A Cone-beam Reconstruction Algorithm for Dose-minimized Short Scan and Super Short Scan

Recently, we proposed an approach that deploys dynamic collimation to shield unnecessary redundant data in circular cone-beam data acquisition, enabling a dose-minimized short scan and super short scan. In this paper, we suggest a new reconstruction algorithm that produces images from these acquisitions that are superior to those restored by traditional short scan FDK-type algorithms. The method involves two stages: First, an initial volume was reconstructed by using a truncation correction method. However, this reconstruction suffers from the streaking artifacts in slices away from the mid-plane due to implicit sharp binary weighting to the cone-beam data. Then, we forward project this initial volume to fill the original projection areas that were shielded by collimation, so that a second reconstruction with a smooth Parker-weighting scheme can be applied to reduce the streaking effects. The evaluation is performed on clinical data with the simulated dose-minimized acquisition scans. The results demonstrate that the proposed algorithm achieves image quality that is comparable to noncollimated FDK short scan reconstruction, with minimized dose to the patient.