Introduction
The inhomogeneous transmit (B₁⁺) fiels in MRI at 7 tesla cause regions of low signal intensity. In addition, B₀ inhomogeneities get more pronounced at higher field strengths, which often result in image artifacts. Dynamic parallel transmit (pTx) pulses can be used to produce uniform flip angle (FA) distributions throughout the head despite of the present field inhomogeneities [1]. Since the design of pTx pulses makes use of subject specific B₁⁺ and B₀ maps, the performance of the pTx pulses depends on the accuracy of these fieldmaps. B₀ maps are typically acquired with multi-echo gradient echo sequences, yet these images themselves may also suffer from the mentioned field inhomogeneities [2]. The aim of this work was to implement pTx excitation pulses into a B₀ mapping sequence. The sequence was equipped with either a circularly polarized (CP) excitation pulse or a pTx universal pulse (UP) [3]. The resulting maps were used for patient specific B₀ shimming, which themselves were used for designing pTx excitation and inversion pulses for use in a 3D MPRAGE sequence.

Methods
Four B₀ maps of a single volunteer were acquired using a 3D sagittally oriented gradient echo sequence (TE1 = 2.39s, TE2 = 4.59s, TE3 = 7.09s; 4mm³; TA = 12s). The sequence made use of either a CP pulse or a pTx UP described below, as well as either default shim currents or a patient specific brain shimming with a vendor provided method. They are denoted respectively, whereby the subscript ‘CP’ or ‘UP’ denotes used excitation pulse, and an added subscript ‘shim’ indicates patient specific shimming: ‘B0_CP/B0_UP/B0_{CP, shim}/B0_{UP, shim}’.
The non-selective pTx UP used in the B₀ mapping sequence was designed using head datasets (B₁⁺, B₀ maps using CP pulses) of eight previously examined healthy volunteers. It was based on an extended SPINS k-space trajectory [4, 5] (target FA: 5°, duration: 1ms). The trajectory parameters and energy regularization weights were defined as ‘Combined Optimization Values’ (COV) for simultaneous optimization (see [5]). Universal pulse shapes were generated based on these COV, using the Variable Exchange algorithm [6].
For evaluation, in total four Fast Online Customized (FOCUS) pTx pulses were designed for excitation and inversion. Their design made use of the different brain shimmed B₀ maps, both B0_CP, shim and B0_UP, shim, as well as B₁⁺ maps of each transmit channel, which were acquired with a magnetization-prepared saturation-recovery TurboFLASH sequence (TA = 35s) [7] using CP pulses. The FOCUS excitation pulses were based on the UP used for B₀ mapping, yet with further optimized RF shapes. The inversion pulses were generated similarly, yet with a k_T point trajectory [8] (COV consisted of k_T point locations, their respective subpulse durations and an energy regularization weight), a pulse duration of 4.8ms and an interior-point based pulse shape optimization algorithm [9].

Results and Discussion
Figure 1 shows the first gradient echo of the B₀ mapping sequences with either CP mode or a UP and their respective simulated FA maps. While the CP pulse shows signal variations throughout the brain and low signal especially in the cerebellum region (NRMSE = 26.9%), the UP shows improved signal homogeneity (NRMSE = 14.2%).
Figure 2 shows an example slice of the maps B0_CP/B0_UP/B0_{CP, shim}/B0_{UP, shim} in off-resonance frequency units [Hz], as well as the differences between B0_{CP, shim} and B0_{UP, shim} with different pulses. Note that for patient specific brain shimming, the difference between B0_CP and B0_UP lead to different shim currents. The effect of the brain shim appears similar for both pulses. The highest offsets in the shimmed maps, as well as highest differences between them mainly occur at tissue interfaces. Motion correction was not performed.
Figure 3 shows FA simulations of FOCUS pulses for excitation and inversion pulses optimized with the shimmed B₀ maps. Differences between the excitation pulses caused by the differences in the B₀ maps are in a range of -0.5° to 0.5°. In the case of the inversion pulses, the one based on B0_{CP, shim} showed an up to 15° lower FA in the sinus region.
Figure 4 shows resulting MPRAGE images, when using FOCUS pulses based on either B0_{CP, shim} or B0_{UP, shim} for excitation and inversion. The FOCUS inversion pulse based on B0_{CP, shim} showed weaker performance in the frontal sinus region, and the corresponding MPRAGE image shows a B₀ related artifact in this region. This artifact is reduced, yet still present when using a FOCUS inversion pulse based on B0_{UP, shim}. Effects caused by the different excitation pulses could not be observed.
Notably, the improved signal strength in the cerebellum, achieved by the UP (Figure 1) did not result in significant differences in the performance of FOCUS pTx pulses in this particular region.

Conclusion
Acquiring B₀ inhomogeneities with a preoptimized pTx UP has minor impact on the behavior of pTx excitation pulses, yet it may be crucial for designing pTx inversion pulses. There B₀ artifacts can be reduced. Further investigations involving gold standard B₀ mapping sequences, as well as measuring B₁⁺ map inaccuracies and reducing them with pTx pulses, are still required. Additionally, the potential of reciprocally improving B₀/B₁⁺ mapping and pTx pulses remains to be explored.

Acknowledgements

No acknowledgement found.