Room: Exhibit Hall | Forum 1
Purpose: Currently, there is increased interest in patient-specific mammographic mean glandular dose (MGD) estimation. Tissue composition is an important MGD determinant, affecting the automatic-exposure control (AEC) performance, and the amount of glandular tissue. Some mammography systems estimate MGD based on tissue composition in the densest breast area, as determined by the AEC. It is the purpose of this work to compare the effect of computing MGD using the maximum density from a breast region to that calculated using the composition of all imaged tissue
Methods: Left or right craniocaudal view mammograms from 124 women, acquired with a GE Senographe Essential were analyzed using the Volpara algorithm, v1.5.4, for % volumetric breast density (VBD), % glandularity by mass (G), maximum VBD in a 1cm² area (maxVBD1cm2), and maximum glandularity in a 1cm² area (maxG1cm2). Compressed breast thickness (CBT) and patient age were recorded. G and maxG1cm2 versus CBT were compared to Dance et al. population-averaged glandularity. MGD was computed using the Dance model.
Results: Patient characteristics with mean ± σ [min, max] are as follows: age = 55 ± 7 [45, 69], CBT = 58 ± 12 [31, 88] mm, VBD = 9.1 ± 6.6 [1.9, 31.8] %, G = 18.0 ± 12.7 [3.8, 61.5] %, and maxG1cm2 = 35.8 ± 18.6 [6.4, 77.2] %. Patient MGD: MGD_G = 1.3 ± 0.2 [0.9, 2.1] mGy, MGD_maxG1cm2 = 1.2 ± 0.2 [0.9, 2.0] mGy, with individual MGD_maxG1cm2 lower than MGD_G by up to 18%.
Conclusion: The maximum glandularity in 1cm² was found to be a poor surrogate for the whole-breast glandularity for most women, and the use of maximum glandularity results in an underestimation of patient dose. The underestimate is greatest for breasts with moderate VBD where the parenchymal pattern can be variable. Patient dose estimation based on whole-breast glandularity is recommended.
Funding Support, Disclosures, and Conflict of Interest: Melissa Hill is a scientific consultant to Volpara Solutions Europe Ltd.