The cone beam computed tomography (CT) imaging system incorporates a closed-loop calibration mechanism based on digital mirroring.

Abstract

The invention provides an advanced cone-beam computed tomography (CT) imaging system incorporating a digital replica module to establish a closed-loop calibration mechanism with a real physical device. The system addresses the time limitations of Monte Carlo simulations through a hybrid simulation structure combining high-resolution, multi-energy analytical projection and low-resolution Monte Carlo simulation, synchronously connected by a primary anchoring mechanism using a scatter-to-primary ratio (SPR) matrix. This highly physically accurate projection simulation data is used to optimize machine operating parameters, ensure radiation safety, and provide a large-scale training data source for physics-based neural network (PINN) models.The PINN model directly applies nuclear physics constraints such as the Klein-Nishina scattering formula and the Beer-Lambert attenuation law to completely eliminate scattering interference, remove beam-hardening spurious images, and support high-quality image reconstruction in sparsely spaced imaging configurations. This invention opens a new chapter in the automation and accuracy enhancement of industrial non-destructive testing and medical imaging.