INTRODUCTION

Chemical exchange saturation transfer (CEST) enables an intrinsic MRI contrast of solutes (i.e. metabolites) by proton exchange with the abundant water pool. Due to the improved spectral resolution, as well as the higher SNR, and generally longer T1 at higher B0 field strengths, CEST effects in tissue are easier to detect, separate and associate with certain chemical shifts [1]. Application of a volumetric snapshot-GRE CEST MRI optimized for 9.4T [3] showed the known effects at +3.5 ppm, +2 ppm and -3.5 ppm, but also showed peaks at +2.7 ppm, +1.2 ppm and -1.7 ppm well-separated for the first time in the human brain in vivo.

METHODS

CEST MRI measurements were performed on a 9.4T whole-body MRI scanner (MAGNETOM, Siemens, Erlangen, Germany) using an in-house build 16Tx/31Rx-coil [2] and a 3D GRE [3] sequence in three healthy volunteers at nominal B1 = 1.2 µT, 0.9 µT, and 0.6 µT. Simultaneous B0 and B1 mapping was performed using the WASABI method [4]. B0 correction was performed pixel-wise using the WASABI B0 map. The 3-point-Z-correction as described by Windschuh et al. [5] was used for B1-inhomogeneity correction. CEST contrast images were calculated from B0- and B1-corrected Z-spectra using Lorentzian difference (LD) MTR_LD as proposed by Jones et al. [6]. MTR_LD was subsequently fitted by a 5-pool Multi-Lorentzian model [5].

RESULTS & DISCUSSION

Figure 1 a) and c) show the acquired highly-resolved Z-spectra at 9.4T for regions in gray and white matter at different effective B1 values. Scans of three different nominal B1 power levels and a B1 map (Figure 4) were used to create the B1-corrected Z-spectrum stack in the whole brain. Figure 1 b) and d) reveal that this correction leads to reproducible gray and white matter Z-spectra in three different volunteers. With validated feasibility and reproducibility more detailed analysis can be performed. By removal of the water and MT background by Lorentzian difference (LD) calculation, the individual CEST peaks can be investigated in the MTR_LD-spectrum in gray and white matter (Figure 2a and b). The +2.7 ppm and -1.7 ppm resonance could be detected reliably in 3 healthy volunteers. Figure 2c shows the averaged spectrum of the three volunteers, presented as mean ± standard deviation. Investigation of the according CEST contrast images (Figure 2d) consistently showed stronger gray matter contrast at positive irradiation frequency offsets and stronger white matter contrast at negative irradiation frequency offsets. To further separate the different resonances a 5 pool Multi-Lorentzian fit model was fitted to the MTR_LD stack. Figure 3 a) and b) shows the fit results in a typical gray and white matter voxel, figure c-e show the isolated resonances at 3.5 ppm, 2 ppm and -3.5 ppm as already detected at 7T [7], while f,g, and h show CEST contrast that were until now not yet reported in vivo. While the 2.7 ppm resonance shows higher signals in grey matter, the 1.2 ppm resonance is especially elevated in the frontal WM regions. The -1.7 ppm peak showed surprisingly strong signals in vessels which might be attribute to red-blood-cell CEST effects [8] and were investigated in detail in a separate abstract. In all contrasts (Figures 2,3), strong noise artifacts still appear in the region of lowest B1 as seen in Figure 4. This large variability in the B1 distribution across the field of view is a major challenge at ultra-high field, however current developments in parallel transmission techniques may allow mitigation of B1 inhomogeneities in the near future and make the presented approach more reliable and faster.

CONCLUSION

To our knowledge, the present study reports the most precise CEST measurement of the human brain in vivo. The high spectral resolution at 9.4T combined with a spectral sampling with 95 offsets revealed as yet unidentified CEST peaks robustly. In addition, maps of the individual peaks could be generated and showed different contrast in the brain, generating possibilities for future use as biomarker.

Acknowledgements

The financial support of the Max Planck and Society, German Research Foundation (DFG, grant ZA 814/2-1, support to M.Z., M.S., and K.H.), and European Union’s Horizon 2020 research and innovation programme (Grant Agreement No. 667510, support to M.Z., A.D.) is gratefully acknowledged.