For the initial step of quality assurance program for the reading of pneumoconiosis, we developed 85 simple chest P-A films by stepwise panel readings of 8 highly qualified radiologists.
Intra-reader reliability(overall kappa value) for the pneumoco...
For the initial step of quality assurance program for the reading of pneumoconiosis, we developed 85 simple chest P-A films by stepwise panel readings of 8 highly qualified radiologists.
Intra-reader reliability(overall kappa value) for the pneumoconiotic findings of 76 films except films for differential diagnosis by 6 radiologists participating in the panel readings was as follows: 0.44 and 0.73 in profusion by complete and short classification, 0.45 and 0.40 in primary and secondary small opacity, 0.62 in extent of small opacity, 0.67 and 0.22 in costophrenic angle(CPA) obliteration and diffuse pleural thickening(DPT). Interreader reliability was as follows: 0.40 and 0.67 in profusion by complete and short classification, 0.48 and 0.43 in primary and secondary small opacity, 0.54 in extent of small opacity, 0.60 and 0.30 in CPA obliteration and DPT.
Pneumoconiotic findings of the 44 films selected from the above 76 films were compared with those of 39 radiologists working in agencies for specific health screening, agency for pneumoconiosis, and other university hospitals.
According to the category, profusion(excluding 0/0 category) agreed completely in 10-40% of the corresponding small opacity and 55-80% of the large opacity by complete classification. Within 1 minor category 55.85% of films agreed. By short classification it agreed in 50-80% and 91.2% of small and large opacity respectively. Shape and size of primary and secondary small opacity agreed completely in 41.0% and 23.2%, and extent of small opacity agreed in 47.9%. Only in the films read as pneumoconiosi by readers, 51.3% and 29.8% of primary and secondary small opacity agreed in shape and size, and 60.4% in extent. In CPA obliteration and DPT, 81.8% and 78.3% of the corresponding films agreed respectively.
Considering the above results, it is necessary to develop educational and quality assurance program for increase of reading agreement.