Abstract
OBJECTIVE: To evaluate possible variability in chest radiologists' interpretations of the Lung Imaging Reporting and Data System (Lung-RADS) on difficult-to-classify scenarios. MATERIALS AND METHODS: Ten scenarios of difficult-to-classify imaginary lung nodules were prepared as an online survey that targeted Korean Society of Thoracic Radiology members. In each question, a description was provided of the size, consistency, and interval change (new or growing) of a lung nodule observed using annual repeat computed tomography, and the respondent was instructed to choose one answer from five choices: category 2, 3, 4A, or 4B, or "un-categorizable." Consensus answers were established by members of the Korean Imaging Study Group for Lung Cancer. RESULTS: Of the 420 answers from 42 respondents (excluding multiple submissions), 310 (73.8%) agreed with the consensus answers; eleven (26.2%) respondents agreed with the consensus answers to six or fewer questions. Assigning the imaginary nodules to categories higher than the consensus answer was more frequent (16.0%) than assigning them to lower categories (5.5%), and the agreement rate was below 50% for two scenarios. CONCLUSION: When given difficult-to-classify scenarios, chest radiologists showed large variability in their interpretations of the Lung-RADS categories, with high frequencies of disagreement in some specific scenarios.