Problem summary

CEST (Chemical Exchange Saturation Transfer) and MT (Magnetization Transfer) have a lot of clinical potential, but the techniques have to be applied with care to avoid misdiagnosis. In many cases, however, the limitations are of a practical nature, which can be overcome in order to make the technique specific, sensitive, and fast.

Background

The theory of CEST and MT will be described in the context of the Bloch/McConnell equations and simulations. Furthermore, the different competing mechanisms that frequently compete with CEST will be described: NOE, and transfer from macromolecules. The CEST mechanism is based on the continuous transfer of saturated spin populations to the water pool. Over time, this accumulated saturation leads to a significant saturation of the water signal, hence the potential enhancement of sensitivity – the molecule with the exchangeable site could often not be detected directly. Since irradiation times are relatively long, relaxation effects during the saturation have to be taken into account, and water T1 is an important parameter influencing the level of observable CEST contrast. MT is a similar process, but refers to the chemical exchange to water from macromolecules. This process is broad-band (as opposed to the narrow-band and frequency-selective appearance of CEST). NOE is a magnetization transfer occurring through space, and is typically found at a site opposite of the CEST sites (-NH2 and –OH). NOE can also play a role in a multi-step transfer in MT effects from macromolecules.

Implementation and Applications

Different MRI sequence implementations will be described, including pulse sequence variations, data treatment, image reconstruction, followed by a discussion of artifacts. Several pulse sequence variations aimed at distinguishing CEST from MT, and NOE will be discussed, as well as techniques for B0 and B1 inhomogeneity correction. The main challenge in CEST is often to separate the signal from direct water saturation. Traditionally, this problem was addressed by taking a difference between experiments performed with irradiation frequencies at opposite sides of the water resonance. This approach can often produce a lot of artifacts, especially for CEST sites close to the water resonance, and in the presence of B0 inhomogeneities. Alternatives exist, for example, based on Lorentzian lineshape fitting, or frequency encoding, which can alleviate this problem to some degree. With either method, a significant overhead is required to sample the response well close to the water resonance. In MT, the choice of the irradiation frequency is less limited, and B0 inhomogeneities are less of a concern as a consequence. Irradiation power, however, is important, and usually the more power, the stronger the contrast. B1 inhomogeneities affect both CEST and MT alike, but strategies based on fitting can help here as well. This aspect is the area where there is perhaps least development at the moment. Most CEST implementations use segmented GRE readouts with slice selection, and some 3D readouts have appeared recently. Several pulse sequences have been developed for separating CEST from NOE and MT, which can be useful in providing more specific contrast. The dependence of CEST on pH is of particular importance for techniques that use pH as a parameter, but also for the visibility of the contrast. The influence of pH and exchange catalysts on the exchange rates will also be discussed.

Summary

The aim of this presentation will be to give the MRI practitioner a good overview of the methods used in CEST and MT imaging, the current state of the art, and to outline the opportunities and limitations of the methods with respect to particular applications.

Acknowledgements

The author acknowledges a grant from the National Institutes of Health, NIBIB, R01 EB016045.