ABSTRACT
BACKGROUND: Ultrasonography (US) is the method of choice for evaluating thyroid nodules. In 2017, the American College of Radiology (ACR) created a classification system based on US characteristics. For the system to be adopted, it must be reproducible. OBJECTIVES: To determine the intraobserver and interobserver variability of the ACR TI-RADS. METHODS: Cross-sectional study; three radiologists with different levels of experience used the ACR TI-RADS to classify 100 nodules on two occasions one month apart, and we calculated the intraobserver and interobserver variability. RESULTS: Regarding intraobserver variability, the first radiologist had nearly perfect concordance for composition, echogenicity, shape, and margins and substantial concordance for echogenic foci; the second radiologist had nearly perfect concordance for composition, echogenicity, shape, and margins and substantial concordance for echogenic foci, and the third radiologist had nearly perfect concordance for composition, echogenicity, and shape and substantial concordance for margins and echogenic foci. The interobserver concordance was calculated for the two readings; the concordance was substantial except for shape in the first reading and for echogenicity and margins in the second reading, which had moderate concordance. CONCLUSIONS: The ACR TI-RADS classification system is reproducible.
Subject(s)
Thyroid Nodule , Cross-Sectional Studies , Humans , Observer Variation , Retrospective Studies , UltrasonographyABSTRACT
BACKGROUND: Ultrasonography (US) is the method of choice for evaluating thyroid nodules. In 2017, the American College of Radiology (ACR) created a classification system based on US characteristics. For the system to be adopted, it must be reproducible. OBJECTIVES: To determine the intraobserver and interobserver variability of the ACR TI-RADS. METHODS: Cross-sectional study; three radiologists with different levels of experience used the ACR TI-RADS to classify 100 nodules on two occasions one month apart, and we calculated the intraobserver and interobserver variability. RESULTS: Regarding intraobserver variability, the first radiologist had nearly perfect concordance for composition, echogenicity, shape, and margins and substantial concordance for echogenic foci; the second radiologist had nearly perfect concordance for composition, echogenicity, shape, and margins and substantial concordance for echogenic foci, and the third radiologist had nearly perfect concordance for composition, echogenicity, and shape and substantial concordance for margins and echogenic foci. The interobserver concordance was calculated for the two readings; the concordance was substantial except for shape in the first reading and for echogenicity and margins in the second reading, which had moderate concordance. CONCLUSIONS: The ACR TI-RADS classification system is reproducible.
