Antenna Array for microwave breast cancer detection

1. 2516 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 10, OCTOBER 2015
Flexible 16 Antenna Array for Microwave
Breast Cancer Detection
Hadi Bahramiabarghouei, Student Member, IEEE, Emily Porter, Student Member, IEEE,
Adam Santorelli, Student Member, IEEE, Benoit Gosselin, Member, IEEE, Milica Popović, Senior Member, IEEE,
and Leslie A. Rusch, Fellow, IEEE
Abstract—Radar-based microwave imaging has been widely
studied for breast cancer detection in recent times. Sensing di-
electric property differences of tissues has been studied over a
wide frequency band for this application. We design single- and
dual-polarization antennas for wireless ultrawideband breast can-
cer detection systems using an inhomogeneous multilayer model
of the human breast. Antennas made from flexible materials are
more easily adapted to wearable applications. Miniaturized flexi-
ble monopole and spiral antennas on a 50-μm Kapton polyimide
are designed, using a high-frequency structure simulator, to be in
contact with biological breast tissues. The proposed antennas are
designed to operate in a frequency range of 2–4 GHz (with reflec-
tion coefficient (S11 ) below –10 dB). Measurements show that the
flexible antennas have good impedance matching when in different
positions with different curvature around the breast. Our minia-
turized flexible antennas are 20 mm × 20 mm. Furthermore, two
flexible conformal 4 × 4 ultrawideband antenna arrays (single and
dual polarization), in a format similar to that of a bra, were devel-
oped for a radar-based breast cancer detection system. By using a
reflector for the arrays, the penetration of the propagated electro-
magnetic waves from the antennas into the breast can be improved
by factors of 3.3 and 2.6, respectively.
Index Terms—Average specific absorption rate (ASAR), bio-
logical tissues, breast cancer detection, dual-polarization, flexible
antenna, single polarization, wideband antenna.
I. INTRODUCTION
THERE has been a great demand for a new nonionizing, re-
liable, cost-effective, and comfortable approach for breast
screening. One such possibility is using microwave imaging
(MWI) methods for early breast cancer screening and diagno-
sis, as a complementary technology to the current method of
X-ray mammography [1]. Two significant studies on ex vivo
tissues report a dielectric contrast between the normal and ma-
lignant tissues in the microwave range. This reported contrast
is the basis for the hypothesis that one can form a microwave
scattering map of the breast tissues, from which the presence of
a malignancy can be inferred [2], [3].
Manuscript received October 28, 2014; revised February 13, 2015, March 24,
2015, and April 25, 2015; accepted April 30, 2015. Date of publication May 19,
2015; date of current version September 16, 2015. This work was supported in
part by the Natural Sciences and Engineering Research Council of Canada, by
Healthcare Support through Information Technology Enhancements. Asterisk
indicates corresponding author.
∗H. Bahramiabarghouei is with the Department of Electrical Engineering,
Laval University, Québec City, QC G1V 0A6, Canada (e-mail: hadi.bahrami.a@
gmail.com).
E. Porter, A. Santorelli, and M. Popović are with McGill University.
B. Gosselin and L. A. Rusch are with Laval University.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TBME.2015.2434956
There are two approaches in MWI: tomography based and
radar based. In microwave tomography, the electrical profile of
the breast is reconstructed by solving a nonlinear and ill-posed
inverse scattering problem [4]. A microwave scattering map can
be obtained from the differences in dielectric properties of the
breast tissues using radar-based imaging [5]. Successful imag-
ing systems of this type have been presented in [6] and [7].
Employing a multistatic array has the advantage of avoiding
the mechanical issues of a scanning antenna; thus, radar-based
breast cancer detection systems based on this idea have been
developed. An example of a successful radar system is given in
[8]. A multistatic system is able to provide more information
about the breast due to the larger number of collected signals
for a given set of sensors [6]. Further, due to the nature of
multistatic systems, they rarely require scanning of the antenna
locations as is often present with monostatic systems. Scanning
of antenna positions is undesirable, particularly for a wearable
array, as extra motorized components would be necessary, as
would methods to accurately position the antennas. Hence, it is
necessary to build even smaller antennas to allow more antennas
to be positioned around the breast simultaneously. In this way,
we maintain comprehensive radiation coverage over the tissue
and reduce the error associated with antenna position and, thus,
improve the accuracy of the system. The image quality provided
by radar-based MWI techniques is affected by the number and
efficiency of the receivers, the synthetic aperture of the antenna
array, and the bandwidth of the probing signal. It is very impor-
tant to obtain a high-performance broadband antenna. Typical
characteristics of the antenna to be used in a radar-based imaging
system are: wide impedance bandwidth, small size, repeatable
and cost-effective fabrication, and ability to efficiently couple
power to the breast [5], [8], [9].
Low frequencies provide deeper penetration (lower loss), but
the higher frequencies offer better range resolution. Smaller
antennas enable a higher number of antennas in the array and
enhance resolution. In the low-frequency band, the number of
the antennas in the array must be reduced to fit the limited
breast surface due to the required larger antenna footprint [9].
The solution to this issue is of great interest for ultrawideband
(UWB) techniques [9].
Miniature antennas in contact with biological tissues have
very different propagation behavior and return loss compared
to those in free space [10]–[13]. The characteristic impedances
of biological tissues vary considerably with frequency and from
tissue to tissue; thus, careful antenna impedance matching is
required. In order to mimic a realistic scenario, the antenna is
0018-9294 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Birla Institute of Technology & Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

2. BAHRAMIABARGHOUEI et al.: FLEXIBLE 16 ANTENNA ARRAY FOR MICROWAVE BREAST CANCER DETECTION 2517
Fig. 1. Overview of a flexible antenna array as a bra for breast cancer detection
modeled in HFSS (single-arm spiral and monopole antenna arrays).
simulated operating in contact with a stack of several layers (in-
homogeneous media) representing different biological tissues.
In this way, the antenna can provide as much energy as possi-
ble so that transmitted signals can be received with reasonable
strengths from breast tissues.
To date, only a few small broadband antennas for breast cancer
detection have been reported [9]–[16]. Planar printed monopole
antennas have been recently considered for breast cancer imag-
ing [14], [15] due to their simple structure, broadband property,
relatively small size, and ease of fabrication. However, these
antennas are not flexible but bulky. To date, different flexible
antennas have been designed for different parts of the body
(except for the breast) by taking into account the effects of bi-
ological tissues for the Industrial Scientific and Medical and
Med-Radio bands [16], [17].
This paper reports a flexible 4 × 4 monopole and single-arm
spiral UWB antenna array operating in the 2–4 GHz spectrum.
This frequency range meets bandwidth requirements of breast
cancer MWI. Further, there is an inherent tradeoff in the upper
and lower frequency limits. At upper frequencies, the resolution
is higher and smaller antennas sizes are possible, with a cost
of high attenuation in the tissues. To date, there is no fixed
bandwidth that is proven best for MWI. However, our studies
have shown that the 2–4 GHz works for our system and our
particular problem [18]. This study is a modified version of
an existing system model [18], with the main difference of
using a flexible antenna array instead of the bulky, fragile, and
expensive traveling wave tapered and loaded transmission line
4 × 4 antenna array.
To the best of our knowledge, a flexible antenna array for
breast cancer screening has not been presented in the literature to
date. Use of a flexible antenna array that mimics a bra has several
advantages, i.e., light weight, low cost, and ease of fabrication
and installation. Sixteen UWB flexible antennas with the same
structure are placed on the same substrate to compose an antenna
array that is expected to form the core of a multistatic imaging
system as shown in Fig. 1.
In Section II, we present the simulated dielectric tissues used
as an inhomogeneous breast media for 3-D simulations in a
high-frequency structure simulator (HFSS) to capture the real
Fig. 2. Multilayer inhomogeneous model of the breast for designing the flex-
ible antenna array in HFSS.
behavior of the antenna propagation in proximity of the breast,
and to design the flexible monopole and the single-arm spi-
ral antenna array. In Section III, we describe the antenna de-
sign methodology for single and dual polarization. Section IV
presents measurement results. Antenna array performance was
measured on a phantom that is representative of actual biologi-
cal tissues. The antenna performance is investigated in Section
V. In Section VI, the use of a reflector is suggested to improve
the penetration of the propagated electromagnetic waves into the
breast. Furthermore, the maximum allowed transmitted power
based on simulated average specific absorption rate (ASAR) for
different positions of the antennas in the array is also discussed.
Finally, conclusions are drawn in Section VII.
II. INHOMOGENEOUS BREAST MODELING
A miniature antenna in contact with biological tissues will
have very different propagation behavior than one in free space
due to the electrical characteristics of biological tissues [19].
The antennas must be designed taking into account the impact
of the proximity to biological tissues which are placed in its near
field. The multiple biological tissues in breast cancer detection
have varying conductivity and dielectric constants leading to
complex RF interaction [13]. The breast as a communication
media is modeled by several biological tissues and each biolog-
ical tissue is defined as a dispersive dielectric in a homogeneous
medium using three electrical parameters: relative permittivity,
loss tangent, and mass density.
By stacking several homogeneous layers, the inhomogeneous
environment is modeled with the HFSS software [19]. The mul-
tilayer model that is used to design the antenna array includes
skin, fat, gland, and muscle as developed in [20], and is shown
in Fig. 2. The dielectric properties (namely, the relative permit-
tivity and conductivity) for these tissues can be found in [2] and
[21] for the 2–4 GHz band. The loss tangent quantifies inherent
dielectric dissipation when interacting with an electromagnetic
wave. The mass density, i.e., the mass of each tissue per volume
unit, is reported in [20] for different breast tissues; this parameter
Authorized licensed use limited to: Birla Institute of Technology & Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

3. 2518 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 10, OCTOBER 2015
TABLE I
PARAMETERS OF BREAST TISSUES
Tissue Thickness (mm) Inner radius (mm) Outer radius (mm) Mass Density (kg/m3)
Skin 2 68 70 1010
Fat 8 60 68 928
Gland 120 0 60 1035
Muscle 8 0 0 1040
is needed for calculation of ASAR. The simulated geometrical
parameters of the breast model are presented in Table I.
To design the wearable antenna, the first step is designing
a 50-Ω transmission line connected to the antenna propagator,
while taking into account the effect of biological tissues. The
next step is finding initial dimensions of the propagator to meet
required specifications (i.e., |S11| below −10 dB) at a single
frequency. Next, the propagator dimensions are optimized to
extend the antenna bandwidth to achieve the bandwidth of in-
terest. One important point, during the design of the antennas,
is accommodating the variation of breast tissues from person to
person. After designing the antenna for the defined thicknesses
and the electrical properties of the different tissues included in
the breast model, the S11 of the antenna is then checked to con-
firm that varying the thicknesses (different size of breasts) and
electrical properties (∼20%) still result in a final |S11| below
−10 dB. Details of design are given in the next section.
III. FLEXIBLE ANTENNA ARRAY
Wearable UWB antennas that touch the breast are subject to
constraints that make their design difficult. These constraints,
among others, are: 1) small footprint, 2) biocompatibility, 3)
high bandwidth, and 4) light weight to be placed on the
breast comfortably (a flexible substrate has this feature). Pla-
nar monopole (single polarization) and spiral (dual polarization)
antennas have the potential to meet these constraints.
Our single-polarization and dual-polarization antennas are
realized as monopole microstrip and single-arm spiral anten-
nas, respectively, and are covered with a biocompatible material
(Kapton polyimide). For all antennas, the feed circuit is a copla-
nar waveguide (CPW) transmission line having 50-Ω impedance
over the 2–4 GHz frequency range.
We use 0.05-mm Kapton polyimide as the biocompatible
substrate for the flexible antennas. The relative permittivity of
Kapton polyimide is 3.5. For both antennas, a superstrate layer
identical to their substrate layers covers the metal antenna traces.
This superstrate layer completely isolated the antenna from the
biological tissues making the antennas fully biocompatible. In
Appendix I, we provide further background related to the envi-
sioned application for the here designed array.
A. Single-Polarization Antennas
There are several single-polarization monopole antennas de-
signed for breast cancer in UWB existing in the literature
[14], [15]. Recently, we designed an implanted and wearable
monopole antenna for neural recording [19]. These antennas
were designed on inflexible substrates that are ill adapted to the
Fig. 3. Induced current on the flexible single-polarization antenna.
curvature of the breast. Microwave substrates that are going to
be used as wearable on the breast should be biocompatible and
as flexible and as soft as possible. In [22], a type of Kapton
polyimide substrate that is utilized in this paper has the two
aforementioned properties. We revisit these designs for breast
tissues by incorporating a flexible substrate to achieve a suitable
antenna array for breast cancer detection.
The first step is to design a transmission line (as shown in
Fig. 10) having 50-Ω impedance over the frequency range of
interest while being fed with a vertical SMA connector. The
dimensions of the transmission line are a function of the elec-
trical properties of the substrate as well as the environment
surrounding the substrate. The transmission width is calculated
for the flexible antenna while it is in contact with the breast. In
the single-polarization antenna, current induced on the antenna
should follow a single axis of propagation.
The second step of designing the antenna is designing the
propagator. The antenna propagator with a rectangular shape
has the highest linearity in induced current. As the rectangu-
lar propagator is much wider than the transmission line, the
return loss will tend to be narrowband. Therefore, the propaga-
tor is tapered to couple to the transmission line at the widest
bandwidth [19]. In order to achieve reflection coefficients (S11)
below −10 dB in the entire frequency range of interest, the
physical dimensions of the antenna are optimized. Fig. 3 shows
the induced current on the antenna. This figure clearly shows
that most of the induced currents on the surface of the antenna
are aligned in the same direction. The S11 of the designed single
antenna should be checked at different positions in the arrays
that are shown in Fig. 1 to confirm that it remains below −10 dB.
It has been shown that truncating the transmission line ground
plane into a staircase shape results in better return loss [19]. By
optimizing the width and the length of the staircase, we can
reach the desired return loss in our frequency range of interest.
In conclusion, to achieve |S11 | below −10 dB, we have optimized
the physical dimensions of the antenna.
B. Dual-Polarization Antennas
Due to the polarization selectivity of breast tissues, we are
interested in a design that covers both X and Y polarizations.
Although such a design is more complex, it may add information
to the collected signal by enabling recording of backscattered
signal of two polarizations.
Authorized licensed use limited to: Birla Institute of Technology & Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

4. BAHRAMIABARGHOUEI et al.: FLEXIBLE 16 ANTENNA ARRAY FOR MICROWAVE BREAST CANCER DETECTION 2519
Fig. 4. Induced surface current on the flexible dual-polarization antenna.
To have a dual-polarization antenna, we use spiral shape to
have induced surface current in both X- and Y-axes. To de-
sign a spiral single-arm antenna, as in the previous section, we
begin with tuning the CPW transmission line to have a 50-Ω
impedance in the frequency range of interest. We incorporate a
spiral propagator structure that induces a current equal in both
the horizontal and the vertical axes of the antenna, as shown in
Fig. 4. The spiral structure of the propagator consists of sev-
eral rectangle-shaped pieces of copper placed on the X- and the
Y-axes with shared current.
To have equal radiated power in the X- and Y-directions, the
total length of the rectangles in each direction must be equal.
We achieve a |S11| below −10 dB in the frequency range of
interest (2–4 GHz) by matching the spiral propagator to the
50-Ω transmission line. To improve S11 bandwidth (i.e., keeping
it below −10 dB for a wider frequency range), the width of
each rectangle in the spiral propagator is tuned. We placed the
rectangular shapes as close as possible to decrease the antenna
size. Fig. 4 shows the induced antenna current. The current is
distributed almost equally in the X- and Y-directions.
IV. MEASUREMENT RESULTS
A. Measurement Setup
In order to verify our simulation results, we fabricated the
arrays on the flexible substrate. Fig. 5 shows the fabricated
arrays which were measured on a phantom. The phantom is
of stable rubber construction based on the design presented
in [23]. It consists of a thin-skin-mimicking layer filled with
adipose-mimicking material. The inside of the phantom is ho-
mogeneous. The dielectric properties are matched to those of
the actual fat and skin tissues. A complete description of the
phantom construction and dielectric property measurements are
provided in [24]. Appendix II summarizes the geometrical pa-
rameters of the designed antennas. A 50-Ω SMA connector
connects the antenna to the measurement equipment. The S11
of different antenna positions are measured with antenna po-
sitions on the array (1–4) shown in Fig. 1. The transmission
coefficient S21 of the antenna pairs between antenna 3 in array
1 and antenna 1–4 in array 2 and 3 shown in Fig. 1 are also mea-
sured. Measurements are made using an Agilent PNA-L Net-
work Analyzer N5232A, 300 kHz–20 GHz for both single and
dual-polarization antennas.
Fig. 5. S-parameter measurement setup, shown here for the spiral antenna.
Same arrangement is used for the monopole antenna array. (a) Antenna array
placed on the dielectric tissue-mimicking phantom, and (b) top view of the array,
identifying the nipple location; view of the individual antenna element with the
included SMA connector.
B. S-Parameter Results
Fig. 6(a) shows the measured S11 of the single-polarization
and Fig. 6(b) shows the measured S11 of the dual-polarization
antennas for different positions of the antennas in the array as
shown in Fig. 1. S11 results show good impedance matching
(|S11| below −10 dB) across the 2–4 GHz frequency range for
each position. Results demonstrate that the flexible antenna is
insensitive to bending and, thus, is a good candidate for use in
the array configuration.
The amplitude and phase of S21 (transfer function) were mea-
sured for several different antenna pairs within the monopole
antenna array. More specifically, the S21 between antenna 3 of
Authorized licensed use limited to: Birla Institute of Technology & Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

5. 2520 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 10, OCTOBER 2015
Fig. 6. Measured S11 for different positions of the antennas in the array
(a) monopole antenna, and (b) single-arm spiral antenna.
array 1 and antennas 1–4 of arrays 2 and 3 (array layouts and
antenna numbering as shown in Fig. 5) is shown in Fig. 7. The
measured S21 results between antenna 3 of array 1 and antennas
1–4 of arrays 2 and 3 for single-arm spiral antenna array are
plotted in Fig. 8. From the S21 results, we conclude that an-
tennas that are further apart communicate with higher insertion
loss. Additionally, because the biological tissues have higher
loss in higher frequencies, the S21 results follow this behav-
ior (less insertion loss in lower frequency). From the measured
S-parameters, it is clear that there are different channel transfer
functions between each transmit–receive antenna pair.
V. ANTENNA PERFORMANCE
In breast cancer applications, the close proximity of the trans-
mitter, receiver antennas, and biological tissues leads to com-
plex propagation and antenna response that is dependent on the
electrical properties of the biological tissues. In this section, we
discuss the efficiency of antennas from a wireless link viewpoint.
We find the received power when a specific signal is transmitted
for each antenna. Furthermore, we investigate fidelity, varia-
tion of group delay versus frequency range of interest, and a
figure-of-merit (FoM) for this application.
A. Power Efficiency and Group Delay
We investigate power efficiency and group delay of the dif-
ferent antenna pairs via simulation by using the measured result
in Figs. 7 and 8 as the received power when our antenna is used
with a common UWB pulse shape. The excitation pulse consists
of a Gaussian-modulated sinusoidal waveform mathematically
Fig. 7. Measured S21 (a) and (b) S21 amplitude of monopole type between
antenna 3 of array 1 and antennas 1–4 of array 2 and 3, respectively, (c) and
(d) S21 phase of monopole one between antenna 3 of array 1 and antennas 1–4
of array 2 and 3.
described by Bahramiabarghouei [25]
V (t) = sin[2πf0 (t − t0)] × e−
( t −t 0 ) 2
2 τ 2
(1)
where f0 = 3 GHz, τ = 450 ps, and t0 = 0 s. This pulse has a
frequency spectrum centered at 3 GHz and a 40 dB bandwidth
from 2 to 4 GHz.
Authorized licensed use limited to: Birla Institute of Technology & Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

6. BAHRAMIABARGHOUEI et al.: FLEXIBLE 16 ANTENNA ARRAY FOR MICROWAVE BREAST CANCER DETECTION 2521
Fig. 8. Measured S21 (a) and (b) S21 amplitude of spiral type between antenna
3 of array 1 and antennas 1–4 of array 2 and 3, respectively (c) and (d) S21
phase of spiral one between antenna 3 of array 1 and antennas 1–4 of array 2
and 3.
MATLAB was used to characterize the performance of the
wireless link between the antenna pairs using the measured S-
parameters of each TX and RX antenna pair. To calculate the
received signal, the spectrum of the pulse is multiplied by S21.
Tables II and III show the portion of the total power which is
received (power efficiency) by the RX antennas. The power level
TABLE II
ANTENNA PERFORMANCE FOR MONOPOLE ANTENNA ARRAY
Pairs Ant. 3 to Fidelity Power Efficiency (%) Group Delay Variation (ps) FoM
Array 2 Ant. 1 0.94 0.001 6.2 0.94
Ant. 2 0.87 0.0009 11.4 0.78
Ant. 3 0.92 0.003 9.1 2.7
Ant. 4 0.97 0.013 7.1 12.6
Array 3 Ant. 1 0.67 0.0004 7 0.27
Ant. 2 0.88 0.00093 9.7 0.81
Ant. 3 0.86 0.0014 10 1.2
Ant. 4 0.95 0.01 10.6 9.5
TABLE III
ANTENNA PERFORMANCE FOR SPIRAL ANTENNA ARRAY
Pairs Ant. 3 to Fidelity Power Efficiency (%) Group Delay Variation (ps) FoM
Array 2 Ant. 1 0.85 0.00035 12.5 0.3
Ant. 2 0.8 0.00034 20 0.27
Ant. 3 0.92 0.00054 6.9 0.5
Ant. 4 0 .84 0.002 4.6 1.7
Array 3 Ant. 1 0.88 0.0003 12.5 0.26
Ant. 2 0.7 0.0012 13 0.84
Ant. 3 0.61 0.00032 19.5 0.2
Ant. 4 0.68 0.0001 11.6 0.07
varies with antenna pair due to the different frequency response
associated with the path of each individual antenna pair.
In Tables II and III, the variation of group delay is calculated
for different antenna pairs based on the measured phase of S21 in
Figs. 7 and 8. As shown, different antenna pairs have different
variations and create different distortions on the transmitted
pulse.
B. Fidelity Factor of the Wireless Links Between the
Antenna Pairs
System performance is optimized when the transmitted wave-
form is received with no distortion. Simple design goals for
amplitude and group delay attempt to minimize the distortion.
Unlike narrowband antennas, UWB antennas can significantly
alter transmitted pulses, due to considerably different behavior
across the wide frequency band [25]. The fidelity factor of the
wireless link captures the similarity between the ideal expected
output waveform of an antenna and the actual radiated wave-
form. The fidelity factor is defined as the maximum cross corre-
lation between the ideal and the actual radiated waveforms when
both waveforms are normalized by their energies [25]. Fidelity
varies between “0 and 1,” with “1” representing the minimum
and “0” representing the maximum distortion introduced by the
antenna.
Let r(t) be the received pulse for an ideal link for a given ex-
citation pulse shape. Let Sr (t) be the actual received waveform
for that excitation pulse shape. The fidelity is defined as the
following, assuming r(t) and Sr (t) have normalized energies
[25]
F = max
τ
+∞
−∞
r(t) × Sr (t + τ) dt. (2)
Authorized licensed use limited to: Birla Institute of Technology Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

7. 2522 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 10, OCTOBER 2015
We examined the different antenna pairs which were pre-
sented in Figs. 7 and 8. We derive Sr (t) by calculating the
inverse Fourier transform of the multiplication of the pulse spec-
trum of (1) by the measured S21 of the antenna pairs. The fi-
delity factors calculated from (2) are given in Tables II and III
showing single-polarization antennas perform better than dual-
polarization antennas in terms of fidelity.
C. Figure of Merit
To compare the designed antennas in this application, we
propose a FoM. Our FoM is based on parameters that play
an important role in the performance of the wireless links: 1)
Pe: power efficiency (%), and 2) F: fidelity factor. The FoM is
defined as
FoM = Pe × F × 1000. (3)
In Tables II and III, FoM is calculated for the links presented
in Figs. 7 and 8. The results show that the single-polarization
antennas have a higher FoM. The reason for not including the
variation of group delays in the FoM is that the fidelity includes
both distortions that are caused by 1) the phase and 2) the
amplitude of the frequency response of the wireless links.
VI. IMPROVING PENETRATION OF PROPAGATED
ELECTROMAGNETIC WAVES INSIDE BREAST AND MAXIMUM
ALLOWED TRANSMITTED POWER
In this section, we show that by using a reflector for the an-
tenna arrays, it is possible to improve penetration of propagated
electromagnetic waves into the breast from the wearable anten-
nas, which causes an improvement in power efficiency of the
links between the antennas, and thus potentially better image
resolution. We also calculate the maximum power that is al-
lowed to be sent by antennas in different positions in the array
shown in Fig. 5, with and without use of a reflector.
A. Improving Penetration of Propagated
Electromagnetic Waves
Electromagnetic waves are used to transport information
through a wireless medium or a guiding structure, from one
point to the other. The quantity used to describe the power den-
sity associated with an electromagnetic wave is the time average
Poynting vector (average power density) [26]:
Wav (X, Y, Z) =
1
2
[E × H∗
] (4)
where E and H are the peak values of instantaneous electric
field intensity (V/m) and instantaneous magnetic field intensity
(A/m), respectively.
The real part of (1) is averaged over the propagated power
density, and the imaginary part represents the reactive (stored)
power density associated with the electromagnetic fields. The
real part is responsible for delivering power from one antenna to
another antenna. Fig. 9 shows the real part of the Poynting vec-
tor when antenna 4 is propagating in the ZX plane for monopole
and spiral antenna arrays. This figure shows that by using re-
Fig. 9. Real part of the Poynting vector propagating in the ZX plane for
antenna 4: without reflector (left side) and with reflector (right side) for (a)
different configurations for reflector, (b) monopole for configuration 1, and (c)
spiral antenna arrays for configuration 1; color map applies to all cases.
flector behind the antenna arrays, it is possible to improve the
penetration of propagated electromagnetic inside the breast.
The reflector should be placed far enough so that it does not
affect the S11 of the antennas. By placing the reflector further
than 1 cm from the antennas, S11 remains unchanged. HFSS
results show that for the spiral antenna using the reflector, the
real part of the Poynting vector in the position of (x = 0, y = 0,
and z = 0) improves by a factor of 3.3 for configuration 1 and
3.7 with configuration 2, and similarly for the monopole, by
a factor of 2.6 for configuration 1 and 3.8 for configuration 2.
This significant penetration of electromagnetic waves inside the
breast improves the power efficiency of the link and the ability
to detect the presence of a tumor. When using a reflector, the
isolation between closest antennas is still more than 25 dB. The
isolation for closest antennas for both cases without reflector
and with reflector is more than 25 dB.
As we explained in Appendix III, we believe the reflectors
would not make this imaging scenario more complicated.
B. Maximum Allowed Transmitted Power by the Antennas
ASAR describes the electromagnetic energy that is absorbed
in biological tissues and is a critical parameter for assessing
the tissue safety of wireless communications in bio applica-
tions. The peak 1-g ASAR spatial distribution versus frequency
Authorized licensed use limited to: Birla Institute of Technology Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

8. BAHRAMIABARGHOUEI et al.: FLEXIBLE 16 ANTENNA ARRAY FOR MICROWAVE BREAST CANCER DETECTION 2523
TABLE IV
MAXIMUM AVERAGED TRANSMITTED POWER FOR MONOPOLE AND SPIRAL
ANTENNAS IN VARIOUS ARRAY POSITIONS
Antenna Monopole Monopole with reflector Spiral (mW) Spiral with reflector
(mW) (mW) (configuration 1) (mW) (configuration 1)
Ant. 1 4.1 4 3.2 3.1
Ant. 2 4.1 4.1 3.2 3.1
Ant. 3 4.0 4.0 3.2 3.1
Ant. 4 3.9 4 3.1 3.1
Fig. 10. Geometric parameters (a) flexible monopole antenna, and (b) flexible
single-arm spiral antenna (the red dotted outline indicates the propagator region
and the blue indicates the transmission line).
TABLE V
GEOMETRIC PARAMETERS OF THE ANTENNAS
Parameters Monopole Antenna (mm) Spiral Antenna (mm)
Polarization Single Dual
A 1 1
B 1.7 9
C 5 8.7
D 8.3 6.2
E 5.8 2.7
F 1.3 8.7
G 4.6 5.5
H 4.1 3
I 1.3 0.8
J 6.5 6.5
K 8.5 .3
L 13 3
M 3.5 1.4
N 5.5 2
O 7.3 1.1
P 15 2.1
Q 15 10.7
R 20 20
S 20 20
is simulated in HFSS for all four positions of the antennas in-
side the array for the monopole and spiral antennas, with and
without use of a reflector. The American National Standards In-
stitute (ANSI) limitations specify a maximum peak 1-g ASAR
of 1.6 W/kg [27].
Based on the maximum simulated ASAR of tissues in HFSS,
the maximum allowed transmitted power from each position of
the antennas in the array is calculated. The results are presented
in Table IV. Sending more power can damage the biological
tissues. The HFSS results show that most radiated power is
absorbed by the skin which is in contact with the antenna arrays.
From Table IV, the maximum averaged transmitted power for
the monopole and the spiral antenna array are around 4 and
3.1 mW, respectively. We observe that the reflector does not
significantly change this power, and neither does changing the
antenna position.
VII. CONCLUSION
We presented a methodology for designing wearable single-
and dual-polarization antennas on a flexible substrate for breast
cancer detection operating over 2–4 GHz frequency bands. The
new array improves on previous microwave radar imaging sys-
tems in that it is highly flexible, cost effective to fabricate, and
light weight. Simulations were carried out with HFSS, exploit-
ing a layered (inhomogeneous) model with different dielectric
constants and loss tangents to capture the effect of surround-
ing tissues. To verify the validity of our model and the antenna
design procedure, the fabricated arrays were measured on a
phantom that is representative of actual biological tissues.
Measurements confirmed that the proposed antenna achieved
our design goals, validating our antenna design methodology
and our biological tissue modeling. Finally, it has been shown
that by using a reflector for the arrays, penetration of the propa-
gated electromagnetic waves can be significantly improved. For
both arrays, we determined the maximum power allowed to be
transmitted from the wearable antenna, by taking into account
the limitation imposed by the ANSI. For our future work, we
will investigate the performance of our wearable arrays (sin-
gle and dual polarizations) on patients, and determine which
antenna is the most practical for our application. We will then
integrate the wearable arrays into a bra-like prototype for im-
proved microwave breast imaging.
APPENDIX I
A home-use breast cancers detection system would be most
advantageous for women who have already been identified as
being at high risk of developing breast cancer. Their breast health
could be checked very frequently, and if there are abnormalities,
they can be identified at an early stage. A home-use detection
system could also be used for monitoring the treatment progress.
Statistics show that between 1/7 to 1/9 women will develop
breast cancer in the United States [28]. If a cancer is detected
in early stage, there is a 99% survival rate. If it has spread to
distant areas of the body, the survival rate drops to only 24%.
But currently, only 61% of breast cancer cases are diagnosed
at the early stage [28]. Home use detection could foster earlier
detection enabling more successful treatment and higher
likelihood of survival.
Currently breast cancer is detected with mammograms. Mam-
mography is usually done 1–2 times per year in the US, but only
for older women/high risk women [28]. For fast growing can-
cers, the one year interval is ineffective. A home-use detection
system would allow frequent monitoring that could be gathered
in a larger database of past breast scans to determine tumor
growth patterns in the patient. Adoption will depend heavily on
Authorized licensed use limited to: Birla Institute of Technology Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

9. 2524 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 10, OCTOBER 2015
availability of a cost-effective device. We are working toward
a custom cancer detection system, with equipment on-chip or
on-board.
APPENDIX II
The geometrical parameters of the antennas are defined in
Fig. 10 and presented in Table V.
APPENDIX III
In this appendix, we explain our methodology for imaging
and tumor detection. For imaging, we reconstruct an image of
the scattering regions of the breast. A short pulse in the time
domain is generated and transmitted into the breast tissues.
It scatters at all interfaces. The antennas collect the scattered
waves at different locations. Then, the collected signals are
used in what is called a delay-and-sum algorithm [29]. In our
work, we use differential signals. So, we take a breast scan
with a healthy breast, and then with a breast with a tumor. We
subtract the collected signals from each breast scan, obtaining
“differential signals.” The differential signals are the input
into the imaging algorithm. Adding reflectors to the antennas
will not add complication to our specific imaging scenario:
since we use differential signals as the input to the imaging
algorithm, any effects that are present in both signals (such as
wave changes caused by a reflector) will be cancelled out. The
same is true for any other sources of noise or clutter that are
inherent to the system and consistent for all breast scans.
For detection, we use signal processing to have a binary out-
put of “healthy” or “cancerous.” In this way, the location of the
tumor is not determined, only whether one exists or not. This is
done using machine learning [30]. We have used support vector
machines and linear discriminant analysis. The literature has
suggested that they offer promise for microwave health appli-
cations [31]. We perform breast scans on many healthy breasts
and many breasts with tumor. The collected signals are used to
train a machine learning algorithm, so we give the algorithm all
of our collected signals, but we also indicate which data set they
belong (i.e., healthy or cancerous). Then, a new breast scan that
is unknown if it is healthy or cancerous can be tested with the
algorithm and it will find (with some small error) if it is healthy
or cancerous. For our machine learning algorithms, we also use
differential signals and the reflectors which were presented in
Section VI-A would not be an issue.
ACKNOWLEDGMENT
The authors would like to thank P. Tome (Flex Circuit Product
Manager of Epec Engineered Technologies) for his significant
help during the antenna fabrication process.
REFERENCES
[1] E. C. Fear et al., “Enhancing breast tumor detection with near-field imag-
ing,” IEEE Microw. Mag., vol. 3, no. 1, pp. 48–56, Mar. 2002.
[2] M. Lazebnik et al., “A large-scale study of the ultrawideband mi-
crowave dielectric properties of normal, benign and malignant breast tissue
obtained from cancer surgeries,” Phys. Med. Biol., vol. 52, pp. 6093–6115,
2007.
[3] T. Sugitani et al., “Complex permittivities of breast tumor tissues obtained
from cancer surgeries,” Appl. Phys. Lett., vol. 104, no. 253702, pp. 1–5,
2014.
[4] S. P. Poplack et al., “Electromagnetic breast imaging: Results of a pi-
lot study in women with abnormal mammograms,” Radiology, vol. 243,
pp. 350–359, 2007.
[5] D. Gibbins et al., “A comparison of a wide-slot and a
stacked patch antenna for the purpose of breast cancer detec-
tion,” IEEE Trans. Antennas Propag., vol. 58, no. 3, pp. 665–674,
Mar. 2010.
[6] J. Bourqui et al., “A prototype system for measuring microwave frequency
reflections from the breast,” Int. J. Biomed. Imag., vol. 2012, pp. 851234–
1–851234-12, 2012.
[7] T. Grzegorczyk et al., “Fast 3-D tomographic microwave imaging for
breast cancer detection,” IEEE Trans. Med. Imag., vol. 31, no. 8,
pp. 1584–1592, Aug. 2012.
[8] M. Klemm et al., “Microwave radar-based differential breast cancer imag-
ing: Imaging in homogeneous breast phantoms and low contrast scenar-
ios,” IEEE Trans. Antennas Propag., vol. 58, no. 7, pp. 2337–2344, Jul.
2010.
[9] T. Sugitani et al., “A compact 4 ∗4 planar UWB antenna array for 3-D
breast cancer detection,” IEEE Antenna Wireless Propag. Lett., vol. 12,
pp. 733–736, Jun. 2013.
[10] Y. Wang et al., “Novel compact tapered microstrip slot antenna for mi-
crowave breast imaging,” in Proc. IEEE Int. Symp. Antennas Propag.,
Spokane, WA, USA, 2011, pp. 2119–2122.
[11] L. Wang et al., “Design of ultra-wideband MIMO antenna for breast tumor
detection,” Int. J. Antennas Propag., vol. 2012, pp. 180158–1–180158-7,
2012.
[12] A. Eesuola et al., “Novel ultra-wideband directional antennas for mi-
crowave breast cancer detection,” in Proc. IEEE Int. Symp. Antennas
Propag., 2011, pp. 90–93.
[13] A. M. A. Shaheen et al., “Modeling and optimizing ultra
wideband antennas for microwave breast cancer detection,” in
Proc. IEEE 9th Int. Multi-Conf. Syst., Signals Devices, 2012,
pp. 1–6.
[14] A. Abbosh et al., “Design of ultra-wideband planar monopole antennas
of circular and elliptical shape,” IEEE Trans. Antennas Propag., vol. 56,
no. 1, pp. 17–23, Jan. 2008.
[15] S. M. Latif et al., “A planar ultra-wideband elliptical monopole antenna
with reflector for breast microwave imaging,” Microw. Opt. Technol. Lett.,
vol. 56, pp. 808–813, 2014.
[16] W. Zheyu et al., “Flexible textile antennas for body-worn communication,”
in Proc. IEEE Int. Workshop Antenna Technol., Mar. 5–7, 2012, pp. 205–
208.
[17] R. Alrawashdeh et al., “Flexible meandered loop antenna for implants in
MedRadio and ISM bands,” Electron. Lett., vol. 49, no. 24, pp. 1515–1517,
2013.
[18] E. Porter et al., “Time- domain multi-static radar system for mi-
crowave breast screening,” IEEE Antennas Wireless Propag. Lett., vol. 12,
pp. 229–232, Feb. 2013.
[19] H. Bahrami et al., “Biological channel modeling and implantable UWB
antenna design for neural recording systems,” IEEE Trans. Biomed. Eng.,
vol. 62, no. 1, pp. 88–98, Jan. 2014.
[20] P. Ghanbari et al., “Finite element analysis of tissue electro-permeability
through the application of electric pulses,” Bioeng. Biomed. Sci., vol. 3,
pp. 1–7, 2013.
[21] S. Gabriel et al., “The dielectric properties of biological tissues: II. Mea-
surements in the frequency range 10 Hz to 20 GHz,” Phys. Med. Biol.,
vol. 41, pp. 2251–2269, 1996.
[22] H. R. Khaleel et al., “A compact polyimide-based UWB antenna for
flexible electronics,” IEEE Antennas Wireless Propag. Lett., vol. 11,
pp. 564–567, May 2012.
[23] J. Garrett and E. Fear, “Stable and flexible materials to mimic the dielectric
properties of human soft tissues,” IEEE Antennas Wireless Propag. Lett.,
vol. 13, pp. 599–602, Mar. 2014.
[24] A. Santorelli et al., “Image classification for a time-domain microwave
radar system: Experiments with stable modular breast phantoms,” pre-
sented at the Eur. Conf. Antennas Propagation, Lisbon, Portugal, Apr.
12–17, 2015.
[25] H. Bahrami, S. A. Mirbozorgi, R. Ameli, L. A. Rusch, B. Gosselin,
“Flexible, polarization-diverse UWB antennas for implantable neural
recording systems,” IEEE Trans. Biomed. Circuits Syst., 2015, doi:
10.1109/TBCAS.2015.2393878
[26] C. A. Balanis, Advanced Engineering Electromagnetics. New York, NY,
USA: Wiley, 1989.
Authorized licensed use limited to: Birla Institute of Technology Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.

10. BAHRAMIABARGHOUEI et al.: FLEXIBLE 16 ANTENNA ARRAY FOR MICROWAVE BREAST CANCER DETECTION 2525
[27] IEEE Standard for Safety Levels With Respect to Human Exposure to Radio
Frequency Electromagnetic Fields, 3 kHz to 300 GHz, IEEE Standard
C95.1-2005, 2006.
[28] (2014). American Cancer Society. Facts figures 2014. [Online]. Atlanta,
GA, USA. Available: http://www.cancer.org
[29] H. B. Lim et al., “Confocal microwave imaging for breast cancer de-
tection: Delay-multiplyand-sum image reconstruction algorithm,” IEEE
Trans. Biomed. Eng., vol. 55, no. 6, pp. 1697–1704, Jun. 2008.
[30] A. Santorelli et al., “Investigation of classifiers for tumor detection with an
experimental time-domain breast screening system,” Prog. Electromagn.
Res., vol. 144, pp. 45–57, 2014.
[31] D. Byrne et al., “Support vector machine-based ultrawide-band breast
cancer detection system,” J. Electromagn. Waves Appl., vol. 25, pp. 1807–
1816, 2011.
Hadi Bahramiabarghouei (S’12) received the B.S.
and M.S. degrees in electrical engineering from the
Isfahan University of Technology, Isfahan, Iran, and
Tarbiat Modares University, Tehran, Iran, in 2005 and
2007, respectively. He is currently working toward
the Ph.D. degree at Biomedical Microsystems Lab,
Department of ECE, Université Laval, Quebec City,
QC, Canada.
His research interests include RFIC, antenna and
propagations, wireless and optical communications,
and photonic integrated circuit.
Emily Porter (S’11) received the B.Eng. and M.Eng.
degrees in electrical engineering in 2009 and 2010,
respectively, both from McGill University, Montreal,
QC, Canada, where she is currently working toward
the Ph.D. degree in electrical engineering.
Her field of Ph.D. research is applied and computa-
tional electromagnetic Her research interests include
medical applications of microwaves, flexible anten-
nas, and the design of realistic breast phantoms and
models. Her current research is on the development
of a wearable prototype for breast health monitoring
using microwave radar.
Adam Santorelli (S’11) received the B.Eng. degree
in electrical engineering and the M.Eng. degree in
electromagnetics both from McGill University, Mon-
treal, QC, Canada, in 2010 and 2012, where he is
currently working toward the Ph.D. degree in electri-
cal engineering.
His research interests include biomedical appli-
cations with the primary goal of increasing the ac-
cessibility to technology in order to improve diag-
nosis, including the optimization and miniaturization
of off-the-shelf components to design and fabricate a
custom-built low-cost microwave systems.
Benoit Gosselin (S’02–M’08) received the M.Sc. and
Ph.D. degrees both in electrical engineering from
École Polytechnique de Montréal, Montréal, QC,
Canada, in 2009.
He is a mixed-signal layout Designer at PMC
Sierra Inc., Montreal, in 2009, and he was an NSERC
Postdoctoral Fellow at the Georgia Institute of Tech-
nology in 2010. He is currently an Associate Pro-
fessor at the Department of ECE, Laval University,
Quebec City, QC, where he is heading the Biomed-
ical Microsystems Lab. His research has generated
more than 70 articles published in journals and international conferences, five
book chapters, and three patent applications, which have been cited more than
600 times per Google Scholar. He organized and chair sessions in several inter-
national conferences, and he served on the Technical Committees of the IEEE
BIOCAS, the IEEE NEWCAS, the IEEE EMBC, and the IEEE ISCAS. He
regularly serves as a Referee for renowned journals and major conferences in
the area of circuits and systems and biomedical engineering. He is the Founder
and the Chair of the CAS/EMB Chapter and the Vice Chair of the Computer
Chapter of the IEEE Quebec Section. He is a Member of the Executive Com-
mittee and a Representative Member on the Board of Directors for Université
Laval of the Microsystems Strategic Alliance of Québec, which regroups 40
researchers and more than 200 students involved in microsystems research.
His research interests include VLSI circuits for bioinstrumentation, wireless
biosensing, implantable electronics, brain–computer interfaces and low-power
analog/mixed-mode integrated circuits design.
Milica Popovic´ (S’95–M’01–SM’07) received the
B.Sc. degree from the University of Colorado, Boul-
der, CO, USA, in 1994, and the Master’s and Ph.D.
degrees in electrical engineering from Northwestern
University, Evanston, IL, USA, in 1997 and 2001,
respectively.
Since 2001, she has been with the Department
of Electrical and Computer Engineering at McGill
University, Montreal, QC, Canada, where she cur-
rently holds the Associate Professorship. She teaches
courses on electromagnetic fields and waves, and an-
tennas and propagation. Her research interests, in large part, include biomedical
applications of electromagnetic theory. She has published works on mobile
device dosimetry, antenna design, and, most dominantly, on numerical and ex-
perimental analysis of breast health screening with low-power microwaves.
Dr. Popovic´ recently received the 2015 William and Rhea Seath Awards in
Engineering Innovation. She is an active Reviewer for a number of reputable
journals on electromagnetic applications. She is a Member of the Professional
Engineers of Ontario and a Fellow of the World Foundation for Innovation.
Leslie A. Rusch (S’91–M’94–SM’00–F’10) re-
ceived the B.S.E.E. degree (with Hons.) from the Cal-
ifornia Institute of Technology, Pasadena, CA, USA,
in 1980, and the M.A. and Ph.D. degrees in electri-
cal engineering from Princeton University, Princeton,
NJ, USA, in 1992 and 1994, respectively.
She has experience in defense, industrial and aca-
demic communications research. She was a Com-
munications Project Engineer for the Department of
Defense from 1980 to 1990. While on leave from
Université Laval, she spent two years (2001–2002)
at Intel Corporation creating and managing a group researching new wireless
technologies. She is currently a Professor at the Department of Electrical and
Computer Engineering, Université Laval, Québec City, QC, Canada, performing
research on wireless and optical communications. She has published more than
110 journal articles in international journals (90% IEEE/IEE) with wide reader-
ship, and contributed to more than 150 conferences. Her articles have been cited
more than 3900 times per Google Scholar. Her research interests include digi-
tal signal processing for coherent detection in optical communications, spatial
multiplexing using orbital angular momentum modes in fiber, radio over fiber
and OFDM for passive optical networks; and in wireless communications, op-
timization of the optical/wireless interface in emerging cloud-based computing
networks, optical pulse shaping for high-bit rate ultrawideband (UWB) systems,
and implantable medical sensors with high-bit rate UWB telemetry.
Dr. Rusch received the IEEE Canada J. M. Ham Award for Graduate
Supervision.
Authorized licensed use limited to: Birla Institute of Technology Science. Downloaded on September 20,2020 at 10:10:11 UTC from IEEE Xplore. Restrictions apply.