基于兴趣度和PC-树的医学图像数据关联且相关规则挖掘  

Associative and Correlative Regular Mining Based on Interestingness and PC-tree for Medical Images Data

作　　者：熊平[1] 李娜[1] 

机构地区：[1]中南大学信息物理工程学院,湖南长沙410083

出　　处：《中国医学物理学杂志》2009年第1期952-954,共3页Chinese Journal of Medical Physics

摘　　要：目的:挖掘出海量医学图像中数据的关联关系。方法:针对目前多数研究偏重对支持度的考查,而忽略了置信度的考查,本文采用兴趣度的两个度量标准—支持度和置信度与时空效率高的PC-树频繁模式关联规则算法相结合对医学图像数据进行分析。结果:通过实验验证,该方法能得到更加有效的关联且相关规则,并消除负相关规则。结论:该方法能为医学诊断提供更大的参考价值。Objective： The purpose of this study was to mine the correlative relationship between massive amounts of medical image data. Methods： Currently most researchers concern about the investigation of support degree, but neglect the confidence degree. This paper associated two aspects, support degree and confidence degree that two measure standard of interestingness, and PC-tree frequent pattern correlation rules algorithm with high time-space efficiency, to analysis medical image data. Results： The experiment showed that this method can gain more efficient associative and correlative rules while eliminate the negative correlation rules. Conclusions： This method can provide more important value for medical diagnose.

关 键 词：数据挖掘 关联且相关规则 兴趣度 医学图像数据 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]