3040

Enabling SENSE Accelerated 2D CSI For Hyperpolarized Carbon-13 Imaging

Ayaka Shinozaki^1,2, Esben S. Hansen³, Juan D. Sanchez-Heredia⁴, Rolf F. Schulte⁵, Duy Anh Dang³, Markus P. Andersen³, Christoffer Laustsen³, Damian J. Tyler^1,2, and James T. Grist^1,2,6
¹Oxford Centre for Clinical Magnetic Resonance Research, University of Oxford, Oxford, United Kingdom, ²Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, United Kingdom, ³Department of Clinical Medicine, Aarhus University, Aarhus, Denmark, ⁴JD Coils, Hamubrg, Germany, ⁵GE HealthCare, Munich, Germany, ⁶Department of Radiology, Oxford University Hospitals, Oxford, United Kingdom

Synopsis

Keywords: Hyperpolarized MR (Non-Gas), Metabolism, carbon-13 imaging, x-nuclei MRI, metabolic imaging, flexible RF coil, CSI

Motivation: For hyperpolarized ¹³C metabolic imaging studies, a challenge is to achieve high temporal resolution without decreasing spatial and/or spectral resolution.

Goal(s): To accelerate hyperpolarized ¹³C MRI by combining a 2D Chemical Shift Imaging (CSI) sequence with SENSitivity Encoding (SENSE) reconstruction.

Approach: Due to the low natural abundance of ¹³C, the sensitivity maps needed for SENSE reconstruction cannot be pre-acquired. As such, in this work, the novel approach of using sodium sensitivity maps was demonstrated.

Results: SENSE reconstruction corrected aliased images, where in-vivo metabolic information was acquired with a 4-fold temporal acceleration.

Impact: As hyperpolarized ¹³C metabolic imaging is clinically translated, there is a need for easy-to-implement, fast, and robust imaging techniques. Therefore, this study implemented a novel ¹³C technique to accelerate Chemical Shift Imaging: a ubiquitous and robust sequence.

Introduction

Hyperpolarized (HP) MRI is a non-invasive clinical technique that detects low-signal metabolites through the injection of pre-polarized ¹³C-labelled substrates^1,2. However, a challenge in HP-¹³C studies is to achieve high temporal resolution while preserving spatial and spectral resolution³. Fast imaging approaches have been developed but all have specific trade-offs: spectral-spatial imaging provides good spatial resolution but offers limited spectral information^4,5; fast spectroscopic imaging offers high temporal resolution but limited spectral bandwidth and spatial resolution due to hardware limitations⁶; multi-echo approaches require well-defined spectral profiles^7,8; and phase-encoded Chemical Shift Imaging (CSI) provides a robust solution but is RF-intensive and slow⁹. Parallel imaging methods, such as SENSitivity Encoding (SENSE), offer acceleration opportunities, where fast acquisition results in a reduced field of view (FOV) and folding artifacts¹⁰. For artifact correction, SENSE requires a pre-acquired sensitivity profile, which is challenging for ¹³C with low natural abundance¹¹. Prior work with a dual-tuned sodium-carbon coil has successfully obtained sodium sensitivity maps representing the carbon profile¹². As such, in this work, a clinically robust and easy-to-use sequence, CSI, was accelerated by leveraging the previously presented sodium-carbon coil while preserving spatial and spectral quality.

Methods

Compliant with Danish law on animal experiments, four female Danish domestic pigs were imaged (figure 1) using a 3T GE MR750 MRI scanner (GE Healthcare, WI). Proton imaging was acquired with the body coil using a T2 fast spin echo sequence. Sodium and carbon imaging were acquired using a clamshell transmit coil (Rapid Biomedical, Germany) and an eight-channel flexible ²³Na/¹³C receive coil (JD Coils, Germany) (figure 1). Pigs received two injections of 30ml of 250mM hyperpolarized pyruvate, prepared in a DNP system (SpinAligner, Polarize) as previously described¹³, administered intravenously at ∼5ml/s and flushed with 20 ml saline. Imaging commenced 20 seconds after injection completion. Injections were separated by ~2 hours. A single slice sodium CSI was acquired (FOV=240mm, number of averages(NEX)=16, flip angle (FA)=90◦, Repetition Time (TR)=38.5ms, bandwidth=5kHz, matrix =24x24, slice thickness=10mm). Either a fully sampled or under sampled ¹³C CSI was acquired after each HP injection (FOV=240 or 120mm, NEX=1, FA=10◦, TR=64.1ms, bandwidth=5kHz, matrix =24x24 or 12x12, slice thickness=10mm, temporal resolution = 36 or 9s). Metabolite signals were extracted from the spectra using a deconvolution spectral fit¹⁴. Reconstructions used sodium sensitivity and carbon images with a 2D SENSE algorithm in MATLAB (R2023b, Mathworks).

Results

In figure 2A, the combination of eight channels aggregates the low signal per loop into a sensitivity image. Figures 2B.1-8 show the corresponding localised sensitivity on the porcine torso closest to each receiver element. Sensitivity drops rapidly away from the coil. Figure 3A shows the anatomy in the axial proton image. Pyruvate was expected to localise at the yellow kidney ROIs. Figure 3B is the pyruvate signal from the fully sampled CSI (SNR = 37 ± 19) showing two high intensity regions corresponding to the kidneys. Figure 3C demonstrates the folded aliasing due to reduced FOV in the under sampled CSI. Figure 3D is the SENSE reconstruction with two bright regions at the kidneys, illustrating the structural similarity between the fully sampled image and corrected aliasing (SNR = 23 ± 14). Figure 4 shows tracked metabolism dynamics (pyruvate, bicarbonate, lactate, and alanine) in the kidneys of a second porcine subject. Increased temporal resolution from the accelerated acquisition is seen in figures 4C/4D compared to that in figures 4A/4B. The imaging scheme in figure 4E illustrates the four-fold temporal acceleration, allowing four under sampled CSIs to be taken in a span of one fully sampled CSI. In figure 5, another subject’s fully sampled CSIs show bright spots in the pyruvate image. The under sampled images illustrate the aliasing due to the reduced FOV. SENSE reconstructed images for pyruvate, lactate, and alanine exhibited two clear bright spots at the kidneys.

Discussion

In vivo accelerated CSI using SENSE reconstruction demonstrated successful aliasing corrections with good SNR. Sodium sensitivity maps eliminate the need for co-registration and additional HP injections. High temporal resolution produced a detailed decay curve, useful for further metabolite kinetic modelling or denoising. In cases with unknown metabolite spectra or limited HP MRI specialist knowledge, this robust acceleration method ensures comprehensive capture of metabolic signals, mitigating the risk of missing spectral data.

Conclusion

The novel approach of using sodium sensitivity profiles with accelerated SENSE-reconstructed CSI demonstrated alias-corrected images, where metabolic information was acquired with a 4-fold temporal acceleration. Further work will look to optimise image post-processing, signal fitting, and k-space acquisitions for increased SNR (spiral, radial and rosette sampling). This robust acceleration technique offers a promising solution for metabolic imaging in diverse clinical scenarios.

Acknowledgements

This work was supported by the Oxford-MRC Doctoral Training Partnership iCASE award, and the Oxford Clarendon Scholarship.

References

1. Kurhanewicz, J. et al. Hyperpolarized 13C MRI: Path to Clinical Translation in Oncology. Neoplasia 21, 1–16 (2019).

2. Wolber, J. et al. Generating highly polarized nuclear spins in solution using dynamic nuclear polarization. Nucl Instrum Methods Phys Res A 526, 173–181 (2004).

3. Grist, J. T. et al. Quantifying normal human brain metabolism using hyperpolarized [1–13C]pyruvate and magnetic resonance imaging. Neuroimage 189, 171–179 (2019).

4. Meyer, C. H., Pauly, J. M., Macovskiand, A. & Nishimura, D. G. Simultaneous spatial and spectral selective excitation. Magn Reson Med 15, 287–304 (1990).

5. Gordon, J. W., Vigneron, D. B. & Larson, P. E. Z. Development of a symmetric echo planar imaging framework for clinical translation of rapid dynamic hyperpolarized 13 C imaging. Magn Reson Med 77, 826–832 (2017).

6. Gordon, J. W. et al. Fast Imaging for Hyperpolarized MR Metabolic Imaging HHS Public Access. J Magn Reson Imaging 53, 686–702 (2021).

7. Bieri, O. & Scheffler, K. Fundamentals of balanced steady state free precession MRI. Journal of Magnetic Resonance Imaging 38, 2–11 (2013).

8. Reeder, S. B. et al. Iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL): application with fast spin-echo imaging. Magn Reson Med 54, 636–644 (2005).

9. Wang, J. X., Merritt, M. E., Sherry, A. D. & Malloy, C. R. Accelerated Chemical Shift Imaging of Hyperpolarized 13C Metabolites. Magn Reson Med 76, 1033 (2016).

10. Pruessmann, K. P., Weiger, M., Scheidegger, M. B. & Boesiger, P. SENSE: Sensitivity Encoding for Fast MRI. doi:10.1002/(SICI)1522-2594(199911)42:5.

11. Arunachalam, A. et al. Accelerated spectroscopic imaging of hyperpolarized C-13 pyruvate using SENSE parallel imaging. NMR Biomed 22, 867–873 (2009).

12. Sanchez-Heredia, J. D. et al. RF Coil Design for Accurate Parallel Imaging on 13 C MRSI using 23 Na Sensitivity Profiles.

13. Laustsen, C. et al. Acute porcine renal metabolic effect of endogastric soft drink administration assessed with hyperpolarized [1-13c]pyruvate. Magn Reson Med 74, (2015).

14. Schroeder, M. A. et al. Measuring intracellular pH in the heart using hyperpolarized carbon dioxide and bicarbonate: a 13C and 31P magnetic resonance spectroscopy study. Cardiovasc Res 86, 82–91 (2010).

Figures

Figure 1 Left: Carbon clamshell transmit coil (Rapid Biomedical, Germany), adapted in this study to transmit at both the carbon and sodium frequencies but had difficulty achieving a 90° flip angle for ²³Na. Middle: An eight-channel array, flexible ²³Na/¹³C receive coil (JD Coils, Germany). Right: A porcine experiment set up with a supine pig inside the clamshell transmit coil and the torso wrapped with the flexible receive coil.

Figure 2A: Coil-combined sodium sensitivity image in the axial direction. The yellow line visualizes the coverage of the flexible receive coil from figure 1. 2B1-8: Coil-wise sodium sensitivity maps, produced by normalising the sodium images, smoothening the signal, and masking out data beyond the receive coil. Higher values (arbitrary unit) indicate higher receive signal and hence stronger sensitivity at that coil element.

Figure 3A: Axial T2 proton image. Kidneys are highlighted in yellow. 3B: Pyruvate signal from fully sampled CSI taken over 36 seconds (SNR = 37 ± 19). 3C: Pyruvate signal from under sampled CSI (2-fold acceleration in both x and y directions) taken over 9 seconds. 3D: Pyruvate SENSE Reconstructed CSI based on under sampled CSI and the sodium sensitivity (SNR = 23 ± 14). Images were normalised to the highest signal. SNR was the ratio of kidney ROI signal and the standard deviation of noise outside the torso of the fully sampled CSI.

Figure 4A and 4B: A dynamic ¹³C spectra of the fully sampled CSI in the kidneys of a pig showing the time course of pyruvate, lactate, alanine and bicarbonate. 4C and 4D: A dynamic ¹³C spectra of the under sampled CSI in the kidneys of a pig showing the time course of pyruvate, lactate, alanine and bicarbonate. Signals in the two kidney ROIs were summed after reconstruction, where data was normalised to the highest pyruvate value. 4E: The fully sampled and under sampled CSI scheme.

Figure 5: An example subject’s fully sampled CSI, accelerated CSI, and SENSE reconstructed ¹³C images demonstrating metabolite distribution. Signal was normalised per metabolite image. The proton image in the upper left corner highlights the kidney ROIs in yellow.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)

3040

DOI: https://doi.org/10.58530/2024/3040