Astrid B. van Rossum, MD #{149} Peter M. T. Pattynama, MD #{149} Erik R. Tjin A. Ton, MD Frank E. E. Treurniet, MD #{149} Jan-Willem Arndt, MD #{149} Berthe van Eck, MD Gerard J. Kieft, MD Pulmonary Embolism: Validation of Spiral CT Angiography in 149 Patients’ Index terms: Angiography, comparative studies, 944.122, 944.1291 #{149} Computed tomography (CT), comparative studies, 944.12912 #{149} Embolism, pulmonary, 60.72 #{149} Lung, radionuclide studies, 60.1216 #{149} Pulmonary angiography, 944.122 #{149} Pulmonary arteries, CT, 944.1291 Abbreviations: CI = confidence interval, PE = pulmonary embolism, PIOI’ED = Progressive In- vestigation of Pulmonary Embolism Diagnosis, V-P = ventilation-perfusion. Radiology 1996; 201:467-470 From the Department of Radiology, Leyenburg Hospital Leyweg 275, 2545 CH The Hague, The Netherlands (A.B.v.R., FEET., G.J.K.); and the Department of Diagnostic Radiology and Nu- clear Medicine, Leiden University Hospital, Leiden, The Netherlands (P.M.T.P., E.R.T.A.T., J.W.A., ByE.). Received February 5, 1996; revision requested March ii; final revision received June 3; accepted June 24. Address reprint requests to A.B.v.R. . RSNA, 1996 467 PURPOSE: To assess the reliability of spiral computed tomographic (CT) angiography of the pulmonary arter- ies in the diagnosis of pulmonary em- bolism (PE). MATERIALS AND METHODS: One hundred forty-nine patients clinically suspected of having PE underwent spiral CT angiography and ventila- tion-perfusion (V-P) scintigraphy. Pulmonary angiography was also performed when the results of the V-P scan were indeterminate. Imag- ing results of spiral CT angiography were compared and validated against a normal perfusion scan in 40 pa- tients, a high-probability scintigram in 53 patients, and a pulmonary an- giogram in 56 patients. RESULTS: Spiral CT angiograms were of satisfactory diagnostic qual- ity in all 149 patients. The sensitivity of spiral CT angiography for the de- tection of PE was 94% (64 of 68 pa- tients) (observer 1) and 82% (56 of 68 patients) (observer 2), and the speci- ficity was 96% (78 of 81 patients) (ob- server 1) and 93% (75 of 81 patients) (observer 2). Good interobserver agreement was obtained for spiral CT angiographic results (K = .774). Spiral CT angiography proved to be effec- tive in the detection of PE in pulmo- nary arteries up to the segmental level but not in the smaller subseg- mental branches. Isolated subseg- mental PE accounted for three false- negative spiral CT angiographic results for both observers. CONCLUSION: Spiral CT angiogra- phy is an accurate method for the de- tection and exclusion of PE, with the exception of isolated subsegmen- tal PE. D IAGNOSIS of pulmonary embolism (PE) remains a problem in clini- cal practice. The clinical diagnosis is unreliable; therefore, the diagnosis of PE relies on imaging modalities, most commonly ventilation-perfusion (V-P) scmntigraphy. Scintigraphy, however, is helpful only if the results are en- tirely normal (14%-40% of patients) or if the results indicate high probabil- ity for PE (10%-20% of patients) (1,2). In the majority of patients, the results of scintigraphy are indeterminate, that is, they are nondiagnostic (1,2). In these patients, the next recommended diagnostic test is pulmonary angiog- raphy. This technique, however, is not widely applied. It has been esti- mated that fewer than 50% of patients with indeterminate V-P scintigrams actually undergo angiography (3). Recently, spiral computed tomo- graphic (CT) angiography has emerged as a potentially useful alternative method for the detection and exclu- sion of PE. The results of initial stud- ies indicate a high accuracy in the evaluation of pulmonary arteries to the fourth- to fifth-order branches (4,5). Thus far, clinical experience with spiral CT angiography has been limited (4-9); therefore, it was our aim to evaluate the accuracy of spiral CT angiography for the detection of PE in a larger group of consecutive pa- tients. MATERIALS AND METHODS From November 1992 to October 1995, 185 patients underwent both V-P scintig- raphy and spiral CT angiography because of clinically suspected acute PE. In addi- tion, pulmonary angiography was per- formed in 48 of 67 patients with indeter- minate V-P scintigrams. (The original protocol did not require patients with in- determinate V-P scintigrams to undergo pulmonary angiography. As this protocol was believed to be insufficient, the study design was changed after inclusion of the first 19 patients with indeterminate scinti- grams.) Pulmonary angiography was also performed in eight patients in whom the results of spiral CT angiography were not consistent with the normal or high-prob- ability results obtained with V-P scintigra- phy (Fig 1). All examinations were per- formed within a 24-hour period to avoid misinterpretation of results because of autothromholysis of PE. Imaging Technique Spiral CT angiography (Somatom Plus 5; Siemens Medical Instruments, Erlangen, Germany) was performed during a 32- second single breath hold. A 5-mm/sec table feed was used to scan a 16-cm vol- ume from a level slightly above the aortic knob in the craniocaudal direction (pitch of 1, 120 kV, 210 mAs). In 21 of 185 patients (11%) who were unable to suspend respi- ration for this 32-second period, scanning was performed during shallow, gentle res- piration. One hundred milliliters of a non-