欢迎来到天天文库
浏览记录
ID:31997202
大小:26.28 MB
页数:37页
时间:2019-01-30
《乳腺实性肿块超声特征的logistic回归分析》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、logit(P)=一9.464+1.12IXl2+3.683Xi+4.315X3“.593Xio一20.407X7+34.429X13对上述模型行似然比检验,具有统计学意义(#=l85.822,P<0.001o利用该回归模型预报这163个乳腺病灶,回归模型能很好地区分这两类病灶,如果在回归值JP大于O.5时预报为恶性,小于等于0.5时预报为良性,则预报正确率高达159/163=97.5%。以模型判断乳腺恶性病灶概率预测值绘$IJRoc曲线,ROC曲线下面积(areaundertheROCcurve,
2、Az)为判断指标检验模型拟合效果,Az为0.996±0.001,尸<0.001,95%置信区间为(O.994,0.998),证明该模型的拟合效果较好,用于预测乳腺肿瘤的良恶性效果也很好。结论1、Logistic回归分析从15个诊断指标中筛选出改良弹性评分、形态、RI、内部回声、后方回声增强、汇聚征,构建的回归模型能很好地区分乳腺良恶性病灶,方程拟合的优度和预测效果比较理想。2、改良弹性评分和形态可作为鉴别诊断乳腺肿瘤良恶性最具价值的项目,在临床工作中,可以首先从这两个项目来评判乳腺肿块的良恶性,再结
3、合其他特征综合分析。3、Logistic回归模型使综合分析各诊断指标更简易、客观,有较好的临床应用价值。Logistic回归模型可将其作为计算机辅助诊断系统的基础。但用它对单个病例进行预报,需考虑人群发病率,必须进一步对模型进行多地域、多中心的前瞻性研究和检验。关键词:超声,乳腺,鉴别诊断,Logistic回归分析一IV—巾山人学硕十学位论文乳腺实性肿块超声特征的Logistic回归分析TheLogisticRegressionAnalysisofUltrasonographicFeatur‘’So
4、lidBreasteaturesInolidBreastMassesJⅥMajor:RadiologyandNuclearMedicineName:ZengJieSupervisor:ProfessorLuoBao—MingABSTRACTWiththeincreaseoftheincidencerateofbreastcancer,detection,diagnosisandtreatmentinearlystagearethekeyforimprovingthesurvivalrateofbre
5、astmalignanttumor.Becauseoftheharmlessness、safety、repeatabilityandcheapness,ultrasonicexaminations,includingconventionaltwodimensionalgrayscaleultrasonography(2D—us),colorDopplerflowimaging(CDFl),ultrasonicelastography(UE)andthreedimensionalultrasonogr
6、aphy(3D—us)arewidelyusedinthediagnosisofbreastdiseases.Thedevelopingultrasonictechnologyprovidesmoreandmorediagnosticinformationofbreastdiseases.Itisthetrendthattogetandcomprehensivelyanalyzethemostimportantdiagnosticinformation.Inrecentyears,thebinary
7、Logisticregressionanalysisiswidelyusedineveryfieldofmedicalresearch,suchasthecase-controlstudyofepidemiologyandetiology,theclinicdiagnosticdiscriminantmodel,theevaluationofcurativeeffect,etc.TheLogisticregressionmodelhasalotofadvantagesinthemultiplevar
8、iableanalysisofetiology.Itcanexactlygivetheprobabilityofthedatawhicharenotbelongedtonormaldistribution,forexample,deathorrecovery(calledbinary-valuedvariables).TheLogisticregressionanalysisWasappliedinultrasoundresearchofbreast.But—V一th
此文档下载收益归作者所有