Logistic回归模型评价超声造影在甲状腺单发结节中的鉴别诊断价值  

Evaluation of Contrast-enhanced Ultrasound in Differentiation Diagnosis of Thyroid Solitary Nodule Using Logistic Regression

作　　者：许永波[1] 姜珏[1] 周琦[1] 余姗姗[1] 魏亚娟[1] 乞艳华[1] 雷小莹[1] 

机构地区：[1]西安交通大学第二附属医院超声研究室,西安市710004

出　　处：《中国超声医学杂志》2013年第4期299-301,共3页Chinese Journal of Ultrasound in Medicine

摘　　要：目的建立以甲状腺单发结节超声诊断特征为变量的回归模型,评价常规超声及超声造影在甲状腺单发结节良恶性鉴别诊断中的价值。方法对129例甲状腺单发结节行常规超声及超声造影检查,以病理诊断为金标准建立Logistic回归模型。绘制ROC曲线,评价Logistic回归模型的预报能力,通过比较各变量的似然比,评价超声造影在甲状腺单发结节良恶性鉴别诊断中的价值。结果运用二分类Logistic回归分析,筛选出对甲状腺单发结节良恶性鉴别诊断中有统计学意义的特征变量,包括钙化、增强程度、达峰时间及最大灌注强度。Logistic回归模型对甲状腺单发结节良恶性预报的正确率为90.70%,敏感性为89.09%,特异性为91.89%。结论超声造影较常规超声更有助于甲状腺单发结节良恶性的鉴别诊断。Objective To evaluate the value of conventional ultrasound and contrast-enhanced ultrasound in the diagnosis of thyroid solitary nodule by applying the binary Logistic regression model. Methods 129 cases of thyroid solitary nodule were examed by conventional ultrasound and contrast-enhanced ultrasound,which were confirmed by surgical pathology. A Logistic model was obtained on the basis of ultrasonographic features. A receiver operator charac- teris（ROC） curve was constructed to assess the performance of the Logistic model. Evaluate the differential diagnosis value of contrast-enhanced ultrasound in thyroid solitary nodule. Results Four uhrasonographic features including cal- cification, enhancement degree,the time to peak and biggest perfusion strength were finally entered into the Logistic model. The percentage of correct prediction,the sensitivity and specificity were 90.07%,89.09% and 91.89% ,respec- tively. Conclusions Contrast-enhanced ultrasound plays more important role in the differential diagnosis of thyroid solitary nodule.

关 键 词：甲状腺 超声造影 LOGISTIC回归分析 ROC曲线 

分 类 号：R581[医药卫生—内分泌] R445.1[医药卫生—内科学]