Ultrasound Radiomics Combined With ACR TI-RADS In Discriminating Benign And Malignant Thyroid Nodules

Objective: To compare the ability of ultrasound radiomics and ACR TI-RADS to predict benign and malignant TNs,and to explore the ability of ultrasound radiomics combined with ACR TI-RADS to predict benign and malignant TNs.Method: Three hundred and ninety-four TNs were enrolled in the training group,of which 150 nodules were randomly selected for the verification group.A Rad-score was developed from the training group.Binary logistic regression was used to establish models based on ACR TI-RADS,the Rad-score,and the Rad-score combined with ACR TI-RADS.The area under the receiver operating characteristic curve(AUC),De Long’s test and decision curve analysis(DCA)were used to identify the optimal model.Results: We finally selected four features to establish the Rad-score(Vertl＿RLNonUni,Vertl＿GLevNonU,WavEnLH＿s4,WavEnHL＿s5).De Long’s test and DCA showed that the Rad-score combined with ACR TI-RADS had the best performance(AUC = 0.913 [95% confidence interval {CI},0.881-0.939] and 0.899 [95% CI,0.840-0.942] in the training and verification groups,respectively),followed by ACR TI-RADS(AUC =0.898 [95% CI,0.863-0.926] and 0.870 [95% CI,0.806-0.919] in the training and verification groups,respectively),and the Rad-score(AUC =0.750 [95% CI,0.704-0.792] and 0.750 [95% CI,0.672-0.817] in the training and verification groups,respectively).Conclusions: The ability of the Rad-score combined with ACR TI-RADS to discriminate benign and malignant TNs is better than that of ACR TI-RADS or the Rad-score alone.The ultrasound Rad-score may play a potential role in improving the differentiation of malignant TNs from benign TNs.