Detection of Anemia on Non-Contrast Chest CT: A Prospective Cross- Sectional Study from Nepal

Abstract

Introduction: Anemia affects 24.8% of the global population. Complete blood count remains the diagnostic gold standard, but laboratory access is limited in rural Nepal. Blood attenuation on noncontrast computed tomography (NCCT) correlates directly with hemoglobin concentration. The study aims to evaluate diagnostic accuracy of subjective signs and objective Hounsfield unit (HU) measurements for detecting anemia on non-contrast chest CT in a Nepalese population.
Methods: In this prospective cross-sectional study (May–December 2024), 329 adults undergoing non-contrast chest CT with hemoglobin assay within 7 days were enrolled. Anemia was defined as hemoglobin <10 g/dL. Two blinded radiologists assessed aortic rim and interventricular septum signs. Mean HU values were measured in the left ventricle, right ventricle, pulmonary trunk, inferior vena cava, and ascending aorta. Diagnostic performance and area under the receiver operating characteristic curve (AUC) were calculated.
Results: Of 329 participants (mean age 58.0±14.2 years; 58% male), 101 (30.7%) were anemic. Aortic rim sign showed sensitivity 88.1% and specificity 75.9% (χ² p<0.001). The interventricular septum sign had a sensitivity 91.1% but a specificity 70.6%. HU measurements correlated significantly with hemoglobin across all sites (all p<0.001), strongest for the left ventricle (r=0.625) and aorta (r=0.555). Left ventricle yielded the highest AUC (0.814), followed by right ventricle (0.808) and aorta (0.790). At a 35 HU cutoff in the left ventricle, sensitivity was 80.2% and specificity 73.7%.
Conclusions: Quantitative HU measurements, particularly in the left ventricle and aorta, demonstrate good diagnostic accuracy for anemia detection on non-contrast chest CT. In resource-limited settings like Nepal, opportunistic anemia screening during routine CT interpretation could prompt timely confirmatory testing and management.