Monte Carlo Methods for Assessment of the Mean Glandular Dose in

Mammography: Simulations in Homogeneous Phantoms

R. M. Tucciariello

, P. Barca

, D. Caramella

, R. Lamastra

, C. Traino

and M. E. Fantacci

1,3

Department of Physics, University of Pisa, Pisa, Italy

Unit of Medical Physics, Pisa University Hospital “Azienda Ospedaliero-Universitaria Pisana”, Pisa, Italy

INFN, Pisa Section, Pisa, Italy

Department of Radiology, Pisa University Hospital “Azienda Ospedaliero-Universitaria Pisana”, Pisa, Italy

c.traino@ao-pisa.toscana.it, maria.evelina.fantacci@unipi.it

Keywords: Digital Mammography, Monte Carlo Simulations, Dosimetry, Glandular Dose, Air Kerma, Breast Imaging,

Skin Model, GEANT4, RADIOMA.

Abstract: The rationale of this study is to perform a personalized dosimetry in digital mammography, using Monte Carlo

simulations. We developed a GEANT4-based application that reproduces mammographic investigations

editable in different setups and conditions. Mean Glandular Dose (MGD) is estimated for different

compressed breast sizes and compositions. Breast compositions are obtained with homogeneous mixture of

glandular and adipose tissues. The simulated setup reproduces the Hologic Selenia® Dimensions®

Mammography System and the TASMIP

tool for deriving the photon fluence from the X-ray source has

been employed. The influence of different skin models is also investigated, deriving the mean glandular dose

in the breast using adipose tissue for different skin thicknesses, from 2 mm to 5 mm, and a dedicated

composition found in literature with the specific thickness of 1.45 mm. We denoted different photon shielding

properties on the MGD values.

1 INTRODUCTION

In European women, breast cancer is the leading

cause of cancer death, causing one in six of all deaths

from cancers in women. Screening mammography is

a low-dose X-ray examination used to detect breast

cancer, even at an early stage, when that cancer is too

small to be felt as a lump. Digital Mammography

(DM) represents the principal technique used to

reduce this mortality rate and is recommended in

women between 50 and 75. Since ionizing radiation

is used in X-ray mammography investigations, there

is a risk of contracting carcinogenesis associated with

the absorption of X-ray in the mammary gland, which

is considered to be the most radiosensitive tissue at

risk.

A DM investigation is made by compressing the

patient breast with a compression paddle and

acquiring two digital images per breast, a cranio-

caudal and a mediolateral oblique view, with a

polychromatic X-ray source. The Mean Glandular

Dose (MGD) is used for the evaluation of radio-

induced cancer risk and, in principle, this value must

be as low as possible, in agreement with the

investigation image quality. Furthermore, patients

have different compressed breast sizes and

percentage of gland tissue, involving different MGD

values associated to the investigations. Moreover,

skin thickness and different mammography units also

have important effects on radiation dose. Thus,

accurate dosimetry is an important goal to achieve in

X-ray breast imaging.

Dose in the gland tissue can’t be measured

directly, but the use of Monte Carlo (MC) simulations

provides a valuable support. In MC simulations

different variables can be investigated and some

assumptions must be taken into consideration to

estimate the dose delivered to the gland. It is

necessary to create a personalized dosimetry able to

evaluate glandular dose for different anatomical

conditions and commercial mammography units.

The rationale of our study is to perform a

personalized dosimetry in mammography, in which

the MC simulations get closer to the real situations.

Using a GEANT4 based code (https://

geant4.web.cern.ch/), which is an object-oriented

242

Tucciariello, R., Barca, P., Caramella, D., Lamastra, R., Traino, C. and Fantacci, M.

Monte Carlo Methods for Assessment of the Mean Glandular Dose in Mammography: Simulations in Homogeneous Phantoms.

DOI: 10.5220/0007482202420249

In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 242-249

ISBN: 978-989-758-353-7

C++ toolkit for the simulation of the passage of

particles through matter. Its areas of application

include high energy, nuclear and accelerator physics,

as well as studies in medical and space science. This

code has been developed at CERN. We developed a

GEANT4-based application, that reproduces

mammographic investigations, editable in different

setups and breast anatomies.

This study is part of the "RADIOMA" project

(RADiazioni IOnizzanti in MAmmografia, ionizing

radiations in mammography) and must be considered

as a preliminary work.

2 MATERIALS AND METHODS

2.1 State of the Art

Monte Carlo methods are computational methods

based on random sampling to obtain numerical

results; multiple possible realizations of the

phenomenon under examination are calculated, with

the weight of the probability of such occurrence. The

rationale is to run a high number of replications; the

greater the number of the events (photons from the X-

ray source in this case), the greater the accuracy of the

simulation.

When MC simulations are adopted for research on

dosimetry, some model assumptions must be

followed, like breast shape, skin model, adopted

materials, glandular tissue percentage, X-ray

polychromatic source etc… Thus, these assumptions

can seriously affect dosimetry values. For concise

reference henceforward, the breast composition is

referred to in terms of the glandular percentage by

weight.

Dose to the breast starts in general from incident

Air Kerma (K

air

), air dose measurement, converted by

dedicated coefficients to obtain a reference value of

dose.

The European mammography dosimetry protocol

employs the model proposed by Dance (Dance,

1990). In the MC simulations the breast is modelled

as a semi-cylinder, with radius of 80 mm and variable

height between 20 and 110 mm, with inside a

homogeneous compound of adipose and glandular

tissues surrounded by a 5 mm thick skin made by

adipose tissue. The conversion factors calculated by

the author are for a breast model of 50% glandularity

and are tabulated as a function of the breast thickness

and the Half Value Layer (HVL) of the X-ray beam.

The formalism used to calculate the Average

Glandular Dose, AGD, is

AGD = K

air

g c s ,

(1)

where K

air

is the incident air kerma without

backscatter at the upper surface of the breast,  is the

conversion factor for a 50% glandularity breast at the

specified HVL, and  and  factors correct for breast

composition and X-ray spectrum choice respectively.

The US protocol follows the Wu’s method (Wu,

1991; Wu, 1994), in which the breast shape is a semi-

cylinder but with a semi-elliptical cross-section. The

breast model has a 5 mm thick skin layer of adipose

tissue (indeed, skin thickness was considered to be 4

mm until the new 2016 ACR Digital mammography

quality control manual), while the inner part is a

homogeneous mixture of adipose and glandular

tissues; the reference relative amounts of glandular

tissue are 0%, 50% and 100%. The Mean Glandular

Dose, MGD, is obtained multiplying K

air

by a factor

denoted as normalized glandular dose ()

MGD = K

air

DgN .

(2)

Using Monte Carlo simulations, the authors

tabulated DgN values for breasts as defined above

and for X-ray spectra derived from a molybdenum

target and molybdenum filter, varying the phantom

breast thickness from 3 to 8 cm.

The maximum dose limits for the “standard

breast” (Yaffe, 2009) are, in digital mammography,

per view, 2.5 mGy in EU protocols and 3 mGy in US

protocols.

In the last years, a trend has emerged to extend

and perform protocols and research groups are

proposing their models and methods (Sechopoulos,

2012; Traino, 2017; Sottocornola, 2018). The

rationale of our study is to perform a personalized

dosimetry in mammography, in which the MC

simulations get closer to the real situations,

simulating real mammography investigations

executed with the Hologic Selenia® Dimensions®

mammography system. For personalized dosimetry

we mean the objective to evaluate both different

breast anatomies and commercial mammography

systems, with other anode/filter combinations, that of

course traduces in different glandular dose values.

The GEANT4-based application developed by

our group reproduces mammographic investigations

in different setups and breast anatomies. The code

provides, in the same run, simulation of mean

glandular dose and incident air kerma at the upper

surface of the breast (backscatter photons are

excluded from the computation). This led to a

reduction in the simulation times. Using an Intel

Core

i7 8700 CPU @ 4.30 GHz (12 threads

Monte Carlo Methods for Assessment of the Mean Glandular Dose in Mammography: Simulations in Homogeneous Phantoms

243

available), 32GB of RAM, multithreading mode

performs 10

events in approximately 10 minutes.

2.2 Code Validation and

Characteristics

The code was validated according to the prescription

of The American Association of Physicists in

Medicine, AAPM Task Group 195 (Case III, for

mammography purposes). The code showed

discrepancies from the reference data of 0.6% with

both monoenergetic and polyenergetic X-ray beams

in MGD scoring, which is computed by

  





  













(3)

where 



 is the G-factor introduced by (Boone,

1999) evaluated for the energy of the jth interacting

photon, 





the energy deposition of this photon and





the glandular mass. For Air Kerma scoring,

results are in agreement with data in literature (Sarno,

2017), using





 





 





















(4)

where 



is the energy of the ith incident photon

passes through the scoring surface S, 







the air mass energy absorption coefficient at the

energy 



(Hubbel, 1995). K

air

computation let to

obtain estimates of dose conversion coefficients to be

used or compared with data in literature (Boone,

2002; Nosratieh, 2015).

This kind of validation is useful due to the

opportunity to compare results with those given by

other groups, that use different MC codes

(Gholamkar, 2016) but the same methodology used in

the protocols.

Furthermore, other research groups propose their

physical phantom models, create to validate their MC

code, using, for example, TLD dosimeters (Wang,

2017; Nigapruke, 2010).

2.2.1 Implemented Geometry

A semi-cylinder was used to simulate a compressed

breast, with a radius of 10 cm, variable heights that

correspond to the different compressed breast

thicknesses, from 2 cm to 10 cm in 1 cm increment,

and 5 breast compositions, 0% 12.5% 25% 50% and

100% glandular fractions. Different glandular

fractions are obtained mixing properly adipose and

glandular tissue, from data provided by (Boone,

1999), in order to obtain homogeneous mixtures with

desired glandularities (Dance, 2016; Sarno 2018).

A skin thickness of 5 mm of adipose tissue was

introduced to the model and two polycarbonate

compression paddles, upper (2.8 mm thick) and lower

(4.1 mm thick).

As prescribed by the AAPM TG195 protocol, a

box made by water is used to consider the scattered

radiation from the patient body (Figure 2).

Figure 1: Breast model adopted. The semi-cylinder radius

is fixed to 10 cm, the range thickness is 2-10 cm while the

skin is investigated changing its thickness.

2.2.2 Polychromatic Source

The X-ray source is positioned following the Hologic

Selenia® Dimensions® mammography system

geometry. The X-ray beam simulates a craniocaudal

view and it is produced by a focal spot of 0.3 mm

a source-to-image receptor distance (SID) of 70 cm.

Since real mammographic exams involve a

polychromatic X-ray beam source, in order to obtain

data that are comparable with real investigations, in

the MC code a polychromatic spectra can be set as a

macro file, in which every energy bin is weighted

with its relative photon fluence; this weight is used in

the computation as a statistical probability to produce

an incidence photon in its energy bin. Thus, we use

TASMIP

, algorithm for tungsten anode material,

provided by (Boone, 1997), to produce photon

fluences referring to the Selenia® Dimensions®

system, to assess glandular dose in digital phantoms.

The algorithm provides spectra for the voltage

applied ranging from 18 to 40 kV; spectral

information is released at 500 eV intervals starting

from 5.5 to 40 keV, with energy bins centered in 5.5,

6.0, 6.5…

BIOINFORMATICS 2019 - 10th International Conference on Bioinformatics Models, Methods and Algorithms

244

Figure 2: Simulation geometry adopted. Source to detector

plane distance is set to the SID of the Hologic Selenia®

Dimensions® system.

Figure 3: Example of spectra obtained using TASMIP

algorithm, in W/Rh anode/filter combination @ 29 kV.

Spectra is normalized to unit Air Kerma.

2.2.3 Physics List

According to the prescriptions provided by the report

of AAPM Task Group 195, the “Option4”

PhysicsList was used in GEANT4, for the

constructors and instances that consider the physics

processes; this model is designed for any application

requiring high accuracy of electrons and it uses the

most accurate standard low-energy models. The

production threshold (“range cut”) fixed for the

secondary particles is expressed in terms of the

distance travelled by the particles in the medium (skin

or breast tissue), converted by GEANT4 in terms of

energy; e.g. the range cuts of 1 mm for photons and 1

µm for electrons correspond respectively to about

2.55 keV and 0.99 keV in 25% glandular breast

tissue. For energies involved in these cases Rayleigh,

photoelectric, Compton and bremsstrahlung are

simulated.

3 RESULTS

With the simulation setup described in the previous

section, MGD and K

air

were obtained. For

monoenergetic beams, we decided to focus in the

energy spectrum between 8 keV and 40 keV, with 0.5

keV step, range in which photons are produced by

mammography X-ray tubes. Figure 4 shows the MGD

per generated photon, for a 5 cm thick breast with

50% glandular fraction and a 5 mm thick skin made

by adipose tissue. For monochromatic purposes we

used 10

simulated events. The simulation model

considers the energy deposited in the breast tissue,

excluding skin; thus, at lower photon energies, the

skin “shields” the breast tissue and the delivered dose

is low.

Figure 4: Dose per photon delivered to glandular tissue to a

50% glandular breast and 5 cm thick, due to both the

primary and the secondary radiation. Each point in the

graph represents one simulation run with a monochromatic

beam with 10

simulated events.

At increasing photon energies, X-ray beam

penetrates the skin layer and deposits dose, up to a

maximum of about 23 keV; then, the total dose to the

glandular breast reduces, due to the decreasing

energy-absorption coefficient.

Of course, we concentred our efforts on

polychromatic spectra; for polyenergetic beams we

simulated 10

primary events. In Figure 5 it is

represented the mean glandular dose versus

Monte Carlo Methods for Assessment of the Mean Glandular Dose in Mammography: Simulations in Homogeneous Phantoms

245

compressed breast thickness for the same X-ray beam

source and 0%, 50% and 100% glandular tissues.

MGD values decrease while increasing compressed

breast thickness and/or glandular fraction, due to a

major glandular mass in the equation (3).

Figure 5: MGD per photon vs. compressed breast thickness

for different glandularities in W/Rh configuration @ 29 kV.

Each point on the graph refers to a single simulation with

events.

Furthermore, beam quality, in terms of half value

layer (HVL, units of mm Al) has to be considered

(Sobol, 1996). Spectra provided by algorithms are not

correct at all, because of the uncertainty of the filter

thickness (usually estimated by the manufacturer on

about 10%) and, of course, due to the algorithm

adopted approximations. This may lead to either an

overestimation, or underestimation, of low energy, or

high energy, of photons, and a consequent mismatch

on dose assessment. To avoid this circumstance a

beam quality estimate has to be performed.

Once one knows the experimental value, the

rationale is to try to reach the correct HVL value on

the algorithm varying the filter thickness. Figure 6

shows the dependence of MGD from the half value

layer of the radiation; a harder beam (i.e. a relatively

major number of photons with higher energy) delivers

more glandular dose to the breast.

Since mammary gland is considered to be the

tissue at risk, one has to consider skin as a shielding

tissue. The greater the thickness, the greater the

shielding. Unfortunately, in literature there are

bucking studies about skin thickness and

composition. The EU and US protocols used until the

2016 different skin thickness, of 5 mm and 4 mm

respectively, made by adipose tissue. The reason is

that skin is composed by three parts, starting from

outside to inside by epidermis, dermis and

hypodermis. It is not possible to differentiate them in

Figure 6: MGD vs. HVL. Different HVL values are

obtained varying the Rh filter thickness from 40 to 60 µm

with 5 µm step.

a clear manner, because thickness and distribution are

very variable, but it is evident that dermis and

hypodermis are mainly composed by adipose tissue.

Nevertheless, only the epidermis, the outer layer, is

evident from breast Computed Tomography (bCT)

images (Huang, 2008), which has a higher density

from adipose tissue. This involves different skin

attenuation and shielding. Because of its obvious

presence in the TC slices, other research groups

(Sarno, 2017; Massera, 2018) tend to consider the

epidermis layer in their respective studies, whose

average thickness is 1.45 ± 0.30.

We wanted to investigate the effect of various skin

models on MGD values, changing thickness and

compositions, simulating, as previously,

monoenergetic and polyenergetic beams. In Table 1

five types of skin model adopted are reported, whose

surround the same 50% phantom glandularity to form

five different 5 cm thick digital phantoms.

Table 1: Skin models adopted. These models are associated

to 50% glandular and 5 cm thick digital breast to form five

different phantoms.

Skin

model

Skin

thickness

Skin

composition

Density

[g/cm

]

1.45 mm

Skin (Boone, 1999)

1.09

2 mm

Adipose tissue

0.93

3 mm

Adipose tissue

0.93

4 mm

Adipose tissue

0.93

5 mm

Adipose tissue

0.93

Obviously, a thicker skin traduces in a major

photon shielding of the mammary gland, but models

#1 and #5 (which is adopted by EU protocol), due to

the different skin compositions and thicknesses,

BIOINFORMATICS 2019 - 10th International Conference on Bioinformatics Models, Methods and Algorithms

246

deserve a brief comparison. In Figure 7 different

shielding properties are denoted. For low energies,

the first breast model is less shielded by its (thin) skin,

showing slightly higher MGD values; a reverse

situation appears for higher energies; this may be

attributed to the property of the thicker skin to

“become a secondary radiation source” at increasing

energies, because of the probability of the Compton

effect (in the skin) increase at the expense of the

photoelectric effect (Sarno, 2017). These results have

been also found by Massera et al. (Massera, 2018),

who used PENELOPE, another Monte Carlo code, to

produce MGD values for monochromatic beams.

Figure 7: MGD vs. photon energy for monoenergetic beams

ranging from 8 to 40 keV, for phantoms #1 and #5.

In order to better outline this behaviour with

polychromatic beams, we reproduce two W/Rh

kilovoltage configurations, 22 kV and 29 kV. It is

evident in figures 8 and 9 that, for the same skin

composition, the thicker is the skin, the lower is the

dose to the gland; nevertheless, if we consider the #1

and the #5 skin models, we see that the last shields

more the breast tissue at @ 22 kV (Figure 8), but less

@ 29 kV (Figure 9). This result confirms the previous

statement in the monoenergetic investigation.

4 DISCUSSION

The aim of this project is to obtain a personalized and

accurate dosimetry for X-ray digital mammography,

for different breasts sizes and composition and

commercial mammography units commonly used.

Figure 8: Dose delivered to glandular tissue versus different

skin thicknesses and compositions. 1.45 mm thick skin has

a different elemental composition (Boone, 1999), while

from 2 mm to 5 mm skin is made by adipose tissue.

Simulations refer to a low-energy examination.

Figure 9: Dose delivered to glandular tissue versus different

skin thicknesses and compositions. Simulations refer to a

higher energy examination.

We developed, and opportunely validated, a

GEANT4 Monte Carlo code for dosimetry in

mammography; adopting some geometry

assumptions and an external algorithm (Boone, 1997;

Hernandez, 2017) for deriving spectral information

for mammography X-ray tubes, the code is able to

replicate different mammography setups for different

geometries and conditions, and different anatomical

women breasts.

In order to assess the Mean Glandular Dose in

digital phantoms for typical mammography

investigations, we used the TASMIP

tool to

reproduce photon fluences referring to the Hologic

Selenia® Dimensions® system, adopted in the

Department of Radiology, University Hospital

“Azienda Ospedaliero-Universitaria Pisana”, Pisa.

Monte Carlo Methods for Assessment of the Mean Glandular Dose in Mammography: Simulations in Homogeneous Phantoms

247

We derived the dependencies of the Mean

Glandular Dose changing breast anatomy and X-ray

beam; MGD decrease with both the increase of

compressed breast thickness and glandular

percentage. Important dependency is represented by

the HVL value (radiation beam quality), which can

lead to an overestimation of the glandular dose in case

of “harder” spectra. The rationale is to know the HVL

experimental value and to find the correct MGD value

referring to the specific radiation.

An important variable not yet permanently

defined in literature is the assessment of skin

thickness and composition. EU and US protocols

used, until 2016, respectively 5 mm and 4 mm thick

skin (now US protocol employs the 5 mm thick skin),

made by adipose tissue but, Huang et al. (Huang,

2008) found a different skin thickness in breast CT

investigations. A comparison between digital breast

phantoms with different skin showed of course

different MGD values. At low-energy investigations

skin 1.45 mm thick “shields” less glandular tissue,

respect to the 5 mm adipose skin, while for higher

energies shields more. This may be attributed to the

property of the thicker skin to “become a secondary

radiation source” at increasing energies, because of

the probability of the Compton effect (in the skin)

increases at the expense of the photoelectric effect

(Sarno, 2017). The choice of an appropriate model of

a digital breast phantom can be a critical aspect and

we reserve the right to continue investigating it.

In the last years, Digital Breast Tomosynthesis

(DBT) is spreading in clinics and represents an

evolution of mammography; this technique let the X-

ray tube to move in an arc over the compressed breast,

acquiring multiple images from different angles.

Images are then reconstructed by a computer forming

three-dimensional images. 3D techniques minimize

tissue overlaps that can hide cancers above the normal

overlapping.

We will improve our MC code implementing the

tomosynthesis set-up for dosimetry purposes.

To achieve the experimental verification of the

MC results, and improve the personalized dosimetry,

our efforts are focused on the creation of physical

phantoms with similar properties of the real breast,

like, of course geometry, but primarily X-ray

attenuation.

ACKNOWLEDGEMENTS

The presented work is part of the RADIOMA project

which is partially funded by "Fondazione Pisa",

Technological and Scientific Research Sector, Via

Pietro Toselli 29, Pisa (Italy). The authors would like

to thank Fondazione Pisa for giving the opportunity

to start this study.

REFERENCES

AAPM REPORT NO. 195. 2015. Monte Carlo Reference

Data Sets for Imaging Research.

Agostinelli, S. et al., 2003. GEANT4 - a simulation toolkit.

Nuclear Instruments and Methods in Physics Research

A 506 (2003) 250–303.

Boone, J.M., Fewell, T.R., Jennings, R.J., 1997.

Molybdenum, rhodium, and tungsten anode spectral

models using interpolating polynomials with

application to mammography. Med Phys. 1997

Dec;24(12):1863-74.

Boone, J.M., 1999. Glandular Breast Dose for

Monoenergetic and High-Energy X-ray Beams: Monte

Carlo Assessment. Radiology. 1999 Oct;213(1):23-37.

Boone, J.M., 2002. Normalized glandular dose (DgN)

coefficients for arbitrary x-ray spectra in

mammography: Computer-fit values of Monte Carlo

derived data. Med Phys. 2002 May;29(5):869-75.

Dance, D.R., 1990. Monte Carlo Calculation of conversion

factor for the estimation of mean glandular breast dose.

Phys Med Biol. 1990 Sep;35(9):1211-9.

Dance, D.R., Sechopoulos, I., 2016. Dosimetry in x-ray

based breast imaging. Phys Med Biol. 2016 Oct

7;61(19):R271-R304.

Gholamkar, L., Mowlavi, L.,A., Sadeghi, M., Athari, M.,

2016. Assessment of Mean Glandular Dose in

Mammography System with Different Anode-Filter

Combinations Using MCNP Code. Iran J Radiol. 2016

October; 13(4):e36484.

Hernandez, A.M., Seibert, J.A., Nosratieh. A,, Boone, J.M.,

2017. Generation and analysis of clinically relevant

breast imaging x-ray spectra. Med Phys. 2017

Jun;44(6):2148-2160.

Huang, S., Boone, J.M., Yang, K., Kwan, A.L.C., Packard,

N.J., 2008. The effect of skin thickness determined

using breast CT on mammographic dosimetry. Med

Phys. 2008 Apr;35(4):1199-206.

Hubbell, J.H., Seltzer, S.M., 1995. Tables of X-Ray Mass

Attenuation Coefficients and Mass Energy-Absorption

Coefficients from 1 keV to 20 MeV for Elements Z = 1

to 92 and 48 Additional Substances of Dosimetric

Interest. NISTIR 5632.

Massera, R.T., Tomal, A., 2018. Skin models and their

impact on mean glandular dose in mammography.

Physica Medica Volume 51, July 2018, Pages 38-47.

Nigapruke, K., Puwanich, P., Phaisangittisakul, N.,

Youngdee, W., 2010. Monte Carlo Simulation of

Average Glandular Dose and an Investigation of

Influencing Factors. Journal of Radiation Research,

2010 51(4), 441–448.

Nosratieh, A., Hernandez, A,. Shen, S.Z., Yaffe, M.J.,

Seibert, J.A., Boone, J.M., 2015. Mean glandular dose

coefficients (DgN) for x-ray spectra used in

BIOINFORMATICS 2019 - 10th International Conference on Bioinformatics Models, Methods and Algorithms

248

contemporary breast imaging systems. Phys. Med. Biol.

60 7179.

Sarno, A., Mettivier, G., Russo, P., 2017. Air kerma

calculation in Monte Carlo simulations for deriving

normalized glandular dose coefficients in

mammography. Phys. Med. Biol. 62 (2017) N337–N349.

Sarno, A., Mettivier, G., Di Lillo, F., Russo, P., 2017. A

Monte Carlo study of monoenergetic and polyenergetic

normalized glandular dose (DgN) coefficients in

mammography. Phys. Med. Biol. 62 (2017) 306–325.

Sarno, A., Mettivier, G., Di Lillo, F., Bliznakova, K.,

Sechopoulos, I., Russo, P., 2018. Homogeneous vs.

patient specific breast models for Monte Carlo

evaluation of mean glandular dose in mammography.

Physica Medica, 2018 Jul; 51, 56–63.

Sechopoulos, I., Bliznakova, K., Qin, X., Fei, B., Feng,

S.S.J, 2012. Characterization of the homogeneous

tissue mixture approximation in breast imaging

dosimetry. Med Phys. 2012 Aug;39(8):5050-5059.

Sobol, W.T., Wu, X., 1996. Parametrization of

mammography normalized average glandular dose

tables. Med Phys. 1997 Apr;24(4):547-54.

Sottocornola C., Traino A.C., Barca P., Aringhieri G.,

Marini C., Retico A., Caramella D., Fantacci M.E.,

2018. Evaluation of Dosimetric Properties in Full Field

Digital Mammography (FFDM) Development of a New

Dose Index. Proceedings of the 11th International Joint

Conference on Biomedical Engineering Systems and

Technologies (BIOSTEC 2018) - Volume 1:

BIODEVICES, 212-217.

Traino A.C., Sottocornola C., Barca P., Marini C.,

Aringhieri G, Caramella D, Fantacci ME. 2017.

Average absorbed breast dose in mammography: a new

possible dose index matching the requirements of the

European Directive 2013/59/EURATOM. European

Radiology Experimental 1:28.

Wang, W., Qiu, R., Ren, L., Liu, H., Wu, Z., Li, C., Li, J.,

2017. Validation of Monte Carlo simulation of

mammography with TLD measurement and depth dose

calculation with a detailed breast model. EPJ Web of

Conferences, 2017 153, 04017.

Wu, X., Barnes, G.T., Tucker, D.M., 1991. Spectral

dependence of glandular tissue in screen-film

mammography. Radiology. 1991 Apr;179(1):143-8.

Wu, X. Gingold, E.L., Barnes, G.T., Tucker, D.M., 1994.

Normalized average glandular dose in molybdenum

target-rhodium filter and rhodium target-rhodium filter

mammography. Radiology. 1994 Oct;193(1):83-9.

Yaffe, M.J., Boone, J.M., Packard, N., Alonzo-Proulx, O.,

Huang, S.Y., Peressotti, C.L., Al-Mayah, A., Brock, K.,

2009. The myth of 50-50 breast. Med Phys. 2009

Dec;36(12):5437-43.

Monte Carlo Methods for Assessment of the Mean Glandular Dose in Mammography: Simulations in Homogeneous Phantoms

249