利用数据挖掘技术构造医学图像分类器的研究

Study of Medical Image Classifier Using Data Mining Technology

基于数据挖掘的医学图像分类方法研究是多媒体数据挖掘的一个重要组成部分。在分析和总结了现有各种特征提取方法的基础上,提出了基于竞争聚类和关联规则的医学图像分类算法和基于关联规则的医学图像分类器框架。该算法先用竞争聚集算法实现医学图像的聚类,利用聚类的结果提取局部特征,基于局部特征用关联规则实现医学图像的分类。实验结果表明,用此方法较好地提高了医学图像分类的准确率,进而为数字化临床诊断提供了有利的证据。

The multi - media data mining is the key part of the whole researches about the method of medical images classification. Under the base of the analysis and conclusion about the methods of deriving the various characteristics previously provided by other researchers, this framework of the medical images classification with the association rules and clustering was provided. At first, the clustering algorithm was used for the characteristic in local areas , and then the medical image classification was realized by the association rides. The results of the experiment showed that the accurate rate could be improved by this method, and better testimony could be provided for digital.diagnosis.