Endocrine Abstracts

Thyroid nodules (TN) are the most common endocrine disorder present in 10 – 67% of adults, with a malignancy risk of 3 – 15%. Sonoelastography (SE) is a novel ultrasound technique providing information on elasticity of TN. Decreased elasticity proved to be a useful predictor of malignancy. SE cannot be regarded as a substitute for fine-needle aspiration biopsy (FNAB), but should rather be considered as a complementary tool facilitating estimation of TN malignancy risk and selection of the region for FNAB. SE is particularly useful as a rule-out technique, thanks to its high negative predictive value, thus might help to reduce the number of FNABs. However, certain limitations of SE must be considered during interpretation of the results (TN size and localization, follicular lesions, calcified or cystic TN, multinodular goiter, subacute/acute thyroiditis, obesity, acromegaly, operator-dependence, lack of standardized method for data reporting and clearly defined cut-off values). The use of semi-quantitative methods is recommended. The TIRADS classification system corrected by SE is more accurate in TN diagnosis. Recently computer-aided diagnosis (CAD) systems based on artificial intelligence algorithms allowing for an automated ultrasound image analysis have been developed to obtain accurate, reproducible and more objective diagnosis of TN. In clinical settings CAD performed similarly or worse than experienced sonographists, but still markedly better than a physician with basic ultrasound skills. It may assist in decision-making and has a potential to be used as a screening device for less-experienced physicians. Future improvements of the technique would increase its diagnostic efficiency. Our research demonstrated that CAD characterizes with high sensitivity and good specificity. The EU-TIRADS classification used by an experienced sonographist allows to identify suspected lesions with high sensitivity but rather low specificity. The best diagnostic performance in estimation of malignancy risk of TN was obtained for the combined model of CAD and EU-TIRADS scale.