New methods and algorithms for the accurate, real-time motion analysis of the left ventricle with MRI-tagging

Abstract Despite its tremendous advantages, magnetic resonance tagging remains unserviceable in clinical conditions due to very tedious image pre-processing tasks. We present here a novel algorithm, fully automatic, real-time and extremely accurate, to perform the tagging pattern extraction. We then address the problem of motion computation and analysis using a novel kinetic deformation class. Very accurate registrations enable to perform automatic ventricle segmentation propagation and a highly sharp motion analysis of the walls. Our work mainly focuses on the CSPAMM protocol, which is briefly recalled. Experiments on acquisitions from 15 healthy volunteers are shown.