Measuring T1 using MP2RAGE in Human Brain at 7T – Effect of B1+ and Inversion Pulse Efficiency

Abstract: Introduction: The MP2RAGE technique has become popular for structural MRI at ultra-high fields. The core idea behind MP2RAGE is to reduce the influence of flip angle inhomogeneity (B +), proton density and T2' by calculating the regularized signed ratio of two complex MP-RAGE images acquired at different inversion times (TI) within the cycle. From this MP2RAGE signal, the underlying T1 relaxation time can be determined by means of a look-up table obtained by forward signal modelling. Nonetheless, persistent B + influence on MP2RAGE may be observed which can reduce the accuracy of the calculated T1 maps. An additional factor that might influence the estimated T1 is the efficiency of the adiabatic inversion pulse. To study (and potentially mitigate) the influence of these variables, the dynamics of longitudinal magnetization during the cycle is simulated. The sensitivity of a given (and later modified) MP2RAGE protocol to B + is studied. The efficiency of the inversion pulse is estimated for different adiabatic pulses through both experiments and simulations. Methods: Healthy adult subjects were examined on a 7T MR system using a dual-transmit head coil with a 32-channel receive array. The standard MP2RAGE protocol featured flip angles of α1/α2 = 5o/3o, TR= 6.8 ms, TI1/TI2 = 900/2750 ms, turbo factor TF=256, and a 5000 ms cycle duration. A DREAM sequence protocol with preparation flip angles of 25°,40°,60°,90° was used for B + mapping. To study changes in inversion efficiency, finv, due to a spatially varying B + field, the maximum amplitude of the inversion pulse (B1-max) was varied between 3 µT and 20 µT while keeping the pulse duration constant. To explore potential improvements in finv and residual B + bias respectively, the standard “Full adiabatic” inversion pulse was replaced by a FOCI pulse and the flip angles were changed to α1/α2 = 4o/5o. Simulations were performed in Python and T1-mapping in MATLAB using a script provided by the authors of MP2RAGE. Results: Simulation of the standard protocol showed optimal T1 accuracy when utilizing a separate B+ map and an inversion factor of 0.94 for the “adiabatic full” inversion pulse at B1-max =18 µT. At B1-max = 6 µT, finv was as low as 0.8. The inversion efficiency increased across the brain with increasing B +. The inversion efficiency also seems to increase with B -max. An empirical dependence of finv on B + was derived, by which the accuracy of the T1-maps can be improved. Additionally, the FOCI pulse yielded higher inversion efficiency at low B + regions in cerebellum and temporal lobes. Lastly, a higher second flip angle seems to result in T1-maps/MP2RAGE images that are less susceptible to B + inhomogeneity. Discussion: The influence of B + inhomogeneity on a high-resolution MP2RAGE protocol at 7T was studied by simulation and experiment. Separately acquired B +-maps improved the accuracy of T1. In particular, the efficiency of the standard “Full adiabatic” inversion pulse depended on B + and the chosen B1-max. A similar investigation has been performed by Hagberg et al. at 9.4T. Notwithstanding better performance of the FOCI pulse in low- B + regions of the brain, B -max of the “Full adiabatic” pulse should be chosen as high as possible. The results prepare for further optimization of the MP2RAGE protocol and T1-mapping at UHF.