The development of a standard set of chest radiographs to test a radiologist's expertise

Six test sets of normal and abnormal chest radiographs were presented to eight radiologists for their interpretation. Based upon their responses to these tests, it was determined that some radiologists consistently overread, some underread, and others do neither. From an analysis of the radiologist's responses, cases from the two best test sets could be combined to produce a standard test set of 50 to 60 chest radiographs representing a variety of pulmonary diseases. This test then could measure a radiologist's proficiency at interpretating the normal and abnormal chest, and those who consistently differ with their colleagues would be so informed.