冠心病气虚血瘀证模型大鼠与临床患者舌象特征的比对研究  

Comparative study of tongue features between model rats and clinical patients with Qi deficiency and blood stasis syndrome of coronary heart disease

作　　者：王瑾茜 雷诗卉 李薇[1] 杨蕙[1] 杨益 刘侃[1] 蔺晓源[1] 胡国恒[2] 郭纯[1] 陈亚[1] WANG Jinxi;LEI Shihui;LI Wei;YANG Hui;YANG Yi;LIU Kan;LIN Xiaoyuan;HU Guoheng;GUO Chun;CHEN Ya(the First Affiliated Hospital of Hunan University of Traditional Chinese Medicine,Changsha 410007,China;Hunan University of Traditional Chinese Medicine,Changsha 410208)

机构地区：[1]湖南中医药大学第一附属医院,长沙410007 [2]湖南中医药大学,长沙410208

出　　处：《中国比较医学杂志》2022年第10期8-16,共9页Chinese Journal of Comparative Medicine

基　　金：湖南省自然科学基金项目(2021JJ40418);湖南省中医药科研项目(2021179,2021172);长沙市科技计划项目(kq2014222,kq2014224);湖南省卫健委科研计划项目(202103010864);湖南中医药大学校级科研基金项目(2019XJJJ057)。

摘　　要：目的本研究旨在寻求CHD气虚血瘀证病证结合动物模型与CHD气虚血瘀证临床患者舌象共通的特异性指标。方法使用matlab软件将冠心病气虚血瘀证临床患者舌图像进行分割并提取相关特征,利用RF模型分类识别临床全样本及动物全样本,制作RF预测模型并评估预测价值,计算各指标AUC以判定其诊断价值。结果与CHD气虚血瘀证临床诊断密切相关的舌象特异性指标有4个,分别是M_(1)、M_(2)、M_(3)、M_(5)。采用RF分类判别结果提示模型预测FPR为15.2%,TPR为83.3%,最佳界值为0.5,敏感度为81.4%,特异性为76.0%,准确率为85.7%,圆周卷积为0.6991,AUC为0.8151。M_(1)、M_(2)、M_(5)的诊断价值较高(AUC分别为0.9040、0.9254、0.9610),M_(3)的诊断价值中等(AUC=0.7708)。在大鼠模型上M_(1)、M_(2)、M_(3)、M_(5)特征值与正常组大鼠比较存在统计学差异,采用RF分类判别的FPR为14.8%,TPR为83.1%,AUC为0.8755,最佳界值为0.5,敏感度为81.8%,特异性为75.6%,准确率为83.1%,圆周卷积为0.6451。M_(1)、M_(2)、M_(5)的诊断价值较高(AUC分别为0.9130、0.9304、0.9088),M_(3)的诊断价值中等(AUC=0.7496)。结论CHD气虚血瘀证病证结合动物模型与CHD气虚血瘀证临床患者的舌象特征拟合度高,M_(1)、M_(2)、M_(3)、M_(5)特征值是CHD气虚血瘀证临床样本和动物模型的共通的特异性指标。Objective This study aimed at finding a common specific indication of tongue images between model rats and patients with Qi deficiency and blood stasis syndrome of CHD.Methods Images of tongues of patients with Qi deficiency and blood stasis syndrome of CHD were divided by matlab software and relevant features were extracted.Whole clinical and animal samples were classified and identified using an RF model and the predictive value of the model was evaluated by the ROC.AUC was calculated to determine the diagnostic value of each index.Results Four indications named M_(1),M_(2),M_(3) and M_(5) were closely related to the diagnosis of CHD with Qi deficiency and blood stasis syndrome.In RF classification,FPR was 15.2%,TPR was 83.3%,the best boundary value was 0.5,sensitivity was 81.4%,specificity was 76%,accuracy was 85.7%,circumferential convolution was 0.6991,and AUC was 0.8151.AUC indicated that all variables had diagnostic value,among which M_(1),M_(2) and M_(5) had high diagnostic values and M_(3) had a moderate diagnostic value.In the rat model,significant differences in the characteristic values of M_(1),M_(2),M_(3) and M_(5) were observed compared with the normal group(P<0.05,P<0.01).In RF classification,FPR was 14.8%,TPR was 83.1%,AUC was 0.8755,the best boundary value was 0.5,sensitivity was 81.8%,specificity was 75.6%,accuracy was 83.1%,and circumferential convolution was 0.6451.AUC suggested that M_(1),M_(2),M_(3) and M_(5) had diagnostic value.Among them,M_(1),M_(2) and M_(5) had high diagnostic values and M_(3) had a moderate diagnostic value.Conclusions The RF method can be used as a method of tongue image recognition in CHD with Qi deficiency and blood stasis syndrome.The characteristic values of M_(1),M_(2),M_(3) and M_(5) can be used as specific indications of clinical samples and animal models of CHD with Qi deficiency and blood stasis syndrome.The rat model of CHD Qi deficiency and blood stasis syndrome is in line with the digital characteristics of tongue diagnosis in patients.

关 键 词：冠心病 气虚血瘀证 病证结合动物模型 舌象特征 随机森林 

分 类 号：R-33[医药卫生]