Nonlinear sinogram filter design for backprojection reconstruction

This work focuses on a tomographic image reconstruction method which will be referred to as nonlinear backprojection (NBP). Rather than explicitly statistically modeling the forward process and the unknown image, the authors train an optimal nonlinear backprojection operator which can be implemented non-iteratively. Under appropriate assumptions, the method forms its estimate by applying nonlinear filters to sinogram data, followed by conventional backprojection. The nonlinear filters are designed through off-line training. Nonlinear backprojection shows promising results relative to both filtered backprojection and MAP Bayesian methods