超声与核素显像对甲状腺单发结节良恶性鉴别诊断价值的Logistic回归分析

The Logistic regression analysis of ultrasonographic and radioisotope scanning features in thyroid solitary nodules

目的:建立以甲状腺单发结节超声及核素静态显像特征为变量的Logistic回归模型,比较超声与核素静态显像在甲状腺单发结节良恶性鉴别诊断中的价值。方法:经手术治疗的98例甲状腺单发结节的患者术前均进行二维超声成像及甲状腺核素静态显像,以病理诊断为金标准,将Logistic回归模型引入对甲状腺单发结节良恶性判断中,建立Logistic回归模型。绘制ROC曲线,评价Logistic回归模型的预报能力,通过比较各变量的优势比,评价超声与核素显像在甲状腺单发结节良恶性鉴别诊断中的价值。结果:经过基于偏最大似然估计的前进法Logistic回归分析,筛选出钙化、回声评分、形态及核素扫描四个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,超声的三个特征变量的优势比均大于核素的优势比。Logistic回归模型对甲状腺单发结节良恶性预报的正确率为94.9%,ROC曲线下面积为0.980。结论:在甲状腺单发结节良恶性鉴别诊断中超声相对核素静态显像具有更高的诊断价值。

Objective To apply the binary Logistic regression model and evaluate ultrasonographic and radioisotope scanning features of thyroid solitary nodules in the differential diagnosis of thyroid solitary nodules.Methods The two-dimensional（2D） ultrasonography and radioisotope scanning were performed in 98 patients with thyroid solitary nodules.Pathology was made as golden standard.A Logistic model was obtained on the basis of ultrasonographic and radioisotope scanning features.A receiver operator characteristic curve（ROC） was constructed to assess the performance of the Logistic model.We also evaluated the value of 2D ultrasonography and radioisotope scanning in the differential diagnosis of thyroid solitary nodules.Results Three ultrasonographic features and radioisotope scanning result were finally entered into the Logistic model;they were shape,calcification、heterogeneous texture and radioisotope scanning result.And the odds ratio of the three ultrasonographic features were higher than the radioisotope scanning.The percentage of correct prediction was 94.9% respectively.The area under ROC curve was 0.980.Conclusions The binary Logistic regression can select out the valuable indexes in the differential diagnosis of thyroid solitary nodules.It can improve the diagnosis accuracy by comprehensively analysing the ultrasonographic features and radionuclide imaging of thyroid solitary nodules.Ultrasonography plays a more important role than radioisotope imaging in the differential diagnosis of thyroid solitary nodules.