The Impact of Mammography Physics Instrumentation On Entrance Air Kerma and Mean Glandular Dose
L Brateman1*, P Heintz2, (1) University of Florida, Gainesville, FL, (2) Univ New Mexico Radiology Dept., Albuquerque, NMWE-G-103-9 Wednesday 4:30PM - 6:00PM Room: 103
Purpose: Recent mammography equipment evaluations for new installations showed discrepancies in measurements of parameters related to MGD with different physics instrumentation. The purpose of this study is to evaluate the reasons for these discrepancies for two major vendors of mammography equipment.
Methods: This study tested GE Essential and Hologic Dimensions units (6 target-filter combinations) with 4 instruments from 3 vendors using techniques specified by ACR and in agreement with manufacturers recommendations. Relative MGDs were calculated from measured entrance air kerma (EAK), kVp and HVL for each target-filter combination at fixed mAs for each instrument. The effect of each parameter on MGD was evaluated independently. Particular attention was paid to the relevant kVps (28-30) and target-filter combinations required for mammography phantom MGD calculations by interpolation from published dose tables.
Results: EAK was found to be nearly independent of field size and varied nearly linearly with kVp for all instruments. In the range of kVps for phantom MGD measurements, a measured error as large as 1 kV had very small (0.5-1.5%) impact on decreasing MGD, whereas +5% error in HVL reduced MGD by 4%. Errors in instrument response for kVp and HVL of these magnitudes were seen and reported separately. For the 6 target-filter combinations and all instruments, the variability in calculated MDG ranged from -5.8% to +5.3%. Analysis of each relevant parameter independently for two instruments showed occasional fortuitous compensations between errors in HVL and EAK that reduced the total error in MGD from 10% to 5%.
Conclusion: Discrepancies in measured EAK are likely to be attributed to the fact that NIST beams for mammography calibrations are limited to Mo/Mo. Errors reported here result from measurements made with extreme care in a well-defined setup. Larger errors in a typical clinical setting could reasonably be expected.