INTRODUCTION

At high magnetic field (≥7T) (i.) B₁-inhomogeneity, (ii.) water and lipid signal-suppression and associated artifacts, (iii.) limitations on in-vivo specific absorption-rate (SAR), and (vi.) chemical-shift displacement artifact (CSDA) are four major factors that impose restrictions on application, interpretation, and quantification of the EPSI-data.¹ However, the use of adiabatic (refocusing) pulses allow minimization of the effect of B₁⁺-inhomogeneity ² and, additionally, the SAR and bandwidth of these type of RF-pulses can be independently controlled by selection of appropriate RF-pulse parameters, and these pulses have therefore a clear advance over amplitude-modulated RF-pulse schemes.³ Finally, CSDA can be minimized by avoiding the use of spatial selective RF-pulses. In order to tackle all 4 points mentioned at the same time, we here propose to replace the slice selective (adiabatic) refocusing pulse (pair) in the EPSI pulse sequence ⁴ by a spectral-selective adiabatic-complex-secant-hyperbolic 2π-pulse pair ⁵ in our adapted EPSI-pulse scheme.

METHODS

The changes we made in the EPSI-sequence.⁴ The slice-selective 180-degree refocusing pulse (Figure 1a) is replaced by two identical chemical shift-selective 180-degree secant hyperbolic adiabatic pulses $$$B_1^+(t) = Ω_0 \cdot sech(βt)^{1+\mu i}$$$ (Figure 1b). Specifically, the bandwidth (BW) of adiabatic pulses is set to 0.8 kHz (i.e. 2.7 ppm), and the carrier frequency was set at 3.0 ppm. The BW of the non-adiabatic excitation pulse is 5.5 kHz (18.5 ppm). In vitro tests were carried out in a spherical brain metabolite phantom (GE), and in vivo tests on one healthy volunteer and one patient (glioma). All measurements were performed on a 7T MAGNETOM Terra MR-scanner (Siemens, Germany) in clinical-mode using the 1Tx head coil.

RESULTS AND DISCUSSION

The original EPSI-implementation ⁴ used a slice-selective Mao refocusing-pulse which is 1.25 kHz (limited by maximum RF-amplitude). The CSDA of the Mao-refocusing is 23.7%. For the non-selective adiabatic pulses, the CSDA is only determined by the excitation-pulse and is 5.4%. Therefore, the overall CSDA-error is reduced by approximately 1 - 5.4/23.7 = 77%. Due to the total absence of in-plane CSDA in case of the proposed chemical shift selective adiabatic refocusing, the amplitude variation of the spectral patterns of metabolites over the total excited 3D-volume is much smaller.

Figure 2 shows the water signal integration maps using both EPSI-variants. Compared to using Mao-refocusing (Figure 2c), the water map of adiabatic pulses (Figure 2d) exciting the water is in good correspondence with the water-only excitation reference maps (Figure a-b): it indicates that despite the B₁-inhomogeneity of the 1Tx-RF-head coil, signal losses due RF-inhomogeneity are almost absent by using the chemical selective adiabatic 2π-pulses. The histograms of the water amplitude over the volume are shown in Figure 2a'-d', and indicate that these 2π-pulses lead to significantly more homogeneous refocusing. The reason for this is that above a certain minimal RF-amplitude these pulses become B₁⁺-insensitive and their flip-angle become nearly independent of the B₁⁺-amplitude. Since the proposed 2π-pulses have a relatively small BW, they operate properly at lower amplitude-threshold and this result in lower SAR than using spatial selective refocusing without the CSDA-drawback.

Three adiabatic 2π-pulses with different BWs were applied to investigate their performance on a spherical phantom (Figure 3). Figure 3c shows nearly perfect water suppression is achieved with water signal suppression factor of ≥1000 compared to Figure 3a. In addition, due to the symmetry of BW around 3 ppm, equivalent suppression performance is expected for the lipid region (0.9 to 1.3 ppm).

Figure 4 shows water and creatine (both 3.7 and 3.0 ppm) integration maps as well as water reference maps using the Mao (Figure 4a'-d') and adiabatic (Figure 4a-d) pulses respectively. See the figure legend for further details.

In vivo studies of healthy brain tissue in two subjects (Figure 5b-c) show a high degree of agreement with the in vitro measurement (Figure 5a). The spectrum within the tumour region (Figure 5d) shows a clearly different pattern than the normal tissue of the same subject (Figure 5c). Furthermore, it is noteworthy that only k-space regridding and the Fourier transformation were performed during pre-and post-processing, and no water removal and baseline correction was used. Therefore, there are still various possibilities to improve the final results by post-processing, such as B₀ and eddy current correction (ECC). The measurement time is about 8 minutes without parallel imaging techniques, which means that even a shorter scan time can be achieved by using GRAPPA or SENSE.

CONCLUSION

Single slab whole-brain EPSI allows the use of chemical selective small BW adiabatic 2π refocusing pulses, and is only possible to perform at high magnetic field strength (≥7T). It offers a way to tackle the B₁⁺ inhomogeneity problem, SAR limitation and the CSDA simultaneously. Additionally, the proposed pulse sequence has excellent homogeneous water (fat) suppression. It has also been demonstrated that the proposed adiabatic 2π-pulses guarantee a uniform refocusing over both the selected chemical range and spatial volume. In vitro and in vivo studies confirmed the expected performance and capability for potential application clinical routine.

Acknowledgements

This work was supported by the Swiss National Science Foundation grant number 182569. Additionally, this project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 813120. Initial development of the EPSI sequence was supported by NIH grant R01EB016064.