Improved liver fat and R 2 * quantification at 0.55 T using locally low-rank denoising

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Physics
• Quantitative Imaging

Background:

• Accurate liver fat quantification is crucial for diagnosing and monitoring liver diseases.
• Low-field MRI (0.55T) offers potential advantages but requires optimized protocols for reliable quantification.

Purpose of the Study:

• To validate an acquisition protocol for improved liver proton density fat fraction (PDFF) and T2* quantification at 0.55 Tesla.
• To investigate the performance of locally low-rank denoising methods for enhancing quantitative MRI accuracy and precision.

Main Methods:

• A Monte Carlo simulation was used to design the 0.55T protocol.
• Robust locally low-rank (RLLR) and random matrix theory (RMT) denoising methods were evaluated.
• Phantom and in vivo liver scans (11 subjects) were performed to assess quantification accuracy and precision.

Main Results:

• Both RLLR and RMT denoising significantly improved accuracy (concordance correlation coefficient >0.992) and precision (>67% decrease in standard deviation) in phantom studies.
• In vivo, RLLR and RMT denoising substantially reduced the standard deviation of PDFF and T2* measurements compared to conventional reconstruction.
• Mean PDFF and T2* values did not differ significantly between methods in vivo.

Conclusions:

• An optimized acquisition protocol for 0.55T liver PDFF and T2* quantification was successfully validated.
• Locally low-rank denoising techniques (RLLR and RMT) demonstrably enhance the accuracy and precision of quantitative liver MRI at 0.55T.