Introduction

Metabolic imaging via MR spectroscopic imaging (MRSI) is extremely valuable in clinical and preclinical applications^1,2. Its routine application remains challenging due to long acquisition times, low signal-to-noise ratio (SNR), development of in-house acquisition sequences and processing pipelines. Enormous advancements have been made in clinical applications regarding the spatial coverage, spatial resolution, acquisition speed, artifact suppression, number of detectable metabolites². In this context, FID-MRSI minimizes the T₂ relaxation and eliminates J-evolution, increasing SNR in a considerably short acquisition time. It also reduces chemical shift displacement errors and sensitivity to B₀ inhomogeneity.
To the best of our knowledge, at the present time, traditional phase encoded MRSI using STEAM, PRESS or SPECIAL excitation schemes (TEs between 2-10ms), is still used for quantitative metabolite mapping for ¹H preclinical applications^3–6. Recently, ¹H-FID-MRSI has been implemented on preclinical scanners providing robust, reproducible and high-quality metabolic maps with significantly reduced technical limitations and time^7,8.
The aim of the present study was to compare side by side PRESS-MRSI with FID-MRSI in preclinical settings at 14.1T in the rat brain.

Methods

All data were acquired in the rat brain at 14.1T (Bruker/Magnex Scientific) using a homemade transmit/receive quadrature surface coil. T₂-weighted Turbo-RARE images (slices=60, thk=0.2mm, TR=3000ms, NA=10, RAREfactor=6, 128x128) were acquired in coronal direction to position the MRSI slice. MAPSHIM was used, first in an ellipsoid covering the full brain, then in a voxel of 10x10x2mm³ centered on the MRSI slice (water linewidths=25-29Hz). Single slice ¹H-FID-MRSI (TR=811ms, 52° flip angle, acquisition delay=1.3ms, acquisition time of 13min) and PRESS-MRSI (TR=2000ms, 90/180/180°, TE=10.2ms, acquisition time of 32min, TR was fixed for reasonable acquisition time) measurements were performed in identical conditions on 3 rats (2mm slice centered on the hippocampus, FOV=24x24mm², matrix=31x31, nominal voxel size=0.77x0.77x2mm³, 1024 spectral points, cartesian k-space sampling). VAPOR water suppression⁹ and saturation slabs or outer volume suppression (OVS) were used to minimize the lipid contamination in FID-MRSI and PRESS-MRSI, respectively.
All data were processed with the MRS4Brain toolbox (in-house developed end-to-end processing pipeline with a graphical user interface for Bruker ¹H-FID-MRSI⁷). Identical processing steps were used for both sequences, except for the retrospective skull lipids removal using the metabolite-lipid spectral orthogonality approach¹⁰ which was not used for PRESS-MRSI due to OVS presence.
LCModel was used combined with acquisition specific metabolites basis-sets (simulated using NMRScope-B/jMRUI¹¹) and in vivo acquired macromolecules (double inversion-recovery) under identical in vivo conditions as metabolites. tCr was used as a reference for both sequences. Semi-automatic quality control steps as displayed in Figure 1 were used. The atlas-based segmentation from MRS4Brain toolbox was used for voxel selection in two brain regions using coronal MRI acquisitions: hippocampus and a mix of striatum and cortex.

Results and Discussion

Figure 1 displays the quality control steps as provided by MRS4Brain toolbox. As can be seen on the water power maps, for the same shimmed volume (10x10x2mm³) FID-MRSI provided an increased brain coverage. The linewidth and ∆B₀ maps showed similar variations, while the SNR per unit time map highlighted ~40% increase for FID-MRSI acquisitions. Overall the quality of the spectra in the middle of the excited VOI was similar for both sequences with an increased SNR for FID-MRSI as shown in Figure 2, in line with the increased SNR per unit of time.
Figure 3 highlights the metabolite maps of three representative metabolites out of eight quantified. FID-MRSI maps showed an increased brain coverage even after the application of quality control criteria (min. SNR, max. FWHM, max. CRLBs limits). Some differences in metabolite concentrations were noted between the two sequences (Figures 3&4&5) which might be related to the lack of metabolite T₁ and T₂ corrections, macromolecules or fitting issues, something requiring further investigations together with the retrospective efficiency of lipid suppression. Despite these differences Figures 4 and 5 show good agreement in metabolite concentration estimates when averaging over 3 rats in striatum and hippocampus, respectively. For PRESS-MRSI a decreased SNR towards the border of the maps was observed leading to the overestimation of some metabolites (i.e. NAA+NAAG).

Conclusion

Both sequences provided robust, reproducible and high-quality metabolic mapping of brain metabolites. Furthermore, our results highlighted an increased brain coverage combined with a significant reduction in the acquisition time when using ¹H-FID-MRSI. Some differences in metabolite concentration were also observed between the two sequences, requiring further investigations.

References

(1) Maudsley, A. A.; Andronesi, O. C.; Barker, P. B.; Bizzi, A.; Bogner, W.; Henning, A.; Nelson, S. J.; Posse, S.; Shungu, D. C.; Soher, B. J. Advanced Magnetic Resonance Spectroscopic Neuroimaging: Experts’ Consensus Recommendations. NMR in Biomedicine 2021, 34 (5). https://doi.org/10.1002/nbm.4309.

(2) Bogner, W.; Otazo, R.; Henning, A. Accelerated MR Spectroscopic Imaging—a Review of Current and Emerging Techniques. NMR in Biomedicine 2021, 34 (5). https://doi.org/10.1002/nbm.4314.

(3) Geppert, C.; Dreher, W.; Leibfritz, D. PRESS-Based Proton Single-Voxel Spectroscopy and Spectroscopic Imaging with Very Short Echo Times Using Asymmetric RF Pulses. MAGMA 2003, 16 (3), 144–148. https://doi.org/10.1007/s10334-003-0016-6.

(4) Delgado-Goñi, T.; Ortega-Martorell, S.; Ciezka, M.; Olier, I.; Candiota, A. P.; Julià-Sapé, M.; Fernández, F.; Pumarola, M.; Lisboa, P. J.; Arús, C. MRSI-Based Molecular Imaging of Therapy Response to Temozolomide in Preclinical Glioblastoma Using Source Analysis. NMR in Biomedicine 2016, 29 (6), 732–743. https://doi.org/10.1002/nbm.3521.

(5) Cudalbu, C. In Vivo Studies of Brain Metabolism in Animal Models of Hepatic Encephalopathy Using 1H Magnetic Resonance Spectroscopy. Metabolic Brain Disease 2013, 28 (2), 167–174. https://doi.org/10.1007/s11011-012-9368-9.

(6) Mlynarik, V.; Gambarota, G.; Kohler, I.; Gruetter, R. Short-TE Proton Spectroscopic Imaging of the Neurochemical Profile in the Rat Brain at 1 ?L Resolution. Proc. ISMRM 2007, 15, 1247.

(7) Simicic, D.; Alves, B.; Mosso, J.; Briand, G.; Lê, T. P.; van Heeswijk, R. B.; Starčuková, J.; Lanz, B.; Klauser, A.; Strasser, B.; Bogner, W.; Cudalbu, C. Fast High-Resolution Metabolite Mapping in the Rat Brain Using 1H-FID-MRSI at 14.1T. arXiv July 6, 2023. https://doi.org/10.48550/arXiv.2307.03261.

(8) Alves, B.; Simicic, D.; Mosso, J.; Lê, T. P.; Briand, G.; Bogner, W.; Lanz, B.; Strasser, B.; Klauser, A.; Cudalbu, C. Noise-Reduction Techniques for 1H-FID-MRSI at 14.1T: Monte-Carlo Validation & in Vivo Application. arXiv September 20, 2023. https://doi.org/10.48550/arXiv.2309.11204.

(9) Tkac, I.; Starcuk, Z.; Choi, I. Y.; Gruetter, R. In Vivo H-1 NMR Spectroscopy of Rat Brain at 1 Ms Echo Time. Magn Reson Med 1999, 41 (4), 649–656. https://doi.org/Doi 10.1002/(Sici)1522-2594(199904)41:4<649::Aid-Mrm2>3.0.Co;2-G.

(10) Klauser, A.; Courvoisier, S.; Kasten, J.; Kocher, M.; Guerquin-Kern, M.; Ville, D. V. D.; Lazeyras, F. Fast High-Resolution Brain Metabolite Mapping on a Clinical 3T MRI by Accelerated 1H-FID-MRSI and Low-Rank Constrained Reconstruction. Magnetic Resonance in Medicine 2019, 81 (5), 2841–2857. https://doi.org/10.1002/mrm.27623.

(11) Starčuk, Z.; Starčuková, J. Quantum-Mechanical Simulations for in Vivo MR Spectroscopy: Principles and Possibilities Demonstrated with the Program NMRScopeB. Analytical Biochemistry 2017, 529, 79–97. https://doi.org/10.1016/j.ab.2016.10.007.