Resolution properties of non-parametric regression sinogram smoothing using an explicit Poisson model

The authors develop and investigate an approach to tomographic image reconstruction in which nonparametric regression (NPR) using an explicit Poisson likelihood model is used to smooth each projection independently prior to reconstruction by unapodized filtered backprojection (FBP). The approach is compared to apodized FBP as well as to a related NPR approach using an objective function based on weighted least squares (WLS) rather than the Poisson likelihood. The authors also investigate the resolution and noise effects of choosing different link functions for the NPR model. The approach is found to lead to improvements in resolution-noise tradeoffs over FBP with a Hanning filter as well as over the WLS approach. The choice of link function is found to influence the resolution uniformity and isotropy properties of the reconstructed images