Dual- energy CT versus single-energy CT for estimation of hematocrit and hemoglobin in the brain: an in vivo analysis

Purpose: Hematocrit (Hct) and hemoglobin (HB) levels in blood are known to be correlated with vascular attenuation values on single-energy computed tomography (SECT). Dual-energy computed tomography (DECT) is likely to have even better correlations than SECT, given its richer information content, but this remains unproven clinically. We compare and contrast DECT and SECT correlations between attenuation in the superior sagittal sinus (SSS) to patient Hct/HB levels, and explore the use of iodine/water decomposition maps for the same purpose. Methods: Brain SECT and DECT were acquired contemporaneously in 83 patients and attenuation was measured in the SSS on SECT, monoenergetic DECT images (40 to 140 keV in 5 keV increments) and DECT material decomposition images (water and iodine). Hct/HB values were from complete blood counts (CBC) within 30 days of imaging. Linear regressions were performed to Hct/HB using the measured attenuations as explanatory variables. Results: Hct and HB were strongly mutually correlated (r = 0.964). Hct/HB were moderately correlated (r = 0.493/0.458) with SSS attenuation on SECT, and moderately to strongly correlated for DECT (Pearson’s r ranging 0.331–0.656) over a range of monoenergetic levels (40 to 140 keV). Above 60 keV, DECT monoenergetic images were better correlated to Hct/HB than SECT, with correlation maximized at 95 keV (r = 0.656, p < 0.001). Material decomposition water images had moderate correlation (r = 0.51), improving to strong correlation (r = 0.659) for a two-variable water and iodine regression, similar to the monoenergetic results. Conclusion: DECT has better correlations to Hct/HB than SECT for all monoenergetic energies above 60 keV, with best correlations at 95 keV.