Studies On Medical Image Segmentation Based On Variational Level Set

With the development of modern medical imaging, medical image segmentation technique has been widely used in clinical cases, but there is importance for further research in this area. In this paper, under the curve evolution theory with variational level set methods, the author introduction two kinds of geometric active contours models that are edge-based and region-based at detail, then analyzes their advantages and weaknesses in image segmentation, and proposes a new variational level model that bases on the edge and region information of a image.For the edge-based model, the author gives Geodesic Active Contours model (GAC) and Chunming Li model which is not needed to re-initialize the level set function for example, they can deal with the topological changes for the curves easily. But the two models are sensitive to noise. For the region-based model, the author gives Mumford-Shah model and Chan-Vese model for example. The two models are suitable to noisy images ,have few iterations and the segmentation results are accurat. But the edge of the segmentation object is not accurate.At last, the author puts forward a new variational level model that is local optimization geometric active contours edge-based . Across many experiments, the author find that the new model inherits the advantage of the two classic models, have few iterations and the segmentation results are accurat. it is worth for further research.