Endocrine Abstracts

The purpose of the study was to evaluate the diagnostic capabilities of systems for assessing ultrasound thyroid imagines. The object of the study was 754 patients with thyroid nodules. Research methods: ultrasound in the framework of image evaluation systems TIRADS, EU-TIRADS, ARC-TI-RADS, BTA, TI-RADS K. For statistical analysis of systems and criteria was performed multidimensional statistical modeling using cluster, factor, discriminant, variance, and ROC analysis. The study found that the systems have different sensitivity and specificity. Data have shown that there are more differences between systems than there are similarities. The reason for this is that there are no uniform rules for the formation of systems. The authors of the systems arbitrarily determined a set of features, their combinations, staging, accounting or not accounting for the size of thyroid nodes. According to the ROC analysis for thyroid cancer, the sensitivity of TIRADS was 91%, specificity-91%, diagnostic efficiency-91%, prognostic value of a positive result-0.92, the likelihood ratio for a positive result-10.6, the likelihood ratio for a negative result-0.03. When detecting thyroid adenomas, TIRADS sensitivity was 62%, specificity-82%, diagnostic efficiency-79%. When detecting goiter, TIRADS sensitivity was 86%, specificity-83%, diagnostic efficiency-79%. ROC analysis found that AUC = 0.972 ± 0.00 484. AUC is an integral measure of diagnostic effectiveness, it can be argued that TIRADS has good predictive capabilities in thyroid cancer, good in goiter, and insufficient effectiveness in adenomas. When comparing the systems, it turned out that the ACR-TI-RADS system, despite the decrease in the number of signs, has the highest prognostic indicators for thyroid cancer (AUC = 0.964) and slightly worse for thyroid adenomas and nodular goiter. The Korean TI-RADS-K system is a test of excellent quality (AUC = 0.959) with good prognostic indicators for various nodular thyroid pathology. In terms of specificity (97%), TI-RADS-K is superior to all other systems. When evaluating EU-TIRADS, it is established that the system is a test of high quality (AUC = 0.826), but has a predictive capability worse than TIRADS. The BTA system has the lowest prognostic indicators of all systems (AUC = 0.823). With Multidimensional Scaling, TIRADS (coordinates: 0.80 043; 0.58 114) and ACR-TI-RADS (coordinates: 0.58 114; 0.52 5725) were close. The Correlation Analysis of a Two-Way Table confirmed the influence of empirical ranking on the results of system evaluations. Gradations related to thyroid cancer have a large diagnostic weight, and the ranking of benign pathology introduces an element of instability in the system, without significantly affecting the final result-the selection of patients for fine-needle aspiration biopsy.