Evaluation of the impact of view differentiation and backprojection weight in circle-plus-line cone-beam tomography

Recently, we proposed a new view differentiation scheme for analytical cone-beam reconstruction formulae that demonstrated a strong robustness to changes in the data acquisition geometry and to coarse view sampling, unlike former differentiation schemes. We incorporated this new scheme into the Katsevich reconstruction formula for the circle-plus-line trajectory. We also implemented an alternative Katsevich formula for the same trajectory, where the view differentiation step was eliminated by using integration by parts. This work evaluates both formulae in terms of resolution performance, noise performance, visual image quality and computational effort. We also evaluate the impact of the z-sampling on the line segment. Experiments are presented from simulated cone-beam data. The experiments show that the view differentiation approach with the new view differentiation scheme achieves similar image quality as the integration-by-part approach while being at the same time much more efficient.