Objective To obtain the elasticity value of solid breast lesions with supersonic shear wave elastrography (SWE) and apply the binary Logistic regression in order to evaluate the value of SWE in differential diagnosis of benign and malignant breast lesions. Methods SWE quantitative elastography was preformed in 91 breast lesions of 91 patients in Zhenghai Longsai Hospital to obtain the maximum and mean elasticity value (Emax, Emean). And receiver operating characteristic (ROC) curves were used to assess the diagnostic performance. A Logistic regression for the gray scale ultrasound and the elastic modulus was conducted with multiple variables including Emax, Emean, border, echo, form, calcification. Results Pathological examination showed 73 benign lesions and 18 malignant lesions. Emax and Emean of malignant lesions were obviously higher than those of benign lesions [(99.73±41.15) kPa vs (38.59±14.28) kPa, (61.45±24.88) kPa vs (23.46±11.44) kPa, t=-15.05,-14.12, both P=0.000]. The area under the ROC curve of Emax and Emean were 0.932 and 0.915. Taking 63.70 kPa as the threshold of Emax, the sensitivity was 77.8%and the speciifcity was 97.3%. Then taking 44.22 kPa as the threshold of Emean, the sensitivity was 83.3%and the speciifcity was 94.5%. The results of Logistic regression analysis showed:the 3 most effective variables were Emax, border of the lesions and Emean. Conclusions The multivariate analysis model of binary Logistic regression can select the valuable indexes of differential diagnosis of benign and malignant breast lesions. SWE plays an important role in differentiating benign and malignant lesions and it is valuable in clinical practice.%目的应用Logistic回归模型评价剪切波弹性成像技术在乳腺病灶良恶性鉴别诊断中的应用价值。方法对2012年3-12月浙江省宁波市镇海龙赛医院经手术病理证实的91例患者共91个乳腺病灶进行剪切波弹性成像检查，并测量其杨氏模量值（最大值Emax及平均值Emean）。以手术病理诊断结果作为金标准，绘制操作者工作特性（ROC）曲线，得出乳腺良、恶性病灶Emax、Emean的诊断临界值；并结合二维灰阶超声检查，建立Logistic回归模型评价Emax、Emean、边缘、内部回声、形态及内部钙化等6个因素对乳腺病灶良恶性的预测效果。结果91个病灶中，良性病灶73个，恶性病灶18个。乳腺良性病灶的Emax、Emean分别为（23.46±11.44）、（38.59±14.28） kPa，乳腺恶性病灶的Emax、Emean分别为（61.45±24.88）、（99.73±41.15）kPa；乳腺恶性病灶的Emax及Emean均大于乳腺良性病灶，且差异均有统计学意义（t值分别为-15.05、-14.12，P均为0.00）。Emax的曲线下面积为0.932，大于Emean的曲线下面积0.915，分别以63.70、44.22 kPa作为乳腺良恶性病灶Emax、Emean的诊断临界值，敏感度和特异度分别为83.3%和94.5%、77.8%和97.3%。Logistic回归分析结果显示，Emax对乳腺病灶良恶性的预测效果最优，其次是病灶边缘，再次是Emean。结论二分类Logistic回归模型筛选出对乳腺病灶良恶性鉴别诊断有意义的特征变量。剪切波弹性成像技术较其他弹性成像技术更有助于鉴别乳腺病灶良恶性，为鉴别乳腺病灶良恶性提供了一种新的方法。
展开▼
