Intratumoral and peritumoral radiomics based on ultrasound for the differentiation of follicular thyroid neoplasm.

Background: Although ultrasound (US) has been widely adopted as the preferred imaging modality for thyroid nodule evaluation, its reliability in distinguishing follicular adenomas from adenocarcinomas based on US features has been a subject of debate. The primary objective of our study was to comprehensively evaluate the efficacy of US-derived intratumoral and peritumoral radiomics in preoperatively differentiating follicular thyroid adenomas from adenocarcinomas, thereby contributing to the ongoing discussion regarding this challenging distinction.

Methods: In total, 195 patients who were pathologically diagnosed with thyroid follicular neoplasm were retrospectively enrolled in this study. Patients were randomly assigned to a training cohort and a test cohort in an 8:2 ratio to develop and evaluate the clinical model, intratumor-region model, peritumor-region model, and combined-region model. Radiomic features from both intratumoral and peritumoral regions were extracted from 2-dimensional (2D) US images, and we used the least absolute shrinkage and selection operator (LASSO) method for constructing the signature within the discovery dataset. Linear regression (LR) model was selected as the foundation for constructing both the radiomics and clinical signature. The prediction performance was evaluated by the area under receiver operating characteristic curve (AUC), sensitivity, and specificity. Decision curve analysis (DCA) was used to assess the clinical applicability of the models. Ultimately, a radiomics-clinical model was developed by integrating clinical information with radiomic features.

Results: A total of 19 radiomics features were selected to develop a radiomics model of intratumoral and peritumoral regions. Compared to the clinical model, the combined radiomics-clinical model showed higher diagnostic accuracy in distinguishing follicular thyroid carcinoma (FTC) in both the training set (AUC: 0.894 0.553) and the validation set (AUC: 0.884 0.540). A radiomics-clinical nomogram was constructed, and its clinical usefulness was validated through DCA.

Conclusions: The radiomics-clinical model that combined the intratumoral and peritumoral radiomics with clinical information had a high diagnostic performance for early identifications of FTC.