Objective: To explore the diagnostic value of the proposed thyroid imaging reporting and data system (TI-RADS) classification in thyroid nodules, and to investigate interobserver variability among different observers using TI-RADS classification for ultrasound.
Method: We reviewed 667 thyroid nodules conventional ultrasound image data from 532 patients who were confirmed by Pathological diagnosis. Those ultrasound images were reviewed by 4 experienced sonographers who can independently assessed the sonographic characteristics and analyzed according to the TI-RADS classification. It is to assess the diagnostic sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of the TI-RADS classification. The multirater kappa statistics were used to assess the interobserver agreement among different observers.
Result: The overall sensitivity, specificity, negative predictive value (NPV) and Accuracy were 94%, 79%, 97% and 81%, respectively. Positive predictive values (PPVs) for categories 4 and 5 were 55% and 92%. There was substantial interobserver agreement for categories 3, 4B and 5 (Kappa = 0.62, 95% CI, 0.58-0.65), and was moderate agreement for category 4A (Kappa = 0.57) and 4B(Kappa = 0.60).
Conclusion: The diagnostic criteria of TI-RADS for differentiating between benign and malignant thyroid nodules have a high diagnostic value. There was substantial interobserver agreement in different experienced sonographers. The TI-RADS diagnostic criteria have a high diagnostic value and has great practical value in making a proper and further treatment plan.