ABSTRACT
PURPOSE: We have developed a classification of high-resolution computed tomography (HRCT) images for screening, surveillance and epidemiological studies of respiratory diseases caused by occupational and environmental factors. The proposed classification consists of three parts: a guideline explaining the elements of the classification scheme, a reading sheet, and reference films to aid in assessing thin-section CT films. We assessed the reliability of the proposed classification system by blinded, independent trial reading. MATERIALS AND METHODS: Seven independent radiologists and pulmonologists performed a trial reading to measure the reliability of the classification system using HRCT films from 27 pneumoconiosis patients and 7 normal controls. RESULTS: The agreement was moderate to good for rounded opacities (weighted kappa=0.68 and 0.64), irregular opacities (0.59, 0.48), honeycombing (0.65, 0.47), emphysema (0.76, 0.62) and large opacities (0.48, 0.52). Ground glass opacities (0.16, 0.20) showed poor to fair agreement. Intra-reader agreement of each of the seven readers was moderate to good (mean: weighted kappa=0.52-0.80) for parenchymal findings, but the agreement was relatively low (mean weighted kappa=0.52) for ground glass opacities. CONCLUSION: The proposed classification is able to describe early dust-related fibrotic changes and provide a semi-quantitative description of the HRCT features of major fibrotic changes in the parenchyma and pleura. Reliability, as measured by inter-reader agreement, was satisfactory.
