Generalized noise power spectrum modeling for robotic CT with arbitrary scan paths

The noise characteristics of cone-beam computed tomography (CBCT) systems with flat-panel detectors significantly affect the image quality and may limit defect detectability in industrial imaging applications. Robotic CBCT systems address imaging scenarios that are not catered to by conventional gantry-based systems. Although cascaded-systems analysis (CSA) provides a useful framework for evaluating CBCT noise performance based on system parameters, conventional formulations are limited to circular scan orbits. In this study, we extended the CSA model to arbitrary scan trajectories, enabling its application to the trajectory design of robotic CBCT systems. Specifically, the backprojection step in the CSA model was modified with an appropriate scaling factor to accommodate non-circular trajectories. The noise-power spectra estimated using the proposed model were in good agreement with the experimental measurements across various scan paths. Theoretical noise-pattern analysis suggests that metal-induced streak artifacts can be reduced by designing scan trajectories that ensure adequate ray coverage orthogonal to the streak direction within the reconstruction plane. The proposed model provides a valuable tool for estimating trajectory-dependent noise characteristics and is expected to play a key role in the optimization of scan trajectories for robotic CBCT systems. • A cascaded-systems NPS model is developed for FBP with arbitrary CBCT scan trajectories. • The proposed model enables trajectory design for robotic CT systems. • Trajectories that suppress metal artifacts are identified by enforcing ray coverage orthogonal to artifact streaks. • Local detectability index estimation using the model facilitates robotic CT trajectory optimization.