Abstract

The main focus of this research is to investigate how effectively a radiomics model based on computed tomography (CT) imaging can discriminate between thyroid cancer and Hashimoto's thyroiditis in patients. The ultimate objective is to introduce fresh approaches and numerical benchmarks that enable precise identification of thyroid cancer even in the presence of Hashimoto's thyroiditis.A grand total of 376 eligible cases were gathered between August 2016 and August 2022.The cases were separated into a group for training purposes and another group for testing purposes. Two radiologists outlined regions of interest in the images and extracted radiomic features. After screening the features, the most statistically significant ones were chosen to create a tumor discrimination model that operates automatically. The evaluation of the model's performance involved analyzing the receiver operator characteristic (ROC) curve, as well as measuring accuracy, sensitivity, and specificity.After an exhaustive analysis, a total of 1316 radiomic characteristics were initially extracted. Among them, 11 optimal radiomic features were further identified through the application of mRMR and LASSO techniques for rigorous selection. The area under the curve (AUC) for the training set was found to be 0.911 ± 0.036 [with a 95% confidence interval of (0.845, 0.942), and a p-value less than 0.001]. The corresponding accuracy, sensitivity, and specificity were determined to be 0.897, 0.889, and 0.862, respectively.In conclusion, this study demonstrates the remarkable diagnostic accuracy of radiomics using CT imaging when differentiating thyroid cancer in the context of Hashimoto's thyroiditis. The diagnostic performance of this approach surpasses that of both ultrasound and subjective evaluations conducted by radiologists, underscoring its substantial practical utility in distinguishing thyroid cancer alongside Hashimoto's thyroiditis.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call