项目名称: 基于不确定图模型的医学图像数据挖掘关键技术的研究
项目编号: No.61272184
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 潘海为
作者单位: 哈尔滨工程大学
项目金额: 85万元
中文摘要: 近年来,以医学图像为对象的图像挖掘形成了一个重要研究领域-医学图像挖掘。医学图像挖掘可以提取出那些隐含在医学图像数据中的知识或模式,起到辅助诊断和医学经验共享的作用,对挖掘知识的质量、挖掘算法的效率和准确率都有更高的要求。传统的医学图像挖掘算法几乎都是将图像转换成事务集合或者多维向量集合,不能充分表达图像的语义,挖掘复杂的知识和模式,而且,很少考虑医学图像中存在的诸多不确定性和领域知识对算法的指导作用,无法满足医学领域越来越高的要求。本项目将针对医学图像的特点,集中研究领域知识指导下的高效高准确度的医学图像挖掘问题,拟提出I2G(Image to Graph)的新方法,利用不确定图模型来表达医学图像,研究利用领域知识指导新型不确定图模型的建立,研究在领域知识指导下,基于不确定图模型的特征子图模式挖掘算法、邻近模式挖掘算法、Top-k图模式挖掘算法、图分类算法,以及医学图像数据挖掘系统框架等
中文关键词: 不确定图;医学图像;数据挖掘;;
英文摘要: In recent years, Image mining on medical images forms an important research field-medical image mining. Medical image mining can deal with the extraction of implicit knowledge or other patterns not explicitly stored in the images and play the important role of aided diagnosis and medical experience sharing. Therefore, higher requirements are brought up to the quality of mining knowledge, the efficiency and accuracy of the mining algorithms. Traditional medical image mining algorithms that almost convert the image into transaction set or multi-dimensional vector set, can not fully express the semantics of the image and find the complex knowledge and mode. These algorithms also pay little attention to the uncertainties existing in the medical image and the guiding roles of domain knowledge. So they can not satisfy the more and more high requirement of the medical field. According to the features of medical image, this project focuses on the high efficient and high accurate medical image mining problems with direction of medical domain knowledge. This project plans to put forward the new method I2G (Image to Graph), use the uncertainty graph model to express medical images, and examine the use of domain knowledge to guide the establishment of the new uncertainty graph model. With the guidance of domain knowledge an
英文关键词: Uncertain graph;Medical image;Data Mining;;