Hyper-CEST data contains a high degree of redundancy when sampling along the spectral dimension. Hence, the pixels in an imaging series show a very correlated behaviour. Although CEST preserves the spectral dimension for detecting different molecules, hyperpolarized nuclei come with a non-renewable magnetization that has to be used as efficient as possible. We therefore work on clever image reconstruction techniques to accelerate xenon MRI.
In particular, we exploit the redeundancies in the data to make use of sub-sampling approaches, thus allowing us to utilize the available magnetization in a much more efficient way. This significantly increases the signal-to-noise-ratio or reduces the acquisition time.