Pulmonary embolism: validation of spiral CT angiography in 149 patients.

PURPOSE: To assess the reliability of spiral computed tomographic (CT) angiography of the pulmonary arteries in the diagnosis of pulmonary embolism (PE). MATERIALS AND METHODS: One hundred forty-nine patients clinically suspected of having PE underwent spiral CT angiography and ventilation-perfusion (V-P) scintigraphy. Pulmonary angiography was also performed when the results of the V-P scan were indeterminate. Imaging results of spiral CT angiography were compared and validated against a normal perfusion scan in 40 patients, a high-probability scintigram in 53 patients, and a pulmonary angiogram in 56 patients. RESULTS: Spiral CT angiograms were of satisfactory diagnostic quality in all 149 patients. The sensitivity of spiral CT angiography for the detection of PE was 94% (64 of 68 patients) (observer 1) and 82% (56 of 68 patients) (observer 2), and the specificity was 96% (78 of 81 patients) (observer 2). Good interobserver agreement was obtained for spiral CT angiographic results (kappa = .774). Spiral CT angiography proved to be effective in the detection of PE in pulmonary arteries up to the segmental level but not in the smaller subsegmental branches. Isolated subsegmental PE accounted for three false-negative spiral CT angiographic results for both observers. CONCLUSION: Spiral CT angiography is an accurate method for the detection and exclusion of PE, with the exception of isolated subsegmental PE.