Total: 1
Reconstructing high-quality images from undersampled dynamic MRI data is a challenging task and important for the success of this imaging modality. To remedy the naturally occurring artifacts due to measurement undersampling, one can incorporate a motion model into the reconstruction so that information can propagate across time frames. Current models for MRI imaging are using the optical flow equation. However, they are based on real-valued images. Here, we generalise the optical flow equation to complex-valued images and demonstrate, based on two real cardiac MRI datasets, that the new model is capable of improving image quality.