A Task-Specific Argument for Variable-Exposure Breast Tomosynthesis

Digital breast tomosynthesis (DBT) is a young technology, and the current imaging protocols are not yet fully optimized. Numerous recent studies have focused on optimizing DBT scan geometries, but the optimal DBT scan geometry is inextricably linked to the exposure delivery scheme. It is possible that alternative, variable-exposure delivery schemes could change our understanding of the optimal DBT scan. There is a need for strategies to evaluate and optimize DBT exposure delivery on a task- and patient-specific basis. To this end, we developed a simulation framework that uses fast, GPU-enabled Monte Carlo simulations and linear observer models to evaluate variable-exposure DBT systems. We tested three different exposure schemes: Equal, Central, and Oblique exposure. Preliminary results indicate that for the specific task of detecting a small signal in low density breast phantoms (15%), the alternative Central and Oblique exposure schemes may increase detectability.