Cardiac video analysis using Hodge Helmholtz field decomposition.

摘要

Ventricular fibrillation (VF) is an extremely rapid, highly irregular heart arrhythmia originating in the ventricles. When the VF occurs, the heart loses its capability of pumping blood, and the patients die within minutes unless the VF is immediately stopped. The mechanisms of the VF are still not completely understood. Several hypotheses suggest that it is important to extract the pure expanding component and the pure rotational component from the cardiac electrical patterns.;In this thesis, we first implement the 2-D discrete Hodge-Helmholtz field decomposition (DHHFD) based on regular triangular grids such that it can be directly used for video analysis. We then analyze the optical flow of the cardiac electrical patterns using the 2-D DHHFD. The pure expanding and the pure rotational motion components of the cardiac electrical signals are extracted. Analyses of the decomposed motion components have shown that the VF might be caused by the strong rotational components of the dynamical cardiac electrical patterns. Techniques have also been developed to detect the dominant critical points such as sources, sinks, and rotational centers in the cardiac electrical patterns. The critical points provide important clues for describing and understanding the abnormal propagation of the cardiac electrical signals. The proposed framework also provides a general tool for analysis of the fluid dynamics.