Simultaneous recovery of left ventricular shape and motion using meshfree particle framework

We have shown that meshfree particle methods possess better flexibility, efficiency, and accuracy in solving domain evolution problems in medical image analysis with complex geometries and discontinuities. In this paper, we present a unified cardiac image analysis framework which simultaneously recovers the shape and motion of the left ventricle, including the endo-, epi-, and mid-wall myocardium. Both segmentation and volumetric motion/deformation results are achieved at the same time through analyzing the spatiotemporal evolution of a set of sampling myocardial nodes bounded by endo- and epicardial boundaries. An meshfree Galerkin formulation provides the representation and computation platform for our effort, and iterative procedures are used to solve the governing equations. Specifically, the external driving forces are individually constructed for each nodal point through the integration of the data-driven edginess measures, the prior spatial distributions of the myocardial tissues, the temporal coherence of the image-derived salient features, the imaging/image-derived Eulerian velocity information, and the cyclic motion model of the myocardial behavior. Robustness and accuracy of the strategy with very promising application results from canine magnetic resonance phase contrast and tagging image sequences are demonstrated.