Application of the binary regression mode in analyzing the ultrasonographic features of solid thyroid micronodules / 中华超声影像学杂志

ABSTRACT

Objective To apply the binary Logistic regression model and evaluate the value of grey scale ultrasonography(US),ultrasond elastography(UE) and contrast-enhanced ultrasound(CEUS).Methods US,UE and CEUS were performed in 130 patients with 142 solid thyroid micronodules (the largest diameter less than 10 mm).Pathology was followed up as golden diagnosis criteria.A Logistic model on the basis of ultrasonographic features was obtained.A receiver operating characteristic(ROC) curve was used to assess the performance of Logistic model.Results Four ultrasonic indexes were finally entered into the Logistic regression model including microcalcification,acpect ratio,advanced CEUS mode and elasticity score.The percentage correction of prediction was 92.1％ (129/142),the area under the ROC curve was 0.944.Conclusions The multivariate analysis model of binary Logistic regression can describe and analyze the process of differential diagnosis of malignant and benign solid thyroid micronodules by ultrasonography,and select out the valuable indexes for differential diagnosis.