METHOD FOR SEGMENTING THREE-DIMENSIONAL IMAGE INCLUDED IN OBJECT

摘要

PROBLEM TO BE SOLVED: To provide a method which requires only a small number of two-dimensional images for the segmentation of a three-dimensional structure and captures a less number of two-dimensional images faster than a three-dimensional image. SOLUTION: This method is concerned in the segmentation of the three- dimensional structure included in an object from >=1 two-dimensional image displaying slices of the object. The method uses a deformable model whose surfaces are formed by a network of meshes mutually connecting network points on the surfaces of the model. Firstly, a mesh crossing at least one image is determined and points on the surface of a structure to be segmented is searched for in the direction of a search line extending in the image across the mesh. Then the position of a network point of the model is newly computed. Those stages are repeated and several models obtained after the deformation are finally considered to be the segmentation of the three-dimensional structure from the two-dimensional images.