乳腺影像报告数据系统（第5版）超声诊断指标的量化研究

发布时间：2018-07-03 13:04

  本文选题：乳腺疾病 + 超声检查　； 参考：《安徽医科大学》2017年硕士论文

【摘要】：目的:应用Logistic回归模型探讨乳腺影像报告数据系统(BI-RADS)中的超声诊断指标在乳腺肿块良、恶性鉴别中的应用价值,基于此建立乳腺超声影像报告数据系统分类(3~5类)评分系统,并对各乳腺肿块评分并分类,旨在为超声医师提供客观有效的评估分类标准。方法:回顾分析在安徽医科大学第二附属医院行乳腺超声检查的401例乳腺肿块超声图像特征,结合手术或穿刺活检病理结果,以第5版乳腺影像报告与数据系统病灶超声特征并加入年龄因素,建立Logistic回归模型;依据回归模型筛选结果及各因素的权重提出BI-RADS 3~5类评分分类标准。另纳入243例乳腺肿块作为测试组,对所有研究病例分别进行评分分类,以病理结果为金标准,计算每类的阳性预测值,与BI-RADS各类别的理论风险范围相比较,观察两者的一致性,评估该评分分类系统的诊断价值。结果:1、多因素回归分析显示最后进入模型的因素共6个:分别为年龄(≥40岁)、方位(不平行)、形态(不规则)、内部回声(不均匀)、边缘(不光整)、肿块内微钙化。2、依据回归模型筛选结果及各因素对肿块良、恶性贡献率不同,制定评分分类系统,BI-RADS 3、4a、4b、4c、5类相对应的分值为6分、7~8分、9~15分,16~22分,≥23分。3、模型病例综合评分的BI-RADS 3~5类的阳性预测值分别为0%、2.67%、16.18%、90.74%、100%;测试病例综合评分的BI-RADS 3~5类的阳性预测值分别为0%、4.17%、21.43%、84.85%、100%;所有研究病例综合评分的BI-RADS 3~5类的阳性预测值为0%、3.25%、17.70%、88.51%、100%。4、测试病例ROC曲线下分析,曲线下面积为0.948。以14分作为乳腺肿块良恶性的Cut-off值(诊断界点),其灵敏度为90.10%、特异度90.14%、约登指数0.80、准确率90.12%。结论:利用多因素建立的乳腺肿块Logistic回归模型并以此制定的BI-RADS评分分类系统,能够对乳腺肿块BI-RADS系统进行客观的分类,为临床评价乳腺肿块的良、恶性提供有效参考依据。
[Abstract]:Objective: to explore the value of ultrasound diagnostic indexes in breast image reporting data system (BI-RADS) in differentiating benign and malignant breast masses with Logistic regression model. The purpose of this paper is to provide an objective and effective criteria for the evaluation and classification of breast masses. Methods: the ultrasonographic features of 401 cases of breast masses performed in the second affiliated Hospital of Anhui Medical University were analyzed retrospectively, combined with the pathological results of surgical or puncture biopsy. A Logistic regression model was established based on the ultrasonic features of breast image report and data system in the 5th edition and the age factors were added, and the BI-RADS 3 / 5 classification criteria were proposed according to the results of the regression model and the weight of each factor. In addition, 243 cases of breast masses were included as test group. All the cases were graded and classified separately. The positive predictive value of each group was calculated according to the pathological results as gold standard, and compared with BI-RADS, the consistency of the two groups was observed. To evaluate the diagnostic value of the scoring classification system. Results the multivariate regression analysis showed that there were six factors that entered the model: age (鈮，

本文编号：2093745