Application Of Logistic Regression Analysis In Differential Diagnosis Of Thyroid Solid Nodule

Objective Constructing the logistic regression model with thyroid solid nodules in multiple ultrasonographic features to evaluate the application value of combined with conventional ultrasound (US), contrast enhanced ultrasound (CEUS) and real-time tissue ultrasound elastography (RTE) in the differential diagnosis of thyroid solid nodules.Methods Retrospective analysis of 199 thyroid solid nodules were performed by US, CEUS and RTE. Contributing the logistic regression model were treated ultrasonographic features as independent variables and postoperative pathological results of benign and malignant as dependent variable. To evaluate the prediction ability of the logistic regression model by drawing ROC curve, and then to calculate the area under the curve (AUC).Results Four malignant ultrasonographic features were finally selected out by the logistic regression analysis which were the internal hypoecho, microcalcification, the early hypoenhancement and high elastic strain score (P <0.05), which were significant in differential diagnosis of thyroid solid nodules. The accuracy of predicting the benign and malignant nodules was 84.4%, AUC was 0.839.Conclusion The combined with CEUS and RTE based on US has a practical value in the differential diagnosis of thyroid solid nodules.