Accurate and precise in vivo liver 3D T1 mapping at 3T.

Abstract

Purpose: To develop an accurate and precise liver 3D $T_1$ mapping method using only scanner-agnostic sequences.

Methods: While the spoiled gradient-recalled echo sequence is widely available on clinical scanners, variable flip angle $T_1$ mapping methods based on this sequence provide biased $T_1$ estimates, with the largest systematic error arising from $B_1^{+}$ inhomogeneities. To correct for this, the flip angle was mapped using a 2D gradient-echo double-angle method approach. To correct for the confounding effect of fat on liver $T_1$ and $B_1^{+}$ , Dixon and fat saturation techniques were used in combination with the variable flip angle and the $B_1^{+}$ map acquisitions, respectively. The $T_1$ and $B_1^{+}$ mapping methods were validated with a $T_1$ -phantom against gold standard methods. An intra- and inter-repeatability study was conducted at 3T in 10 healthy individuals' livers.

Results: The developed 3D $T_1$ mapping method achieved an excellent agreement with the gold standard, with a weighted root mean squared normalized error below 2.8%. In vivo, a median $T_1$ standard deviation of 31 ms and an interquartile range of [27, 39] ms was achieved across all measurements, including the intra- and inter-repeatability study data. A within-subject standard deviation for $T_1$ of 21 ± 5 ms had a corresponding repeatability coefficient of 60 ms. The measured $T_1$ values agree well with MOLLI and SASHA $T_1$ mapping methods, with average $T_1$ differences of 5%.

Conclusion: Accurate and precise 3D $T_1$ liver measurements can lead the way to the wider adoption of a clinically feasible $T_1$ measurement as a marker of hepatic fibro-inflammation.