Automated adjustment of energy-dependent window settings for virtual monoenergetic images in dual-energy CT

Virtual monoenergetic image (VMI, 40-190 keV) exhibits substantial variations in CT number compared to conventional 120 kVp CT, necessitating energy-dependent adjustments of display window settings. This study aimed to develop a practical method to automate such adjustments.

A decomposition formula for CT number, expressed in Hounsfield Unit (HU), was derived as HUE=ρ∼_e+fE∙Z∼ , where ρ∼_e and Z∼ are determined by material properties, and f(E) represents an energy-specific factor. The theoretical values of f(E) (E = 40-190 keV) were obtained by parametric fitting of the attenuation coefficients from the NIST-XCOM database. Scanner-dependent values of f(E) were measured following a calibration procedure using a tissue characterization phantom. A universal representation of human tissues in the parameter space (ρ∼_e,Z∼) was constructed using the human tissue composition data from the ICRU-46 Report and was generalized as a piece-wise linear curve. Using established window settings for conventional CT as reference, the corresponding VMI window settings were derived by mapping to the same range in the (ρ̃_e, Z̃) space with altered energy-specific factors. Clinical images were employed to illustratethe performance of the auto-adjusted VMI window settings.

Energy-specific window settings were generated for representative anatomical regions including brain, neck, lung, abdomen, pelvis, spine and bone. In the evaluated cases, the adjusted VMI windows produced improved and more consistent image appearance across different monoenergetic levels.

The proposed method potentially provides a practical preset and automation tool for VMI window settings, facilitating standardized and efficient VMI visualization in clinical practice.