利用输入阻抗预测体动脉狭窄的仿真研究  

Simulation Investigation of Arterial Stenosis Prediction by Using Input Impedance of Systemic Arterial Tree

作　　者：肖汉光[1,2] 何为[1] 刘兴华[1] 李松浓[1] 毕喜飞[1] 

机构地区：[1]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400030 [2]重庆理工大学光电信息学院,重庆400054

出　　处：《重庆理工大学学报（自然科学）》2011年第6期63-68,123,共7页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(50877082)

摘　　要：通过仿真验证动脉树输入阻抗预测动脉狭窄的可行性,为动脉狭窄的无创检测提供一种新方法。在已建立的55段人体动脉树分布式电网络模型的基础上,通过设定不同狭窄位置和狭窄程度的动脉段和输入阻抗的递归计算,建立具有多样性的输入阻抗仿真病例数据库。利用K近邻分类方法对动脉狭窄进行了十次交叉验证的分类预测。讨论了不同狭窄位置和不同狭窄程度对预测准确率的影响。预测结果表明:K近邻分类方法对动脉狭窄的平均准确率为89.5%,平均特异度为95.8%,平均灵敏度为85.4%;随着狭窄位置的离心距离的减少,总准确率从90%逐渐提高到95.5%;随着狭窄程度的增加,总准确率由约80%提高到99.4%。因此,利用人体动脉树的输入阻抗和K近邻分类方法预测动脉狭窄在理论上是可行的。To propose a novel noninvasive method for the prediction of arterial stenosis by using input impedance of systemic arterial tree,and to verify the feasibility of the method by simulation.Based on the built distributed electric network model of 55 segment arterial tree,a diverse case database of input impedance was established by setting different position and degree of arterial stenosises and calculating them with the recursive algorithm of input impedance.K-Nearest Neighbor Classifier（KNN） was used to classification and prediction with the 10 Cross Validation（CV）.The effects of stenosis position and extent on the accuracies of prediction were discussed.The results show the mean specificity,sensitivity and overall accuracy of KNN is respectively 95.8%,85.4% and 89.5%.With the decrease of distance from heart,the overall accuracy is improved gradually from 90% to 95.5%.When increasing the stenosis degree from 10% to 90%,the overall accuracy is enhanced from 80% to 99.4%.The simulation demonstrated the theoretical feasibility of the proposed method for the prediction of arterial stenosis by using input impedance of systemic arterial tree and KNN.

关 键 词：输入阻抗 动脉狭窄 狭窄预测 动脉树 K近邻分类方法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] R318.01[自动化与计算机技术—计算机科学与技术]