www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No.

ORIGINAL RESEARCH

Automated modification and fusion of voxel models to

construct body phantoms with heterogeneous breast
tissue: Application to MRI simulations
Joseph V. Rispoli∗1,2,3 , Steven M. Wright1,4 , Craig R. Malloy5,6 , Mary P. McDougall1,4
1
Department of Biomedical Engineering, Texas A&M University, College Station, Texas, United States of America
2
Weldon School of Biomedical Engineering, Purdue University, West Lafayette, Indiana, United States of America
3
School of Electrical and Computer Engineering, Purdue University, West Lafayette, Indiana, United States of America
4
Department of Electrical & Computer Engineering, Texas A&M University, College Station, Texas, United States of America
5
Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America
6
Department of Radiology, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America

Received: December 22, 2016 Accepted: February 15, 2017 Online Published: February 26, 2017
DOI: 10.5430/jbgc.v7n1p1 URL: https://doi.org/10.5430/jbgc.v7n1p1

A BSTRACT
Background: Human voxel models incorporating detailed anatomical features are vital tools for the computational evaluation of
electromagnetic (EM) fields within the body. Besides whole-body human voxel models, phantoms representing smaller heteroge-
neous anatomical features are often employed; for example, localized breast voxel models incorporating fatty and fibroglandular
tissues have been developed for a variety of EM applications including mammography simulation and dosimetry, magnetic
resonance imaging (MRI), and ultra-wideband microwave imaging. However, considering wavelength effects, electromagnetic
modeling of the breast at sub-microwave frequencies necessitates detailed breast phantoms in conjunction with whole-body voxel
models.
Methods: Heterogeneous breast phantoms are sized to fit within radiofrequency coil hardware, modified by voxel-wise extrusion,
and fused to whole-body models using voxel-wise, tissue-dependent logical operators. To illustrate the utility of this method,
finite-difference time-domain simulations are performed using a whole-body model integrated with a variety of available breast
phantoms spanning the standard four tissue density classifications representing the majority of the population.
Results: The software library uses a combination of voxel operations to seamlessly size, modify, and fuse eleven breast phantoms
to whole-body voxel models. The software is publicly available on GitHub and is linked to the file exchange at MATLAB R
Central. Simulations confirm the proportions of fatty and fibroglandular tissues in breast phantoms have significant yet predictable
implications on projected power deposition in tissue.
Conclusions: Breast phantoms may be modified and fused to whole-body voxel models using the software presented in this
work; user considerations for the open-source software and resultant phantoms are discussed. Furthermore, results indicate
simulating breast models as predominantly fatty tissue can considerably underestimate the potential for tissue heating in women
with substantial fibroglandular tissue.

Key Words: Electromagnetic fields, Anatomic models, Computer simulations, Magnetic resonance imaging, Breast

∗ Correspondence: Joseph V. Rispoli, Ph.D.; Email: jrispoli@purdue.edu; Address: Weldon School of Biomedical Engineering, Purdue University,

West Lafayette, Indiana, 47907, United States of America.

Published by Sciedu Press 1

www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No. 1

1. I NTRODUCTION need for anatomically-correct heterogeneous breast phan-

Human voxel models incorporating detailed anatomical fea- toms in conjunction with whole body voxel models, van
tures are vital tools for the computational evaluation of elec- der Velden et al. fused a body model with 3D image data
tromagnetic (EM) fields within the body.[1] One example acquired from five healthy volunteers and ultimately noted
research domain in which simulation of human voxel models that the observed disparity of simulated SAR distributions
has become a basis for ensuring patient safety is magnetic among these models was due to indeterminate variations
resonance imaging (MRI) studies. The IEC and FDA guide- in size and tissue makeup.[29] Breast size and tissue den-
lines on maximum power deposition, quantified by specific sity have great variability among the patient population; to
absorption rate (SAR, with units W kg−1 ), provide the basis classify the extent of breast tissue density across women, ra-
for setting safe MRI scanner parameters controlling input diologists largely have embraced the four tissue composition
power to the radiofrequency (RF) coil.[2, 3] To comply with categories prescribed by the American College of Radiology
these guidelines for human studies, it is common practice to (ACR) in the Breast Imaging Reporting and Data System
characterize RF transmit coils and medical implants using (BI-RADS R ).[30] Using this method, breast density is de-
full-wave electromagnetic modeling in order to establish the fined as (a) almost entirely fat, (b) scattered fibroglandular
parameters that ensure safe power and SAR levels.[4] Com- tissue, (c) heterogeneous fibroglandular tissue, and (d) ex-
monly, finite-difference time-domain (FDTD) based methods treme fibroglandular tissue. An analysis of projected SAR
are utilized owing to the ease of incorporating SAR calcula- encompassing these four classifications is used as a test bed
tions with heterogeneous body models.[5–7] in this study.
In addition to whole-body human voxel models, phantoms In order to address the need to augment existing whole-body
representing regional anatomical features are often employed voxel models to account for prone imaging and the neces-
for EM simulations. While isolated regional phantoms suf- sity to use heterogeneous breast tissue models, this work
fice for certain applications, more thorough results may be details a software procedure to seamlessly modify and fuse
obtained in some circumstances by fusing these smaller phan- heterogeneous breast phantoms to whole-body voxel mod-
toms with a whole-body model. For example, localized els. First, the breast phantoms are resized and appropriately
breast voxel models have been developed for a variety of EM situated with respect to the whole-body voxel model, and
techniques utilizing short (relative to the body) wavelengths, subsequent voxel extrusion and subtraction operations effec-
including radiation dosimetry and ultra-wideband microwave tively fuse the breast anterior to the pectoral muscles of the
imaging;[8–11] however, the frequencies currently associated whole-body voxel models. To illustrate the utility of this
with human MR scanners (i.e., 64-450 MHz) correspond to method, we performed simulations of 7 tesla MRI radiofre-
longer EM wavelengths with which the effects of the remain- quency excitation of a whole-body voxel model fused with
der of the body cannot be ignored. Thus, simulating RF field nine high-resolution, anatomically-correct breast phantoms
behavior during breast MRI scans would ideally incorporate spanning the four BI-RADS tissue density classifications,
a female whole-body voxel model into the FDTD mesh. Al- thus demonstrating the importance of being able to modify
though several adult female body models are available to the few existing whole body models for particular studies.
the research community, all are oriented in the standing or Results confirm the proportions of lipid and fibroglandular
supine positions;[12–16] unfortunately, these orientations limit tissues in breast phantoms have significant yet predictable
the existing voxel models’ applicability for simulating the implications on projected SAR. Finally, usage of the publicly-
filling factor of breast RF coils designed for women in the available software and resultant phantoms is discussed.
prone position, as is typically performed with MRI to avoid
respiratory artifacts.[17, 18] 2. M ETHODS
Furthermore, the accuracy of breast phantoms is confounded 2.1 Voxel models and software
by the anatomical variability of fatty (lipid) and fibroglan- Of the presently reported adult non-pregnant female whole-
dular tissues and their distinct material properties that af- body models, “Ella” from the Virtual Population is the
fect electromagnetic exposure.[8] Rudimentary efforts to highest-resolution and most comprehensive tissue model
model a prone breast either replaced or supplemented the of the adult female body.[12] Disadvantages from the stand-
body model with geometric volumes representing homo- point of breast modeling include the facts that Ella 1) rep-
geneous breast tissues,[19, 20] while improved breast phan- resents a woman in the supine position and 2) does not in-
toms with heterogeneous fibroglandular and adipose tissue clude any fibroglandular tissue in the breast region. For this
structures have been generated by using image data conver- work, we adapted nine heterogeneous 0.5-mm resolution
sion[11, 21, 22] or software algorithms.[23–28] Recognizing the pendant breast phantoms produced by the Hagness group at
2 ISSN 1925-4008 E-ISSN 1925-4016
www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No. 1

the University of Wisconsin–Madison (UW) for modeling into MATLAB for joining to the 0.5-mm breast phantoms.
microwave interactions between 0.5 and 20 GHz.[11] These In accord with recent studies of MRI data segmentation be-
phantoms were labelled according to the subject’s radiologist- tween the breast and the pectoral muscle, the anterior extent
assigned BI-RADS classification, denoted BI-RADS-a, -b, of Ella’s chest wall remained flat and aligned on the coro-
-c, and -d.[30] At least two phantoms were developed for nal plane.[34, 35] To position the breast phantoms on the Ella
each category, with the exception of a single phantom for BI- model before the fusing operation detailed below, two voxels
RADS-d. All nine breast models included tissues defined as were manually selected for centering the posterior coronal
fatty, fibroglandular, transitional, and skin. Within the fatty layer of all left and right breast phantoms. These locations
and fibroglandular classes, the preceding Wisconsin–Calgary were determined by orienting the base of each breast phan-
study of microwave dielectric properties identified increasing tom on the inmost posterior coronal plane in Ella’s anterior
incongruence of conductivity and permittivity values above thorax that solely overlapped breast tissue, fat, and skin,
500 MHz and accordingly divided both tissue classes into i.e., without encroaching on the pectoral fascia or adjacent
three tiers, each defined by single-pole Cole-Cole and Debye muscle tissue. As a result, the medial extent of each breast
relaxation models.[31] As these discrepancies in dielectric phantom’s base was situated underneath Ella’s skin layer,
properties have not been observed in the RF band, in this whereas the lateral extent of the base extended in free space
work the three tiers for each tissue type were merged into above the anterior skin layer, as shown in Figure 1a. The
single definitions as fat or fibroglandular tissues. The fat, locations for positioning the breast phantoms may be revised
fibroglandular, and skin tissues were assigned dielectric prop- by providing the software function alternate coordinates with
erties using the four-pole Cole-Cole equations stipulated by respect to the body phantom.
the IT’IS Foundation’s tissue properties database.[32] Vox-
els designated as transitional were assigned dielectric and
physical properties averaging the values from breast fat and
gland tissues. The resulting tissue dielectric properties at
7 T were σ = 0.033 S/m and εr = 5.5 for breast fat, σ =
0.85 S/m and εr = 62 for breast gland, σ = 0.44 S/m and
εr = 34 for transitional tissue, and σ = 0.64 S/m and εr =
50 for skin. For comparison to the approach of utilizing
Figure 1. Renderings of breast phantom modification and
homogeneous phantoms, a hemispherical cylinder was also
fusion with whole-body voxel model
generated to represent the breast in the two extreme cases; Two-dimensional axial mesh slices illustrate the (a) initial
that is, this geometry included a 1.5-mm skin layer with the discontinuity and (b) subsequent joining of BI-RADS-a UW
remainder uniformly assigned either fatty or fibroglandular Phantom 1 and Ella, with skin tissue labelled brown, glandular
tissue. tissue green, transitional tissue blue, and fatty tissues pink and
yellow.
The volumetric tissue data from the nine image-derived phan-
toms[33] were imported into MATLAB R (MathWorks, Nat-
To automate integration of breast phantoms with a body
ick, MA, USA). Since these voxel models only included
model, we developed a software function to extrude and trim
heterogeneous tissue from a single breast, with the thorax
voxels from each breast phantom. First, voxels on the pos-
modeled as uniform two-dimensional slabs of skin, fat, and
terior coronal layer separated from Ella by free space were
muscle, modifications were necessary to enable seamless
individually extruded toward the body model, penetrating
integration with the Ella body model. The voxels from the
the skin layer until encountering a voxel of adjacent tissue.
thorax slabs were removed and each phantom was duplicated
By the same token, breast phantom voxels defined as skin
for bilateral breast simulation. The breast phantoms were
that were situated inside the Ella phantom were removed
then enlarged using a nearest-neighbor interpolation routine
from the phantom. Any remaining voxels from the hetero-
to achieve a maximum 13.3-cm diameter in order to equally
geneous breast phantoms that were co-located with the body
fill the RF coil. These dimensions were selected as a worst
model took meshing precedence in simulations, overwriting
case to position the breast tissue 10 mm from the nearest
Ella voxels assigned skin, breast (fat), subcutaneous adipose
coil conductor, as is the case with the actual RF coil. The
tissue, and average infiltrated fat. These extrusion and trim-
resulting size of each breast phantom may be adjusted by
ming functions were applied independently for both the left
providing the software routine with input parameters defin-
and right breast positions. To illustrate this integration, the
ing the desired dimensions. The finest-resolution Ella v1.3
Ella voxel model fused with BI-RADS-b UW Phantom 2 is
voxel model, with 0.5-mm isotropic resolution, was imported
illustrated in Figure 1b.
Published by Sciedu Press 3
www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No. 1

2.2 Simulations netic flux density (|B1 + |, with units T) was computed inside
2.2.1 Mesh setup the volume RF coil; only tissue voxels were included and
All breast phantoms were exported in a file format to facili- voxel size was taken into account during averaging. As MR
tate import into commercial FDTD software (XFdtd R 7.4, scanners typically monitor SAR with respect to the trans-
Remcom, State College, PA, USA). A previously-described mit magnitude, all SAR results in this study were scaled to
quadrature 1 H breast volume coil (as reported in Ref.[19] ) achieve an average |B1 + | of 1 µT at 100% duty cycle. Ac-
was positioned over each of the 11 right breast phantoms. cordingly, the raw and 10-g spatial-average SAR for each
As illustrated in Figure 2, the coil conductors were centered case were gauged in watts per kilogram per squared mi-
on the breast phantom and separated from the nearest skin crotesla (W/kg/µT2 ). To assess relative transmit coil power
voxel by 10 mm, as is the case in the actual coil former. The requirements as it relates to breast density, the net input (i.e.,
coil was driven by two sinusoidal feeds with appropriate forward minus reflected) power required to achieve average
90◦ phase shifts to simulate quadrature operation. The fre- |B1 + | of 1 µT was also noted.
quency was set to 298 MHz to model operation at 7 T; the
simulation frequency may be adjusted to model other field 3. R ESULTS
strengths, e.g., 128 MHz for 3 T. To account for the different Figure 3 displays raw and 10-g average SAR plots on the
impedances of the quadrature pair, current source feed ampli- central sagittal slice. As anticipated, the raw SAR plots re-
tudes were adjusted to deliver equal input power to the two veal localized maxima in regions with higher conductivity
quadrature channels with less than 0.5% (-23 dB) variation. and permittivity as compared to adipose tissue. Notably,
Cell gridding was adaptively set between 0.5 and 1.0 mm for all BI-RADS-c and -d phantoms resulted in maximum 10-
the coil and breast phantom, and all curved conductor geome- g average SAR incorporating gland regions interior to the
tries utilized the software’s conformal meshing capabilities. breast, whereas the maxima in all BI-RADS-a, -b, and uni-
The entire mesh was surrounded by a quarter-wavelength of form phantoms were in the skin layer on the breast surface.
free space padding cells and the boundary comprised seven Furthermore, note the higher-density phantoms are more in-
perfectly matched layers. Simulations were run on a work- clined to have internal SAR maxima as opposed to surface
station with two linked NVIDIA GeForce GTX 780 GPU localization in the mostly-fatty cases.
cards, each with 3 GB RAM and 2,304 cores, resulting in
an average computational time of 73 minutes. Convergence SAR data and coil efficiency for all cases are shown in Table
was determined by transients dissipating to -50 dB devia- 1. Compared to the mostly-fatty BI-RADS-a UW phantoms,
tion from the pure sinusoidal wave. Steady-state field data the maximum local 10-g average SAR was threefold higher
were calculated for all tissue voxels interior to the RF coil in the denser BI-RADS-d UW Phantom 1. Note the extreme
and globally on a central sagittal plane, and SAR data were case of a uniform cylinder of dense gland tissue resulted
calculated throughout the phantoms. in almost fourfold the SAR of the otherwise worst-case BI-
RADS-d UW Phantom 1. As expected with more substan-
tial dielectric loads, coil transmit efficiency decreased with
denser breast phantoms, as indicated by the increased power
requirements to achieve an average |B1 + | of 1 µT throughout
the breast.

4. D ISCUSSION
The MATLAB functions developed for this study collectively
facilitated 1) three-dimensional resizing, 2) placement with
respect to the whole-body voxel model, and 3) voxel extru-
sion and subtraction to fuse all peripheral breast phantoms
with the whole-body voxel model. In this study, the functions
were performed on breast phantoms to illustrate the utility
Figure 2. The Ella voxel model fused with BI-RADS-c UW
Phantom 2, together surrounded by the quadrature volume of fusing anatomical voxel models. The software may be
breast coil applied to other regions, e.g., tumors, patient-specific fea-
tures, or richly-detailed extremity phantoms. As an added
2.2.2 Analysis benefit of this software method, the fusing operation does
Steady-state data output from FDTD simulations were im- not modify the body model itself, nor require distribution
ported into MATLAB for analysis. Mean transmit-field mag- with the body model. Therefore, any resulting secondary
4 ISSN 1925-4008 E-ISSN 1925-4016
www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No. 1

phantoms may be independently distributed under separate bundled readme documentation details the function names
licensing terms. The software developed for this work is and descriptions, input/output variables, and recommended
open-source, publicly-available in the GitHub repository, execution sequence.
and linked to the file exchange at MATLAB Central.[36] The

Figure 3. Raw and 10-g average SAR plots through the right breast’s central sagittal slice for the 11 phantom cases;
extreme average SAR in the uniform glandular phantom required a separate scale bar for the single case

Table 1. Coil maximum 10-g average SAR and power efficiency data
Maximum 10-g Average SAR Input Power
Breast Phantom Tissue Classification
[W/kg/μT2] [W] +
BI-RADS-a UW Phantom 1 almost entirely fat 0.30 0.63
BI-RADS-a UW Phantom 2 almost entirely fat 0.30 0.64
Uniform fat with skin layer almost entirely fat 0.45 0.67
BI-RADS-b UW Phantom 1 scattered fibroglandular tissue 0.25 0.62
BI-RADS-b UW Phantom 2 scattered fibroglandular tissue 0.30 0.65
BI-RADS-b UW Phantom 3 scattered fibroglandular tissue 0.50 0.69
BI-RADS-c UW Phantom 1 heterogeneous fibroglandular tissue 0.61 0.72
BI-RADS-c UW Phantom 2 heterogeneous fibroglandular tissue 0.58 0.68
BI-RADS-c UW Phantom 3 heterogeneous fibroglandular tissue 0.69 0.72
BI-RADS-d UW Phantom 1 extreme fibroglandular tissue 0.89 0.75
Uniform gland with skin layer extreme fibroglandular tissue 3.4 1.8

The simulation study findings demonstrated significant varia- produced threefold higher maximum 10-g average SAR as
tion among breast phantoms with varying levels of fat and compared to mostly fatty breasts (BI-RADS-a). The geomet-
fibroglandular breast tissues. During RF transmission, breast ric phantoms representing homogeneous tissue illustrated the
tissues with higher electrical conductivity and permittivity expected RF field interference patterns in a uniform, elec-
values, i.e., skin and glands, intensify the electric field and trically large dielectric.[37] The intensity of this effect was
consequently are subject to greater localized SAR. For in- especially apparent with the homogeneous gland tissue phan-
stance, a dense heterogeneous (BI-RADS-d) breast model tom and its symmetric SAR distributions as shown in Figure

Published by Sciedu Press 5

www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No. 1

3; thus, the suitability of uniform phantoms for modeling the simulated SAR data for predicting tissue temperature
in vivo breast scans is problematic. This study’s findings rise.
indicate simulating the breast as predominantly-fatty tissue
This work addressed two problems relevant to modeling the
can considerably underestimate the potential for tissue heat-
breast accurately—the need to augment existing body voxel
ing in women with extensive fibroglandular breast tissue.
models to account for prone imaging and to use heteroge-
This line of reasoning suggests to conservatively character-
neous breast tissue models. The software developed to fuse
ize SAR for regulatory compliance, simulation analyses of
breast and body phantoms is publicly available and may be
breast coils should incorporate a heterogeneous breast model
employed to join other anatomical features. The ensuing
largely consisting of fibroglandular tissue.
breast MRI simulation study presented herein incorporated
To facilitate a straightforward comparison, this study as- several high-resolution, anatomically-correct breast phan-
sumed equivalent breast sizes and positioning within the RF toms. Simulation results indicate breasts with predominantly
coil structure. Future research may provide further insight fibroglandular tissue are susceptible to significantly increased
into the effects on SAR with varying breast size and posi- SAR as compared to breasts mostly composed of fat.
tioning. This study was also limited to a single validated
coil design; simulations featuring other RF coils may be ACKNOWLEDGEMENTS
informative. Finally, this simulation study quantified SAR, This work was supported in part by CPRIT con-
which is currently the standard metric for safety regulatory tract RP100625 and NIH awards R21EB016394 and
compliance. Thermometry studies may further complement P41EB015908.

R EFERENCES [10] Parker MS, Hui FK, Camacho MA, et al. Female breast radiation
[1] Hand JW. Modelling the interaction of electromagnetic fields (10 exposure during CT pulmonary angiography. Am J Roentgenol. 2005;
MHz–10 GHz) with the human body: methods and applications. Phys 185: 1228-1233. https://doi.org/10.2214/AJR.04.0770
Med Biol. 2008; 53: R243-R286. https://doi.org/10.1088/00 [11] Zastrow E, Davis SK, Lazebnik M, et al. Development of anatom-
31-9155/53/16/R01 ically realistic numerical breast phantoms with accurate dielec-
[2] Technical committee 62 (Electrical equipment in medical practice), tric properties for modeling microwave interactions with the hu-
subcommittee 62B (Diagnostic imaging equipment). Medical electri- man breast. IEEE T Bio-Med Eng. 2008; 55: 2792-2800. https:
cal equipment—Part 2-33: Particular requirements for the safety of //doi.org/10.1109/TBME.2008.2002130
magnetic resonance equipment for medical diagnosis. 3.1 ed. Geneva [12] Christ A, Kainz W, Hahn EG, et al. The Virtual Family—development
(Switzerland): International Electrotechnical Commission; 2013. Re- of surface-based anatomical models of two adults and two children
port No. 60601-2-33. for dosimetric simulations. Phys Med Biol. 2010; 55: N23-N38.
[3] Center for Devices and Radiological Health. Guidance for the sub- https://doi.org/10.1088/0031-9155/55/2/N01
mission of premarket notifications for magnetic resonance diagnostic [13] Dimbylow P. Development of the female voxel phantom, NAOMI,
devices. Rockville (MD): Food and Drug Administration; 2016. Re- and its application to calculations of induced current densities and
port No. 340. electric fields from applied low frequency magnetic and electric fields.
[4] Nyenhuis J. RF device safety and compatibility. eMagRes. 2010. Phys Med Biol. 2005; 50: 1047-1070. https://doi.org/10.108
https://doi.org/10.1002/9780470034590.emrstm1160 8/0031-9155/50/6/002
[5] Collins CM, Li S, Smith MB. SAR and B1 field distributions in [14] Liu W, Collins CM, Smith MB. Calculations of B1 distribution,
a heterogeneous human head model within a birdcage coil. Magn specific energy absorption rate, and intrinsic signal-to-noise ratio
Reson Med. 1998; 40: 847-856. https://doi.org/10.1002/mr for a body-size birdcage coil loaded with different human sub-
m.1910400610 jects at 64 and 128 MHz. Appl Magn Reson. 2005; 29: 5-18.
[6] Dimbylow P. FDTD calculations of the whole-body averaged SAR https://doi.org/10.1007/BF03166953
in an anatomically realistic voxel model of the human body from [15] Nagaoka T, Watanabe S, Sakurai K, et al. Development of realistic
1 MHz to 1 GHz. Phys Med Biol. 1997; 42: 479-490. https: high-resolution whole-body voxel models of Japanese adult males
//doi.org/10.1088/0031-9155/42/3/003 and females of average height and weight, and application of models
[7] Jin J, Chen J. On the SAR and field inhomogeneity of birdcage coils to radio-frequency electromagnetic-field dosimetry. Phys Med Biol.
loaded with the human head. Magn Reson Med. 1997; 38: 953-963. 2004; 49: 1-15. https://doi.org/10.1088/0031-9155/49/1
https://doi.org/10.1002/mrm.1910380615 /001
[8] Dance DR, Skinner CL, Alm Carlsson G. Breast dosimetry. Appl [16] Stabin MG, Emmons MA, Segars WP, et al. Realistic reference adult
Radiat Isot. 1999; 50: 185-203. https://doi.org/10.1016/S0 and paediatric phantom series for internal and external dosimetry.
969-8043(98)00047-5 Radiat Prot Dosim. 2012; 149: 56-59. https://doi.org/10.109
[9] Hassan AM, El-Shenawee M. Review of electromagnetic techniques 3/rpd/ncr383
for breast cancer detection. IEEE Reviews in Biomedical Engineer- [17] Orel SG, Schnall MD. MR imaging of the breast for the detection,
ing. 2011; 4: 103-118. https://doi.org/10.1109/RBME.2011. diagnosis, and staging of breast cancer. Radiology. 2001; 220: 13-30.
2169780 https://doi.org/10.1148/radiology.220.1.r01jl3113

6 ISSN 1925-4008 E-ISSN 1925-4016

www.sciedupress.com/jbgc Journal of Biomedical Graphics and Computing 2017, Vol. 7, No. 1

[18] Hylton NM, Frankel SD. Imaging techniques for breast MR imaging. [28] Ma AKW, Gunn S, Darambara DG. Introducing DeBRa: a detailed
Magn Reson Imaging Clin N Am. 1994; 2: 511-525. PMid: 7489305. breast model for radiological studies. Phys Med Biol. 2009; 54(14):
4533. https://doi.org/10.1088/0031-9155/54/14/010
[19] McDougall MP, Cheshkov S, Rispoli J, et al. Quadrature transmit coil
for breast imaging at 7 tesla using forced current excitation for im- [29] van der Velden TA, Italiaander M, van der Kemp WJM, et al. Ra-
proved homogeneity. J Magn Reson Imaging. 2014; 40: 1165-1173. diofrequency configuration to facilitate bilateral breast 31P MR spec-
https://doi.org/10.1002/jmri.24473 troscopic imaging and high-resolution MRI at 7 tesla. Magn Reson
Med. 2015; 74: 1803-1810. https://doi.org/10.1002/mrm.25
[20] Winkler SA, Rutt BK. Practical methods for improving B1+ homo- 573
geneity in 3 tesla breast imaging. J Magn Reson Imaging. 2015; 41: [30] Morris E, Comstock C, Lee C. ACR BI-RADS R Atlas, Breast Imag-
992-999. https://doi.org/10.1002/jmri.24635 ing Reporting and Data System. Reston (VA): American College of
[21] Klifa C, Carballido-Gamio J, Wilmes L, et al. Quantification of Radiology; c2013. ACR BI-RADS R magnetic resonance imaging;
breast tissue index from MR data using fuzzy clustering. Proceed- p. 1-165.
ings of the 26th Annual International Conference of the IEEE Engi- [31] Lazebnik M, Popovic D, McCartney L, et al. A large-scale study of
neering in Medicine and Biology Society; 2004 Sep 1-5, San Fran- the ultrawideband microwave dielectric properties of normal, benign
cisco, CA. Piscataway (NJ): IEEE; c2004. p. 1667-1670. https: and malignant breast tissues obtained from cancer surgeries. Phys
//doi.org/10.1109/IEMBS.2004.1403503 Med Biol. 2007; 52: 6093-6115. https://doi.org/10.1088/00
31-9155/52/20/002
[22] Samani A, Bishop J, Yaffe MJ, et al. Biomechanical 3-D finite ele-
ment modeling of the human breast using MRI data. IEEE T Bio-Med [32] Hasgall P, Neufeld E, Gosselin M, et al. IT’IS database for thermal
Eng. 2001; 20: 271-279. https://doi.org/10.1109/42.92147 and electromagnetic parameters of biological tissues. Zurich (Switzer-
6 land): IT’IS Foundation, c2013. http://dx.doi.org/10.13099/
ViP-Database-V2.4
[23] Bliznakova K, Bliznakov Z, Bravou V, et al. A three-dimensional [33] Zastrow E, Davis SK, Lazebnik M, et al. Database of 3D grid-
breast software phantom for mammography simulation. Phys Med based numerical breast phantoms for use in computational elec-
Biol. 2003; 48(22): 3699. https://doi.org/10.1088/0031-9 tromagnetics simulations. Madison (WI): University of Wisconsin-
155/48/22/006 Madison; c2014 [cited 2014 Sep 21]. Available from: https:
[24] Bliznakova K, Sechopoulos I, Buliev I, et al. BreastSimulator: A //uwcem.ece.wisc.edu/phantomRepository.html
software platform for breast x-ray imaging research. J Biomed Graph [34] Ertas G, Doran SJ, Leach MO. A computerized volumetric segmenta-
Comput. 2012; 2(1): 1. https://doi.org/10.5430/jbgc.v2n tion method applicable to multi-centre MRI data to support computer-
1p1 aided breast tissue analysis, density assessment and lesion localiza-
tion. Med Biol Eng Comput. 2016. https://doi.org/10.1007/
[25] Bakic PR, Zhang C, Maidment ADA. Development and charac-
s11517-016-1484-y
terization of an anthropomorphic breast software phantom based
upon region-growing algorithm. Med Phys. 2011; 38: 3165-3176. [35] Mustra M, Grgic M, Rangayyan RM. Review of recent advances
https://doi.org/10.1118/1.3590357 in segmentation of the breast boundary and the pectoral muscle
in mammograms. Med Biol Eng Comput. 2016; 54: 1003-1024.
[26] Chen B, Shorey J, Saunders Jr RS, et al. An Anthropomorphic Breast https://doi.org/10.1007/s11517-015-1411-7
Model for Breast Imaging Simulation and Optimization. Acad Radiol. [36] Rispoli J. Breast-body-fusion [software]. San Francisco (CA):
2011; 18(5): 536-546. https://doi.org/10.1016/j.acra.201 GitHub; c2015 [cited 2015 Apr 14]. Available from: https://
0.11.009 github.com/jrispoli/breast-body-fusion
[27] Li CM, Segars WP, Tourassi GD, et al. Methodology for generating [37] Yang QX, Wang J, Zhang X, et al. Analysis of wave behavior in
a 3D computerized breast phantom from empirical data. Med Phys. lossy dielectric samples at high field. Magn Reson Med. 2002; 47:
2009; 36: 3122-3131. https://doi.org/10.1118/1.3140588 982-989. https://doi.org/10.1002/mrm.10137

Published by Sciedu Press 7