CT attenuation as a biomarker of mild-to-moderate anemia: development of a non-contrast chest CT prediction model

Non-contrast thoracic computed tomography (CT) may incidentally demonstrate imaging features associated with anemia and could complement laboratory testing, particularly in mild-to-moderate cases that are often clinically unrecognized. This study aimed to evaluate the relationship between visual signs and attenuation-based measurements on non-contrast chest CT and mild-to-moderate anemia, and to develop a combined prediction model using these imaging features.

This retrospective study included 141 adult patients who underwent non-contrast thoracic CT and hemoglobin testing within a 6-hour interval. Qualitative CT features, including the aortic ring sign and interventricular septum sign, were assessed together with quantitative attenuation measurements of the left ventricular cavity and descending aorta, expressed in Hounsfield units (HU). The diagnostic performance of individual imaging features was evaluated using receiver operating characteristic (ROC) analysis. A multivariable logistic regression model incorporating selected imaging parameters was constructed, followed by internal validation using a 7:3 training-validation split. A nomogram was derived from the final model.

The aortic ring and interventricular septum signs demonstrated moderate discriminative performance, with AUCs of 0.748 (95% CI: 0.659-0.838) and 0.763 (95% CI: 0.672-0.853), respectively. Quantitative attenuation measurements showed higher discrimination, with AUCs of 0.903 (95% CI: 0.842-0.963) for left ventricular attenuation and 0.917 (95% CI: 0.861-0.972) for descending aortic attenuation. The combined nomogram yielded AUCs of 0.980 in the training cohort and 0.953 in the validation cohort, with acceptable calibration.

Non-contrast thoracic CT may provide imaging information associated with mild-to-moderate anemia and could support incidental detection during routine examinations.