高频超声及弹性成像对甲状腺结节性质鉴别诊断的logistic回归分析

发布时间：2018-05-02 18:34

本文选题：甲状腺结节 + logistic回归分析　；参考：《中国地方病防治杂志》2015年04期

【摘要】：目的采用二分类logistic回归分析,筛选对甲状腺结节良恶性有显著鉴别作用的超声指标并建立回归模型。方法应用高频超声对357例甲状腺结节进行检查(良性245例,恶性112例),观察其二维灰阶、彩色多普勒超声表现及弹性成像特点。以超声引导下甲状腺结节术后病理及穿刺活检的病理结果作为诊断标准,将结节的良恶性作为因变量,以超声表现作为自变量,采用二分类logistic回归分析筛选出对甲状腺结节良恶性有显著鉴别作用的超声指标并建立回归模型。结果经二分类logistic回归分析筛选进入回归模型且回归系数有统计学意义的有4个自变量:结节后方回声、结节内部回声、结节内钙化、弹性成像比值。回归模型诊断甲状腺结节性质的特异性93.88%,敏感性81.25%,准确率89.91%。ROC曲线下面积(area under curve,AUC)=0.946。结论采用二分类logistic回归分析能够筛选出对甲状腺结节良恶性鉴别有显著作用的超声表现,且回归模型对甲状腺结节的良恶性能做出准确的判断。
[Abstract]:Objective to establish a regression model of benign and malignant thyroid nodules by using two classification logistic regression analysis. Methods 357 cases of thyroid nodules (245 benign and 112 malignant) were examined by high frequency ultrasound. The features of two-dimensional gray scale, color Doppler ultrasonography and elastic imaging were observed. The pathological results of postoperative and puncture biopsy of thyroid nodules under the guidance of ultrasound were taken as diagnostic criteria, the benign and malignant nodules were regarded as dependent variables, and the ultrasonographic manifestations were used as independent variables. Two classification logistic regression analysis was used to screen the ultrasonic indexes which could distinguish benign and malignant thyroid nodules and to establish a regression model. Results four independent variables were selected by two-classification logistic regression analysis and the regression coefficients were statistically significant: echo behind the nodules, internal echo, calcification in the nodules, and the ratio of elastic imaging. The specificity of the regression model for the diagnosis of thyroid nodule was 93.88 and the sensitivity was 81.25. The accuracy rate of the area under the 89.91%.ROC curve was 0.946. Conclusion two classification logistic regression analysis can be used to screen the sonographic features of benign and malignant thyroid nodules, and the regression model can make an accurate diagnosis of benign and malignant thyroid nodules.
【作者单位】：一汽总医院;吉林大学中日联谊医院;
【基金】：吉林省科技发展计划项目(编号:20130206020SF)
【分类号】：R445.1;R581.3

【参考文献】