Accuracy of unenhanced CT in the diagnosis of cerebral venous sinus thrombosis.

ABSTRACT

OBJECTIVES: To evaluate the diagnostic performance of unenhanced brain CT (NECT) in identifying patients with cerebral venous sinus thrombosis (CVT). METHODS: Forty-eight consecutive patients with CVT and 48 healthy controls were included in our retrospective study. All patients underwent NECT and CT/MR angiography within 24 h. Two radiologists independently evaluated NECT images for the presence of sinus hyperdensity; discrepancies were solved by consensus. Sinus attenuation was measured in seven sites. The obtained data were compared with the presence of CVT at CT/MR angiography and with patients' hematocrit. RESULTS: Interobserver agreement in sinus hyperdensity detection was good (k = 0.64). The presence of sinus hyperdensity at NECT enabled to detect patients with CVT with 81% sensitivity, 77% specificity, 78% PPV, and 80% NPV. Mean attenuation was significantly higher in sinus segments involved by CVT than in patent ones (62.4 ± 10 versus 55.6 ± 6 HU, p < 0.0001). ROC analysis showed that a cutoff value of 63 HU enables to detect patients with CVT with 52% sensitivity and 88% specificity. Hematocrit values were significantly correlated with patent sinus segments attenuation (r = 0.19). CONCLUSIONS: The presence of sinus hyperdensity at NECT enables to detect patients with CVT with 81% sensitivity and 77% specificity. A sinus attenuation cutoff value of 63 HU can be used in order to increase specificity, but lowering sensitivity.