医学影像可视化算法及集成化医学影像算法平台的研究
文献类型：学位论文
| 作者 | 赵明昌 |
| 学位类别 | 工学博士 |
| 答辩日期 | 2004-05-01 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 田捷 |
| 关键词 | 医学影像 基于点的重建 基于点的绘制 表面绘制 体绘制 算法平台 图像分割 图像配准 Medical Imaging Points Based Reconstruction Points Based Rendering Surface Rendering. Volume Rendering Algorithms Toolkit Image |
| 其他题名 | Research on Medical Image Visualization Algorithms and Integrated Medical Imaging Algorithms Toolkit |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 自从X射线发明以来，CT（计算机断层成像）、MRI（核磁共振成像）、CR（计算机X线成像）、B超、电子内窥镜等现代医学影像设备先后出现，使得传统的医学诊断方式发生了革命性的变化。随着现代计算机科学技术的发展，使用计算机对医学影像进行处理与分析，从而辅助医生进行更准确的珍断这一技术越来越多地受到人们的重视，现在已经成为一门新兴的发展迅速的交叉科学领域。21世纪是以人为中心的世纪，如何更全面地掌握和更有效地利用人的信息将会非常关键，与全社会人民的医疗保健和健康事业息息相关的医学影像学也将在21世纪得到快速的发展及广泛的应用。 在医学影像的各个研究分支中，可视化无疑是非常重要的一个领域。在传统的影像医疗诊断中，医生主要是通过观察CT、MRI的二维断层图像去发现病变体，这往往依赖于医生的阅片经验，而通过医学影像可视化技术，可以为医疗人员提供具有真实感的病变区域的三维模型，便于他们从多角度、多层次进行观察，并对病变体或感兴趣的区域进行定量的分析，使医生看得更清楚，看得更准确，从而可以大大提高医疗诊断的正确性和准确性。医学影像学要得到广泛的应用，除了在基础算法理论上的研究以外，在算法支撑平台上的研发也是必不可少的。在国际上，很多科研机构对丌发高质量的医学影像算法研发平台非常重视：美国国家卫生院下属的国立医学图书馆近几 年投入巨资支持开发医学影像分割与配准算法平台ITK；SPIE Medical Imaging和MICCAI近两年的年会上都有专门的分会来探讨未来在医学影像领域内高质量软件，尤其是算法研发平台（Algorithms Toolkit）的研究问题。 本文的主要工作集中在两个方面：第一个方面是可视化算法的研究方面：第二个方面是集成化的医学影像算法平台的研究与实现。本文工作主要内容包括： 1）提出了一种基于单层表面跟踪的三维重建算法，可以快速地处理高解析度的医学图像。通过三个主要的步骤：单层的表面跟踪，数掘缓冲机制和二角带（Triangle Strips）生成，算法可以在当前主流的PC硬件配置和中等三维加速卡的条件下，快速地处理高分辨率的医学图像，并且实时、真实地显示重建出来的三维数据。克服了传统的三维重建算法占用内存大和速度慢的缺点，在几个实际医学数掘集上的实验也证明了这一点。 2）提出了一种基于点的重建算法，和计算机图形学领域内的基于点的绘制相结合，同时考虑到医学影像数据集的特点，形成了一个新的算法：基于点的重建与绘制（Points Based Reconstru |
| 英文摘要 | With the application of modem medical imaging device, such as CT, MRI, CR and Electronic endoscopy, the traditional diagnosis methods have revolutionary changes. The techniques which use computer to assist doctors to process and analyze the medical images are called medical image processing and analyzing. Because 20st century is a century in which scientific technology related to human-self will get great progress, we can believe that after the subject formation and development process in 20st century, medical image processing and analyzing will get into a widely application process. 3D visualization is a very important research area in medical image processing and analyzing. It can provide doctors realistic 3D human organ model and multi-view, multi-layer displaying. This technique can improve the accuracy of medical diagnosis notably. In addition to the effort on algorithms research, the research and development of algorithms support platform is also important to promote the application of techniques. Many international conferences have devoted to this area. For example, in the conference of SPIE Medical Imaging 2004, there is a session called Visualization Toolkits to devote to the toolkits development; in the conference of MICCAI 2003, there was a workshop called Software Development Issues for Medical Imaging Computing & Computer Assisted Interventions to devote to the similar topic. This thesis focuses on the above two research areas. The main research works are listed as follow. 1) A fast surface reconstruction algorithm is proposed for processing large scale medical images with high resolution. Through the following main steps: surface tracking in single layer, data caching mechanism and triangle strips generation, the algorithm can extract and represent isosurface efficiently at current mainstream PC. 2) We proposed a new Points Based Reconstruction and Rendering algorithm which can interactively reconstruct and render very large volume dataset in near real time on common PC. The PBRR algorithm uses the points as basic primitive of surface reconstruction and rendering to reduce the computational cost. By utilizing the special characteristic of medical image datasets and making use of modern CPU SIMD instructions, we obtain a fast and efficient algorithm. The experimental results on Visible Man dataset show that this algorithm is feasible and efficient. 3) We proposed a data flow based computational framework and designed and implemented an integrated medical imaging algorithm toolkit, called MITK, based on this framework. MITK combines the functions provided by VTK and ITK, and provides a consistent framework to integrate medical image segmentation, registration and visualization algorithms. To promote the wide application of MITK, we distribute MITK as Freeware and release it on Internet. The related researchers can downlo |
| 语种 | 中文 |
| 其他标识符 | 811 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/5815]  |
| 专题 | 毕业生_博士学位论文 |
| 推荐引用方式 GB/T 7714 | 赵明昌. 医学影像可视化算法及集成化医学影像算法平台的研究[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2004. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。