Ultrasound Ventricular Contour Extraction Based on an Adaptive GVF Snake Model

In order to accurately extract the ultrasound ventricular contour, an adaptive contour extraction algorithm based on a kind of active contour model was put forward against the shortcomings of the traditional GVF snake model that was difficult to converge to the deep concave edges and noise sensitivity. The algorithm was based on the traditional GGVF snake, which added a pressure pointing to the normal direction of snake contour in the part of external force, and the value of the force was adaptive to the mean gradients of the neighbor contour points. The experimental results show that the algorithm can make the contour converge to the deep cavity border fleetly and improve the robustness to noise. Furthermore, it also can maintain the ability of weak edge extraction which has a good performance of the ultrasound ventricular contour extraction.