基于贝叶斯网络的乳腺肿瘤超声诊断定量分析  

Quantitative Analysis for Ultrasound Diagnosis of Breast Tumors Based on Bayesian Network

作　　者：黄金华[1] 刘烁 苏福再 丁雪梅 杨洪钦[3] HUANG Jin-hua;LIU Shuo;SU Fu-zai;DING Xue-mei;YANG Hong-qin(Department of Computer Engineering,Zhangzhou Institute of Technology,Zhangzhou 363000,China;College of Mathematics and Informatics,Fujian Normal University,Fuzhou 350117 ,China;Key Lab of Optoelectronic Science and Technology for Medicine of Ministry of Education,Fujian Normal University,Fuzhou 350007 ,China)

机构地区：[1]漳州职业技术学院计算机工程系,福建漳州363000 [2]福建师范大学数学与信息学院,福建福州350117 [3]福建师范大学医学光电科学与技术教育部重点实验室,福建福州350007

出　　处：《福建师范大学学报（自然科学版）》2019年第3期36-41,共6页Journal of Fujian Normal University：Natural Science Edition

基　　金：国家重点基础研究发展计划(973计划;2015CB352006);国家自然科学基金重点资助项目(61335011)

摘　　要：利用贝叶斯网络建模方法,定量分析了1 500例良性肿瘤和500例恶性肿瘤临床乳腺肿瘤超声检查相关参量的诊断参考价值以及各参量之间的关联程度.研究结果表明,在超声检查中,形态检查参量的诊断价值最高(40.3%),其次分别为阻力指数(25.0%)、钙化灶(18.4%)和血流信号(16.3%)等检查参量.此外,阻力指数与血流信号之间的关联性比较强,约为0.432.贝叶斯概率模型在乳腺肿瘤超声智能诊断的前期应用研究,有助于帮助医生根据各检查参量诊断参考价值和各参量之间关联程度分析,实现乳腺肿瘤超声的智能诊断,提高诊断准确率.This paper quantitatively analyzed breast tumor ultrasound data which included 1 500 benign tumors and 500 malignant tumors using Bayesian network modelling approach,in order to discover different ultrasound features's contributions to diagnosis and their dependency.The results showed that the shape was the most important ultrasonic feature for breast cancer diagnosis (40.3%),following by resistance index (25.0%),calcification (18.4%) and blood signal (16.3%).Besides,resistance index and blood signal have strong dependence (0.432).This study a preliminary study for the intelligent diagnosis of breast cancer based on Bayesian network probabilistic model.It will provide assistance for clinical expert to make more objective and reasonable analysis and estimation.Diagnostic effectiveness can be improved according to the value of different features and the relationships between those features.

关 键 词：贝叶斯网络 乳腺肿瘤 超声检查 智能诊断 

分 类 号：TN29[电子电信—物理电子学]