BI-RADS4~5类乳腺小肿块多模态超声特征研究

Multimodal Ultrasound Characteristics of Small Breast Masses in Breast BI-RADS Category 4 to 5

目的:联合使用常规超声、超声弹性成像及超声造影建立并验证乳腺BI-RADS4~5类乳腺小肿块(≤2cm)的良恶性预测模型。方法:采集200例经病理证实的乳腺小肿块的超声特征,并进行Logistic回归分析,建立多模态超声诊断预测模型。另选59例乳腺肿块,验证预测模型并绘制ROC曲线。结果:多因素回归分析显示最后进入Logistic模型的7个特征分别为患者年龄(≥40岁),肿块的形状、方向、微钙化、弹性应变比、增强后病灶范围增大、增强后局部灌注缺损。结论:多模态超声在预测乳腺BI-RADS4~5类良恶性病灶方面显示出良好的诊断性能。

Objective:Combined conventional ultrasound,ultrasound elastography and contrast-enhanced ultrasound were used to establish and validate apredictive model for the benignity and malignancy of small breast masses(≤2cm)in breast BI-RADS category 4 to 5.Methods:Ultrasound characteristics of 200 pathologically confirmed small breast masses were collected and subjected to Logistic regression analysis to establish amultimodal ultrasound diagnostic prediction model.A total of 59 patients with breast masses were also selected to validate the prediction model and plot ROC curves.Results:Multi-factor regression analysis showed that the last 7 features that entered the Logistic model were patient age(≥40years),shape andorientation of the mass,microcalcification,elastic strain ratio,increased lesion extent after enhancement,and local perfusion defect after enhancement.Conclusion:Multimodal ultrasound showed good diagnostic performance inpredicting benign and malignant lesions in breast BI-RADS category 4 to 5.