Table 1 showed the acquisition parameters and D_g displayed during mammography examinations according to the thickness of the compressed breast. A comparison of the two calculation methods based on the Boone and Dance models showed a statistically significant increase (p<0.05) in doses calculated by the Dance method, regardless of breast thickness (Figure 1). Acquisition parameters were not statistically different (p=1). The greatest differences were observed for the smallest breast thicknesses (<65mm). Results for 85-95mm breast thickness were not representative as few patients make up this group.

Table 1: Acquisition parameters and displayed Average Glandular Doses (Dg) in relation to the thickness of the breast according to calculations 's model on patients. Abbreviations: CC: craniocaudal incidence; MLO: medio-lateral oblique incidence.

Fig 1: Comparison of mean glandular dose (Dg) values displayed according to craniocaudal (CC) and medio-lateral oblique (MLO) incidences using the Boone and Dance calculation methods during patient mammography examinations.

Figure 2 illustrated the temporal evolution of the cumulative doses from the craniocaudal and the medio-lateral oblique incidences for standard breast thicknesses. When the calculation method was changed from Boone to Dance, the average increase of diplayed D_g was 21%.  The mean entrance dose (ESD) increased by 29% from 5.2 ± 1.2 mGy to 6.7 ± 0.5 mGy due to the change in the calculation reference point between the two software versions. However, this was not statistically different (p=0.15).

Fig 2: Temporal evolution of the doses for clinical examinations in 2D mammography for standard breast thickness (45-55mm). The red vertical line indicates the date of change of software version and Dg calculation method associated.

Table 2 showed the impact of the change to the new EFOMP/AAPM formalism on the displayed D_g on a phantom study. The results suggested that when switching from Dance's model to this new formalism, it was expected that the displayed D_g would decrease by a significant amount (p<0.05). This decrease from the thinnest to the thickest breast thickness ranged from 21% to 102%. The results also showed a significant decrease in D_g when switching from Boone's model to EFOMP/AAPM formalism (p<0.05). This decrease became more pronounced with increasing breast thickness. Compared to switching between Dance and EFOMP/AAPM, D_g was reduced less, ranging from -5% to -29%. HVL and K_air were not statistically different between both models: (p=0.2). 

Table 2: Comparison of three breast glandular dose (Dg) calculation methods by breast thickness on phantom.

The question of how well the estimate of D_g with PMMA phantoms during external quality control agrees with the average patient D_g is of interest⁴. For the 45-55 mm range of clinically compressed breast thickness, the ratios between clinical D_g and the value obtained on the phantom for 45 mm thickness with 26% glandularity were 1.4 and 1.7 for Boone and Dance, respectively. For each of the calculation methods (Boone and Dance), the D_g values calculated from the phantom measurements were lower than the D_g displayed in the clinic, regardless of the thickness studied.