Motion estimation and correction in cardiac CT angiography images using convolutional neural networks

Tanja Lossau, Hannes Nickisch, Tobias Wissel, Rolf-Dieter Bippus, Holger Schmitt, Michael M. Morlock, Michael Grass

Abstract: Highlights•Coronary Motion Forward Artifact model for CT data (CoMoFACT) generates motion perturbed image patches along with underlying motion vectors.•Supervised learning for Coronary Motion estimation by Patch Analysis in CT data (CoMPACT).•First iterative motion compensation pipeline based on CoMPACT.•Either global artifact reduction or fast local processing of few coronary centerline points.•Significantly reduced motion artifact levels, especially in image data with severe artifacts.