摘要
针对医学图像数据的特殊性,提出了一种适合挖掘大量医学图像数据的关联分类算法。该算法以频繁模式树为基础,通过引入双支持度,排除一部分对分类无意义且存在干扰的项,以提高分类正确率。实验结果表明,当用于医学图像分类时,该算法可以取得同样的基于关联规则的分类算法CMAR更高的执行效率及更好的分类效果。
According to the characteristic ofmedical image dataset, new associative classification algorithm is introduced which suitable for mining huge medical image dataset. The new algorithm is based on FP-growth, which introduce double-support to eliminate items which interfere with classification. The experiments show that when used for medical image classification the method has better efficiency and classification accuracy than other reported associative classification methods.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第12期3234-3236,共3页
Computer Engineering and Design
基金
国家自然科学基金项目(60572112)
关键词
数据挖掘
关联规则
分类
频繁模式树
医学图像
data mining
associative rules
classification
FP-growth
medical image
