Tracking the endocardial border in artifact-prone 3D images

Echocardiography is a commonly-used, safe, and noninvasive method for assessing cardiac dysfunction and related coronary artery disease. The analysis of echocardiograms, whether visual or automated, has traditionally been hampered by the presence of ultrasound artifacts, which obscure the moving myocardial wall. In this study, a novel method is proposed for tracking the endocardial surface in 3D ultrasound images. Artifacts which obscure the myocardium are detected in order to improve the quality of cardiac boundary segmentation. The expectation-maximization algorithm is applied in a stationary and dynamic, cardiac-motion frame-of-reference, and weights are derived accordingly. The weights are integrated with an optical-flow based contour tracking method, which incorporates prior knowledge via a statistical model of cardiac motion. Evaluation on 35 three-dimensional echocardiographic sequences shows that this weighed tracking method significantly improves the tracking results. In conclusion, the proposed weights are able to reduce the influence of artifacts, resulting in a more accurate quantitative analysis.