乳腺良恶性结节超声特征的Logistic回归分析.pdfVIP

齐齐哈尔医学院学报2012年第33卷第18期 ·2449· 乳腺良恶性结节超声特征的Logistic回归分析 隋东奎 【摘要】 目的 应用Logistic回归模型对乳腺实性结节的超声特征进行分析，筛选出对其良恶性 鉴别诊断具有统计学意义的特征变量。方法 对95例乳腺结节患者共106个实性结节进行常规二维 超声成像和剪切波弹性成像，记录其常规超声声像图表现及弹性模量值特征，所有结节均经手术或穿刺 活检获得病理结果，建立Logistic回归模型，对各个特征指标进行分析评价。结果经过Logistic回归 分析，选取进入回归方程的特征变量有：形态、边界、RI、内部回声、后方回声衰减和弹性模量值六个指 标。结论Logistic回归分析能有效地筛选出对乳腺实性结节良恶有鉴别意义的指标。 【关键词】 超声 RI弹性模量 乳腺实性结节Logistic模型 ofultrasoniccharactersinbreastsolidnodulesSUI (Uhrason— Logisticregressionanalysis Dong—tmi． ic First P．R．China．) Department，thePeoplesHospital，Lianyungang，Jiangsu222002，the theultrasoniccharactersofbreast [Abstract]0bjectiveByLogisticregressionmodel，toanalyze solid outoftheirdifferentialof and characterswithstatisti— nodule．Screened diagnosisbenignmalignant variables．Methods95 with106breastsolidnodulesforconventionaltwo—di— tallysignificant patients mensionalultrasoundandshearwave the ofconvention— imaging elasticityimaging。recordedperformance al featuresandelasticmodulus．Allthenoduleswereconfirmed and sonographic bysurgicalpathologyand established modelto thevariouscharactersofindicators．ResultsThechar— Logisticregression analyze actersofselectedvariablesintothe regressionequation echo，RI， are：Morphology，boundary，internal rearechoandelasticmodulus．Conclusions filteroutthe Logistic can regressionanalysiseffectively and charactersofbreastsolid meaningfulbenignmalignant nodules．