“…Some techniques introduce post-processing steps for the discretization of the result of a continuous reconstruction algorithm [13,14,15]. Other approaches introduce steering mechanisms into the process of continuous reconstruction methods to gain discrete results [11,12,19,32,48], or reformulate the problem as an energy minimization task, and approximate the solution with some stochastic [6,7,8,16,30,41,51,52,68], or deterministic [44,45,55,56,67,68] optimization strategy. Moreover, the difficulties described at the Continuous Reconstruction problem -i.e., the possible inconsistency of the projections, and the non-uniqueness of the results -can still be present in the discrete case, which makes an even bigger need for approximate solutions capable of handling inconsistent and incomplete projection data.…”