东北大学学报（自然科学版）

东北大学学报(自然科学版) ›› 2013, Vol. 34 ›› Issue (8): 1100-1104.DOI: -

• 信息与控制 • 上一篇    下一篇

基于代价敏感SVM优化组合算法的微钙化簇识别

曹鹏1,2，李博1,2，刘鑫1,2，赵大哲1,2   

  1. (1东北大学信息科学与工程学院，辽宁沈阳110819；2东北大学医学影像计算教育部重点实验室，辽宁沈阳110179)
  • 收稿日期:2013-01-04 修回日期:2013-01-04 出版日期:2013-08-15 发布日期:2013-03-22
  • 通讯作者: 曹鹏
  • 作者简介:曹鹏(1982-)，男，辽宁沈阳人，东北大学博士研究生；赵大哲(1960-)，女，辽宁沈阳人，东北大学教授，博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61001047)；中央高校基本科研业务费专项资金资助项目(N110618001).

Microcalcification Clusters Recognition Based on Optimized CostSensitive SVM Combinational Algorithm〓

CAO Peng1,2， LI Bo1,2， LIU Xin1,2， ZHAO Dazhe1,2   

  1. 1. School of Information Science & Engineering， Northeastern University， Shenyang 110819， China; 2Key Laboratory of Medical Image Computing of Ministry of Education， Northeastern University， Shenyang 110179， China.
  • Received:2013-01-04 Revised:2013-01-04 Online:2013-08-15 Published:2013-03-22
  • Contact: CAO Peng
  • About author:-
  • Supported by:
    -

RichHTML

1

PDF (PC)

758

摘要： 微钙化簇是乳腺癌一个重要的早期发现，现有的检测技术为了达到高敏感性要求，产生很多假阳性数据.根据微钙化簇特点，提出一种整体和局部相组合的分类识别策略，并根据真假阳性样本错分代价的不同，使用代价敏感SVM方法进行分类学习.在构造分类器模型过程中利用粒子群进行分类器的参数优化及特征集合的选择，以提升分类学习的泛化能力.该算法在保证高敏感性的同时，降低了过多的假阳性数据，并删除了冗余和不相关的特征.实验结果表明，基于粒子群优化的代价敏感SVM组合分类算法提高了传统方法的识别能力.

关键词: 微钙化簇检测, 计算机辅助诊断, 代价敏感学习, 组合分类, 粒子群优化, 特征选择

Abstract: Microcalcification clusters(MCs)are important signs for early breast cancer detection. The existing initial detection methods result in lots of false positive data because of the requirements of high sensitivity. A classification strategy combining with global and local views was proposed based on MCs characteristics. The costsensitive SVM classification algorithm was employed according to different misclassification costs of true and false positive instances. In the construction of classification model， the parameters and feature subset were optimized with particle swarm method to enhance the generalization performance. The ensemble method reduces excessive false positive data but with high sensitivity， and removes redundant and irrelevant features. Experimental results show that the proposed method improves the performance of traditional methods on microcalcification clusters recognition.

Key words: microcalcification cluster detection, computeraided diagnosis, costsensitive learning, ensemble classification, particle swarm optimization, feature selection

中图分类号: