论文部分内容阅读
目的:建立基于超声征象预测卵巢良恶性肿瘤的Logistic回归模型,并探讨该预测模型在鉴别诊断卵巢良恶性肿瘤中的应用价值.方法:回顾性收集重庆市渝北区人民医院2013年1月至2016年3月经手术病理证实的189例卵巢肿瘤患者,根据病理结果分为良性组(120 例)和恶性组(69 例),比较两组彩色多普勒超声各项指标特征,以病理诊断作为金标准,建立Logistic回归预测模型,计算预测模型准确率、灵敏度、特异度等指标,绘制ROC曲线并计算曲线下面积.结果:单因素及多因素 Logistic 回归分析结果显示:形态(OR =7. 149)、内部回声( OR =7. 085)、血流(OR=8. 908)、RI(OR=13. 224)是卵巢良恶性肿瘤鉴别诊断的主要超声影像特征指标. Logistic回归模型对卵巢良恶性肿瘤的预测正确率为93. 7% (177／189),灵敏度92. 5% (111／120),特异度95. 7% (66／69),阳性预测价值97. 4% (111／114),阴性预测价值88. 0% (66／75). ROC曲线下面积为0. 945 ± 0. 019,P<0. 001,95% CI:0. 910~0. 976.结论:基于超声征象的Logistic预测模型对于鉴别卵巢良恶性肿瘤具有较高的价值,可用于指导临床实践.“,”Objective: To establish a Logistic regression model to predict benign and malignant ovarian tumors based on ultrasound images, and evaluate the value of Logistic model in the differential diagnosis of benign and malignant ovarian tumors. Methods: We retrospectively selected 189 cases of ovarian tumors at The People’s Hospital of Yubei District in Chongqing from January 2013 to March 2016. Among them, 69 cases was malignant and 120 cases was benign. Ultrasound features of benign and malignant ovarian tumors were compared. With pathologic diagnosis as gold standard, a Logistic model was established to calculate the accuracy, sensitivity, specificity and other indicators of the prediction model. A receiver op-erating characteristic (ROC) curve was drawn to calculate the area under the curve. Results: Univariate and multivariate Logistic regression analysis showed that morphology (OR=7. 149), internal echo (OR=7. 085), blood flow (OR=8. 908) and RI (OR=13. 224) were the main ultrasonographic features in differential diagnosis of benign and malignant ovarian tumors. The accuracy, sensitivity, specificity, positive predictive value and negative predictive value of Logistic regression model were 93. 7% (177／189), 92. 5% (111／120), 95. 7% (66／69), 97. 4% (111／114) and 88% (66／75), respec- tively. The area under the ROC curve was 0. 945 ± 0. 019 ( P<0. 001, 95% CI: 0. 910 ~0. 976). Conclusion: Logistic model based on ultrasound features for the differential diagno- sis of benign and malignant ovarian tumors is highly valuable, and can be used to guide clinical practice.