Introduction

Dynamic Contrast-Enhanced MR (DCE-MR) perfusion assessment with gadolinium contrast agents is currently the most widely used cerebral perfusion MR method¹. The method relies on the acquisition of a dynamic series of T1-weighted images during administration of the contrast agent, and as such, the contrast is converted to an apparent concentration, which is then used to model the cerebral perfusion². However, increasing awareness on the deposition of gadolinium contrast in brain tissue supports the pursuit of novel perfusion measures such as arterial spin labelling (ASL). ASL is, however, inherently hampered by the low signal to noise ratio. In contrary, hyperpolarized MR is the signal source in itself and thus simplifies the quantification³. A limitation, however, is the need for 13C-molecular contrast agents, which limits spatial and temporal resolutions significantly. Recently, hyperpolarized water has been demonstrated in rodents^4,5 and in the porcine kidney⁶, showing superior spatial resolution compared to 13C angiographies. Here we investigated the potential for utilizing hyperpolarized water in the porcine brain, in conjunction with a model-free deconvolution, to map the cerebral perfusion⁷.

Methods

A healthy female Danish domestic pig weighing 31 kg was anaesthetized via continuous intravenous infusion of both Propofol and Fentanyl. Catheterizations were performed in the left femoral vein for blood sampling and the right and left femoral arteries for administration of hyperpolarized water, Percutaneous Intervention (PCI) -procedure and invasive blood pressure monitoring. 15 mL hyperpolarized water was injected over 5 s, initiated approximately 22 s after dissolution. A sample of 1 mL 30 mM TEMPO (2,2,6,6-Tetramethylpiperidine 1-oxyl, 98%) in H2O/glycerol 1:1 (w/w) was prepared. Dissolution medium consisted of D2O with 1 mM calcium disodium ethylenediaminetetraacetic acid (EDTA) and 9 g/L NaCl (all from Sigma Aldrich, Denmark). The sample vial was rapidly transferred from the liquid nitrogen bath to the 5 T magnet in the polarizer (SPINlab, GE Healthcare, Denmark). The sample was irradiated with app. 50 mW microwaves at 139.923 GHz for one hour. After dissolution, the radical was removed by extraction in 25 mL heptane (Sigma Aldrich, Denmark) in a separatory funnel. MRI imaging was performed on a 3 T GE HDx with an 8-channel cardiac array receiver coil (GE Healthcare, USA). Axial and coronal images were acquired to cover the brain to allow for planning of the angiography slab for hyperpolarized water imaging. The sequence was as follows: a 3D T1 weighted sequence with a standard SSFP (TE = 1.1 ms, TR = 2.7 ms, FA = 35°, matrix 256x256, FOV = 340x340 mm2, in-plane resolution of 1.3 mm and slice thickness of 3 mm). Angiographies were acquired in a coronal plane covering the brain using a gradient echo sequence: FA = 5°, slice thickness of 40 mm, TR = 3.4 ms, TE = 0.984 ms, matrix 256x256 and FOV = 140x140 mm². The acquisition time for each frame was 870 ms and 60 frames were acquired. Brain perfusion maps were calculated in OsiriX (Pixmeo, Geneva, Switzerland) using an open source OsiriX plug-in for T1-DCE-MRI perfusion analysis⁸. A pixel-by-pixel model-free, fast deconvolution was applied, assuming a direct signal enhancement of the hyperpolarized water, a hematocrit of 0.45 and a regularization of 0.15. Signal-to-Noise-Ratio (SNR) was measure as peak signal versus baseline noise in the input ROI.

Results

Injection of hyperpolarized water into the brain allowed for visualization of both large and small vessel as well as for acquisition of perfusion values from brain tissue. Figure 1 shows the acquired frames and presents the amount of details from vasculature we have obtained with this setup. The calculated perfusion and MTT maps are shown in Figure 2. The results from the perfusion ROI analysis are (ROI mean value): Anterior, 255 (mL/100mL/min); Posterior, 220 (mL/100mL/min); Circle of Willis, 1895 (mL/100mL/min); Input, 2645 (mL/100mL/min). The results from the MTT ROI analysis are (ROI mean value): Anterior, 1.09 (s); Posterior, 1.77 (s); Circle of Willis, 0.80 (s); Input, 1.00 (s). Figure 3 shows the dynamic evolution of hyperpolarized water entering the four ROIs and the difference between peak signal from input to anterior and posterior part of the brain is clearly observed. SNR was measured to be 273.

Conclusion

In this study, we demonstrate the feasibility of hyperpolarized water for visualizing the brain vasculature of a large animal in a clinically relevant setting. Furthermore, perfusion values within normal physiological range are obtained from the dynamic acquisition. We believe this to be a promising case for further sequence and method optimization. We expect that the polarization of the hyperpolarized water can be increased by a factor 5-10 relative to the current study.

Acknowledgements

Funded by The Danish Diabetes Academy supported by the Novo Nordisk Foundation.