Accelerated dynamic Fourier velocity encoding by exploiting velocity-spatio-temporal correlations

To describe how the information content in a Fourier velocity encoding (FVE) scan can be transformed into a very sparse representation and to develop a method that exploits the compactness of the data to significantly accelerate the acquisition.

For validation, fully sampled FVE datasets were acquired in phantom and in vivo experiments. Fivefold and eightfold acceleration was simulated by using only one fifth or one eighth of the data for reconstruction in the proposed method based on the k-t BLAST framework. Reconstructed images were compared quantitatively to those from the fully sampled data.

Velocity spectra in the accelerated datasets were comparable to the spectra from fully sampled datasets. The detected peak velocities remained accurate even at eightfold acceleration, and the overall shape of the spectra was well preserved. Slight temporal smoothing was seen in the accelerated datasets.

A novel technique for accelerating time-resolved FVE scan is presented. It is possible to accelerate FVE to acquisition speeds comparable to a standard time-resolved phase-contrast scan.