The modified Look-Locker inversion recovery (MOLLI) T1 mapping sequence has been extensively used in cardiovascular MRI and also proved to be useful in the characterization of hepatic tissue¹. Livers typically contain fat in the range of 1% to 50% (with the 10-20% range being far from rare), which is known to have short T1. Since water has long T1, the simplest assumption would be that the T1 of a fatty liver would be a weighted sum of the T1s of water and fat. This is not seen in practice when using MOLLI at 3T: higher T1s are measured in fatty livers. The purpose of this study was to explore the effect of hepatic lipid concentrations (HLC) on MOLLI T1 maps at 3T.MOLLI uses balanced steady-state free precession (bSSFP) readouts after inversions. When using a bSSFP TR close to the reciprocal of the fat-water chemical shift difference at 3T, i.e. TR≈2.3 ms, the fat and water signals will be out of phase (fig. 1). According to the partial volume model, the resulting signal from a voxel containing both water and fat will be the sum of the signals of the two components. Fitting the signal recovery of this mixture will result in a T1 value that is biased.

Simulations: Bloch equations were simulated for a MOLLI sequence with one inversion pulse followed by five bSSFP readouts. A two-compartment model of the liver was built, including liver parenchyma and fat vacuoles, and signals for these compartments were simulated separately. The resulting signals were combined by varying fat fraction from 0% to 100% in steps of 1%. Since iron deposits reduce relaxation times of tissue, their effect was also included in the simulation^2,3 by a Lorentzian weighting of isochromats over a -4000 Hz to 4000 Hz frequency range. Normal iron concentration was set to 1.2 mg/g. The fat signal was simulated using a six-peak spectral model⁴. Imaging parameters were: 192×144 matrix, FA=35°, TR=2.3 ms, TI_min=100 ms, TI_max=3600 ms, RR=900 ms. Differences between the MOLLI T1 for different TR values found in the literature were also explored^1,5.

Phantoms: Margarine with 30% fat content was used (Flora Light, Unilever). A sample of margarine was centrifuged to separate water and fat components. T1 values for the three samples were determined with spin echo inversion recovery (SE-IR) and from MOLLI maps at 3T. SE-IR sequence parameters were: 128×128 matrix, TR/TE=10000/7.4 ms, 11 TI values between 50 ms and 6500 ms. The MOLLI parameters were the same as in simulations, except TI_max=4100 ms, RR=1000 ms.

Patients: N=8 patients undergoing bariatric surgery were scanned pre- and six months post-operatively. MOLLI T1 maps, T2* maps and STEAM ¹H spectra were acquired in each case.

Simulations have shown that MOLLI T1 increases with increasing levels of HLC (fig. 2) but is less sensitive to TR changes (fig. 3). MOLLI T1 also showed a dependence on off-resonance frequencies, due to the multiple peak fat model and the response of the adiabatic inversion pulse used. Phantom MOLLI scans resulted in T1_margarine=3874 ms, T1_fat=291 ms and T1_water=2266 ms while T1s determined from SE-IR images were T1_margarine=1313 ms, T1_fat=325 ms and T1_water=2448 ms. Patient measurements revealed agreement with simulation-predicted T1 behavior in patients with absolute change in HLC larger than 1% (fig. 4). One patient was excluded because fitting of pre-operative data failed, due to HLC=39%.This study shows that increasing fat fractions cause an unexpected increase of MOLLI T1s in the liver at 3T, when using bSSFP TR=2.3 ms. At 3T, small variations of TR do not change drastically the behavior of T1: there is 0.98% relative difference between the TR=2.14 ms and TR=2.3 ms curves and 1.96% relative difference between the TR=2.3 ms and TR=2.6 ms curves. Patient measurements were in agreement with simulations, however, patient livers may have some level of inflammation or fibrosis causing elevations in T1 (fig. 4).Although the T1 of fat is short, it has an additive effect on overall MOLLI T1 when using bSSFP TR=2.3 ms at 3T. Since fatty liver is a common occurrence, it is essential to develop a correction mechanism to counterbalance its effect.