Bone CT sequence image oriented grid model adaptive reconstruction method

Abstract

The invention provides a bone CT sequence image oriented grid model adaptive reconstruction method. The method includes following steps: extracting two-dimensional profile data of bones in each layer of a CT image in sequence; converting a formed multi-layer two-dimensional profile data set to three-dimensional point clouds with normal vectors; constructing an implicit surface of a bone model; performing adaptive primitive sampling on the bone implicit surface; and performing triangularization on an adaptive primitive set to form a bone grid model. According to the method, the generation of a lot of redundant triangular grids and narrow and long triangular grids can be prevented in the bone model, the high-quality triangular grid model can be adaptively reconstructed according to the curvature of the bone surface, reasonable transition between the triangular grids of different dimensions is realized, the smoothness effect of the grids is good, and important application values are achieved in the field of digital osteology.

Description

technical field [0001] The invention relates to image processing, in particular to a grid model adaptive reconstruction method for bone CT sequence images. Background technique [0002] CT scanning, as an X-ray computed tomography technique, can non-destructively and accurately obtain images of internal tissues and organs of a living body. In the clinical application of bones, whether the three-dimensional mesh model of bones can be accurately reconstructed is directly related to the subsequent computer-aided manufacturing of bones, finite element analysis of bones, and 3D printing effects. At present, the methods for reconstructing bone CT sequence images into mesh models mainly include: Marching Cubes algorithm and direct triangulation method based on contour set. The moving cube algorithm processes the voxels in the CT volume data field one by one, and approximates the isosurface inside the voxel with triangular patches. Since it can generate voxel-level grids, the skele...

AI Technical Summary

Problems solved by technology

The method of direct triangulation based on the contour set is to form a segmented triangular mesh by directly connecting adjacent contour lines, and then sequentially connect the triangular meshes between all adjacent contours to finally generate an overall triangular mesh model. However, when there are multiple contour lines to be reconstructed in adjacent contours, there will be problems of branching and multivariate matching, especially when reconstructing complex bone models, topological connection errors will occur, and the reconstructed mesh model contains a large number of narrow and long triangular meshes. And the transition between triangular meshes of different sizes cannot be reasonable, and the surface smoothing effect is poor

[0003] In addition, neither of the above two methods can adaptively divide the mesh according to the curvature characteristics of the bone surface. A large number of small-sized triangular meshes are generated in the low-curvature surface area, and local fine-grained mesh cannot be performed in the high-curvature surface area, which cannot effectively guarantee The discrete precision and grid quality of the grid seriously affect the subsequent analysis and use

Images