The main objective of this study is to propose a simple formulation (which we called DEEDZ) for deriving effective atomic numbers (Zeff) via electron density (ρe) calibration from dual-energy (DE) CT data. We carried out numerical analysis of this DEEDZ method for a large variety of materials with known elemental compositions and mass densities using an available photon cross sections database. The new conversion approach was also applied to previously published experimental DECT data to validate its practical feasibility.Methods
We performed numerical analysis of the DEEDZ conversion method for tissue surrogates that have the same chemical compositions and mass densities as a commercial tissue-characterization phantom in order to determine the parameters necessary for the ρe and Zeff calibrations in the DEEDZ conversion. These parameters were then applied to the human-body-equivalent tissues of ICRU Report 46 as objects of interest with unknown ρe and Zeff. The attenuation coefficients of these materials were calculated using the XCOM photon cross sections database. We also applied the DEEDZ conversion to experimental DECT data available in the literature, which was measured for two commercial phantoms of different shapes and sizes using a dual-source CT scanner at 80 kV and 140 kV/Sn.Results
The simulated Zeff’s were in excellent agreement with the reference values for almost all of the ICRU-46 human tissues over the Zeff range from 5.83 (gallstones-cholesterol) to 16.11 (bone mineral-hydroxyapatite). The relative deviations from the reference Zeff were within ± 0.3% for all materials, except for one outlier that presented a −3.1% deviation, namely, the thyroid. The reason for this discrepancy is that the thyroid contains a small amount of iodine, an element with a large atomic number (Z = 53). In the experimental case, we confirmed that the simple formulation with less fit parameters enable to calibrate Zeff as accurately as the existing calibration procedure.Conclusions
The DEEDZ conversion method based on the simple formulation proposed could facilitate the construction of ρe and Zeff images from acquired DECT data.