Résumé
Objective To evaluate the clinical value of ultrasonography in diagnosis of benignPHand malignant thyroid nodules by binary Logistic regression model.Methods A retrospective analysis of 240 cases of thyroid nodules confirmed by pathology after operation excision was conducted.173 nodules were benign as control group,67 nodules were malignant as observation group.The ultrasonic features of thyroid nodules were collected.The factor had statistics significance by x2 test between the two groups were analyzed by binary Logistic regression.A logistic regression model was created.The ROC curve was drawn and the area under the curve was calculated.Results There were statistically significant differences among boundary,aspect ratio,shape,internal echo,calcification,acoustic halo,cervical lymphadenectasis between the control group and observation group(x2=45.392,7.590,30.039,24.168,37.406,6.893,16.078,all P<0.01).And after Logistic regression analysis,six variables that entered the regression equation were boundary(P=0.000,OR=8.437),aspect ratio(P=0.000,OR=12.816),microcalcification of the nodules(P=0.000,OR=8.893),shape(P=0.000,OR=8.791),internal echoes(P=0.001,OR=8.313)and cervical lymphadenectasis(P=0.001,OR=6.891).The accuracy of the model was 90.7%(223/246)in predicting malignant nodules in thyroid.The area under the ROC curve was(0.904±0.031).Conclusion The binary Logistic regression can be used to differentiate malignant and benign thyroid nodules,and make a more accurate judgment for the nodules in thyroid.