This study developed and validated a combined CT-based radiomics model to differentiate between benign and malignant thyroid nodules. Using data from 378 patients and 408 resected nodules, the model incorporated patient age, three morphologic CT features, and 28 radiomic features from noncontrast, arterial, and venous CT phases. In external validation, this combined model demonstrated superior performance with an AUC of 0.923, significantly outperforming other models. The findings suggest that this model could improve the management of thyroid nodules detected on CT.
0