CASIA OpenIR  > 毕业生  > 博士学位论文
医学图象处理与分析中图象分割与绘制技术研究
诸葛婴
学位类型工学博士
导师田捷
1999-07-01
学位授予单位中国科学院自动化研究所
学位授予地点中国科学院自动化研究所
学位专业模式识别与智能系统
关键词医学可视化 图象分割 体绘制 动态规划 模糊连接 距离变换 Medical Visualization Image Segmentation Volumc Rcndcring Dynamic Programming Fuzzy Connectedness Distance Transform
摘要医学图象处理与分析近些年来得到迅速的发展,它可以辅助医生对病变或 各种正常组织进行定性甚至定量的分析,并进行直观、有效的显示,提高医学 诊断的准确性和科学性。同时在手术规划与模拟、医学教学和医学研究中也发 挥着重要的作用。 本文主要对医学可视化领域的图象分割及绘制技术进行了深入的研究,提 出了一些创新性的方法。主要包括以下几个方面: 1.针对医学图象的特点,本文改进了一种基于动态规划的交互式图象分割方 法,着重研究了以下两方面的内容:(1)采用适当的训练机制,提高图象分 割的抗噪音能力;(2)对目标物体边缘局部模糊的情况,将区域分割方法 与动态规划方法相结合,以提高分割的准确性。 2.改进了关于图象的“模糊连接”描述方法,定义了一个新的图象模糊场, 提出了基于模糊连接的图象分割方法,并应用于人脑磁共振图象中肿瘤的 分割,取得了较好的实验效果。 3.提出了一种三维欧氏距离变换算法,在三维空间中可以得到完全欧氏距离。 算法的时间复杂度为O(n3logn)。并将三维欧氏距离变换算法应用于医学CT 图象内部软组织的显示,并取得了较好的实验效果。 4.本文提出了一种基于体绘制的三维数据场多表面显示方法,可以很好的反 映三维数据场的全局状况。不仅加快了图象显示的速度,还提高了投影成 象法的图象显示质量。 5.与课题组成员一起开发了基于微机Windows 95/Windows NT的三维医学图 象处理与分析系统3DMed。3DMed具有丰富的医学图象处理与分析功能, 实现了二维、三维医学图象数据管理、二维断层图象预处理、断层图象重 组、三维面绘制、三维体绘制、手术模拟等一系列功能。 本文工作受到国家自然科学主任基金及国家863计划基金资助。3Dmed系 统获“98年度亚洲CT十大进展”奖。
其他摘要Medical image processing and analyzing has been developed rapidly in recent years. It can assist doctors to diagnose and analyze all kinds of tissues of interest qualitatively and quantitatively, and display them effectively with 3D style, which can largely improve the accuracy of diagnosis. Moreover, it can play important role in the field of operation planning and simulation, medical education and medical research. In this thesis, some novel ideas and methods on medical image segmentation and volume rendering have been presented, which mainly include as follow: 1. According the features of medical image, we improved an image segmentation method based on dynamic programming. First, we append a training mechanism to enhance the robustness of our approach to local noise. Second, we integrate the region-based segmentation with DP to improve the accuracy of image segmentation. 2. We modified the concept of fuzzy connectedness, and defined a new image fuzzy field, which can not only make the aimed object having high fuzzy degree of cormectedness with start point, but also can represent locally the degree that pixels belong to the aimed object. We presented a segmentation method of tumors in human brain MR image based on fuzzy connectedness, and made some experiments. 3. We presented a new method of 3D Euclidean distance transform, which can get the true Euclidean distance, and time complexity of the method reach O (n31ogn). We applied the method to render the soft tissue in 3D medical CT image. 4. We presented a multi-surface rendering method based on volume rendering technique for 3D data sets, which can represent the whole data sets better. The method can not only make the rendering fast, but also can improve the image quality. 5. We developed a 3D medical image processing and analyzing system based on PC Windows 95/Windows NT environment, which has powerful functions such as image preprocessing, reslice, surface rendering, volume rendering, manipulation. Our research work was sponsored by National Natural Science Fund and National 863 High-tech Fund. And the 3D medical image processing and analyzing system 3Dmed awarded Asian 10 CT scientific and technological Advances in 1998.
馆藏号XWLW534
其他标识符534
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/5702
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
诸葛婴. 医学图象处理与分析中图象分割与绘制技术研究[D]. 中国科学院自动化研究所. 中国科学院自动化研究所,1999.
条目包含的文件
条目无相关文件。
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[诸葛婴]的文章
百度学术
百度学术中相似的文章
[诸葛婴]的文章
必应学术
必应学术中相似的文章
[诸葛婴]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

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。