Introduction and objectivesThyroid nodules frequently require ultrasound and Fine Needle Aspiration Cytology (FNAC) evaluation. However, FNA cytology does not allow differentiation between follicular adenoma and carcinoma on Bethesda type IV lesions. This situation leads to many unnecessary surgical procedures because it is not possible to assure the benignity of the lesions, even when most of the specimens correspond to adenomas or even other benign lesions.The objective is this study is to establish if there are any US characteristics that would help us to predict the risk of malignancy of nodules with a pathological diagnosis of follicular neoplasm in order to achieve a more conservative management for non-suspicious nodules. Material and methodsWe studied 61 nodules in 61 patients (51 women and 10 men) that underwent thyroid surgery and had histopathological results of either follicular adenoma or carcinoma.Different US characteristics of the nodules were analysed (composition, echogenicity, margin, calcification status, the presence of halo and overall observer suspicion of malignancy) and were correlated with the histopathological analysis. ResultsWe have found a statistically significant association between the presence of calcifications, ill-defined borders and overall observer suspicion or impression (defined by well-known suspicious for malignancy ultrasonographic features, such as calcification, poorly defined margin, and a markedly hypoechoic solid nodule; and benign ultrasonographic features, such as predominantly cystic echogenic composition and the presence of a perinodular hypoechogenic halo) with follicular carcinoma. However all those features have shown low sensitivities in the present study (30%, 30% and 50%, respectively). On the other hand, the absence of halo sign has shown a sensitivity of 100% and a negative predictive value (NPV) of 100% in our study. ConclusionsThe presence of calcifications, ill-defined borders and the overall impression or suspicion of malignancy associate with a higher risk for follicular carcinoma in Bethesda type IV thyroid nodules but their absence do not allow to predict benignity in these nodules. Inversely, when a halo sign lesion is observed, benign follicular neoplasm should be considered.