Qualitative Diagnostic Value Of CT Radiomics For Thyroid Nodules In TI-RADS Category 4A And 4B

PART 1CT radiomics for discriminating benign and malignant thyroid nodules in TI-RADS category 4A and 4BObjective:To evaluate the diagnostic value of CT radiomics model for distinguishing benign from malignant thyroid nodules of thyroid imaging reporting and data system(TI-RADS)category 4A and 4B.Methods:This retrospective study included 660 cases with TI-RADS 4A and 4B nodules(328 benign,332 malignant)undergoing preoperative CT and surgery from January 2018 to June 2022.We extracted radiomics features from CT images to build plain scan,arterial phase,venous phase,delayed phase and multi-phase combined(plain scan + arterial phase + venous phase + delayed phase)radiomics model,integrated the most effective radiomics model and the clinical features to construct the synthetic diagnostic model and draw nomogram.The model performance was determined by the area under the receiver operating characteristic curve(AUC).The clinical usefulness was assessed by decision curve analysis(DCA).The TI-RADS 4A and 4B subgroups were hierarchically analyzed.Results: The diagnostic performance of multi-phase combined radiomics model(training set AUC vs.test set AUC : 0.828 vs.0.745)is better than that of single phase radiomics model(plain scan: 0.752 vs.0.698;arterial phase : 0.822 vs.0.708;venous phase : 0.803 vs.0.702;delayed phase : 0.792 vs.0.708);compared with the clinical model(training set AUC: 0.672;test set AUC: 0.815)and multi-phase combined radiomics model,the synthetic diagnostic model had better diagnosis effects in both the training set(AUC: 0.857)and the test set(AUC: 0.861).The decision curve analysis showed that the combine model had higher clinical application value.Moreover,the synthetic diagnostic model can also effectively distinguish benign and malignant thyroid micro-nodules in the TI-RADS 4A and 4B subgroups.Conclusions: CT radiomic model had good discrimination for the preoperative prediction of the benign from malignant thyroid nodules in TI-RADS category 4A and4B.PART 2Contrast-enhanced CT-based radiomics for distinguishing benign from malignant thyroid micro-nodules in TI-RADS category 4A and 4BObjective:To evaluate the diagnostic value of contrast-enhanced CT-based radiomics model for distinguishing benign from malignant thyroid micro-nodules of thyroid imaging reporting and data system(TI-RADS)category 4A and 4B.Methods:This retrospective study included 300 patients with TI-RADS 4A and 4B micro-nodules(115 benign,185 malignant)undergoing preoperative contrast-enhanced CT and surgery from January 2018 to January 2022.We extracted radiomics features from contrast-enhanced CT images to build arterial phase,venous phase,delayed phase and multi-phase combined(arterial phase + venous phase +delayed phase)radiomics model,integrated the most effective radiomics model and the clinical features to construct the synthetic diagnostic model and draw nomogram.The model performance was determined by the area under the receiver operating characteristic curve(AUC).The clinical usefulness was assessed by decision curve analysis(DCA).The TI-RADS 4A and 4B subgroups were hierarchically analyzed.Results: The diagnostic performance of multi-phase combined radiomics model(training set AUC vs.test set AUC : 0.814 vs.0.718)is better than that of single phase radiomics model(arterial phase : 0.730 vs.0.601;venous phase : 0.794 vs.0.859;delayed phase : 0.793 vs.0.622);compared with the clinical model(training set AUC:0.732;test set AUC: 0.766)and multi-phase combined radiomics model,the synthetic diagnostic model had better diagnosis effects in both the training set(AUC: 0.876)and the test set(AUC: 0.813).The decision curve analysis showed that the combine model had higher clinical application value.Moreover,the synthetic diagnostic model can also effectively distinguish benign and malignant thyroid micro-nodules in the TI-RADS 4A and 4B subgroups.Conclusions: Contrast-Enhanced CT-based radiomic model had good discrimination for the preoperative prediction of the benign from malignant thyroid micro-nodules in TI-RADS category 4A and 4B.