A new curve-filtered FDK-type algorithm for circular cone-beam CT reconstruction

Circular cone-beam CT is one of the most popular configurations in both medical and industrial applications. The FDK algorithm is the most popular method for circular cone-beam CT. However, with increasing cone angle the cone-beam artifacts associated with the FDK algorithm deteriorate because the circular trajectory does not satisfy the data sufficiency condition. Along with an experimental evaluation and verification, this paper proposed a new curve-filtered FDK-type (C-FDK) algorithm. The key idea of C-FDK is data rebinning along a set of concave curves. First, cone-parallel projections are rebinned from the native cone-beam geometry in two independently rebinning directions. The projections of C-FDK are rebinned and filtered along different curves from the T-FDK in the central virtual detector plane. Without any other extra trajectories supplemental to the circular trajectory, C-FDK has a visible image quality improvement than the FDK and T-FDK algorithm by the experiments. In the end, numerical experiments are done to validate the effectiveness of the proposed algorithm by comparing with both FDK and T-FDK algorithm.