Diagnosis of diaphragmatic trauma with helical CT in a swine model.

Abstract

The purpose of this study was to determine whether diaphragmatic injury can be accurately diagnosed with helical CT in a swine model. The hypothesis of our study was that thin-section helical CT with sagittal and coronal reformations can reliably detect injury of the diaphragm. The study was performed in a swine model because of the similarity of the swine thorax to the human thorax. Ten swine had a limited abdominal helical CT (enteral contrast; 3-mm collimation; pitch, 1) before and after surgical creation of a 6-cm posterolateral laceration in the left hemidiaphragm. A repeat scan was obtained after 5 cm of gastric fundus was sutured through the laceration. The gastric fundus was used because it is the most commonly herniated viscus in human diaphragmatic injury. No IV contrast was used. Control, laceration, and herniation scans were reconstructed with 1.0-mm overlap and reformated in axial, sagittal, and coronal planes. Three observers scored each reformation as control or injury (defined as laceration or herniation) in a blinded and randomized fashion. Using helical CT, the observers were able to distinguish diaphragmatic injury from controls (p < .0001). The sensitivity and specificity were 92% and 87%, respectively, for sagittal reformations; 85% and 87%, respectively, for coronal reformations; and 73% and 80%, respectively, for axial reformations. Sagittal reformations proved superior to coronal or axial reformations (p = .01). The results were independent of individual observers: We found no significant difference in accuracy among the three observers. Helical CT can accurately detect diaphragmatic injury in a swine model.