Logistic回归模型评价自动乳腺全容积扫描联合常规超声诊断乳腺影像报告及数据系统3～5类结节  

Logistic regression model in evaluation on diagnosis of breast nodules classification of breast imaging reporting and data system3-5category with automated breast volume scanner combined with conventional ultrasound

作　　者：李静敏 邵玉红[1] 孙秀明[1] 张惠[1] 李鹏[1] 王彬[1] LI Jingmin;SHAO Yuhong;SUN Xiuming;ZHANG Hui;LI Peng;WANG Bin(Department of Ultrasound,Peking University First Hospital,Beijing 100034,China)

机构地区：[1]北京大学第一医院超声科,北京100034

出　　处：《中国介入影像与治疗学》2018年第5期282-285,共4页Chinese Journal of Interventional Imaging and Therapy

摘　　要：目的探讨多因素Logistic回归分析评价自动乳腺全容积扫描(ABVS)联合常规超声鉴别诊断乳腺影像报告及数据系统(BI-RADS)3～5类结节良恶性的价值。方法 216例患者(247个BI-RADS 3～5类乳腺结节)接受常规超声及ABVS检查。以穿刺活检或术后病理为金标准,建立多因素Logistic回归模型,筛选出诊断乳腺恶性肿瘤的主要超声特征,并评价其鉴别诊断乳腺BI-RADS 3～5类结节的价值。绘制ROC曲线,评价Logistic回归模型的诊断效能。结果多因素Logistic回归筛选出4个诊断BI-RADS 3～5类结节良恶性的独立危险因素,为"皱缩征"[X_1,优势比(OR)=12.03,P<0.01)、"微分叶征"(X_2,OR=6.00,P<0.01)、结节边界(X_6,OR=11.48,P=0.01)和纵横比(X_8,OR=4.09,P=0.01),其中"皱缩征"的OR最大。回归方程为Logit(P)=-4.43+2.49 X_1+1.79 X_2+2.44 X_6+1.41 X_8(χ~2=196.32,P<0.01)。Logistic回归模型预测乳腺恶性结节的ROC曲线下面积为0.95(P<0.01)。结论 Logistic回归模型鉴别诊断BI-RADS 3～5类良恶性结节有较高价值。Objective To explore the value of Logisticregression model in evaluation on automated breast volume scanner（ABVS）combined with conventional ultrasound in differential diagnosis of breast imaging reporting and data system（BIRADS）3—5 category breast nodules.Methods Totally 216 patients（247 BI-RADS 3—5 category breast nodules）underwent ABVS and conventional ultrasound.Taking puncture biopsy or postoperative pathology as the gold standards,Logistic regression model was established and main ultrasound characteristics in diagnosis of breast malignant nodules of BI-RADS 3—5 category were screened and evaluated.Finally,ROC curve of evaluating the diagnostic efficiency of Logistic regression model was drawn.Results There were 4 independent variables in diagnosis of breast malignant nodules of BIRADS 3—5 category,including ＂ Retraction phenomenon＂（X1,odds ratio [OR]=12.03,P〈0.01）, ＂ microlobulated margin＂（X2,OR=6.00,P〈0.01）,border（X6,OR=11.48,P=0.01）and anteroposterior to transverse ratio（A/T,X8,OR=4.09,P=0.01）.OR of ＂retraction phenomenon＂ was the highest.The regression equation was Logit（P）=-4.43＋2.49 X1＋1.79 X2＋2.44 X6＋1.41 X8（χ2=196.32,P〈0.01）,and the area under the ROC curve was 0.95（P〈0.01）.Conclusion Logistic regression model has great value in differential diagnosis of breast malignant nodules of BIRADS 3-5 category.

关 键 词：超声检查 乳腺肿瘤 LOGISTIC回归 

分 类 号：R737.9[医药卫生—肿瘤] R445.2[医药卫生—临床医学]