高频超声预测甲状腺微小癌的临床研究

Clinical study of high frequency ultrasound in predicting thyroid microcarcinoma

目的探讨甲状腺微小癌的恶性超声征象,评估其对甲状腺微小癌的诊断价值。方法回顾性分析筛选经病理结果证实的197个结节(其中恶性126个、良性71个),通过Logistic回归分析筛选出独立恶性征象,建立Logistic回归方程。结果最后,进入Logistic回归模型的五个指标分别是:微小结节的形态、内部结构、纵横比、钙化类型及结节与甲状腺被膜的关系,其ROC曲线下面积为0.972 6。以预测概率P=0.5作为阈值,其预测准确率为92.39%。结论形态不规则、实性、纵横比≥1、结节与甲状腺被膜关系模糊、微钙化是甲状腺微小癌的独立危险因素,由上述五个指标建立的Logistic回归模型有助于鉴别甲状腺微小结节的良恶性。

Objective To explore the malignant ultrasound signs of thyroid microcarcinoma(TMC)and evaluate its diagnostic value for TMC.Methods Retrospective analysis was performed to screen 197 nodules confirmed by pathological results(including 126 malignant and benign 71).Logistic regression analysis was used to screen out independent malignant signs and logistic regression equation was established.Results The five indicators that entered the logistic regression model were:morphology,internal structure,aspect ratio,calcification type,and relationship between nodules and thyroid capsules.The area under the ROC curve was 0.9726.With the prediction probability P=0.5 as the threshold,the prediction accuracy is 92.39% .Conclusions Irregularity,solidity,aspect ratio≥1,the relationship between nodules and thyroid tunica,and microcalcification are independent risk factors for TMC.A Logistic regression models established by the above five indicators are helpful to identify TMC which is malignant or benign.