Acceleration of chemical shift encoding-based water fat MRI for liver proton density fat fraction and T2* mapping using compressed sensing

Fabian Lohöfer, Georgios Kaissis, Christina Müller-Leisse, Daniela Franz, Christoph Katemann, Andreas Hock, Johannes Peeters, Ernst J. Rummeny, Dimitrios C. Karampinos, Rickmer BrarenmedRxiv 2019 · DOI

Introduction

Non-Alcoholic Fatty Liver Disease (NAFLD) and hepatic iron overload represent significant global health burdens, often serving as precursors to cirrhosis and hepatocellular carcinoma. Magnetic Resonance Imaging (MRI) has emerged as the non-invasive gold standard for assessing these conditions through Proton Density Fat Fraction (PDFF) and T2* mapping. However, the clinical utility of these quantitative metrics is often hampered by patient-related factors—specifically, the ability to maintain a stable breath-hold.

Standard multi-echo Dixon sequences typically require breath-holds of approximately 15 seconds. For elderly, pediatric, or severely ill patients, this duration is often unattainable, leading to respiratory motion artifacts that can degrade image quality and bias quantitative measurements. This study evaluates a hybrid acceleration approach combining Parallel Imaging (SENSE) with Compressed Sensing (CS) to reduce scan times below the critical 10-second threshold without sacrificing diagnostic accuracy.

Methodology

The researchers conducted a comparative study on a 3T MRI scanner using a six-echo Dixon sequence. Two distinct acquisition strategies were evaluated:

  1. Standard SENSE: A traditional parallel imaging technique utilizing an acceleration factor of 2.64.
  2. CS+SENSE (Proposed): A hybrid approach combining Compressed Sensing and SENSE to achieve a higher total acceleration factor of 4.0.

Compressed Sensing leverages the mathematical "sparsity" of medical images, allowing for the undersampling of k-space (raw frequency data). The full image is then reconstructed using iterative algorithms. In this study, a two-step reconstruction was employed: first, reconstructing multi-echo complex images via the CS+SENSE algorithm, followed by water-fat separation to generate the final PDFF and T2* maps.

Clinical Evaluation

The method was validated on a clinical cohort of 89 patients (average age 62) undergoing routine liver MRI at the Technical University of Munich. To ensure the study focused on diffuse liver tissue quantification, patients with primary or metastatic liver tumors were excluded. Quantitative measurements were performed and compared across all eight liver segments to ensure spatial consistency.

Key Findings

1. Significant Temporal Acceleration

The integration of Compressed Sensing reduced the total scan duration from 14.5 seconds (SENSE) to 9.3 seconds (CS+SENSE). This 35% reduction in scan time is clinically significant, as it brings the breath-hold requirement well within the capabilities of most patients.

2. Drastic Reduction in Motion Artifacts

The most striking clinical benefit was the impact on data integrity. Significant motion artifacts were present in 12.3% of the standard SENSE acquisitions. In contrast, the accelerated CS+SENSE acquisitions showed motion artifacts in only 1.1% of cases. By shortening the scan, the researchers effectively eliminated the primary source of image degradation.

3. Statistical Equivalence in Quantification

Crucially, the increased acceleration did not come at the cost of precision. There were no statistically significant differences (P > 0.05) in PDFF or T2* values between the two methods across any liver segments. For instance, in segment VII, the mean PDFF was 6.37% with SENSE compared to 6.28% with CS+SENSE. The data also successfully preserved known physiological trends, such as the slightly higher fat distribution typically found in the right liver lobe.

Discussion and Limitations

This study demonstrates that CS-based acceleration is "transparent" to underlying physiological measurements. Unlike retrospective simulations often found in literature, this prospective implementation on a standard vendor platform suggests the technology is ready for immediate clinical adoption.

However, some limitations remain. The study lacked histopathological validation (biopsy), though MRI-based PDFF is already widely accepted as a surrogate. Additionally, the research was conducted at a single center on a 3T system; further validation is required to confirm these acceleration factors at 1.5T or across different hardware manufacturers. Finally, while an acceleration factor of 4 was successful, the "ceiling" at which CS reconstruction might begin to smooth out fine details or bias T2* measurements remains to be determined.

Conclusion

Combining Compressed Sensing with Parallel Imaging allows for rapid liver PDFF and T2* mapping in under 10 seconds. This approach maintains high quantitative accuracy while virtually eliminating motion artifacts, making robust liver fat and iron quantification accessible to a broader, more vulnerable patient population.

Summary updated

Abstract

Abstract Objectives To evaluate proton density fat fraction (PDFF) and T2* measurements of the liver with combined parallel imaging (sensitivity encoding, SENSE) and compressed sensing (CS) accelerated chemical shift encoding-based water-fat separation. Methods Six-echo Dixon imaging was performed in the liver of 89 subjects. The first acquisition variant used acceleration based on SENSE with a total acceleration factor equal to 2.64 (acquisition labeled as SENSE). The second acquisition variant used acceleration based on a combination of CS with SENSE with a total acceleration factor equal to 4 (acquisition labeled as CS+SENSE). Acquisition times were compared between acquisitions and proton density fat fraction (PDFF) and T2*-values were measured and compared separately for each liver segment. Results Total scan duration was 14.5 sec for the SENSE accelerated image acquisition and 9.3 sec for the CS+SENSE accelerated image acquisition. PDFF and T2* values did not differ significantly between the two acquisitions ( paired Mann-Whitney and paired t-test P>0.05 in all cases). CS+SENSE accelerated acquisition showed reduced motion artifacts (1.1%) compared to SENSE acquisition (12.3%). Conclusion CS+SENSE accelerates liver PDFF and T2*mapping while retaining the same quantitative values as an acquisition using only SENSE and reduces motion artifacts. Strengths of this study Compressed sensing allows accelerated imaging with reduction of motion artifacts without alteration of quantitative measurements Robust results in fat and iron quantification in a heterogeneous patient cohort Limitations of this study No histopathological validation of the MR findings was performed The study was not performed at different field strengths
BibTeX
@misc{lohofer2019acceleration,
  author = {Fabian Lohöfer and Georgios Kaissis and Christina Müller-Leisse and Daniela Franz and Christoph Katemann and Andreas Hock and Johannes Peeters and Ernst J. Rummeny and Dimitrios C. Karampinos and Rickmer Braren},
  title = {{Acceleration of chemical shift encoding-based water fat MRI for liver proton density fat fraction and T2* mapping using compressed sensing}},
  year = {2019},
  doi = {10.1101/19000927},
  howpublished = {medRxiv},
}
Figure