Detection of lung perfusion abnormalities using computed tomography in a porcine model of pulmonary embolism.

The purpose of this study was to identify perfusion defects of the lung using computed tomography (CT). A balloon catheter was placed in a lobar pulmonary artery of six anesthetized, ventilated, juvenile pigs to simulate occlusive segmental embolus. Contrast medium was injected via a central venous catheter at rates of 1.5, 3, 4.5, and 9 ml/s in each pig. A 40-second single-level cine CT was acquired distal to the inflated balloon during suspended inspiration. Three computer-manipulated images (time to maximal enhancement, change in maximal attenuation, maximal contrast minus precontrast subtraction) were generated using custom software and compared with the unmodified maximal enhancement and precontrast images. Two independent observers identified perfusion defects and scored the level of confidence (5-point scale) on all five images. Regions of interest were drawn in perfused and nonperfused lung and time-attenuation curves were generated. Perfusion defects were accurately (99.8 +/- 0.3%) and confidently (4.5 +/- 0.6) detected and there was excellent interobserver agreement (Kappa 0.99 +/- 0.02) on all computer-manipulated images. There was a significant increase in confidence (p < 0.05) between contrast medium injection rates of 1.5 and 9 ml/s. A linear relationship exists (r = 0.88) between injection rate and change in maximal attenuation. In conclusion, perfusion defects of the lung are seen using computer-manipulated CT images.