基于粗糙神经网络的医学图像分类新方法

被引:6
作者
蒋芸
李战怀
王勇
张龙波
机构
[1] 西北工业大学计算机学院
关键词
粗糙神经网络; 粗糙集理论; 乳腺X光图像;
D O I
暂无
中图分类号
TP391.41 [];
学科分类号
080203 ;
摘要
由于乳腺X光图像的复杂性,直接从图像中看出肿瘤及其良、恶性质是比较困难的,因此建立高效的肿瘤自动诊断系统是非常必要的。文章将粗糙集理论中基于信息增益的约简方法和神经网络相结合,提出了粗糙神经网络算法RNN,将其应用于乳腺X光图像分类。实验结果表明,该方法的分类精确度可达到92.37%比单独使用神经网络方法的分类精确度(81.25%)要高,同时所花费的时间也明显减少。
引用
收藏
页码:151 / 153
页数:3
相关论文
共 10 条
[1]  
Rough sets andintelligent data analysis. Pawlak Z W. Infor-mation sciences . 2002
[2]  
Evaluation of Texture Methods for I mage Analysis. Sharma M,,Singh S. Proceedings of the7thAustralian and New Ze-land Intelligent Information Systems Conference . 2001
[3]  
Wavelet Based Automatic Thresh-olding for I mage Segmentation. Zhang Xiao-Ping,,Desai M D. Proceedings of the ICIP’97conference . 1997
[4]  
Data Mining Via Generalization,Discretiza-tion and Rough Set Feature Selection. Hu X,Cercone N. Knowledge and Infor-mation System:An International Journal . 1999
[5]  
MammogramScreening Using Multi-resolution-basedI mage Segmentation. Brazokovic D,,Neskovic M. International Journal of Pattern Recognition and Artificial Intelligence . 1993
[6]  
Application of the Wave-let Transformto Automated Detection of Clustered Microcalcifica-tions in Digital Mammograms. Yoshida H,,Doi K,Nishikawa R,et al. Academic Reports of To-kyo Institute of Polytechnics . 1994
[7]  
Application of data mining techniques for medical i mage classification. Antonie ML,,Zaiane O R,Coman A. Proc.of Sec-ond Intl Workshop on Multi media Data Mining(MDM/KDD’2001)in Conjunction with Seventh ACM SIGKDD . 2001
[8]  
Markov RandomFieldfor Tumor Detectionin Digital Mammography. Li H,et al. IEEE Trans Medical I maging . 2000
[9]  
Feature Extraction from Mammographic I mages Using Fast Matching Methods. Bottigli U,,Golosio B. Nuclear Instruments . 2002
[10]  
A Novel Self-Opti mi-zing Approachfor Knowledge Acquisition. Pan Dan,,Zheng Qi-Lun,Zeng An,et al. IEEE Trans on Sys-tems,Man and Cybernetics-Part A:Systems and Humans . 2002