close

Вход

Забыли?

вход по аккаунту

?

Imaging breast tumors with microwaves: Simulation-based assessment of detection capabilities of a broadband antenna-sensor

код для вставкиСкачать
Imaging Breast Tumors with Microwaves:
Simulation-Based Assessment of Detection
Capabilities of a Broadband Antenna-Sensor
Negar Tavassolian
Department o f Electrical & Computer Engineering
McGill University
Montreal, Canada
June 2006
A thesis submitted to McGill University in partial fulfillment o f the
requirements for the degree o f Master o f Engineering
© Negar Tavassolian 2006
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Library and
Archives Canada
Bibliotheque et
Archives Canada
Published Heritage
Branch
Direction du
Patrimoine de I'edition
395 Wellington Street
Ottawa ON K1A 0N4
Canada
395, rue Wellington
Ottawa ON K1A 0N4
Canada
Your file Votre reference
ISBN: 978-0-494-28627-2
Our file Notre reference
ISBN: 978-0-494-28627-2
NOTICE:
The author has granted a non­
exclusive license allowing Library
and Archives Canada to reproduce,
publish, archive, preserve, conserve,
communicate to the public by
telecommunication or on the Internet,
loan, distribute and sell theses
worldwide, for commercial or non­
commercial purposes, in microform,
paper, electronic and/or any other
formats.
AVIS:
L'auteur a accorde une licence non exclusive
permettant a la Bibliotheque et Archives
Canada de reproduire, publier, archiver,
sauvegarder, conserver, transmettre au public
par telecommunication ou par I'lnternet, preter,
distribuer et vendre des theses partout dans
le monde, a des fins commerciales ou autres,
sur support microforme, papier, electronique
et/ou autres formats.
The author retains copyright
ownership and moral rights in
this thesis. Neither the thesis
nor substantial extracts from it
may be printed or otherwise
reproduced without the author's
permission.
L'auteur conserve la propriete du droit d'auteur
et des droits moraux qui protege cette these.
Ni la these ni des extraits substantiels de
celle-ci ne doivent etre imprimes ou autrement
reproduits sans son autorisation.
In compliance with the Canadian
Privacy Act some supporting
forms may have been removed
from this thesis.
Conformement a la loi canadienne
sur la protection de la vie privee,
quelques formulaires secondaires
ont ete enleves de cette these.
While these forms may be included
in the document page count,
their removal does not represent
any loss of content from the
thesis.
i *i
Bien que ces formulaires
aient inclus dans la pagination,
il n'y aura aucun contenu manquant.
Canada
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
The work reported in this thesis is motivated by the need for new screening techniques
for detecting early-stage breast tumors. In recent years, pulsed microwave imaging in the
gigahertz range has been suggested as a promising complementing methodology to the
currently existing detection and imaging modalities. This technique is based on
significant electrical contrast between the cancerous and healthy breast tissue in the
microwave range. To exploit this electrical contrast for imaging purposes, a broadband
trans-receiving antenna is placed near the breast surface. The antenna launches a pulse
and then collects the backscattered response, used for detection o f the potentially
underlying tumor. In our work, we examine tumor detection capabilities o f the “Dark
Eyes” antenna, recently reported in the literature and suggested as antenna o f choice for
pulsed microwave breast imaging due to its compact size, ease o f fabrication and costeffectiveness. The simulation tool, SEMCAD-X, is based on the finite-different timedomain method and is used throughout this work to construct the realistic hemi-spherical
breast model and analyze its interaction with the microwave radiated from the antenna
source.
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sommaire
Le travail presente dans cette these est motive par le besoin de nouvelles techniques de
criblage permettant de detecter les premieres phases des tumeurs du sein. Ces demieres
annees, des techniques d ’imagerie micro-onde pulsee dans la gamme des gigahertz ont
ete proposees pour complementer les modalites d ’imagerie et de detection existantes.
Ces techniques sont basees sur le contraste electrique significatif qui existe dans la
gamme des micro-ondes entre le tissu cancereux et sain. Afin d ’exploiter ce contraste
electrique pour la detection d ’images, une antenne de transport-reception a bande large
est placee pres de la surface du sein. L'antenne lance une impulsion et capte la reponse,
celle-ci etant alors utilisee pour la detection potentielle de la tumeur sous-jacente. Notre
travail rend compte des capacites de l’antenne « Dark Eyes » a detecter des tumeurs,
celle-ci ayant ete decrite recemment dans la litterature et suggeree comme antenne de
choix pour l’imagerie micro-onde pulsee du sein en raison de sa dimension compacte, de
sa facilite de fabrication ainsi qu’ a sa rentabilite. Notre outil de simulation, SEMCAD-X,
base sur la technique des differences finies dans le domaine temporel (FDTD), est utilise
tout le long de ce travail pour construire un modele hemispherique realiste du sein et
pour analyser son interaction avec l’onde micro-onde rayonnee a partir de l’antenne
source.
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
I am grateful to my supervisor, Dr. Milica Popovic for her kind support and patience
during m y work.
I would like to thank Mr. Houssam Kanj in the Electromagnetics group, who helped me
constantly, especially during the first months o f my work. He always generously shared
with me his knowledge and his most interesting ideas.
I am deeply indebted to my mother, who has not only given me endless love and support,
but has also been a symbol o f strength and integrity for me.
I wish to thank my best friend, Behnood Gholami who has always understood and cared
for me. His constant encouragement and valuable advice have given me motivation to
work hard.
The staff o f Schmid and Partner Engineering AG provided and assisted us with
SEMCAD software. We appreciate their help.
This work was supported in part by Natural Sciences and Engineering Research Council
(NSERC) o f Canada.
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Contents
1
Introduction ..........................................................................................................................1
2
Background Review ........................................................................................................... 3
2.1) Types of Breast Cancer .................................................................................... 4
3
2.2)
Overview o f Main Breast Cancer Imaging Methods ...................................5
2.2.1. Mammography ............................................................................... 5
2.2.2. Ultrasound ....................................................................................... 7
2.2.3. Magnetic Resonance Imaging (MRI) ...........................................7
2.2.4. Microwave Breast Cancer Imaging .............................................. 8
2.3)
Ultra Wideband Antennas ...............................................................................10
11
2.3.1. The Exciting Pulse .......................................
2.3.2. Resistively Loaded Antennas .......................................................11
2.3.3. The “Dark Eyes” A ntenna.............................................................12
2.4)
Finite-Difference Time-Domain Method ......................................................13
2.4.1 Maxwell’s Equations .....................................................................14
2.4.2 Discretization: the Yee Cell .......................................................... 15
2.4.3 Boundary Conditions .....................................................................17
2.4.4 Numerical Stability......................................................................... 17
the Breast Model ................................................................................................................18
3.1) Geometry and Parameters ...............................................................................18
3.2)
Simulation Outline .......................................................................................... 22
3.3)
SEMCAD-X .....................................................................................................22
4 Parametric Studies- Results and Discussion ................................................................... 24
4.1) On-Center Tumor ............................................................................................ 24
4.1.1 Effect o f Glandular Tissue on Detection Level .......................24
4.1.2 Sensitivity o f Tumor Response Levels to Electrical Parameters
o f the Central (Tumor Containing) Gland ................................26
4.2)
Presence o f an Off-Center Tumor: Effect o f Glandular Tissue on Detection
Levels ...............................................................................................................28
4.3)
Mutual Coupling ............................................................................................. 31
4.3.1 On-Center Tumor .........................................................................32
4.3.2 Off-Center tumor .........................................................................33
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
Conclusions and Directions for Future Work ............................................................... 35
References ................................................................................................................................. 37
A Related Publications.............................................................................................................41
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List o f Figures
2-1 Anatomy o f the breast, showing lymph nodes and lymph vessels [37] .....................3
2-2 (a) Mammography equipment [39] (b) Mammography procedure [40] .....................6
2-3 (a) The basic microwave imaging problem involves illuminating the breast with
microwaves and detecting energy traveling through or reflected from the breast.
(b) With a tumor present, the reflected and transmitted waves change compared
To the tumor-free case [1] ..............................................................................................9
2-4 The exciting pulse used in our simulations ..................................................................11
2-5 Geometry o f the microstrip-fed resistively loaded “Dark Eyes” antenna [6] ......... 13
2-6 A three-dimensional standard Yee cell. Primary and secondary grids are shown
[3 1 ].....................................................................................................................................16
3-1 The hemispherical breast model (total diameter o f 14cm) .........................................19
4-1 (a) Central cross-sectional plane o f the breast model o f Fig. 3-1, parallel to the chest
wall. The antenna is located in this plane and is rotated to eight different locations
for 45° increments o f angle $ . (b) Tumor response levels calculated for each
location o f the antenna. For reference, tumor response levels are shown also for the
case o f no glandular tissue surrounding the main (tumor-containing) gland and for
tumor being embedded in the fat tissue only .............................................................. 25
4-2 Variation o f tumor response with (a) conductivity and (b) relative permittivity o f the
central gland (in which the tumor is embedded) .........................................................27
4-3 The breast model with off-center tumor .......................................................................29
4-4 (a) Central cross-sectional plane o f the breast model o f Fig. 4-3, parallel to the chest
wall. The antenna is located in this plane and is rotated to eight different locations
for 45° increments o f angle <F. (b) Tumor response levels calculated for each
location o f the antenna. Tumor response levels are shown for the cases o f complete
model, fat-only model, and one-gland only model .................................................... 30
4-5
On-center model with three antennas .........................................................................32
4-6
Off-center model with three antennas .........................................................................33
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List o f Tables
3-1 Gland properties ..............................................................................................................20
3-2 Breast tissue properties ...................................................................................................21
4-1 Tumor response levels for each location o f the antenna, and for the fat-only and one
gland only m o d e ls.............................................................................................................26
4-2 Tumor response levels for each location o f the antenna, for the cases o f complete
model, fat-only model and one-gland only model .......................................................31
4-3 Tumor Response o f the three antennas in Fig.4-5 ......................................................... 33
4-4 Tumor response o f the three antennasin Fig. 4-6 .........................................................34
vm
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 1
Introduction
Breast Cancer is the most frequently diagnosed cancer (other than skin cancer) in women
[12]. Canadian cancer statistics show that in 2005, an estimated 21,600 women were
diagnosed with breast cancer with 25% mortality rate [10].
An efficient strategy for reducing breast cancer mortality is the detection o f earlystage tumors [7]. When breast cancer is found early, the five-year survival rate is reported
to be 96% [38]. Over two million breast cancer survivors are alive in the United States of
America today [38]. Currently, the most commonly used method for detecting breast
tumors is X-ray mammography, known as the “Gold standard Method for breast
imaging” [1]. The technique has been in use for about thirty years [13]. However, this
method exhibits several limitations in detecting breast tumors at an early stage. These
difficulties arise from the small (few percent) intrinsic contrast between diseased and
normal tissue at X-ray frequencies. Further, the association o f X-ray mammography with
uncomfortable or painful breast compression and exposure to low levels o f ionizing
radiation may reduce patient compliance with screening recommendations. These
concerns motivate the search for techniques that image other physical tissue properties or
metabolic changes [41].
Microwave detection o f breast cancer is particularly motivated by the significant
contrast in the dielectric properties at microwave frequencies o f normal and malignant
breast tissue suggested by published measured data [17] -[19]. Furthermore, microwave
attenuation in normal breast tissue is low enough to make signal propagation through
even large breast volumes feasible [8]. In addition, microwave technology would be non­
ionizing and non-invasive. For these reasons, microwave breast imaging has the potential
to overcome some o f the limitations o f conventional breast cancer screening modalities
[41].
In this thesis, the performance o f the “Dark Eyes” antenna for microwave breast
cancer detection is investigated. The “Dark Eyes” antenna has recently been suggested
and has several advantages over other broadband antennas used for this purpose.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specifically, the antenna design demands constant surface resistive loading for costeffectiveness and practicality, and broadband behavior is achieved by favorable antenna
geometry [5]. The antenna is compact (overall dimensions o f 20mm x 21.75mm x
1.4mm), making it a favorable option for an antenna array. [6].
The Finite-Difference Time-Domain (FDTD) method is used as a numerical tool
for all reported simulations. The hemispherical breast model is developed using the
commercial software SEMCAD-X
[28].
Several glands with varying electrical
parameters are randomly embedded in the model, forming a non-homogeneous breast
model. The behavior o f the “Dark Eyes” antenna in the vicinity o f the breast model is
observed for several configurations, in order to assess tumor detectability and its level in
the presence o f gland tissue.
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
Background Review
Breast Cancer is the most frequently diagnosed cancer (other than skin cancer) in women
[12]. Canadian cancer statistics show that in 2005, an estimated 21,600 women were
diagnosed with breast cancer with about 25% mortality rate [10].
The general anatomy o f the breast is shown in Fig. 2-1. Breast cancer usually
begins in the “lobules”, the milk-producing glands o f the breast, or in the “ducts” that
bring the milk to the nipple. The lobules and ducts are surrounded by fatty and connective
tissue, nerves, blood vessels, and lymphatic vessels, which carry lymphatic fluid to and
from the lymph nodes (Fig. 2-1). About 80% o f breast cancers start in the ducts; the rest
start in the lobules.
1.obtik's.
Duett
■Areola
Figure 2-1: Anatomy o f the breast, showing lymph nodes and lymph vessels [37],
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
All forms o f cancer are characterized by uncontrolled cell growth. Cancer cells
generally mass together, forming a lump or “tumor”. When they migrate to other sites in
the body, creating new cancers, the process is known as “metastasis” [35]. When breast
cancer is diagnosed early, the cancer cells may be very small and found only in the ducts
or lobules (“in situ" term refers to the type o f breast cancer that has remained “in place”
and has not spread from its point o f origin to other tissues or organs). Ideally, the
cancerous tissue is detected while still confined to the mammary ducts, so that its
removal eliminates the risk o f its spreading to the surrounding tissue. However, even if
the cancerous cells invade the healthy tissue (invasive carcinoma), the tumor can be
removed if detected at its early stage [10], [35].
2.1 Types of breast cancer
The likelihood o f breast cancer spreading to other tissues or parts o f the body depends on
the type o f cancer involved [35].
1. Ductal carcinoma in situ (DCIS): The most common type o f in situ cancer,
DCIS is a breast cancer at its earliest and most curable stage, still confined to the
ducts. In screening centers, 20-30% o f newly diagnosed breast cancers are DCIS.
DCIS often occurs at several points along a duct, and appears as a cluster of
calcifications, or white flecks, on a mammogram. Because o f its potential to recur
or become invasive, DCIS is treated with excision (if the area o f DCIS is small) or
mastectomy (when the disease is more extensive), possibly followed by radiation.
2.
Invasive (Infiltrating) ductal carcinoma (IDC): IDC is the most common type
o f invasive breast cancer. It begins in a duct, breaks through the duct wall, and
invades fatty tissue in the breast. From there, it can metastasize to other parts of
the body. IDC is usually detected as a mass on a mammogram or as a palpable
lump during a breast exam.
3. Invasive (Infiltrating) lobular carcinoma: About 10-15% o f invasive cancers
are o f this type. The malignant cells have grown through the wall o f the lobule
and can spread to other parts o f the body by way o f the lymphatic channels or
bloodstream.
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4. Medullary, tubular, and mucinous carcinoma: These are less common types o f
ductal carcinoma, together accounting for less than 10% o f breast cancers.
Medullary and tubular carcinomas are both invasive but may have better
prognoses than invasive ductal or invasive lobular carcinomas.
5. Inflammatory breast cancer: This unusual and aggressive form o f breast cancer
starts with breast swelling and red skin that feels warm, somewhat like an
infection. The skin is red because cancer cells are blocking the lymphatic
channels, and fluid cannot drain from the lymph vessels o f the skin.
2.2 Overview of main breast cancer detection methods
An efficient strategy for reducing breast cancer mortality is the detection o f early-stage
tumors [7]. When breast cancer is found early, the five-year survival rate is about 96%.
Over two million breast cancer survivors are alive in America today [38]. The
following information describes the current imaging techniques that are in use or are
being studied as potential breast tumor imaging alternatives.
4.1.1
Mammography
In this method, the breast is exposed to a strictly controlled dose o f x-ray radiation to
produce an image o f internal breast tissue. The image o f the breast is produced as a result
o f a portion o f the x-rays being absorbed, while others pass through the breast to expose
either a film
(conventional mammography) or digital image receptor (digital
mammography). The exposed film is placed in a developing machine, producing images
much like the negatives from a camera; digital images are stored on a computer [40].
A mammography unit is a rectangular box that houses the tube in which x-rays
are produced. The unit is used exclusively for x-ray exams o f the breast, with special
accessories that allow only the breast to be exposed to the x-rays. Attached to the unit is a
device that holds and compresses the breast and positions it so that images can be
obtained at different angles (Fig. 2-2).
X-ray mammography is currently the most commonly used method for detecting
breast tumors, known as the “Gold standard Method for breast imaging” [1]. The
technique has been in use for about thirty years [13]. However, this method exhibits
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
several limitations in 1) assessing dense glandular tissues and regions located close to the
chest wall or underarm and 2) imaging very early-stage tumors that do not yet exhibit
micro-calcifications. This causes a high rate o f false negatives (when the tumor does not
show up on a mammogram) and false positives (when the mammogram identifies an
abnormality that looks like a tumor, but turns out to be normal) [8]. About 15% o f breast
tumors are missed in mammography. Also nearly 75% o f the tumors detected as
malignancies are found to be benign [21]. Diagnosis often involves waiting for further
imaging or biopsies.
The ultimate diagnosis o f all types o f breast disease depends on a biopsy. A
biopsy is an invasive procedure to remove and examine tissue or cells for the presence of
cancer. In most cases the decision for a biopsy is based on mammography findings.
During the mammography procedure, the breast is compressed in order to even
out the tissue, increase image quality, and to hold the breast still (preventing motion
blur). Consequently, many patients find mammography uncomfortable or painful.
Further, the tissue is exposed to ionizing radiation which has a potentially hazardous
effect.
(a)
(b)
Figure 2-2: (a) Mammography equipment [39] (b) Mammography procedure [40].
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2.2. Ultrasound
Ultrasound sends high-frequency sound waves through the breast and converts them into
images on a viewing screen. It is used as an adjunct technique to mammography and can
increase the overall sensitivity o f conventional breast imaging. If an abnormality is seen
on mammogram or felt by physical exam, ultrasound is a good way to find out if the
lump is a (solid) malignant mass or a (liquid) benign cyst.
Mammograms can be difficult to interpret in young women because their breasts
tend to be dense. Older women's breasts tend to be more fatty and are easier to evaluate.
In mammograms, glandular tissues look dense and white— much like a cancerous tumor.
Also, the risks o f radiation from mammography are most significant in young women.
Therefore, for women under 30 years o f age, ultrasound may be recommended before
mammography [14].
However, ultrasound is not a substitute for a screening mammogram because o f a
variable false-negative rate o f up to 47% and the operator-dependent nature o f efficacy.
The image quality is poor and the penetration depth limited [14].
2.2.3. Magnetic Resonance Imaging (MRI)
MRI is a powerful diagnostic tool that uses strong DC magnetic fields in combination
with radio frequency radiation to create images o f the body. During an M RI procedure,
the patient lies still and is moved in and out o f a narrow tube as the machine creates
images o f the body. The value o f MRI for breast cancer detection remains uncertain.
Some doctors believe MRI can distinguish a breast cancer from normal breast gland
tissue better than other techniques. However, MRI is expensive, and requires highly
specialized equipment and highly trained experts. Relatively few MRI centers exist,
especially outside o f major cities. Even at its best, MRI produces many uncertain
findings. Some radiologists call these "unidentified bright objects," or UBOs. MRI cannot
detect calcifications. Finally, M RI’s strong fields can interfere with proper functioning of
some implanted devices, such as pacemakers, or dislodge metal implants. It is therefore
unlikely that MRI will be used as a general screening tool for breast cancer [13].
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2.4.
Biological
Microwave Breast Cancer Imaging
tissue
interactions
with
fields
are
defined
by
their
complex
permittivity £ r (co ) , which consists o f the dielectric constant (the real part), e r and the
loss factor (the imaginary part)£-". The dielectric constant determines the ability o f the
material to store the electric field energy, while the loss factor indicates how much
energy is converted into heat and dissipated [1]. The complex permittivity £ r (co ) can
be written as:
£*r (o)) = £r +
a
j(0£ o
(2-1)
Microwave breast cancer detection exploits the contrasts in electromagnetic
properties at microwave frequencies between healthy and malignant tissues located in the
breast. Low-water-content tissues, such as bone, fat, lung and the outer layer o f skin,
have lower permittivity £rand conductivity a values than high water content tissues
such as muscle, blood, brain, and internal organs [16]-[18]. According to the literature,
there is an order-of-magnitude dielectric-property contrast between normal and malignant
breast tissues at microwave frequencies. Normal breast tissues display properties similar
to fatty tissues, while tumor properties are similar to muscle [8], [19], [20]. The dielectric
properties o f malignant tumors show no significant variation with tumor age, suggesting
that the large contrast exists at the earliest stages o f tumor development [21], This makes
this method suitable to identify early-stage tumors, while X-ray mammography cannot
detect them since early-stage tumors do not yet exhibit micro-calcifications.
Microwave imaging is defined as “seeing” the internal structure o f an object by
means o f electromagnetic fields at microwave frequencies (300MHz-30GHz) [1], A
transmitter is used to illuminate the breast with microwaves, which travel through the
breast and may be detected at receivers located at the opposite side o f the breast.
Reflections may also be recorded at the transmitting antenna (which is the case in our
simulations). With a tumor present, waves traveling through the breast encounter a
change in electrical properties, causing the incident wave to scatter. The scattering
changes the energy detected at the receivers and the transmitter. Images are formed using
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
information contained in the detected energies. The physics o f the problem is illustrated
in Fig. 2-3.
©
Transmit
©' Receivers
O
Breast
Tumor
©
O
Transmitter
b)
®
Receivers
Figure 2-3: (a) The basic microwave imaging problem involves illuminating the breast with microwaves
and detecting energy traveling through or reflected from the breast, (b) With a tumor present, the reflected
and transmitted waves change compared to the tumor-free case [1].
Microwave frequencies are a favorable frequency option for breast cancer detection
for two major reasons [8]:
6. The high water content o f malignant tumors causes them to have significantly
larger microwave scattering cross section than normal fatty breast tissues that
have low-water content. The vascularization (naturally occurring vascular
network around tumors) o f malignant tumors further increases the scattering
cross-section.
7. Microwave attenuation in normal breast tissue is low enough (less than 4dB/cm
up to 10 GHz)
To obtain backscatter signals from the breast, we illuminate the breast with an ultrawideband pulse. For breast imaging, we are interested in the l-10GHz regime because it
appears to balance the conflicting demands o f better spatial resolution (higher
frequencies) and better penetration depth (lower frequencies) [36]. The ultra-wideband
nature o f this approach enhances spatial resolution [9].
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Illuminating the breast with this signal demands an antenna with reasonable ultrawideband performance. At the same time, the antennas must be small enough to be
placed on or near the breast. In our studies, a Gaussian pulse with a central frequency o f
6GHz and a bandwidth o f 11 GHz is used in conjunction with the miniaturized “Dark
Eyes” antenna. The details are explained later in this chapter.
Microwave detection is more attractive for breast cancer than for other cancer types.
One reason is that the contrast between healthy and diseased breast tissues is likely much
greater than the contrast in these permittivity values for other tissues. The small size and
physical accessibility o f the breast compared to internal organs is also an advantage. For
example, accessing the breast with the microwaves does not involve penetrating layers o f
muscle, which results in scattering and attenuation o f the incident field before it reaches
the object o f interest (e.g. an internal organ) [1].
From the patient’s viewpoint, microwave breast imaging is attractive because both
ionizing radiation and breast compression are avoided. This results in safer and more
comfortable exams. From a technical point o f view, microwave breast tumor detection
has the potential to detect small tumors. It is also expected to be less expensive than
methods such as MRI and nuclear medicine. This is because microwave equipment costs
a fraction o f the equipment needed for MRI and nuclear medicine installations (hundreds
o f thousands vs. millions). The imaging process is also anticipated to be rapid, sensitive
(detect most tumors in the breast) and specific (detect only cancerous tumors). The key to
sensitivity, specificity and the ability to detect small tumors is the reported electrical
property contrast between malignancies and normal tissues [1].
2.3 Ultra Wideband (UWB) Antennas
Recently, a good deal o f research has been made on ultra wideband (UWB) antennas,
motivated in part by a growing interest in the use o f short-pulse radar techniques in
remote and subsurface sensing applications at radio, microwave and terahertz frequencies
[22]. Compared to narrowband systems, UWB radars offer potentially higher imaging
resolution and better target characterization.
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Typical UWB antenna design requirements include a broad operating bandwidth
for short pulse radiation, a compact radiation pattern to avoid interference caused by
backscatter from undesired directions and a small size to make the system portable [22],
Pulsed microwave biological imaging can benefit from improved compact broadband
antenna designs [26], as will be discussed further.
2.3.1
The Exciting Pulse
The excitation waveform used in the simulations is a Gaussian modulated sinusoidal
pulse described by [5]:
v(0 = sin[2^f0(t - t 0)]e x p [-(r- t 0) 2 / 2 r 2]
(2-2)
with / 0 = 6 GHz, r = 80 ps, and t0 = 5 r . In the frequency domain, this pulse has a center
frequency o f 6 GHz and a bandwidth o f 11 GHz.
This pulse shape and its parameters follow from the intended application o f the antenna
for breast cancer detection, as described in the previous section [5].
0.5
(/>
o
>
>
-0.5
0.5
2.5
t(nsec)
Figure 2-4: The exciting pulse used in our simulations.
2.3.2
Resistively Loaded Antennas
Resistive loading has been proposed as a method for improving the bandwidth o f planar
antennas. Resistively loaded conical and bow-tie antennas have been suggested for breast
tumor detection [23], [25]. The basic idea behind the resistive loading is to reduce the
unwanted reflections that occur along the antenna and the associated distortion o f the
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
radiated signal. The radiated pulse from the resistive antenna more closely resembles the
excitation at the feed than for a comparable metallic antenna [23].
Another problem to be addressed is that small or weakly scattering tissue
structures adjacent to an impulsively excited antenna can be obscured by the reflections
from the ends o f the antenna. The resistively loaded antennas mentioned above have end
reflections 40-50 dB below the exciting pulse. This reflection is too high for detecting
tumors in the breast [8].
The bowtie antenna suggested by Hagness et al. [8], [24] demonstrates end
reflections o f 106 dB below the exciting pulse. Although this may be sufficient for the
breast tumor detection, the antenna uses variable resistive loading as a part o f the
broadband design [8]. This technique implies challenges and potential cost increase in the
antenna fabrication [5].
2.3.3
The “Dark Eyes” Antenna
We can choose to limit our design technique to constant surface resistive loading for costeffectiveness and practicality, and aim to achieve broadband behavior by favorable
antenna geometry [5].
Fig. 2-3 shows the antenna geometry. The proposed design is symmetrical with
respect to the feeding point. Each half o f the antenna consists o f two sections. The first is
a metallic bowtie like, located at the apex o f the antenna. The second section is tapered
and o f constant surface resistive loading. It is this section that is used to minimize the
signal reflections off the antenna ends and hence improve its broadband behavior [5]. The
antenna feed is a modification o f the broadband uniplanar microstip-to CPS transition [6].
In this design, the balun is scaled down to operate at 6GHz and is buried under a resistive
layer, which has shown not only to reduce the radiation from the microstrip feed itself,
but also to help suppress the resonance [6], [26]. The whole antenna is immersed in a
lossless dielectric medium o f a relative permittivity e r = 10.2 for impedance-matching, as
this value is close to that o f relative permittivity o f fatty tissue. The electric conductivity
o f the constant resistive loading and the resistive layer is chosen to be 277 S/m.
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The performance o f the Dark Eyes antenna is shown to be as good as or better
than the bowtie antenna in several prospects [4]. The fabrication process is much simpler
and cheaper.
The antenna was given its name as the described geometry was reminiscent o f the
eye shape [27].
Ground Plane
Resistive Layer
Dielectric Substrate
o f thickness
Resistivi
Loading
Antenna
Apex
|
i
2Ws+Sj
Figure 2-5: Geometry o f the microstrip-fed resistively loaded “Dark Eyes” antenna. The Dimensions o f the
structure are (in mm): Wx - W 3 = W 4 = 0.75 ,W 2 = 1 .5 ,W S = S , = S 2 = 0.3 7 5 ,
L x = L s = 1 .5 ,L 2 = L 3 = 3 .7 5 , L 4 - 5 .6 2 5 ,L 6 = L i = 2 .2 5 ,L 7 = 4 .5 , L g = 8 , L l0 = 6
£=20, W= 12.75, and f=0.64 [6],
2.4 Finite-Difference Time-Domain (FDTD) Method
The Finite-Difference Time-Domain (FDTD) method is a popular electromagnetic
modeling technique. It is intuitive and straight-forward to implement. Since it is a timedomain technique, it can cover a wide frequency range with a single simulation run.
FDTD is particularly well-suited to computing transient responses [34]. FDTD computes
the electric field (E-field) and magnetic field (H-field) explicitly.
Yee [32] first introduced the novel approach o f replacing M axwell’s equations
with finite difference equations. Taflove [31] further developed the method and since its
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conception, the FDTD method has dramatically progressed [34]. FDTD Method has
successfully been applied to many different problems in electromagnetics such as
scattering, radiation o f antennas, optical applications and guided wave propagation [15].
The main drawback o f the FDTD method is its potential need for large computational
sources [15]. The spatial grid refinement, resulting from constraints imposed by the
wavelength and the problem geometry, can yield a large number o f unknowns. This, in
turn, leads to long computation time. For example, models with long, thin features (like
wires) are difficult to model in FDTD because o f the excessively large computational
resources required. Similar problems can be encountered for calculation o f fields at a
location physically far from radiation source, as this would create a large computational
domain [15].
2.4.1 Maxwell’s Equations
FDTD solves Maxwell’s equations by discretizing them in both time and space.
Faraday’s and Ampere’s law are given as:
-
dt
= - — 'V x E - £ - H
p
p
dE
1
<t
dt
s
s
(2' 3)
(2‘4)
Where E is electric field intensity in V/m, H is magnetic field in A/m , e is electrical
permittivity in F/m, p is magnetic permeability in H/m, a is electrical conductivity in
S/m, p ' is equivalent magnetic resistivity in Q/m.
In an isotropic medium we can express Maxwell’s equations in Cartesian coordinates as:
dt
p
dH
dt
dz
dy
l dE2
•=
- ( -
p
dx
dz
dE,
dH.
dt
dE,
p
dy
dx
~ P ' H X)
■ p'Hy )
- p 'H2)
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2-5)
(2 -6)
(2-7)
The
subscripts
in
the
above
equations
imply
a
component
of
the
field.
These six equations form the basis for the FDTD algorithm for modeling propagation
through general media.
2.4.2 Discretization: The Yee cell
In Yee’s algorithm, Maxwell’s equations in differential form are approximated by
second-order central-differences. The relevant central-difference equations are as
follows:
dF(i,j,k,n)
F i+\/2,j,k - F
dx
dF ( i, j, k ,n )
dt
d 2F(i, j , k,ri)
Sx
+ 0[Sx ]
n - 1/2
n+l/2
F i,j,k - F
— + 0[S'2]
Sx
”-ij,k ~ ^ F \ "yj,k +
dx2
Sx2
m j„
i,k
+ 0[Sx4]
(2- 11)
(2- 12)
(2-13)
With F " a s the Electric ( E ) or magnetic ( H ) field at timen.St , where St is the time step.
i,j,kaxe. the indices o f the spatial lattice, and 0[(Sx)2] and 0[(St)2] are error terms.
Using equations (2-12)-(2-14), equations (2-4) and (2-5) are discretized. The resulting
equations are solved in a “leap-frog” manner [32], [33]. E and H components are
positioned along the side o f the unit lattice cell as shown in Fig. 2-4. In the primary grid,
the H-field is computed by using the circulating E-fields. In the secondary grid, E-field is
computed using the circulating H-fields.
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Primary Grid
i>E.
Node
S eco n d a ry
Figure 2-6: A three-dimensional standard Yee cell. Primary and secondary grids are shown [31]
The resulting expressions for the fields are:
p'{i,j +^,k +)^Ai
1n+ -
1
„
. l , i7
2M ( t , j + - , k + ~)
1
j ,
n_i
jf-w , H.'.y+- ,r +- ) +
H xx 2(i,j
K ’ J + -2 , >k + -2 )> =
p V J + - , k +-)At
1+
-
„
At
. i , rr
1+
E;(ij+±,k+i)-E;(ij+±,k)
+
Az
E:(ij,k+~)-E?(i,j+i,k+^)
Ay
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2-
2.4.3 Boundary Conditions
Absorbing boundary conditions (ABC) are necessary to truncate the problem to a finitesized computational domain which is similar to infinite space in a numerical manner. In
all our simulations, the ABCs are a perfectly matched layer (PML) [31].
2.4.4 Numerical Stability
For the explicit finite-difference time-domain scheme to yield a stable solution, the time
step, St used for the updating must be limited according to the Courant-Friedrich-Levy
(CFL) criterion. For the FDTD formulation o f Maxwell’s equations, this requirement is:
8<
(2-16)
Where 8x, Sy and dz are the mesh steps o f a Cartesian coordinate system and c the speed
o f light within the material o f a cell.
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3
The Breast Model
3.1
Geometry and Parameters
For our simulations, we develop a realistic breast model using the SEMCAD-X software.
The model is considered to be a hemisphere with a diameter o f 14 cm. Several tissue
types are included, such as 5 different gland types, a chest wall and a 2-mm thick skin.
The breast model is adapted from [l]-[3]. The electrical parameters o f the glands are
considered at 6GHz [l]-[4], [15].
The details o f the glands are explained in Table 3-1. Five different gland types
with different electrical parameters (relative permittivity s r and electrical conductivity a )
are considered. Each type consists o f a number o f glands with different shapes (spherical
or cylindrical). Other tissue details (shapes and electrical parameters) are explained in
Table 2-2. The whole model is immersed in a non-conductive immersion liquid matched
to the antenna dielectric substrate. This helps reduce unwanted reflections.
Since the Gaussian pulse used for exciting the antenna is composed o f a wide
range o f frequencies, electrical parameters o f the breast will change in this frequency
range. Several methods can be used to handle dispersive materials in a broad range o f
frequency range, such as the “four-term (lOHz-lOOGHz) Cole-Cole parametric dispersion
model” [18], [36] for the complex permittivity o f infiltrated fat and muscle-two
biological tissues that mimic the lower and upper bounds on the relative permittivity ( s r)
and conductivity [ a (S/m)] o f tissue in the breast:
. a
£r ~ J
(3-1)
msQ
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In-gland Tumor
The antenna plane
Chest Wal' *
Figure 3-1: The hemispherical breast model (total diameter o f 14cm). (a) General view, showing the “Dark
Eyes” antenna located near the model, in the x-y plane located centrally between the chest wall and the
nipple, (b) Cross-sectional side view, showing the tissues considered in the model.
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3-1: Gland properties
Electrical
Gland Type
Parameters
e r = 15
#1
a =0.4 S/m
Dimensions
1-1) Cylindrical, r=2mm, h - 27mm*
1-2) Cylindrical r=4mm, h=24mm
1-3) Spherical r=8.5mm**
2-1) Cylindrical
#2
s r = 14
r = 18 mm, A=16mm
2-2) Cylindrical r=9mm, A=27mm
2-3) Cylindrical r=9mm, /i=12mm
cr =0.4
S/m
2-4) Cylindrical r=5mm,
h=12mm
2-5) Cylindrical r=5mm, A=12mm
s r =13
#3
cr =0.4 S/m
Cylindrical r=5m m , h=2Am m
er = U
#4
a =0.5 S/m
=11
#5
<7=0.5 S/m
Spherical r = 12.5 mm
5-1) Cylindrical r=5mm, h=12mm
5-2) Cylindrical r=10mm, h=12mm
5-3) Cylindrical r=15mm, hr=l4mm
5-4) Cylindrical r=5mm, h=12mm
*r: radius, and h: height
**r: radius
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3-2: Breast tissue properties
Tissue
Electrical Parameters
Geometrical Shape
s r =9
Fat
=0.4 S/m
<7
Hemispherical: 68-mm diameter
er =36
Skin
cr - 4 S/m
2-mm thick
er =50
Tumor
ll
cr =7 S/m
Nipple
cr =5 S/m
Spherical: 6-mm diameter
Cylindrical:
r=10mm, A=5mm*
s r =50
Chest Wall
cr - 7 S/m
Cubic: 90mmx90mmx20mm
e r =10.2
Immersion Liquid
cr =0 S/m
*r: radius, and h: height
The parameters for infiltrated fat are
c
s x = 2 .5 , a s = 0 .0 3 5 — ,A e x = 9.0,A£-2 = 35,A £3 = 3 .3 x l0 4,A£-4 = l x l 0 7
m
Tl =7.96ps , r 2 = 15.92ns, r 3 = 159.15ps,r4 = 15.915ms
a , = 0 .2 0 ,a 2 = 0 .1 0 ,a 3 = 0 .05 ,a4 =0.01
While the parameters for muscle are:
= 4 , a s = 0 . 2 — , A s 1 = 5 0 , A s 2 = 7 . 0 x l 0 3, A f 3 = 1 . 2 x l 0 6, A s 4 = 2 . 5 x l 0 7
m
r, = 7 .2 3 p s,r2 = 353.68ns, r 3 = 318.31ps,r4 = 2.274ms
or, = 0.10 , a 2 = 0.10 , a 3 = 0.05, a 4 = 0.0
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, the use of these models increases the simulation time. We choose to use fixed
electrical parameters at 6 GHz, the frequency at which the pulse is centered. This gives us
acceptably accurate results without overburdening computational resources.
3.2
Simulation Outline
The tumor is considered to be a sphere, embedded inside a gland, as is often encountered
in medical diagnosis (in-ductal carcinoma), [35].
In the first part o f our simulations (section 4-1), we consider the case o f an “oncenter” tumor. The tumor is a 6mm diameter sphere, embedded at the center o f the breast,
inside gland 1-2 ( e r = 15, a = 0 - 4 ^ ^ ) (Fig. 3-1). Several simulations are done for this
case. First, the electrical parameters o f the glands are kept constant, and the antenna is
rotated around the breast model. Tumor responses are assessed and analyzed. Next, the
antenna is kept fixed and the electrical parameters ( s r and cr) o f the gland within which
the tumor is embedded (glands #1, #2) are changed.
As a subsequent study, we consider the “off-center” case. The tumor is a 4-mm
diameter sphere, embedded inside gland #3 ( s r = 13, a = 0 .4 ^ ^ ) . Again, the gland
parameters are kept constant, as the antenna is rotated around the breast.
In the last part o f our investigation, the tumor is an “on-center” 6-mm diameter
sphere. Three coplanar antennas are considered, with an angle o f 90° between adjacent
antennas. Mutual effects between antennas are analyzed and discussed.
3.3 SEMCAD-X
SEMCAD-X is a 3-D full-wave simulation environment based on the FDTD method,
developed and provided by Schmid and Partner Engineering Company [28]. SEMCAD-X
is an all-purpose electromagnetic solver, which offers extended capabilities for numerical
dosimetry and the simulation o f antennas. Its ACIS-based three-dimensional solid
modeling interface combined with a flexible automatic mesh generator enables fast and
easy design o f complex, arbitrary shaped geometrical structures.
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SEMCAD-X provides standard solvers as well as the first commercial ADIFDTD solver and solutions for coupled EM-thermal simulations.
Properties such as
absorbing, conducting and periodic boundary conditions (Perfectly Matched Layer
(PML), Mur, etc.), or various source models (lumped edge, plane wave, waveguide)
enable the optimization o f antennas, design o f microwave circuits or the treatment o f a
range o f scattering problems [15]. Materials can be specified as dielectric, PEC (perfect
electric conductor) and PMC (perfect magnetic conductor). Solids can be set as Solid,
Thin Sheet, or Thin Wire, defining appropriate numerical modeling treatment o f the
geometry in question.
In the here reported simulations, a non-uniform mesh was used with a minimum
and maximum grid resolution o f 0.1 and 1.2 mm in x and y directions, respectively, and
35 pm and 1mm in the z direction (the axes are shown in Fig. 3-1), respectively. The
antenna was fed at its center with a 1-V, 50-fl resistive gap source and the grid domain
was terminated with an 8-cell PML absorbing boundary.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 4
Parametric Studies — Results and Discussion
4.1 “On-Center” Tumor
For the first part of our simulations, we consider the tumor to be a 6-mm diameter sphere,
embedded at mid-breast cross-sectional plane, inside a gland (Fig. 3-1). The antenna is
also located in the same cross-sectional plane parallel to the chest wall, as depicted in
Fig. 3-1 (a).
4.1.1
The effect of glandular tissues on detection levels
For this part, the antenna was rotated to eight locations defined by uniform increments o f
45° o f angle <P, as shown in Fig. 4-1 (a). Tumor response o f the antenna was recorded at
each location. The tumor response was calculated in the following manner: First the
complete model with the tumor present was radiated, and the back-scattered signal was
collected at the antenna location. Next, the breast model without the tumor present was
excited and the back-scattered signal again collected. The second signal was subtracted
from the first, and the maximum absolute value o f the resulting signal (in dB) was
defined as the tumor response. Since the main radiation lobe o f the antenna will “see”
different glandular arrangement from each location, it is expected that the tumor response
will vary.
For comparison, we also record the tumor response for the case where no glands
are present and for the case where only the central gland (gland 1-2, in which the tumor is
embedded) is present. Since the tumor is located at the center o f the breast, the antenna
“sees” the same structure as it rotates around the “fat-only” and “one-gland only” model.
Only one simulation is thus needed for each o f these cases.
Results are shown in Fig. 4 -1(b) and explained in Table 4-1. These results
demonstrate that the changing gland arrangement relative to the antenna location results
in tumor response that varies around ±5dB with respect to the case o f no glands
surrounding the tumor in the central gland. The level o f tumor detection response when
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the tumor is located in the fat tissue only is approximately 4dB higher than the response
o f a tumor hidden in the main gland.
(a)
-116 r
••
C o m p le te M o d el
Fat-O nly (N o g la n d s)
-118 ■
Only o n e gland
C
O -120
*o
<D
£ -122 h
O
CL
<0
<D
™
a : -124
i—
o
E
| 2 -126
-128
-130_l
45
90
135
180
225
270
315
(b)
Figure 4-1: (a) Central cross-sectional plane o f the breast model o f Fig. 3-1, parallel to the chest wall. The
antenna is located in this plane and is rotated to eight different locations for 45° increments o f angle $ .
(b) Tumor response levels calculated for each location o f the antenna. For reference, tumor response levels
are shown also for the case o f no glandular tissue surrounding the main (tumor-containing) gland and for
tumor being embedded in the fat tissue only.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4-1: Tumor response levels for each location o f the antenna, and for the fat-only and one gland only
models.
Rotation Angle (°)
Tumor Response (dB)
0
-118.5
45
-125.63
90
-117.2
135
-128.4
180
-120
225
-125.5
270
-122.6
315
-125.5
Fat-only
-119.5
One Gland Only
-123.4
These results also suggest that the presence o f multiple glands o f different shapes,
sizes and electrical parameters, with varying arrangement with respect to the sensing
antenna, cause not only attenuation o f the signal and then, in turn, degradation o f the
tumor response, but also multiple reflections which can yield a response that could be
falsely interpreted as a response from a larger tumor. It is key for this fact to be taken into
account when image-construction algorithms are designed to process signals from either
array-elements or multiple locations o f mechanically-scanned antennas.
4.1.2 Sensitivity o f tumor response levels to electrical parameters of the
central (tumor-containing) gland
Next, we keep the antenna at a fixed location (0° rotation) and vary the electrical
parameters o f gland 1-2, in which the tumor is located. The goal is to determine tumor
response sensitivity to the parameters o f the tumor-containing gland.
In the first part, we keep the relative permittivity ( s r ) constant at its assumed
value ( e r = 1 5 ) and vary the conductivity (cr) in the range 0 .1 ^ / < a < 0.7*5/
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(±75% from its assumed value o f 0.4 S/m). Fig. 4-2 (a) shows the degradation in the
tumor response level from -117.9dB to -1 18.8dB ascr changes from 0 . 1 ^ / to 0.7$ / .
Next, the conductivity is kept at its assumed value ( a - 0 . 4 ^ / ) and we change relative
permittivity about ±25% around the previously assumed value o f 15, i . e . l l < £ r <19.
Fig. 4-2(b) graphs the deterioration o f the tumor response from -117.2dB to -119dB as£r
changes from 11 to 19.
-117.5
m
T>
a>
v>
c
o
a.
CO
a>
i—
o
E
-118.5
3
I-
0.3
0.4
0.5
0.6
0.7
a (SI m)
(a)
-117
a=0.4 S/m
TJ
T -117.5
S ’ -118
-118.5
-119
(b)
Figure 4-2: Variation o f tumor response with (a) conductivity and (b) relative permittivity o f the central
gland (in which the tumor is embedded).
We obtain two conclusions from these results. First, the tumor response is less
pronounced as the electrical parameters o f the gland containing it approach those o f the
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tumor itself. Second, even though this deterioration is noticed, it does not exceed ~2dB,
implying that the variation o f gland electrical parameters may not have a significant
effect on the sensitivity o f the tumor detection process with microwaves.
4.2 Presence of an “Off-Center” Tumor: Effect of Glandular
Tissue on Detection Levels
In this section, we consider the tumor to be a smaller sphere (4-mm diameter), embedded
inside gland 3 (2cm away from the center) at mid-breast level. The antenna is again
located at the mid-breast cross-sectional plane parallel to the chest wall. Fig. 4-3 shows
the geometry under investigation.
The antenna is then rotated to eight locations defined by uniform increments o f 45° of
angle 0 , as shown in Fig. 4-4 (a). Tumor response levels are shown in Fig. 4-4 (b). For
comparison, the tumor responses for the case where no glands are present and for the case
where only gland #3 (in which the tumor is embedded) is present are also shown. Results
are offered in Table 4-2. As we can see in Fig. 4-4, the tumor response o f the complete
model is generally lower than that o f the fat-only and one gland only models. The
response for the one gland only model is almost always very close to that o f the complete
model. This shows that the gland in which the tumor is embedded has a significant effect
on the tumor response.
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
Nipple
▼
In-gland Tun mi
Glands
The antenna plane
Chest Wall
(b)
Figure 4-3: The Breast model with off-center tumor.
(a)
General view o f the model near the antenna, (b) Cross-Sectional side view (x-z plane).
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-105
- Only O n e Gland
- Fat-Only
C o m p le te M odel
-11
-115
w -120
-125
-130
\ \ \
-135
-140,
100
150
n
200
250
300
(b)
Figure 4-4: (a) Central cross-sectional plane o f the breast model o f Fig. 4-3, parallel to the chest wall. The
antenna is located in this plane and is rotated to eight different locations for 45° increments o f angle 4>. (b)
Tumor response levels calculated for each location o f the antenna. Tumor response levels are shown for the
cases o f complete model, fat-only model, and one-gland only model.
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4-2: Tumor response levels for each location o f the antenna, for the cases o f complete model, fatonly model and one-gland only model
Tumor Response (dB)
Rotation Angle (°)
Complete Model
“Fat-only”
“One Gland only”
Model
Model
0
-110.4
-108.3
-110.3
45
-117.1
-114.8
-116.5
90
-131.8
-126.9
-129.4
135
-136.5
-135.3
-136.3
180
-131.2
-138.1
-139
225
-130.0
-131.7
-132.8
270
-125.5
-121.9
-123.2
315
-113.1
-110.7
-112.6
4.3 Mutual Coupling
The final goal in designing the “Dark Eyes” antenna is to design a suitable element for an
antenna array for breast tumor detection. Mutual coupling between elements is an
important factor to be considered when designing an array.
In order to study the impact o f one antenna element on the other elements in an
array, we investigate the following. Three antennas are located around the breast, again at
mid-breast level. Adjacent antennas are separated by an angle o f 90°. We excite the first
antenna, but keep the other two antennas passive, and we record the tumor response at all
three antenna locations. In this way, we could observe the impact o f an excited antenna
on the passive elements, and the impact o f passive elements on the tumor response
detected by the active element.
Two cases are studied: in the first case, the tumor is a 6-mm diameter sphere
located at the center o f the breast (as described in section 4-1). In the second case, the
tumor is a 4-mm diameter sphere, located off-center (section 4-2).
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.1
On-center tu m o r
The problem geometry is shown below. Antenna (1) is excited by a 1-V, 50-12 voltage
source. Antennas (2) and (3) are passive.
Figure 4-5: On-center model with three antennas.
The tumor response at antenna (1) is now -118.673dB. When compared to the
case where no passive elements are present (tumor response o f -1 18.526dB), we observe
that tumor response has changed less than 0.2 dB. This shows that the presence o f
antennas (2) and (3) does not have a significant effect on the tumor response o f antenna
(1).
Antennas (2) and (3) show tumor responses o f -121.460dB and -117.608dB
respectively. This is comparable with the tumor response o f antenna (1), and is almost the
same as the responses o f these antennas when they act as active elements with no passive
element present (Section 4-1). These results are summarized in Table 4-3.
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4-3: Tumor Response o f the three antennas in Fig. 4-5
Antenna #
Tumor Response of the
Tumor Response (dB)
antenna alone (active)
Antenna 1: Active
Antennas 2,3: Passive
(dB)
4.3.2
1
-118.5
-118.7
2
-122.6
-121.5
3
-117.2
-117.6
Off-center tumor
Now the tumor is located off-center, same place as in Section 4-2. Again antenna (1) is
active and antennas (2) and (3) are passive (Fig. 4-6).
Figure 4-6: Off-center model with three antennas.
The tumor response o f antenna (1) is -110.385dB. This is exactly the same as the tumor
response o f antenna (1) with no other elements present. This shows that antennas (2) and
(3) have practically no effect on the tumor response o f antenna (1). The tumor response
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of antennas (2) and (3) are -126.549dB and -133.132 dB respectively. When these
antennas are excited with no other antennas present, their tumor responses are found to be
-125.474dB and -131.784dB, respectively. This demonstrates weak impact o f antenna (1)
on antennas (2) and (3). The results are summarized in Table 4-4. They suggest that, with
the given spacing o f the antenna-elements, mutual coupling between them would not
present an obstacle in the tumor-detection process.
Table 4-4: Tumor response o f the three antennas in Fig. 4-6
A ntenna #
T um or Response of the
T um or Response (dB)
antenna alone (active)
Antenna 1: Active
Antennas 2,3: Passive
(dB)
1
-110.4
-110.4
2
-125.5
-126.5
3
-131.8
-133.1
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 5
Conclusion and Directions for Future Work
This work focuses on issues in the development o f a novel breast cancer detection
methodology using microwave radiation. In contrast to the reviewed current detection
modalities, this technique promises an imaging procedure that would be comfortable to
the patient, non-ionizing in nature and cost-effective.
Several antennas have been suggested for microwave breast cancer detection.
Our investigations focused on the performance o f the “Dark Eyes Antenna”. This antenna
has been recently reported in the literature as a potentially winning design for the cancer
detection application, due to its compact size and cost-effectiveness, achieved through
constant resistive loading and a favorable geometry.
A realistic breast model was developed in order to examine the “Dark Eyes”
antenna behavior in the vicinity o f a complex, inhomogeneous tissue structure. Antenna
behavior was simulated with SEMCAD-X, commercial software based on the finitedifference time-domain (FDTD) method. Several geometries o f interest were simulated.
In all our simulations, the tumor was embedded inside a glandular tissue as in real cases
o f breast cancer. We first considered an “on-center” tumor (tumor located in the central
gland, below the nipple), and calculated the tumor response for different antenna
locations. In a subsequent study, the antenna location was fixed and the effect o f breast
gland variation on the tumor response level was examined.
Next, we considered a smaller tumor located in one o f the off-center glands. We
examined tumor response for different antenna locations. The resulting data demonstrate
that the gland in which the tumor is embedded has a significant effect on the tumor
response o f the on-center and off-center tumor. More importantly, these studies suggest
that the mammary tissue, despite the fact that it introduces significant heterogeneity to the
geometry and electrical parameter variation, does not threaten the sensitivity o f the
microwave breast cancer detection method.
Often, sensing techniques with antennas involve more than one sensing element.
In anticipation that the here investigated imaging technique may use an antenna array, we
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
aim the last simulation studies reported in the work at the mutual effect between two
“Dark eyes” antenna by arranging three antennas around the breast model, with an angle
o f 90° between every two adjacent antennas. All three antennas were located in the plane
at mid-breast level, parallel to the chest wall. In each simulation, we excited one o f the
antennas, but kept the other two as passive elements, and tumor responses o f all three
antennas were recorded and analyzed. On-center and off-center tumors were taken into
consideration. We have observed negligible mutual coupling between adjacent antenna
elements. This shows that with the given spacing o f the antenna-elements, mutual
coupling would not present an obstacle in the tumor-detection process.
This work has demonstrated promising behavior o f the “Dark eyes” antenna in
microwave tumor detection o f a realistic 3-D breast model. Many possible avenues for
future work branch from this fascinating topic:
•
Further miniaturization and optimization o f the sensor-element;
•
Investigations o f closer element spacing within a hemi-spherical array;
•
Numerical modeling and simulation o f the microwave breast cancer detection
process on models developed from realistic geometries, e.g. MRI and ultrasound
images;
•
Identification o f difficulties and related strategies when multiple tumors are
present in the breast;
•
Development o f signal-processing image-formation schemes with noise
suppression and enhancement o f the tumor image
Microwave breast cancer detection and imaging is a novel technique still at its infant
stage. We hope that our work will contribute to a development o f a safe, cost-effective
imaging technique that will complement the existing modalities and thus help combat one
o f the most feared illnesses by women today.
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
References
[1] E.C. Fear, P.M. Meaney and M.A. Stuchly, “Microwaves for Breast Cancer
Detection?,” IEEE Potentials, vol. 22, Issue 1, Feb-Mar 2002, pp: 12-18.
[2] E.C. Fear and Michael Okoniewski, “Confocal Microwave imaging for breast tumor
detection: application to a hemispherical breast model”, Microwave Symposium
digest, 2002 IEEEM TT-SInternational, vol. 3, 2-7 June 2002, pp. 1759-1762.
[3] E.C. Fear and J.M. Sill, “Preliminary investigations o f tissue sensing adaptive radar
for Breast Tumor Detection”, Proceedings o f the 25th Annual International
Conference o f the IEEE Engineering in Medicine and Biology Society, Cancun,
Mexico, Sep 2003.
[4] M. Popovic, H. Kanj, Q. Han, "Comparison o f two broadband antennas for
microwave breast cancer detection," Proceedings o f the 3rd International Workshop
on Biological Effects o f Electromagnetic Fields, 4-8 October 2004, Kos, Greece, pp.
294-301.
[5] H. Kanj and M. Popovic: “Microwave-range broadband “Dark Eyes” antenna:
detailed analysis and design,” Antennas and Wireless Propagation Letters, vol. 4,
2005, pp: 262-265.
[6] H. Kanj and M. Popovic: “Miniaturized Microstrip-Fed Dark Eyes Antenna for
Pulsed Microwave Probing: Return Loss, Efficiency and Fidelity,” IEEE
International Symposium o f Antennas and Propagation Society, 2005.
[7] J. R. Harris, M. E. Lippman, U. Veronesi, and W. Willet, “Medical progress: Breast
Cancer,” the New England Journal o f Medicine, vol. 327, no. 5, pp. 319-328, 1992.
[8] S. C. Hagness, A. Taflove, and J. E. Bridges, "Three-dimensional FDTD analysis o f a
pulsed microwave confocal system for breast cancer detection: Design o f an antennaarray element," IEEE Transactions on Antennas and Propagation, vol. 47, pp. 783791, May 1999.
[9] E. C. Fear, S. C. Hagness, Paul M. Meaney, Michael Okoniewski, and Maria Stuchly,
“Enhancing Breast Tumor Detection with Near-Field Imaging,” Microwave
Magazine, vol. 3, Issue 1, March 2002, pp: 48 - 56.
[10] “Canadian Cancer Society Website,” http://www.cancer.ca, 2006.
[11] “Canadian Cancer Statistics 2005,” Report, 2005.
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[12] X. Yun, E.C. Fear and Ronald H. Johnston, “Compact Antenna for Radar-Based
Breast Cancer Detection,” IEEE Transactions on Antennas and Propagation, vol. 53,
no. 8, August 2005.
[13] “Breast Cancer Website,” http://www.breastcancer.org. 2005.
[14] “Medical Instrument Analysis and Machine Intelligence Group (MIAMI) Website,
University o f Waterloo,” http.7/www.ece.uwaterloo.ca/, 2006.
[15] Qingsheng Han, “FDTD Analysis o f Antennas Close to Human Tissue,” Master
Thesis, August 2004.
[16] C.Gabriel, S. Gabriel, and E. Corthout, “The dielectric properties o f biological
tissues— I: Literature Survey,” Physics in Medicine and Biology, vol. 41, no. 11, pp.
2231-2249, Nov. 1996.
[17] S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties o f biological
tissues— II: Measurements on the frequency range 10 Hz to 20 GHz,” Physics in
Medicine and Biology, vol. 41, no. 11, pp. 2251-2269, Nov. 1996.
[18] S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties o f biological
tissues— III: Parametric models for the dielectric spectrum o f tissues,” Physics in
Medicine and Biology, vol. 41, no. 11, pp. 2271-2293, Nov. 1996.
[19] K. R. Foster and J. L. Schepps, “Dielectric properties o f tumor and normal tissue at
radio through microwave frequency,” J. Microwave Power, vol. 16, pp. 107-119, 1981.
[20] J. A. Rogers, R. J. Shepard, E. H. Grant, N. M. Bleehen, and D. J. Honess, “The
dielectric properties o f normal and tumor mouse tissue between 50 MHz and 10 GHz,”
British Journal o f Radiology, vol. 56, pp. 335-338, 1983.
[21] E J. Bond, X.Li, S.C. Hagness, and B. D. Van Veen, “Microwave Imaging via
space-time beamforming for early detection o f breast cancer,” IEEE Transactions on
Antennas and Propagation, vol 51, pp.1690-1705, August 2003.
[22] X Li, S. C. Hagness, M.K. Choi, and D.W. van der Weide, “Numerical and
experimental investigation o f an ultra wideband ridged pyramidal horn antenna with
curved launching plane for pulse radiation,” IEEE Antennas and Wireless Propagation
Letters, vol. 2, Issue 1, 2003, pp:259-262.
[23] K. L. Shlager, G. S. Smith, and J. G. Maloney, “Optimization o f bow-tie antennas
for pulse radiation,” IEEE Transactions on Antennas and Propagation, vol. 42, no. 7,
July 1994.
[24] S.C. Hagness, A. Taflove, J.E. Bridges, “Wideband ultralow reverberation antenna
for biological sensing,” Electronic Letters, vol. 33, Issue 19, Sept. 1997, pp:1594 - 1595.
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[25] J.G. Maloney and G.S. Smith, “Optimization o f a conical antenna for pulse
radiation: An efficient design using resistive loading,” IEEE Transactions on Antennas
and Propagation, vol. 41, pp. 940-947, July 1993.
[26] H. Kanj and M. Popovic, “Miniaturized Microstrip-Fed “Dark Eyes” Antenna for
Near-Field Microwave Sensing,” IEEE Antennas and Wireless Propagation Letters, vol.
4, 2005.
[27] H. Kanj and M. Popovic, “FDTD Analysis o f Resistively loaded Broadband Dark
Eyes Antenna,” Proceedings o f IEEE Antennas and Propagation Society Symposium,
2004, vol. 4,20-25 June 2004, pp. 4527-4530.
[28] Semcad Website: http://www.semcad.com. 2006.
[29] X. Li and S.C.Hagness, “A confocal microwave imaging algorithm for breast cancer
detection,” IEEE Microwave Wireless Components Letters, vol. 11, pp. 130-132, Mar.
2001 .
[31] A. Taflove and S.C Hagness, Computational Electrodynamics: the finite-differemce
time-domain method, 2nd edition, Boston: Artech House, 2000.
[32] K.S. Lee, “Numerical solution o f initial boundary value problems involving
Maxwell’s equations in isotropic media,” IEEE Transactions on Antennas and
Propagation, vol. 14, no.3, pp. 302-307, May 1966.
[33] D.M. Sullivan, Electromagnetic Simulation Using the FDTD method, IEEE Press
Series on RF and Microwave Technology, 2000.
[34] Lawrence Duong, “Computational Electromagnetics in Microwave Hyperthermia,”
Master Thesis, June 2005.
[35] Breast Cancer Update, Part I, ''''Harvard Women’s Health watch,” Oct. 2000.
[36] X. Li, S.K. Davis, S.C. Hagness, D.W. van der Weide and B.D. Van Veen,
“Microwave Imaging via Space-Time Beamforming: Experimental Investigation of
Tumor Detection in Multilayer Breast Phantoms,” IEEE Transactions on Microwave
Theory and techniques ,vol.52, no.8, Aug. 2004.
[37] “Memorial Hermann health library website,” http://www.memorialhermann.org,
2006.
[38] “National Breast Cancer Foundation website,” http://www.nationalbreastcancer.org/,
2006.
[39] “Bremen Hospital website,” http://www.bremenhospital.com/, 2006.
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[40] “Radiology Info website,” http://www.radiologvinfo.org/. 2006.
[41] E. C. Fear, X Li, S. C. Hagness, and M. A. Stuchly, “Confocal Microwave Imaging
for Breast Cancer Detection: Localization o f Tumors in Three Dimensions,” IEEE
Transactions on Biomedical Engineering, vol. 49, no. 8, pp. 812-822, August 2002.
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix
Related Publications
•
M. Popovic, H, Kanj and N. Tavassolian, "Antenna Design Issues and Effects of
Tissue Properties on Sensing Breast Tumors with Microwaves", Presented at
the International Conference on Electromagnetic fields, Health and Environment,
Madeira Island, Portugal, April 2006
•
N. Tavassolian, H. Kanj and M. Popovic, "The Effect o f Breast Glands on
Microwave Tumor Sensing with Dark Eyes Antenna" (Accepted for IEEE
International Symposium o f the Antennas and Propagation Society, Albuquerque,
New Mexico, July 2006)
•
N. Tavassolian, H. Kanj and M. Popovic, "Assessment of Dark Eyes Antenna
Radiation in the Vicinity of the Realistic Breast Model" (Accepted for the 12th
International Symposium on Antenna Technology and Applied Electromagnetics
(ANTEM) Montreal, Canada, July 2006)
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
International Conference on Electromagnetic Fields, Health and Environment,
Madeira Island, Portugal, April 2006
Antenna Design Issues and Effects of Tissue Properties on
Sensing Breast Tumors with Microwaves
Milica Popovic
Houssam Kanj ('2>and Negar Tavassolian ^
Department of Electrical and Computer Engineering
McGill University, Montreal, CANADA H3A 2A7
(1)poppy@ece.mcgill.ca
(2)hkani @,po-box.mc gill .ca
(3)ntavas@po-box.mcgill.ca
Abstract
In recent years, detection of breast tumours with microwaves has been proposed as a
new approach that could complement the existing detection and imaging modalities.
In this paper, we present finite-difference time-domain analysis of the “Dark Eyes”
antenna in the vicinity of complex tissue structures to assess the potential problems
associated with near-field tumour sensing. First, we observe response of a 5-mm
tumour embedded in a simplified tissue-layered half-space for three different radial
distances from the antenna to estimate the response drop-off with tumour location
depth. Then, we place the antenna next to a realistic hemi-spherical breast model to
determine the effect of gland property variation on detection capabilities.
Keywords: microwaves, breast tumours, antenna design, sensing
1. Introduction
In recent years, pulsed microwave imaging has been suggested for early detection o f
breast cancer [1]. These findings are based on the dielectric contrast between the
cancerous and fatty breast tissue. Due to the higher water content in malign tissues,
permittivity and conductivity o f tumors at microwave frequencies is almost an order o f
magnitude larger than that o f the surrounding healthy breast tissue. In addition, the
attenuation o f microwave-range signals in fatty breast material does not exceed 4dB/cm,
making the fatty medium “transparent” and the embedded tumor “visible” to microwaves.
To ensure transmission and reception o f significant Fourier components o f the pulse
spectrum, the sensing antenna needs to exhibit broadband behavior in the frequency
range o f interest (~6GHz).
2. Problem geometry
In this paper, we use a finite-difference time-domain tool (SEMCAD) to numerically
examine the “Dark Eyes” antenna [2] radiation in the vicinity o f simple, tissue-layered
structure (Fig.l), which hides a 5-mm embedded tumour. The antenna is excited with a
120-ps differentiated Gaussian pulse, and the backscatter is observed at the source
location. Backcattered signal is simulated for the no-tumor geometry and subtracted from
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
International Conference on Electromagnetic Fields, Health and Environment,
Madeira Island, Portugal, April 2006
the result obtained with the embedded tumour to calcultate the tumour-only response for
each location o f tumour noted in Fig. 1. The antenna is immersed in a matching medium,
having the same dielectric constant as the antenna substrate.
\C o n sta n t su rface
resistive loading Rs
Ground plan e
Metallic
Matching
medium
Skin
Fat
d
(r, 6, <t>)
T
Lu
om
caotirons
Fig. 1. Simplified geometry o f tissue layers used to simulate “Dark Eyes” antenna radiation with complex
media in its near-field region. A 5-mm tumor is embedded in the fatty tissue at selected indicated locations
for three radial distances from the antenna apex: r = 3cm, 4cm and 5cm. The electrical parameters, relative
permittivity and conductivity, at the central frequency o f 6 GHz are as follows: for skin, er = 36 and <7 = 4
S/m; for fatty breast tissue, er = 9 and a = 0.4 S/m; for the tumor, er = 50 and a = 7 S/m eT= 36, [1]. The
antenna [2] is a dual-layer PCB structure (dielectric substrate er = 10.2) with overall dimensions o f 20mm
x 21.75mm x 1.4mm.
Further simulations are carried for the antenna located in the middle horizontal
cross-sectional plane o f the realistic hemi-spherical breast geometry depicted in Fig. 3 (a)
and adopted from [3]. The glands randomly embedded in the hemispherical breast model
are o f cylindrical and spherical shapes with varying electrical parameters (ll< e r<15,
0.4<ff< 0.5). A 6-mm tumor is located in the central mammary gland. Tumor response is
calculated for antenna rotated in the x-y plane around the central axes o f the
hemispherical geometry in 45-angle increments.
3. Results and conclusions
Fig.2 shows the tumour response level maps, obtained by interpolation o f results for
points indicated in Fig. 1 for three radial distances from the antenna apex. These results
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
International Conference on Electromagnetic Fields, Health and Environment,
Madeira Island, Portugal, April 2006
illustrate two main points. First, the slight asymmetry o f the main antenna radiation lobe
must be taken into account when interpreting the backscatter data from more complex
geometries. Second, the tumour response level is decreased by ~10dB for each additional
centimetre o f tumour location away from the sensing element.
r
L
r=
4cm
r = 5cm
Fig. 2. Simulated tumor response level interpolated from results obtained for selected locations indicated in
Fig. 1 for three radial distances from the antenna apex. The results demonstrate the slight asymmetry in
antenna radiation and approximate drop o f ~10dB in tumor response level per centimeter o f tumor location
depth.
Fig.3.(b) tabulates tumour-response levels as a function o f angle by which the antenna is
rotated around the axes o f the hemispherical breast model o f Fig.3(a). Results suggest
that the overall decrease in the tumour response level due to multiple reflections from the
heterogeneities introduced by the presence o f mammary glands does not exceed ~12dB
with respect to the case o f the tumour embedded in the same location but with no
mammary glands present.
A __________
0, no glands
0°
45°
90°
135°
O
O
00
225°
270°
'315°
Tumor response (dB)
- 119
-118.5
- i26
-117
-128.5
-120
-125
- 122.5
-124.5
(b)
(a)
Fig. 3. (a) Hemispherical breast geometry adopted from [3] and the calculated tumour-response
levels for a 6-mm tumour embedded in the central mammary gland.
Acknowledgements
The authors are grateful to Nature Sciences and Engineering Council o f Canada
(NSERC) for their support and the Schmidt & Partner Eng. for usage o f SEMCAD tool.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
International Conference on Electromagnetic Fields, Health and Environment,
Madeira Island, Portugal, April 2006
References
[1] E.C. Fear, X. Li, S.C. Hagness, and M. A. Stuchly, “Confocal microwave imaging for breast
cancer detection: localization of tumors in three dimensions,” IEEE Trans. Biomed. Eng., vol. 49,
pp 812-822, Aug. 2002.
[2] H. Kanj and M. Popovic, “Miniaturized microstrip-fed ‘Dark Eyes’ antenna for near-field
microwave sensing,” IEEE Antennas and Wireless Prop. Lett., vol. 4, pp. 397-401, 2005.
[3] E. C. Fear and M. Okoniewski, “Confocal microwave imaging for breast tumor detection:
application to a hemispherical breast model”, 2002 IEEE MTT-SInternational Symposium, vol. 3,
2-7 June 2002, pp. 1759-1762.
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IEEE AP-S International Symposium, Albuquerque, N ew M exico, July 2006
The Effect of Breast Glands on
Microwave Tumor Sensing with “Dark Eyes” Antenna
Negar Tavassolian* (1), Houssam Kanj(2) and Milica Popovic(3)
Department o f Electrical and Computer Engineering, McGill University
Montreal, Canada H3 A 2A7
(1 )negar.tavassolian@mcgill.ca
(2)houssam .kanj @mail .mcgill .ca
(3) poppy@ece.mcgill.ca
In t r o d u c t io n
In recent years, pulsed microwave imaging has been suggested for early detection o f
breast cancer [1] [2]. These findings are based on the dielectric contrast between the
cancerous and fatty breast tissue. Due to the higher water content in malign tissues,
permittivity and conductivity o f tumors at microwave frequencies is almost an order o f
magnitude larger than that o f the surrounding healthy breast tissue. In addition, the
attenuation o f microwave-range signals in fatty breast material does not exceed 4dB/cm,
making the fatty medium “transparent” and the embedded tumor “visible” to microwaves.
To ensure transmission and reception o f significant Fourier components o f the pulse
spectrum, the sensing antenna needs to exhibit broadband behavior in the frequency
range o f interest (~6GHz).
In this paper, we present finite-difference time-domain analysis o f the “Dark Eyes”
antenna [3] [4] in the vicinity o f complex tissue structures to assess the potential
problems associated with near-field tumour sensing. With the antenna placed next to a
realistic hemi-spherical breast model, we observe the response o f a 6-mm tumour
embedded in a gland. The main parameters o f our study are the location o f the antenna
relative to asymmetrical inhomogeneous cluster o f glands and electrical parameters o f the
gland within which the tumour is assumed to be embedded.
G eo m etry o f the pro blem and param eters
The hemispherical breast model used in the analysis is shown in Fig. 1. In order to assess
the variation o f glandular tissue on the detection levels (tumor response), differently sized
glands o f spherical (radius 8.5mm < r < 12.5mm) and cylindrical (radius 2mm < r <
18mm, height 12mm < h < 27mm) shape were embedded in the fatty tissue. In addition,
their electrical parameters (relative permittivity er and conductivity a) varied in the range
11< e r-giand <15 and 0.4 S/m < a giand < 0.5 S/m [1] [2]. The electrical parameters o f the
remaining tissues assumed in the simulation model were: fat er.fat = 9, <7f-at = 0.4 S/m; skin
(2-mm) €r-skin = 36, <7skjn = 4 S/m [3]; tumor (diameter 6mm) e r-tum or = 50, fftum or = 7 S/m
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IEEE AP-S International Symposium, Albuquerque, N ew M exico, July 2006
[3], nipple ^-nipple= 45, anippit - 5 S/m, chest wall (estimated 9cm*9cmx2cm) 6r.chest ^ 50,
Ochest = 7 S/m [2], non-conductive immersion liquid matched to the antenna dielectric
substrate er-iiquid = 10.2, <7iiquid = 0 [4]. For the main central gland, located below the
nipple and containing the tumor, er-tumor-giand “ 15 and Otumor-giand - 0.4 S/m for the study of
varying antenna location, but its properties served as a varying parameters in the
subsequent study when the antenna location was fixed.
Details o f antenna design and the excitation pulse (differentiated Gaussian, 120ps,
centered at 6 GHz) can be found in [4], The analysis was performed with SEMCAD-X
[5], a simulation tool based on the finite-difference time-domain (FDTD) method.
In -g lan d T u m o r
Chest Wall i
(a)
(b)
Figure 1: The hemispherical breast model (total diameter o f 14cm), (a) general view, showing the antenna
located near the model and (b) cross-sectional side view. Adapted from [1], [2] and [6].
E f f e c t o f g l a n d u l a r t is s u e o n d e t e c t io n l e v e l s
For the here presented results, antenna located at the mid-breast cross-sectional plane
parallel to the chest wall, as depicted in Fig. 1(a). The antenna was then rotated to eight
locations defined by uniform increments o f 45° o f angle $ , as shown in Fig. 2(a). Since
the main radiation lobe o f the antenna will “see” different glandular arrangement from
each location, it is expected that the tumor response will vary. Indeed, Fig. 2(b)
demonstrates that the changing gland arrangement relative to the antenna location results
in tumor response that varies around ±5dB with respect to the case o f no glands
surrounding the tumor in the central gland. For reference, the graph o f Fig. 2(b) also
shows the level o f tumor detection response when the tumor is located in the fat tissue
only, which is an approximate 4dB higher than the response o f a tumor hidden in the
main gland.
The results graphed in Fig. 2(b) suggest that the presence o f multiple glands o f different
shapes, sizes and electrical parameters, with varying arrangement with respect to the
sensing antenna, cause not only attenuation o f the signal and then, in turn, degradation of
the tumor response, but also multiple reflections which can yield a response that could be
falsely interpreted as a response from a larger tumor. It is key for this fact to be taken into
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IEEE AP-S International Symposium, Albuquerque, N ew M exico, July 2006
account when image-construction algorithms are designed to process signals from either
array-elements or multiple locations o f mechanically-scanned antennas.
-116
„ -1 1 8
--------Only One Gland
0
/
\
g -122
A
Q.
if)
2
Complete Model
-------- No glands
; \
co
"O 1
— -120
<
if)D
-0
9
.
“124
V
I -126
3
l _ -128
-130
0
45
90
135
180
1
1
22 5
270
315
♦H
(b)
(a)
Figure 2. (a) Central cross-sectional plane o f the breast model o f Fig. 1, parallel to the chest wall. The
antenna is located in this plane and is rotated to eight different locations for 45 "-increments o f angle <F. (b)
Tumor response levels calculated for each location o f the antenna. For reference, tumor response levels are
shown also for the case o f no glandular tissue surrounding the main (tumor-containing) gland is present and
for tumor being embedded in the fat tissue only.
S e n s it iv it y o f t u m o r r e s p o n s e l e v e l s t o e l e c t r ic a l
PARAMETERS OF THE CENTRAL (TUMOR-CONTAINING) GLAND
Next, we investigate the fixed-location antenna scenario with varying electrical
parameters o f the central gland in which the tumor is embedded.
-117
-117.5
w
-118
8- -us
Z *118.5
-118.5
-
11 !
' 118.1
0.2
0.3
0.4
0.5
0.6
0.7
a (SI m)
(a)
(b)
Figure 4. Variation o f tumor response with (a) relative permittivity and (b) conductivity o f the central gland
in which the tumor is embedded.
Fig. 4(a) graphs the deterioration o f the tumor response from -117.2dB to -119dB as the
permittivity o f the central gland varies ±25% around the previously assumed value, from
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IEEE AP-S International Symposium, Albuquerque, N ew M exico, July 2006
= H to 6 r -tumor-gland “ 19, while itS Conductivity WaS kept constant at <7tumot-gland =
0.4 S/m. Fig. 4(b) shows the degradation in the tumor response level from -1 17.9dB to 118.8dB as the conductivity o f the central gland varies in the range 0.1 S/m < Otumor-giand <
0.7 S/m (±75% from its assumed value) and the permittivity is kept at er-tumor-giand = 1 5 .
These results suggest two conclusions. First, the tumor response is less pronounced as the
electrical parameters o f the gland containing it approach those o f the tumor itself.
Second, even though this deterioration is noticed, it does not exceed ~2dB, implying that
the variation o f gland electrical parameters may not have a significant effect on the
sensitivity o f the tumor detection process with microwaves.
A tu m o r-g la n d
A cknow ledgem ent
This work was funded by Natural Science and Engineering Research Council (NSERC)
o f Canada. The authors are grateful for their support. We would like to thank Schmid &
Partner Engineering, AG, for providing and helping us with the SEMCAD software tool.
R eferences
[1] E.C. Fear, P.M. Meaney and M.A. Stuchly, “Microwaves for Breast Cancer
Detection?” IEEE Potentials, vol. 22, Issue 1, Feb-Mar 2003, pp:12-18
[2] E.C. Fear and Michael Okoniewski, “Confocal Microwave Imaging for breast tumor
detection: application to a hemispherical breast model”, Microwave Symposium
digest, 2002 IEEE MTT-SInternational, vol. 3, 2-7 June 2002, pp. 1759-1762.
[3] M. Popovic, H. Kanj, Q. Han, "Comparison o f two broadband antennas for
microwave breast cancer detection", Proc. 3rd International Workshop on Biological
Effects o f Electromagnetic Fields, 4-8 October 2004, Kos, Greece, pp. 294-301.
[4] H. Kanj and M. Popovic, “Miniaturized Microstrip-Fed "Dark Eyes" Antenna for
Near-Field Microwave Sensing,” IEEE Antennas and Wireless Propagation Letters,
vol. 4, pp. 397—401, 2005.
[5] http://www.semcad.com
[6] E.C. Fear and J.M. Sill, “Preliminary investigations o f tissue sensing adaptive radar
for breast tumor detection”, Proc. 25th International Conference o f the IEEE EMBS,
Cancun, Mexico, September 17-21, 2003.
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ANTEM International Symposium, Montreal, Canada, July 2006
“ D A RK EY ES” ANTENNA FO R M IC RO W A V E TU M O R IM A G IN G :
ASSESSM ENT O F M UTUAL C O U PLIN G B ETW EEN ARRAY ELEM EN TS
Negar Tavassolian* and Milica Popovic**
McGill University, Montreal, Canada
Email:
* negar.tavassolian@mcgill.ca
** DODPV@.ece.mcgill.ca
INTRODUCTION
•
skin (2-mm) er.skin = 36, ffskin = 4 S /m ;
In recent years, the contrast o f the microwave
electric properties between the malign lesions
and the healthy breast tissue has been studied as
a foundation o f novel ways to image early breast
tumors [1] [2], One o f the key factors in pulsed
microwave (~6 GHz) breast cancer detection is
the design o f a suitable broadband antennaelement. We continue our study o f the “Dark
Eyes” antenna [3] [4] for this purpose.
•
tumor (diameter 4mm)
S /m ;
•
nipple 6r.nippie = 45, ffnipple = 5 S /m ;
“z
In this paper, we focus our attention on a
prelimiry investigaiton o f the antenna as a part of
an antenna array. The finite-difference timedomain (FDTD) model is used to analyse the
radiation and tumor detection properties o f the
“Dark Eyes” antenna with realistic breast tissue
geometry placed in its near-field region.
Specifically, after trasmitting a pulse from the
antenna, we observe the response o f a 4-mm
tumour embedded in an off-center gland. As the
location o f the antenna changes in relation to
asymmetrical inhomogeneous cluster o f glands,
we quantify the effect o f gland presence on the
tumour response level. Next, we examine the
case o f three antennas located around the breast
model. Mutual coupling between antennas is
examined for two cases: when the tumor is
located in the central gland, and when it is
situated in the gland off the axis o f the
hemispherical breast model.
(a)
Nipple
In-gland Tum or
B t Tj
Chest Wall
GEOMETRY OF THE PROBLEM
The hemispherical breast model used in the
analysis is shown in Fig. 1. In order to assess the
variation o f glandular tissue on the detection
levels (tumor response), differently sized
spherical (radius 8.5mm < r < 12.5mm) and
cylindrical (radius 2mm < r < 18mm, cylinder
height 12mm < h < 27mm) glands were
embedded in the fatty tissue. In addition, their
electrical parameters (relative permittivity er and
conductivity a) varied in the range 11< er.giand
<15 and 0.4 S/m < <rgland < 0.5 S/m [1] [2], The
electrical parameters o f the remaining tissues
assumed in the model were as follows [2] [3] [4]:
•
= 50, atamor= 7
fat er.fat = 9, fffat = 0.4 S /m ;
(b)
Figure 1: The hemispherical breast model (total
diameter of 14cm). (a) General view, showing the
“Dark Eyes” antenna located near the model and (b)
cross-sectional side view. Adapted from [1], [2] and
[6],
•
chest wall (9cmx9cmx2cm)
= 7 S/m;
•
€ r-chest
= 50, ffchest
non-conductive immersion liquid matched to
the antenna dielectric substrate e r -iiquid =
10.2, (Tljquid = 0 [4].
In a previous study, we reported a 6-mm
diameter spherical tumor embedded inside a
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ANTEM International Symposium, Montreal, Canada, July 2006
gland at the center o f the breast [7], with the
focus o f the work being the analysis o f the effect
o f breast gland parameter values on the tumor
response level. For the first part o f the here
presented work, a 4-mm tumor is embedded in
an off-center cylindrical gland o f £r.iumor-giand =13
and Otumor-giand = 0.4 S/m. The dimensions o f the
gland are r = 10mm, h = 24mm. The tumor
response is observed as we change the location
o f the antenna with respect to the asymmetrical
gland distribution (and the tumor location) by
rotating it around the main hemispherical axis
within the horizontal mid-plane o f the breast
model. The subsequent section reports on mutual
coupling effect on tumor response level. Two
cases are considered: on-center tumor (6-mm
diameter) and off-center tumor (4-mm diameter).
locations), exceeding the level o f response o f the
complete model by 2-4 dB.
These results suggest that the asymmetrical
distribution o f varying glands has the potential to
affect the backscattered signal (the tumor
response) in two ways. First, the signal can be
attenuated, degrading the signal-to-noise ration.
Second, the alternative effect is a consequence o f
multiple reflections and scattering, which can
yield a higher-level response, potentially
erroneously interpreted as a response from a
larger or closer tumor. It is important for these
issues to be taken into account when imageconstruction algorithms are designed to process
signals from either array-elements or multiple
locations o f mechanically-scanned antennas.
Details o f antenna design and the excitation
pulse (differentiated Gaussian, 120ps, centered at
6 GHz) can be found in [4]. The analysis was
performed with SEMCAD-X [5], a simulation
tool based on the finite-difference time-domain
(FDTD) method.
EFFECT OF ASYMMETRICALY
DISTRIBUTED GLANDULAR TISSUE ON
DETECTION LEVELS
For the here presented results, the antenna is
located at the mid-breast cross-sectional plane
parallel to the chest wall, as depicted in Fig. 1(a).
The antenna was then rotated to eight locations
defined by uniform increments o f 45° o f angle # ,
as shown in Fig. 2(a). Fig. 2(b) demonstrates the
tumor response variation with respect to the
rotation angle. Tumor response variation is
almost 26dB for the range o f locations that cover
a full rotation o f the sensor around the main
hemispherical axis. For comparison, two more
cases are examined. First, all the glands except
the tumor-containing gland are removed, in order
to observe the role o f the tumor-containing gland
on tumor response level. In the next scenario, the
tumor-containing gland is also removed and the
tumor is embedded in fat only. Fig. 2(b)
compares these three cases.
Only O n e G land
Fat-Only
C o m p le te M odel
♦n
(b)
The results in Fig. 2(b) demonstrate that the
tumor response in the one-gland only model
closely follows that o f the complete model.
However, the response is lower in the one-gland
only model for a rotation angle o f 180 degrees or
more. The fat-only model has the highest tumor
response for most rotation angles (i.e. antenna
Figure 2: (a) Central cross-sectional plane of the breast
model of Fig. 1, parallel to the chest wall. The antenna
is located in this plane and is rotated to eight different
locations for 45” increments of angle <F. (b) Tumor
response levels calculated for each location of the
antenna. Tumor response levels are shown for the
cases of the complete model, fat-only model, and onegland only model.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ANTEM International Symposium, Montreal, Canada, July 2006
MUTUAL COUPLING
Tumor responses for the three antennas are
summarized in Tables I and II. For reference, the
tumor response o f each antenna in its active
mode (with no other elements present) is also
shown.
As a preliminary study o f an array configuration,
we report investigations o f mutual impact o f
adjacent antenna-elements placed around the
breast. Three antennas are located around the
breast in the mid-plane parallel to the chest wall.
Adjacent antennas are separated by an angle o f
90° with respect to the central hemispherical
axis. We excite the first antenna, but keep the
other two passive, and we record the tumor
response at all three antenna locations. Two
cases are studied: in the first case, the tumor is a
6-mm sphere located at the center o f the breast
(as in [7]) inside a cylindrical gland o f e r-tu„ior-giand
=15 and f f tm„ 0r-giand = 0.4 S/m. The gland is o f
radius r = 8mm and is h = 24 mm high. In the
second case, the tumor is a 4-mm diameter
sphere, encapsulated in an off-center gland, as
noted in the previous section.
For the on-center tumor, tumor response o f the
active element decreases as little as 0.15 dB
when the passive elements are present. This
shows the small effect o f the passive elements on
the tumor response o f the active antenna. The
passive elements receive a response which is
very close to their response for their active mode.
For the off-center tumor, the active antenna
shows no change in tumor response when the
passive elements are present. This result is not
surprising, since, in the geometry studied, the
off-center tumor is closer to the radiating (active)
antenna than to the passive elements. The
passive elements show a small decrease in tumor
response level compared to the case when they
radiate alone, as expected. These results suggest
that the mutual coupling, in the arrangement
studied, has little effect on tumor response levels
o f the “Dark Eyes” antenna.
Figs. 3 and 4 show the antenna locations for both
the on-center and the off-center cases. In both
cases, antenna (1) is excited, and antennas (2)
and (3) are passive.
T a b le
I.
T u m o r respon se lev el d etected by th e three
ANTENNAS IN F lO . 3 , WITH TUMOR IN THE CENTRAL
GLAND OF THE BREAST MODEL.
Tumor
Antenna #
response
(dB)
Figure 3: On-center tumor with three antennas.
Tumor response
of the antenna in
the active mode
without the other
two antennas
present (dB)
-118.526
1 (Active)
-118.673
2 (Passive)
-121.460
-122.571
3 (Passive)
-117.608
-117.234
Figure 4: Off-center tumor with three antennas
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ANTEM International Symposium, Montreal, Canada, July 2006
Table II.
Finally, the authors would like to extend their
gratitude to Mr. Houssam Kanj for his invaluable
help on accurate modeling o f the “Dark Eyes”
antenna.
T u m o r respon se level d etected by the th ree
ANTENNAS IN FlG . 4 , WITH TUMOR IN THE OFF-CENTER
GLAND OF THE BREAST.
1 (Active)
-110.385
Tumor response
of the antenna in
the active mode
without the other
two antennas
present fdB)
-110.385
2 (Passive)
-126.549
-125.474
3 (Passive)
-133.132
-131.784
Tumor
Antenna #
response
(dB)
REFERENCES
[1] E.C. Fear, P.M. Meaney and M.A. Stuchly,
“Microwaves for Breast Cancer Detection?”
IEEE Potentials, vol. 22, Issue 1, Feb-Mar
2003, pp. 12-18.
[2] E.C. Fear and Michael Okoniewski,
“Confocal Microwave Imaging for breast
tumor
detection:
application
to
a
hemispherical breast model”, Microwave
Symposium digest, 2002 IEEE MTT-S
International, vol. 3, 2-7 June 2002, pp.
1759-1762.
[3] M. Popovic, H. Kanj, Q. Han, "Comparison
o f two broadband antennas for microwave
breast cancer detection", Proc.
3rd
International Workshop on Biological
Effects o f Electromagnetic Fields, 4-8
October 2004, Kos, Greece, pp. 294-301.
[4] H. Kanj and M. Popovic, “Miniaturized
Microstrip-Fed "Dark Eyes" Antenna for
Near-Field Microwave Sensing,” IEEE
Antennas and Wireless Propagation Letters,
vol. 4, pp. 397-401,2005.
[5] http://www.semcad.com
[6] E.C. Fear and J.M. Sill, “Preliminary
investigations o f tissue sensing adaptive
radar for breast tumor detection”, Proc. 25th
International Conference o f the IEEE
EMBS, Cancun, Mexico, September 17-21,
2003.
[7] N. Tavassolian, H. Kanj and M. Popovic,
"The Effect o f Breast Glands on Microwave
Tumor Sensing with Dark Eyes Antenna",
IEEE AP-S International Symposium,
Albuquerque, New Mexico, July 2006
CONCLUSION
This paper reports on two issues related to the
microwave breast cancer detection, both
investigated for the “Dark Eyes” antenna sensor.
First, the simulations suggest significant effect o f
gland containing the tumor and the surrounding
glandular tissue variation on the tumor response
level. Second, for three coplanar antennas
surrounding part o f the breast in the plane
parallel to the chest wall, with adjacent elements’
axes angled with respect to each other at 90°,
mutual coupling was shown to have little effect
on the performance o f the “Dark Eyes” antenna
sensor. Immediate near-future work includes
simulations o f tumor response based on MRIderived three-dimensional breast model and
study o f signal-processing schemes needed for
image construction.
ACKNOWLEDGEMENT
This work was funded by Natural Science and
Engineering Research Council (NSERC) o f
Canada. The authors are grateful for their
support. We would like to thank Schmid &
Partner Engineering, AG, for providing and
helping us with the SEMCAD software tool.
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
2 162 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа