using Prior Shape and Points in Medical Image Segmentation

In this paper we propose a new variational framework for image segmentation that incorporates the information of expected shape and a few points on the boundary into geodesic active contours. The energy functional in this model consists of three terms. One measures high image gradients, the other two measure the disparity in shape between the interface and prior, and the distance from the prior points to the interface, respectively. The model is presented using the variation level set formulation. The existence of the solution to the proposed minimization problem is also discussed. We also report experimental results on synthetic images and ultrasound images, and compare them with the results of using the model in [C] that only incorporates shape prior into active contours.