Evaluation of ultrasound elastography in diagnosis of thyroid small nodules using binary logistic regression / 中华超声影像学杂志

ABSTRACT

Objective To select sonogram features for the differential diagnosis of benign and malignant thyroid small nodules by Logistics regression analysis,and to contribute the binary Logistic regression model of sonogram features as independent variable and evaluate the value of conventional ultrasonography and ultrasound elastography (UE) in the differential diagnosis of benign and malignant thyroid small nodules.Methods 166 thyroid nodules (≤ 10 mm) in 140 patients were reviewed and analyzed by 2D ultrasound,color Doppler flow imaging and UE.A Logistic model was obtained based on pathology as golden diagnosis criteria.The odds ratio of variables in the equation were compared to assess various variables,especially the efficacy of elastography in the diagnosis.Results Four statistically significant features were finally entering the Logistic stepwise regression model,including shape,calcification,the internal component of nodules and elasticity score.And the odds ratio of the elasticity score was higher than other features.Conclusions The analysis of binary Logistic regression can select the valuable variables for the diagnosis of pathological nature of thyroid small nodules.UE has much more dominances than other features.The combinated application of UE and 2D ultrasonic features plays a great clinical role in the final diagnosis of thyroid small nodules.