Diagnostic accuracy of noncontrast CT imaging markers in cerebral venous thrombosis.

Abstract

To assess the added diagnostic value of semiquantitative imaging markers on noncontrast CT scans in cerebral venous thrombosis (CVT). In a retrospective, multicenter, blinded, case-control study of patients with recent onset (<2 weeks) CVT, 3 readers assessed (1) the accuracy of the visual impression of CVT based on a combination of direct and indirect signs, (2) the accuracy of attenuation values of the venous sinuses in Hounsfield units (with adjustment for hematocrit levels), and (3) the accuracy of attenuation ratios of affected vs unaffected sinuses in comparison with reference standard MRI or CT angiography. Controls were age-matched patients with (sub)acute neurologic presentations. We enrolled 285 patients with CVT and 303 controls from 10 international centers. Sensitivity of visual impression of thrombosis ranged from 41% to 73% and specificity ranged from 97% to 100%. Attenuation measurement had an area under the curve (AUC) of 0.78 (95% confidence interval [CI] 0.74-0.81). After adjustment for hematocrit, the AUC remained 0.78 (95% CI 0.74-0.81). The analysis of attenuation ratios of affected vs unaffected sinuses had AUC of 0.83 (95% CI 0.8-0.86). Adding this imaging marker significantly improved discrimination, but sensitivity when tolerating a false-positive rate of 20% was not higher than 76% (95% CI 0.70-0.81). Semiquantitative analysis of attenuation values for diagnosis of CVT increased sensitivity but still failed to identify 1 out of 4 CVT. This study provides Class II evidence that visual analysis of plain CT with or without attenuation measurements has high specificity but only moderate sensitivity for CVT.