Investigating the influence of individual physical effects on dimensional measurement deviations using XCT simulations

This study investigates individual sources of measurement deviations associated with physical effects in X-ray computed tomography (XCT) imaging, including focal spot blur effect, detector modulation transfer function, Poisson-Gaussian noise, and virtual gain correction. Key parameters were modeled to reflect XCT device characteristics, such as X-ray tube properties as well as detector attributes. Using a calibrated object designed for XCT metrology, namely a hole plate, a series of experiments were conducted to reproduce real measurement setups in a virtual simulation environment. XCT scans are compared with simulated XCT data generated under various conditions, including different numbers of projections (120 and 1000) and with physical effects both enabled and disabled. Deviations for both real and simulated XCT were determined by measuring hole diameters and distances between cylinder holes and comparing these measurements to reference values from the calibrated object, allowing quantification of the impact of different physical effects on measurement accuracy. The analysis revealed that virtual gain correction and focal spot blur had the most significant impact on measurement accuracy, with deviations of 0.018 mm and 0.016 mm, while Poisson–Gaussian noise had a minor influence of about 0.008 mm. The simulation framework accurately replicates XCT behavior (SSIM = 0.9645 for a simulated 1000 projections) and isolates individual physical effects. This study provides insight into the relative contributions of different physical effects to measurement deviations in industrial XCT. The findings have shown potential to guide future efforts in improving measurement accuracy, estimating measurement uncertainty, and developing more realistic simulation tools for XCT metrology. • Simulated physical effects are critical for replicating XCT uncertainty and deviations. • VGC and focal spot blur cause the most deviation; noise has minimal impact. • Simulations with 1000 projections closely resemble real XCT data. • Projection number strongly affects deviations in real XCT; effect is smaller in simulations. • Edge sharpness errors stem mainly from focal spot blur and VGC, affecting deviations.