高频彩超及弹性成像对甲状腺结节良恶性的鉴别诊断价值

目的：对甲状腺小结节高频彩超及弹性成像表现进行多因素 Logistic 回归分析，探讨其对良恶性鉴别诊断的价值。方法：回顾性分析经手术病理确诊的126个直径≤2 cm 的甲状腺结节的超声表现，根据每个结节的高频超声及弹性成像声像图特征进行分类，采用二分类多因素 Logistic 回归分析，筛选出评价甲状腺恶性结节的因素，建立概率方程。结果：根据多因素 logistic 回归分析结果，边界不清、纵横比≥1、微钙化、极低回声、弹性成像评分是鉴别甲状腺小结节良恶性的关键因素。回归模型预测超声诊断甲状腺结节良恶性的符合率为89．7％，其中良性结节的诊断符合率为90．0％，恶性结节的诊断符合率为89．3％，ROC 曲线下面积为0．959。结论：多因素 Logistic 回归模型可筛选出对甲状腺小结节良恶性有鉴别诊断意义的特征性变量，综合评价各个变量，有利于甲状腺结节良恶性的鉴别诊断。

Differential diagnosis of benign and malignant thyroid nodules using high-frequency color doppler ultrasonography with elas-tosonography

Objective:The aim of the present study was to investigate the high frequency ultrasound and elastography charateristcs in terms of the differential diagnosis of thyroid nodules by use of multivariate logistic regression analysis and to assess the application value of these features.Methods:Sonographic features of 126 thyroid nodules (≤20mm)confirmed by pathology were analyzed retrospectively.High frequency ultrasonic and elastography features of each nodule were classified and then analyzed with multivariate logistic regression analysis.Factors of malignant nodules were screened to establish a probability equation.Results:In light of the results of multivariate logistic regression analysis,five features with statistical significance,including obscure boundary,A/T ratio>1,micro calcification,very low echo and elasticity score,were entered into the logistic stepwise regression model,presenting as the key factors to differentiate benign and malignant nodules.An overall diagnostic accuracy of 89.7% was obtained using this model,with 90.0% benign and 89.3% malignant nodules,re-spectively.The area under the ROC curve was 0.959.Conclusions:The model of multivariate logistic regression analysis can select the valuable variables in the diagnosis of small thyroid nodules.Comprehensive evaluation of each variable facilitates the differentiation of benignancy or malignancy.