Applications Of Active Shape Models In Medical Image Processing

Image segmentation is very important in medical image processing. It can provide guidable information in a clinical medical plan. This paper introduces traditional image segmentation methods and the theory of deformable models, and gives a brief introduction for parametric defoimable models. The major part of the paper describes the theory of the active shape model, gives some examples in medical image segmentation and some extensions to the ASM.The Active Shape Models (ASM) uses a different approach to incorporate prior shape information. The prior model described in this paper is Point Distribution Model (PDM). At first, we select some shape examples to be a training set. In order to model a shape, we represent it by a set of points by placing points around the boundary. This must be done for each shape in the training set. Then we align the training set and capture the statistics of a set of aligned shapes. In order to make use of the statistics of the grey level in regions around each of the labeled model points, we should model the grey level appearance. In image search, we calculate a suggested movement for each model point and compute changes in the pose and shape parameters, then update the pose and shape parameters. The procedure is iterative, it repeats until no significant change results. The final shape is the shape we wanted.In this paper, extension methods are introduced. A method of combining Snake with Active Shape Models is used to segment the left ventricle. In addition, this paper introduces the AAM(Active appearance models )and a method of using the idea of the ASM to improve the external forces of the snakes.With the development of 3D technology of the computer vision, segmenting 3D image slices of medical image will attract more attention. Medical image segmentation faces to clinic applications. Accuracy is needed. But present development is not enough. There isn't a general method .The work is going on for better solutions.