Citation
Numerical modeling for non-invasive imaging using ultrahigh frequency electromagnetic waves

Material Information

Title:
Numerical modeling for non-invasive imaging using ultrahigh frequency electromagnetic waves
Creator:
Alzuhiri, Mohand ( author )
Language:
English
Physical Description:
1 electronic file (132 pages). : ;

Subjects

Subjects / Keywords:
Nondestructive testing ( lcsh )
Three-dimensional imaging ( lcsh )
Electromagnetic testing ( lcsh )
Electromagnetic testing ( fast )
Nondestructive testing ( fast )
Three-dimensional imaging ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Nondestructive evaluation (NDE) is an important field in industry that has a wide range of applications to guarantee the quality of new products, maintain the safety of working equipment and perform medical diagnosis. In this thesis work, we are presenting two noninvasive NDE imaging techniques, microwave-induced thermoacoustic imaging for breast cancer diagnosis and structured light 3D characterization for gas pipes internal surface reconstruction. Thermoacoustic imaging is an emerging imaging technique that combines both high contrast of microwave imaging and sub-millimeter resolution of ultrasound imaging. This imaging technique is highly affected by acoustic and electrical properties of the imaged target. Other important factors are the properties of the excitation source such as its electric field pattern and pulse duration. This thesis developed a numerical model to simulate the signal generation of near-field microwave-induced thermoacoustic imaging. The study simulates both the electromagnetic interaction of microwaves with the imaged object, as well as the generation of acoustic signals and their propagation from the target to the transducers. The study also investigates the effect of different design parameters that determine the intensity and frequency of generated acoustic signals for imaging. Various parameters are investigated, such as target’s electrical properties, microwave frequencies, and the microwave source pulse width. Another major contribution of this thesis work is on data reconstruction for optical NDE methods. Structured light has been widely used to create 3D models of objects. In this thesis, we propose to use this technique to characterize and analyze damage and defects inside natural gas pipelines. The study aims to develop and implement 3D reconstruction algorithms to characterize the defects and deformations while the data are acquired using the optical sensors developed at CU LEAP group. The study develops a reconstruction algorithm for single laser ring laser ring scanner and evaluates its performance.
Thesis:
Thesis (M.S.)--University of Colorado Denver.
Bibliography:
Includes bibliographic references
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Mohand Alzuhiri.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
953734692 ( OCLC )
ocn953734692

Downloads

This item has the following downloads:


Full Text
NUMERICAL MODELING FOR NON-INVASIVE IMAGING USING ULTRAHIGH
FREQUENCY ELECTROMAGNETIC WAVES
by
MOHAND ALZUHIRI
B.S., University of Baghdad, 2011
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Electrical Engineering
2016


2016
MOHAND ALZUHIRI
ALL RIGHTS RESERVED


Ill
This thesis for the Master of Science degree by
Mohand Alzuhiri
has been approved for the
Electrical Engineering
By
Yiming Deng, Chair
Mark Golkowski
Stephen Gedney
May 4, 2016


IV
Alzuhiri, Mohand (M.S. Electrical Engineering)
Numerical Modeling for Non-invasive Imaging Using Ultrahigh Frequency Electromagnetic
Waves
Thesis Directed by Assistant Professor Yiming Deng
ABSTRACT
Nondestructive evaluation (NDE) is an important field in industry that has a wide range
of applications to guarantee the quality of new products, maintain the safety of working
equipment and perform medical diagnosis. In this thesis work, we are presenting two
noninvasive NDE imaging techniques, microwave-induced thermoacoustic imaging for breast
cancer diagnosis and structured light 3D characterization for gas pipes internal surface
reconstruction. Thermoacoustic imaging is an emerging imaging technique that combines both
high contrast of microwave imaging and sub-millimeter resolution of ultrasound imaging. This
imaging technique is highly affected by acoustic and electrical properties of the imaged target.
Other important factors are the properties of the excitation source such as its electric field
pattern and pulse duration. This thesis developed a numerical model to simulate the signal
generation of near-field microwave-induced thermoacoustic imaging. The study simulates both
the electromagnetic interaction of microwaves with the imaged object, as well as the generation
of acoustic signals and their propagation from the target to the transducers. The study also
investigates the effect of different design parameters that determine the intensity and frequency
of generated acoustic signals for imaging. Various parameters are investigated, such as targets
electrical properties, microwave frequencies, and the microwave source pulse width. Another
major contribution of this thesis work is on data reconstruction for optical NDE methods.
Structured light has been widely used to create 3D models of objects. In this thesis, we propose
to use this technique to characterize and analyze damage and defects inside natural gas


V
pipelines. The study aims to develop and implement 3D reconstruction algorithms to
characterize the defects and deformations while the data are acquired using the optical sensors
developed at CU LEAP group. The study develops a reconstruction algorithm for single laser
ring laser ring scanner and evaluates its performance.
The form and content of this abstract are approved. I recommend its publication.
Approved: Yiming Deng


VI
DEDICATION
I dedicate this work to my father, mother and my family who supported me
throughout my life and motivated me to challenge the difficulties of life and become the
person who I am.


ACKNOWLEDGMENT
This work was supported by the higher committee of education development in Iraq. I
want to thank them for giving me this great opportunity to finish my master degree in the
United States. I want to thank to my LEAP group for their kind support and advice. Without
our teamwork I wouldnt have achieved my goals. I also want to thank my partner Ryan Jacob
for his great help with the experimental work. Special thanks to my advisor Dr. Yiming Deng
for his invaluable advices and guidance through the thesis work. Finally, I want to thank Dr.
Golkowski and Dr Gedney for their help and advice on electromagnetic theory and simulations.


Vlll
TABLE OF CONTENTS
Chapter
1. Introduction....................................................................1
1.1. Breast cancer...............................................................1
1.2. Microwave-induced thermoacoustic imaging....................................4
1.3. Pipelines internal surface reconstruction...................................6
1.4. Structured light............................................................8
1.5. Thesis scope................................................................9
1.6. Motivation.................................................................10
2. Theoretical background.........................................................11
2.1. Electromagnetic waves propagation and absorption...........................11
2.1.1. Electromagnetic spectrum..............................................11
2.1.2. EM waves propagation..................................................13
2.1.3. Boundary conditions...................................................15
2.1.4. Attenuation...........................................................16
2.1.5. Near field and far field..............................................17
2.1.6. Rectangular waveguide.................................................18
2.1.7. Microwave heating.....................................................19
2.2. Acoustic propagation theory................................................20
2.2.1. Wave equation.........................................................21
2.2.2. Acoustic impedance....................................................22
2.2.3. Acoustic wave propagation in inhomogeneous media......................22
2.2.4. Acoustic boundary conditions..........................................23
2.3. Thermoacoustic equation....................................................24
2.4. Light tri angulation:......................................................26
3. Numerical modeling of thermoacoustic signal generation.........................28
3.1. Experimental Imaging model.................................................28
3.2. COMSOL.....................................................................29
3.3. Electromagnetic waves simulation...........................................30
3.3.1. Simulation geometry...................................................30
3.3.2. Simulation parameters.................................................32
3.3.3. Acoustic propagation model............................................33


IX
3.3.4. Simulation parameters....................................................34
3.4. Model validation.............................................................34
3.5. Open ended waveguide vs horn antenna......................................37
3.6. Case studies.................................................................39
3.6.1. Effect of frequency......................................................39
3.6.2. Effect of material conductivity..........................................39
3.6.3. Effect of microwave pulse width..........................................39
3.6.4. Optimal target analysis..................................................40
4. Image reconstruction..............................................................43
4.1. K-Wave.......................................................................43
4.2. Reconstruction challenges....................................................44
4.2.1. Homogeneity of sound speed...............................................44
4.2.2. Estimation of sound speed................................................44
4.2.3. Size of transducer.......................................................45
4.2.4. Effect of transducer orientation.........................................45
4.3. Image reconstruction with time reversal......................................46
4.4. Simulation study of image reconstruction with time reversal..................48
4.4.1. Effect of MW pulse width.................................................50
4.4.2. Biological samples embedded in a conductive medium.......................51
4.5. Reconstruction of experimental data........................................52
.4.5.1 Uniform piece of meat....................................................52
4.5.2. Rectangular biological tissue with copper wire...........................53
4.5.3. The effect of inserting highly conductive tissue.........................55
4.5.4. Two separate pieces of meat..............................................56
5. Pipe internal surface reconstruction..............................................58
5.1. S canner setup...............................................................58
5.2. Extraction of 3D information.................................................62
5.3. Experimental results.........................................................66
5.4. Resolution and accuracy of structured light reconstruction...................69
6. Conclusion and future work........................................................71
References............................................................................73


X
LIST OF TABLES
Table
1.1: Sound speed in breast tissues...................................................4
2.1: Electric properties of different tissues at 2.5GHz..............................12
3.1: Variation of ethylene glycol properties with microbubbles concentration.........35
4.1: Properties of simulated geometry................................................49
4.2: Properties simulated materials..................................................52


XI
LIST OF FIGURES
Figure
1.1: Normal human breast.............................................................3
1.2: Generation of thermoacoustic signals............................................5
1.3: External inspection of pipelines................................................7
1.4: Internal inspection of pipelines................................................8
1.5: Structured light based depth calculation........................................9
2.1: Electromagnetic spectrum.......................................................12
2.2: Field regions around antenna...................................................17
2.3: Rectangular waveguide..........................................................19
2.4: Optical triangulation..........................................................26
3.1: Modeling overview for MITAI....................................................28
3.2: Experimental setup of MITAI....................................................29
3.3: 3D geometry of the EM simulation...............................................31
3.4: Cross section plane through the 3D geometry....................................32
3.5: Meshing of 3D geometry.........................................................33
3.6: Acoustic simulation region.....................................................34
3.7: Meshing of acoustic region.....................................................35
3.8: Comparison between simulation and experimental results.........................36
3.9: Comparison with other simulation work..........................................37
3.10: Electric field distribution for horn and open ended WG antennas..............38
3.11: Effect of frequency on acoustic signal amplitude.............................40
3.12: Effect of material conductivity of acoustic signal amplitude.................41
3.13: Effect of MW pulse width of the wavelenght of acoustic signal................41
3.14: Effect of changing permittivity and conductivity over wire range of values...42
4.1: Effect transducer orientation on image reconstruction..........................46
4.2: Time reversal reconstruction of simulation data................................49
4.3: Reconstruction with different MW pulse widths..................................50
4.4: Reconstruction of samples that are embedded in another conductive medium.......51
4.5: Imaging of a uniform piece of meat.............................................53
4.6: Imaging of a uniform piece meat with copper wire...............................54
4.7: Effect of using highly conductive tissue......................................55
4.8: Imaging of two separate pieces of meat........................................56
5.1: Laser source and camera orientation in setup 1.................................59
5.2: Laser source and camera orientation in setup 2.................................59
5.3: Initial setup with endoscopic camera..........................................60
5.4: LEAP experimental setups with fisheye camera..................................60
5.5: Comparison between images of endoscopic and fisheye cameras...................61
5.6: Flowchart of the 3D reconstruction............................................63
5.7: Comparison between red and green layers.......................................64
5.8: Power law transformation......................................................64
5.9: Laser ring after applying thinning process....................................66


xii
5.10: 3D reconstruction from commercial scanner............................................67
5.11: Artificial damages in the pipes wall................................................68
5.12: 3D reconstruction with endoscopic camera.............................................68
5.13: Surface reconstruction with fisheye camera (setup 2).................................69


1
1. Introduction
Nondestructive evaluation is the science that deals with testing and analyzing the material
properties of objects without damaging them or altering their physical content. NDE is applied
in fields where material failure could result in high possibility of hazard or economic losses.
NDE is widely used in industry to test the quality of new products and inspection of aging
critical infrastructures. In medicine, similar NDE methodologies are used to inspect the
damage that happens to the human body like bone fractures, tumors or to monitor physiological
process that is undergoing inside the body like pregnancy examination using ultrasonic
imaging. NDE depends on the responses of the material to electrical, mechanical and chemical
energy interactions in the testing process. Some of the well-known techniques are, X-rays,
ultrasonography, magnetic resonance imaging, optical microscopy and microwave imaging,
etc. All of those methods should leave no harmful effects on the tested object if the tests are
conducted under proper conditions or sometimes the effects are negligible. In this study, two
nondestructive evaluation methods are developed and discussed for both medical and industrial
applications.
1.1. Breast cancer
Breast cancer is a malignant tumor that starts in a Breast cells, then grows out of control.
It is very rare in males and most of the cases are diagnosed in females [1], Studies show that it
threatens the life of millions of women around the world [2], In 2014, more than 232 thousand
cases of breast cancer were diagnosed in the United States only [2] and studies show that those
patients have the second highest death rates among other cancer patients[3]. Cancer tumors
can be classified according to their spreading nature into two categories. The first type is known
as in situ carcinoma, this type of tumors is confined and does not spread to its neighbor and


2
this is the reason of considering it as a noninvasive cancer. The second type is invasive
carcinoma. In this type cancer cells spread through the breast and other parts of the body. It is
considered as the most dangerous type of cancer cells because it cannot be confined and
continues to spread and destroy all the body systems. As the cancerous cells spread through
the body, the amount of damage increases and controlling the situation becomes harder. As a
result, the patient survival highly depends on the cancer stage when the patient is diagnosed,
i.e., early detection of cancer cells plays an important role in cancer treatment and saving the
patients life.
The human breast can be divided into two main regions according to the type of dominant
tissues as shown in Figure 1.1 [4], The first region is dominant with mammary glands(lobules)
that are responsible for milk production while the second region is dominant with fatty
tissues[5]. In addition to those regions, the breast has large number of milk ducts that transfer
the milk from the gland to the nipple and the whole breast structure is shielded with skin from
outside. The dominant type of breast cancer is originated from ducts and is called ductal
cancer[5]. This cancer type represents 80 percent of invasive cancer cases[6]. Another type of
cancer cell can be originated from the lobules and is called lobular cancer. This type represents
10 percent of the invasive cancer cases[6]. Breast cancer can also originate from other tissues
but they are uncommon.


3
Figure 1.1: Normal human breast
X-ray mammography is widely used for breast cancer screening and is considered as the
standard method for initial diagnostics[3]. X-ray proved to be an effective tool for reducing the
number of cancer fatalities since its application but it suffers from some drawbacks. Studies
show that about 15% of the cases were detected improperly[3] and some of the cases cannot
be detected if the tumor is embedded inside radiographically dense breasts (breast with a high
rate of glandular and connective tissues)[7]. Other cases were reported while there were no
real cancer cells. This is added to the fact that x-ray relies on ionizing radiation in the process
of imaging which increases the probability of creating new cancer cells due to the ionizing
radiation[3]. Another drawback with mammography is the applied pressure to the breast during
the diagnostic process, which is a source of discomfort to the patient.


4
On the other hand, MRI is a capable imaging technique that gives superior images but it
is considered to be an expensive solution for routine screening[8]. For this reason, it is mostly
used for further evaluations in cancer diagnostic. In addition to cost, MRI also suffers from
high false-positives and long imaging time[7]. Recently, Microwave imaging was proposed to
detect cancer tumors. This method proved that it can provides good contrast in between breast
tissues due to the variation of electrical properties of those tissues. But this method suffers
from the low resolution of acquired images due to the long wavelength in the microwave
region[3]. On the contrary, Ultrasound offers good resolution (sub-millimeter) but suffers from
poor contrast in biological tissues due to the small variation of sound speed in breast tissues.
Table 1.1 shows the sound speed in different breast tissues[2].
Table 1.1: Sound speed in breast tissues
Tissue type Sound speed (m/s)
Breast Fat 1457
Breast Gland 1470
Tumor tissue 1509
Skin 1624
1.2. Microwave-induced thermoacoustic imaging
The deficiencies of the current imaging methodologies encouraged the development of
alternative solutions that are efficient and cost effective and this led to the introduction of
thermoacoustic imaging. Since the ninth decade of last century, Microwave Induced
Thermoacoustic Imaging (MITAI) has attracted wide attention in the research community and
some breakthroughs have already been achieved to enhance signal generation and
reconstruction quality [9],
Microwave induced thermoacoustic imaging is a hybrid imaging methodology that
combines microwave imaging with ultrasonography. Hybrid imaging methodologies combine


5
more than one imaging technique to achieve the best qualities of both of them. The result of
this combination in MITAI is a methodology with the high contrast of microwave imaging and
fine resolution of ultrasonography. It is described as either ultrasound enhanced
electromagnetic modality or an ultrasound modality that take the advantage of electromagnetic
wave contrast[9], MITAI signals intensity depends of the amount of heat generation inside the
sample, i.e the images represent the spatially variant electromagnetic loss distribution inside
the imaged object[9].
Pulsed Radiation In Acoustic Energy Out
(I Anatomy
Figure 1.2: Generation of thermoacoustic signals
MITAI signals are generated by sending strong short microwave pulses into the imaged
object as shown in Figure 1.2[10], The microwave pulse absorption inside the electrically lossy
target results in a temporal heat gradient inside the imaged object. The heat gradient excites a
thermo-elastic expansion that generates acoustic waves in the range of ultrasound frequencies.
Then those acoustic waves are used to reconstruct the imaged objects structure using standard
tomography approaches. The frequency of the acoustic waves is inversely proportional to
width of the electromagnetic excitation pulse.


6
Photoacoustic and thermoacoustic imaging share the same physical concept. The naming
depends on the type of employed excitation source that is used for heat generation.
Photoacoustic imaging employs high power lasers while thermoacoustic imaging depends on
microwave pulses to initiate the thermal expansion. With photoacoustic, the imaged object
needs to be scanned by a thin laser beam while in thermoacoustic imaging the whole target can
be scanned at one time if the right setup is available. MITAI is attractive as an imaging solution
for breast cancer due to its low cost, simplicity, high resolution and good contrast. Breast
cancer is one of the potential application for MITAI but it can also be employed in other
nondestructive evaluation processes like detection explosives embedded in biological
tissues[ll][12]. For this application, explosive devices appear as dark objects in the images
due to their small conductivity (0.001 S/m)[12], With this conductivity contrast explosives can
be easily differentiated biological tissues.
1.3. Pipelines internal surface reconstruction
With the current industrial revolution, energy became a crucial factor in the success of
any business. In order to satisfy this demand for energy, gas and oil pipelines networks have
extended over thousands of kilometers over land and under sea. These huge networks require
routine maintenance in order to guaranty the safety and durability of the networks. In order to
guarantee that, many nondestructive evaluation methods are being used to check the safety of
these pipes. Most of the current NDE methods can be classified into two categories, external
and internal methods. External methods are used to conduct the tests from outside the pipe
without the need to take the pipe to an offline state. This can be performed by ultrasonic guided
wave testing or simple visual inspection as shown in Figure 1.3[13][14],


7
In the case of internal inspection methods, passive or active sensors are deployed inside the
pipe. Those methods require the pipes to be offline in order to go inside the pipe and do in
depth analysis of the problem. These tests can be conducted by smart pipeline inspection
gauges (pigs) and laser scanners as shown in Figure 1.4.a and b respectively. Pigs can be
equipped with multiple nondestructive evaluation tools like magnetic flux leakage and
ultrasound sensors for metal pipes health monitoring. Laser scanners are used to detect cracks,
defects and internal surface deformations. Most of the current internal diagnostic devices are
designed for large size pipes with a diameter larger than 6 inches, which leaves the area of
small diameter pipes with no efficient solutions.
(a) (b)
Figure 1.3: External inspection of pipelines
Multiple approaches are available for 3D characterization. Lidar is a laser-based
technology that depends on the time of flight for depth calculation. This technology is robust
and accurate but the equipment has large sizes and is relatively expensive. Another approach
is passive stereovision where two cameras are used to capture the seen and the shift between
the two images is used for depth calculations. This method has been tested but the results were


8
not good enough due to the poor illumination inside the pipe and the distance between the
cameras were very small due to the pipe size.
Thats one smart pig
Energy companies use a special pipeline inspection
gauge, known as a smart pig, that contains electronic
and magnetic sensors to check the interior condition
of pipe walls. If they detect cracks or other problems,
workers can further inspect sections of the pipeline, or
dig it up for repairs or replacement The pig is
launched through traps off the main pipeline and
propelled by the flowing oiL
STAFF GRAPHIC I HCHA&RSHER
(a)
Figure 1.4: Internal inspection of pipelines
1.4. Structured light
Structured light is a method to acquire depth images from the deformations of projected
light patterns by employing triangulation techniques. The depth acquisition is implemented
by using a camera and light projection device as shown in Figure 1.5. A predefined light
pattern with sharp edges is projected on the object and imaged by a camera that is placed on a
well-known baseline. The dimensions of images are converted to their values in the real
plane and the depth of the image is calculated by determining the amount of the shift of lines
in the recorded image. Different light sources and pattern can be employed for depth
reconstruction. The source can be a single laser source that projects single sharp line or a
projector that projects different kind of patterns. The density of the projected patterns
determines the number of acquired 3D points from each frame but at the same time decreases


9
the possibility of detecting the edges correctly. In this study, a laser source that projects a
single colored ring is used as the main projector. The ring pattern is chosen because the
sensor is designed to work in a circular geometry. Both of the projectors and camera are
miniaturized to be suitable to inspect pipelines with diameters smaller than 3 inches. This
approach is chosen because it is simple, cost effective and can be miniaturized easily.
1.5. Thesis scope
This thesis aims to create a high-fidelity numerical model to simulate the electromagnetic
and acoustic interactions in thermoacoustic imaging. The model is designed to optimize the
acoustic signal generation and enhance the electromagnetic wave absorption inside the imaged
objects. The study also focuses on the reconstruction algorithms and implements and tests
different types to compare their performance. For thermoacoustic imaging the goals can briefed
Light stripe
S
prc
Triangulation base
Figure 1.5: Structured light based depth calculation
as bellow


10
Develop an integrated model that simulates the process
Validate the model with experimental results
Study the properties of thermoacoustic signals
Analyze the suitability of the method for different applications
Apply different reconstruction algorithms to reconstruct the data.
The study also develops structured light-based reconstruction algorithms for the
characterization of pipes internal surface reconstruction. It analyzes the suitability and stability
of those reconstruction methods and compare their performance.
1.6. Motivation
The motivation of the study is the need to develop new imaging techniques that are able
to avoid the drawbacks of current imaging methodologies and provide efficient and cost
effective solutions for industrial applications.


2. Theoretical background
Different types of electromagnetic and mechanical waves are used in this study.
Thermoacoustic imaging employs two types of waves in the imaging process. The first type is
the electromagnetic waves that are used in the heating of the imaged sample. The second type
is the acoustic waves that are used to extract the characteristics of the imaged object. While
structured light employs visible light in the inspection process. Another issue is the conversion
between the microwave heating and the acoustic excitation (thermoacoustic effect). This
chapter gives a brief introduction to the wave propagation, scattering, reflection and present
the basics of each imaging system.
2.1. Electromagnetic waves propagation and absorption
Thermoacoustic imaging depends on microwaves as the main source of the heating in
the imaging system. Hence, the propagation of EM waves in materials and in free space plays
important role in modeling the imaging system. In this section, we are giving a brief review of
the mechanism of the EM wave propagation and scattering in materials. This section also
provides a brief introduction to transmission lines and some antenna concepts needed in the
modeling process.
2.1.1. Electromagnetic spectrum
Electromagnetic (EM) waves are transverse electric and magnetic waves where the
electric and magnetic fields are orthogonal to each other and to the direction of propagation.
The EM spectrum expands from zero hertz to infinity and the waves characteristics and their
interaction with materials varies according to their frequency. Figure 2.1 shows different
frequency bands of EM spectrum where each band has a different kind of wave that has its
own special characteristics. For the thermoacoustic signals, we are mainly interested in the


12
microwave region. This region is chosen because the cancerous cells show relatively good
contrast from other breast tissues at this band of EM spectrum. Table 2.1 [15] shows the
electrical permittivity and conductivity of different biological tissues at 2.5 GHz. Another
important factor is the availability of the required efficient sources, amplifiers and other
components in this region of electromagnetic spectrum.
Wavelength (Meters)
10* 10J 10* 0.5x10* 10* 10* 101i
Radio Microwave Infrared ; Visible Ultraviolet X-ray Gamma R;
^x\y\/\AA!mm
10* 10* 1011 101* 10* 10* 10*
Frequency (Hz)
Figure 2.1: Electromagnetic spectrum
Table 2.1: Electric properties of different tissues at 2.5GHz
Tissue name Permittivity Conductivity(S/m)
Blood 56-60 2.5
Bone 12 0.4
Brain (Grey matter) 45 2
Fat (Not infiltrated) 4-5 0.07-0.1
Heart 55 2.3
Kidney (Cortex) 55 2.5
Liver 42 1.8
Lung 20 0.7-0.8
Muscle 50-55 1.8-2.2
Skin (Dry) 38 1.5
Skin (Wet) 43 1.8
Spleen 52 2.2
Tendon 42 1.8
For the case of structured light based internal surface reconstruction the waves are in the visible
light region. Those waves can provide very good resolution for our system due to their short


13
wavelength (400-700nm). The equipment in this region are well developed and the cameras
can provide superior resolutions.
2.1.2. EM waves propagation
The EM wave propagation is governed by the medium electrical properties. The
medium is electrically characterized by three factors and they are, conduction, polarization,
and magnetization. The dominance of any one of those factors specifies the type of the material
if it is a conductor or dielectric or magnetic material. The effect of those factors can be shown
clearly by examining Maxwells equations and their constitutive relations. Maxwells
equations govern the EM wave propagation in any medium. The equations are given by the
following formulas:
_ dB ^
V x E = + M
at
(2.1)
_ - dD
VxH=j + 1F
(2.2)
V D = pv (2-3)
V B = 0 (2.4)
Where E, D, H, B, pi, e,J, M and a are the electric field intensity, electric flux density, magnetic
field intensity, magnetic flux density, magnetic permeability, electric permittivity, electric
current density, magnetic current density and electric conductivity, respectively. Equation (2.1)
is Faradays law and it states that any time varying magnetic field is associated with a spatially
varying electric field. Equation (2.2) is Amperes law and it states that any time varying
electric field is associated with a spatially varying magnetic field. Equation (2.3) is Gauss law
of electricity and it states the net electric flux radiated out of any close surface is equal to the


14
amount of charge enclosed by the surface. Equation (2.4) is Gauss law of magnetism and it
states the net magnetic flux radiated out of any close surface equals to zero. Another auxiliary
equation is called the continuity equation and it is used to define the relationship between the
electric current and charge density and it is given by
d
(2.5)
Maxwells equations are supported by three extra constitutive relations that define the
relationship between the fields components. Those relations are given by
Tc = (2.6)
D = eE = e0£rE (2.7)
B =iiH = iiQ iirH (2.8)
/i0 and /ir are free space electric permittivity and magnetic permeability respectively. £r and
Hr represent the atomic and molecular dipoles of the material and magnetic dipole of the atoms
in the medium[16]. By using the phasor form, the derivatives are exchanged with ja> and
equation (2.1) and (2.2) are written as:
V X E = -juiiH + M (2.9)
VxH =ja)£E +J (210)
By taking the curl of (2.9) and using (2.10) and the constitutive relations we get
VxVxl = joifxV x H = 0)2fX£E (2.11)
Moreover, by using the vector identity, (2.11) can be written as
V2E + a)2[i£E = 0.
(2.12)
This expression is called the electromagnetic wave equation.


15
2.1.3. Boundary conditions
Behavior of field components at any boundary that separate two different mediums is
governed by a group of boundary conditions. The boundary conditions at a charge free surface
are given by
n x (~El £^) = 0 (2.13)
nx(jV1-lT2) = Q (2.14)
n.(Dl-~D2) = 0 (2.15)
= 0 (2.16)
The equations show that the tangential components of E and H are continuous across the
boundary. They also show that the normal components of D and B are continuous across the
boundary, i.e., any incident wave that travels toward a boundary is ether reflected from the
boundary or pass through to the other medium or suffers from both reflection and transmission.
The amount of reflection depends on the wave impedance difference between the two media.
Wave impedance is the ratio between the transverse electric and magnetic components of the
wave in medium. Wave impedance is given by
E
n =-jj, \v\ =
>Qr
(2.17)
The wave transition between two media, medium one and medium two is described in terms
of reflection and transmission coefficients. Reflection coefficient (r) represents the ratio
between the reflected and incident wave and it is given by


16
r =
T}2 cos 9t Tji COS 0j
(2.18)
fh COS 0j + Vj2 cos 9t
?71and ?72 are the wave impedances of the first and second mediums. 0j and 6t are the incident
and transmission angles. Transmission coefficient(T) represents the ratio between the
transmitted and incident wave and it is given by
T =
2r]2 cos 9i
(2.19)
J7X cos 0j + r]2 cos 9t
and the summation of reflection and transmission coefficients equals to 1 in the lossless case.
2.1.4. Attenuation
The wave amplitude is degraded throughout the its propagation into the medium. This
degradation is either due to the spread of the wave energy over a large area or due to the
losses in the medium. The losses can be accounted for in the wave equation. The lossy
version of wave equation is written as
V2£ + yE = 0 ,y = a + j(3 (2.20)
y, a, (3 are the propagation constant, attenuation constant and phase constant, a and (3 are
expressed by
a) is the angular frequency of the applied wave.
(2.21)
(2.22)


17
2.1.5. Near field and far field
The radiation region around an antenna is divided into three main regions illustrated in
Figure 2.2. The first region is the reactive near field. It starts from the region exactly next to
Id3
the antenna and extends to 0.62 I) is the largest dimension of the antenna and X is the
"v A
transmitted wavelength. This region is mainly dominated by the reactive field. The second
region is the radiating near field and it is surrounded by the reactive near field and far field
regions. This region is a transitional stage between the near and far fields and it is dominated
by radiation fields, and the angular field distribution is governed by the distance from the
antenna. If the antenna has a maximum dimension that is not large compared to the wavelength,
this region may not exist. The third region is the far field region where the angular field
distribution is independent of the distance from the antenna. In general, this region starts from
2D2/X to infinity with some exception for some types of antennas.
Far-field (Fraunhofer)
region
Radiating near-field (Fresnel) region
Reactive
near-field region
\
Figure 2.2: Field regions around antenna


18
2.1.6. Rectangular waveguide
The waveguide is described as a structure that is used to transfer EM power from one
point to another. In this study, an open-ended rectangular waveguide is used to transfer the
power from the source to the imaged target. Rectangular waveguides are widely used in the
microwave region of the EM spectrum due to their ability to handle high amounts of power
with high efficiency. Rectangular waveguides consist of four conductive plates. Each two
plates are parallel to each other perpendicular to the other two plates as shown in
Figure 2.3 [17], The wave propagation inside rectangular waveguide is governed by the
reflections from its highly conductive walls. As a result, a specific number of guided modes
are excited between the conductive walls. The number of those modes depends on the
waveguides dimensions and the highest frequency launched inside the waveguide. Transverse
electromagnetic waves cant propagate inside rectangular waveguides[17]. Therefore, the
modes are either transverse electric or magnetic. The waveguide acts like a high pass filter,
and the minimum frequency that can propagate inside it is governed by the following
equation[17]
n,m represent the number of excited modes. For the first dominant mode TE10, the electric
and magnetic fields are given by
(2.23)
(2.24)


19
C1 is an arbitrary constant. y,a and P are the propagation constant, attenuation constant and
phase constant. In computational electromagnetic, waveguides are simply modeled with four
PEC boundaries. At those boundaries, all the tangential electric fields are set to zero.
Figure 2.3: Rectangular waveguide
2.1.7. Microwave heating
Microwave heating has a broad range of applications in industry and household
appliances. It has being widely used in home kitchens and food processing. Recently
microwave heating has had many medical applications such as thermoacoustic imaging tumor
ablation. The microwave heating is generated by focusing a strong microwave beam on a
target. The propagated EM waves suffer from losses inside the targeted object. As a result,
those EM power losses are converted into heat. The amount of resulted heat depends on
multiple factors like material conductivity, permittivity, and density (related to heat transfer).
It also depends on the location of the target from the microwave source. The location affects


20
the intensity and the distribution of electromagnetic power inside the Target. From Maxwell
equations, the instantaneous power loss density is given by the following equation[3]:
power loss density = PL(x,y,z, t) =
o{x,y, z)\E{x,y, z, t)|:
(2.25)
The instantaneous power loss density is integrated over time to calculate the total dissipated
power inside the target. The total dissipated power inside the heated target is given by[3]
total dissipated power = Ptl I PL(x,y,z,t) dt (2.26)
r is the total simulation time and dt is the simulation time step.
2.2. Acoustic propagation theory
Soundwaves are one of the well-known types of waves due to their wide range of
applications. In addition to the fact that it is an important communication method used by
humans and animals, it also has many other medical and industrial applications. It is used for
nondestructive evaluation of materials, medical imaging and as a destructive tool (high-
intensity focused ultrasound) for therapeutic purposes.
Acoustic waves are mechanical longitudinal pressure waves that require the
propagation medium to have both compressibility and inertia. Due to their mechanical nature,
acoustic waves cannot propagate in vacuum or free space. The speed of the sound waves
mainly depends on the medium mechanical properties. Their spectrum expands over wide
range of frequencies and can be divided into three main [18],
Infrasonic region: less than 20 Hz
Audio frequency region: from 20 Hz to 20 kHz


21
Ultrasonic region: higher than 20 kHz
Human beings can only recognize frequencies in the audio region while some other animals
can hear further in the ultrasonic region. This study is mainly interested in the ultrasound waves
due to their short wavelength. The propagation of those waves manly depends on:
Particle displacement ((): is the displacement of the mediums particles around their mean
during the wave propagation
Particle velocity (it): is the velocity of particles inside the medium. It consists of two
components and they are non- oscillatory net flow velocity (u0) and the oscillatory velocity
(ua). The oscillatory velocity refers to the derivative of the particles displacement ua =
The total particle velocity is the summation of both of them, i.e. u = u0 + ua.
Acoustic pressure (p): represents the amount of compression of media. It also consists of an
ambient pressure and an oscillatory pressure due to the wave propagation. The total pressure
is the summation of them (p = p0 + pa).
Material density (p): represents the amount of mass per unit volume. The wave propagation
through the medium leads to a fluctuation in the ambient mass due to the particles
displacement. The total density is written as p = p0 + pa.
2.2.1. Wave equation
Propagation of acoustic wave in any medium is governed by the acoustic wave
equation. For heterogeneous mediums, it is given by
(2.27)


22
Wherep is the acoustic pressure and co is the sound speed in the medium. For a homogenous
medium where the ambient density has a fixed value everywhere, the equation is reduced to
1 d2pa
Cq dt2
- V2pa = 0
(2.28)
2.2.2. Acoustic impedance
The acoustic impedance represents the amount of opposition of the medium to the
acoustic pressure flow. There are two types of acoustic impedances:
Specific acoustic impedance (zn): is the ration between the acoustic pressure and the particle
velocity in a specific direction
*=o (229>
n is a unit vector in the direction of interest.
Characteristic acoustic impedance (Z): is a constant that represents the relation between the
pressure and particle velocity in a plane wave[18],
Z = p0c0 (2.30)
2.2.3. Acoustic wave propagation in inhomogeneous media
The propagation of acoustic waves in any media is encountered by many restrictions.
The first one is the attenuation of the signal due to spatial spread or due to the abortion inside
the propagation medium. The second factor is the reflection of the acoustic waves from
boundaries between different materials due to the difference in the characteristic impedance.
The difference of sound speed does not only cause the wave to reflect, a portion of the wave is
also transmitted through the boundary and some are scattered by the object if the objects size


23
is small with respect to the wavelength. I.e. the size of the object and the amount of variation
in the wave speed and incidence angle results in different scattering effects.
2.2.4. Acoustic boundary conditions
The wave incidence on a boundary is classified according to the angle of incidence into
two types, normal and oblique incidence. By starting with normal incidence, when a wave
reaches a boundary, it is either reflected in the opposite direction or transmitted through the
boundary wall. The reflection and transmission through boundary are better explained in the
terms of reflection and transmission coefficients. The reflection coefficient (R) is the ratio
between reflected and incident waves. Transmission coefficient (T) represents the ratio
between transmitted and incident waves. They are written as
Pu Pr> Pt represent the pressure of incident, reflected and transmitted waves. Boundary
condition at any acoustic boundary are given by
ut, ur> ut are the particle velocities of incident, reflected and transmitted waves. Dividing the
equations in 22 yields to:
Pr Pt
R=, T =
Pi Pi
(2.31)
Pi+Pr= Pt. Ui+Ur = Ut
(2.32)
Pi+Pr = Pt
U-i + Ur Uf
(2.33)
In plane-waves p = u and this yields to
Pr
ur
Pt _
PlTl, P2C2
ut
t
(2.34)


24
Using the relation 1 + R = T (pressure continuity), the transmission and reflection
coefficients become:
R =
T =
P2C2 Plcl ^2
P2c2+Plcl Z2+Z1
2p2c2 2Z2
P2c2+Plc2 Z2+Z,
For oblique incident, the angle of incident is taken into account while all the boundary
conditions still hold and the reflection and transmission coefficients become:
(2.35)
(2.36)
_ P2C2 cos(0j) -Picx cos(0t) Z2 cosjdi) Zx
1 p2C2 cos(0j) + p1c1 cos(0t) Z2 cos(0j) + Zx
TWO =
2p2c2 cos(0j)
2Z2 cos(0j)
p2c2 cos(0j) + p1c2 cos(0t) Z2 cos(0j) + Zx cos(0t)
(2.37)
(2.38)
2.3. Thermoacoustic equation
The thermoacoustic wave generation is a mechanical phenomenon that results from
thermal expansion of mediums. Those thermal expansions result in a series of mechanical
waves (acoustic waves) that are govern by the wave equation. Thermoacoustic signal
generation and propagation are described by thermoacoustic equation and it is given by[9]:
V2P(r, t)
1 d2P(r,t) p d2T(r,t)
(2.39)
c2 dt2 kc2 dt2
c,p,k are the sound speed, volume expansion coefficient and the isothermal compressibility
of the medium. P(r, t) is the instantaneous pressure at location r. vs is the sound speed in the
medium. T(r,t) is the temperature of the medium. The left side of the equation represents the
acoustic wave equation while the right side of the equation represents the photoacoustic source.
The photoacoustic source depends on the second derivative of time varying temperature,
volume expansion coefficient and heat capacity. The heat capacity and volume expansion


25
coefficient are consonants that depend on mediums properties and the only varying term in
the source is the acceleration of temperature. When thermal confinement condition is met, the
source part becomes equals to[9]
dT(r, t)
pCv^ = H(r, t),
(2.40)
p, Cv are the material density and specific heat capacity of the medium. H(r,t) is the heating
function that represents the amount of dissipated energy per unit volume and time by the
excitation pulse and it is also called the specific absorption rate (SAR) of the material. Thermal
confinement condition assumes that the excitation pulse is short enough so that the acoustic
wave is excited before the occurrence of any significant heat conduction. This condition
requires that the electromagnetic pulse to has an excitation period that satisfies the following
condition
r <
4 a
(2.41)
th
Where dc and ath are characteristic dimension (m) and the thermal diffusivity of the heated
region. The heating function can be written as a two separate terms
H(r, t) = A(r)/(t) (2.42)
Where A(r) is the absorbed energy density and I(t) is the temporal envelope of the excitation
pulse. As mentioned in section 2.1.7, the absorbed energy density is given by
A(r) =
a(r)\E(r)y
2p(r)
(2.43)
Where electric field inside the target. By combing those relations, the thermoacoustic equation is
rewritten as


26
1 d2P(r,t) /? dH(r,t)
(2.44)
Where Cp = pckCv is the specific heat capacity of the medium under constant pressure.
2.4. Light triangulation
Triangulation is the process of calculating the position of a point in a 3D space by using
the angles to it from a specific baseline. Figure 2.4 shows a simple 2D triangulation and the
imaging system orientation for depth calculation. Here we have a light source that sends its
signals at a certain angle toward an object. Another observation point is placed at a distance d
from the source to monitor the location of the projected light spot. A straight line is drawn
from the observation point to the location of the incidence to calculate the angle (3. After
calculating /3, y is calculated directly (y + (3 + a = 180). By knowing all the angles and the
base length the depth calculation becomes handy process.
Figure 2.4: Optical triangulation
By using the normal to d we get


27
dsin(a) = bsin(y)
And by using the relation sin(y) = sin(n + (3)
b sin(a)
sin(n + (3)
By knowing d, the point coordinate is given by
x = d. cos (3
h = d. sin (3.
(2.45)
(2.46)
(2.47)


28
3. Numerical modeling of thermoacoustic signal generation
Thermoacoustic imaging modeling consists of a forward model that simulates the
signals generation and a backward model that is responsible for the reconstruction of final
images. An overview the generation and reconstruction modeling is explained in Figure 3.1
where the forward model is implemented by COMSOL Multiphysics software and the
backward model is implemented by using K-wave toolbox. This chapter discusses the
modeling process and signals characteristics in the forward model. The model simulates the
microwave propagation, electromagnetic absorption and heat generation, and the
thermoacoustic signal generation and propagation. The model also studies the effect of
different experimental parameters on signal characteristics and investigates the optimal
experimental conditions to generate the strongest acoustic signals.
Figure 3.1: Modeling overview for MITAI
3.1. Experimental Imaging model
Our experiment employs the standard experimental setup of thermoacoustic signals
generation with some extra modifications. The imaging system consists of the following parts
as shown in Figure 3.2


29
Lucas Epsco PG5KB pulsed microwave source with a maximum output voltage of 5kV.
The device is responsible for providing the required excitation signal to initiate the
thermoacoustic effect. It is tuned to sends 4.5Kv pulses with a carrier frequency of
2.45GHz and pulse width varies from 0.3 to 3 microseconds.
A coupling device to transfer the signal of the microwave source to the imaged target.
In our experiment, this is represented by an open-ended waveguide due to its narrow
beam (comparison with horn antenna is provided later).
An Olympus acoustic transducer with a center frequency of 2.45 MHz is used to detect
the thermoacoustic signals.
Peripheral devices for synchronization, amplification, filtering and post-processing of
the received acoustic signal.
Figure 3.2: Experimental setup of MITAI
3.2. COMSOL
COMSOL Multiphysics is a commercial simulation software that is used to solve
coupled phenomena problems. It is a Finite Element Method (FEM) based solver in time and
frequency domain. The software is divided into several types of modules where each module


30
is specified for solving a certain phenomenon type with the ability to couple those modules. In
this study, we mainly use two types of modules
RF Module: this module is responsible for simulating the electromagnetic propagation
through the target and calculating the amount and distribution of the EM losses inside
the imaged object.
Acoustic module: this module is responsible for simulating the generation of
thermoacoustic waves inside the target and its propagation from the target to the
detectors.
In addition to that, a Multiphysics link is created to couple the EM losses results from the RF
module to the acoustic module
3.3. Electromagnetic waves simulation
3.3.1. Simulation geometry
The model employs both 2D and a 3D geometries that simulate the same experimental
setup. The main components in the model are
Open Ended waveguide: WR-340, D band waveguide with dimensions of 86.36 x
43.18mm. It is modeled as a hollow box that filled with air and shielded with perfect
electric conductor boundary condition. The waveguide is excited with a wave port that
has an electric field parallel to the short side of the waveguide.
Acrylic tanks: a cuboid box of acrylic with a wall thickness of 3mm is placed at the end
of the waveguide and filled with safflower oil. The bottom part of the acrylic tank is
shielded with perfect electric conductor to be similar to the experimental setup.


31
Target: the target is placed directly above the acrylic layer and the end of the
waveguide.
All The simulation domain except the bottom of the tank (shielded with PEC) is
terminated with a perfectly matched layer to reduce the size of simulation domain.
Figure 5 shows the 3D simulation geometry of the model. To enhance the computation speed
in some cases the 3D model is reduced to 2D model by taking a plane at 7=0 as shown in
Figure 3.4.
z
Wx
Figure 3.3: 3D geometry of the EM simulation


32
Figure 3.4: Cross section plane through the 3D geometry
3.3.2. Simulation parameters
The simulation employs tetrahedral meshes in case of 3D simulations and triangular
meshes in case of 2D simulations to achieve an accurate characterization of the actual geometry
shape. Two different mesh sizes are used in the EM simulations. The general simulation
geometry uses variable mesh size with maximum element size equals to 1 cm since the
minimum wavelength equals to 12 cm. The target is meshed with different mesh size to achieve
more precise result for the electric field loss distribution inside it. It is meshed with a resolution
similar to that of the acoustic simulation (will be explained later) with maximum element size
of 0.1mm. 3D meshing of the geometry can be seen in Figure 3.5. The simulated carrier


33
frequency equals to 2.45 GHz and target electrical properties varies according to the simulation
requirements.
(3
x
Figure 3.5: Meshing of 3D geometry
3.3.3. Acoustic propagation model
The EM loss distribution inside the target is forwarded to the acoustic module to act as
a monopole acoustic source. The simulation geometry is reduced by taking only a finite area
around the target and use suitable boundary conditions to terminate the simulation domain. An
example can be seen in Figure 3.6 where the simulation domain is terminated with absorbing
boundary condition (ABC).


34
160 165 170 175 180 185 1
Figure 3.6: Acoustic simulation region
3.3.4. Simulation parameters
Acoustic simulation geometry is meshed with different mesh than that used in the EM
simulation due to the difference in the simulated wavelength. The geometry is meshed with
triangular mesh structures as showed in Figure 3.7. The maximum frequency that we are
interested in is 2 MHz Therefore the maximum element size is decide to be 0.1mm (7.5 samples
per wavelength).
3.4. Model validation
The described model of thermoacoustic signal generation is validated by comparing the
simulation results with experimental work by Mashal et.al [19] and simulation results by Deng
and Golkowski [3], The model simulates the generation of thermoacoustic signals by a circular
target with a radius of 6mm. The target is immersed in mineral oil and an acoustic transducer
is aligned with the center line that pass through the target center. The imaged target material


35
consists of ethylene glycol with different concentrations of air micro bubbles. Different
concentrations of microbubbles produce different conductivities and sound speeds inside the
imaged object. Table 3.1 gives a detailed description about the change in materials properties
with bubbles concentration.
Figure 3.7: Meshing of acoustic region
Table 3.1: Variation of ethylene glycol properties with microbubbles concentration
Solution number Bubbles concentration Average £r Effective conductivity (S/m) Sound speed
Solution 1 0% 14.03 2.32 1660.7
Solution 2 20% 9.66 1.33 1729.8
Solution 3 30% 8.34 1.11 1805.6
Solution 4 35% 7.38 0.95 1922.6
Solution 5 40% 6.86 0.84 2049.4
Solution 6 ??% NA NA Na


36
From Table 1.1, it can be seen that the material conductivity decreases with the increase of the
microbubbles concentration. A comparison between the model and the experimental results is
shown in Figure 3.8.a and b respectively. The model shows good agreement with the
experimental results. The amplitude of the signal decreases with the increase of the
microbubbles concentration (decrease of conductivity) and this normal due to the decrease of
the electromagnetic losses inside the material. It also shows that the signal period is decreased
with the increase of microbubbles concentration due to the increase of sound speed.
Variation of MITAI signal with bubbles concentration
Figure 3.8: Comparison between simulation and experimental results
Figure 3.9 shows a comparison to the simulation results of [3], The signal amplitude and time
period of the signal are changing in the same with a noticeable difference in the speed of signal
decay. This difference could be related to the to the difference in the computational method
and meshing techniques since their simulation employs FDTD method with square meshes.


37
Figure 3.9: Comparison with other simulation work
3.5. Open ended waveguide vs horn antenna
MITAI system requires an antenna that can handle high amount of power with high
efficiency because the system needs to transfer very strong pulses (larger than 5KV) to the
imaged target. The antenna should also have relatively narrow near field beam in order to
concentrate the transmitted power inside the imaged object. Two options are available to for
the experiment. The first option is a standard 20db horn antenna. The second option is WR340
open ended waveguide. A numerical model is created for the antennas to investigate the the
field distribution and intensity at the vicinity of the antennas. The geometry of horn antenna
and open ended waveguide can be shown in Figure 3.10.a and b respectively. In both cases,
the air is used as propagation medium and the simulation domain is terminated with a perfectly
matched layer. The models are simulated with a frequency equal to 2.45 GHz and a variable
mesh size with maximum element size equals to 10mm. Electric field intensity distribution at
a distance equals to 3 mm from the antenna end is shown in Figure 3.10.C and d. The results
show that the open ended waveguide has much higher field intensity in the near field region


38
while the horn antenna shows more homogenous distribution of the field but with lower
intensity. This is due to the distribution of the field over larger area in the case of horn antenna.
The simulation also shows that the reflection coefficient of the open ended waveguide is much
higher than that of the horn antenna. Under the same testing conditions, Sll=-12db for the
open ended waveguide vs -22db for the horn antenna.
Figure 3.10: Electric field distribution for horn and open ended WG antennas


39
3.6. Case studies
Different experimental parameters can affect thermoacoustic signal intensity and
characteristics. In this section, different scenarios are studied to check the effect of excitation
signal and imaged object properties on the signal characteristics.
3.6.1. Effect of frequency
This study shows the effect of changing the frequency on the amplitude of the generated
thermoacoustic signal. The results are shown in Figure 3.11. It shows that changing the
frequency over a small range has negligible effect on the generated thermoacoustic signal. This
study is conducted over limited frequency range due to the waveguide bandwidth restrictions.
Increasing frequency could have a noticeable effect within high range as was reported by [8],
3.6.2. Effect of material conductivity
In this study, the change of mechanical properties of the material in Table 3.1 is
neglected and only the conductivity is changed. The results are shown in Figure 3.12. It shows
that increasing the conductivity within a certain range increases the amplitude of the generated
thermoacoustic signals. This effect is expected and it is a result of the increase of the dissipated
losses by electromagnetic waves according to Equation (2.25). However, increasing the
conductivity above certain threshold limits the penetration ability of EM waves, which will be
shown later.
3.6.3. Effect of microwave pulse width
Different microwave pulses were sent with different pulse widths and the simulation
environment is kept the same. The gaussian pulse standard deviation is changed from 0.3 to
0.9 micro-seconds. The results are shown in Figure 3.13. It shows that increasing the pulse
period directly increases the wavelength of the generated thermoacoustic signal. This means


40
that decreasing the microwave pulse width directly enhances the resolution of the
thermoacoustic images.
3.6.4. Optimal target analysis
In this study, a rectangular target with a dimension of 25 by 10mm is used as the new
target. The permittivity and conductivity are changed over wide range of values as shown in
Figure 3.14.a and b. The results represent the integration of power loss inside the target for
each permittivity and conductivity. The power loss inside the sample increases with increasing
the conductivity then starts to drop again. The increase is a results of the increase of losses in
the medium according to Equation (2.25) but after a certain threshold the EM wave lose its
ability to propagate through the imaged obj ect and the amount of delivered energy to the target
drops again. The optimal conductivity and permittivity is not constant because it depends on
the target size and shape.
Figure 3.11: Effect of frequency on acoustic signal amplitude


41
Time(/zs)
Figure 3.12: Effect of material conductivity of acoustic signal amplitude
Time(/iS)
Figure 3.13: Effect of MW pulse width of the wavelenght of acoustic signal


Absorbed energy in the target Absorbed energy in the target
42
101 10 101 102
Conductivity
(a)
10'1 10 101 102
Conductivity
(b)
Figure 3.14: Effect of changing permittivity and conductivity over wire range of values


43
4. Image reconstruction
Image reconstruction is considered as one of the challenging tasks in thermoacoustic
imaging. The data requires high amount of post processing in order to get the required images.
Many algorithms are proposed to reconstruct the data and their performance varies depending
on the target material properties and acquisition geometry. The homogeneity of sound speed
simplifies the reconstruction problem because it eliminates the need for any extra sound speed
estimation tools while the electric properties are the main engine behind the image contrast.
For the acquisition system, the orientation of the transducers around the target and their size
are crucial factors in determining the resolution of the imaging system. This chapter discuss
those challenges and their effect on the acquired images.
4.1. K-Wave
K-wave is an open source Matlab toolbox that is built for the simulations of acoustic
waves[20]. This toolbox solves the acoustic wave equation by employing a pseudo-spectral
time domain solver. It has special modules for the simulation and reconstruction of
photoacoustic imaging. It implements multiple reconstruction methods for different sensor
geometries in 2D and 3D. As thermoacoustic and photoacoustic imaging follow the same
concept, we are adopting this toolbox for the reconstruction of our simulated and experimental
data. This toolbox is adopted instead of doing the reconstruction by using COMSOL due the
following reasons. The first reason is avoiding the inversion crime of using the same solver
configurations for the forward and inverse simulations. The second is the increase of
computation speed due to the use of pseudo spectral time domain (PSTD) since it requires only
two samples per wavelength. Finally, the toolboxes are flexible and their functions can be
easily modified to suit the experimental requirements.


44
4.2. Reconstruction challenges
As mentioned earlier, reconstruction of MITAI is not an easy task and there are many
factors that need to be taken into consideration during the reconstruction process. These factors
vary according to the reconstruction algorithm but there are some common problems that are
shared between most them.
4.2.1. Homogeneity of sound speed
The homogeneity of sound speed in the medium plays an important role in simplifying
the reconstruction problem. The reconstruction of the image of any object requires that the
solution should vanish inside the space after a finite time T in order to have a known initial
value (zero initial values) to start the reconstruction process. This means that all the acoustic
waves should leave the imaged target after time T. This kind of mediums is known as non-
trapping medium because it allows all the waves to leave after a certain time and no wave is
trapped inside the medium. The inhomogeneity of sound speed highly effects the
reconstruction of the object and the estimation sound speed becomes a necessity in some cases
when there are strong abrupt changes in sound the speed.
4.2.2. Estimation of sound speed
Some cases require estimation of sound speed if there is large variation in the medium
sound speed. Some algorithms are proposed to estimate it from the acquired thermoacoustic
signals but most of them are instable or only works under strict conditions [9], The current best
method to estimate sound is to use pulse eco mode data to enhance the thermoacoustic images
in the case of acoustically inhomogeneous mediums.


45
4.2.3. Size of transducer
The size of transducer is required to be as small as possible because it controls the
resolution of the reconstructed image. Most of the reconstruction algorithms assume point like
Omni directional transducers but real transducers are different. The real transducer has finite
size, and the recorded signal represents an integration of orthogonal components of the incident
waves on the transducer surface and it is given by
pd = jS.dA (4.1)
Where S, dA are the incident signal on the transducer and the surface unit area, i.e. the
transducers signals represent the sum of all the signal that are crossing a specific area. This
limits the resolution of the acquired images and makes it dependent on the transducer size and
properties. The effect can reduced by either using a special transducers that have thin and long
structure or de-convolute the shape and size of the transducer with the reconstructed image
[21].
4.2.4. Effect of transducer orientation
Ideal reconstruction of thermoacoustic signals requires that the target is surrounded by
360 degrees of transducers [9], This is an expensive approach and difficult to implement in
practical situations. In case of homogenous mediums, the rule states that in order to completely
reconstruct a point, then any line that passes through that point should intersect with the
transducers plane at least one time[9]. This rule becomes more complicated in the case of
inhomogeneous mediums because the waves are deflected due to medium inhomogeneity.
Figure 4.1 shows the reconstruction results with different transducers orientations. The
simulated geometry is shown in a while b and c show the reconstruction results with a line and
L shape arrays. The results show that the edges that are normal to the transducers plane are


46
reconstructed clearly while the other edges are blurred and this is the reason behind the
enhancements in c.
Figure 4.1: Effect transducer orientation on image reconstruction
4.3. Image reconstruction with time reversal
Current reconstruction methods neglect the heterogeneities in electric field and sound
speed and assume that both of them are distributed uniformly inside the imaged object.
Different methods are proposed for the reconstructions like filtered back projection,
Eigenfunction expansion method, and time reversal methods. This project implements the time
reversal due to its robustness, flexibility[22],
Time reversal is a robust reconstruction method that exploits the reciprocity property
of the wave equation. According to Huygens principles, for any initial source that has a
bounded support, there is a finite time for the wave to leave the domain[22]. According to this
theory, a solution P vanishes inside the domain after a finite amount of time T. After time T, a
zero initial value can be imposed to the domain and the wave can be retransmitted to the


47
domain in reversed timing sequence to reconstruct the source[22]. Below are the required
formulations for time reversal. By using greens functions and by assuming that the source is
a Dirac delta function, equation (2.28) can be solved as bellow[23]
p(r)= JJJ G(r,r') H(r',t')dr' P \r-r'\ (4.2)
H(j',t') = I0A(r') (4.3)
J?(t') = <5(t') (4.4)
G,A,rj ,/0are the medium Greens function, microwave loss distribution, temporal profile of
the microwave pulse and amplitude of microwave pulse, respectively. By exploiting the
reciprocity of Greens functions, the solution can be written as a convolution between the
Greens function of the medium and the wave source.
p(f) = G(f, f') pa(f) (4.5)
BI o
Pa 00 = ---
(4.6)
Where pa is the induced pressure from A that is enclosed in the sphere |f f'|. As |f f'|
go to zero, pa can be given by [23]
ps(f')df
Ps(f)
Bln ^
Lp
(4.7)
(4.8)


48
The goal of the reconstruction is to get the function A from the pressure p, The retransmission
of the wave again into the media with the same received pressure and boundary conditions
results in the reconstruction of the initial source. I.e., the detected pressure is reconvoluted with
the medium.
pT(f) = G(r,r') 0 p5(f) 0 G(r',r) (4.9)
G(r, r') 0 G{f', f) is the self-correlation function of the medium.
4.4. Simulation study of image reconstruction with time reversal
Three biological samples with different shapes are placed in safflower oil to study
MITAI system as shown in Figure 4.2.a. The samples are a circle with radius of 1mm, square
with side length of 2mm, and a rectangle with dimensions of 1.5 by 3 mm. The forward model
employs the same experimental setup mentioned in chapter 3. The model is excited with a
pulse that has a carrier frequency of 2.45 GHz and a gaussian shape with a standard deviation
of 40 nanoseconds. All the samples are set have the same electrical and acoustical properties
as explained in Table 4.1. The acoustic transducers are represented by an array of point like
transducers that are placed at the top of the simulation domain. Figure 4.2.b shows a semi
uniform electromagnetic loss distribution inside the samples which indicates that the samples
regions can be reconstructed evenly. The same forward simulation model is recreated by using
K-wave toolbox and the signals are retransmitted in a reverse time sequence. Time reversal
results are shown in Figure 4.2.c and a thresholded version of it is shown in Figure 4.2.d. The
positions and sizes of the objects are reconstructed correctly. The case is the same for the edges
that are normal to the transducer plane while we can see that the horizontal edges are blurred
due to the orientation of the detection plane.


49
Table 4.1: Properties of simulated geometry
Medium property Value
Electrical conductivity 0.4
Relative permittivity 9
Relative permeability 1
Sound speed 1537
Material density 1041
Heat capacity 3510
Coefficient of thermal expansion 3e-4
(a)
(b)
1.5
1
0.5
0
-0.5
-1
-2
-2.5
0.5
0
(c)
(d)
Figure 4.2: Time reversal reconstruction of simulation data


50
4.4.1. Effect of MW pulse width
The simulation is repeated with different MW pulse widths with a standard deviation
of 0.6, 0.8, 1.2 and 4 microseconds. The results are shown in Figure 4.3. a, b, c and d
consequently. The results shows that the objects edges are blurred with increasing the pulse
width due to the decrease of the excited acoustic waves frequency.
0.5
Figure 4.3: Reconstruction with different MW pulse widths


51
4.4.2. Biological samples embedded in a conductive medium
In this case the same sample shapes were imbedded inside another biological tissue
that has lower conductivity and different sound speed. The experiment is conducted with
same conditions as in 4.4.1 and the materials properties are given in Table 4.2. The results
are shown in Figure 4.4. The embedded biological samples can be easily identified from
other biological tissues in the background, but they are blurred due to the variations in sound
speed in the medium.


(c)
(d)
Figure 4.4: Reconstruction of samples that are embedded in another conductive medium


52
Table 4.2: Properties simulated materials
Medium property Material 1 Material 2
Electrical conductivity 0.4 0.749
Relative permittivity 9 5
Relative permeability 1 1
Sound speed 1537 1580
Material density 1041 1041
Heat capacity 3510 3510
Coefficient of thermal expansion 3e-4 3e-4
4.5. Reconstruction of experimental data
The same time reversal algorithm is used for the reconstruction of experimental data.
A geometry similar to the experimental one is created in K-wave with PML boundary
conditions. The simulation sound speed is set to be equal to the average sound speed in
safflower oil and the imaged sample and the recorded data is interpolated to match the time
step of the simulation. The measured transducer signal is filtered by median filter with order
of 20 to reduce the noise effect on the reconstructed signal.
4.5.1. Uniform piece of meat
A uniform piece of pork meat is used as test sample target. This meat type is chosen
due to its availability and relatively good conductivity. Figure 4.5.a, b, c, d show the imaged
meat sample, ultrasound images, and thermoacoustic images before and after time reversal
respectively. The results show that the meat boundaries are reconstructed correctly when it is
compared to the ultrasound images.


53
Figure 4.5: Imaging of a uniform piece of meat
4.5.2. Rectangular biological tissue with copper wire
In this test, a 4mm diameter copper wire shown in Figure 4.6.a is inserted into the
meat sample as shown in Figure 4.6.b and the imaging process is repeated again. The
reconstructed thermoacoustic image in Figure 4.6.d and e show that the upper boundary of
meat is reconstructed correctly. The copper wire effect can be seen as a dark spot in the
thermoacoustic image. Moreover, the figure shows that the lateral details are blurred due to
the orientation of the transducers.


Depth (unitless)
54
-20 -15 -10 -5 0 5 10 15 20
y-position [mm]
(e)
Figure 4.6: Imaging of a uniform piece meat with copper wire


55
4.5.3. The effect of inserting highly conductive tissue
This experiment shows the effect of inserting highly conductive (electrically) tissue
into the piece of meat. The new tissue is inserted as shown in Figure 4.7.a. The results are
shown in b, c, d. The results show that the highly conductive tissue produces no acoustic signals
because most of the EM are reflect and they are not able to penetrate the tissue.
Figure 4.7: Effect of using highly conductive tissue


56
4.5.4. Two separate pieces of meat
In this experiment we are using two separate pieces of the of the same meat type as
shown in Figure 4.8.a. The results are shown in c, d, e in the same sequence as before. We can
see that we have good separation between the two pieces but the side edges are still blurred
due to the orientation of the transducer plane.
Figure 4.8: Imaging of two separate pieces of meat
In all our experimental images, a wavy pattern appears inside the imaged samples as
shown in Figure 4.8.d. This phenomenon is under investigation and the early indications
refers to electrical standing waves inside the meat samples due to the mismatching with its


57
boundary. More work is being done to investigate this phenomenon and enhance the
electrical matching between the sample and it background.


58
5. Pipe internal surface reconstruction
Pipe internal surface reconstruction employs the structured light concept with a
modified triangulation for circular geometries. This chapter presents a brief introduction to the
laser scanner setup and develops 3D reconstruction algorithm for defects characterization.
5.1. Scanner setup
Two scanners are developed with each one has different camera and laser source
orientation. In the first setup, a laser source that projects a thin laser ring is aligned in parallel
with the pipe main axes as shown in Figure 5.1. A small camera is placed beside the laser
source to record the laser ring deformations along the pipes internal wall. The second setup
employs the same concept but the camera and the projector are placed in opposite directions
as shown in Figure 5.2. A preliminary setup shown in Figure 5.3 employs a simple endoscopic
camera due to its small diameter. The experiments on this cameras have proved that this type
of cameras cant be employed due to its limited view angle that cause large shadow area in the
acquired images. Therefore, the endoscopic camera is changed with a fisheye 180-degree
camera in order to capture the whole scene as shown in Figure 5.4. The figure shows a two
3D printed holders where a and b represents a direct implementation of setup one and two
respectively. Figure 5.5.a and b shows a comparison between the images that are acquired by
endoscopic and fisheye camera respectively. With the fisheye camera, we are able to minimize
the amount of the shadow that was caused by the setup.


59
Figure 5.1: Laser source and camera orientation in setup 1
Figure 5.2: Laser source and camera orientation in setup 2


60
Figure 5.3: Initial setup with endoscopic camera
Figure 5.4: LEAP experimental setups with fisheye camera


61
Figure 5.5: Comparison between images of endoscopic and fisheye cameras


62
5.2. Extraction of 3D information
The main task in this process is the segmentation of the laser ring from the recorded
camera frames. Figure 5.6 shows a flowchart of the main steps in the segmentation and
extraction of camera frames. The algorithm can be briefied by the following steps:
Extract the best layer of the RGB frame
One layer is chosen according to the color of the projected ring color and the color of the
pipes internal walls. For example, if a white pipe and red laser ring is used, the experimental
tests showed that green layer has better contrast and less amount of noise than the other two
layers. Figure 5.7 shows a comparison between the red and green layers of a single frame.
Image de-noising
In some cases, the recorded camera images contain high amount of noise due the low
amount of illumination inside the pipe. In order to reduce the noise, a Gaussian low pass filter
is applied to the image to reduce the high frequency components.
Power law transformation
This transformation is used to enhance the contrast of the denoised image. It increases
the intensity of high values in the image and separate them from low values. The transformation
is applied according to the following formula:
impi = imL (5.1)
Where im and impl are the input and output images. L is the value of the power that is used to
transform the image. Figure 5.8.a[24] explains the effect of using different values of L on the
image values. In our case, we are using positive value for L to increase the contrast of the large
values in the image. The result of using power law transformation is shown in Figure 5.8.b.


63
The result shows that most of the low values are suppressed and the ring has better contrast
from the background.
Figure 5.6: Flowchart of the 3D reconstruction


64
(a) (b)
Figure 5.7: Comparison between red and green layers
50
100
150
200
250
300
350
450
/-/4 L/2 M/4 l. I 100 200 300 400 500 600
Input pray level, r
(a) (b)
Figure 5.8: Power law transformation
Where im and impi are the input and output images. L is the value of the power that is used to
transform the image. Figure 5.8.a[24] explains the effect of using different values of L on the
image values. In our case, we are using positive value for L to increase the contrast of the large
values in the image. The result of using power law transformation is shown in Figure 5.8.b.


65
The result shows that most of the low values are suppressed and there is an improvement in
the contrast of the final image.
Thresholding
Thresholding is used to divide a certain data set into two different regions according to
their intensity. Thresholding is applied to separate large values of the ring region from other
image regions. The value of the threshold is decided by doing a quick analysis to the first frame
of the recorded video. In this process the greyscale image is converted to a binary image where
the high values are set to one and the low values are set to zero.
Line extraction
As explained before the laser ring has a finite thickness, therefore, it doesnt appear as
single pixel line in the image. The options to introduce a single pixel line from the ring are
either looking for the maxima of the line or finding the middle point in the line. As finding the
maxima is highly effected by the light distortions, we opted to go with finding the middle point
of the line. A Morphological line thinning is used to find the center of the laser ring and the
results can be shown in Figure 5.9. In some cases, the thinning process results in high amount
of spurs. Therefore, in order to guaranty a clean and smooth ring reconstruction, a spurs
cleaning process is implemented to remove any unwanted spurs.
3D point cloud registration
After extracting the locations of the ring pixels in the image, those locations are mapped
to their real locations in the real plane by applying the triangulation process. Moreover, by
assuming a constant scanning speed, the frames are positioned in the cloud as equally spaced
3D rings along the scanning line. The spacing between the rings is calculated according to the
speed of moving the camera setup.


66
i
\
\,
Figure 5.9: Laser ring after applying thinning process
5.3. Experimental results
The proposed algorithm were applied to different experimental setups. At the beginning
the algorithm were applied to a part of laser scanning video of large parameter pipe. The video
is available from maverickinspection.com[25]. The results are shown in Figure 5.10 and the
algorithm was able to characterize the surface correctly. After that the algorithm were applied
to the scanning videos from our LEAP sensors that were shown before. Currently the scanners
are dealing with artificial pipe damages (holes and metal screws) that were created in the
laboratory as shown Figure 5.11. The results from the initial setup with the endoscopic camera


67
are shown in Figure 5.12 and both the screws and holes appears like holes in the pipes surface.
This is because of the strong reflection from the surface of the screws. Another scanning result
is shown in Figure 5.13 by using the second model with the fisheye camera.
Surface
deformation
300
200 1
100
Figure 5.10: 3D reconstruction from commercial scanner


68
'** } <1 > M :

Figure 5.12: 3D reconstruction with endoscopic camera


69
Figure 5.13: Surface reconstruction with fisheye camera (setup 2)
5.4. Resolution and accuracy of structured light reconstruction
The accuracy and density of the 3D reconstruction in structured light system is
governed by multiple factors. The first factor is the resolution of the acquired image. The size
of the image pixel in real plane defines the maximum resolution that the system can achieve.
However, this factor is also restricted by the processing speed of the acquisition device because
larger images require higher processing time, which means less number of processed frames
per second and higher acquisition time. The second factor is the width of the acquired edges.
In the case of laser scanners, it is represented by the thickness of the projected laser ring that
is applied to the imaged object. Decreasing the thickness of the laser ring increases the
resolution but also decreases the ring detectability. Another important factor is the color
distortion from the pipes walls. For example, red walls that are scanned with red laser scanner


70
suffers from high amount of distortion due to the wall color similarity with the projected ring.
In order to overcome this issue, multicolor patterns are recommended for those surfaces to
increase the probability of detecting the ring despite the variations in surface color.


71
6. Conclusion and future work
Nondestructive evaluation is an important field in the industry and the demand to develop
new techniques is growing due to introduction of new materials and the need for more precise
detection methods. In this study, we have introduced two NDE methods. The first one is
proposed for the detection of breast cancers while the other one is dedicated for the defects
detection in gas pipelines.
Thermoacoustic is a hybrid imaging methodology that combines the advantages of
microwave imaging and ultrasonography. A numerical model is created to simulate the signal
generation of thermoacoustic imaging systems. The model shows that the acquired
thermoacoustic images represent the microwave loss distribution inside the imaged target. The
detected signal level is increased by increasing the target conductivity until it reaches a certain
threshold when the MW signals starts to lose the ability to penetrate through the target due to
its high conductivity. In same manner increasing the frequency increase the signal level due to
the increase of losses with increasing the frequency but this however, decreases the penetration
ability of the MW signal. The study also shows that the MW pulse period has direct effect on
the frequency of the generated acoustic signals. I.e. decreasing the pulse width increases the
frequency of acoustic wave and results in high-resolution images.
Time reversal reconstruction method is used for the reconstruction thermoacoustic
images due to its robustness. The method exploits the reciprocity of the wave equation by
retransmitting the detected acoustic signals back into the medium with a reverse time sequence
to reconstruct the acoustic source. The method is implemented successfully to reconstruct both
simulation and experimental data. For the simulation, different scenarios were implemented
for the target shapes and transducers orientation. The study shows that in order to reconstruct


72
a point then any line passes though that point should intersect with the transducers plane. For
the experimental data, time reversal is implemented after applying filtering to reduce the effect
of noise. Different biological samples were tested with different shapes and conductivities.
The boundaries of the samples were clearly reconstructed but the internal details were blurred
due to the large transducer size and the high amount of noise in the experimental
measurements.
Structured light based reconstruction is an effective 3D characterization method that
employs the triangulation concept to acquire the depth from 2D images. In this study we have
implemented a laser ring based structured light system to reconstruct the defects inside small
gas pipelines. The resolution of this system is governed by the camera resolution and the
thickness of laser ring and the image is highly effected by the pipe color.
Future work
For the thermoacoustic system, the next step is to enhance the signal to noise ratio of
the current experiment and develop a tomographic system that is capable giving 3D
characterization of the image object. The method is also proposed for the testing of composite
materials due to their conductive nature.
For structured light-based system, a multi colors multi rings system is under
development. The speed the segmentation process can be enhanced by using parallel
processing and more efficient segmentation methods. The misalignment of the sensor with pipe
can be eliminated by using an inertial measurement unit to give the real orientation of the
sensor and the 3D registration can be done automatically with ICP.


73
REFERECES
[1] E. Bash, Experimental study of thermoacousti imaigng system, 2015.
[2] T. Hopp and N. V Ruiter, Breast Tissue Characterization by Sound Speed :
Correlation with Mammograms using a 2D / 3D Image Registration, vol. im, pp.
937-940, 2012.
[3] Y. Deng and M. Golkowski, Innovative biomagnetic imaging sensors for breast
cancer: A model-based study, J. Appl. Phys., vol. Ill, no. 7, pp. 7-10, 2012.
[4] Why Should I Use Twitter? [Online], Available:
https://www.osfhealthcare.org/media/filer_public/c4/ee/c4ee865d-e6ce-4f85-b39d-
16858a6adeab/breastfeeding_tips_l-7_english.pdf. [Accessed: 27-Mar-2016],
[5] American cancer society, what is breast cancer. [Online], Available:
http://www.cancer.org/cancer/breastcancer/detailedguide/breast-cancer-what-is-breast-
cancer.
[6] Breast cancer. [Online], Available:
http://www.cancer.org/cancer/breastcancer/detailedguide/breast-cancer-breast-cancer-
types.
[7] X. Wang, Thermoacoustic applications in breast cancer" University of Arizona, vol.
225, no. 3, pp. 61-78, 2014.
[8] X. Wang, D. Bauer, R. Witte, and H. Xin, Microwave-Induced Thermoacoustic
Imaging Model for Potential Breast Cancer Detection, IEEE Trans. Biomed. Eng.,
vol. 06, no. 01, p. 1350001, 2012.
[9] O. Scherzer, Handbook of Mathematical Methods in Imaging. .
[10] Rakruger, Thermoacoustic imaging, Wikipedia. .
[11] Hidden agenda: IED detection using ultra-wideband radar. [Online], Available:
http://www.army-technology.com/features/featureied-detection-ultra-wideband-
microwave-radar-darpa-jieddo/.
[12] T. Qin, X. Wang, H. Meng, Y. Qin, B. Webb, G. Wan, R. S. Witte, and H. Xin,
Microwave-induced thermoacoustic imaging for embedded explosives detection,
IEEE Antennas Propag. Soc. AP-S Int. Symp., pp. 1917-1918, 2014.
[13] Sprialboy, Guided wave testing. [Online], Available:
https://en.wikipedia.Org/wiki/Guided_wave_testing#/media/File:Guided_wave_testing
_GWT.jpg.
[14] visual inspection. [Online], Available: http://thesurge.com/stories/need-job-hiring-
60-natural-gas-pipeline-inspectors-cali.
[15] T. Gunnarsson, Microwave Imaging of Biological Tissues: applied toward breast
tumor detection, 2007.


74
[16] D. Voltmer, Fundamentals of Electromagnetics, vol. 2, no. 1. 2007.
[17] D. Pozar, Microwave Engineering Fourth Edition. 2005.
[18] B. Cox, Acoustics for Ultrasound Imaging, Lect. Notes, no. January, pp. 49-56,
2013.
[19] A. Mashal, J. H. Booske, and S. C. Hagness, Toward contrast-enhanced microwave-
induced thermoacoustic imaging of breast cancer: an experimental study of the effects
of microbubbles on simple thermoacoustic targets. Phys. Med. Biol., vol. 54, no. 3,
pp. 641-650, 2009.
[20] B. E. Treeby and B. T. Cox, k-Wave: MATLAB toolbox for the simulation and
reconstruction of photoacoustic wave fields, J. Biomed. Opt., vol. 15, no. 2, p.
021314, 2010.
[21] M. Haltmeier, O. Scherzer, and G. Zangerl, A reconstruction algorithm for
photoacoustic imaging based on the nonuniform FFT, IEEE Trans. Med. Imaging,
vol. 28, no. 11, pp. 1727-1735, 2009.
[22] Y. Hristova, P. Kuchment, and L. Nguyen, Reconstruction and time reversal in
thermoacoustic tomography in acoustically homogeneous and inhomogeneous media,
Inverse Probl., vol. 24, no. 5, p. 55006, 2008.
[23] G. Chen, Z. Zhao, Z. Nie, and Q. H. Liu, Computational study of time reversal mirror
technique for microwave-induced thermo-acoustic tomography, J. Electromagn.
Waves Appl., vol. 22, no. 16, pp. 2191-2204, 2008.
[24] R. C. Gonzalez and R. E. Woods, Digital Image Processing (3rdEdition). 2007.
[25] Maverick Inspection Ltd., Laser Profiling in PVC Test Pipe with Video Pipe
Camera. [Online], Available: https://www.youtube.com/watch?v=KEspkl5LIv0.


Full Text

PAGE 1

NUMERICAL MODELING FOR NO N INVASIVE IMAGING USING ULTRAHIGH FREQUENCY ELECTROMAGNETIC WAVES by MOHAND ALZUHIRI B.S., University of B aghdad 2011 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fu lfillment of the requirements for the degree of Master of Science Electrical Engineering 201 6

PAGE 2

ii 2016 MOHAND ALZUHIRI ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Science degree by M ohand A lzuhiri has been approved for the Elect rical Engineering B y Yiming Deng Chair Mark Golkowski Stephen Gedney May 4 2016

PAGE 4

iv Alzuhiri Mohand (M.S. Electrical Engineering) Numerical Modeling for Non invasive Imaging U sing Ultrahigh F requency Electromagnetic Waves Thesis Directed by Assist ant Professor Yiming Deng ABSTRACT Nondestructive evaluation (ND E ) is an important field in industry that has a wide range of applications to guarant ee the quality of new products, maintain the safety of working equipment and perform medical diagnos is In this thesis work we are presenting two noninvasive ND E imaging techniques microwave induced thermo acoustic imaging for breast cancer diagnosis and structured light 3D characterization for gas pipes internal surface reconstruction. Therm o a coustic imaging is an emerging imaging technique that combines both high contrast of microwave imaging and sub millimeter resolution of u ltrasound imaging. This imaging technique is highly a ffected by acoustic and electrical prope rties of the imaged target. O ther importa nt factors are the properties of the excitation source such as its electric field pattern and pulse duration. This thesis developed a numerical model to simulate the signal generation of near field microwave induced thermoacoustic imaging. The study simula tes both the electromagnetic interaction of microwaves with the imaged object as well as the generation of acoustic signals and their propagation from the target to the transducers The study also investigates the effect of different design parameters tha t determine the intensity and frequency of generated acoustic signals for imaging Various parameters are investigated, such as electrical properties, microwave frequenc ies and the microwave source pulse width. Another major contribution of this thesis work is on dat a reconstruction for optical NDE methods. S tructured light has been widely used to create 3D models of objects. In this thesis, we propose to use this technique to characterize and analyze damage and defects inside natural gas

PAGE 5

v pipeline s. The study aims to develop and implement 3D reconstruction algorithms to characterize the defects and deformations while the data are acquired using the optical sensors developed at CU LEAP group The study develops a reconstruction algorithm for single laser ring laser ring scanner and evaluates its performance. The form and content of this abstract are approved. I recommend its publication. Approved: Yiming Deng

PAGE 6

vi DEDICATION I dedicate this work to my father, mother and my family who supported me thro ughout my life and motivated me to chall enge the difficulties of life and become the person who I am

PAGE 7

vii ACKNOWLEDGMENT This work was supported by the higher committee of education development in Iraq I want to thank them for giving me this great opportuni ty to finish my master degree in the United States. I want to thank to my LEAP group for their kind support and advice W ithout our teamwork I also want to thank my partner Ryan Jacob for his great help with the experimen tal work Special thanks to my advisor Dr Yiming Deng for his invaluable advice s and guidance through the thesis work. Finally, I want to thank Dr Golkowski and Dr Gedney for their help and advice on electromagnetic theory and simulations.

PAGE 8

viii TABLE OF C ONTENTS Chapter Introduction ................................ ................................ ................................ ....................... 1 Breast cancer ................................ ................................ ................................ .............. 1 Microwave ind uced thermoacoustic imaging ................................ ............................ 4 Pipelines internal surface reconstruction ................................ ................................ .... 6 Structured light ................................ ................................ ................................ ........... 8 Thesis scope ................................ ................................ ................................ ............... 9 Motivation ................................ ................................ ................................ ................ 10 Theoretical background ................................ ................................ ................................ ... 11 Electromagnetic waves propagation and absorption ................................ ................ 11 2.1.1. Electromagnetic spectrum ................................ ................................ ................. 11 2 .1.2. EM waves propagation ................................ ................................ ..................... 13 2.1.3. Boundary conditions ................................ ................................ ......................... 15 2.1.4. Attenuation ................................ ................................ ................................ ........ 16 2.1.5. Near field and far field ................................ ................................ ...................... 17 2.1.6. Rectangular waveguide ................................ ................................ ..................... 18 2.1.7. Microwave heating ................................ ................................ ............................ 19 Acoustic propagation theory ................................ ................................ .................... 20 2.2.1. Wave equation ................................ ................................ ................................ .. 21 2.2.2. Acoustic impedance ................................ ................................ .......................... 22 2.2.3. Acoustic wave pro pagation in inhomogeneous media ................................ ...... 22 2.2.4. Acoustic boundary conditions ................................ ................................ ........... 23 Thermoacoustic equation ................................ ................................ ......................... 24 Light triangulation: ................................ ................................ ................................ ... 26 Numerical modeling of thermoacous tic signal generation ................................ .............. 28 Experimental Imaging model ................................ ................................ ................... 28 COMSOL ................................ ................................ ................................ ................. 29 Electromagnetic waves simulation ................................ ................................ ........... 30 3.3.1. Simulation geometry ................................ ................................ ......................... 30 3.3.2. Simulation parameters ................................ ................................ ...................... 32 3.3.3. Acoustic propagation model ................................ ................................ ............. 33

PAGE 9

ix 3.3.4. Simulation parameters ................................ ................................ ...................... 34 Model valida tion ................................ ................................ ................................ ...... 34 Open ended waveguide vs horn antenna ................................ ................................ .. 37 Case studies ................................ ................................ ................................ .............. 39 3.6.1. Effect of frequency ................................ ................................ ........................... 39 3.6.2. Effect of material conductivity ................................ ................................ ......... 39 3.6.3. Effect of microwave pulse width ................................ ................................ ...... 39 3.6.4. Optimal target analysis ................................ ................................ ..................... 40 Image reconstruction ................................ ................................ ................................ ....... 43 K Wave ................................ ................................ ................................ ................... 43 Reconstruction challenges ................................ ................................ ........................ 44 4.2.1. Homogeneity of sound speed ................................ ................................ ............ 44 4.2.2. Estimation of sound speed ................................ ................................ ................ 44 4.2.3. Size of transducer ................................ ................................ .............................. 45 4.2.4. Effect of transducer orientation ................................ ................................ ........ 45 Image reconstruction with time reversal ................................ ................................ .. 46 Simulation study of image reconstruction with time reversal ................................ .. 48 4.4.1. Effect of MW pulse width ................................ ................................ ................. 50 4.4.2. Biological samples embedded in a conductive medium ................................ ... 51 Reconstruction of experimental data ................................ ................................ ........ 52 Uniform piece of meat ................................ ................................ ...................... 52 4.5.2. Rectangular biological tissue with copper wire ................................ ................ 53 4.5.3. The effect of inserting highly conductive tissue ................................ ............... 55 4.5.4. Two separate pieces of meat ................................ ................................ ............. 56 Pipe internal surface reconstruction ................................ ................................ ................ 58 Scanner setup ................................ ................................ ................................ ............ 58 Extraction of 3D information ................................ ................................ ................... 62 Experimental results ................................ ................................ ................................ 66 Resolution and accuracy of structured light reconstruction ................................ ..... 69 Conclusion and future work ................................ ................................ ............................ 71 References ................................ ................................ ................................ .............................. 73

PAGE 10

x LIST OF TABLES Table 1. 1: Sound speed in breast tissues ................................ ................................ ............................. 4 2.1: Electric properties of different tissues at 2.5GHz ................................ ............................ 12 3.1: Variation of ethylene glycol pr operties with microbubbles concentration ...................... 35 4.1: Properties of simulated geometry ................................ ................................ .................... 49 4.2: Properties simulated materials ................................ ................................ ......................... 52

PAGE 11

xi LIST OF FIGURES Figure 1.1: Normal human breast ................................ ................................ ................................ ........ 3 1.2: Generation of thermoacoustic signals ................................ ................................ ............... 5 1.3: External inspection of pipelines ................................ ................................ ........................ 7 1.4: Internal inspection of pipelines ................................ ................................ ......................... 8 1.5: Structured light based depth calculation ................................ ................................ .......... 9 2.1: Electromagnetic spectrum ................................ ................................ ............................... 12 2.2: Field regions around anten na ................................ ................................ .......................... 17 2.3: Rectangular waveguide ................................ ................................ ................................ ... 19 2.4: Optical triangulation ................................ ................................ ................................ ....... 26 3.1: Modeling overview for MITAI ................................ ................................ ....................... 28 3.2: Experimental setup of MITAI ................................ ................................ ......................... 29 3.3: 3D geometry of the EM simulation ................................ ................................ ................ 31 3.4: Cross section plane through the 3D geometry ................................ ................................ 32 3.5: Meshing of 3D geometry ................................ ................................ ................................ 33 3.6: Acoustic simulation region ................................ ................................ ............................. 34 3.7: Meshing of acoustic region ................................ ................................ ............................. 35 3.8: Comparison between simulation and experimental results ................................ ............. 36 3.9: Comparison with other simulation work ................................ ................................ ........ 37 3.10: Electric field distribution for horn and open ended WG antennas ............................... 38 3.11: Effect of frequency on acoustic signal amplitude ................................ ......................... 40 3.12: Effect of material conductivity of acoustic signal amplitude ................................ ....... 41 3.13: Effect of MW pulse width of the wavelenght of acoustic signal ................................ .. 41 3.14: Effect of changing permittivity and conductivity over wire range of values ............... 42 4.1: Effect transducer orientation on image reconstruction ................................ ................... 46 4.2: Time reversal reconstruction of simulation data ................................ ............................. 49 4.3: Reconstruction with different MW pulse widths ................................ ............................ 50 4.4: Reconstruction of samples that are embedded in another conductive medium .............. 51 4.5: Imaging of a uniform piece of meat ................................ ................................ ................ 53 4.6: Imaging of a uniform piece meat with copper wire ................................ ........................ 54 4.7: Effect of using highly conductive tissue ................................ ................................ ......... 55 4.8: Imaging of two separate pieces of meat ................................ ................................ .......... 56 5.1: Laser source and camera orientation in setup 1 ................................ .............................. 59 5.2: Laser source and camera orientation in setup 2 ................................ .............................. 59 5.3: Initial setup with endoscopic camera ................................ ................................ .............. 60 5.4: LEAP experimenta l setups with fisheye camera ................................ ............................ 60 5.5: Comparison between images of endoscopic and fisheye cameras ................................ 61 5.6: Flowchart of the 3D r econstruction ................................ ................................ ................ 63 5.7: Comparison between red and green layers ................................ ................................ ..... 64 5.8: Power law transformation ................................ ................................ ............................... 64 5.9: Laser ring after applying thinning process ................................ ................................ ..... 66

PAGE 12

xii 5.10: 3D reconstruction from commercial scanner ................................ ................................ 67 ................................ ................................ ............ 68 5.12: 3D reconstruction with endoscopic camera ................................ ................................ .. 68 5.13: Surface re construction with fisheye camera (setup 2) ................................ .................. 69

PAGE 13

1 Introduction Nondestructive evaluation is the science that deals with testing and analyzing the material properties of objects without damaging them or alte ring their physical content. NDE is applied in fields where material failure could result in high possibility of hazard or economic loss es NDE is widely used in industry to test the quality of new product s and inspection of aging critical infrastructures In medicine, simila r NDE methodolog ies are used to inspect the damage that happens to the human body like bone fractures tumors or to monitor physiological process that is undergoing inside the body like pregnancy examination using ultrasonic imaging NDE depends on the res ponse s of the material to electrical, mechanical and chemical energy interactions in the testing process Some of the well known techniques are, X rays, ultrasonography, magnetic resonance imaging, optical microscopy and microwave imaging, etc. All of thos e methods should leave no harmful effects on th e tested object if the test s are conducted under proper conditions or sometimes the effects are negligible. In this study, two nondestructive evaluation methods are developed and discussed for both medical and industrial applications Breast cancer Breast cancer is a malignant tumor that starts in a Breast cells then grows out of control. It is very rare in males and most of the cases are diagnosed in females [1] Studies show that it threatens the life of millions of women around the world [2] In 2014, more than 232 thousand cases of breast cancer were diagnosed in the United States only [2] and studies show that those patients have th e second highest death rates among other cancer patients [3] C ancer tumors can be classified according to their spreading nature into two categories. The first type is known as in s itu carcinoma this type of tumors is confined and d oes not spread to its neighbor and

PAGE 14

2 this is the reason of considering it as a noninvasive cancer The second type is i nvasive carcinoma I n this type cancer cells spread through the breast and other parts of the body I t is considered as the most dangerous type of cancer cells because it ca nnot be confined and continues to spread and destroy all the body systems As the cancerous cells spread through the body the amount of damage increase s and controlling the situation becomes harder. As a result the patie nt survival highly depends on the cancer stage when the patient is diagnosed i .e. early d etection of cancer cells plays an important role in cancer treatment and saving the patient life The h uman breast can be divided into two main regions according t o the type of dominant tissues as shown in Figure 1 1 [4] The first region is dominant with mammary glands (lobules) that are respons ible for milk production while t he second region is d ominant with fatty tissues [5] I n addition to those regions, the breast has large number of milk ducts that transfer the milk from the gland to the nipple and the whole breast structure is shielded with skin from outside The dominant type of breast cancer is originated from ducts and is called ductal cancer [5] This cancer type represents 80 percent of invasive cancer cases [6] Another type of cancer cell can be originated from the lobules and is called lobular cancer This type represents 10 percent of the invasive cancer cases [6] Breast cancer can also originate from other tissues but they are uncommon.

PAGE 15

3 Figure 1 1 : Normal human breast X ray mammography is widely used for breast cancer screening and is considered as the standard met hod for initial diagnostics [3] X ray proved to be a n effective tool for reduc ing the number of cancer fatalities since its application but it suffers from some drawbacks. Studies show that about 15% of the cases were detected improperly [3] and s ome of the cases canno t be detected if the tumor is embedded in si de radiographically dense breasts (breast with a high rate of gl andular and connective tissues) [7] Other cases were reported while there were no real cancer cells. This is added to the fac t that x ray relies on ionizing radiation in the process of imaging which increases the probability of creating new cancer cells due to the ionizing radiation [3] Another drawback with mammography is the applied pressure to the breas t during the diagnostic process which is a source of discomfort to the patient.

PAGE 16

4 On the other hand, MRI is a capable imaging technique that give s superior i mages but it is considered to be an expensive solution for routine screening [8] For this reason it is mostly used for further evaluations in cancer diagnostic. In addition to cost, MRI also suffers from high false positives and long imaging time [7] Recently Microwave imaging was proposed to detect cancer tumors. This method proved that it can provides good contrast in between breast tissues due to the variation of electrical propert ies of those t issues But this method suffers from the low resolution of acquired images due to t he long wavelength in the microwave region [3] On the contrary Ultrasound offers good resolution (sub millimeter) but suffers from poor contrast in b iological tissues due to the small variation of sound speed in breast tissues Table 1 1 shows the sound speed in different breast tissues [2] Table 1 1 : S ound speed in breast tissues Tissue type Sound speed (m/s) Breast Fat 1457 Breast Gland 1470 Tumor tissue 1509 Skin 1624 Microwave induced thermoacoustic imaging The deficiencies of the current imaging methodolog ies encouraged the development of alternative solutions that are efficient and cost effective and this led to the introduct ion of thermoacoustic imaging Since the ninth decade of last century Microwave Induced Thermoacoustic Imaging ( MITAI ) has attracte d wide attention in the research community and some breakthroughs have already been achieved to enhance signal generation and reconstruction quality [9] Microwave i nduced t hermoacoustic i maging is a hybrid imaging methodology that combines microwave imaging with ultrasonography. Hybrid imaging methodo logies combine

PAGE 17

5 more than one imaging technique to achieve the best qualities of both of them. The result of this combination in MITAI is a methodology with the high contrast of microwave imaging and fine resolution of ultrasonography It is described as ei ther ultrasound enhanced electromagnetic modality or an ul trasound modality that take the advantage of electromagnetic wave contrast [9] MITAI signals intensity depe n ds of the amount of heat generation inside the sample, i.e the images represent the spatially variant electromagnetic loss distribution insi de t he imaged object [9] Figure 1 2 : G eneration of thermoacoustic signals MITAI signals are generated by sending strong short microwave pulses into the imaged object as shown in Figure 1 2 [10] The microwave pulse absorption inside the ele ctrically lossy target results in a temporal heat gradient inside the imaged object. The heat gradient excites a thermo elastic expansion that generates acoustic wave s in the range of ultrasound frequencies Then those acoustic waves are used to reconstruc tomography approaches. The frequency of the acoustic waves is inversely proportional to width of the electromagnetic excitation pulse.

PAGE 18

6 P hotoacoustic and thermoacoustic imaging share the same physical concept The nam ing depends on the type of employed excitation source that is used for heat generation Photoacoustic imaging employs high power lasers while thermoacoustic imaging depends on microwave pulses to initiate the thermal expansion. With photoacoustic the imaged object need s to be scanned by a thin laser beam while in thermoacoustic imaging the whole target can be scanned at one time if the right setup is available. MITAI is attractive as an imaging solution for breast cancer due to its low cost, simpli city, high resolution and good contrast. Breast cancer is one of the potential application for MITAI but it can also be employed in other nondestructive evaluation processes like detection explosives embedded in biological tissues [11] [12] For this application explosive devices appear as dark objects in the images due to their small conductivity (0.001 S/m) [12] With this conductivi ty contrast explosives can be easily differentiated biological tissues. Pipelines internal surface reconstruction With the current industr ial revolution, energy became a crucial factor in the success of any business. In order to satisfy this demand for ene rgy, gas and oil pipelines networks have extended over thousands of kilometers over land and under sea Th e se huge networks require routine maintenance in order to guaranty the safety and durability of the networks. In order to guarant ee that many nondest ructive evaluation methods are being used to check the safety of these pipes. Most of the current NDE methods can be classified into two categories, external and internal methods. External methods are used to conduct the tests from outside the pipe without the need to take the pipe to an offline state. This can be performed by ultrasonic guided wave testing or simple visual inspection as shown in Figure 1 3 [13] [14]

PAGE 19

7 In the case of internal inspection methods, passive or active sensors are deployed inside the pipe. Those methods require the pipes to be offline in order to go insi de the pipe and do in depth analysis of the problem. These test s can be conducted by smart pipeline inspection gauges (pigs) and laser scanners as shown in Figure 1 4 .a and b respectively Pigs can be equipped wit h multiple nondestructive evaluation tools like magnetic flux leakage and ultrasound sensors for metal pipes health monitoring Laser scanners are used to detect cracks, defects and internal surface deformation s Most of the current internal diagnostic dev ices are designed for large size pipes with a diameter larger than 6 inches which leaves the area of small diameter pipes with no efficient solutions. Figure 1 3 : External inspection of pipelines Multiple approaches are available for 3D characterization. Lidar is a laser based technology that depends on the time of flight for depth calculation. This technology is robust and accur ate but the equipment has large sizes and is relatively expensive. Another approach is passive stereovision where two cameras are used to capture the seen and the shift between the two images is used for depth calculations. This method has been tested but the results were (a) (b)

PAGE 20

8 not good enough due to the poor illumination inside the pipe and the distance between the cameras were very small due to the pipe size. Figure 1 4 : Internal inspection of pipelines Structured light Structured light is a method to acquire depth images from the deformations of projected light patterns by employing triangulation techniques. The depth acquisition is im plemented by using a camera and light projection device as shown in Figure 1 5 A predefined light pattern with sharp edges is projected on the object and imaged by a camera that is placed on a well known baseline. The dimensions of image s are converted to their values in the real plane and the depth of the image is calculated by determining the amount of the shift of lines in the recorded image. Different light sources and pattern can be employed for depth reconstruction. The sou rce can be a single laser source that projects single sharp line or a projector that projects different kind of patterns. The density of the projected patterns determines the number of acquired 3D points from each frame but at the same time decreases (a) (b)

PAGE 21

9 the p ossibility of detecting the edges correctly. In this study, a laser source that projects a single colored ring is used as the main projector The ring pattern is chosen because the sensor is designed to work in a circular geometry. Both of the projectors a nd camera are miniaturized to be suitable to inspect pipelines with diameters smaller than 3 inches. This approach is chosen because it is simple, cost effective and can be miniaturized easily. Figure 1 5 : Structured light based depth calculation T hesis scope This thesis aims to create a high fidelity numerical model to simulate the electromagnetic and acoustic interactions in thermoacoustic imaging. The model is design ed to optimize the acoustic signal generation and enhance the electromagnetic wave absorption inside the imaged objects. The study also focuses on the reconstruction algorithm s and implement s and test s different types to compare their performance. For ther moacoustic imaging the goals can briefed as bellow

PAGE 22

10 Develop an integrated model that simulate s the process Validate the model with experimental results Study the properties of thermoacoustic signals Analyze the suitability of the method for different applic ations Apply different reconstruction algorithms to reconstruct the data The study also develops structured light based reconstruction algorithms for the characterization of pipe s internal surface reconstruction. It analyzes the suitability and stability of those reco nstruction methods and compare their performance. M otivation The motivation of the study is the need to develop new imaging technique s that are able to avoid the drawbacks of current imaging methodologies and provide efficient and cost effecti ve solution s for industrial applications.

PAGE 23

T heo retical background Different types of electromagnetic and mechanical waves are used in this study. Thermoacoustic imaging employs two types of wave s in the imaging process. The first type is the electromagn etic waves that are used in the heating of the imaged sample The second type is the acoustic waves that are used to extract the characteristics of the imaged object. While structured light employs visible light in the inspection process. Another issue is the conversion between the microwave heating and the acoustic excitation (thermoacoustic effect). This chapter gives a brief introduction to the wave propagation, scattering, reflection and present the basics of each imaging system. E lectromagnetic waves p ropagation and absorption Thermoacoustic imaging depends on microwave s as the main source of the heating in the imaging system. Hence the propagation of EM waves in materials and in free space plays important role in modeling the imaging system In thi s section we are giving a brief review of the mechanism of the EM wave propagation and scattering in materials. This section also provides a brief introduction to transmission lines and some antenna concepts needed in the modeling process 2.1.1. Electromagnetic spectrum Electromagnetic (EM) waves are transverse electric and magnetic waves where the electric and magnetic fields are orthogonal to each other and to the direction of propagation. The EM spectrum expa nds from zero hertz to infinity and the acteristics and their interaction with materials varies according to their frequency. Figure 2 1 shows different frequency bands of EM spectrum where each band has a different kind of wave that has its own special characteristics F or the thermoacoustic signals we are mainly interested in the

PAGE 24

12 microwave region. This region is chosen because the cancerous cells show relatively good contrast from other breast tissues at this band of EM spectrum Table 2 1 [15] shows the electrical permittivity and conductivity of different biological tissues at 2.5 GHz. Another important factor is the availability of the required efficient sources, amplifiers and other component s in this region of electromagnetic spectrum. Figure 2 1 : Electromagnetic spectrum Table 2 1 : Electric properties of different tissues at 2.5 GHz Tissue name Permittiv ity Conductivity(S/m) Blood 56 60 2.5 Bone 12 0.4 Brain (Grey matter) 45 2 Fat (Not infiltrated) 4 5 0.07 0.1 Heart 55 2.3 Kidney (Cortex) 55 2.5 Liver 42 1.8 Lung 20 0.7 0.8 Muscle 50 55 1.8 2.2 Skin (Dry) 38 1.5 Skin (Wet) 43 1.8 Spleen 52 2.2 Tendon 42 1.8 For the case of structured light based internal surface reconstruction the waves are in the visible light region. Those wave s can provide very good resolution for our system due to their short

PAGE 25

13 wavelength (400 700nm). The equipment in this region are well developed and the cameras can provide superior resolution s 2.1.2. EM waves propagation The EM wave propagation is governed by the medium electrical properties. The medium is electrically characterized by three factors and they are conducti on, polarization and magnetization The dominance of any one of those factors specifies the type of the material if it is a conductor or dielectric or magnetic material. The effect of those factors can be shown their constitutive relations. equations govern the EM wave propagation in any medium The equations are given by the following formulas : ( 2 1 ) ( 2 2 ) ( 2 3 ) ( 2 4 ) Where are the electric field intensi ty electric flux density, magnetic field intensity magnetic flux density, magnetic permeability, electric permittivity, electric current density magnetic current density and electric conductivity respectively Equation ( 2 1 ) is Faraday s law and it states that any time varying magnetic field is associated with a spatially varying electric field. Equation ( 2 2 ) is law and it states that any time varying electric field is associated with a sp atially varying magnetic field. Equation ( 2 3 ) is Gauss law of electricity and it states the net electric flux radiated out of any close surface is equal to the

PAGE 26

14 amount of charge enclosed by the surface. Equation ( 2 4 ) is Gauss law of magnetism and it states the net magnetic flux radiated out of any close surface equals to zero. Another auxiliary equation is called the continuity equation and it is used to define the relationship between the electric cur rent and charge density and i t is given by ( 2 5 ) that define the relation ship between the field components Those r elations are given by ( 2 6 ) ( 2 7 ) ( 2 8 ) and are free space electric permittivity and magnetic permeability respectively. and represent the atomic and molecular dipoles of the material and magnetic dipole o f the atoms in the medium [16] By using the phasor form, the derivative s are exchanged with and equation ( 2 1 ) and ( 2 2 ) are written as: ( 2 9 ) ( 2 10 ) By taking the curl of ( 2 9 ) and us ing ( 2 10 ) and the constitutive relations we get ( 2 11 ) Moreover, by using the vector identity ( 2 11 ) can be writte n as ( 2 12 ) This expression is called the electromagnetic wave equation

PAGE 27

15 2.1.3. Boundary conditions B ehavior of field components at any boundary that separate two different mediums is gove rned by a group of boundary conditions. The boundary conditions at a charge free surface are given by ( 2 13 ) ( 2 14 ) ( 2 15 ) ( 2 16 ) The equations show that the tangential components of E and H are continuous across th e boundary. They also show that the normal components of D and B are continuous across the boundar y, i .e. a ny incident wave that travels toward a boundary is ether reflected from the boundary or pass through to the other medium or suffers from both reflec tion and transmission The amount of reflection depend s on the wave impedance difference between the two medi a Wave impedance is the ratio between the transverse electric and magnetic components of the wave in medium. Wave impedance is given by ( 2 17 ) The wave transition between two media medium one and medium two is described in terms of reflection and transmiss ion coefficients. Reflection coefficient ( ) represents the ratio between the reflected and incident wave and it is given by

PAGE 28

16 ( 2 18 ) are the wave impedances of the first and second mediums. and are the incident and transmission angles. Transmission coefficient ( T ) represents the ratio between the transmitted and incident wave and it is given by ( 2 19 ) a nd the summation of reflection and transmission coefficients equals to 1 in the lossless case. 2.1.4. Attenuation The wave amplitude is degrade d throughout the its propagation into the medium. This degradation is either due to the spread of the wave energy over a large area or due to the losses in the medium. The loss es can be account ed for in the wave equation. The lossy version of wave equation is written as ( 2 20 ) are the propagation constant, attenuation constant and phase constant. and are expressed by ( 2 21 ) ( 2 22 ) is the angular frequency of the applied wave.

PAGE 29

17 2.1.5. Near field and far field The radiation region around an antenna is divided into thr ee main regions illustrated in Figure 2 2 The first region is the reactive near field It starts from the region exactly next to the antenna and extend s to D is the largest dimension of the antenna and is the transmitted wavelength. This region is mainly dominated by the reactive field. The second region is the radiating near field and it is surrounded by the reactive near field and far field regions. This region is a transitional stage between the nea r and far field s and it is dominated by radiation f ields and the angular field distribution is governed by the distance from the antenna If the antenna has a maximum dimen sion that is not large compared to the wavelength, this region may not exist. The third region is the far field region where the angular field distribution is independent of the distance from the antenna. In general, this region starts from to infinity with some exception for some types of antennas Figure 2 2 : F ield regions around antenna

PAGE 30

18 2.1.6. R ectangular waveguide The w aveguide is described as a structure that is used to transfer EM power from one point to another. In this study, an open ended rectangular waveguide is used to tra nsfer the power from the source to the imaged target. Rectangular waveguides are widely used in the microwave region of the EM spectrum due to their ability to handle high amounts of power with high efficiency Rectangular waveguide s consist of four conduc tive plates. Each two plates are parallel to each other perpendicular to th e other two plates as shown in Figure 2 3 [17] The wave propagation inside rectangular waveguide is governed by the reflections from its highl y conductive walls. As a result, a specific number of guided modes are excited between the conductive walls. The number of those modes depends on the electromagnetic [17] Therefor e, the modes are either transverse electric or magnetic. The waveguide acts like a high pass filter, and the minimum frequency that can propagate inside it is governed by the following equation [17] ( 2 23 ) n,m represent the number of excited modes For the first dominant mode TE10, the electric and magnetic fields are given by ( 2 24 )

PAGE 31

19 is an arbitrary constant. are the propagation constant, attenuation constant and phase constant. In computational electromagnetic, waveguides are simply modeled with four PEC boundaries. At those boundaries, all the tangential electric fields are set to ze ro. 2.1.7. M icrowave heating Microwave heating has a broad range of applications in industry and household appliances. It has being widely used in home kitchens and food processing. Recently microwave heating has had many medical applications such as thermoa coustic imaging tumor ablation. The m icrowave heating is generated by focusing a strong microwave beam on a target. The propagated EM waves suffer from losses inside the targeted object. As a result, those EM power losses are converted into heat. The amoun t of resulted heat depends on multiple factors like material conductivity, permittivity, and density (related to heat transfer). It also depends on the location of the target from the microwave source. The location a ffects Figure 2 3 : Rectangular waveguide

PAGE 32

20 the intensity and the distributio n of electromagnetic power inside the Target. From Maxwell equations, the instantaneous power loss density is given by the following equation [3] : ( 2 25 ) The instantaneous power loss density is integrated over time to calculate the total dissipated power inside th e target. The total dissipated power inside the heated target is given by [3] ( 2 26 ) is the t otal simulation time and is the simulation time step A coustic propagation theory S oundwave s are one of the well known types of waves due to their wide range of applications. In addition to the fact that it is an important communication method used by humans and animals it also has many other medical and industrial applications. It is used for nondestructive evaluation of material s medical imaging and as a destructive tool ( high intensity focused ultrasound) for therapeutic purposes. A coustic waves are mechanical longitudinal pressure wave s that require the propagation medium to have both compressibility and inertia. Due to their mechanical nature acoustic waves cannot propagate in vacuum or free space. The s peed of the sound waves mainly depends on the medium mechanical properties. Their spectrum expands over wide range of frequencies and can be divided into three main [18] Infrasonic region: less than 20 Hz Audio frequency region: from 20 Hz to 20 kHz

PAGE 33

21 Ultrasonic region: higher than 20 kHz Human being s can only recognize frequencies in the audio region while some other animals can hear further in the ultrasonic region This study is mainly interested in the ultrasound waves due to their short wavelength The propagation of those waves manly depends on : Particle displacement ( during the wave propagation Particle velocity ( ): is the velocity of particles inside the medium. It consis ts of two components and t he y are non oscillatory net flow velocity ( ) and the oscillatory velocity ( ). The oscillatory velocity refers to the derivative of the particles displacement The total particle velocity is the summation of both of them, i.e. Acoustic pressure ( p ): repre sents the amount of compression of media. It also consists of an ambient pressure and an oscillatory pressure due to the wave propagation. The total pressure is the summation of them ( ). Material density ( ): represents the amount of mass per unit volume. The wave propagation through the medium leads to a fluctuation in the ambient mass due to the particles displacement. The total density is written as 2.2.1. Wave equation P ropagation of acoustic wave in any medium is govern ed by the acoustic wave equation. For heterogeneous mediums it is given by ( 2 27 )

PAGE 34

22 Where p is the acoustic pressure and c 0 is the sound spee d in the medium. For a homogenous medium where the ambient density has a fixed value everywh ere, the equation is reduced to ( 2 28 ) 2.2.2. Acoustic impedance The acousti c impedance represents the amount of opposition of the medium to the acoustic pressure flow. There are two types of acoustic impedances: Specific acoustic impedance ( ): is the ration between the acoustic pressure and the particle velocity in a spec ific direction ( 2 29 ) is a unit vector in the direction of interest. C haracteristic acoustic impedance ( ) : is a constant that represent s the relation between the pressure and particle velocity in a plane wave [18] ( 2 30 ) 2.2.3. Acoustic wave propagation in inhomogeneous media The propagation of acoustic waves in any media is encountered by many restriction s The first one is the atte nuation of the signal due to spat ial spread or due to the abortion inside the propagation medium. The second factor is the reflection of the acoustic waves from boundaries between different materials due to the difference in the characteristic impedance The difference of sound speed does not only cause the wave to reflect, a portion of the wave is also transmitted through the boundary and some are scattered by the object if the object size

PAGE 35

23 is small with respect to the wavelength. I .e. the size of the object and the amount of variat ion in the wave speed and incidence angle result s in different scattering effect s 2.2.4. Acoustic boundary conditions The wave incidence on a boundary is classified according to the a ngle of incidence into two types, normal and oblique incidence. By starting wi th normal incidence, when a wave reaches a boun dary, it is either reflected in the opposite direction or transmitted through the boundary wall. The reflection and transmission through boundary are better explained in the terms of reflection and transmissio n coefficients. The reflection coefficient (R) is the ratio between reflected and incident waves. Transmission coefficient (T) represent s the ratio between tr ansmitted and incident waves. They are written as ( 2 31 ) r epresent the pressure of incident, reflected and transmitted wave s Boundary condition at any acoustic boundary are given by ( 2 32 ) are the particle velocities of incident, reflected and transmitted wave s Dividing the equations in 22 yield s to: ( 2 33 ) In plane waves and this yield s to ( 2 34 )

PAGE 36

24 Using the relatio n ( pressure continuity ) the transmission and reflection coe ffi cient s become: ( 2 35 ) ( 2 36 ) For oblique incident the angle of incident is taken in to account while all the boundary conditions still hold and the reflection and transmission coefficient s become : ( 2 37 ) ( 2 38 ) T hermoacoustic equation T he t hermoacoustic wave generation is a mechanical phenomenon that results from thermal expa nsion of mediums. Those thermal expansions result in a series of mechanical waves (acoustic waves) that are govern by the wa ve equation. Thermoacoustic signal generation and propagat ion are described by thermoacoustic equation and it is given by [9] : ( 2 39 ) are the sound speed, volume expansion coefficient and the isothermal compressib ility of the medium. is the instantaneous pressure at location r i s the sound speed in the medium. T(r,t) is the temperature of the medium. The left side of the equation represent s the acoustic wave equation while t he right side of the e quation represents the photoacoustic source. The photoacoustic source depends on the second derivative of time varying temperature, volume expansion coefficient and heat capacity. The heat capacity and volume expansion

PAGE 37

25 coefficient are consonant s that depen and the only varying term in the source is the acceleration of temperature. When thermal confinement condition is met, the source part becomes equal s to [9] ( 2 40 ) are the material den sity and specific heat capacity of the medium. H(r,t) is the heating function that represents the amount of dissipated energy per unit volume an d time by the excitation pulse a nd it is also called the specific absorption rate (SAR) of the material. Thermal confinement condition assumes that the excitation pulse is short enough so that the acoustic wave is excited before the occurrence of any significant heat conduction. This condition requires that the electromagnetic pulse to has an excitation period that satisfies the following condition ( 2 41 ) Where and are characteristic dimension (m) and the thermal diffusivity of the heated region The heating function can be writt en as a two separate terms ( 2 42 ) Where is the absorbed energy d ensity and I(t) is the t emporal envelope of the excitation pulse. As mentioned in section 2.1.7 t he abso rbed energy density is given by ( 2 43 ) Where and are the effective material conductivity and the amplitude of applied electric fiel d inside the target. By combing those relations, the thermoacoustic equation is re written as

PAGE 38

26 ( 2 44 ) Where is the specific heat capacity of the medium under constant pressure. Light triangulation Triangulation is the process of calculating the position of a point in a 3D space by using the angle s to it from a specific baseline Figure 2 4 shows a simple 2D triangulation and the imaging system orientation for depth calculation. Here we ha v e a light source that sends it s signals at a certain angle toward an object. Another observation point is placed at a distance d from the source to monitor the location of the projected light spot. A straight line is drawn from the observation point to the location of the incidence to calculate the angle After calculating is calculated directly ( ) By knowing all the angl es and the base length the depth calculation becomes handy process. Figure 2 4 : O ptical triangulation By using the normal to d we get

PAGE 39

27 ( 2 45 ) And by using the relation ( 2 46 ) By knowing d, the point coordinate is given by ( 2 47 )

PAGE 40

28 Numerical m odeling of thermoacoustic signal generation Thermoacoustic imaging modeling consists of a forward model that simulate s the signals generation and a backward model that is responsible for the reconstruction of final image s An overview the generation and reconstruction modeling is explained in Figure 3 1 where the forward model is implemented by COMSOL Multiphysics software and the backward model is implemented by usin g K wave toolbox This chapter discusses the modeling process and signals characteristics in the forward model. The model simulates th e microwave propagation, electromagnetic absorption and heat generation, and the thermoacoustic signal generation and propagation. The model also studies the effect of different experimental parameters on signal characteristics and investigates the optimal experimental conditions t o generate the strongest acoustic signals Figure 3 1 : Modeling overview for MITAI E xperimental Imaging model Our experiment employs the standard experimental setup of thermoacoustic signals generation with some extra modifications. The imaging system consists of the follo wing parts as shown in Figure 3 2

PAGE 41

29 Lucas Epsco PG5KB pulsed microwave source with a maximum output voltage of 5k V The device is responsible for providing the required excitation signal to initiate the thermoacoust ic effect. It is tuned to send s 4.5Kv pulses with a carrier frequency of 2.4 5 GHz and pulse width varies from 0.3 to 3 microseconds. A coupling device to transfer the signal of the microwave source to the imaged target. In our experiment this is represent ed by an open ended waveguide due to its narrow beam (comparison with horn antenna is provided later). An Olympus acoustic transducer with a center frequency of 2.45 MHz is used to detect the thermoacoustic signals Peripheral devices for synchronization, amplification, filtering and post processing of the received acoustic signal Figure 3 2 : E xperimental setup of MITAI COMSOL COMSOL Multiphysics is a commercial simulation software that is used to solve co upled phenomena problems It is a Finite Element Method (FEM) based solver in time and frequency domain. The software is divided in to several types of modules where each module

PAGE 42

30 is specified for solving a certain phenomenon type with the ability to couple t hose modules In this study we mainly use two types of modules RF Module: this module is responsible for simulating the electromagnetic propagation through the target and calculating the amount and distribution of the EM losses inside the imaged object A coustic module : this module is responsible for simulating the generation of thermoacoustic waves inside the target and its propagation from the target to the detector s In addition to that, a Multiphysics link is created to couple the EM losses results fr om the RF module to the acoustic module Electromagnetic waves simulation 3.3.1. Simulation geometry The model employs both 2D and a 3 D geometries that simulate the same experimental setup. The main components in the model are Open Ended waveguide: WR 340 D ban d waveguide with dimensions of 86.36 x 43.18mm It is modeled as a hollow box that filled with air and shielded with perfect electric conductor boundary condition. The waveguide is excited with a wave port that has an electric field parallel to the short s ide of the waveguide. Acrylic tanks: a c uboid box of acrylic with a wall thickness of 3mm is placed at the end of the waveguide and filled with safflower oil. The bottom part of the acrylic tank is shielded with perfect electric conductor to be similar to the experimental setup.

PAGE 43

31 Target: the target is placed directly above the acrylic layer and the end of the waveguide. All The simulation domain except the bottom of the tank (shielded with PEC) is terminated with a perfectly matched layer to reduce the s ize of simulation domain Figure 5 shows the 3D simulation geometry of the model To enhance the computation speed in some cases the 3D model is reduced to 2D model by taking a plane at Y =0 as shown in Figure 3 4 Figure 3 3 : 3D geometry of the EM simulation

PAGE 44

32 Figure 3 4 : C ross section plane through the 3D geometry 3.3.2. Simulation parameters The simulation employs tetrahedral meshes in case of 3D simulations and triangular meshes in case of 2D simulations to achieve an accurate characterization of the actual geometry shape. Two different mesh sizes are used in the EM simulations. The general simulation geometry use s variable mesh size with maximum element size equals t o 1 cm since the minimum wavelength equals to 12 cm The target is meshed with different mesh size to achieve more precise result for the electric field loss distribution inside it It is meshed with a resolution similar to that of the acoustic simulation (will be explained later) with maximum element size of 0.1mm. 3D meshing of the geometry can be seen in Figure 3 5 The simulated carrier Imaged target PML Excitation port Acrylic PEC

PAGE 45

33 frequency equals to 2.45 GHz and tar get electrical properties varies according to the simulation requirements. Figure 3 5 : M eshing of 3D geometry 3.3.3. A coustic propagation model The EM loss distribution inside the target is forwarded to the ac oustic module to act as a monopole acoustic source. The simulation geometry is reduced by taking only a finite area around the ta rget and use suitable boundary conditions to terminate the simulation domain. An exampl e can be seen in Figure 3 6 where the simulation domain is term inated with absorbing boundary condition (ABC)

PAGE 46

34 Figure 3 6 : Acoustic simulation region 3.3.4. Simulation parameters Acoustic simulation geometry is mesh ed with different mesh than that used in the EM simulation due to the difference in the simulated wavelength. The geometry is meshed with triangular mesh s tructures as showed in Figure 3 7 T he maximum frequency that we are interested in is 2 MHz Therefore the maximum element size is decide to be 0.1mm (7.5 samples per wavelength) Model validation The described model of thermoacoustic signal generation is validated by comparing the simulation results with experimental work by Mashal et.al [19] and simulation results by Deng and Golkowski [3] The model simulates the generation of thermoacoustic signals by a circular target with a radius of 6mm. The target is immersed in mineral oil and an acoustic transducer is align ed with the center line that pass through the target center. The imaged target material ABC Imaged target Coupling medium

PAGE 47

35 consists of ethylene glycol with different concentrations of air micro bubbles. Different concentrations of microbubbles produce different conductivities and sound spee ds inside the imaged object. Table 3 1 gives a detailed description about the change in materials properties with bubbles concentration. Figure 3 7 : Meshing of acoustic region Table 3 1 : Var iation of ethylene glycol properties with microbubbles concentration Solution number Bubbles concentration Average Effective conductivity (S/m) Sound speed Solution 1 0% 14.03 2.32 1660.7 Solution 2 20% 9.66 1.33 1729.8 Solution 3 30% 8.34 1.11 1805.6 Solution 4 35% 7.38 0.95 1922.6 Solution 5 40% 6.86 0.84 2049.4 Solution 6 ??% NA NA Na

PAGE 48

36 From Table 1 1 it can be seen that the material conductivity decrease s with the increase of the microbubbles concentration. A c omparison between the model and the experimental results is shown in Figure 3 8 .a and b respectively The model shows good agreement with the experimental results. The amplitude of the signal decreases with the increase of the mi crobubbles concentration (decrease of conductivity) and this normal due to the decrease of the electromagnetic losses inside the material. It also shows that the signal period is decreased with the increase of microbubbles concentration due to the increase of sound speed Figure 3 8 : C omparison between simulation and experimental results Figure 3 9 shows a comparison to the simulation results of [3] The signal amplitude and time per iod of the signal are changing in the same with a noticeable difference in the speed of signal decay. This difference could be related to the to the difference in the computational method and meshing techniques since th eir simulation employs FDTD method wi th square meshes. (a) (b)

PAGE 49

37 Figure 3 9 : Comparison with other simulation work Open ended waveguide vs horn antenna MITAI system requires an antenna that can handle high amount of power with high efficiency because the system needs to transfer very strong puls es (larger than 5KV) to the imaged target. The antenna should also have relatively narrow near field beam in order to concentrate the transmitted power inside the imaged object. Two option s are available to for the experiment. The first option is a standar d 20db horn antenna. The second option is WR340 open ended waveguide. A numerical model is created for the antennas to investigate the the field distribution and intensity at the vicinity of the antennas. The geom etry of horn antenna and open ended wavegui de can be shown in Figure 3 10 .a and b respectively In both cases the air is used as propagation medium and the simulation domain is terminated with a perfectly matched layer. The models are simulated with a frequency equal to 2.4 5 GHz and a variable mesh size with maximum element size equals to 10mm. Electric field intensity distribution at a distance equals to 3 mm from the antenna end is shown in Figure 3 10 .c and d The results show that the open e nded waveguide has much higher field intensity in the near field region (a) (b)

PAGE 50

38 while the horn antenna shows more homogenous distribution of the field but with lower intensity This is due to the distribution of the field over larger area in the case of horn anten na. The simulation also shows that the reflection coefficient of the open ended waveguide is much higher than that of the horn antenna. Under the same testing conditions, S 11= 12db for the open ended waveguide vs 22db for the horn antenna. Figure 3 10 : Electric f ield distribution for h orn and open ended WG antenna s (a) (b) (d) (d)

PAGE 51

39 C ase studies Different experimental parameters can affect t hermoacoustic signal intensity and characteristics. In this section, different sce narios are studied to check the effect of excitation signal and imaged object properties on the signal characteristics. 3.6.1. E ffect of frequency This study shows the effect of changing the frequency on the amplitude of the generated thermoacoustic signal. The r esults are shown in Figure 3 11 It shows that changing the frequency over a small range has negligible effect on the generated thermoacoustic signal. T his study is conducted over limited frequency range due to the waveguide bandwidth restrictions Increasing frequency could have a noticeable effect within high range as was reported by [8] 3.6.2. E ffect of material conductivity In this study, the change of mechanical properties of the material in Table 3 1 is neglected and only the conductivity is changed T he results are shown in Figure 3 12 It shows that increasing the conductivity within a certain range increases the amplitude of the generate d thermoacoustic signal s This effect is expected and it is a result of the increase of the dissipated losses by electromagnetic waves according to E quation ( 2 25 ) However, increasing the conductivity above certain threshold lim its the penetration ability of EM waves which will be shown later 3.6.3. E ffect of microwave pulse width Different microwave pulses were sent with different pulse widths and the simulation environment is kept the same. The gaussian pulse standard deviation is c hanged from 0.3 to 0.9 micro seconds. The results are shown in Figure 3 13 It shows t hat increasing the pulse period directly increase s the wave length of the generated thermoacoustic signal. This means

PAGE 52

40 that decreasing the microwave pulse width directly enhances the resolution of the thermoacoustic images. 3.6.4. O ptimal target a nalysis In this study, a rectangular target with a dimension of 25 by 10mm is used as the new target. T he permittivity and conductivity are changed over wide range of v alues as shown in Figure 3 14 a and b. The results represen t the integration of power loss inside the target for each permittivity and conductivity The power loss inside the sample increases with increasing the conductivity then starts to drop again. The increase is a results of the increase of losses in the medi um according to E quation ( 2 25 ) but after a certain threshold the EM wave lose its ability to propagate through the imaged object and the amount of delivered energy to the target drops again The optimal conductivity and permittivity is not constant because it depends on the targ et size and shape. Figure 3 11 : Effect of frequency on acoustic signal amplitude

PAGE 53

41 Figure 3 12 : E ffect of material conductivity of acoustic signal amplitude Figure 3 13 : Effect of MW pulse width of the wavelenght of acoustic signal

PAGE 54

42 (a) (b) Figure 3 14 : Effect of changing permittivity and conductivity over wire range of values

PAGE 55

43 Image reconstruction Image reconstruction is considered as one of the challenging tasks in thermoacoustic imaging. The data requires high amount of post processing in order to get the required image s Many algorithms are proposed to r econstruct the data and their performance var ies depending on the target material properties and acquisition geometry. The homogeneity of sound speed simplifies the reconstruction problem because it eliminates the need for any extra sound speed estimation tools while the electric properties are t he main engine behind the image contrast. For the acquisition system, the orientation of the transducers around the target and their size are crucial factors in determining the resolution of the imaging system This chapter discuss those challenges and their effect on the acquired images. K W ave K wave is an open source Matlab toolbox that is built for the simulations of acoustic waves [20] This toolbox solves the acoustic wave equation by employing a pseudo spectral time domain solver It has special modules for the simulation and reconstruction of photoacous tic imaging. It implements multiple reconstruction methods for different sensor geometries in 2D and 3D. As thermoacoustic a nd photoacoustic imaging follow the same concept, we are adopting this toolbox for the reconstruction of our simulated and experimen tal data. T his toolbox is adopted instead of doing the reconstruction by using COMSOL due the following reasons. The first reason is avoiding the inversion crime of using the same solver configurations for the forward and inverse simulations. The second is the increase of computation speed due to the use of pseudo spectral time domain ( PSTD ) since it requires only two samples per wavelength Finally the toolboxes are flexible and the ir functions can be easily modified to suit the experimental requirements.

PAGE 56

44 Reconstruction challenges As mentioned earlier, reconstruction of MITAI is not an easy task and there are many factors that need to be taken into consideration during the reconstruction process. These factor s vary according to the reconstruction algorithm but there are some common problem s that are shared between most them 4.2.1. H omogeneity of sound speed The homogeneity of sound speed in the medium plays an important role in simplifying the reconstruction problem. The reconstruction of the image of any object requires that the solution should vanish inside the space after a finite time T in order to have a known initial value (zero initial values) to start the reconstruction process This means that all the acoustic waves should leave the imaged target after ti me T. This kind of mediums is known as no n trapping medium because it allows all the waves to leave after a certain time and no wave is trapped inside the medium The inhomogeneity of sound speed highly eff ects the reconstruction of the object and the est imation sound speed becomes a necessity in some cases when there are strong abrupt changes in sound the speed 4.2.2. Estimation of sound speed Some cases require estimation of sound speed if there is large variation in the medium sound speed. Some algorithms ar e proposed to estimate it from the acquired thermoacoustic signals but most of them are instable or only work s under strict conditions [9] The current best method to estimate sound is to use pulse eco mode data to enhance the thermoacoustic images in the case of acoustically inhomogeneous mediums.

PAGE 57

45 4.2.3. Size o f transducer The s ize of transducer is required to be as small as possible because it controls the resolution of the reconstructed image. Most of the r econstruction algorithms assume point like Omni directional transducer s but real transducers are differen t. The real transducer has finite size and the recorded signal represents an integration of orthogonal components of the incident waves on the transducer surface and i t is given by ( 4 1 ) Where S, dA are the incident signal on the transducer and the surface unit area i .e. the s represent the sum of all the signal that are crossing a specific area. This l imits the resolution of the acquired images and make s it dependent on the transducer size and properties. The effect can reduced by either us ing a special transducers that have thin and long structure or d e convolute the shape and size of the transducer w i th the reconstructed image [21] 4.2.4. Effect of tra nsducer orientation Ideal reconstruction of thermoacoustic signals requires that the target is surrounded by 360 degrees of transducers [9] This is an expensive approach and difficult to implement in practical situation s In case of homogenous mediums, the rule states that in order to completely reconstr uct a point, then any line that pass es through that point should intersect with the transducer s plane at least one time [9] This rule becomes more complicated in the case of inhomogeneous mediums because the waves are deflected due to medium inhomogeneity Figure 4 1 shows the reconstruction results with different transducers orientations. The simulated geometry is shown in a while b and c show the reconstruction resu lts with a line and L shape array s The results show that the edges that are normal to the trans ducers plane are

PAGE 58

46 reconstructed clearly while the other edges are blurred and this is the reason behind the enhancements in c Figure 4 1 : Effect transducer orientation on image reconstruction I mage reconst ruction with time reversal C urrent reconstruction methods neglect the heterogeneities in electric field and sound speed and assume that both of them are distributed uniformly inside the imaged object. Different methods are proposed for the reconstructions like filtered back projection, Eigenfunction expansion method and time reversal methods. This project implements the time reversal due to its robustness, flexibility [22] Time reversal is a robust reconstruction metho d that exploit s the reciprocity property of the wave equation. According to Huygens principles, for any initial source that has a bounded support, there is a finite time for the wave to leave the domain [22] According to this theory a solution P vanishes inside the domain after a finite amount of time T. After time T, a zero initial value can be imposed to the domain and the wave can be retransmitted to the (a) (b) (c) Transducers Transducers

PAGE 59

47 domain in rever s ed timing sequence to reconstruct the source [22] Below are the required and by assuming that the source is a Dirac delta function equation ( 2 28 ) can be solved as bellow [23] ( 4 2 ) ( 4 3 ) ( 4 4 ) are the medium G the microwave pulse and amplitude of microwave pulse respectively By exploiting the reciprocit y of G reen functions the solution can be written as a convolution between the G reen function of the medium and the wave source. ( 4 5 ) ( 4 6 ) Where is the induced pressure from A that is enclosed in the sphere As go to zero can be given by [23] ( 4 7 ) ( 4 8 )

PAGE 60

48 The goal of the reconstruction is to get the function A from the pressure p r The retrans miss ion of the wave again in to the media with the same received pressure and boundary conditions results in the reconstruction of the initial s ource I.e., the detected pressure is reconvoluted with the medium ( 4 9 ) is the self correlation function of the medium. Simulation study of image reconstruction with time rever sal T hree biological samples with differen t shapes are placed in safflower oil to study MITAI system as shown in Figure 4 2 .a T he sample s are a circle with radius of 1mm, square with side length of 2mm and a rectangle with dime nsions of 1.5 by 3 mm. The forward model employs the same experimental setup mentioned in chapter 3 The model is excited with a pulse that has a carrier frequency of 2.45 GHz and a gaussian shape with a standard deviation of 40 nanoseconds. All the samples are set have the same electrical and acoustical properties as explained in Table 4 1 The acoustic transducers are represented by an array of point like transducers that are placed at the top of th e simulation domain. Figure 4 2 .b shows a semi uniform electromagnetic loss distribution inside the samples which indicates that the samples regions can be reconstructed evenly. The same forward simulation model is recreated by using K wave toolbox and the signal s are retransmitted in a reverse time sequence. Time reversal results are shown in Figure 4 2 .c and a thresholded version of it is shown in Figure 4 2 .d. Th e position s and sizes of the objects are reconstructed correctly The case is the same for the edges that are normal to the transducer plane while we can see that the horizontal edges are blurred due to the orientation of the detection plane.

PAGE 61

49 Table 4 1 : Properties of simulated geometry Medium property V alue Electrical conductivity 0.4 Relative permittivity 9 Relative permeability 1 Sound speed 1537 Material density 1041 Heat capacity 3510 Coefficient of thermal ex pansion 3e 4 Figure 4 2 : Time reversal reconstruction of simulation data (a) (b) (c) (d)

PAGE 62

50 4.4.1. Effect of MW pulse width The simulation is repeated with different MW pulse widths with a standard deviation of 0.6, 0.8, 1.2 and 4 microseconds. The results are shown in Figure 4 3 a, b, c and d consequently. The results shows that the objects edges are blurred with increasing the pulse width due to the decrease of the excited acoustic frequency. Figure 4 3 : Reconstruction with different MW pulse widths

PAGE 63

51 4.4.2. Biological samples embedded in a conductive medium In this case the same sample shapes were imbedded inside another biological tissue that has lower conductivity and diff erent sound speed. The experiment is conducted with same conditions as in 4.4.1 and the materials properties are given in Table 4 2 The results are shown in Figure 4 4 The embedded biological samples can be easily identified from other biological tissues in the background, but they are blurred due to the variations in sound speed in the medium. Figure 4 4 : Reconstructio n of samples that are embedded in another conductive medium (a) (b) (c) (d) Material 2 Material 1

PAGE 64

52 Table 4 2 : Properties simulated materials Medium property Material 1 Material 2 Electrical conductivity 0.4 0.749 Relative permittivity 9 5 Relat ive permeability 1 1 Sound speed 1537 1580 Material density 1041 1041 Heat capacity 3510 3510 Coefficient of thermal expansion 3e 4 3e 4 Reconstruction of experimental data The same time reversal algorithm is used for the reconstruction of experiment al data. A geometry similar to the experimental one is created in K wave with PML boundary conditions. The simulation sound speed is set to be equal to the average sound speed in safflower oil and the imaged sample and the recorded data is interpolated to match the time step of the simulation. The measured transducer signal is filtered by median filter with order of 20 to reduce the noise effect on the reconstructed signal. 4.5.1. Uniform piece of meat A uniform piece of pork meat is used as test sample target. T h is meat type is chosen due to its availability and relatively good conductivity. F igure 4 5 .a, b, c, d show the imaged meat sample, ultrasound images, and thermoacoustic images before and after time reversal respectively The results show that the meat boundaries are reconstructed correctly when it is compared to the ultrasound images

PAGE 65

53 F igure 4 5 : Imaging of a u niform piece of mea t 4.5.2. Rectangular biological tissue with copper wire In this test, a 4 mm diameter copper wire shown in Figure 4 6 .a is inserted into the mea t sample as shown in Figure 4 6 .b and the imaging process is repeated again. The reconstructed thermoacoustic image in Figure 4 6 .d and e show th at the upper boundary of meat is reconstructed correctly. The copper wire effect can be seen as a dark spot in the thermoacoustic image. Moreover, t he figure shows that the lateral details are blurred due to the orientation of the transducer s (a) (b) (c) (d)

PAGE 66

54 Figure 4 6 : Imagin g of a u niform piece meat with copper wire (b) (a) (c) (d) (e)

PAGE 67

55 4.5.3. The effect of inserting highly conductive tissue This experiment shows the effect of inserting highly conductive (electrically) tissue into the piece of meat. The new tissue is inserted as shown in Figure 4 7 a. The results are shown in b c, d The results show that the highly conductive tissue produces no acoustic signals because most of the EM are reflect and they are not able to penetrate the tissue. Figure 4 7 : Effect of using highly conductive tissue (a) (b) (c) (d)

PAGE 68

56 4.5.4. Two separate pieces of meat In this experiment we are using two separate pieces of the of the same meat type as shown in Figure 4 8 .a The results are shown in c, d, e in the same sequence as before. We can see that we have good separation between the two pieces but the side edges are still blurred due to the orientation of the transducer plane. Figure 4 8 : Imaging of two separ ate pieces of meat In all our experimental images, a wavy pattern appears inside the imaged samples as shown in Figure 4 8 .d. This phenomenon is under investigation and the early indications refers to electrical standing waves inside the meat samples due to the mismatching with its (a) (b) (c) (d) Wavy pattern s

PAGE 69

57 boundary. More work is being done to investigate this phenomenon and enhance the electrical matching between the sample and it background.

PAGE 70

58 Pipe internal surface reconstruction Pipe inte rnal surface reconstruction employs the structured light concept with a modified triangulation for circular geometries. This chapter presents a brief introduction to the laser scanner setup and develops 3D reconstruction algorithm for defects characterizat ion Scanner setup Two scanners are developed with each one has different camera and laser source o rientation In the first setup a laser source that projects a thin laser ring is aligned in parallel with the pipe main axes as shown in Figure 5 1 A small camera is placed beside the laser source to record the laser ring deformations along the pipes internal wall. The second setup employs the same concept but the camera and the projector are placed in opposite directions as shown in Figure 5 2 A p reliminary setup shown in Figure 5 3 employs a simple endoscopic camera due to its small diameter The experiment s on this cameras have proved that this type limited view angle that cause large shadow area in the acquired images Therefore, the endoscopic camera is changed with a fisheye 180 degree camera in order to capture the whole scene as shown in Figure 5 4 The figure shows a two 3D printed holders where a and b represents a direct implementation of setup one and two respectively. Figure 5 5 .a and b shows a comparison between the images that are acquired by endoscopic and fisheye camera respectively With the fisheye camera, we are able to minimize the amount of the shadow that was caused by the setup

PAGE 71

59 Figure 5 1 : L aser source and camera orientation in setup 1 Figure 5 2 : L aser source and camera orientation in setup 2 Pipe axis Source Camera Source Camera

PAGE 72

60 Figure 5 3 : Initial setup with endoscopic camera Figure 5 4 : LEAP experimental se tups with fisheye camera Camera Laser source (a) (b) C amera Laser source

PAGE 73

61 Figure 5 5 : Comparison between images of endoscopic and fisheye cameras (a) (b)

PAGE 74

62 Extraction of 3D information The main task in this process is the segmentation of the laser ring from the re corded camera frames. Figure 5 6 shows a flowchart of the main steps in the segmentation and extraction of camera frames. The algorithm can be briefied by the following steps: Extract the best layer of the RGB frame One layer is chosen according to the color of the projected ring color and the color of the pipes internal walls. For example if a white pipe and red laser ring is used t he experimental tests showed that green layer has better contrast and less amount of noise than the other two layers Figure 5 7 shows a comparison between the red and green layers of a single frame. Image de noising In some cases, the recorded camera images contain high amount of noise due the low amount of illumination inside the pipe. In order to reduce the noise, a Gaussia n low pass filter is applied to the image to reduce the high frequency components. Power law transformation This transformation is used to enhance the contrast of the denoised image. It increases the intensity of high values in the image and separate them from low values. The transformation is applied according to the following formula: ( 5 1 ) Where and are the input and output images. L is the value of the power that is used to transform the image. Figure 5 8 .a [24] exp lains the effect of using different values of L on the image values. In our case, we are using positive value for L to increase the contrast of the large values in the image. The result of using power law transformation is shown in Figure 5 8 .b.

PAGE 75

63 The result shows that most of the low values are suppressed and the ring has better contrast from the background. Figure 5 6 : Flowchart of the 3D reconstruction

PAGE 76

64 Figure 5 7 : C omparison between red and green layers Figure 5 8 : Power law transformation Where and are the input and output images. L is the value of the power t hat is used t o transform the image. Figure 5 8 .a [24] explains the effect of using different values of L on the image values In our case, we are using positive value for L to increase the contrast of the large values in the image. The result of using power law transformation is shown in Figure 5 8 .b. (a) (b) (a) (b)

PAGE 77

65 T he result shows that most of the low values are suppressed and there is an improvement in the contrast of the final image. Thresholding Thresholding is used to divide a certain data set into two different regions according to their intensity Th resholding is applied to separate large value s of the ring region from other image regions. The value of the threshold is decided by doing a quick analysis to the first frame of the recorded video. In this process the greyscale image is converted to a bina ry image where the high values are set to one and the low values are set to zero. Line extraction As explained before the laser ring has a finite thickness, therefore, single pixel line in the image. The options to introduce a single pixel line from the ring are either looking for the maxima of the line or finding the middle point in the line. As finding the maxima is highly effected by the light distortions we opted to go wit h finding the middle point of the line. A Morphological l in e thinning is used to find the center of the laser ring and the results can be shown in Figure 5 9 In some cases, the thinning process results in high amount of spurs. Therefore, i n order to guaranty a clean and smooth ring reconstruction, a spurs cleaning process is implemented to remove any unwanted spurs. 3D point cloud registration After extr acting the locations of the ring pixels in the image, those locations are mapped to their real locations in the real plane by applying the triangulation process Moreover, by assuming a constant scanning speed, the frames are positioned in the cloud as equ ally spaced 3D rings along the scanning line The spacing between the rings is calculated according to the speed of moving the camera setup.

PAGE 78

66 Figure 5 9 : Laser ring after applying thinning process Experimental results The proposed algorithm were applied to different experimental setups. At the beginning the algorithm were applied to a part of laser scanning video of lar ge parameter pipe. The video is available from maverickinspection.com [25] The results are shown in Figure 5 10 and the algorithm was able to characterize the surface correctly After that the algorithm were applied to the scanning videos from our LEAP sensors that were shown before Currently the scanners are dealing with artificial pipe damages (holes and metal screws) that were created in the laboratory as shown Figure 5 11 The results from the initial setup with the endoscopic camera

PAGE 79

67 are shown in Figure 5 12 and both the screws and holes appears like holes in the pipes surface. This is because of the strong reflection from the surface of the screws. Another scanning result is shown in Figure 5 13 by using the second model with the fisheye camera. Figure 5 10 : 3D reconstruction from commercial scanner Surface deformation

PAGE 80

68 Figure 5 11 : Artificial damages Figure 5 12 : 3D reconstru ction with endoscopic camera

PAGE 81

69 Figure 5 13 : S urface reconstruction with fisheye camera ( setup 2) Resolution and accuracy of s tructured light reconstruction The accuracy and dens ity of the 3 D reconstruction in structured light system is governed by multiple factors The first factor is the resolution of the acquired image. The size of the image pixel in real plane define s the maximum resolution that the system can achieve. However, t his factor is also restri cted by the processing speed of the acquisition device because larger images require higher processing time which means less number of processed frames per second and higher acquisition time. The second factor is the width of the acquired edges. In the ca se of laser scanners, it is represent ed by the thickness of the projected laser ring that is applied to the imaged object. Decreasing the thickness of the laser ring increase s the resolution but also decreases the ring detectability Another important fac tor is the color distortion from the pipes walls. For example, red walls that are scanned with red laser scanner

PAGE 82

70 suffers from high amount of distortion due to the wall color similarity with the projected ring. In order to overcome this issue, multicolor p atterns are recommended for those surfaces to increase the probability of detecting the ring despite the variations in surface color

PAGE 83

71 Conclusion and future work Nondestructive evaluation is an important field in the industry and the demand to develop new techniques is growing due to introduction of new materials and the need for more precise detection methods. In this study, we have introduced two NDE methods T he first one is proposed for the detection of breast cancers while the other one is de dicated fo r the defects detection in gas pipelines. Thermoacoustic is a hybrid imaging methodology that combines the advantages of microwave imaging and ultrasonography. A numerical model is created to simulate the signal generation of thermoacoustic imaging systems The model shows that the acquired thermoacoustic images represent the microwave loss distribution inside the image d target. The detected signal level is increased by increasing the target conductivity until it reaches a certain threshold when the MW sign als starts to lo se the ability to penetrate through the target due to its high conductivity. In same manner increasing the frequency increase the signal level due to the increase of losses with increasing the frequency but this however decrease s the penet ration ability of the MW signal. The study also shows that the MW pulse period has direct effect on the frequency of the generated acoustic signals. I.e. decreasing the pulse width increas es the frequency of acoustic wave and results in high resolution ima ges. Time reversal reconstruction method is used for the reconstruction thermoacoustic images due to its robustness The method exploit s the reciprocity of the wave equation by retransmitting the detected acoustic signals back into the medium with a revers e time sequence to reconstruct the acoustic source The method is implemented successfully to reconstruct both simulation and experimental data. For the simulation, different scenarios were implemented for the target shapes and transducers orientation The study shows that in order to reconstruct

PAGE 84

72 a point then any line passes though th at point should intersect with the transducers plane. For the experimental data time reversal is implemented after applying filtering to reduce the effect of noise. Different biological samples were tested with different shapes and conductivities. The boundaries of the sample s were clearly reconstructed but the internal details were blurred due to the large transducer size and the high amount of noise in the experimental measu rements Structured light based reconstruction is an effective 3D characterization method that employs the triangulation concept to acquire the depth from 2D images In this study we have implemented a laser ring based structured light system to reconstru ct the defects inside small gas pipelines. T he resolution of this system is governed by the camera resolution and the thickness of laser ring and the image is highly effected by the pipe color F uture work For the thermoacoustic system, the next step is t o enhance the signal to noise ratio of the current experiment and develop a tomographic system that is capable giving 3D characterization of the image object. The method is also proposed for the testing of composite materials due to their conductive nature For structured light based system, a multi colors multi rings system is under development. The speed the segmentation process can be enhanced by using parallel processing and more efficient segmentation methods. The misalignment of the sensor with pipe c an be eliminated by using an inertial measurement unit to give the real orientation of the sensor and the 3D registration can be done automatically with ICP.

PAGE 85

73 R EFERECES [1] Experimental study of therm oacousti imaigng system [2] 937 940, 2012. [3] canc er: A model J. Appl. Phys. vol. 111, no. 7, pp. 7 10, 2012. [4] https://www.osfhealthcare.org/media/filer_public/c4/ee/c4ee865d e6ce 4f85 b39d 16858a6adeab/breastfeeding_tips_1 7_english.pdf. [Accessed: 27 Mar 2016]. [5] http://www.cancer.org/cancer/breastcancer/detailedguide/breast cancer what is breast cancer. [6] http://www.cancer.org/cancer/breastcancer/detailedguide/breast cancer breast cancer types. [7] Thermoacoustic applications in breast cancer" University of Arizona, vol. 225, no. 3, pp. 61 78, 2014. [8] Induced Thermoacoustic IEEE Trans. Biomed. Eng. vol. 06, no. 01, p. 1350001, 2012. [9] O. Scherzer, Handbook of Mathematical Methods in Imaging [10] Wikipedia [11] http://www.army technology.com/features/featureied detection ultra wideband microwave radar darpa jied do/. [12] T. Qin, X. Wang, H. Meng, Y. Qin, B. Webb, G. Wan, R. S. Witte, and H. Xin, IEEE Antennas Propag. Soc. AP S Int. Symp. pp. 1917 1918, 2014. [13] https://en.wikipedia.org/wiki/Guided_wave_testing#/media/File:Guided_wave_testing _GWT.jpg. [14] job hiring 60 natural gas pipeline inspectors cali. [15]

PAGE 86

74 [16] D. Voltmer, Fundamentals of Electromagnetics vol. 2, no. 1. 2007. [17] D. Pozar, Microwave Engineering Fourth Edition 2005. [18] Lect. Notes no. January, pp. 49 56, 2013. [19] enhanced microwave induced thermoacoustic imaging of breast cancer: an experimental study of the effects Phys. Med. Biol. vol. 54, no. 3, pp. 641 650, 2009. [20] Wave: MATLAB toolbox for the simulation and J. Biomed. Opt. vol. 15, no. 2, p. 021314, 2010. [21] ction algorithm for IEEE Trans. Med. Imaging vol. 28, no. 11, pp. 1727 1735, 2009. [22] thermoacoustic tomography in acoustic Inverse Probl. vol. 24, no. 5, p. 55006, 2008. [23] technique for microwave induced thermo J. Electromag n. Waves Appl. vol. 22, no. 16, pp. 2191 2204, 2008. [24] R. C. Gonzalez and R. E. Woods, Digital Image Processing (3rd Edition) 2007. [25] www.youtube.com/watch?v=KEspk15LIv0.