In hyperpolarized ¹³C MRI, quantification of metabolites or conversion rate constants is used as a marker for assessing metabolic activities in vivo^1,2. In these quantitative applications, high vasculature signal can interfere with surrounding tissues, leading to quantification error. To reduce the contamination coming from vessels, previous research used bipolar gradient to suppress the vasculature signal^3,4. However, the velocity of vessel can vary depending on the anesthetic level as well as pulsation. In addition, the delayed data acquisition by the insertion of bipolar gradient could cause severe signal loss in ultra-high field (9.4T) due to short ¹³C T₂* relaxation time (~20ms). In this study, simulations were performed to optimize the bipolar gradient with respect to velocity and T₂*, then the velocity-optimized bipolar gradient was implemented in conventional chemical shift imaging (CSI) for application in hyperpolarized ¹³C mouse liver experiment.

Simulation

Signal suppression of flowing spins by bipolar gradient was modeled by $$$\phi_{v}=\gamma m_{1} v(x)$$$ and $$$S_{flow}=S_{0}(1-\int_{voxel}^{}exp(-t/T_2^*)\times exp(-i\phi_{v})dv)$$$, where $$$\phi_{v}$$$ is a velocity-dependent phase accrual, $$$m_{1}$$$ is the 1^st gradient moment, $$$v(x)$$$ is velocity at position $$$x$$$ in voxel, and $$$S_{0}$$$ is the signal in the absence of bipolar gradient.³ The velocity within the voxel was assumed to follow a laminar distribution. Also, T₂* related signal loss of static spins was simulated by $$$S_{T_2^*}=S_{0}exp(-t/T_2^*)$$$. ¹³C T₂* of 21.2ms was used, which was obtained from separate scans of ¹³C CSI (data not shown). Since there is a trade-off between flow suppression and T₂* relaxation, an efficiency term for flow suppression was defined as $$$E_{fs}=(S_{flow}/S_{0})\times (S_{T_2^*}/S_{0})$$$. The simulation was performed with varying bipolar gradient pulse width (0.01~4.5ms) and velocity (0.01~4.5cm/s) (Fig.1b). In this simulation, fixed gradient amplitude of 352mT/m, corresponding to 80% of maximum gradient strength, was used to minimize bipolar gradient duration.

In vivo Experiments

All experiments were performed on a 9.4T small animal imaging system (Bruker BioSpin MRI GmbH, Germany) equipped with ¹H-¹³C dual-tuned mouse volume transmit/receive coil. [1-¹³C] pyruvic acid doped with 15mM Trityl radical and 1.5M Dotarem was polarized for 1h using HyperSense polarizer (Oxford Instruments, Oxford, UK). Samples were dissoluted with Tris/EDTA-NaOH solution, and approximately 350ul of pyruvate was injected into Balb/c mouse through tail vein catheter over duration of 5s. The bipolar gradient was inserted in alternate TRs in the direction of slice-selection after slice-selective excitation as shown in Fig.1a. Prior to flow suppression experiments, velocities of vessels flowing through the selected slice were measured to optimize the bipolar gradient using phase-contrast MRI (PC-MRI). For validation of flow suppression, ¹H fast low-angle shot (FLASH) with interleaved acquisition was performed (TR=250ms, TE₁/TE₂=5.3/13.3ms, FA=30, G_amp=88mT/m, δ=3.9ms). For hyperpolarized ¹³C experiments, flow-suppressed CSI in the presence (TR₁/TE₁=89/8.1ms) and absence (TR₂/TE₂=81/1.1ms) of bipolar gradient was performed on mouse liver. The slowest velocity of vessels was measured as 2.2cm/s. The optimal pulse width (δ) of bipolar gradient for the slowest velocity was determined to be 3.3ms (Fig.1c). Other scan parameters were set as follows: FoV=28×28mm², matrix size=16×16, FA=10°, spectral bandwidth=6510Hz, sampling point=512, NEX=2, G_amp=352mT/m. The scan was started at 30s after injection of hyperpolarized [1-¹³C] pyruvate.

Flow suppression using bipolar gradient was validated in ¹H imaging as shown in Fig.2. High signal intensity in the vasculature was observed in the absence of bipolar gradient (Fig.2a). With the implemented bipolar gradient, the signal from flowing spins was suppressed (Fig.2b). Peak SNR (PSNR) maps of each metabolite were obtained from the flow-suppressed CSI data (Fig.3). The signal of pyruvate and lactate in the vasculature was suppressed by 88%. In liver tissue, pyruvate and lactate PSNR was reduced by 34.7% and 29.4%, respectively, resulting in PSNR of 23.4±4.4 for pyruvate, and 14.8±3.7 for lactate. While the signal reduction of lactate signal was comparable to the simulation result as expected from the T₂* relaxation (27% reduction), increased reduction of pyruvate signal was observed. The large reduction of pyruvate was attributed to decreased contamination by flow suppression.Signal contamination from vasculature was reduced using bipolar gradient implemented for hyperpolarized ¹³C CSI in mouse liver. To minimize signal loss arising from T₂* and mitigate the variability in each experiments, the bipolar gradient was optimized with respect to velocity and T₂*. Although additional signal loss is introduced in stationary tissues by both T₂* relaxation and diffusion sensitization, diffusion related signal loss was ignored in this study because the b-value used for flow suppression was relatively low compared to diffusion weighted imaging (DWI)³. As a future work, this technique can also be applied to other organs which require accurate quantification of metabolism in hyperpolarized ¹³C studies.