Vessel Segmentation Based on Region-Scalable Fitting Energy
The region-scalable fitting energy (RSF) model is one of the classical active contour models, it performs well while dealing with the images with intensity inhomogeneity. However, complex vessels with more branches and different sizes make it disabled for image segmentation. In order to overcome the difficulties in complex vessels segmentation, we propose an improved region-scalable fitting energy. It consists of area term, length term, and penalty term. Due to the length term and penalty term both smoothing the zero level contour, they may prevent the level set function from evolving. Therefore, we use the arctangent function to reduce the impact of penalty term and length term to further evolve. And we modify the area term to denoise. The experiments show that the proposed method can have a better segmentation performance while dealing with complex vessels with intensity inhomogeneity. The proposed method is less sensitive to initial contour and robust to illumination.
KeywordsIntensity inhomogeneity Level set Active contour model Image segmentation
- 1.Mcauliffe, M.J., Lalonde, F.M., Mcgarry, D., Gandler, W.: Medical image processing analysis and visualization in clinical research. In: IEEE Symposium on Computer-Based Medical Systems (2010)Google Scholar
- 3.Li, C., Xu, C., Gui, C., Fox, M.D.: Level set evolution without re-initialization: a new variational formulation. In: Conference 2005, CVPR, vol. 1, pp. 430–436. IEEE (2005)Google Scholar
- 15.Li, C., Kao, C.Y., Gore, J.C., Ding, Z.: Implicit active contours driven by local binary fitting energy. In: IEEE Conference on Computer Vision & Pattern Recognition (2007)Google Scholar