MRI suffers from intrinsic low sensitivity because it is related to very small energy transitions of the spin ensemble which is characterized by almost equal population of the spin energy levels. This yields almost vanishing spin polarization at ambient temperatures (see Fig. 1). Large spin pools like that of the abundant water in tissue are therefore the signal source in conventional imaging. However, the concept of the expanding field of molecular and metabolic imaging is based on progress in the understanding of many diseases on a molecular level and requires a significantly improved sensitivity to detect highly dilute compounds. Increasing the detectable magnetization through special preparation of the spin system prior to encoding and detection is therefore a continuing endeavour in the field of NMR and MRI. Hyperpolarization is a technique that increases the sensitivity of NMR and MRI by a special preparation of the system that shifts it far away from thermal equilibrium. Though being of transient nature, it is a powerful approach that has enabled many applications for molecular and cellular imaging. Drug development and therapeutic monitoring will also benefit significantly from specific contrast agents or metabolites with enhanced sensitivity since a biochemical or cell biological response to a certain treatment usually occurs much earlier than mesoscopic changes in morphology.¹ MRI will make important contributions because the noninvasiveness of this modality for repetitive studies of an organism is an important aspect in this context which encourages ongoing efforts for solving the sensitivity issue.²

Considerable effort has been undertaken to improve NMR and MRI methods to make them part of the molecular imaging toolbox and to complement other modalities which themselves exhibit serious limitations. As most studies rely on the detection of the NMR signal with radiofrequency (RF) coils through classical Faraday induction, a macroscopic magnetization of the dilute marker molecules is still necessary. The common approach is an ex situ manipulation of the magnetization beyond thermal equilibrium to allow the detection of dilute spins. Such systems with populations largely deviating from this equilibrium are called ‘hyperpolarized’ and can be generated for various molecules of biomedical interest. The advantage of NMR is that the origin of the detectable signal, i.e. the macroscopic magnetization, does not influence the pharmacokinetic and pharmacological behavior, regardless of the achieved spin polarization. In this context, hyperpolarization of several different nuclei has been achieved, including carbon, which has always been of great interest.

Efficient approaches rely on employing interactions on the atomic level to push the ensemble far beyond equilibrium. Other spin systems that are more easily polarized can be manipulated to function as a polarization precursor. Then, through various interactions, the polarization can be transferred onto the eventually detected nuclei. This can extend the range of achievable polarization from thermal polarization, P = 10^-5–10^-6, all the way to the theoretical limit of P = 1. Three main approaches are discussed with their advantages and limitations.³ Their basic principles of interactions between the precursor and the eventually detected nuclei are depicted in Fig. 2.

For Dynamic Nuclear Polarization (DNP), the thermal polarization of unpaired electrons is used as the initial source of magnetization. This is at any given field strength and temperature much higher than for ¹H protons or any other nuclei. DNP exploits the spin of unpaired electrons of free radicals to polarize NMR-active nuclei. If the electron couples with nearby nuclei through scalar or dipolar interactions, by irradiating at or around the electron Larmor frequency, the electron polarization can be transferred to the nuclei. The upper limit of the polarization of the nuclei is determined by the thermal polarization of the electrons.

Techniques based on chemical incorporation of para-hydrogen (PHIP and SABRE) rely on the fact that hydrogen exists in two different isomers, para- and orthohydrogen, that have remarkably different NMR properties. Para-hydrogen represents the antisymmetric energy eigenstate and has a total spin of zero, and, as such, cannot be detected directly in an NMR experiment. However, its spin order can be converted into detectable magnetization and it is easy to increase the population of para-hydrogen from (the high-temperature limit of) 25% to 90% just by cooling to < 50 K. To create a detectable NMR signal, the symmetry of the hydrogen molecule must be broken. This can be achieved through a hydrogenation chemical reaction. The two hydrogen atoms must bind to magnetically inequivalent sites on the target molecule, breaking their symmetry, whilst remaining coupled. Different methods are used to transfer the spin order generated by PHIP into longitudinal magnetization of, e.g., ¹³C of different tracers, including adiabatic field cycling and RF pulse sequences.

The intrinsic polarization of photons is used to generate hyperpolarized noble gases through Spin Exchange Optical Pumping (SEOP). It found early applications in lung imaging using ³He and ¹²⁹Xe. In particular, Xe is a promising candidate for many applications beyond gas phase imaging, including molecular sensing. The initial step for SEOP is the production of circularly polarized light, which is then used to pump a specific transition in an alkali metal vapor that serves as an effective one-electron system. Such electrons can later undergo spin–spin interactions with Xe nuclei when brought into close contact. This transfers the polarization onto the noble gas.

All of the techniques yield enhancement factors of up to 10⁵, hence they typically achieve polarization levels in the range of a few percent to almost 100%, depending on the application. Most initial studies originated in the context of physical chemistry. The increasing number of animal applications and clinical studies required adaptations of entire protocols to specific in vivo conditions. The basic configurations of clinical whole body scanners, in particular, and sometimes also animal scanners are not necessarily sufficient to perform such studies and make hardware adaptations necessary. Further challenges for the translation of initial proof of principle studies mainly include the production of pharmacologically safe hyperpolarized agents and the mastering of in vivo relaxation conditions. Fast encoding of the transiently available magnetization is crucial for making optimum use of the signal enhancement.

Based on recent progress, metabolic imaging with tracers such as [1-¹³C]pyruvate has made it already into clinical applications in oncology.⁴ Studies involving para-hydrogen are currently exploring different tracers for biomedical applications.⁵ Hyperpolarized Xe finds an increasing number of applications in lung disease diagnostics⁶ and in fundamental research to allow for ultra-sensitive imaging of cell surface markers at nanomolar concentrations.^7,8