28论著影像研究与医学应用2024年4月第8卷第7期二元逻辑回归分析在MSCT鉴别诊断机化性肺炎及周围型肺癌中的价值魏微微，邹佳妮（通信作者）（中部战区总医院放射科湖北武汉437000）ThevalueofbinarylogisticregressionanalysisinthedifferentialdiagnosisoforganizedpneumoniaandperipherallungcancerusingMSCTWEIWeiwei,ZOUJiani(Correspondingauthor)DepartmentofRadiology,CentralTheaterGeneralHospital,Wuhan,Hubei437000,China【Abstract】ObjectiveToexplorethevalueofbinarylogisticregressionanalysisinthedifferentialdiagnosisoffocalorganizingpneumonia(FOP)andperipherallungcancer(PLC)usingmulti-slicespiralCT(MSCT).Methods53patientswithfocalorganizingpneumoniaconfirmedbypathologywereselectedasthestudygroup,and61patientswithperipherallungcancerwereselectedasthecontrolgroup.TheCTsignsofthetwogroupswereanalyzedandcompared,andaregressionmodelwasconstructedbyusingstatisticallysignificantfeatures.Theareaunderthecurve,sensitivity,andspecificitywerecalculatedbyreceiveroperatingcharacteristic(ROC)curves.ResultsBinarymultivariatelogisticregressionanalysisshowedthatunclearboundaries,longthorn,shortthorn,andairbronchogramwereindependentpredictivefactorsfordistinguishingFOPfromPLC;TheareaundertheROCcurveofthemodelis0.945,withasensitivityof79.20%andaspecificityof98.40%.ConclusionUnclearboundaries,longthorn,shortthorn,andairbronchogramsignsarehelpfulforthedifferentialdiagnosisofFOPandPLC.Thepredictivemodelcanimprovethediagnosticefficiencyandprovideimportantauxiliaryinformationforclinicalpractice.【Keywords】Multi-slicespiralCT;Focalorganizingpneumonia;Peripherallungcancer;Sensitivity;Specificity【摘要】目的：探讨二元逻辑回归分析在多层螺旋CT（MSCT）鉴别诊断局灶性机化性肺炎（FOP）及周围型肺癌（PLC）中的价值。方法：选取2016年3月—2022年5月中部战区总医院经病理证实的53例局灶性机化性肺炎患者作为研究组，61例周围型肺癌患者作为对照组，分析比较两组病变CT征象，对差异有统计学意义的定性征象构建回归模型，通过受试者工作特征（ROC）曲线计算曲线下面积、灵敏度、特异度。结果：二元多因素Logistic回归分析显示边界不清、长毛刺、短毛刺...
