Assessing repeatability and confounding factors of magnetic resonance fingerprinting (MRF) for quantitative liver imaging.

backgroundAdvanced radiation therapy techniques are increasingly used to treat liver cancers. However, the lack of robust imaging approaches for reliable assessment of early-phase treatment response limits full potential of these therapeutic methods. PURPOSE: To evaluate the performance of a new quantitative MR imaging technique, named magnetic resonance fingerprinting (MRF), in terms of repeatability, reproducibility, and motion sensitivity for quantitative T1 and T2 assessment in hepatic tissues.

methodsA rapid 2D liver MRF technique based on a steady-state free precession sequence was adapted and performed on 3T MR scanners. Quantitative T1 and T2 relaxation time maps with a spatial resolution of 1.6 mm and slice thickness of 5 mm were acquired simultaneously in one breath-hold of 15 s per imaging slice. The accuracy of the liver MRF method was first validated using the standardized water phantom. The repeatability and reproducibility of the liver MRF were further evaluated on five healthy volunteers. Each volunteer was scanned once per week for three consecutive weeks. The effects of dietary status and respiratory motion were further evaluated on three subjects. Finally, a pilot study was performed on one patient with multiple metastatic lesions.

resultsPhantom experiments show the liver MRF method achieved an accurate assessment for T1 up to 2000 ms and T2 up to 300 ms. High-quality quantitative T1 and T2 maps were successfully obtained from all five subjects. Bland-Altman plots show that both T1 and T2 obtained from MRF are highly repeatable, with a coefficient of repeatability of 86 ms for T1 and 9 ms for T2. The quantitative measurements are also highly reproducible with a coefficient of variation of 1.5% ± 0.2% for T1 (range, 1.3%-1.7%) and 4.1% ± 1.3% for T2 (range, 2.9%-5.5%). The MRF-derived T1 and T2 measurements are independent of dietary status. However, clear motion artifacts and vessel blurring were noted in motion-contaminated scan scenarios, which had instructions for 10 s initial breath-holding followed by 5 s shallow breathing. Finally, higher T1 and T2 values were observed in the metastatic lesions as compared to the values obtained in normal liver parenchyma.

conclusionThis pilot study demonstrates that 2D liver MRF can provide high-quality, repeatable, and reproducible T1 and T2 mapping for hepatic tissues. These findings suggest potential applications in tissue/lesion characterization and longitudinal treatment response monitoring in the liver.