Introduction

Hyperpolarized [1-¹³C]Pyruvate (Pyr) has shown considerable promise for measuring in-vivo glucose metabolism with the resulting [1-¹³C]Lactate (Lac) signal reflecting glycolysis and that from ¹³C-bicarbonate (Bic) reflecting Pyr conversion to Acetyl-coA. In contrast, hyperpolarized [2-¹³C]Pyr offers the potential to measure both lactate and TCA cycle intermediates. While the two most popular dynamic imaging methods for hyperpolarized [1-¹³C]Pyr are fast spectroscopic imaging¹ and interleaved metabolite-selective imaging², unfortunately, neither of these approaches is directly applicable for [2-¹³C]Pyr studies. The large chemical shift of [2-¹³C]Lac (almost 150 ppm relative to [2-¹³C]Pyr) results in a large chemical shift spatial misregistration error in slice-selective spectroscopic imaging acquisitions³ (Fig. 1). While this misregistration error can be eliminated by sequential selective imaging of [2-¹³C]Lac, the density of spectral peaks from pyruvate, glutamate, citrate, acetoacetate, and acetyl-carnitine are too close together to allow efficient selective excitation. Here we propose a hybrid approach for which crowded spectral peaks around 170-210 ppm are measured using spectroscopic imaging, while the highly shifted [2-¹³C]Lac peak is measured using selective excitation with a conventional fast imaging readout.

Methods

Interleaving a spectroscopic readout for the peaks around 170- 210 ppm with an imaging readout for selectively excited [2-¹³C]Lac is relatively straightforward. The major complication is that the [2-¹³C]Lac peak is a doublet with a 140 Hz J-coupling constant, which results in an unacceptable k-space weighting during readout. For example, as shown in Fig. 2 and Fig.3, there will be a cosine weighting primarily in the phase encode direction(k_y) due to the C₂-H J-coupling during a bidirectional echo planar imaging(EPI) acquisition. In general, this k-space weighting can be modeled as: I1(k_x,k_y) = I(k_x,k_y) cos(πJt) . Simlarly, a sin(πJt) weighted image can be generated using a readout shifted by (1/2J). The complex combination of these two “Quadrature” components results in: I’(k_x,k_y) = I(k_x,k_y) cos(πJt) + i.I(k_x,k_y) cos(πJ(t+1/2J)) = I(k_x,k_y) { cos(πJt) + i.sin(πJt) } = I(k_x,k_y) e^i(πJt). Hence, combining two acquisitions with 1/2J shifted readouts (1/2J = 3.6 ms for [2-¹³C]Lac), an artifact-free, but spatially shifted, image can be reconstructed, with the spatial shift then easily removed during final image reconstruction. Simulations were performed in MATLAB (MathWorks Inc. MA, USA) by tracing an EPI k-space trajectory of a [2-¹³C]Lac Shepp-Logan phantom with parameters suitable for hyperpolarized pyruvate imaging in human brain (FOV=32cm, 32x32 imaging matrix, 10mmx10mmx20mm voxel size,) (Fig. 4). A Lac image, via quadrature combination of acquisitions from two echo times, TE₁=10.6ms and TE₂=14.2ms, was then computed to recover a shifted version of the original image. A correction for this spatial shift, computed as (Treadout*J_CH/2), was then applied as the final step in image reconstruction.

Results

The cosine modulation of k-space due to J-coupling creates severe ghosting and blurring artifacts for metabolite-selective [2-¹³C]Lac imaging. These artifacts were effectively removed by quadrature combination of TE₁=10.6 ms and TE₂=14.2 ms acquisitions. The reconstruction from simulated single-shot EPI data sets collected at TE₁=10.6ms and TE₂=14.2ms of a [2-¹³C]Lactate Shepp-Logan phantom are presented in Fig. 4a & 4b respectively. Comparison of the original phantom (Fig. 4c) and the result of quadrature reconstruction (Fig. 4d) highlights the effectiveness of this new technique in recovering the original lactate image in the presence of J-modulation, even in the case of long readout times (Fig. 5).

Conclusion

In this work we show that by acquiring EPI images at two different echo times separated by 1/2J, J-coupling artifacts can be eliminated using a quadrature detection reconstruction technique. With this method, a hybrid imaging sequence that incorporates spectroscopic imaging readout for pyruvate, glutamate, citrate, acetoacetate, and acetyl-carnitine and spectrally selective fast EPI (or other fast imaging sequence such as spirals) for [2-¹³C]Lac can be developed for robust dynamic imaging of hyperpolarized [2-¹³C]Pyr and downstream products.

Acknowledgements

The Lucas Foundation, GE Healthcare, and NIH grants R01 CA17683603, R01 EB01901802, R01 MH110683, and P41 EB01589121.