A New Formula for Hemoglobin Level Prediction in Non-enhanced Brain Computed Tomography Scan

Abstract

Background: For many patients with neurological complaints, a non-enhanced brain computed tomography (CT) scan is the first workup. In some of these patients, there is no pathological finding. Anemia is a condition that can present with neurological symptoms without any imaging findings. The correlation of dural venous sinus density with hemoglobin (Hb) level has been shown in some recent studies. Objectives: This study aimed to propose a strategy to predict the level of Hb and investigate the possibility of underlying anemia based on dural venous sinus density to facilitate treatment. Patients and Methods: The CT scans of selected patients, who were referred to Faghihi Hospital in Shiraz, Iran, from October 2018 until February 2019, were reviewed in this study; the complete blood count (CBC) was measured for cases without any findings. The data of 78 patients, including CBC parameters and the mean Hounsfield unit (HU) in the superior sagittal sinus (SSS), torcula herophili (TH), and transverse sinuses (TS), were also analyzed. Results: A relatively strong direct linear correlation was found between the Hb level and HU. The Hb level was calculated based on the following formula: Hb level = 0.2 × SSS HU + 1.2 × sex factor - 0.01 × age (where sex factor is zero for females and one for males). Besides, measurement of the cutoff point for the mean HU of SSS, based on the ROC curve to predict anemia, showed that with SSS HU ≤ 50, anemia could be predicted with 84.62% sensitivity, 75.38% specificity, and 75.64% accuracy in the general population. Conclusion: A significant positive correlation was found between the Hb level and the mean HU of dural venous sinuses. Therefore, the level of Hb is predictable based on HU, and differential diagnoses are limited.