计算解剖学在腹部疾病诊断及手术中的应用研究

项目名称： 计算解剖学在腹部疾病诊断及手术中的应用研究

项目编号： No.81460274

项目类型： 地区科学基金项目

立项/批准年度： 2015

项目学科： 医药、卫生

项目作者： 张学军

作者单位： 广西大学

项目金额： 47万元

中文摘要： 计算解剖学是将存在个体差异的人体解剖结构用数理统计记述的方法进行模型 化的新兴学科，其含盖了外科手术学、影像学和计算机科学在医学领域的临床研究。本课题 将着重研究计算解剖学在腹部器质性疾病诊断及手术中的应用，建立庞大的腹部图象数据库 以导出适合壮族地区人群的正常腹部器官的分类模型，进而实现对病变器官的检测；设计实现新的脏器抽出及三维重建算法，利用重建的三维图像可进行外科手术的规划和虚拟，也可用于手术教学和训练。我们提出的基于图像的集成学习算法将不仅克服目前传统分割算法中对不同器官要采用不同分割算法的不足，还能有效地解决病变器官的抽出问题。研究内容包括：腹部器官，血管及组织的全自动抽出；三维可视化及虚拟手术的实现；基于自动寻找特征点的非刚性配准等。

中文关键词： 计算解剖学；腹部手术；器质性疾病；计算机辅助诊断；图像处理

英文摘要： The discipline of computational anatomy covers wide spectrum of medical, biological and mathematical interests. One of the most important features of our project is, however, the certain application-oriented aspect to the computer assisted diagnosis (CAD) and computer-assisted surgery (CAS) on abdominal images. For such applications, computational understanding of medical images based on image segmentation techniques with high accuracy and robustness is expected. Our advancement of medical image segmentation is mainly based on ensemble learning and majority voting that is differ from traditional methods, making the segmentation of organs easier and faster, and most important point is to uniform all the segmentation in one algorithm. Large scale of CT database covered Zhuang's people will be established for the ensemble learning to construct the normal model of abdominal body, which would be also used for surgical planning, simulation and resident's training. By clinical image understanding via such organic anatomy models, abnormal would be detected by such as changes of volume or surface shapes. This work mainly contains several parts: Automatic segmentation of the organ region, vessels and tissues; Establishing the shape model to describe the irregularity of organic surface; 3D visualization and visual surgical planning; and Non-rigid registration based on automaticlly found landmarks.

英文关键词： computational anatomy;abdominal surgery;organic disease;computer-aided diagnosis(CAD);image processing