Research On Segmentation Algorithm Of Abdominal Wall Based On Active Shape Model

In the abdominal cavity tissue lesions detection and diagnosis system based on abdominal CT images, the abdominal wall is segmented. Because of the adhesion of the abdominal wall and some tissues in the abdominal cavity, the gr ay level is very close to the segmentation. The traditional gray based segmentation algorithm is difficult to achieve the ideal segmentation results. To solve this problem, the segmentation algorithm based on the active shape model is applied to segment th e abdominal wall.The basic idea of the segmentation algorithm based on the active shape model is to use the prior knowledge of the shape and shape of the sample, to constrain and guide the segmentation of the target image. The algorithm includes three parts: the establishment of the active shape model, the determination of the initial shape, the segmentation search of the model. There are large differences between the abdominal cavity, middle and lower part and different cases, and a number of models are established. Specific method is: on the abdomen of the upper, middle and lower part of the models are established respectively subsection integral model, and then establish sub packet model of grouping sections of the case. Moreover, the feature extraction of the multi-scale local feature model is extracted from the sample images of different scale. Then, the segmentation search of the first stage is carried out by using the abdominal contour and the segmented integral model. After the segmentation result, the initial shape of the second stage segmentation is determined according to the grouping sub model and the position of spine. In the segmentation search let each feature point with reference to their local feature model to move to their optimal positions to obtain the desired shape, and then use the shape model constraints to avoid illegal shape, repeat these two steps until the search shape does not change. In the search process, the multi-scale search method is used to search the hierarchical search, and the accuracy of the search is improved by the gradient weighting.The segmentation of the abdominal CT sequence images of a number of cases is segmented using the segmentation algorithm based on the active shape model. The results show that the algorithm can segment the abdominal wall accurately.