Image reconstruction method for segmental limited-angle CT based on coupled relative structure

Reducing scanning time or radiation doses is a primary demand in computed tomography (CT) imaging. Few-view CT and limited-angle CT are considered as two effective imaging ways to meet this demand. However, they both face different challenges in practical applications, such as difficulties in technical implementation and image reconstruction. This study focuses on a special imaging strategy called segmental limited-angle (SLA) CT, where the scanning angular range is composed of multiple segments. This strategy helps to avoid difficulties in technical implementation and reduces the complexity of CT reconstruction. However, SLA projections inherit the properties of limited-angle projections, and the reconstructed images may encounter pronounced shading artifacts. This paper proposes a reconstruction model based on coupled relative structure (CRS) and presents an algorithm to solve this model. The CRS method utilizes mutual structure features between the prior image and target image to guide image reconstruction, rather than relying solely on the prior image. To demonstrate its effectiveness, we conduct numerical simulation experiments and real CT data experiments. The reconstruction results indicate that the target image can inherit prior structures and keep away from reconstruction errors that may be introduced by inconsistent structures. Compared to other methods, the images reconstructed by our method are closer to the reference images.