【作者】 崔宁；

【导师】 张彩明；

【作者基本信息】 山东大学 ， 计算机应用技术， 2007， 硕士

【摘要】 虚拟内窥镜技术是随着计算机技术、计算机图形学、计算机图像处理尤其是虚拟现实等学科的发展而逐步形成的一种独特的技术。它属于医学虚拟现实的范畴，是医学技术在虚拟现实中的推广。与传统的光学内窥镜相反，它不是把镜体植入到人体中去，而是想办法将人体的管腔组织在计算机上显现出来，让内窥镜直接在计算机上进行漫游。具体办法是，病人在检查时首先作CT、MRI或超声等扫描，所得数据传入计算机后进行处理，生成便于计算机显示的数据，然后利用类似虚拟现实的手段对所要检查的区域进行显示，在计算机的屏幕上观察是否有病变。显然，整个检查过程中病人只需要做一次扫描就可以，所以病人的痛苦大大地减低了。由于在检查的过程中减少了病人的参与，从而也很大程度的降低了医疗事故的发生率以及医疗成本。所以，虚拟内窥镜技术在医学上有着广泛的应用前景。本文针对虚拟内窥镜系统的关键技术和实现方法进行了深入研究。首先，文章分析了在CT、磁共振、医院PACS系统中应用广泛的医学图像数据DICOM文件格式，实现对DICOM文件的正确解读。解读DICOM文件是三维重建的前提。只有准确的提取出文件中的信息，如病人的检查部位、图像的窗宽窗位以及切片厚度等，才能够建立人体数据的三维模型，从而为虚拟内窥镜以及其他的医学应用奠定基础。接下来利用提取出的信息进行三维重建，本文简要介绍了几种常用的表面绘制算法，并对Marching Cube算法的实现流程作了说明。其次，文章详细讨论了虚拟内窥镜系统的相关内容，包括虚拟内窥镜的发展现状，基本原理，实现步骤以及各个步骤的实现方法。虚拟内窥镜系统实现的一个关键步骤是路径规划。文章对此部分进行了深入的研究。路径规划包括图像二值化，二值图像骨架化，路径生成等三个步骤。其中骨架化是虚拟内窥镜的核心。骨架化的意义在于提取出管腔组织的中心点，连接这些中心点从而生成漫游的路径。因此中心点的提取速度和准确度直接影响着虚拟内窥镜系统的实用性。本文详细介绍了几种骨架化算法，然后针对其中的拓扑细化方法提出了改进算法。拓扑细化算法是一种有效的提取中心路径的方法，该方法原理简单，得到的结果基本上能够正确的保持物体的拓扑结构，但是耗时却很长，而用于医学处理的数据集往往又是十分庞大的，所以严重的影响了该算法的实用性。本文针对这一缺点提出了几种改进方法，在不同程度上有效地提高了算法的处理速度，减少耗时，从而可以使该算法更好的应用到实际的应用之中。实验证明，改进后的算法可以快速稳定的处理体数据，很大程度上起到了优化的作用。最后，由于直接沿骨架化算法的结果进行漫游容易出现图像跳跃现象，文章采用了NURBS曲线对路径进行平滑处理。文中给出了权因子变化对NURBS曲线形状的影响规律，从力学角度揭示了权因子的物理意义，为更有效地利用权因子提供的额外自由度来调整曲线形状提供了理论依据。更多还原

【Abstract】 VE(Virtual Endoscopy) is a new special technique which developed quickly these years. It is an integration of computer technique, computer graphics, computer image processing and virtual reality, and it is an application of medical virtual reality. Contrary to the optical endoscopy, VE does’t put de endoscopy into the body , but use computer processing of 3D datasets to create simulated views of specific organs similar or equivalent to those produced by standard endoscopic procedures,so we can observe the pathological findings on the computer. The patients who need physical examination only need to make a CT or MRI scan to get the slices, then we can use the slice data to build the models of apparatus on the computer. VE works with the models. For the patients ,VE is painless. For the hospital ,VE also reduce the cost of the examines. Just because of these merits, VE is one of the most promising medical applications.This article research into the virtual endoscopy technique First, DICOM file is analyzed. DICOM file is widely applied in CT, MRI and PACS systems. After getting the information from DICOM files, the 3D reconstruction could be going on. Then, some 3D reconstructed algorithms based on surface rendering algorithm are introduced, and the realization process of Marching Cube algorithm is addressed briefly.Secondly, we introducing the related knowledge of Virtual Endoscopy, including the development of the VE, the theory of the VE, the process of carrying out and the approaches of each step. Path programming is a most important technique of VE. We discussed this part in detail. The path programming including three parts .First ,translating images into the binary digital images, second, skeletonization, and last, path-smoothing. The skeletonization algorithm is the core. Its significance is extracted the centerline of the tube and the camera will move along the centerline to get the observation. So the speed and accuracy of skeletonization algorithms affects the whole system’s utility.This paper describes several kinds of skeletonization algorithm and optimize the topology thinning algorithm. Topology thinning algorithm is an effective method to getting the skeleton. This method is simple and the results are accuracy. But it is a time-consuming algorithm. Usually.the data base of medicine is very large, which seriously influence the practicality of the algorithm. This paper proposed several improvements to reduce the time-consuming. The improved algorithm works better and the experiments show that it can get the centerline rapidly and stably.Finally, practical path-smoothing algorithm based on NURBS is given since image jumping often occurs when flying along the result of skeletonization. Physical meaning of weight is uncovered and then theoretic evidence is provided to modify curve shape by using weight effectively.更多还原