摘要
目的探究手部异常特征与冠状动脉粥样硬化性心脏病(简称“冠心病”)的相关性,为数字化中医手诊提供临床数据支持。方法利用基于NVIDIA Jetson平台和Qt框架的手诊信息采集技术,采用手掌关键点预测算法,通过图像分析和数据挖掘,自动捕捉、采集手掌和指甲的细微特征。收集2023年12月—2024年4月于北京中医药大学东直门医院心内科接受冠状动脉CT血管成像(CACTA)检查的心脏病患者,共纳入438例。根据CACTA结果分为冠心病组(148例)和非冠心病组(290例),比较两组患者的手部特征,并通过随机森林分析及单因素、多因素logistic回归分析筛选出与冠心病相关的异常手部特征。结果综合单因素logistic回归分析及随机森林分析结果,筛选出与冠心病有关并纳入多因素logistic回归分析的变量,包括大鱼际形态、大鱼际是否有皱褶、指甲形态、指甲纹理、中指血管长度。经多因素logistic回归分析,大鱼际肥厚[比值比(OR):3.049,95%CI:1.430~6.503,P=0.004],大鱼际有皱褶(OR:2.206,95%CI:1.119~4.348,P=0.022),指甲有灰黑色竖条纹(OR:1.981,95%CI:1.173~3.347,P=0.011),甲面凹凸(OR:3.130,95%CI:1.822~5.378,P<0.001),以及甲面内扣(OR:5.727,95%CI:1.812~18.102,P=0.003)与冠心病呈正相关;中指血管超过指节的1/3(OR:0.405,95%CI:0.234~0.702,P=0.001)与冠心病呈负相关。结论冠心病与某些手部特征存在相关性,此发现可为数字化中医手诊提供临床数据支持。
Objective This study aimed to explore the correlation between abnormal hand features and coronary atherosclerotic heart disease(CHD)to provide clinical data support for digitalized traditional Chinese medicine(TCM)hand diagnosis.Methods A palm key point prediction algorithm was used to automatically capture and collect detailed features of the palm and nails through image analysis and data mining using the hand diagnosis information collection technology based on the NVIDIA Jetson platform and Qt framework.A total of 438 cardiac patients who underwent coronary artery computed tomography angiography(CACTA)in the Department of Cardiology,Dongzhimen Hospital,Beijing University of Chinese Medicine,from December 2023 to April 2024 were included and divided into CHD(148 cases)and non-CHD groups(290 cases)based on the CACTA results.The hand features of the two groups were compared,and abnormal hand features associated with CHD were screened using random forest analysis as well as univariate and multivariate logistic regression analyses.Results Based on the results of univariate logistic regression and random forest analyses,a set of hand-related features associated with CHD were identified and subsequently included in the multivariate logistic regression analysis.These features included the morphology of the thenar eminence,the wrinkles of the thenar eminence,nail shape,nail texture,and the length of the blood vessel in the middle finger.Multivariate logistic regression analysis revealed that hypertrophic thenar eminence[odds ratio(OR):3.049,95%confidence interval(CI):1.430-6.503,P=0.004],presence of wrinkles on the thenar eminence(OR:2.206,95%CI:1.119-4.348,P=0.022),presence of gray-black vertical stripes on the nails(OR:1.981,95%CI:1.173-3.347,P=0.011),uneven nail surface(OR:3.130,95%CI:1.822-5.378,P<0.001),and inward-bending nail surface(OR:5.727,95%CI:1.812-18.102,P=0.003)were positively associated with CHD.In contrast,the blood vessel in the middle finger longer than 1/3 of the phalanx was negatively associated with CHD(OR:0.405,95%CI:0.234-0.702,P=0.001).Conclusion Certain hand features are significantly associated with CHD,providing the valuable clinical data to support for the digitalization of TCM hand diagnosis.
作者
周晖
李筱
关之玥
王爽秋
李建雨
陈琪
王豪
胡孔法
徐雪
ZHOU Hui;LI Xiao;GUAN Zhiyue;WANG Shuangqiu;LI Jianyu;CHEN Qi;WANG Hao;HU Kongfa;XU Xue(School of Medicine,Wuhan University of Science and Technology,Wuhan 430065,China;Dongzhimen Hospital,Beijing University of Chinese Medicine,Beijing 100700,China;School of Artificial Intelligence and Information Technology,Nanjing University of Chinese Medicine,Nanjing 210023,China)
出处
《北京中医药大学学报》
北大核心
2025年第8期1044-1051,共8页
Journal of Beijing University of Traditional Chinese Medicine
基金
国家重点研发计划项目(No.2022YFC3502300,No.2022YFC3502302)。
关键词
冠状动脉粥样硬化性心脏病
手部特征识别
中医手诊仪
智能诊断
随机森林分析
coronary atherosclerotic heart disease
hand features recognition
traditional Chinese medicine hand diagnostic instrument
intelligent diagnosis
random forest analysis
