1Pattern Recognition Lab, Friedrich-Alexander-University Erlangen-Nrnberg, Erlangen, Germany; 2Erlangen Graduate School in Advanced Optical Technologies (SAOT), Friedrich-Alexander-Universitt Erlangen-Nrnberg, Erlangen, Germany; 3Pattern Recognition Lab, Department of Computer Science, Friedrich-Alexander-Universitt Erlangen-Nrnberg, Erlangen, Germany; 4Healthcare Sector, Siemens AG, Erlangen, Germany
In the current work, a method to compensate for respiratory motion by incorporating a motion model into a compressed sensing reconstruction is presented. Therefore, the input data were binned into respiratory phases first and, then, all bins with sufficient data were reconstructed iteratively using a weighted CS reconstruction. Based on these sub-volumes, a motion model was determined using non-rigid inverse-consistent registration. Finally, the resulting deformation fields were used in the reconstruction of a combined dataset without motion artifacts. Compared to the original weighted CS reconstruction the images feature more signal and a sharper delineation of the coronary arteries.