Value of spontaneous hyperdensity of cerebral venous thrombosis on helical CT

Abstract

Background Excluding a cerebral venous thrombosis (CVT) through imaging is a frequent request in the emergency setting. This assessment often starts by an unenhanced brain computed tomography (CT). Purpose Re-evaluate the value of spontaneous hyperdensity of CVT on helical unenhanced brain CT. Methods Multicentric retrospective study on CVT probability based on visual assessment of spontaneous hyperdensity of cerebral venous system, performed by four blinded radiologists, individually then collectively, on a population including 14 helical unenhanced brain CTs with CVT and 102 unenhanced brain CTs without CVT (all confirmed by CT or magnetic resonance [MR] venography). Exclusion criteria: no DICOM image, symptoms >15 days, CVT indirect signs on unenhanced CT. A fifth radiologist set 768 regions of interest to measure and to compare the density within the normal venous sinuses and the CVTs. Results After consensus reading, sensitivity of this sign for CVT diagnosis was 100%, specificity 95.1%, and negative predictive value (NPV) 100%, with high individual NPV (99–100%). Area under the receiver-operating characteristic curve was 0.992 after consensus (0.976–0.986 individually). The spontaneous density was significantly different ( P <0.05) between normal sinuses and CVTs, with a density >70 HU reported only within the CVTs, except for the horizontal part of the superior sagittal sinus (hSSS). Conclusion The dense triangle sign on helical unenhanced brain CT has an excellent NPV to exclude a sinus thrombosis during the first 2 weeks. However, we believe that visual assessment of spontaneous hyperdensity is not sufficient for the diagnosis of CVT, with possible false-positive of the hSSS on unenhanced CT.