乳腺MRI特征及ADC值预测乳腺影像报告和数据系统4类良恶性病变

MRI descriptors and ADC value in prediction of benign and malignant breast lesions of breast imaging reporting and data system category 4

目的探讨乳腺MRI特征及ADC值对乳腺影像报告和数据系统(BI-RADS) 4类良恶性病变的预测能力,并尝试建立Logistic回归预测模型。方法收集MRI诊断为BI-RADS 4类病变、并取得病理结果的79例乳腺病变患者(82个病变)。采用单因素二元Logistic回归及两独立样本t检验分析各MRI特征和ADC值鉴别良恶性乳腺病变的统计学意义,并建立多因素Logistic回归预测模型,绘制ROC曲线评价回归模型预测BI-RADS 4类病变良恶性的效能。结果肿块型病变中,将边缘、内部强化及ADC值纳入Logistic回归预测模型中(P均<0.05,伪R^2=0.62),其诊断良恶性乳腺病变的ROC曲线AUC为0.981,敏感度为87.80%,特异度为100%。非肿块型病变中,无预测变量纳入建立Logistic回归预测模型(P均>0.1)。结论乳腺MRI特征(边缘、内部强化)及ADC值对预测肿块型BI-RADS 4类病变的良恶性具有一定意义;Logistic回归预测模型可有效鉴别BI-RADS 4类肿块型病变性质。

Objective To investigate the performance of MRI characteristics and ADC value in prediction of benign and malignant breast imaging reporting and data system (BI-RADS) category 4 lesions,and to establish Logistic regression predictive models. Methods Totally 79 patients with 82 BI-RADS 4 breast lesions confirmed with pathological results were enrolled.Univariate binary Logistic regression analysis and two-sample t -test were performed to analyze the difference of MRI characteristics and ADC values between benign and malignant breast lesions.The multivariate Logistic predictive model was established,and the ROC curve was drawn to evaluate the efficacy in prediction of benign and malignant lesions of BI-RADS 4. Results In mass lesions,the Logistic regression model was established based on margin,internal enhancement and ADC value (all P <0.05,Cox & Snell R 2 =0.62),with the AUC of ROC curve of 0.981,the sensitivity of 87.80% and the specificity of 100%.There was no statistically significant index in Logistic regression prediction model in non-mass lesions (all P >0.1). Conclusion Some MRI descriptors (margin and internal enhancement) and ADC value have a good predictive performance for benign and malignant mass lesions of BI-RADS 4.The established Logistic regression predictive model can effectively differentiate the character of BI-RADS 4 mass lesions and has potential clinical value.