ABSTRACT
OBJECTIVE: The objective of our study was to determine the relative sensitivity and the lowest threshold of bleeding detectable with digital subtraction angiography (DSA) and with MDCT using an in vitro physiologic system. MATERIALS AND METHODS: A closed pulsatile cardiopulmonary bypass circuit was connected to tubes traversing a water bath to simulate the abdominal aorta and inferior vena cava. Three smaller interconnecting acrylic plastic tubes were connected as branches to the aortic tubing to simulate branch vessels. One of the three tubes, the control, had no holes in it, one had a 100-microm hole, and one had a 280-microm hole. The leakage rates were predetermined with a cardiac output of 2 and 4 L/min and with a mean arterial pressure (MAP) ranging from 30 to 100 mm Hg for each hole size. The following studies were performed for each of the predetermined leakage rates. For study 1, 16-MDCT was performed using bolus tracking after 35 mL of contrast medium had been injected into a simulated peripheral vein. For study 2, DSA was performed using a 4-French straight catheter placed 10 cm proximal to the holes (selective first aortic branch cannulation). For study 3, DSA was performed with a catheter placed in the small branch at the site of the hole (highly superselective). For study 4, 16-MDCT was performed with a catheter placed as in study 2, 10 cm proximal to the holes, for the detection of lower leakage rates. Cine loops of MDCT and DSA images were examined by two blinded observers to detect extravasation from the holes in the tubes (i.e., the branch arteries). Interobserver agreement was studied using Cohen's kappa statistic. RESULTS: The threshold to detect bleeding was as follows for each study: For IV contrast-enhanced MDCT (study 1), it was 0.35 mL/min; DSA with a catheter 10 cm proximal to the holes (study 2), 0.96 mL/min; DSA with a catheter at the holes (study 3), 0.05 mL/min [corrected] or lower; and intraarterial selective MDCT (study 4), 0.05 mL/min [corrected] or lower. The ease of detection improved with increasing MAPs and larger volumes of leakage. Interobserver correlation was excellent. CONCLUSION: In vitro, i.v. contrast-enhanced MDCT is more sensitive than first-order aortic branch-selective DSA in detecting active hemorrhage unless the catheter position is highly superselective and is close to the bleeding artery. These results suggest that MDCT can be used as the initial imaging technique in the diagnosis of active hemorrhage if the clinical condition of the patient allows.
