超声造影定性与定量分析鉴别乳腺肿块的对比研究

Comparison study between qualitative analysis and quantitative analysis of contrast-enhanced ultrasound to differential diagnosis of breast masses

目的比较超声造影定性和定量分析鉴别乳腺肿块的价值。方法回顾性分析已取得病理诊断的73例乳腺肿块的超声造影定性指标（采用目测法获得）和定量指标（根据时间一强度曲线获得），行单因素和多因素分析，用Logistic回归模型产生方程，并生成新的预测值pre-1和pre-2，以此绘制出两个ROC曲线，评价两个ROC曲线下面积。结果超声造影定性指标产生的回归方程中包含增强特征、径线扩大、达峰强度分级三项指标，其对乳腺良恶性肿块的预测准确率为91．8％；而定量指标产生的回归方程中仅有相对达峰强度一项指标，其对乳腺良恶性肿块的预测准确率仅为61．5％。2个新预测值所对应的ROC曲线下面积分别为91．3％与75．7％，经Z检验，两者差异有统计学意义（P〈0．05）。结论超声造影定性分析对乳腺肿块的鉴别能力优于定量分析。

Objective To assess the diagnosis ability of contrast-enhanced ultrasound （CEUS） between qualitative analysis and quantitative analysis in breast masses. Methods The contrast-enhanced ultrasound （CEUS） imagings of 73 cases breast masses （41 malignant cases,32 benign cases） were analysed retrospectively,including qualitative indexes（obtained by eyes-watching method） and quantitative indexes （obtained by time-intensity curve）. The relationship between these CEUS signs and pathology were evaluated by single variety and multiple variety analysis. The two logistic models on the basis of CEUS signs were obtained. According to the logistic regression results to create new variants pre-1 and pre-2, the two receiver operating characteristic （ROC） curves were created to assess the performance of the logistic models. Results By qualitative analysis, the lesion enhanced features,lesion diameter expanding or not and the grades of lesion enhancement at the peak time entered the logistic regression. And the prediction accuracy to differential diagnosis the breast masses was 91.8% and the area under the ROC was 91.30//oo. However,by quantitative analysis, only the relative peak intensity entered the regression, the prediction accuracy was 61.5% and the area under the ROC was 75.7%. According to Z test, the areas difference under the two ROCs had statistically significant （ P〈0.05）. Conclusions To differential diagnosis the breast masses by CEUS,the qualitative indexes have more useful than the quantitative indexes.