3496
Towards Whole Brain Diffusion Tensor Spectroscopic Imaging1Neurology Department, University of New Mexico, Albuquerque, NM, United States, 2Nuclear Engineering Department, University of New Mexico, Albuquerque, NM, United States, 3Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, 4Physics and Astronomy Department, University of New Mexico, Albuquerque, NM, United States
Synopsis
Here we describe the development of a high-spatial resolution multi-slice single-shot Diffusion-Tensor proton-echo-planar-spectroscopic-imaging (PEPSI) technique using a binomial spatial-spectral refocusing RF pulse for mapping the diffusion tensor of water, Cho, Cr and NAA. Analysis of NAA and water diffusion in white matter obtained in a 4-slice acquisition with 1cc voxel size in 6 minutes (bmax = 2430.7 mm2/s) showed a mean ADC value of 0.15*10-3 and 0.66*10-3 mm2/s, which is consistent with values reported in the literature. Single-shot high spatial resolution diffusion sensitive MR spectroscopic imaging has the potential to probe metabolite diffusion across extended brain regions.
Introduction
Characterization of brain metabolite diffusion can provide intracellular information that is associated with aging, a diverse range of pathologies, such as brain tumors1,2, multiple sclerosis3 and infarcts4, and it may be applicable to developmental disorders, such as autism spectrum disorder. However, to date, most studies use single voxel methods. We have previously demonstrated the feasibility of Diffusion-Tensor Spectroscopic Imaging (DTSI) to map the age-dependence of Choline, Creatine and NAA diffusion in young children and adults5. However, phase-encoded DTSI techniques5,6 are very sensitive to head movement and currently limited to a single slice. Here we describe the development of a multi-slice high-spatial-resolution single-shot DTSI technique that has the potential for whole-brain mapping of the diffusion tensor of water, Cho, Cr and NAA.Methods
Single-shot DTSI was developed as a module of a general-purpose proton-echo-planar-spectroscopic-imaging (PEPSI) pulse sequence7 and implemented on a 3T Siemens Prisma scanner equipped with a 32-channel head array coil (Figure 1). Spatial-spectral encoding was performed using a series of double-sampled EPI readout modules with ramp sampling for separate processing of even and odd echo data as described in8. For single-slice DTSI a frequency selective refocusing RF pulse was used to suppress spectral signals outside of the spectral width of the echo-planar spatial-spectral encoding module. For multi-slice DTSI a binomial spatial-spectral refocusing RF pulse was used. Two different localization modules were implemented: (i) single-spin echo excitation with frequency selective refocusing in which the navigator-based phase correction5. is based on the metabolite signals within the selected frequency range and (ii) double spin echo excitation in which the second refocusing RF pulse is frequency selective, diffusion gradients are applied before and after the first refocusing RF pulse and the navigator-based phase correction5 uses the water signal from the first refocusing RF pulse to increase SNR for the navigator-based phase correction. Subcutaneous fat was suppressed using 8 outer volume suppression slices. Water suppression was not applied. Pulse sequence parameters were as follows: TR/TE: 3000/83ms in case of single spin echo and 3000/64ms in case of single-spin echo acquisition, spatial matrix: 32x32 or 16x16, voxel size: 7x7x20mm3 or 14x14x20mm3, spectral bandwidth: 40.5 or 95Hz, spectral points: 32 or 64, spectral resolution: 1.3 or 1.5Hz. Peripheral pulse gating was performed in-vivo. Spectra were reconstructed online with navigator-based phase correction as described previously7. Data were acquired with up to 6 gradient directions and maximum b-value of 2430.7 s/mm2. Post-processing of the data was performed using custom MATLAB tools and LCModel9. Linear interpolation to increase the spectral points from 32 to 64 was performed in the time domain and circular spectral shifting to center the metabolite peak(s) were performed prior to LCModel fitting using analytically modeled basis sets. Metabolite maps for each of the 7 diffusion gradients were masked to discard background voxels and input to MedINRIA 1.9.0 software to calculate ADC and FA maps.Results
Spatial ghosting in a phantom was negligible and uniform ADC and FA maps were obtained (Figure 2). Eddy current effects were minor. Mean Trace/3 ADC values of water and NAA in a phantom at 200C were consistent with previous single voxel studies, e.g.10 (Figure 3). FA values of water were close to zero and the FA of NAA was < 0.2 on average. The 32x32 matrix considerably improved the uniformity of NAA in vivo compared with the 16x16 matrix. Comparison of NAA signal intensities with different trigger delays showed maximum signal at a delay of 800 ms. Comparison of single-spin echo and double spin echo NAA mapping for b = 2430.7 mm2/s with diffusion encoding along the AP direction showed comparable localization performance, suggesting similar performance of navigator correction. Comparison of NAA signal intensities with different diffusion gradient directions showed maximum signal for diffusion encoding along the AP direction. Preliminary analysis of NAA and water diffusion in white matter obtained in a single-slice acquisition with 2cc voxel size in 6 minutes with diffusion encoding along the AP direction (bmax = 2430.7 mm2/s) showed a mean ADC value of 0.15+/-0.073*10-3 and 0.66+/-0.26*10-3 mm2/s, which is consistent with values reported in the literature. The four-slice acquisition using 1cc voxel size showed an NAA ADC of 0.22+/-0.12*10-3 mm2/s (Figures 4 and 5).Discussion
Single-shot DTSI substantially reduces motion sensitivity compared with phase encoded DTSI techniques. Decreasing voxel size reduces the sensitivity to rotational movement-related k-space signal shifts. The resulting decrease in spectral bandwidth introduces spectral aliasing of the peak of interest, which can be corrected by circular spectral shifting provided the spectral bandwidth of the frequency selective refocusing RF pulse is sufficiently large to capture the frequency shifts across the slice without aliasing other spectral peaks. However, slice-specific offsets of the spatial-spectral refocusing pulse still need to be implemented to increase volume coverage. The characterization of the full diffusion tensor of metabolites in vivo is in progress. This approach of spectrally selective DTSI is also suitable for concurrent DTSI and fMRI as described in our recent study7.Conclusion
Single-shot high spatial resolution diffusion sensitive MR spectroscopic imaging has the potential to probe metabolite diffusion across extended brain regions.Acknowledgements
Supported by 1R21CA241714. We gratefully acknowledge Alex Avram and Peter Basser for sharing MATLAB code to quantify b-values of diffusion pulse sequences.References
[1] M. Harada, M. Uno, F. Hong, S. Hisaoka, H. Nishitani, and T. Matsuda, "Diffusion-weighted in vivo localized proton MR spectroscopy of human cerebral ischemia and tumor," NMR Biomed, vol. 15, pp. 69-74, Feb 2002.
[2] J. M. Hakumaki, H. Poptani, A. M. Puumalainen, S. Loimas, L. A. Paljarvi, S. Yla-Herttuala, and R. A. Kauppinen, "Quantitative H-1 nuclear magnetic resonance diffusion spectroscopy of BT4C rat glioma during thymidine kinase-mediated gene therapy in vivo: Identification of apoptotic response," Cancer Research, vol. 58, pp. 3791-3799, Sep 1 1998.
[3] E. T. Wood, I. Ronen, A. Techawiboonwong, C. K. Jones, P. B. Barker, P. Calabresi, D. Harrison, and D. S. Reich, "Investigating axonal damage in multiple sclerosis by diffusion tensor spectroscopy," Journal of Neuroscience, vol. 32, pp. 6665-6669, May 9 2012, 3360480.
[4] W. Dreher, E. Busch, and D. Leibfritz, "Changes in apparent diffusion coefficients of metabolites in rat brain after middle cerebral artery occlusion measured by proton magnetic resonance spectroscopy," Magnetic Resonance in Medicine, vol. 45, pp. 383-389, Mar 2001.
[5] K. Fotso, S. R. Dager, A. Landow, E. Ackley, O. Myers, M. Dixon, D. Shaw, N. M. Corrigan, and S. Posse, "Diffusion tensor spectroscopic imaging of the human brain in children and adults," Magn Reson Med, vol. 78, pp. 1246-1256, Oct 2017, PMC5409876.
[6] A. E. Ercan, A. Techawiboonwong, M. J. Versluis, A. G. Webb, and I. Ronen, "Diffusion-weighted chemical shift imaging of human brain metabolites at 7T," Magn Reson Med, Jul 1 2014.
[7] S. Posse, B. Sa De La Rocque Guimaraes, T. Hutchins-Delgado, K. Vakamudi, K. Fotso Tagne, S. Moeller, and S. R. Dager, "On the acquisition of the water signal during water suppression: High-speed MR spectroscopic imaging with water referencing and concurrent functional MRI," NMR Biomed, p. e4261, Jan 30 2020.
[8] S. Posse, K. F. Tagne, and S. R. Dager, "Single-Shot Diffusion Tensor Spectroscopic Imaging in Human Brain," in Proc. International Society for Magnetic Resonance in Medicine (ISMRM) Toronto, Canada, 2015, p. 614.
[9] S. W. Provencher, "Estimation of Metabolite Concentrations from Localized in-Vivo Proton Nmr-Spectra," Magnetic Resonance in Medicine, vol. 30, pp. 672-679, Dec 1993.
Figures
