用Logistic回归分析法对直径＜2cm孤立肺癌结节CT征象的回顾性分析

摘要

目的 运用L o g istic回归分析法回顾性分析直径＜2 cm肺部孤立结节的CT征象对早期肺癌的诊断意义.方法 选取156例病理证实的直径＜2 c m的肺部小结节病例进行回顾性分析,对各C T征象进行统计学赋值,并进行二分类L o g istic回归分析筛选出可能是肺癌结节的CT征象并分析其为早期肺癌的危险度.结果 6个CT征象进入L o g istic方程模型,依次为"磨玻璃征"(8.12)、"分叶征(6.72)"、"血管纠集征"(6.02)、"毛刺征"(5.07)、"空洞坏死"(3.41)、空泡征(1.02).结论 "磨玻璃征"是肺癌结节危险度最高的征象,以CT征象建立的L o g istic模型对鉴别孤立肺癌小结节有较大的参考意义.%Objective ToanalyzethecharacteristicmanifestationsandearlydiagnosticvalueofCTinsolitarypulmonarynodules (SPNs)lessthan2cmusing L o g istic regressionanalysis.Methods 156patientswithSPNlessthan2cmconfirmedbypathology werecollected.Statisticalassignment was performed and binary L o g istic regression wasimplemented for CT manifestations.Those features,whichmightbesignsoflungcancer,wereextractedfromtheCTimagesandtheirrisklevelswerealsoanalyzed.Results SixCTsignsincluding "ground-glasssign"(8.12),"lobulationsign"(6.72),"vascularconvergencesign"(6.02),"spiculesign"(5.07),"necrosis and cavitation "(3.41 ),and "vacuole sign (1.02 )" were enrolled in the L o g istic equation.Conclusion "Ground-glass sign "is associated with the highest risk level for lung cancer nodules.T he L o g istic m odel constructed fro m C T m anifestations is helpful for identifyingsolitarylungcancernodules.