OBJECTIVE: The purpose of this study was to compare the diagnostic reliability of hard-copy and soft-copy interpretation of radiographs obtained in the emergency department using a methodology for evaluating imaging systems when independent proof of the diagnosis is not available. MATERIALS AND METHODS: We collected radiographs from a stratified sample of 100 patients seen in the emergency department. The images were obtained using computed radiography, and the digital images were printed on film and stored for display on a workstation. A group of seven experienced radiologists reported the cases using both film and the workstation display. The results were analyzed using mixture distribution analysis (MDA). RESULTS: The reliability expressed as the percentage of agreement of a typical observer relative to the majority was computed from the MDA. The result was 90% for both hard copy and soft copy with bootstrap confidence intervals of 86-94%. CONCLUSION: We conclude that, in the emergency department, soft-copy interpretation is as reliable as hard-copy interpretation. The strength of this conclusion depends on the validity of the MDA approach as well as the extent to which the observer sample and case sample are representative of the emergency department.
OBJECTIVE: The purpose of this study was to compare the diagnostic reliability of hard-copy and soft-copy interpretation of radiographs obtained in the emergency department using a methodology for evaluating imaging systems when independent proof of the diagnosis is not available. MATERIALS AND METHODS: We collected radiographs from a stratified sample of 100 patients seen in the emergency department. The images were obtained using computed radiography, and the digital images were printed on film and stored for display on a workstation. A group of seven experienced radiologists reported the cases using both film and the workstation display. The results were analyzed using mixture distribution analysis (MDA). RESULTS: The reliability expressed as the percentage of agreement of a typical observer relative to the majority was computed from the MDA. The result was 90% for both hard copy and soft copy with bootstrap confidence intervals of 86-94%. CONCLUSION: We conclude that, in the emergency department, soft-copy interpretation is as reliable as hard-copy interpretation. The strength of this conclusion depends on the validity of the MDA approach as well as the extent to which the observer sample and case sample are representative of the emergency department.