摘要
目的：探究全自动乳腺超声冠状面影像的诊断价值。方法选取2010年6月至2012年12月来本院就诊的病例1100例，根据BI-RADS第5版超声术语和相关文献选取影像特征并进行筛选，采用Logistic回归分别以筛选后横切面和横切面结合冠状面影像特征与病理诊断结果建立诊断模型。结果单因素分析显示肿物回声、形状、方向、边界、钙化、后方回声、冠状面汇聚征、冠状面边界差异均有统计学意义。在仅有横切面因素时所建诊断模型AUC为0.862（95%CI：0.838～0.886），而纳入冠状面特征因素后则提高至0.910（95%CI：0.892～0.928），经过Z检验证实差异有统计学意义（P〈0.001）。结论加入冠状面影像特征后的诊断模型诊断效率提高，说明冠状面影像特征可以在判断病灶良恶性中提供额外信息。
Objective To explore the diagnostic value of coronal imaging of three dimensional breast ultrasound. Methods 1 100 women patients from June 2010 to December 2012 admitted into our hospital were included in the study. Based on the fifth edition of BI-RADS ultrasound lexicon and related literature,features of imaging in both transverse and coronal planes were selected and their statistical significance was assessed. Diagnostic models were established using Logistic regression analysis as one consisted only of features on the transverse plane and the other on both planes. Results AUC of the diagnostic model with transverse imaging features was 0.862（95%CI：0.838-0.886）and was improved to 0.910（95%CI：0.892-0.928）after addition of coronal features. The difference between the two AUCs analyzed using Z test and the P value was less than 0.001. Conclusion Characteristics in the coronal plane help to improve diagnostic efficiency with providing additional information for differentiating benign from malignant lesions.
出处
《中华普通外科学文献（电子版）》
2016年第3期195-199,共5页
Chinese Archives of General Surgery(Electronic Edition)
关键词
乳腺肿瘤
超声检查
模型
结构
Breast neoplasms
Ultrasonography
Model,structural
