CT perfusion image of the lung: value in the detection of pulmonary embolism in a porcine model

RATIONALE AND OBJECTIVES: We assess the value of computed tomography perfusion image (CTPI) obtained by postprocessing the CT data in the diagnosis of pulmonary embolism. METHODS: An experimental pulmonary embolism model was made in 6 pigs by injecting 2 types of emboli into the pulmonary arteries. For each pig, 5 type-A (diameter 3.5 x 8 mm) and 5 type-B (diameter 2.5 x 6 mm) emboli were injected through a catheter with the distal tip located in the inflow tract of the right atrium. After obtaining precontrast and postcontrast CT data during a single breath-hold using a 4-slice multidetector CT, perfusion images were generated by data subtraction. Approximately 150 to 180 mL of contrast material was injected at an injection rate of 6 mL/s to obtain postcontrast CT. Three independent observers twice analyzed CT images for the presence of emboli: once with postcontrast CT scans (CT angiography: CTA) alone and again with both CTA and CTPI. The locations of the emboli in the pulmonary arteries were confirmed by examining the killed porcine lungs. RESULTS: The sensitivity and positive predictive value in the detection of pulmonary emboli with CTA alone were 59% (106/180) and 87% (106/122), respectively. The sensitivity and positive predictive value with both CTA and CTPI were 87% (156/180) and 85% (156/184), respectively. For type-A emboli, the sensitivity with both CTA and CTPI (76/90, 85%) was better than that with CTA alone (63/90, 70%) (P < 0.001). For type-B emboli, the sensitivity with both CTA and CTPI (80/90, 89%) was also better than that of CTA alone (43/90, 48%) (P < 0.001). CONCLUSIONS: CTPI could be obtained using digital subtraction of the CT data. It appeared to be an adjunct in enhancing the diagnostic accuracy of pulmonary embolism, particularly when detecting small pulmonary emboli.