New iterative reconstruction technique for large field of view digital tomosynthesis

Large field of view (LFOV) X-ray digital tomosynthesis is a powerful medical imaging tool and can nondestructively provide the internal physical and biochemical properties of the specimens. Suffering from the incomplete projection dataset, LFOV X-ray digital tomosynthesis has structure-overlapping and projection-truncated artifacts when the conventional filtered back-projection (FBP) algorithm is used to reconstruct the slice images. In this paper, we report on an algebraic iterative reconstruction (AIR) technique for LFOV digital tomosynthesis. This work comprises a numerical study of the algorithm and its experimental verification using a dataset measured at an experimental setup. Different from FBP, AIR is fault tolerant and considers the reconstruction process as the optimization of a discrete representation of the object function to satisfy a system of equations that describes the LFOV imaging modality. The incomplete or corrupted dataset will be repeatedly used to reach the best and consequently the structure-overlapping and projection-truncated artifacts are suppressed. As an additional novel aspect, this method can also reduce the beam hardening artifacts in LFOV digital tomosynthesis. The proposed reconstruction approach may find applications in digital tomosynthesis system.