应用Logistic回归模型评价剪切波弹性成像在甲状腺结节鉴别诊断中的价值

Evaluation of shear wave elastography in discriminating thyroid nodules using Logistic regression

目的 建立以甲状腺结节超声诊断特征为变量的Logistic回归模型,评价常规超声及剪切波弹性成像在甲状腺结节良恶性鉴别诊断中的价值.方法 对在我院接受甲状腺手术或甲状腺结节活检的193例患者进行常规超声和剪切波弹性成像检查.以手术病理或活检病理为金标准.结果 纳入研究的235个结节中,157个为良性结节,78个为恶性结节.1 57个良性结节的平均杨氏模量为（28.7±16.7） kPa,78个恶性结节的平均杨氏模量为（66.8±44.6）kPa（P=0.000）.杨氏模量值诊断结节良恶性的ROC曲线下面积为0.802,最优诊断界点为39.2 kPa.利用Logistic回归分析评价各个超声特征,筛选出回声类型、纵横比、微小钙化及杨氏模量等4个在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,其中杨氏模量的优势比（6.211）远高于其他变量.结论 剪切波弹性成像测得的杨氏模量值能定量评价结节硬度,较常规超声检查更有助于甲状腺结节的良恶性诊断.

Objective To contribute the binary Logistic regression model to evaluate conventional ultrasound and shear wave elastography （SWE） on differentiating malignant thyroid nodules.Methods A total of 193 patients scheduled to undergo core-biopsy or surgery of thyroid nodule were enrolled.Both conventional ultrasound and SWE were performed,which were then correlated with pathological results.Results There were 157 benign thyroid nodules and 78 malignant thyroid nodules.The mean of SWE value of benign nodules was （28.7 ± 16.7）kPa,and it was （66.8 ± 44.6）kPa in malignant nodules （P =0.000）.The area under the operator characteristic curve （AUC） was 0.802,with the optimal cut-off value of 39.2 kPa.Four ultrasound characteristics finally entering the multivariate Logistic model were echogenicity,aspect ratio,microcalcification and SWE values.The odd ratio of SWE value （6.211） was higher than all the conventional ultrasound characteristics.Conclusions SWE provides quantitative information about the tissue stiffness,which plays a more important role in discriminating malignant thyroid nodules than conventional ultrasound characteristics.