A quantitative metrology for performance characterization of breast tomosynthesis systems based on an anthropomorphic phantom

Common methods for assessing image quality of digital breast tomosynthesis (DBT) devices utilize uniform, simplified or otherwise unrealistic phantoms. This study proposes a new imaging performance paradigm using a three-dimensional, anthropomorphic breast phantom and novel quantitative metrology. Images of the phantom were acquired and computationally characterizing in terms of a contrast index, a contrast variability index, a contrast to noise ratio, and a background non-uniformity index. Indices were measured at ROI sizes of 10 mm and 37 mm. The method was applied to two commercial systems, referred to as system X and system Y. All images were acquired at fixed dose of 1.5 mGy AGD. Results showed differences between systems. At all window sizes, system X had a greater contrast index than Y, where median values were 8% and 5%, respectively. Contrast variability tracked with contrast but change slightly with window size. At 10 mm, systems X and Y showed CVIs of 6% and 4%, respectively, which at 37 mm became 4% and 3%. Like the contrast index, CNR was found to be independent of window size. CNR was 0.8 for system X and 0.5 for system Y. Lastly, NI at 10 mm was 7% for system X and 6% for Y. At 37mm these reduced to 3% and 2%, respectively. The metrics showed system-dependencies correlated with tissue contrast and visual appearance of clutter from out-of-plane artifacts. The findings illustrate that the anthropomorphic phantom can be used as an image quality tool with results that are quantitative and are targeted to be reflective of clinical performance of breast tomosynthesis systems.