改进Marching Cubes算法的椎骨CT图像三维重建与可视化  

3D Vertebrae CT Images Reconstruction and Visualization with Optimized Marching Cubes Algorithm

作　　者：刘侠 刘欢 王淼淼 王波 LIU Xia;LIU Huan;WANG Miao-miao;WANG Bo(School of Automation,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学自动化学院图像处理与模式识别研究室,哈尔滨150080

出　　处：《哈尔滨理工大学学报》2019年第6期84-92,共9页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61172167);黑龙江省科学基金青年科学基金(QC2017076);黑龙江省普通本科高等学校青年创新人才培养计划面上项目(UNPYSCT-2015045);黑龙江省教育厅科学技术研究项目(12541105).

摘　　要：在临床医学实际操作和教学研究中,尤其在椎弓根螺钉置入手术中,需要对椎骨进行三维重建,针对椎骨模型重建时间过长,清晰度、光滑度较低的问题,依据三维重建的面绘制理论,首先对54片分辨率为512×512的DICOM格式人体椎骨CT图像采用基于特征选择的双边滤波去噪算法进行预处理,在传统移动立方体(Marching Cubes,MC)算法基础之上进行改进,先选择种子体素,利用区域生长思想,提取出全部包含等值面的体素,再利用黄金分割法代替传统的线性插值法计算等值点,减少公共棱边计算次数,由可视化工具包VTK和OpenGL借助GPU并行处理,完成椎骨图像的快速、精准三维重建与可视化,算法的执行效率整体提高。In the actual operation of clinical medicine and teaching research,especially in the pedicle screw implant surgery,a three-dimensional reconstruction of the vertebrae is required,for the time of reconstruction is too long,and the resolution is low,based on the surface rendering theory of 3D reconstruction.First the 54 DICOM vertebrae CT images that have the resolution of 512×512 were preprocessed with the bilateral-filter denoising algorithm.On the basis of the traditional Marching Cubes(MC)algorithm to improve,we first select the seed voxel,with the region growing algorithm,extract all the voxels that contain isosurfaces,and then through the golden segmentation instead of the traditional linear interpolation to calculate the equivalent point,reduce the calculations of public edge,and then use VTK and OpenGL to implement the 3D reconstruction and visualization of the vertebrae on GPU quickly and accurately,thus the efficiency of the algorithm is improved dramatically.

关 键 词：椎骨CT MARCHING CUBES 三维重建 可视化 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]