Backprojection-filtration Reconstruction for Helical Cone-beam CT with Curved Detectors

Abstract

Helical CT scanners with multi-row detectors have gained wide popularity in clinics. With increasing interest in extending the number of detector rows, the cone angle must be accounted for in image reconstruction algorithms. As the cone angle increases, artifacts in reconstructed images caused by approximate algorithms, such as the widely used FDK-based algorithms, become more of a factor that degrades the image quality. Recently, a novel reconstruction algorithm for helical cone-beam CT (CB-CT) has been proposed, which is referred to as the backprojection-filtration algorithm. This algorithm requires theoretically minimum data to reconstruct a volume image. The original BPF algorithm was presented by assuming a flat-panel detector. However, most current multi-row detectors employed in clinic CT scanners are curved detectors. In this work, we modify the backprojection-filtration (BPF) algorithm to allow for reconstruction from data collected with a curved detector. We perform simulation studies using the Shepp-Logan phantom to evaluate the modified BPF algorithm.