腮腺多形性腺瘤及腺淋巴瘤的多因素Logistic回归分析  

Multi-factor Logistic regression analysis of parotid pleomorphic adenoma and adenolymphoma

作　　者：李毓红[1] 彭建春[2] 代月黎[2] 

机构地区：[1]南华大学附属第一医院超声科,湖南衡阳421001 [2]南华大学附属第二医院放射科,湖南衡阳421001

出　　处：《中国医学影像技术》2016年第5期713-716,共4页Chinese Journal of Medical Imaging Technology

摘　　要：目的观察腮腺多形性腺瘤及腺淋巴瘤的声像图特征,构建Logistics回归模型鉴别腮腺多形性腺瘤及腺淋巴瘤。方法回顾性分析第一组患者的超声图像,建立多因素Logistic回归模型,以第二组患者资料验证模型的可行性。结果经多分类Logistic回归分析,年龄、边界、后方回声3个自变量有助于鉴别多形性腺瘤,以P=0.5为分界,ROC曲线下面积为0.826±0.030(P<0.001)。性别、年龄、无回声区、边界四个自变量有助于鉴别腺淋巴瘤,以P=0.5为分界,ROC曲线下面积为0.969±0.014(P<0.001)。模型预测多形性腺瘤的准确率为56.48%(48/85),敏感度为88.00%(22/25),预测腺淋巴瘤的准确率为58.82%(50/85),敏感度为100%(19/19)。结论 Logistic回归筛选自变量,建立简便、有效的回归模型,对预测腮腺多形性腺瘤及腺淋巴瘤有良好的临床应用价值。Objective To construct the multi-factor Logistic regression model and obtain the differentiat method through analyzing the ultrasonographic features of the parotid pleomorphic adenoma and adenolymphoma. Methods The Logistic regression models was established through reviewing the sonographic features of the first group, imaging data of the second group were analyzed prospectively to verify the usefulness of the Logistic regression models. Results The paramenters named ＂age＂, ＂boundary＂, ＂posterior acoustic enhancement＂ helped to identify pleomorphic adenoma. Taking P= 0.5 as the boundary, the area under the ROC curve was 0. 826 ± 0. 030 （P〈0. 001））. Similarly, the four independent variables named ＂gender＂, ＂age＂,＂ anechoic area＂, ＂boundary＂ helped to identify adenolymphoma. The area under the ROC curve was 0. 969±0. 014 （P〈0. 001））. The accuracy of the model in forecasting pleomorphic adenoma was 56.48% （48/85）, the sensitivity was 88.00% （22/25） the accuracy of the prediction adenolymphoma was 58.82% （50/85）, the sensitivity was 100% （19/19）. Conclusion Logistic regression model have good value of clinical applications to identify parotid pleo- morphic adenoma and adenolymphoma through screening variables.

关 键 词：腺瘤 多形性 腺淋巴瘤 超声检查 回归分析 

分 类 号：R739.8[医药卫生—肿瘤] R445.1[医药卫生—临床医学]