This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Breast, Radiation physics, Computer applications, Mammography, Dosimetry, Equipment, Quality assurance
Authors:
G. Gennaro1, I. Sechopoulos2, L. Gallo3, V. Rossetti4, R. Highnam5; 1Padova/IT, 2Atlanta, GA/US, 3Castelfranco Veneto/IT, 4Torino/IT, 5Wellington/NZ
DOI:
10.1594/ecr2015/C-1576
Aims and objectives
The mean glandular dose (MGD) is the parameter used to estimate the amount of radiation dose absorbed by the breast during mammography.
MGD is calculated by multiplying the air-kerma measured at the breast surface entrance by appropriate conversion factors obtained from Monte Carlo calculations [1-3] or analytical models [4].
Depending on the considered model,
conversion factors count for the energy characteristics of the X-ray spectrum selected for breast exposure,
usually represented by the half value layer (HVL),
and for the breast absorption “properties”,
i.e.
compressed breast thickness and “composition”,
where the composition is given as percentage of glandular tissue,
assuming that the breast is composed of a certain fraction of glandular tissue and a complementary fraction of adipose tissue [1-2].
While the HVL and the compressed breast thickness can be experimentally determined,
the glandular tissue fraction,
often reported as “breast density” can only be hypothesized.
In the past,
breast density (BD) was visually assessed by radiologists through the estimation of the amount of “white” in the overall distribution of “black and white” of the mammography images.
Breast density was considered to range between 0% and 100%,
where 0% characterized breasts composed by adipose tissue which appeared as “transparent” when imaged by X-rays,
while 100% was assigned to breasts consisting essentially of glandular tissue and appearing as “fully opaque” at X-rays.
In this context,
the intermediate composition,
50% adipose – 50% glandular,
was considered a “typical value”,
characterizing the majority of breasts [1].
For this reason,
when breast density assessment by radiologists was not available,
the estimation of MGD was done assuming that all breasts had the same fifty-fifty composition.
This general assumption produced a large uncertainty on the MGD estimation,
not counting for the wide,
individual differences in breast composition.
The introduction of digital technologies in mammography has favored the development of software tools able to quantify the percentage breast density from the digital mammograms [5-6].
The use of such tools have also indicated that the fifty-fifty assumption usually made for MGD estimation is very far from the results obtained computing breast density by volumetric methods obtained from the digital images,
applying physical models to reconstruct the effective breast absorption [7-8].
In this work,
a large number of mammography images collected from different digital mammography equipment was processed by a commercial software and volumetric breast density (VBD) measured.
Each VBD value was used to determine the dose conversion factor associated to breast composition,
as required by the Dance’s dose model [1],
and mean glandular dose was estimated.
MGD “adjusted” by the compositional factor was compared with MGD values provided by manufacturers and stored in the image file headers.