Pancreas Segmentation using Level-set Method based on Statistical Shape Model

As Computer Aided Diagnosis (CAD) was widely used in medical field, it was necessary to be studied deeply, especially in image segmentation. The complex anatomy structure around the pancreas made segmentation of pancreas become more difficult. In this paper a new method for medical image segmentation was proposed which used level-set method based on statistical shape model to solve above problem and improve the performance of segmentation. The level-set method was used firstly to obtain the segmentation result of the pancreas from CT images. Then some anatomical information was added into segmentation process by using a better shape of the interest object obtained by statistical shape model. Several experiments were carried out to demonstrate the validation of our proposed algorithm, and the level-set and region-grow methods were chosen as the contrast algorithm. We also gave quantitative results which showed that our method had a good robustness and better precise of segmentation than other methods.