Accelerated MR parameter mapping with low-rank and sparsity constraints.

Area of Science:

• Medical Imaging
• Biophysics
• Computational Science

Background:

• Accelerated data acquisition is crucial for efficient magnetic resonance (MR) parameter mapping.
• Sparse sampling techniques offer potential for faster MR imaging but require advanced reconstruction methods.
• Existing methods often rely on single constraints, limiting performance with high acceleration.

Purpose of the Study:

• To develop and evaluate a novel constrained reconstruction method for accelerated MR parameter mapping.
• To leverage recent advances in constrained imaging with sparse sampling for improved accuracy.
• To enable faster and more precise MR parameter mapping.

Main Methods:

• A new constrained reconstruction method is proposed, integrating low-rank and joint sparsity constraints.
• The method unifies these constraints within a single mathematical formulation for contrast-weighted image sequences.
• A convex optimization problem is solved using an alternating direction method of multipliers algorithm.

Main Results:

• The method was evaluated for T2 mapping (human brain) and T1 mapping (rat brain) at moderate and high acceleration levels.
• Compared to methods using single constraints, the proposed approach demonstrated superior accuracy, especially with highly undersampled data.
• Accurate parameter estimation was achieved even with significantly accelerated data acquisition.

Conclusions:

• Simultaneously applying low-rank and sparsity constraints significantly enhances the accuracy of fast MR parameter mapping.
• The proposed method offers a robust solution for accelerated MR imaging with sparse sampling.
• This approach advances the field of quantitative MR imaging.