基于支持向量机的肝纤维化CT图像分类  

Classification based on support vector machine for the CT image of hepatic fibrosis

作　　者：李宁[1] 孙铭[1] 王磊[1] 刘兴龙[1] 童隆正[1] 

机构地区：[1]首都医科大学生物医学工程学院,北京100069

出　　处：《北京生物医学工程》2007年第1期40-43,共4页Beijing Biomedical Engineering

基　　金：北京市教育委员会科技发展项目(01KJ-096)资助

摘　　要：基于统计学习理论中结构风险最小化原则的支持向量机是易于小样本的机器学习方法。本文使用支持向量机和二叉树的方法对肝纤维化CT图像进行分类,并与k近邻法和BP神经网络等其它算法进行比较,结果显示对于肝纤维化图像,支持向量机的分类效果和鲁棒性要高于其他两种算法。Support vector machine （ SVM ） based on the structural risk minimization of statistical learning theory is a method of machine learning for small sample set. SVM and binary tree were used to classify the CT images of patients with hepatic fibrosis, and the result was compared with those of k-nearest neighbor algorithm and BP neural network in this paper. The comparative result shows that SVM is more effective than k-nearest neighbor and BP neural network algorithms in classification and robust.

关 键 词：肝纤维化 CT图像 支持向量机 最优分类超平面 二叉树 

分 类 号：R318.04[医药卫生—生物医学工程]