A comparison between Patient and Phantom Mean Glandular Dose for a Quality Assurance initiative

Abstract

Introduction This study investigated and compared mean glandular dose (MGD) quality assurance measurements to clinical data, using the Dance dosimetry model [1] to calculate MGD. The aim was to establish if the mammography QA for BreastScreen QLD (BSQ) provides an accurate representation of the clinical MGD. Method Anonymised data from the PACS from all BSQ mammography systems (9 Hologic, 43 Siemens and 10 GE) were extracted from the DICOM header for 500,000 screening examinations. QA data were also collected from testing conducted once a year or after detector/tube change. MGD was calculated from the QA and clinical data using the Dance method. The differences in MGD between the measured QA and the clinical data were calculated for each manufacturer. Results There were notable differences between the QA data and patient MGD for all compressed thicknesses and all manufacturers (Table 1; Figure 1). Conclusion The QA data shows that the range of breast thicknesses simulated by PMMA does not reflect the full range of breast thicknesses encountered clinically. Differences between clinical and QA data can be explained by the thickness of PMMA not matching the equivalent breast thickness, resulting in clinical and QA data being acquired at a different kV and target filter combination. Another reason for the difference is the use of a LUT for the entrance surface air kerma (ESAK) on the DICOM header, which was used to calculate clinical MGD. A LUT does not take into account the variability in the characteristics between mammography units, whereas ESAK is measured for every unit during QA. This study provided ways to more precisely optimise diagnostic procedures to improve patient outcomes through QA to allow for identification of deviations from optimum performance of mammographic systems and clinical practice.L6338747622021-01-14 <br />