CDMAM phantom scoring is widely used to assess the detectability performance of mammography systems. We propose to study the impact of structured background on this performance assessment, using simulated CDMAM phantom images with flat and textured backgrounds. Three dose levels have been investigated, ranging from -50% to +50% around the reference dose computed by the acquisition system. For textured backgrounds, the simulated projected breast corresponds to a 50mm thick, 60% glandular breast, with a texture generated by a power-law filtered noise model. Images have been scored by four image quality experts. For the smaller insert sizes, Image Quality Factor (IQF) scores obtained in textured backgrounds are lower than and well correlated with those obtained in flat backgrounds. IQF values increased with dose. For the larger insert sizes, detectability performance in textured background is even more degraded and is not as dose dependent as it is in flat backgrounds.
展开▼