Radiographic detectability of occipital and temporal-parietal fractures induced in cadaver heads.

Diagnostic performance based on a double-blind evaluation of test and control radiographs produced from nine different fractured cadaver skulls was compiled for six radiologists. The data resulting from two different diagnostic tasks were analyzed statistically for accuracy and confirmed interpretability including an assessment of task-related performance differences. Performance was found to be relatively independent of differences among radiologists and lesion location. Little overall correlation was found to exist between demonstrated ability to detect lesions radiographically, and visual confirmation of known lesions from the same films. These results suggest that overall performance was probably limited by a number of undefined clinical factors which influence the amount of meaningful information available from the resulting radiographs. The distribution of these degrading influences was not uniform within the population sampled so that certain radiographs proved to be particularly misleading when considered in terms of their apparent diagnostic value. As a result, even relatively large depressed fractures were in some instances considered radiographically uninterpretable.