FBP initialized few-view CT reconstruction algorithm using similar prior image constraint.

Few-view CT reconstruction attracts wide attention in the purpose of radiation dose reduction. For analytical algorithm filtered back projection (FBP), the reconstructed image would have extremely severe artifacts. Prior information could be used in few-view CT reconstruction to improve image quality, and similar prior images of the same person could provide much more reliable information. However, in most cases, the previous CT scans might be obtained a long time ago and not much the same as the current one. Applying the entire prior image could result in irreversible structure distortions in iterative reconstruction. In our work, we proposed an FBP initialized similar prior image constrained technique (FISPICT) to utilize general distribution information from FBP in few projection views and guide the search of distribution regions from prior images. Matched regions from similar prior image are adopted to approximate the real image in order to suppress structure differences and streak artifacts. The simulation experiments show better performance and detail preservation. The algorithm also has a high robustness while prior image differs.