Inter-exam agreement and diagnostic performance of the Korean thyroid imaging reporting and data system for thyroid nodule assessment: Real-time versus static ultrasonography

Objective To investigate the inter-exam agreement for thyroid nodule between real-time ultrasonography (US) assessment and retrospective US interpretation and to compare the diagnostic performance between two methods by using the most recently published guidelines for the US-based management of thyroid nodules, the Korean Thyroid Imaging Reporting and Data System (K-TIRADS). Methods The study included 253 nodules in 238 patients for the inter-exam agreement and 201 nodules in 190 patients for the diagnostic performance. Real-time and retrospectively static US images were analyzed according to the US descriptors and final categories of the K-TIRADS. Inter-exam agreements between real-time US assessments and static US image interpretation were analyzed, as was the diagnostic performance of both methods. Results Overall inter-exam agreements were almost perfect for orientation (κ = 0.868); and substantial for composition, spongiform appearance, echogenicity, shape, margin, calcification, and final K-TIRADS categories (κ = 0.754, 0.786, 0.747, 0.670, 0.666, 0.778, and 0.754, respectively). Specifically, moderate agreements were observed for predominantly cystic composition and ill-defined margin. The overall diagnostic performances of both real-time US assessment and retrospective US interpretation using the K-TIRADS were comparable. Conclusions Overall inter-exam agreements between real-time and retrospective US image interpretation for thyroid nodules using the K-TIRADS were equal or more than a substantial. Therefore, the use of K-TIRADS can provide consistent description and assessment for thyroid US regardless of the timing of interpretation.