VOLUMETRIC MESHFREE FRAMEWORK FOR JOINT SEGMENTATION AND MOTION TRACKING OF THE LEFT VENTRICLE

Meshfree particle method (MPM) exhibits improved flexibility and accuracy in dealing with problems with large deformation, complex geometry and material discontinuities. In this paper, we present a MPM based framework for the simultaneous shape recovery and motion tracking of the left ventricle. The myocardium is modeled as an anisotropic elastic body accounting for the fiber directions, represented by sampling nodes bounded by endocardial and epicardial boundaries. Cardiac dynamics is then driven by external force constructed individually for each node through integration of measures of image edgeness, image derived salient feature coherence, prior tissue distribution model, and temporal motion model. The displacement field throughout the cardiac cycle is obtained when the total energy of the elastic body is minimized to reach equilibrium state. Experiments with 3D canine MR images illustrate the benefits and potentials of such effort.