A 3-D Multi-modality Image Framework for Left Ventricle Motion Analysis

In this paper, a multiplatform computing tool for left ventricle (LV) motion analysis from three-dimensional cardiac images is described. The tool uses as input data, LV geometric representations constructed from multislice computed tomography databases, and LV 3D models reconstructed from biplane angiography projections. A non-rigid bidimensional correspondence algorithm is used to track a set of anatomical landmarks extracted from LV surfaces. The objective is to establish the correspondence between contours on consecutive instants of time. From the estimated correspondence, parameters associated with cardiac dynamical function are calculated. These parameters are useful for assessing cardiac pathologies. Among parameters considered are: torsion, radial contraction, longitudinal contraction, and the static motion representation based on LV longitudinal and radial slices, and their integration into a polar scheme known as bull's eye.