Intracranial stenoocclusive disease: Double-detector helical CT angiography versus digital subtraction angiography

BACKGROUND AND PURPOSE: To our knowledge, no large-scale studies comparing the accuracy of CT angiography (CTA) to intraarterial digital subtraction a ngiography (DSA) of intracranial stenosis have been reported. We attempted to determine the diagnostic value of intracranial CT angiography (CTA) of n ormal vasculature and variants as well as of stenoocclusive disease. METHODS: One-hundred and twelve patients underwent CTA and intraarterial an giography, and 2205 vascular segments were examined to ascertain presence, visibility, and degree of arterial stenoses (n = 105) as well as anatomic v ariants. Source, maximum intensity projection (MIP), and MIP-generated mult iplanar reformatted (MPR) images were evaluated. RESULTS: All 55 anatomic variants were identified correctly, Visibility of small-vessel segments was increased from 75% to 83% by using source images, MPR was helpful in differentiating distal vertebral hypoplasia from stenos is and in overcoming artifacts. All 43 occlusive segments were graded corre ctly (sensitivity = 100%, predictive value = 93.4%) as follows: severely st enotic ([n = 23], sensitivity = 78%, predictive value = 81.8%); moderately stenotic ([n = 36], sensitivity = 61%, predictive value = 84.6%); and mildl y stenotic ([n = 3], sensitivity = 66%, predictive value = 28%), Normal seg ments (n = 2100) had a sensitivity of 99.5%, and CTA evinced a specificity of 99% for detecting stenoocclusive disease. Approximately one-third of wro ng assessments were related to the petrous segment of the carotid artery. CONCLUSION: CTA with double-detector technology and advanced postprocessing algorithms, including MPR, is about as reliable as MRA in depicting the va sculature of the anterior and posterior circulation and in grading intracra nial stenoocclusive lesions, with the exception of the petrous segment of t he carotid artery, CTA might be superior to MRA in the evaluation of postst enotic low-flow segments.