A Monte Carlo study of monoenergetic and polyenergetic normalized glandular dose (DgN) coefficients in mammography.

We investigated the influence of model assumptions in GEANT4 Monte Carlo (MC) simulations for the calculation of monoenergetic and polyenergetic normalized glandular dose coefficients (DgN) in mammography, focussing on the effect of the skin thickness and composition, of the role of compression paddles and of the bremsstrahlung processes. We showed that selecting a skin thickness of 4 mm instead of 1.45 mm produced DgN values with deviations from 9% to 32% for x-ray spectra routinely adopted in mammography. Consideration of the bremsstrahlung radiation had a weak influence on monoenergetic DgN. Simulations (in the range 8-40 kVp) which included consideration of bremsstrahlung radiation, a skin thickness of 1.45 mm and a 2 mm thick compression paddles produced polyenergetic DgN coefficients up to 19% higher than corresponding literature data. Adding a 2 mm thick adipose layer between the skin layer and the radiosensitive portion of the breast produces polyenergetic DgN values up to 15% higher than those routinely adopted. These findings provide a quantitative estimate of the influence of model parameters on the calculation of the mean glandular dose in mammography.