| Objective:(1)To explore the value of conventional ultrasound(US),virtual touch tissue imaging quantification(VTIQ)and contrast-enhanced ultrasound(CEUS)in the differential diagnosis of benign and malignant nodules in Thyroid Imaging Reporting and Data System(TI-RADS);(2)Comparison of diagnostic performance with TI-RADS classification by constructing a nomogram prediction model for quantitative diagnosis of multimodal ultrasound image features.Methods: A total of 145 patients with 164 thyroid nodules diagnosed with TI-RADS category 4 from October 2021 to October 2022 at our hospital were retrospectively collected,and all nodules were confirmed to be benign or malignant by surgery or fine needle aspiration biopsy.The patients were randomly divided into training and validation groups according to 7:3.In the training group,US,VTIQ and CEUS were screened for risk factors associated with malignant thyroid nodules by univariate analysis,and US,VTIQ,CEUS,US+VTIQ,US+CEUS and multimodal ultrasound(US+VTIQ+CEUS)prediction models were constructed and validated by multivariate Logistic regression analysis,respectively.The differential diagnostic efficacy of each model for the differential diagnosis of the 4 types of thyroid TIRADS nodules was assessed by comparing the area under curve(AUC)of the receiver operating characteristic curve(ROC)between models,and the model with the best diagnostic efficacy was used to construct the nomogram.The AUC and calibration curves were used to verify the discrimination and consistency of the nomogram model,and the decision curves were used to assess the clinical applicability of the nomogram models.The AUC of thyroid nodules diagnosed by nomogram model and TI-RADS classification was compared,and the consistency between the predicted results of nomogram model and TIRADS classification and pathological results was tested by Kappa test.Results:(1)The results of univariate analysis showed that the differences in US features(aspect ratio,microcalcifications,margins,extraperitoneal invasion),VTIQ parameters(SWVmax,SWVmin,SWVratio1,SWVmean),CEUS qualitative features(enhancement temporal phase,enhancement degree,enhancement uniformity,post-enhancement border)and CEUS quantitative parameters(TTP,PI)were statistically significant between benign and malignant thyroid nodules.(2)Multimodal ultrasound features multivariate Logistic regression analysis showed that poorly defined margins,presence of microcalcifications,extraperitoneal invasion,high SWVmean,high TTP,and low PI were independent risk factors for malignant thyroid nodules(OR of 4.908,4.987,6.818,3.570,4.612,0.397,respectively,all P < 0.05).In the training group,the AUC of multimodal ultrasound(US+VTIQ+CEUS)model in diagnosing benign and malignant thyroid TI-RADS 4 nodules was higher than that of US,VTIQ,CEUS,US+VTIQ and US+CEUS model(0.938 vs 0.790,0.938 vs 0.803,0.938 vs 0.832,0.938 vs 0.894,0.938 vs 0.897,all P < 0.05),which showed that multimodal ultrasound model has the best diagnostic efficiency;(3)A multimodal ultrasound nomogram prediction model was constructed,the AUC,sensitivity,specificity,positive predictive value and negative predictive value of the nomogram model in the training group were0.938,0.972,0.791,0.886 and 0.944 respectively,and they were 0.977,0.970,0.938,0.970 and 0.934 respectively in the verification group.The calibration curve and decision curve suggested better model consistency and clinical applicability,respectively.(4)The AUC of the multimodal ultrasound nomogram model for identifying malignant thyroid nodules was higher than the TI-RADS classification in both the training and validation groups(0.938 vs 0.845 in the training group and 0.977 vs 0.848 in the validation group,all P < 0.05).The consistency test showed that the nomogram prediction model was more consistent with pathological outcomes(Kappa: 0.789 in the training group,0.907 in the validation group,all P < 0.05)than the TI-RADS classification(Kappa: 0.639 in the training group,0.600 in the validation group,all P <0.05).Conclusion: Multimodal ultrasound has a high differential diagnostic value for benign and malignant thyroid TI-RADS class 4 nodules,and the multimodal ultrasound nomogram model has better diagnostic efficacy than the TI-RADS classification,and VTIQ and CEUS techniques can be used as a supplement to conventional ultrasound to provide more diagnostic information and provide a reference basis for clinicians’ treatment decisions. |