Diagnostic Accuracy and Procedural Efficiency of ACR TI-RADS, EU-TIRADS, and K-TIRADS in ≤ 10 mm Thyroid Nodules Using Bethesda Cytology: A Cohort of 998 Nodules

OBJECTIVES: To compare the diagnostic performance and procedural efficiency of American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS), European Thyroid Imaging Reporting and Data System (EU-TIRADS), and Korean Thyroid Imaging Reporting and Data System (K-TIRADS) in ≤ 10 mm nodules using Bethesda cytology as the reference standard. METHODS: This retrospective study included 998 nodules ≤10 mm undergoing ultrasound-guided fine-needle aspiration biopsy (FNAB). Primary analyses included Bethesda II (negative) and Bethesda V-VI (positive) nodules excluding Bethesda I/III/IV, yielding 604 nodules (540 negative; 64 positive). Categories 1-3 were considered low risk and 4-5 high risk under a harmonized threshold Diagnostic performance and classification patterns were compared. RESULTS: K-TIRADS showed the numerically highest sensitivity (93.8%) and negative predictive value (NPV) (97.9%), although sensitivity did not differ significantly between systems, with lower specificity (35.4%) and accuracy (41.6%).EU-TIRADS showed intermediate performans (sensitivity 68.8%, specificity 73.7%, accuracy 73.2%). ACR TI-RADS achieved the highest specificity (83.3%) and accuracy (81.0%) with lower sensitivity (60.9%). Low-risk classification rates were 78.6% (ACR), 69.2% (EU), and 32.3% (K) of nodules; false-positive high-risk classifications among Bethesda II nodules were 16.7%, 26.3%, and 64.6%, respectively. CONCLUSION: In ≤ 10 mm nodules, the three systems demonstrated distinct diagnostic and classification profiles under a harmonized TR4-5 threshold, with different trade-offs between sensitivity and specificity. ADVANCES IN KNOWLEDGE: Applying a harmonized TR4-5 threshold in ≤ 10 mm nodules reveals substantially differences in classification patterns among the three systems, demonstrating that they are not interchangeable under a uniform risk threshold.