MEMoRIAL-M1.7 | Model-based MRI reconstruction

The acquisition of MR images might run considerably slow due to the one-dimensional character of the signal and the need to consecutively measure many data points for a single image. Classically, an image cannot be uniquely reconstructed if the number of measured data points deceeds the number of points in the image.

In this project, prior knowledge derived from other sources than the MR acquisition itself will be used to uniquely reconstruct MR images from less-than-complete measurement data, particularly aiming at faster acquisition in moving organs. Therefore, (prior) knowledge such as information on the position of interventional instruments or the subject's breathing motion (deforming abdominal organs whereas not entirely changing the object itself) will be exploited and incorporated into mathematical models - the latter describing these objects and in turn being parameterised based on measurement data.