e17588 Background: A univocal ultrasound (US) classification system for characterising thyroid nodules is still lacking. The aim was to create an algorithm (TMC-RSS) using US features in combination with Color Doppler (CD), TIRADS, elastography (ES) and test its diagnostic performance. Methods: Analysis of prospectively collected data over 10 months. All studies were performed on single equipment using standard US protocol with TIRADS scoring. Gold standard was pathology. 340 nodules in 260 patients of which 316 nodules included for final analysis. 148 (46.8%) benign and 168 (53.2%) Malignant. TMC-RSS Multiple Logistic Regression Model. Malignant characteristics:Plus (+) 3 points for: ES score 3/4, malignant nodes. (+) 1 point for: Taller than wider, microcalcification, hypoechogenicity, solid composition, ill-defined margins, central +/- peripheral vascularity. (+) 0.5 point for: irregular halo, size > 1 cm. Benign characteristics:Minus (-) 3 points for: Purely cystic, ES score-1; (-) 1 point for: spongiform, comet tail artefact, complete halo. (-) 0.5 point for: peripheral vascularity. Final TMC-RSS Score calculated by addition of all the points. Results: Mean age was 47.3 yrs. Mean size was 2.3 +/- 1.5 cm. Diagnostic performance of combined US, ES, TIRADS and CD (sensitivity: 96%, specificity: 95%, PPV: 95%, NPV: 96%, kappa: 0.911) was significantly higher (p < .001) than either combination of US, TIRADS, ES, CD. On multivariate analysis all US features except taller than wider were independent predictor of malignancy (p < .01). TMC-RSS Model had 90% sensitivity, 89% specificity and 91% accuracy for characterising nodules. On the ROC curve the cut-off for best performance of TMC-RSS score was 5.75. The cumulative risk of malignancy based on TMC-RSS score was 2.4% for score < 3, 18% for score ≥ 3 - < 6 and 80% for score ≥ 6. Conclusions: The proposed TMC-RSS is easy-to-use, robust, reproducible and provides higher degree of confidence to a non-expert radiologist. It can be the first step in standardization of reporting lexicon to allow effective communication between the radiologists, pathologists, and clinicians; helping avoid unnecessary interventions.
