实性孤立性肺结节数学诊断模型建立  

Establishment of mathematical diagnostic model of solid solitary pulmonary nodules

作　　者：王欢[1] 钱立勇[1] 曹捍波[1] 王善军[1] 罗清泉 乐涵波[1] WANG Huan;QIAN Liyong;CAO Hanbo(Department of Cardiothoracic Surgery,Zhoushan Hospital of Zhejiang Province,Zhoushan 316021,CHINA)

机构地区：[1]浙江省舟山医院心胸外科,316021

出　　处：《江苏医药》2020年第9期931-934,共4页Jiangsu Medical Journal

基　　金：舟山市科技计划项目(2017C31102)。

摘　　要：目的建立实性孤立性肺结节(SPN)的数学诊断模型。方法选取经胸部高分辨率计算机断层扫描(HRCT)检查发现并通过手术病理确诊的实性SPN患者396例(A组)作为训练样本,采用多因素logistic回归分析筛选出恶性实性SPN的独立预测因子,建立预测恶性实性SPN的logistic回归数学诊断模型。选取经胸部HRCT检查发现并通过手术病理确诊的实性SPN患者254例(B组)作为建立数字诊断模型的验证样本,将B组患者相应数据代入回归方程,用于验证诊断效能。结果A组良性和恶性实性SPN患者在年龄、性别构成比、肿瘤家族史、空气支气管征、空泡征、分叶征、毛刺征、血管集束征、胸膜凹陷上均有统计学差异(P<0.05)。多因素logistic回归分析显示,女性、肿瘤家族史、分叶征、毛刺征、血管集束征是恶性实性SPN的独立预测因素(P<0.01)。恶性实性SPN的诊断模型为P=e^X/(1+e^X),X=3.346+(1.880×性别)+(3.311×分叶征)+(1.702×毛刺征)+(1.739×血管集束征)+(0.915×肿瘤家族史)。将B组患者相应数据代入回归方程,诊断实性SPN的准确率为85.4%(217/254),灵敏度为90.8%(118/130),特异度为79.8%(99/124),阳性预测值为82.5%(118/143),阴性预测值为89.2%(99/111)。结论数学诊断模型能较为准确地预测实性SPN的良、恶性,为临床提供参考。Objective To establish the mathematical diagnostic model of solid solitary pulmonary nodules(SPN).Methods A total of 396 cases with solid SPN(group A)found by chest high-resolution computed tomography(HRCT)and confirmed by surgical pathology was selected.Multiple factor logistic regression analysis was used to screen out the independent predictors of malignant solid SPN.The logistic regression mathematical diagnostic model to predict malignant solid SPN was established.Additionally,254 cases with solid SPN(group B)found by chest HRCT and confirmed by surgical pathology were applied to regression equation to verify the diagnostic efficiency.Results The age,gender,family history of tumors,air bronchus sign,vacuole sign,lobulation sign,spicule sign,vascular cluster syndrome and pleural depression in group A were different between the patients with benign solid SPN and those with malignant solid SPN(P<0.05).Multiple factor logistic regression analysis showed that female,family history of tumors,lobulation sign,spicule sign and vascular cluster syndrome were the independent predictors of malignant solid SPN(P<0.01).The diagnostic model of malignant solid SPN was P=e^x/(1+e^x),χ=3.346+(1.880×gender)+(3.311×sign of lobulation)+(1.702×spicule sign)+(1.739×vascular cluster syndrome)+(0.915×family history of tumors).The data of group B were applied to regression equation.The diagnostic accuracy was 85.4%(217/254),sensitivity was 90.8%(118/130),specificity was 79.8%(99/124),positive predictive value was 82.5%(118/143),and negative predictive value was 89.2%(99/111),respectively.Conclusion The mathematical diagnostic model can accurately assess the benign and malignant solid SPN and providea reference for clinical practice.

关 键 词：孤立性肺结节 数学诊断模型 

分 类 号：R563[医药卫生—呼吸系统]