Synopsis

Dynamic nuclear polarization and dissolution offer the exciting possibility of imaging biochemical reactions in vivo, including some of the key enzymatic reactions involved in cellular metabolism. For MR metabolic imaging using 13C‐labeled compounds and DNP, the desired information lies in both the spectral domain, with the relative amplitudes of the different chemical shift species, as well as in the spatial domain. This necessitates some form of spectral encoding together with the acquisition of imaging data, which strongly influences the design of pulse sequences for this application. We will discuss how to efficiently use the limited available hyperpolarized magnetization in conjunction with available imaging pulse sequences.

Background

Dynamic nuclear polarization and dissolution offer the exciting possibility of imaging biochemical reactions in vivo, including some of the key enzymatic reactions involved in cellular metabolism. The signal amplification of the "hyperpolarized" state is typically 10,000-fold of the thermal equilibrium signal. However, this amplification is transient, meaning that there are certain design constraints which must be considered. For MR metabolic imaging using 13C‐labeled compounds and DNP, the desired information lies in both the spectral domain, with the relative amplitudes of the different chemical shift species, as well as in the spatial domain. This necessitates some form of spectral encoding together with the acquisition of imaging data, which strongly influences the design of pulse sequences for this application.

General principles and design constraints

The general principles and design constraints for this application are:

1. The short-lived nature of the hyperpolarized state, which decays exponentially to thermal equilibrium with the T1 relaxation time. For [1-13C]pyruvate, this is approximately 40 seconds in vivo, which limits scan times typically to minutes after injection of the substrate.
2. The consumption of the hyperpolarization by each acquisition and RF pulse application.
3. The consideration of optimal acquisition timing, as there is a delay from when the substrate is administered, subsequent tissue perfusion, and when appreciable labeling of the downstream metabolites occurs. For [1-13C]pyruvate, these metabolic products are typically lactate, alanine, and bicarbonate.

Data acquisition

This talk will discuss acquisition strategies for hyperpolarized substrates. We will discuss how to efficiently use the limited available hyperpolarized magnetization in conjunction with available imaging pulse sequences.

In spectroscopic acquisitions, the raw signal is viewed as a 4-dimensional volume (3 spatial dimensions plus time). MRSI relies on the periodic sampling of the same spatial k-space points at intervals corresponding to the spectral width, and subsequent Fourier transformation to obtain spectra at each spatial location. The most basic implementation is chemical shift imaging (CSI) where dense sampling of each spatial k-space position allows for reconstruction of a grid of spectra. More efficient k-space sampling can be achieved by using time-varying gradients during each readout. For example, echo-planar spectroscopic imaging (EPSI) can be used to encode an entire plane of k-space per excitation. As mentioned in design constraints, the hyperpolarized experiment typically lasts for a few minutes. In order to maximize the information gained from each experiment, it is possible to exploit prior knowledge about the chemical shift species to be investigated, as well as their spatial relationship and temporal dynamics. Popular approaches include variable flip angle schemes, multi-echo (“model-based”) acquisition, spectral-spatial excitation to image single metabolites, and constrained reconstructions based on parallel imaging and/or compressed sensing. Combinations of these methods can be used to design robust hyperpolarized imaging acquisitions.

Acknowledgements

References