甲状腺结节的超声声像图多元回归分析

Ultrasonographic diagnosis of thyroid nodules by multiple Logistic regression

摘要: 目的:探讨超声多元参数对甲状腺良恶性结节判别的意义,以筛选鉴别结节良恶性的敏感指标。方法:对301个经病理确诊甲状腺结节的超声声像图参数进行多元逐步回归分析。结果:经逐步回归的多变量二分类Logistic回归分析,引入方程的4个超声检查指标包括结节是否伴有微小钙化、结节内部回声、包膜的完整性以及结节的边界。结论:以超声特征诊断甲状腺结节的Logistic回归模型有助于鉴别甲状腺良、恶性病变。

Abstract: Objective: To explore the significance of ultrasound multiple parameters in distinguishing benign and malignant thyroid nodules,and to screen the sensitive indexes for the differential diagnosis.Methods: Three hundreds and one thyroid nodules were enrolled in this study.The ultrasonograms were retrospectively studied with multiple stepwise regression analysis.Results: With stepwise regression analysis of multivariable binary logistic regression analysis,4 ultrasound evaluation indexes were enrolled:microcalcification,internal echo of nodule,integrity of capsule and nodule boundary.Conclusions: The Logistic model can help differentiate benign from malignant thyroid nodules.