Liver fibrosis occurs as a result of long standing chronic liver disease of various etiologies (ranging from infectious to immunological)¹ . Reversibility of liver fibrosis has generated considerable attention lately². Hence, early detection of increased liver stiffness would potentially guide towards more effective treatments. Although liver biopsy remains the gold standard for assessing fibrosis, it is invasive, expensive, and suffers from poor reproducibility³. MR-Tagging^4,5 is a promising non-invasive technique for assessment of liver mechanics using cardiac-induced motion of the liver without the need for an external device. Tracking the deformation of the tag lines in the liver showed significant differences in measured strain between healthy and cirrhotic patients^6-8. In this study, we developed an accelerated fast harmonic phase⁹ (fastHARP) MR tagging sequence for liver strain quantification with ramped flip angle and larger volumetric coverage for a tolerable and better assessment of liver mechanics.

Data Acquisition

Nine healthy adult volunteers (5 females, 4 males, age = 39.8 ± 15.1 y.o., BMI = 23.6 ± 3.6 Kg/m²) were included. Data were acquired on a 3T Siemens Verio scanner. An in-house cardiac-gated gradient echo based 1-1 spatial modulation of magnetization (SPAMM) sequence was developed incorporating ramping of the excitation RF flip angles to compensate for tag fading as the cardiac cycle progresses. Prior to tagging, an inversion recovery MOLLI sequence was utilized for T1 quantification in the liver. An average T1 estimate from three ROIs in the left lobe of the liver was incorporated into the iterative calculation of the ramped RF flip angles in the tagging protocols. Three coronal view slices passing through the left ventricle of the heart were selected at the location where the heart is seen most impacting the liver and imaged using the full SPAMM tagging and fastHARP tagging protocols. Imaging parameters for the full SPAMM sequence were TR/TE = 59/3.54 ms, FOV = 350 mm, flip angle = 10°, matrix size = 180 x 162 for an in-plane resolution of 2.2 x 1.9 mm, slice thickness = 8 mm, pixel bandwidth = 201 Hz/Pixel, tagging distance = 7 mm, and cardiac phases = 12. Imaging parameters for the fastHARP sequence were as follows: TR/TE = 60/1.58 ms, FOV = 400 mm, flip angle = 10°, matrix size = 64 x 64, slice thickness = 8 mm, pixel bandwidth = 1953 Hz/Pixel, tagging distance = 7 mm, and heart phases = 12. Fat suppression was enabled for all protocols.

Analysis

The harmonic peak containing all the tag motion information was isolated using a band-pass filter in the frequency domain. Motion and Eulerian principle strain components were calculated using harmonic phase analysis (HARP) in multiple 1 cm² user-defined regions in the upper left lobe near the heart-liver contact. Peak principle strain was determined using MATLAB. Correlation between SPAMM and fastHARP measurements and Bland-Altman assessment were performed using MedCalc.

All nine subjects completed the MR scans successfully. The full SPAMM tagging sequences took 6 breath-holds total (each breath-hold ≈ 14-18 seconds) for the 3 slices to acquire horizontal and vertical tags (for total acquisition time ≈ 90 seconds). The fastHARP sequence only took 2 breath-holds that were substantially shorter (each ≈ 8-9 seconds) to capture all 3 slices with horizontal and vertical tag lines (for total acquisition time ≈ 18 seconds). Each slice took approximately 3-4 heartbeats to acquire with fastHARP. Fig. 1 shows a representative subject’s tagged liver with the myocardium at end-systole with zoomed insets of the area where the myocardium borders the left lobe of the liver at end-diastole (left inset) and at end-systole (right inset). The tag lines are seen pulled upwards as the left ventricle contracts. Fig. 2 shows the same slice when acquired with the fastHARP sequence. The left inset shows when the myocardium is at end-diastole whereas the right inset shows when the myocardium is at end-systole. The SNR is highest at the first cardiac phase and decreases gradually as the cardiac phases progress. Fig. 3 shows the strain maps using fastHARP (left panels) and HARP (right panels). Peak strain values using fastHARP and HARP from the three acquired slices across all subjects showed a strong correlation with a high significance (Correlation coefficient R = 0.79 , p <0.05) as shown in Fig. 4, left plot, with Bland-Altman assessment, right plot.Utilizing fastHARP for accelerated acquisition of liver tagging MRI to assess liver strain allows larger volumetric coverage and a substantially shorter acquisition time (≈ 80% reduction of total acquisition) than conventional tagging whilst delivering comparable results.