Lisette van Dam

Imaging modalities for diagnosing cerebral vein thrombosis 7 115 follow up), found a sensitivity 41-100% and specificity of 77-100%.( Appendix 2 ) 19-26 The best reported diagnostic accuracy of the attenuated vein sign on NCCT for the diagnosis of deep cerebral vein thrombosis was a sensitivity of 100% and specificity of 99%. 20 It is important to note that this was found in a small single center retrospective study, including only 8 patients diagnosed with deep cerebral vein thrombosis. 20 The reported accuracy for cerebral sinus thrombosis is a sensitivity of 50-100% and a specificity of 83-100%. 20-22,24,25 An explanation for these wide ranges may be the different scan technologies and acquisition measures used, with a generally lower sensitivity in older studies. 23 Poor interrater variability of the evaluation of direct and indirect signs on NCCT may also play a role. 26 In the specific setting of isolated cortical vein thrombosis, the reported sensitivity and specificity are 25% (95%CI 18-25%) and 100% (95%CI 92-100%), respectively. 21 In these studies, a thrombus was often missed since the “cord sign” or “string sign”, a direct sign of cortical vein thrombosis on CT, is difficult to detect due to its location next to the skull. 21 Because of the linear association between the attenuation of blood and haematocrit levels 27 , high haematocrit values can result in a false positive CVT diagnosis. 15 On the other hand anaemia may result in false negative diagnosis. Moreover, subacute thrombosis may also result in false negative diagnoses since the density of thrombi attenuated over time and becomes isodense or even hypodense after approximately 7-14 days. 15,28 Therefore, studies have evaluated whether quantitative assessment of the attenuation and attenuation values compared to haematocrit (H:H ratio) improved the diagnostic accuracy of NCCT for the diagnosis of CVT, but did not find superior sensitivity (64-95%) or superior specificity (54-100%). 13,15,24-26,29-31 Notably, after the administration of a contrast agent, direct/indirect signs on CT scan can still be absent in up to 30% of the CVT cases ( Appendix 2 ). 9,32-39 Recently, a meta-analysis summarizing the diagnostic accuracy of CT (non- contrast- and contrast-enhanced) for CVT has been published. 40 Twenty-four eligible publications, including 48 studies with varying study designs and diagnostic standards were included, for a total of 4595 individual patients. Overall, CT was found to have a reasonable diagnostic accuracy with a pooled sensitivity of 79% (95%CI 76-82%) and a pooled specificity of 90% (95%CI 89%-91%). For the diagnosis of cerebral sinus thrombosis, the pooled sensitivity and specificity of CT were 81% (95%CI 78-84%) and 89% (95%CI 88-91%), respectively. Subgroup analyses showed no significant difference of the diagnostic accuracy in suspected acute, sub-acute