Characteristic-constrained accelerating MR T1rho mapping with blockwise infimal convolution of matrix elastic-net...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Quantitative MRI

Background:

• Magnetic Resonance Parameter Mapping (MRPM) is crucial for clinical and research applications.
• Accelerating MRPM acquisition remains a significant challenge for improving its utility.

Purpose of the Study:

• To develop an accelerated Magnetic Resonance T1rho (MR T1rho) mapping technique using a novel undersampled k-space based joint multi-contrast image reconstruction approach.
• Introduce the CC-IC-LMEN method for enhanced MR T1rho imaging.

Main Methods:

• The CC-IC-LMEN method employs a blockwise low-rank assumption in the characteristic-image series (c-p space).
• Infimal convolution (IC) is used to leverage generalized low-rank properties across different c-p spaces for improved accuracy.
• Matrix elastic-net (MEN) regularization is incorporated for stable solutions under high acceleration and noisy conditions.
• The reconstruction problem is solved using an alternating direction method of multipliers (ADMM) algorithm.
• T1rho maps are generated via nonlinear least-squares (NLSQ) fitting.

Main Results:

• The CC-IC-LMEN approach demonstrated superior accuracy compared to existing methods.
• Performance was validated using relative L2-norm error (RLNE) and structural similarity (SSIM) metrics.
• The method proved effective even with heavy undersampling and noisy observations.

Conclusions:

• The proposed CC-IC-LMEN method offers accurate and robust solutions for accelerated MR T1rho mapping.
• This advancement has the potential to enhance clinical applications and biomedical research utilizing MR T1rho imaging.