| The generation of multi-tissue anatomic structures from medical images is a basic but important work in many applications of computer assisted medical fields.This thesis proposes a new method to generate multi-tissue tetrahedral meshes for medical images. It mainly includes two steps:constrained Delaunay tetrahedral meshes with tissue labels generation and mixed-tissue tetrahedrons division.Firstly,this thesis does a deeply research to basic principles,methods and characteristics of multi-tissue tetrahedral meshes generation technology from medical images,mainly focusing on grid-based and Delaunay criterion-based tetrahedralization.Then,a method extended from constrained Delaunay algorithm is proposed.The extended algorithm takes outer boundary surface and inner sampling points of anatomic structures as input.According to tissue labels of input points and tissue labels calculation rules,tissue labels of generated mesh elements and newly inserted Steiner points can be determined.The output is a tissue-labeled constrainded Delaunay tetrahedral mesh.However,limited to the tetrahedralization algorithm kernel of geometric distance of points,the tissue-labeled constrained Delaunay tetrahedral mesh contains mixed-tissue tetrahedrons that cross inner tissue boundaries. Mixed-tissue tetrahedral division algorithm calculates locations and tissue labels of dividing points,and remeshes them into pure-tissue tetrahedrons.As a result,a multi-tissue tetrahedral mesh with different classfication pure-tissue tetrahedrons is generated.This method has been implemented using Visual C++ 6.0 on Windows 2000.The resulted multi-tissue tetrahedral models of human mandible and human head from Chinese Visible Human and Korea Visible Human datasets are accepted.This ease new method which complicated models can be generated provides new solutions to complicated modeling problems and applications,such as compute assisted tissue engineer.It also can be generalized to other fields. |
PDF Full Text Request