Logistic Regression Analysis Of Papillary Thyroid Microcarcinoma With Multi-parameter Ultrasound Diagnosis

Objective: To screen out the independent malignant sonographic features of papillary thyroid microcarcinoma(PTMC)and to assess their value in the diagnosis of PTMC.Methods: The gray-scale ultrasound,color Doppler flow imaging and shear wave elastography Young's modulus of 145 thyroid micronodules were retrospected which included 70 benign nodules and 75 malignant nodules by surgical pathology or fine needle aspiration pathology.A Logistic model was established for predicting PTMC and making ROC curve.Results: The sonographic characteristics of echogenic foci type,internal echo,nodule composition,boundary and taller-than-wide were selected into the Logistic regression model.If apply P=0.5 as cut-off,the diagnostic accuracy rate reaches 95.2%,and the area under the ROC curve is 0.981.Conclusion: Punctate echogenic foci,mixed echogenic foci,very hypoechoic,solidity,irregular border and taller-than-wide?1 are the independent malignant ultrasound indicators for PTMC.Certainly,the Logistic regression model can benefit the Ultrasound diagnosis of PTMC.