A reverse scatter correction method for CT images of nuclear graphite components

• Novel reverse scatter correction method for CT imaging of nuclear graphite. • Predicts transmitted photons, estimates scattered photons via low-frequency distribution. • Superior scatter artifact reduction vs. Monte Carlo and machine learning. • Enhances image uniformity, improves defect recognition and detection. • Computationally efficient, suitable for high-temperature gas-cooled reactor lines. The structural integrity of high-temperature gas-cooled reactors (HTGRs) is reliant on over 3000 carbon/graphite components, each of which necessitates computed tomography (CT) scanning for non-destructive testing prior to deployment, as stipulated by the reactor’s technical specifications. Despite the critical role of CT scans, they are frequently marred by significant scatter artifacts due to the detection of scattered photons, which compromises image uniformity and diminishes the system’s ability to detect defects. Our research presents a novel inverse scatter photon prediction model that addresses the shortcomings of traditional models by utilizing empirical data and the inherent properties of nuclear graphite/carbon components. This method begins by mapping and subtracting the transmitted photons from the projection data, followed by estimating the scattered photon distribution based on their low-frequency characteristics. The experimental results confirm that our method not only surpasses Monte-Carlo simulations and machine learning in reducing scatter artifacts but also meets the efficiency requirements for online detection. Our quantitative analyses indicate that our approach achieves the highest Defects Recognition Performance (DRP) value, which underscores a notable enhancement in the system’s ability to detect defects. We are confident that our findings will significantly enhance the defect detection capabilities of the CT scanning system, thereby contributing to the overall safety and reliability of HTGR operations.