Proposing an Acquisition Geometry That Optimizes Super-Resolution in Digital Breast Tomosynthesis

Abstract

In digital breast tomosynthesis (DBT), oblique x-ray incidence shifts the image of an object in sub-pixel detector element increments with each projection angle. Our previous work has shown that DBT is capable of super-resolution as a result of this property. Although super-resolution is achievable over a broad range of positions for frequencies parallel to the chest wall side of the breast support, it is feasible at fewer positions for frequencies perpendicular to the chest wall. This finding arises because translational shifts in the image between projections are minimal in the chest wall-to-nipple direction. To optimize super-resolution, this work proposes an acquisition geometry in which the detector is translated in the chest wall-to-nipple direction between projections. At various increments of detector translation, we calculate the reconstruction of a sine input whose frequency is greater than the detector alias frequency. The model gives a proof-of-principle justification that detector translation promotes super-resolution.