摘要
目的建立以超声特征诊断乳腺恶性病变的Logistic回归模型。方法对手术病理证实的475个乳腺病灶的二维超声、彩色多普勒超声、超声弹性成像特征进行多因素回归分析,建立Logistic模型。用ROC曲线法评价Logistic模型的预报能力。结果多因素回归分析显示最后进入Logistic模型的7个特征分别为:弹性成像评分(OR=5.735)、病灶边缘(OR=8.421)、病灶内微小钙化,(OR=8.755)、CDFI分级(OR=1.767)、内部回声不均匀(OR=0.276)、后方回声衰减(OR=3.282)、后方回声增强(OR=0.396)。Logistic模型的ROC曲线下面积为0.979。结论以超声特征诊断乳腺恶性病变的Logistic回归模型有助于鉴别乳腺良、恶性病变。
Objective To establish a logistic model for predicting breast malignancy on the basis of ultrasonographic features. Methods The features of gray-scale ultrasonography （US）, color Doppler flow imaging （CDFI） and ultrasonic elastography （UE） were evaluated in 475 breast lesions confirmed by surgical pathology. A Logistic model for predicting breast malignancy on the basis of ultrasonographic features was obtained. A receiver operating characteristic （ROC） curve was used to assess the performance of the Logistic model. Results Seven ultrasonic features were finally entering the Logistic model, they were： elasticity score （odds ratio, OR= 5. 735）, margin （OR= 8. 421）, microcalcification （OR=8. 755）, color Doppler flow grade （OR=1. 767） ,heterogeneous texture （OR=0. 276）, shadowing （OR=3. 282）, and enhanced transmission （OR =0. 396）. The area under the ROC curve was 0. 979. Conclusion The Logistic model can help differentiate benign from malignant breast lesions.
出处
《中国医学影像技术》
CSCD
北大核心
2007年第1期88-90,共3页
Chinese Journal of Medical Imaging Technology
作者简介
沈建红（1971-），女，江苏人，硕士，主治医师（现在广州中医药大学附属第二医院工作）．E-mail：xzyezi@yahoo．com．cn
[通讯作者]罗葆明，中山大学附属第二医院超声科，510120．E-mail：bmluo2005@126．com
