
• The calibration-less extension (CaL JS CS SENSE) provides a robust solution for advanced pMRI applications.