Mean glandular dose estimation in population-based breast screening in Queensland, Australia: Retrospective...

To compare mean glandular dose (MGD) estimates using the Dance and AAPM/EFOMP TG282 methodologies in a large breast screening dataset from Queensland, Australia and to establish population-level baselines that can inform future analysis and development of diagnostic reference levels (DRLs).

This retrospective study analysed 1,280,343 two-dimensional mammographic exposures acquired between April 2022 and March 2025 on Hologic 3Dimensions and Selenia Dimensions, and Siemens MAMMOMAT Inspiration and Revelation systems. MGD was estimated using Dance and AAPM/EFOMP TG282 methodologies. Percentage differences were calculated per compressed breast thickness (CBT) interval and compared using the Wilcoxon signed-rank test. Volumetric breast density (VBD) was estimated using Volpara software.

AAPM/EFOMP TG282 produced lower MGD estimates than Dance, with overall reduction of 24.4 % for craniocaudal (CC) views and 23.5 % for mediolateral oblique (MLO) views (p < 0.001). Differences were minimal at low CBT but increased with thickness, with discontinuities corresponding to spectral transitions in automatic exposure control (AEC) mode of operation. Mean CBT values (56.2 mm CC, 59.2 mm MLO) were comparable to previously reported values. VBD distributions aligned closely with AAPM/EFOMP TG282 reference values.

The TG282 methodology yielded lower MGD estimates than Dance, consistent with international findings and reflecting more anatomically realistic breast modelling. Results emphasise the importance of dosimetry methodology and CBT reference values in setting DRLs. The data provide population-level baselines to support future national DRLs for mammography in Australia.