Plain and computed radiography for detecting experimentally induced pneumothorax in cadavers: implications for detection in patients.

Pneumothorax was induced in cadavers to determine the effects of patient positioning and imaging modality (conventional screen-film and computed radiography) on radiographic findings. Chest radiography, with cadavers in the supine frontal, erect frontal, and left lateral decubitus positions, was performed at baseline and after injection of incremental quantities of air into the pleural space. Five radiologists independently interpreted each radiograph. The ability of the radiologists to diagnose pneumothorax varied by cadaver position and depended on volume of air. Overall, the left lateral decubitus view was most sensitive (88%) for diagnosis of pneumothorax, followed by the erect (59%) and supine (37%) views. Receiver operating characteristic curves and multiple repeated measures analysis of variance revealed no statistically significant difference between diagnostic proficiency with conventional screen-film radiography and that with computed radiography. The authors conclude that the lateral decubitus view is superior to the erect and supine views for pneumothorax detection and that conventional and computed radiography perform similarly in pneumothorax detection.