Effect of experimental parameters on X-ray Micro-CT imaging for Spectral CT applications

Abstract

Computed tomography (CT) is recognized as a reliable method for structural characterization of different materials. Recently, Spectral CT methods have emerged as a tool to describe the sample composition through assessment of mass density (ρ) and effective atomic number (Zeff) assessment from CT results. These techniques depend on measuring the linear attenuation of a set of standards samples. However, besides ρ and Zeff, these results are influenced by factors such as X-ray energy and may be affected by experimental factors not accounted for in the foundational equation. Thus, in this study, we propose to quantify the effects of two factors: uniaxial compression and image pixel size on measurements of linear attenuation assessed through the registered mean gray value and its standard deviation. Initially, a group of metallic oxides pellets were compressed under different pressures and scanned at various magnification factors. Subsequently, based on factorial design theory, Yates’ method was applied to evaluate the significant effects of the factors. In a second phase, using a group of known samples, a Dual-Energy CT (DECT) measurement of a validation sample was conducted to verify the quantified effects. As expected initially, we observed that uniaxial compression positively influences attenuation, as it directly affects mass density. However, we found that an increase in image pixel size leads to a decrease in registered attenuation as it reduces the relationship between radiation attenuation and detected photons. Indeed, for the leveraged variation, the linear effect of pixel size variation proved to be the most influential on the registered mean gray value. When considering these effects in DECT application, no statistical difference was observed between loose powder and pellet measurements. Furthermore, no trend was observed in the results regarding image pixel size. However, it is possible that more significant changes in these factors may yield different results, and therefore, operators applying Spectral CT methods must ensure that the same experimental conditions are applied to both the standard sample set and the sample under investigation.