Ultra-high field strengths offer the benefit of higher signal to noise ratio as well as improved separation of metabolites in spectroscopy. This improvement is expected to enable easier observation and characterization of metabolites that are on the downfield side of the spectrum, as they are often more difficult to see and separate at lower field strengths^[1,2]. In particular, the downfield spectrum may provide information on metabolites that are not present or not so easily separated upfield^[3,4], and furthermore can provide information on exchanging species, which is of great interest for chemical exchange dependent saturation transfer experiments. In the current work, the metabolite-cycling technique^[5,6] is implemented at 9.4T in order to evaluate the downfield part of the human brain spectrum without water suppression, as well as to investigate any effects due to exchange after water suppression.

Three healthy volunteers were scanned using a 9.4T Siemens whole-body human MRI scanner and a four-channel transceiver array surface coil^[7]. A 20x20x20mm³ region was selected in a mix of grey and white matter at the back of the brain. A Siemens FASTERMAP B₀ shimming implementation was used prior to running metabolite-cycling scans. A 22ms asymmetric adiabatic metabolite cycling pulse was applied in the TM period of a STEAM sequence with TR/TE/TM=3000/8/45ms^[8]. For each of the volunteers, 256 averages were acquired for the non-water suppressed metabolite-cycling sequence, as well as 256 averages for the same sequence with a SODA water suppression sequence^[9] applied before STEAM, with an acquisition time of 12.8min for each scan. In one volunteer, only 128 averages were reconstructed as the other 128 showed motion artifacts. During the measurement, shots were stored individually; following acquisition, offline processing was performed: spectral alignment, scaling, and eddy current correction using the water reference signal; and averaging and coil combination.

Qualitative comparisons were also performed between the current study’s data and previous data acquired at 7T (Philips Achieva) using a conventional STEAM sequence with VAPOR water suppression for seven volunteers (256 averages each, quadrature transceiver surface coil (Rapid Biomedical) at TR/TE/TM = 4000/13/26ms^[10].

Non-water suppressed and water suppressed spectra from one volunteer are shown in Fig. 1. The averages over our initial three volunteers’ spectra are shown in Fig. 2, along with the difference spectrum visualizing the effect of exchange-related signal drop due to water suppression. Fig. 3 shows the averages of water-suppressed spectra from 3 volunteers at 9.4T and from 7 volunteers at 7T.

The spectra in Fig. 1 are representative of the data acquired at 9.4T for all individuals, including peak separation and resolution; the individual spectra show some increase in separation of metabolites compared to 3T^[1] and 7T (Fig. 3) and indicate differences between non-suppressed and water-suppressed data.

The averaged spectra in Fig. 2 show clear differences between water suppressed and non-water suppressed spectra due to chemical exchange of protons between the respective molecules and water. The respective difference spectrum indicates the most prominent exchanging peaks, which are similar to the exchanging peaks determined at 3T when comparing to Ref.[1]. There is some increased structure in the 8.2-8.5ppm region, both in the averaged spectra and the difference spectrum, and the peak at 5.8ppm is much stronger than at both lower field strengths (not expected to be fully accountable by differences in water suppression pulse bandwidths); furthermore, the two peaks at 6.0 and 6.1ppm appear stronger and better separated at 9.4T. Overall, the 9.4T spectra confirm the 3T findings on exchanging peaks, and improve slightly on metabolite separation.

In Fig. 3, major differences are highlighted between field strengths, in particular indicating peaks that are not as easily visible at 7T. Some of these differences may be due to the shorter SODA water suppression sequence used at 9.4T (130ms), compared to the VAPOR sequence (700ms). The longer water pre-suppression delay in VAPOR may well accentuate the exchange-related signal loss. However, the peak at approximately 6.1ppm does not appear to be greatly affected by exchange at 9.4T, but still is not visible at 7T. The other difference between the data is that the 9.4T data was collected in a voxel of mixed white and grey matter, compared to the 7T data which is from mainly grey matter.

The current work is a preliminary investigation into further elucidating the downfield part of the human brain spectrum and into measuring effects of exchange at ultra-high field strengths, and indicates that the higher field strength improves metabolite separation, allowing for better identification and quantification of exchanging peaks.