Research project

The acquisition of the data is a process, which can take several minutes and is thus prone to patient motion. The motion results into an acquisition of the patient with a different orientation. Due to this misalignment of the stepwise measurement and the acquisition in in Fourier domain, motion can result in artifacts which spread across the whole image. The image might not be usable and a costly reacquisition might be required. Approaches to compensating the patient motion often requires a change in the data acquisition, additional hardware or consists of a time consuming iterative post-processing. In this project artificial neural network based methods are developed, which, solely using the acquired data, allows a fast and reliable motion correction.