Vessel segmentation using centerline constrained level set method
Abstract
Vascular related diseases have become one of the most common diseases with high mortality, high morbidity and high medical risk in the world. Level set is a kind of active contour model, and can be used to extract vessel structures. However, the applications of level set methods in vessel segmentation suffer from two problems. The first problem is the error caused by the false inclusion of some non-vessel structures. The second one is the sensitivity of the level set evolution to the initialization condition. In this paper, we propose an algorithm termed Centerline constrained level set (CC-LS) for vessel segmentation which utilizes centerline information to improve the evolution of level set. Using centerline information as the initial level set condition leads to improved evolution efficiency and extraction accuracy. Additionally, a new centerline modulated velocity term can be used in the level set evolution function to avoid the wrong inclusion of non-vessel structures. Performance of the proposed CC-LS algorithm is well validated using both 2D and 3D coronary images in different types. The proposed method is able to attain satisfactory results on both 2D and 3D coronary data.