A Momentum-Based Constraint On Optical Flow Tracking Of The Endocardial Surface

Echocardiography is the standard of care for the evaluation of cardiac function in a variety of clinical scenarios. Despite the increasing availability of RT3D imaging, its utility remains limited due to a lack of tools available to analyze 3D+t datasets. In previous work, we have proposed and validated optical flow as an effective correlation-based technique to track myocardial motion and deformation in RT3D datasets. However, OF's ability to track small regions of tissue (e. g. the endocardial surface) is diminished in less optimal acquisitions. Our goal, therefore, is to develop additional constraints on OF-estimated motion in order to increase the robustness of endocardial surface tracking. We present several modifications to OF-based tracking including motion field smoothing and momentum correction that results in improved OF tracking.