ProTCT: projection quantification and fidelity constraint integrated deep reconstruction for Tangential CT

Computed tomography (CT) is a useful tool for non-destructive testing of large-diameter samples, such as oil pipelines and rockets. However, the projections are truncated along the detector direction, resulting in degraded images with radial artefacts. Meanwhile, existing methods fail to achieve satisfactory reconstruction quality because of the ill-defined sampling conditions in the projection domain and over-smoothing in the image domain. In this paper, we propose a Projection quantification and fidelity constraint integrated deep reconstruction method for Tangential CT (ProTCT) to improve the reconstruction quality for large-diameter object imaging. Specifically, the sampling conditions for reconstruction are analysed, offering practical guidelines for TCT firmware design. In addition, a deep artefact-suppression network and a fidelity-constraint module are introduced to operate across both projection and image domains to remove artefacts and restore edge details. As demonstrated on simulated and real datasets, ProTCT demonstrates good performance in structure restoration and detail retention. This work advances the understanding of the sampling conditions and the reconstruction methods of TCT, further promoting the application of CT in industrial inspection.