Synopsis

Single-voxel ¹H MRS in body applications often suffers from respiratory and other motion induced phase and frequency shifts, which lead to incoherent averaging and hence to suboptimal results. Here we show the application of metabolite cycling (MC) for liver ¹H MRS on a 7T parallel transmit system, using 8 transmit-receive fractionated dipole antennas with 16 additional, integrated receive loops. The preserved water signal in MC-MRS allowed for robust phase and frequency correction of individual acquisitions before coil combination and averaging, which resulted in in-vivo liver spectra with an improved spectral resolution as compared with MRS with VAPOR water suppression.

INTRODUCTION

METHODS

All measurements were performed at a 7T whole body system (Philips, Best, the Netherlands) using eight parallel transmit channels, each connected to a transmit-receive fractionated dipole antenna⁵ (MR Coils BV, Drunen, the Netherlands) and 16 additional receive loops integrated with the antennas (2 per antenna). Data were acquired in a phantom containing a lipid emulsion and in the liver of a lean, healthy volunteer, who gave written informed consent according to local ethics regulations.

First, T₁ weighted images were acquired. Then B₁ shimming and B₀ shimming were performed for the region of interest using an in-house written MATLAB script (for transmit phase optimization) and the MRCode software (MR Code BV, Zaltbommel, the Netherlands), respectively. For both the phantom and the in-vivo measurement, the B₁ shimming resulted in a B₁⁺ of 18 μT in the region of interest. In the liver, a DIXON scan was performed after B₁ and B₀ shimming. STEAM spectra were acquired in a 20x20x30 mm³ voxel. Spectra were acquired with water suppression using VAPOR (150 Hz bandwidth for the phantom and 200 Hz bandwidth for the liver measurements) and without water suppression using MC. For the latter, the STEAM sequence was modified to include an adiabatic MC inversion pulse in the mixing period (MC pulse duration = 22.4 ms, MC pulse offsets = +100 Hz and -100 Hz for odd and even scans, respectively). Other scan parameters were as follows: BW=4000 Hz, data points=1024, TE=12/10 ms (phantom/liver), TM=38 ms, TR=2000/2500 ms (phantom/liver), N_avg=64/128 (phantom/liver). In the liver, also a MC-STEAM spectrum was recorded using a navigator for respiratory gating (N_avg=64). Spectra of the 24 channels (8 transmit-receive fractionated dipole antennas and 16 receive loops) were individually phase and frequency corrected and combined before averaging.

RESULTS AND DISCUSSION

Figure 1 compares VAPOR-STEAM and MC-STEAM spectra recorded on the phantom with a lipid emulsion. Using MC-STEAM, the subtraction of downfield and upfield inverted spectra led to a spectrum free of gradient modulation sidebands, with good water suppression. However, the olefinic (-CH=CH-) lipid signal, which is close to the water peak, could not be observed in the MC-STEAM spectrum. The signal intensity of the lipid methylene (-CH₂-) peak was comparable for VAPOR-STEAM and MC-STEAM, indicating that the inversion was complete.

Figure 2 shows the voxel positioning for the in-vivo liver measurements, as well as the results for the MC-STEAM acquisition. In Figure 2B the downfield and upfield inverted spectra (after individual phase and frequency correction, coil combination and averaging) are separately displayed, showing again an excellent efficiency of the MC inversion pulse. The sum of the downfield and upfield inverted spectra (Figure 2C) shows a clean water spectrum, whereas the difference of the downfield and upfield inverted spectra represents the metabolite spectrum.

Figure 3 compares VAPOR-STEAM and MC-STEAM spectra acquired in vivo in the liver from the voxel indicated in Figure 2. The spectral resolution for MC-STEAM (Figure 3B) was slightly better than for VAPOR-STEAM (Figure 3A). For the MC-STEAM data, the individual phase and frequency correction can be performed on the water signal, which is preserved in each scan. This gives more reliable results and can explain the improved linewidths. Adding navigator-based respiratory gating to MC-STEAM (Figure 3C) led to an even better spectral resolution.

CONCLUSION

Acknowledgements

References

[1] W. Dreher et al., Magn Reson Med 54, 190-195 (2005)

[2] I.-A. Giapitzakis et al., Magn Reson Med 79, 1841-1850 (2018)

[3] A. Hock et al., Magn Reson Med 69, 1253-1260 (2013)

[4] A. Fillmer et al., Scientific Reports 7, 16898 (2017)

[5] A.J.E. Raaijmakers et al., Magn Reson Med 75, 1366-1374 (2016)