Quality assurance in modern X-ray breast imaging

One of the requirements for breast cancer screening is high-quality mammography. Quality assurance programs are thus important and strict quality standards need to be set, monitored and met. Moreover, the risk associated with this screening exam should be small since asymptomatic women are examined. Due to these demands, manufacturers constantly improve and up-date X-ray systems to increase performance. This results in better and more complex X-ray systems. This thesis evaluates current physical technical performance measurements on these modern X-ray breast imaging systems.
Part 1 of this thesis evaluates and optimizes procedures for phantom based average glandular dose (AGD) assessment. We proposed alternative phantoms assuming a homogeneous distribution of the glandular tissue. The estimated AGD values obtained with these phantoms were in reasonable agreement with patient AGD values. However, due to the homogeneous nature of these phantoms, variations in densities of real breasts are not taken into account, which may lead to differences depending on the automatic exposure control system. Such differences can be reduced by including local densities in the phantoms.
In part 2 of this thesis we propose objective measurements of image quality of digital mammography systems. Currently, objective image quality measurements are performed on unprocessed images. We evaluated whether model observers, mathematical computer-based observers, have potential to evaluate image quality of processed images. We found that model observers can predict human observer performance for a simple task independent of the image quality level which suggests that model observers can be used for objective image quality assessment.