Background: Cerebral venous sinus thrombosis (CVST) is a recognized cause of childhood and neonatal stroke with high morbidity and mortality and a challenging diagnosis in the pediatric population.
Objective: We hypothesize that measuring Hounsfield units (HU) of blood in venous sinuses is a more reliable method to diagnose CVST and that normalizing the measured HU in relation to the patient's hematocrit levels may further improve detection of CVST in the pediatric population.
Materials and methods: We performed a retrospective chart review of 15 pediatric patients with acute CVST and 31 control patients. Regions of interest (ROIs) were plotted to measure HU values within the venous sinuses of each patient. Hounsfield unit to hematocrit (HU:Hct) ratios were also calculated. In patients with CVST, HU values were determined in thrombosed and non-thrombosed venous sinuses. Statistical analysis was performed to calculate the differences between patient and control groups and to determine optimal cutoff values for HU and HU:Hct measurements in diagnosing CVST on non-contrast brain computed tomography (CT).
Results: A statistically significant difference in sinus attenuation and HU:Hct ratio was found between thrombosed (66.2±5.3 HU, 1.96±0.4) and non-thrombosed sinuses (47.2±4.5 HU, 1.38±0.25) in the patient group (P<0.0001), with the average attenuation difference being 19 HU. A statistically significant difference was also found between thrombosed sinuses in the patient group and sinuses (48.9±3.13 HU, 1.3±0.12) in the control group (P<0.0001).
Conclusion: Optimal cutoff values of 58 HU and HU:Hct ratios of 1.4 lead to sensitivities of 100% in diagnosing CVST.
Keywords: Cerebral venous sinus thrombosis; Children; Computed tomography; Hematocrit; Hounsfield unit.