Synopsis

An optimized GRASE sequence was developed for dynamic hyperpolarized ¹³C Imaging and demonstrated the feasibility for human brain research. The multi-echo signal can either be used to encode a 3D volume or as a 2D multi-echo approach to increase SNR or estimate T2. Varied crusher gradients were investigated to reduce the loss of longitudinal hyperpolarized ¹³C magnetization.

Introduction

Methods

The proposed GRASE sequence (Fig. 1a) was implemented on the RTHawk system (HeartVista, CA). A spectral spatial RF pulse⁶ was used for excitation, followed by a train of spectrally selective refocusing pulse (TBW 4.5, pulse duration 14ms) and EPI readouts. The refocusing pulse is designed to minimally perturb pyruvate signal while refocusing lactate (Fig. 1b). A pair of crusher gradients was placed before and after each refocusing pulse. Fig. 1c shows a varied crusher scheme⁷, which has been used to eliminate the stimulated echo path in the proton FSE sequence.

Studies were performed on a clinical GE 3T scanner. To compare constant crushers with varied crushers, phantom experiments were performed with an enriched [¹³C] NaHCO₃ phantom (T2 ~1s, T1 ~20s). For each crusher scheme, 2D acquisitions were performed using the proposed GRASE (1s TR, 12 echoes) for four time points with 30° flip angle. The same experiment was repeated at a relative RF power of 90%, 100% and 110%, respectively.

Two hyperpolarized ¹³C experiments were performed on a tumor-bearing mouse with the injection of [1-¹³C]pyruvate pre-polarized in a HyperSense DNP system. Each time point has one 3D stack-of-EPI gradient-echo pyruvate acquisition and one single-shot 3D lactate acquisition using the proposed GRASE (12 echoes). Constant and varied crushers were used in the first and second experiment, respectively. At the end of the second experiment, lactate signals were acquired without z phase encoding to measure signal phase, which was used for phase correction in 3D reconstruction. Scan parameters are: 3s temporal resolution, 14x14x12 matrix size, 24mm excitation slab, 6° and 30° flip angle for pyruvate and lactate, respectively, 3x3x4mm and 4x4x4mm resolution for the first and second experiment, respectively.

A human brain study was performed with an injection of hyperpolarized [1-¹³C]pyruvate prepared using a SPINlab polarized and methods described in a prior study8. Each time point has one 2D gradient-echo pyruvate acquisition and one multiecho lactate acquisition using the proposed GRASE (12 echoes) with varied crushers. Scan parameters are: 1.5x1.5x2cm resolution, 24x24cm FOV, 3s temporal resolution, 20s° and 30° flip angle for pyruvate and lactate, respectively.

All experiments were triggered at bolus peak within the tissue of interest for real-time frequency and B1 calibration⁸.

Results & Discussion

Results of phantom studies are shown in Fig. 2. Faster signal decay is observed in experiments using constant crushers (Fig. 2a) compared to using varied crushers (Fig. 2b), indicating that eliminating stimulated echo path with varied crushers could reduce the loss of the longitudinal magnetization. When RF power changes, varied crushers (Fig. 2f) showed more stable phase over echoes than constant crushers (Fig. 2e). The robustness of phase over echoes to RF power change is necessary for applying phase encoding over echoes when imaging in B1 inhomogeneous regions.

Results of animal studies using constant or varied crushers are shown in Fig. 3. 3D lactate images of both experiments matches with the anatomical details and the 24mm excitation slab which corresponded to 6 slices. As expected, similar decay rates are found between gradient-echo pyruvate signals of two experiments. Faster decay rate of lactate signals was found in the experiment using a constant crusher gradient, which agrees with the phantom results discussed above.

Fig. 4 shows the dynamic ¹³C images of pyruvate, lactate and bicarbonate in the human study. Combining echo images shows an improved SNR compared to a single-echo image. Fig 5. shows lactate and bicarbonate T2 results of the human study. Bicarbonate T2 values are predominantly about 500ms, while lactate T2 is substantially longer, with half of lactate T2 values greater than 1000ms. The fitting examples in Fig. 5 demonstrate the goodness of T2 fitting. Although the duration of the spin-echo train (440ms) of 12 echoes is relatively short compared to the fitted T2 values (500-1500ms), our results still provide a preliminary estimate of ¹³C lactate and bicarbonate T2 values in the human brain.

Conclusion

Acknowledgements

References

1. Golman K, in’t Zandt R, Thaning M. Real-time metabolic imaging. Proc Natl Acad Sci U S A 2006; 103:11270–5. 


2. Yena Y, Roux P, Mayerm D, et al. T2 relaxation times of 13C metabolites in a rat hepatocellular carcinoma model measured in vivo using 13C-MRS of hyperpolarized [1-13C]pyruvate. NMR Biomed. 2010; 23: 414–423

3. Reed GD, von Morze C, Verkman A, et al. Imaging Renal Urea Handling in Rats at Millimeter Resolution using Hyperpolarized Magnetic Resonance Relaxometry. Tomography. 2016; 2(2):125-135. 4. Milshteyn E, von Morze C, Reed GD, et al. Development of high resolution 3D hyperpolarized carbon-13 MR molecular imaging techniques. Magn Reson Imaging. 2017; 38:152-162.

5. Wang J, Wright A, Hu D, et al. Single Shot Three-Dimensional Pulse Sequence for Hyperpolarized 13C MRI. Magn Reson Med. 2017; 77:740–752 .

6. Gordon JW, Hansen RB, Shin PJ, Feng Y, Vigneron DB, Larson PEZ. 3D hyperpolarized C-13 EPI with calibrationless parallel imaging. J Magn Reson 2018;289:92–99.

7. Poon CS, Henkelman RM. Practical T2 Quantitation for Clinical Applications. J Magn Reson 1992; 2:541-553.

8. Tang S, Milshteyn E, Reed GD, Gordon JW, Bok RA, Zhu X, Zhu Z, Vigneron DB, Larson PEZ. A regional bolus tracking and real-time b1 calibration method for hyper- polarized 13c mri. Magn Reson Med 2018; p. In Press.