Automatic segmentation technique without user modification for 3d visualization in meical images

Abstract

It is necessary to analyze an image from CT or MR and then to segment an image of a certain organ from that of other tissues for 3D (Three-Dimensional) visualization. There are many ways for segmentation, but they have a somewhat ineffective problem because they are combined with manual treatment. In this study, we developed a new segmenting method using a region-growing technique and a deformable modeling technique with control points for more effective segmentation. As a result, we try to extract the image of liver and identify the improved performance by applying the algorithm suggested in this study to two-dimensional CT image of the stomach that has a wide gap between slices. © Springer-Verlag 2004.