A standardized approach to technical image quality assessment in digital breast tomosynthesis using a commercial...

This study aimed to establish standardized methodologies for analysing Digital Breast Tomosynthesis (DBT) images acquired during technical image quality control tests using the CIRS Model 021 DBT Phantom.

Procedures were developed to compute the z-Point Spread Function (z-PSF) and the z-Response Function (z-RF) using a tungsten wire tilted relative to the detector plane, the Artifact Spread Function (ASF) from a grid of ceramic beads, the Signal-Difference-to-Noise Ratio (SDNR) from a set of spheroidal masses. Phantom slabs were assembled in different configurations, with each exposure repeated seven times on three DBT system models to assess reproducibility.

All methods demonstrated reproducibility within 5%, with ASF measurements showing variability of 1%. The z-resolution of a 40-degree acquisition angle DBT system was 1.22 ± 0.05 mm for stacks of reconstructed planes that were 1 mm thick. The correspondence between the visually identified plane of focus in the reconstructed images and the peak of the ASF profile was within 0.5%. In plane geometric distortion values were verified to be within 1%.

The ASF metric offers an efficient means for evaluating z-geometric distortion, while the z-PSF and z-RF metrics provide a more appropriate quantification of z-resolution. The investigated z-RF method enhances reproducibility and removes the need to calculate the wire's tilt angle, offering a simplified and reliable approach for routine quality control.