Multi scale lung extraction based on an improved feature-guided geodesic active contour model

Object extraction is usually a hot and challenging problem in medical area. Within this area, variational methods are used largely when showing their stunning performance. However, they are still often confronted with the obstacles of local minima issues, which prevent the optimization process converging to the right optima significantly. In this paper, an improved multi-scale object extraction based on feature-guided active contour model with its application in lung segmentation is proposed, which is based on novel constrained variational framework. The experimental results show that the proposed algorithm has a better performance over traditional relative methods.