Item

Abstract

Purpose/Introduction: Inversion-recovery bSSFP has first been proposed for T1-quantification [1], and afterwards for simultaneous T2-quantification [2]. However, the latter is highly sensitive to B0- and B1-inhomogeneity/errors [2, 3], making its quantification currently impossible at ultra high-field. T1-quantification is much more stable against these effects, especially for low flip angles [2, 3]. Thus, we restrict this work’s focus on improving T1-quantification at UHF. Subjects and Methods: IR-bSSFP can be described by a threeparameter exponential relaxation [1, 2]: S0 (signal after inversion), SStSt (steady-state signal), and T1 * (apparent relaxation). T1 and T2 can be calculated from these parameters [2, 4]. S0 can also be obtained from a low-flipangle pre-scan [6] or replaced by a T1/T2- assumption. Experiments were performed at 9.4T on a healthy volunteer with ethics approval. Whole-brain IR-bSSFP and a 4-S0-reference scan were performed using a custom-built head-coil [6]: nominal FA = 10, TR = 3 ms, 0.8 mm resolution, 14 frames, 72 shots (each: 5 s acq. + 4 s relaxation delay), TA = 10:48 min. For comparison, a MP2RAGE protocol [7, 8] was acquired. T1-maps were calculated based on a flip angle map [9] and with an inversion efficiency of 90 (from simulation). Results: Fig. 1 shows T1 * and estimated T1 vs. true T1 depending on T1/T2 (simulation). Experimental results are shown in Fig. 2 and Table 1. The T1/T2-assumption-based T1-quantification shows higher B0-sensitivity than the other IR-bSSFP data. Discussion/Conclusion: The IR-bSSFP T1-maps are comparable to the MP2RAGE’s. However, WM-T1 for the S0-reference-based data is higher, probably due to MT-effects that are lower in the lowflipangle reference. Furthermore, GM-T1 is lower in the T1/T2-assumption-based quantification (accuracy may be improved using a lower flip angle). A general problem is, that it is difficult to capture the fast dynamics after inversion, necessary for accurate S0-extrapolation. Thus, the prescan and the assumption-based approach potentially allow for sparser sampling of the recovery curve as only T1 * and SStSt are obtained from the fit. Likewise, these approaches do not require a relaxation delay between inversion pulses.