Research On Marching Cubes Reconstruction Based On 3D Segmentation

Application and research of surface rendering has been used more often in various technology of medical image visualization, for it's simple, less calculation and relatively quick. In medical data isosurface does not always represent the surface of an object. In order to solve this problem, to reconstruct a better surface of interesting object, the representative Marching Cubes algorithm of surface rendering is integrated with 3D segmentation to reconstruct human organ partly.In this dissertation, filtering is done to reduce impact of noise on 3D segmentation. Linear interpolation is done, when the spacing between slices is much more than the spacing between pixels in slice, to avoid distortion. In the research of 3D segmentation, regiongrow algorithm is adopted to make prepare for using Marching Cubes algorithm for surface rendering.In this dissertation, by the analysis of Marching Cubes algorithm and its improved algorithms, Marching Cubes algorithm is composed by two parts: "data segmentation" and "surface fitting", so 3D segmentation algorithm could be embedded in Marching Cubes algorithm. Besides, Marching Cubes algorithm could expand from threshold segmentation to regiongrow segmentation, so as to fit for various data object and different segmentation goal.Marching Cubes algorithm based on 3D segmentation is implemented on 3DMed benchmark using the plug-ins interfaces that is exoteric. The reconstruction results of CT data of head and both leg and abdominal is given. The result shows that Marching Cubes algorithm embedded with region growing could reconstruct small regions like tumor in head. And the algorithm embedded with 3D threshold segmentation could realize reconstruction of thin structure like crura bones.