Method for automatically generating a two-dimensional triangular mesh of a bone from a CT image considering its density heterogeneity

초록

To perform numerical analyses, such as bio-mechanical analyses, on bones, it is necessary to construct the mesh model of the shape of the bone. In previous studies, the mesh has been generated by assuming that bones are homogeneous in terms of material distribution. However, bones are inherently heterogeneous. Therefore, the mesh has to be generated by considering the variation of the properties of bones. In this study, we propose a method for automatically generating a bone mesh by considering its heterogeneity. However, only two-dimensional triangular meshes are considered. In the proposed method, the boundary polylines of a bone are extracted from its computed tomography (CT) image using the marching squares method. Nodes are randomly generated on and inside the boundary polylines, and triangles are generated from the nodes by Delaunay triangulation. When the nodes are generated on and inside the boundary polylines, the density of the nodes is controlled by threshold functions defined by the magnitude of the gradient vector of the Hounsfield unit values of the CT image. The proposed method was implemented using C++ and visualization toolkit, and experiments were performed using CT images on the right femur, hip bones, and vertebra. We also verified that the proposed method can generate meshes reflecting the variation of the properties of bones.

키워드