University of Southern California Dissertations and Theses

Array transducers for high frequency ultrasound imaging

(USC Thesis Other)

Array transducers for high frequency ultrasound imaging

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ARRAY TRANSDUCERS FOR HIGH FREQUENCY ULTRASOUND IMAGING

by

Hyung Ham Kim

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOMEDICAL ENGINEERING)

May 2010

Copyright 2010 Hyung Ham Kim

DEDICATION

TO MY PARENTS, MRS. YANGGYOON JEON AND MR. KYUNGSOOK KIM
ii

ACKNOWLEDGMENTS
It was at a dinner table with my team seniors at a restaurant in Shanghai, China. It goes
back to one day in fall 2003. We were energetic, passionate, and reckless. I was the youngest and
wanted to break the status quo. I realized that I would repeat the same cycle of thoughts if I would
not stop my career in the medical imaging industry and return to school to learn ultrasonic
transducers. Now, in 2010, I can recall that it was one of my life changing moments and I thank
God for guiding me all the way to here today.
I would like to thank my advisor, Dr. K. Kirk Shung for his support and guidance not
only in my research but also in my life. He trusted me from when I first met him with my resume
in his office at Pennsylvania in 2001 and encouraged me with an exciting comment, “out-of-box
thinking” when I started my PhD research project, high frequency convex arrays. My special
thanks to Dr. Jonathan Cannata for his continued support and teaching during my graduate studies.
He helped me identify what are really important in transducer engineering and taught me with all
possible details. I thank many people who helped my research in different ways. Mr. Jay
Williams offered me clear answers all the time when I had any questions on array fabrication. Dr.
Jesse Yen provided me valuable comments that made my dissertation stronger and Dr. Ellis Meng
and Dr. Eun Sok Kim showed their great interests on the positive impact of this research to high
frequency ultrasound imaging.
It is always my pleasure to work with good people. I thank all of my colleagues in NIH
Resource Center of Medical Ultrasonic Transducer Technology. Morning coffee meetings with
Dr. Jin Ho Chang and Dr. Jungwoo Lee were a fountain of our joy, concern and new ideas. I also
iii

iv

thank Dr. Jong-Seob Jeong, Mr. Jinhyoung Park and Mr. Changyang Lee for their help to finish
my convex array project. Dr. Qifa Zhou and Dr. Chang-Hong Hu deserve my thanks for sharing
productive ideas for medical ultrasound and scientific research.
When I made a decision to return to school, Mr. Sung Hee Park was a strong supporter
from the beginning. During my graduate school days, for a senior student looking for a good
research topic, he always pointed out that I should do something practical, something can be led
to a real product. He was one of team seniors at that dinner table in Shanghai, China. I wish him
all the best.
I could not finish this single piece of accomplishment without my family’s support and
love. I wish to express my deepest thanks to my wife, Jeein and our children, Seaok (Diana),
Seaon (Karen), Jaehyun (Daniel) and one more to come in November 2010. Jeein and I have
known each other now for more than 20 years. She is my friend, my advisor and my life. Each of
children has made sacrifices in her or his own way and they are our joy all the time. I also thank
Jeein’s parents, Mrs. Taeim Shin and Mr. Sangyeol Jeong for their unconditional loving support.
Dr. Harrison Kim, my brother and Mrs. Minju Kim, my sister deserve my special thanks for their
love and support.
Lastly, I want to dedicate my dissertation to my mother, Mrs. Yanggyoon Jeon and my
father, Mr. Kyungsook Kim with all my respect and love. Their lives showed me the value of
honesty, integrity and diligence. I wish my achievement would be a bit of help for my mom to get
over her illness. “Mom, you will be fine. You are strong and God is always with you.”

TABLE OF CONTENTS
Dedication ii
Acknowledgments iii
List of Tables vii
List of Figures viii
Abstract xiv
CHAPTER 1 Introduction 1
1.1 Medical Ultrasound 1
1.2 High Frequency Ultrasound Imaging 5
1.3 High Frequency Ultrasound Ophthalmic Imaging 8
1.4 High Frequency Ultrasound Transducers 9
1.5 Objective of Research 11
CHAPTER 2 High Frequency Dual Element Transducers 13
2.1 Dual Element Transducers for Harmonic Imaging 13
2.2 Design of Dual Element Transducers 14
2.2.1 Transducer Geometry and Structure 14
2.2.2 KLM Modeling 17
2.3 Fabrication of Dual Element Transducers 18
2.4 Performance of Dual Element Transducers 19
2.5 Clinical Evaluation of Dual Element Transducers 29
2.6 Advantages of Dual Element Transducers for Harmonic Imaging 33
2.7 Extended Applications of Dual Element Transducers 33
v

vi

CHAPTER 3 High Frequency Convex Array Transducers 37
3.1 Convex Array Transducers 37
3.2 Design of Convex Arrays 40
3.2.1 Array Geometry 40
3.2.2 Elevation Focusing 49
3.2.3 1-3 Composite 49
3.2.4 Array Structure 55
3.2.5 KLM Modeling 55
3.2.6 Flexible Circuits 57
3.3 Fabrication of Convex Arrays 59
3.3.1 Fabrication Options 59
3.3.2 1-3 Composites by Interdigital Pair Bonding 65
3.3.3 Matching Layers and Backing Blocks 69
3.3.4 Bonding and Conforming Layers 70
3.3.5 Finished Arrays 72
3.4 Characterization of Convex Arrays 74
3.4.1 Quantitative Testing of Convex Arrays 74
3.4.2 Electrical Impedance 75
3.4.3 Pulse-Echo Response 76
3.4.4 Crosstalk 80
3.4.5 Insertion Loss 81
3.4.6 Images 81
3.5 Advantages of Convex Arrays 90
CHAPTER 4 Summary and Future Work 91
4.1 Summary 91
4.2 Future Work 92
Bibliography 94

LIST OF TABLES
Table 2.1. Design Parameters for Dual Element Transducers 15
Table 3.1. Features Comparison of Single Element Transducers and Convex Arrays 40
Table 3.2. Array Geometry and Acoustic Parameters of 20 MHz Convex Array
Transducers 45
Table 3.3. Properties of Piezoelectric Materials 50
Table 3.4. Properties of Composites 53
Table 3.5. Design Parameters of 20 MHz Convex Array Transducers 56
Table 3.6. Summary of Pulse-Echo Test Results of a 20 MHz Convex Array Transducer 78
vii

LIST OF FIGURES
Figure 1.1. 2D/Color/Pulsed Wave Doppler combined mode image of the umbilical cord
(courtesy of Medison Co., Ltd.). 2
Figure 1.2. Sonosite S-Nerve
™
, a mountable, zero-footprint ultrasound system used in the
anesthesia room (courtesy of SonoSite, Inc. and Duke University Medical Center). 3
Figure 1.3. GE Healthcare Vscan™, a pocket size handheld ultrasound imaging system
(courtesy of GE Healthcare, Inc.). 4
Figure 1.4. Real-time display of thermal (TI) and mechanical (MI) indices at the upper
right side on the screen (Courtesy of Philips Healthcare, Inc.). 5
Figure 1.5. Clinical images of the eye: (a) choroidal tumor and retinal attachment, (b)
retinal detachment, (c) iris crypt, (d) accommodative lens movement (Courtesy of
Sonomed, Inc.) 9
Figure 1.6. VisualSonics Vevo
®
2100 Imaging System: (a) an imaging system, (b) a 30
MHz linear array, (c) an image of vasculature and lobes of the mouse liver (Courtesy of
VisualSonics, Inc.) (VisualSonics, 2010) 11
Figure 2.1. Dual element transducers: (a) photograph of finished device, (b) cross-
sectional drawing of the transducer (not to scale) 16
Figure 2.2. Transmit characteristic of the 20 MHz outer ring of the dual element
transducer (Option 2: D
out
= 12 mm): (a) a measured waveform by a hydrophone at the
focal depth, (b) its spectrum magnitude. 20
viii

Figure 2.3. Transmit characteristic of the 40 MHz single element source transducer: (a) a
measured waveform by a hydrophone at the focal depth of the source transducer (z = 12
mm), (b) its spectrum magnitude (‘at z = 12 mm’), attenuation curve vs. frequency in the
water at the distance of 30 mm (‘attenuation’) and attenuated spectrum at the surface of
receive element, z = 42 mm (‘at z = 42 mm’). 22
Figure 2.4. Receive characteristic of the 40 MHz inner circular element of the dual
element transducer: (a) a received waveform by the receive element, (b) an attenuated
spectrum of the signal from the source transducer and a spectrum of the received signal
of receive element. 23
Figure 2.5. Pulse-echo test results using the 20MHz outer ring element for transmission
and reception (Option 2: D
out
= 12 mm): (a) an echo waveform from the soft silicone
rubber target at the focal depth, (b) its spectrum magnitude. 25
Figure 2.6. Pulse-echo test results for the dual element transducer using the 20 MHz outer
ring element for transmission and the 40 MHz inner circular element for reception
(Option 2: D
out
= 12 mm): (a) an echo waveform from the soft silicone rubber target at
the focal depth, (b) its spectrum magnitude. 26
Figure 2.7. Lateral beam profiles obtained by the transmit waveforms at the focus from
the 20 MHz outer ring of the dual element transducer (Option 1: D
out
= 10 mm) and the
20 MHz circular shape single element transducer, measured by a hydrophone and
simulated by Field II. 27
Figure 2.8. Lateral beam profiles obtained by the transmit waveforms at the focus of dual
element transducers for different outer diameters of the 20 MHz outer ring element
(Option 1: D
out
= 10 mm vs. Option 2: D
out
= 12 mm). 28
Figure 2.9. Schlieren images of 20 MHz circular and ring shape elements (Option 2: D
out

= 12 mm), uncompressed and linearly mapped to a 256-level gray scale: (a) circular, at
the focal depth, (b) ring, at the focal depth. 29
ix

Figure 2.10. Images of the posterior segment of excised pig eye, 3.0 mm × 1.5 mm in the
vicinity of the focal depth of 30 mm, logarithmically compressed to a dynamic range of
60 dB and linearly mapped to a 256-level gray scale: (a) fundamental imaging using the
single element transducer, (b) harmonic imaging using the dual element transducer
(Option 2: D
out
= 12 mm). 30
Figure 2.11. Images of a choroidal nevus of the human eye: (a) images by SLO and OCT,
(b) fundamental imaging using the single element transducer, (c) harmonic imaging using
the dual element transducer (Option 1: D
out
= 10 mm; thin arrow: greater shadowing by
the lesion, thick arrow: improved depiction of retina/sclera border, ON: optic nerve),
logarithmically compressed to a dynamic range of 30 dB and linearly mapped to a 256-
level gray scale 32
Figure 2.12. Image of an excised pig eye obtained by a 20 MHz/40 MHz dual element
transducer of an excised pig eye with frequency compounding of four sub-band signals
from Chang et al. (Chang, Kim, Lee, & Shung, 2010) 34
Figure 2.13. Photoacoustic imaging system setup using a 20 MHz receive, ring shape
ultrasonic transducer and a 532-nm wavelength laser transmitter from Kong et al. (Kong,
et al., 2009) 35
Figure 2.14. Images of a ciliary body of an excised pig eye. Pulse-echo (left) and
photoacoustic (right) images. The B-scans were made in the plane perpendicular to the
ciliary processes. The photoacoustic images reveal individual processes with high
resolution and clarity not obtainable with the pulse-echo 20 MHz ultrasound from Kong
et al. (Kong, et al., 2009) 36
Figure 3.1. Convex arrays for ophthalmic imaging: (a) a convex array placed over sclera
of the human eye (Wikipedia, 2010), (b) a trapezoidal imaging plane with the focal zone
at the posterior segment of the eye 41
Figure 3.2. Two-way curved aperture of 20 MHz 192 element convex array transducers
created for Field II simulation 43
Figure 3.3. 64 channel, 2-way planar beam profile of 20 MHz convex array transducers
by Field II simulation 43
x

Figure 3.4. 64 channel, 2-way lateral beam profile of 20 MHz convex array transducers
by Field II simulation 44
Figure 3.5. 64 channel, 2-way axial beam profile of 20 MHz convex array transducers by
Field II simulation 44
Figure 3.6. Wire phantom images simulated by Field II: (a) 20 MHz 192 element linear
array, (b) 20 MHz 192 element convex array 46
Figure 3.7. H&E stain image of a dog eye processed to the gray scale with the
corresponding scattering value for Field II simulation 47
Figure 3.8. Simulated images of the dog eye by Field II with (a) the linear array, (b) the
convex array 48
Figure 3.9. Measured impedance of 1-3 composite test pieces of (a) TRS HK1-HD
(dimension: 2.0 mm × 1.0 mm × 0.075 mm) with 14 µm kerf, (b) TFT L-155N
(dimension: 2.0 mm × 1.0 mm × 0.075 mm) with 7 µm kerf (both resonating in the air,
magnitude: solid line, phase: dashed line) 52
Figure 3.10. Design of 1-3 composites: (a) Option 1 for double index dicing, (b) Option 2
for interdigital pair bonding, (c) Option 3 for interdigital pair bonding with a wider dicing
blade (length in µm) 54
Figure 3.11. The KLM pulse-echo response (solid line) and its spectrum (dashed line) for
a single 20 MHz array element. 56
Figure 3.12. A group of traces to generate laser slits in the flexible circuit to serve as a
strain relief and ground patterns in the final assembly for 20MHz convex arrays. 58
Figure 3.13. “Bending and bonding” fabrication option: (a) conformed layers to be
bonded, (b) a composite conforming structure, (c) a conformed composite. 60
xi

Figure 3.14. “Backbone” fabrication option: Pre-bonded and diced composite / flex
circuit / backplate layers to be conformed by a conforming fixture. 62
Figure 3.15. “Side attachment” fabrication option: Half-cut flex circuit to be bonded on
the sputtered, pre-diced and conformed backplate. 63
Figure 3.16. “Bonding and bending” fabrication option: Pre-bonded matching / composite
/ flex circuit layers to be conformed. 64
Figure 3.17. Interdigital pair bonding technique to fabricate 1-3 composites (W
c
: ceramic
dicing width, W
p
: polymer dicing width, W
k
=(W
p
-W
c
)/2: kerf width) from Liu et al. (Liu,
Harasiewicz, & Foster, 2001) 66
Figure 3.18. Pictures of finished 1-3 composites: (a) SEM image of Option 1: double-
index dicing, (b) microscope image of Option 3: interdigital pair bonding 67
Figure 3.19. Micro-cracks at grain boundaries: (a) from (Nix, Corbett, Sweet, & Ponting,
2005), (b) Option 1 68
Figure 3.20. Process to make a backing block: (a) an aluminum positive surrounded by
dams, (b) degassed RTV silicone rubber poured in the space, (c) a finished RTV mold,
(d) a finished backing block 69
Figure 3.21. Picture of a composite bonded on the flex circuit, signal traces of the flex
circuit aligned with an array of elements of the composite 71
Figure 3.22. Conforming and bonding of matching layer/composite/flex circuit onto a
backing block 72
Figure 3.23. A picture of the finished 20 MHz convex array successfully built to create
two-way curved aperture 73
xii

xiii

Figure 3.24. Stainless steel curved target for pulse-echo test of convex arrays with the
radius of curvature of 53.6 mm to place the reflector at the focal depth 75
Figure 3.25. Measured impedance magnitude and phase of an element of a finished 20
MHz convex array (Option 2) 76
Figure 3.26. Pulse-echo test results of an element of a finished 20 MHz convex array
(Option 2): (a) an echo waveform at the focus, (b) its spectrum 78
Figure 3.27. Pulse-echo test results for all 192 elements of 20 MHz convex array
transducers: Loop sensitivity [dB], -6 dB center frequency [MHz], -6 dB bandwidth
[MHz], -6 dB fractional bandwidth [%], -20 dB pulse width [µs], time of flight [µs] vs.
element number 79
Figure 3.28. A simplified drawing of a wire phantom used for synthetic imaging
experiments 83
Figure 3.29. Synthetic aperture images of a wire phantom by (a) measured RF data and
(b) Field II simulation. The dynamic range of the image is 40 dB and the display uses a
linear gray scale for mapping. 84
Figure 3.30. Line spread functions for the center wire of the image, Figure 3.29(a)
synthetic aperture image from measured RF data: (a) lateral and (b) axial. 85
Figure 3.31. Line spread functions for the center wire of the image, Figure 3.29(b)
synthetic aperture image from Field II simulation: (a) lateral and (b) axial. 86
Figure 3.32. Synthetic aperture image of a wire phantom reconstructed using a half-angle
of 9°and no apodization or thresholding. The dynamic range of the image is 40 dB and
the display uses a linear gray scale for mapping. 88
Figure 3.33. Synthetic aperture images of a porcine eye: (a) fundamental imaging with a
19 MHz tone burst, (b) pulse inversion tissue harmonic imaging with a 11 MHz tone
burst 89
ABSTRACT
Ultrasound transducer solutions were proposed for imaging the posterior segment of the
human eye. In one approach, concentric annular type dual element transducers for second
harmonic imaging at 20 MHz / 40 MHz were designed and fabricated for imaging the posterior
segment of the eye. The outer ring element was designed to transmit the 20 MHz signal and the
inner circular element was designed to receive the 40 MHz second harmonic signal. Multiple
prototype transducers were fabricated and characterized quantitatively. Images of a posterior
segment of an excised pig eye and a choroidal nevus of human eye were obtained and the
advantages of dual element harmonic imaging were demonstrated. In another approach, 20 MHz
192 element convex array transducers have been designed, fabricated, and characterized. It was
demonstrated that convex array transducers of 20 MHz with an aperture curved both in azimuth
and elevation direction could be fabricated with an acceptable uniformity in element-by-element
performance. All 192 elements of the array were fully characterized by the pulse-echo test,
crosstalk measurement, insertion loss test and synthetic aperture imaging. Average pulse-echo
loop sensitivity was –63.7 dB and average –6 dB fractional bandwidth 69.2%, which are
acceptable for imaging purpose. Ringing of echoes was substantially reduced by using a lower
composite kerf by the interdigital pair bonding for fabricating 1-3 composites. Created images of
ex vivo porcine eye tissues with a trapezoidal field of view were shown to a wider view angle
than linear arrays.
xiv

CHAPTER 1 INTRODUCTION
1.1 Medical Ultrasound
Ultrasound is one of the most often used medical imaging modalities due to its real-time
imaging capability, mobility, and safety (Shung, 2006). Among other medical imaging modalities,
such as X-ray, computed tomography (CT), magnetic resonance imaging (MRI), nuclear imaging
or positron emission tomography (PET), some modalities like digital x-ray have a real-time
imaging option as an add-on feature. However, ultrasound is still the only option which uses the
real-time capability as a fundamental feature. Most imaging modes are based on real-time
imaging. Real-time imaging gives users much more comfort in diagnosis of moving targets, either
it is a slow object like a fetus in the third trimester or a fast one like valves of the mouse heart.
Real-time imaging also enables users to obtain multiple slices of images during the image
acquisition. It helps to render 3-D images and results in a significant improvement in locating and
evaluating the lesion. For certain modes, ultrasound gives audible as well as visual information.
For instance, in diagnosing the flow in the umbilical cord in fetal imaging, as shown in Figure 1.1,
the umbilical cord was acquired with 2D mode, the flow was detected by color flow mode and the
direction and speed of the flow was quantized by the pulsed Doppler mode. In addition, the audio
sound of the flow is given to the user by speakers attached to the system. This multi-mode,
combined information gives a plenty of flexibility in diagnosing disease or evaluating the lesion
and therefore leads to the enhanced accuracy of diagnosis.
1

Figure 1.1. 2D/Color/Pulsed Wave Doppler combined mode image of the umbilical cord
(courtesy of Medison Co., Ltd.).
Ultrasound has the highest mobility of medical imaging modalities. Different from other
modalities that require patients to move to the imaging system for scanning, it is portable and so
easily accessible to the patients. This type of mobility allows the imaging room to be moved
virtually everywhere. Ultrasound can be used at the bed side of the patient in emergency room to
check quickly the status of a fetus of a pregnant woman delivered from the car accident, for
example. It can be also used at a ranch to image the tendon of a horse. Even in the battlefield, it
can be used to locate the region of internal bleeding of a soldier from explosion (Gawande, 2004).
Sonosite, Inc., one of medical ultrasound system manufacturers specialized in handheld
ultrasound, released a mountable ultrasound system which can be mounted to a cart, wall or
ceiling (SonoSite, 2010). As shown in Figure 1.2, this type of minimal footprint ultrasound
2

system can be used to locate a needle in the target area in the anesthesia procedure at the bed side.
This type of mountable ultrasound imaging system can be accepted as a basic monitoring device
which should be available near the patient’s bed.

Figure 1.2. Sonosite S-Nerve
™
, a mountable, zero-footprint ultrasound system used in the
anesthesia room (courtesy of SonoSite, Inc. and Duke University Medical Center).
Currently, the portability reached to the pocket size ultrasound system available from GE
Healthcare as shown in Figure 1.3 (GE Healthcare, 2010). It is the smallest medical diagnostic
ultrasound system in the world at the present time and its purpose is to provide users with earlier
and faster clinical assessment at the point of care. The system has an incorporated 2-4 MHz
phased array probe and provides 2D mode and color flow imaging. It operates for an hour with
the integrated battery and displays the image with 3.5 inch LCD with 240 x 320 pixels. It is
expected to be used like an ultrasonic ‘stethoscope’ and if it is successful, it will show the
extreme end of its mobility.

3

Figure 1.3. GE Healthcare Vscan™, a pocket size handheld ultrasound imaging system (courtesy
of GE Healthcare, Inc.).
Ultrasound does not produce an ionizing radiation and is considered safe if the intensity
and power level is under the limit determined by regulatory bodies such as International
Electrotechnical Commission (IEC) (International Electrotechnical Commission, 2007) or Food
and Drug Administration (FDA) (Food and Drug Administration, 1997). Medical diagnostic
ultrasound systems with color and Doppler capability must display the thermal and mechanical
indices in real time under the NEMA output display standard (National Electrical Manufacturers
Association, 2004). These indices give users estimation of the thermal increase of the tissue and
the likelihood of cavitations by ultrasound. An example of displayed indices is shown at the upper
right side of the image in Figure 1.4. As long as the acoustic output is under the limit, it is
4

considered safe. Ultrasound is the preferred imaging method for fetal imaging since ionizing
radiation is not recommended for fetus.

Figure 1.4. Real-time display of thermal (TI) and mechanical (MI) indices at the upper right side
on the screen (Courtesy of Philips Healthcare, Inc.).
Real-time imaging capability, mobility and safety among other advantages of ultrasound
enabled to rank it at the 2nd place of diagnostic imaging modality by the number of clinical trials
and it is next to conventional X-ray (Shung, 2006).
1.2 High Frequency Ultrasound Imaging
One of major research topics in medical ultrasound today is the high frequency
ultrasound imaging. High frequency ultrasound uses the frequency range of 15 to 120 MHz
5

whereas the conventional low frequency ultrasound uses 2 to 15 MHz. The main advantage that
can be achieved by using the higher frequency is the improvement of spatial resolution. However,
the penetration is limited by increasing the frequency and therefore the trade-off should be made
between the desired spatial resolution and depth of penetration in designing imaging systems and
transducers upon each application.
The lateral and axial resolution at the focal point can be predicted by (1.1) and (1.2)
(Foster, et al., 2002).
λ ⋅ =
#
f R
lateral
(1.1)
BW
c
R
axial
⋅
=
2
(1.2)
where the f-number, ( : focal length, D : the diameter of transducer), the
wavelength,
D z f
f
/
#
=
f
z
c
f c / = λ ( : the speed of sound, : the center frequency), c is the speed of
sound and BW the –6 dB bandwidth of the received ultrasound pulse. The lateral resolution can
be improved by increasing the center frequency and/or the aperture size of a transducer. The axial
resolution can be improved by increasing the center frequency and the bandwidth of a transducer.
c
c
f
The depth of penetration is mainly determined by the attenuation. The pressure of a plane
monochromatic wave propagating in z-direction decreases exponentially as a function of z as
shown in (1.3) (Shung, 2006).
z
e z p z p
α −
= = ) 0 ( ) ( (1.3)
where is the pressure at ) 0 ( = z p 0 = z and α is the pressure attenuation coefficient. The
attenuation coefficients are different for tissue types but are directly proportional to the frequency.
6

Therefore, as the ultrasound wave travels deeper in the tissue, the center frequency of transmitted
ultrasound is shifted down and it results in poorer axial resolution due to down-shifted frequency
and lowered bandwidth as well as the decreased magnitude of echo signal.
In addition, the depth of focus, , is defined by the region that its intensity of the beam
is within –3 dB of the maximal intensity of the focus. It is theoretically calculated by:
f
D
λ ⋅ =
2
#
f D
f
(1.4)
where is the f-number and
#
f λ the wavelength. By improving the lateral and axial resolution by
increasing the center frequency, the depth of focus gets worse.
In determining design parameters of transducers such as the aperture size, focal depth,
center frequency and others, the expected performance should be simulated to see if they meet all
performance goals.
High frequency ultrasound imaging is expanding its clinical applications to unexplored
area continuously. It includes but not limited to ophthalmology (Foster, et al., 1993) (Coleman, et
al., 2004), dermatology (Turnbull, et al., 1995) (Passmann & Ermert, 1996), intravascular
imaging (Liu & Goldberg, 1999), small animal studies (Turnbull, 1999) (Foster, et al., 2002) and
molecular imaging applications (Baddour, Sherar, Hunt, Czarnota, & Kolios, 2005) (Liang &
Blomley, 2003).

7

1.3 High Frequency Ultrasound Ophthalmic Imaging
High frequency ultrasound is one of major diagnostic tools in ophthalmology along with
optical coherence tomography (OCT). Ultrasound has better depth of penetration whereas OCT
has better spatial resolution.
Since its inception for high resolution imaging by Huang et al. in 1991 (Huang, et al.,
1991), OCT has been continuously improved and posterior segment imaging is one of main
applications. OCT has a superior resolution on the order of 10 μm than 20 MHz ultrasound,
which has a nominal lateral resolution of 250 μm. However, the depth of penetration of OCT is
approximately 1 mm. It is also degraded by the opacity like cataract or hemorrhages in the eye
and can only be performed in the fundus. That means the bottom part of the eye visible through
the pupil can only be imaged. However, ultrasound can visualize retinal, choroidal and scleral
layers and also the optic nerve without such limitations. Although the resolution is not yet at the
best level, the capability of ultrasound differentiating deeper tissue layers has a great benefit for
better diagnosis.
High frequency ultrasound ophthalmic imaging can be categorized into two parts,
anterior segment imaging and posterior segment imaging of the eye. Currently available
commercial ultrasonic biomicroscopy (UBM) systems use 35-50 MHz transducers for imaging
the anterior segment consisting of the cornea, anterior chamber, iris, ciliary body, and anterior
lens (Sonomed, 2006). UBM systems have contributed significantly in evaluation of pathologies
including glaucoma, corneal scars, hypotony and tumors. For imaging the posterior segment of
the eye, 10-20 MHz transducers are used. Posterior segment imaging shows greater need for
improved spatial resolution with enough penetration for correct diagnosis of retinal disease such
8

as age related macular degeneration, detached retina and diabetic retinopathy (Silverman,
Coleman, Ketterling, & Lizzi, 2005). Figure 1.5 shows examples of clinical images of the eye
obtained by a commercial ophthalmic imaging system.

Figure 1.5. Clinical images of the eye: (a) choroidal tumor and retinal attachment, (b) retinal
detachment, (c) iris crypt, (d) accommodative lens movement (Courtesy of Sonomed, Inc.)
1.4 High Frequency Ultrasound Transducers
There are two major trends in the development of high frequency ultrasound transducers.
One is to develop clinically useful, commercially usable, and reliable transducers in the frequency
range of 20 to 50 MHz. The other is to break the upper limit of the operating frequency of
transducers.
9

A variety of single element or array transducers for high frequency ultrasound imaging
have been reported to achieve the first goal; clinically useful and reliable transducers. 20–80 MHz
lithium niobate single element transducers have been developed and used for ultrasonic
biomicroscopy (UBM) imaging (Cannata, Ritter, Chen, Silverman, & Shung, 2003) and 20 MHz /
40 MHz dual element transducers were reported to be used for harmonic imaging of the posterior
segment of the eye (Kim, Cannata, Liu, Chang, Silverman, & Shung, 2008). PVDF 40 MHz
annular array transducers were also developed (Kettering, Aristizabal, Turnbull, & Lizzi, 2005)
and verified with the imaging of the bovine eye (Kettering, Ramachandran, & Aristizabal, 2006).
A 30 MHz linear array with 2-2 composite built by the stack-and-bond method (Ritter, Shrout,
Turwiler, & Shung, 2002), 35 MHz linear array transducers with 2-2 composite fine-grain
piezoceramic (Cannata, Williams, Zhou, Ritter, & Shung, 2006) and a 40 MHz linear array with
1-3 PZT – polymer composite with geometric elevation focusing (Brown, Foster, Needles, Cherin,
& Lockwood, 2007) have been reported. Recently, as shown in Figure 1.6, Foster et al. have
introduced a linear array based high frequency ultrasound system (Foster, et al., 2009) which
targets preclinical imaging. The laser-diced 2-2 composite linear array transducers offer uniform
fine resolution over the field of view and color/Doppler flow images which is hard to implement
with single element transducers.
To reach the highest possible operating frequency of ultrasound transducers, many groups
have tried to break its limit with new materials, new fabrication techniques, and new structures.
Sputtered zinc oxide (ZnO) 100 MHz self-focused single element transducer was proposed and
the image of zebra fish was acquired (Cannata, et al., 2008). A 100 MHz ZnO linear array
transducer was also introduced (Ito, Kushida, Sugawara, & Takeuchi, 1995).
10

Figure 1.6. VisualSonics Vevo
®
2100 Imaging System: (a) an imaging system, (b) a 30 MHz
linear array, (c) an image of vasculature and lobes of the mouse liver (Courtesy of VisualSonics,
Inc.) (VisualSonics, 2010)
1.5 Objective of Research
Diagnostic medical ultrasound market is dominated by radiology, cardiology and
obstetrics and gynecology (OB/Gyn). The combined market portion of these three major
applications in the United States is larger than 80%. Market portions of other applications are
much lower than those of three major applications. The market size of ophthalmic imaging is very
small, less than 1% (Klein Biomedical Consultants, Inc., 2007). OCT is a strong competitor in
ophthalmic imaging since it has superior resolution. However, deeper penetration of ultrasound
should be able to address its needs in ophthalmic imaging. More importantly, most transducers
commercially available now are single element transducers. Mechanically translated single
11

12

element transducers have limitations in achieving higher frame rates, color flow imaging and
Doppler duplex imaging. Array transducers well suited to desired applications with matched
ultrasound imaging systems will be of significant help to create a new market.
The goal of this research is to provide transducer solutions for imaging the posterior
segment of human eye. First, concentric annular type 20 MHz/40 MHz dual element transducers
for harmonic imaging were designed, fabricated, and tested. Harmonic imaging with 20 MHz
transmit and 40 MHz receive demonstrated superior image quality over fundamental imaging of
20 MHz. Second, 20 MHz convex array transducers have been designed, fabricated, and
characterized. Two-way curving both in elevation and azimuth directions were possible by the
flexibility of 1-3 composite. Its curved aperture could generate wider field of view than do linear
arrays.
CHAPTER 2 HIGH FREQUENCY DUAL ELEMENT TRANSDUCERS
2.1 Dual Element Transducers for Harmonic Imaging
Tissue harmonic imaging has been accepted as one of standard imaging modalities in
many applications since its introduction to medical ultrasound imaging in 1990’s (Averkiou,
2000). Especially in cardiac and abdominal studies, tissue harmonic imaging is very often used
for diagnostics along with fundamental imaging. By utilizing the second harmonic component of
the receive signal, images can be improved by reducing near field reverberation, decreasing phase
aberration error, and improving border delineation (Averkiou, 2000). Recently, harmonic imaging
has been used in ophthalmic (Silverman, Coleman, Ketterling, & Lizzi, 2005), urologic (Merks,
Bouakaz, Bom, Lancee, van der Steen, & de Jong, 2006), and intravascular ultrasound (IVUS)
(Vos, et al., 2005) imaging studies.
A variety of specially designed transducers dedicated to harmonic imaging have been
reported recently. A lithium niobate plate transducer with a local ferroelectric inversion layer
produced by titanium diffusion and heat treatment was proposed for 50MHz transmit and 100
MHz receive harmonic imaging (Nakamura, Fukazawa, Yamada, & Saito, 2006). A multilayer
transducer with a PZT layer used for transmission and a PVDF layer used for reception was
reported for the bladder volume measurement (Merks, Bouakaz, Bom, Lancee, van der Steen, &
de Jong, 2006). An oval shape PZT transducer with additional passive mismatching layers for
intravascular ultrasound (IVUS) was also reported for 20 MHz / 40 MHz harmonic imaging (Vos,
et al., 2005).
13

Recently, broadband lithium niobate (LiNbO
3
) single element transducers operating at
about 20 MHz have been used for imaging the posterior segment of the eye (Coleman, et al., 2004)
but were limited due to the spatial resolution at that frequency. As an alternative to fundamental
imaging, harmonic imaging was also examined (Silverman, Coleman, Ketterling, & Lizzi, 2005).
Unfortunately, transducers operating at 20 MHz may not provide the spatial resolution needed to
adequately delineate layers on the posterior segment of the human eye and on the contrary those
operating in the higher frequency range do not provide sufficient depth of penetration. In this
study, we propose to resolve this problem by creating a concentric annular type dual element
transducer for second harmonic imaging of the posterior segment of the eye. The outer ring
element is used for transmit and the inner circular element for receive. A ring shaped outer
element produces higher sidelobes than does a circular element of the same diameter, but this is
to some degree compensated for by the inherently lower sidelobes in the harmonic compared to
the fundamental (Duck, 2002).
Multiple prototype dual element transducers for harmonic imaging have been fabricated
and tested. Different apertures, different matching layer thicknesses, and different fabrication
techniques have been experimented with to optimize performance. Clinical images of various
cases were obtained and compared to fundamental imaging.
2.2 Design of Dual Element Transducers
2.2.1 Transducer Geometry and Structure
A concentric annular type dual element structure was chosen to satisfy the needs to have
two separate transducers in one housing. Those two elements should have the identical beam axis
14

and focal point to obtain echoes from the same image plane while functioning at different
frequencies. The transducer also should have two separate connectors since their use is dedicated
for transmit and receive respectively.
Figure 2.1(a) shows that the outer ring element and the inner circular element were
placed in a brass housing and have separate connectors. The outer ring element was designed to
transmit a 20 MHz signal and the inner circular element was designed to receive the second
harmonic signal at 40 MHz. Figure 2.1(b) is the cross-sectional drawing of this transducer. The
outer diameter of the transmit ring element was set to be 10.0 mm or 12.0 mm (See Options 1 and
2 in Table 2.1). The inner diameter was 5.4 mm. The center was filled by the 5.0 mm diameter
receive element.
Table 2.1. Design Parameters for Dual Element Transducers
Layer Material
Option 1
(D
out
= 10 mm)
Option 2
(D
out
= 12 mm)
Thickness [ μm] Thickness [ μm]
20 MHz
Transmit
40 MHz
Receive
20 MHz
Transmit
40 MHz
Receive
Piezo LiNbO
3
150 77 150 77
1
st
M/L Silver epoxy I
a
23 12 23 12
2
nd
M/L Parylene 19 19 14 14
Backing Silver epoxy II
b
N/A N/A N/A N/A
a
Insulcast 501 + 2-3 μm silver particles
b
E-Solder 3022 (centrifuged)
M/L: Matching layer

15

Figure 2.1. Dual element transducers: (a) photograph of finished device, (b) cross-sectional
drawing of the transducer (not to scale)

16

2.2.2 KLM Modeling
KLM modeling software (PiezoCAD, Sonic Concepts, Woodinville, WA) was used to
determine the aperture size, and the proper thicknesses of the lithium niobate (LiNbO
3
), inner
silver epoxy matching layer and parylene outer matching layer. A more detailed description of
the materials used in the design can be found in Cannata et al. 2003 (Cannata, Ritter, Chen,
Silverman, & Shung, 2003). The diameter of each element was chosen to match its electrical
impedance at the given frequency to 50 Ω. As described by Cannata et al. 2003 (Cannata, Ritter,
Chen, Silverman, & Shung, 2003), lithium niobate is as good material to use in large-aperture
high frequency transducer designs since it has a low relative clamped dielectric constant (
/
)
of 39, a high longitudinal sound speed ( ) of 7340 m/s, and a comparable thickness mode
electromechanical coupling coefficient ( ) of 0.49 to PZT-5H. The radius of curvature for both
transmit and receive elements was set to 30 mm to place the focal point at the retina in immersion
mode scanning.
The thicknesses of the lithium niobate layer and 1st matching layer of the transmit
element were optimized for 20 MHz and those of the receive element for 40 MHz. The thickness
of the second matching layer was chosen to be intermediate to the ideal calculated for 20 MHz
and 40 MHz (See Option 1 in Table 2.1), since the same layer thickness would be applied to both
the inner element and outer ring to ease construction. The thickness of parylene optimized at 40
MHz (See Option 2 in Table 2.1) was also investigated to improve the performance of receive
element.
17

2.3 Fabrication of Dual Element Transducers
36° rotated Y-cut lithium niobate plates (Boston Piezo-Optics, Bellingham, MA) with
thickness of 500 μm each were prepared, one for the 20 MHz transmit and the other for the 40
MHz receive element. Lithium niobate plates were lapped to the designed thicknesses, 150 μm
for 20 MHz and 77 μm for 40 MHz respectively, and electroplated with 1500 Å chrome / gold
(Cr / Au) layer on both sides by an NSC-3000 automatic sputter coater (Nano-Master, Austin,
Tx). The silver epoxy inner matching layer, made from a mixture of Insulcast 501 epoxy
(American Safety Technologies, Roseland, NJ) and 2-3 μm silver particles (Aldrich Chemical
Co., Milwaukee, WI) with the weight ratio of 2.5 to 6, was cast over the lithium niobate plates
after applying the adhesion promoter (Chemlok AP-131, Lord Corp., Erie, PA). The epoxy was
then centrifuged at 3000 rpm for 10 minutes. After curing, the matching layers were lapped down
to the desired thickness. The lapped piezo / 1
st
matching stack plates were mechanically diced
into square pieces with the size that would encompass a ring or a circular element. A conductive
silver epoxy (E-Solder 3022, Von Roll Isola Inc., New Haven, CT) backing layer was then cast
and centrifuged on to back side of the electroplated pieces of lithium niobate. After curing, the
backed/matched pieces of lithium niobate were carefully drilled to the desired inner diameter of
the ring and turned down to the outer diameter of the ring on a lathe to create the 20 MHz
transmit element. The 40 MHz inner circular element was also turned down to the proper
diameter. A protective piece of machinable ceramic was waxed to the top of the stack to reduce
chipping during the drilling and turning down stages. A thin brass cylinder was fabricated and
placed between transmit and receive elements to provide RF shielding. Machined transmit and
receive acoustic stacks were placed in the brass housing concentrically and the gaps between
stacks were filled in by an insulating epoxy (Epo-Tek 301, Epoxy Technologies, Billerica, MA).
18

Press-focusing was used to generate a mechanical focus at 30 mm for both transmit and
receive elements (Cannata, Ritter, Chen, Silverman, & Shung, 2003). The transducer surface was
sputtered with Cr / Au electrodes to make ground connection between the elements and the brass
housing. The outer parylene layer was then deposited using a PDS 2010 Labcoater (SCS,
Indianapolis, IN). The finished transducer elements were connected by two separate SMA
connectors, one for transmit and the other for receive.
In addition, 20MHz single element transducers of the same total diameter and similar
acoustic characteristics as the transmit ring element were fabricated for comparison.
2.4 Performance of Dual Element Transducers
The transmit and receive elements were characterized by a one-way sound field
measurement with a hydrophone, two-way pulse-echo test with a soft target for harmonic
generation, and a Schlieren image test for visualizing the beam profile (Zanelli & Howard, 2006).
The transmit sound field data for dual element transducers were compared with those of single
element transducers to determine the effect of the ring shape. Round trip lateral beam profiles of
dual element transducers with harmonic imaging were simulated and compared with fundamental
imaging with single element transducers.
The transmit beam of the 20 MHz outer ring element at the focal point was measured
with a needle hydrophone (HGL-0085, Onda Corp., Sunnyvale, CA). In this measurement, an
ultrasonic analyzer (5900PR, Panametrics Inc., Waltham, MA) was used to excite the transducers
and a digital oscilloscope (LC534, LeCroy Corp., Chestnut Ridge, NY) was used to record the
waveform acquired by the hydrophone. The measurement was performed in a degassed /
deionized water tank. After aligning the hydrophone to be perpendicular to the beam axis of a
19

transducer, the maximum peak-to-peak voltage point along the beam axis was located. The time-
domain waveform at the maximum peak-to-peak voltage point was recorded and its spectrum was
calculated. Figure 2.2 shows the measured results for the 20 MHz transmit element. Its –3 dB
center frequency is 21.0 MHz and –3 dB fractional bandwidth is 56 %.

Figure 2.2. Transmit characteristic of the 20 MHz outer ring of the dual element transducer
(Option 2: D
out
= 12 mm): (a) a measured waveform by a hydrophone at the focal depth, (b) its
spectrum magnitude.

20

The receive characteristic of the 40 MHz inner circular element was measured by
utilizing a separate source transducer. Special test settings were configured to check if the receive
element has proper center frequency and bandwidth to receive the second harmonic signal
centered at 40 MHz. A 40 MHz single element transducer (diameter = 4 mm, focus = 12 mm)
was used as a source and the receive element as a receiver. The distance between source and
receiver for this measurement was determined by the sum of focal distances from each
transducer. The focused beam at 12 mm from the source travels 30 mm to reach the surface of
receiver. Therefore, a comparison of the spectra acquired by the hydrophone and the receive
element would yield the receive characteristic of the receive element. However, with the
hydrophone, it was difficult to acquire enough signal strength at that distance (42 mm) since the
beam diverges significantly after the focal point. Instead, the beam was measured with the
hydrophone placed at the focal point of 12 mm as shown in Figure 2.3(a). Its spectrum was
calculated, and the attenuation of water (2.2×10
-3
dB/cm/MHz
2
) (Lockwood, Turnbull, & Foster,
1995) was taken into account to approximate the signal spectrum incident to the surface of the
receive element. Figure 2.3(b) shows the spectrum of Figure 2.3(a) (‘at z = 12 mm’), the
attenuation curve vs. frequency in the water at the distance of 30 mm (‘attenuation’) and
attenuated spectrum at the surface of receive element, z = 42 mm (‘at z = 42 mm’). The resultant
spectrum is shown again in Figure 2.4(b) as a dotted line. The waveform acquired by the receive
element at 42 mm is shown in Figure 2.4(a) and its spectrum is shown as a solid line in Figure
2.4(b). The spectrum of the received signal is centered at 30.5 MHz. Comparing the input (a
dotted line) and the output (a solid line) of the receive element, it was found that the receive
element has acceptable bandwidth to pass spectral components centered at approximately 30
MHz. This frequency characteristic of the receive element allows extracting the second harmonic
21

signal with maximum achievable signal to noise ratio by a digital bandpass filter. Although the –3
dB frequencies in the one-way spectrum of the receive element are approximately 20 MHz and 43
MHz, the receive signal in the range between 25 MHz and 45 MHz is mainly amplified as shown
in Figure 2.4(b). Therefore, we designed a 33-tap FIR bandpass filter with cutoff frequencies of
25 MHz and 45 MHz, which was used to produce harmonic images for clinical evaluation.

Figure 2.3. Transmit characteristic of the 40 MHz single element source transducer: (a) a
measured waveform by a hydrophone at the focal depth of the source transducer (z = 12 mm), (b)
its spectrum magnitude (‘at z = 12 mm’), attenuation curve vs. frequency in the water at the
distance of 30 mm (‘attenuation’) and attenuated spectrum at the surface of receive element, z =
42 mm (‘at z = 42 mm’).
22

Figure 2.4. Receive characteristic of the 40 MHz inner circular element of the dual element
transducer: (a) a received waveform by the receive element, (b) an attenuated spectrum of the
signal from the source transducer and a spectrum of the received signal of receive element.
Figure 2.5 and 2.6 demonstrate the advantage of dual element structure over single
element for second harmonic imaging. The pulse-echo test data for the dual element transducer
show clearly identified second harmonic component which is not visible with single element
transducer. A soft silicone rubber target was used to generate harmonic components. When the
silicone rubber target was used, the higher voltage could be applied for the transmit signal than
when the quartz target was used. The echo signal was easily saturated from the quartz target since
23

it has a stronger reflection than the silicone rubber. Therefore, the higher transmit voltage with
the silicone rubber target resulted in the increased nonlinearity of the echo since the level of the
second harmonic component (the degree of nonlinearity in tissue harmonic imaging) is
proportional to the mechanical index (MI) (Averkiou, 2000). The MI is defined as
, where is the peak rarefactional pressure in MPa derated by
0.3dB/cm-MHz to the point on the beam axis, , where the pulse intensity integral is
maximum, and is the center frequency in MHz (Food and Drug Administration, 1997) and the
MI is directly proportional to the transmit voltage.
) /( ) (
2 / 1
3 . c sp r
f z p MI =
c
f
) (
3 . sp r
z p
sp
z
Figure 2.5 shows a pulse-echo signal and its spectrum obtained transmitting and
receiving with a 20-MHz outer ring element only. The spectrum has a center frequency of 15.6
MHz and a very low second harmonic component because of an insufficient bandwidth of the
transducer for the second harmonic imaging. Figure 2.6 is a pulse-echo signal and its spectrum
obtained by transmitting with the 20 MHz outer ring element and receiving with 40 MHz inner
circular element. In the spectrum obtained with the dual element transducer in Figure 2.6(b), the
magnitude of the second harmonic component centered at 30.5 MHz is 7 dB lower than that of
the fundamental component. These results demonstrate that the dual element transducer is
capable of efficiently receiving the second harmonic component and thus improving image
quality due to its adequate sensitivity and bandwidth.
24

Figure 2.5. Pulse-echo test results using the 20MHz outer ring element for transmission and
reception (Option 2: D
out
= 12 mm): (a) an echo waveform from the soft silicone rubber target at
the focal depth, (b) its spectrum magnitude.
25

Figure 2.6. Pulse-echo test results for the dual element transducer using the 20 MHz outer ring
element for transmission and the 40 MHz inner circular element for reception (Option 2: D
out
= 12
mm): (a) an echo waveform from the soft silicone rubber target at the focal depth, (b) its spectrum
magnitude.
In order to evaluate the transmit sidelobe level of the 20 MHz ring element, its lateral
beam profile was also measured by a needle hydrophone at its focal point and compared with that
of full aperture 20 MHz single element transducer. As shown in Figure 2.7, the sidelobe level for
the ring element is higher than that of full aperture single element transducer. Measured results
were also compared with the simulated lateral beam profiles generated by Field II (Jensen, 1996).
As is visible in the Figure 2.7, the simulated and measured results are fairly well matched. The
26

sidelobe level of dual element transducer measured at x = 0.33 mm was -9.6 dB whereas that of
single element transducer was -14.7 dB. Figure 2.8 shows the effect of outer ring diameter to the
lateral beam pattern. By increasing the outer diameter from 10 mm to 12 mm, the mainlobe
energy was increased and the sidelobe level was reduced significantly.
Although the transmit element was ring shaped and the increased sidelobe level was
anticipated, harmonic imaging lead to a reduction of the sidelobe level and resulted in comparable
improvement in lateral resolution (Duck, 2002).

Figure 2.7. Lateral beam profiles obtained by the transmit waveforms at the focus from the 20
MHz outer ring of the dual element transducer (Option 1: D
out
= 10 mm) and the 20 MHz circular
shape single element transducer, measured by a hydrophone and simulated by Field II.
27

Figure 2.8. Lateral beam profiles obtained by the transmit waveforms at the focus of dual element
transducers for different outer diameters of the 20 MHz outer ring element (Option 1: D
out
= 10
mm vs. Option 2: D
out
= 12 mm).
Schlieren imaging is a more efficient tool to visualize the sound field (Zanelli & Howard,
2006). Multiple images were obtained with a high frequency Schlieren imaging system
(Optison® Beam Analyzer – High Frequency Option, Onda Corp., Sunnyvale, CA). From the
transmit beam patterns of circular and ring shape transducers at the focus shown in Figure 2.9,
which is an uncompressed 256 gray-scale image, it is clearly seen that the ring shape transducers
generate higher sidelobe levels than circular shape transducers. This result correlates well with
the previous evaluation with the lateral beam profiles obtained with the hydrophone measurement
and Field II simulation.
28

Figure 2.9. Schlieren images of 20 MHz circular and ring shape elements (Option 2: D
out
= 12
mm), uncompressed and linearly mapped to a 256-level gray scale: (a) circular, at the focal depth,
(b) ring, at the focal depth.
2.5 Clinical Evaluation of Dual Element Transducers
The performance of the dual element transducer was evaluated with images of the
posterior segment of the excised pig eye. The excised pig eye was cut in half to remove the
anterior segment with and only the posterior segment placed in the water tank. The Panametrics
5900PR was used to excite the 20 MHz ring element and the LC534 digital oscilloscope was used
to record the echo signals received by the 40 MHz circular element obtained by sampling at 500
MHz. In order to acquire 400 scan lines for harmonic imaging, a stepper motor repeatedly
translated the dual element transducer by 25 μm along the horizontal direction. For the
29

comparison, another 400 scan lines were also acquired with a 20 MHz single element transducer
in the same manner. The two sets of scan lines were used to form a harmonic and a fundamental
image respectively. Figure 2.10 shows images of the posterior wall of the pig eye, which cover an
area of 3.0 mm × 1.5 mm in the vicinity of the focal depth of 30 mm. From Figure 2.10, it is
clearly seen that a harmonic image produced by the dual element transducer (b) has better spatial
resolution and better border delineation capability than a fundamental image by the single
element transducer (a).

Figure 2.10. Images of the posterior segment of excised pig eye, 3.0 mm × 1.5 mm in the vicinity
of the focal depth of 30 mm, logarithmically compressed to a dynamic range of 60 dB and
linearly mapped to a 256-level gray scale: (a) fundamental imaging using the single element
transducer, (b) harmonic imaging using the dual element transducer (Option 2: D
out
= 12 mm).
30

A study of in-vivo human eyes was also performed with a custom built imaging system
(Coleman, et al., 2004) (Reinstein, et al., 2000). This study adhered to the tenets of the
Declaration of Helsinki and was approved by the Institutional Review Board of the Weill Cornell
Medical College.
Figure 2.11 shows the images of a choroidal nevus. The scanning laser ophthalmoscope
(SLO) image shows the retinal vasculature and a dark, disrupted and roughly round area towards
the left of the image. The OCT C-scan (upper right) is, like the SLO image, an en face
representation, but in this case of a plane cutting through the retinal pigment epithelium, which is
highly reflective and represented in red in the color-scale OCT images. The peninsular
appearance of the region in the C-scan indicates that it is an elevated lesion. The OCT B-scan
(bottom) is in the plane of the horizontal red line drawn through the SLO image and is 1.5 mm in
depth. It shows the retinal layers, including the pigment epithelium (red) overlying the lesion.
The lesion itself is seen only as an elevation of the retinal contour due to absorption of light by
pigment. The macula is seen as a small dip to the right of the tumor.
Figure 2.11(b) and (c) show the appearance of the lesion on the B-scan with a single
element 20 MHz transducer by fundamental imaging (b) and the dual element transducer by
harmonic imaging (c). The image, which is 6.3 mm in depth, was constructed from RF echo data
acquired at a 250 MHz sample rate. We can observe greater shadowing by the lesion (thin arrow)
and the improved depiction of the overlying retina and sclera border (thick arrow) in the
harmonic image. The optic nerve (ON) cup is more clearly depicted in the harmonic image as
well.
31

Figure 2.11. Images of a choroidal nevus of the human eye: (a) images by SLO and OCT, (b)
fundamental imaging using the single element transducer, (c) harmonic imaging using the dual
element transducer (Option 1: D
out
= 10 mm; thin arrow: greater shadowing by the lesion, thick
arrow: improved depiction of retina/sclera border, ON: optic nerve), logarithmically compressed
to a dynamic range of 30 dB and linearly mapped to a 256-level gray scale
32

2.6 Advantages of Dual Element Transducers for Harmonic Imaging
20 MHz / 40 MHz dual element transducers for high frequency ophthalmic imaging were
designed, fabricated and evaluated. The concentric type dual element structure packaged in a
single housing worked well as a dedicated harmonic imaging transducer. Although the ring shape
of the transmit element generated increased sidelobe levels, the advantage of harmonic imaging in
suppressing sidelobes more than sufficiently compensated its shortcoming to achieve an
improved lateral resolution.
Harmonic imaging with 20MHz transmit and 40 MHz receive showed superior capability
than fundamental imaging at 20 MHz to diagnose the retinal disease at the posterior segment of
the eye. The center frequencies of transmit and receive elements of dual element transducers can
be further optimized to match the designed center frequencies to support a larger dynamic range.
The aperture size of transmit and receive elements can also be optimized with further
experimentation to achieve the best combination of transmit and receive efficiency.
2.7 Extended Applications of Dual Element Transducers
A dual element transducer is basically a transducer which has two elements of different
center frequencies in a monolithic device. Although the primary purpose was the harmonic
imaging with 20 MHz transmit and 40 MHz receive, any combination of transmit and receive
with two different frequencies could be implemented. Broad bandwidth achieved by this
concentric annular type dual element structure could be used to implement frequency
compounding to reduce speckle interference while minimizing the loss of axial resolution (Chang,
Kim, Lee, & Shung, 2010). In frequency compounding, frequency division is carried out to obtain
sub-band images containing uncorrelated speckles. A compounding image is then produced by
33

adding or averaging the sub-band images. However, this frequency division degrades the axial
resolution and often leads to the degradation of total signal-to-noise ratio (SNR) when the
speckle’s SNR is not improved as much as the degraded axial resolution. Two different bands of
a dual element transducer could avoid this issue by providing a much wider bandwidth (>100%),
which is not practical to achieve by a single transducer. Improved SNR was observed from in
vitro imaging experiments of an excised pig eye as shown in Figure 2.12.

Figure 2.12. Image of an excised pig eye obtained by a 20 MHz/40 MHz dual element transducer
of an excised pig eye with frequency compounding of four sub-band signals from Chang et al.
(Chang, Kim, Lee, & Shung, 2010)
Photoacoustic imaging which combines optics and ultrasound has been recently emerged
as a complementary imaging technique. Ring shape 20 MHz ultrasonic transducers with the same
design parameters as a transmit element of a dual element transducer were built and used as a
receiver in a compact monolithic device, which allows focused laser and ultrasonic beams to be
launched collinearly (Kong, et al., 2009). The structure of the imaging system is shown in Figure
2.13 and the images obtained in Figure 2.14. This concentric structure enabled a far extended
34

working distance and a significantly improved lateral resolution which made imaging
choroid/retina layers of the posterior segment of the eye better than conventional ultrasound
(Silverman, et al., 2010).

Figure 2.13. Photoacoustic imaging system setup using a 20 MHz receive, ring shape ultrasonic
transducer and a 532-nm wavelength laser transmitter from Kong et al. (Kong, et al., 2009)
35

Figure 2.14. Images of a ciliary body of an excised pig eye. Pulse-echo (left) and photoacoustic
(right) images. The B-scans were made in the plane perpendicular to the ciliary processes. The
photoacoustic images reveal individual processes with high resolution and clarity not obtainable
with the pulse-echo 20 MHz ultrasound from Kong et al. (Kong, et al., 2009)

36

CHAPTER 3 HIGH FREQUENCY CONVEX ARRAY TRANSDUCERS
3.1 Convex Array Transducers
Since its inception to the real-time ultrasound scanner in the early 1980s (Woo, 1998),
convex array transducers have been widely used in medical ultrasound imaging, especially in
radiology, obstetrics and gynecology. Convex array transducers are linear arrays curved in a
convex shape along the azimuth direction. Given the same pitch and the same number of
elements, i.e. the same array length, convex arrays give a wider field of view than do linear arrays
since the curved aperture can cover a trapezoidal shaped area whereas linear arrays can cover a
rectangular shaped area only. Phased arrays have been used for imaging that requires a small
footprint for transducers. Cardiac imaging is one of main applications that have adopted phased
arrays since the array transducer should be placed in the narrow access window, intercostal
spaces. Phased arrays generate 90° triangular-shaped images by steering beams and therefore can
create a wide field of view with a small aperture. However, their lateral resolution in the far field
and at the side of images degrades rapidly. Phased arrays also require the finer pitch of less than
half-wavelength to avoid the grating lobes, which is a limiting factor for fabrication for higher
frequency array transducers. The finer pitch of half-wavelength limitation for phased arrays also
means a smaller size of array elements that leads to a smaller aperture given the same number of
channels of the beamforming circuits. Due to a smaller aperture, a lower sensitivity and a higher
lateral beamwidth are expected. Limited field of view for linear arrays and degradation of
resolution for phased arrays can be resolved by increasing the number of elements but that
requires more channels for beamforming and thus increases manufacturing and operating costs.
37

Obstetrics and gynecology is the application most benefited from convex arrays. Long
aperture linear arrays for fetal imaging in early 80’s had been replaced by more compact convex
arrays and they provided better contact on the patient’s abdomen and a wider field of view.
Endocavity probes with a smaller radius of curvature and a long shaft design fit inside a small
cavity enabled fetal imaging in the first trimester. Pregnant women can now obtain ultrasound
images of the fetus from the very first visit to clinic. The early stage of fetus in the uterus can be
imaged by this tightly curved convex array transducer.
The fabrication of convex arrays is more difficult than linear or phased arrays due to the
curvature of the aperture. The radius of curvature and the pitch on the array surface must be
accurately maintained in order to minimize the error in the beamforming. Prior to curving in the
azimuth direction, the ceramic should be diced along the elevation direction to prevent cracks.
For the curved aperture, 2-2 or 1-3 composites are preferred since pre-diced ceramic posts do not
create cracks in conforming. If bulk ceramics are used, lens is preferred for elevation focusing.
The center frequency of currently available convex array transducers for medical
diagnostic ultrasound ranges from 2.0 to 7.5 MHz. Lower frequency arrays operating in the 2.0 to
5.0 MHz range are mainly used for abdominal or cardiac imaging and higher frequency ones
operating in the 5.0 to 7.5 MHz range are used for endocavity or neonatal applications. No
convex array transducers above 10 MHz are currently available since linear arrays can cover most
of clinical needs in that frequency range and few applications have been found with high
frequency convex arrays. Difficulties in fabrication of convex arrays above 7.5 MHz also
contribute to their limited availability.
38

Ophthalmic imaging is one of main applications of high frequency ultrasound along with
small animal studies, skin imaging, and intravascular applications (Shung, 2006). Most high
frequency ophthalmic imaging systems are based on fixed focus single element transducers. Arc
scanning is mainly used to image along the contour of the anterior segment of the eye, whereas
sector scanning is used for imaging the posterior segment. The major drawback to these
techniques is that the aperture is translated mechanically, which results in a fairly low frame rate.
Annular array transducers can achieve better spatial resolution over a larger depth of field
(Kettering, Aristizabal, Turnbull, & Lizzi, 2005) but they still have the same limitation of
mechanical translation. Doppler can be also implemented by single element transducers or
annular arrays but color flow mapping is very difficult to implement with those transducers.
Linear array transducers can achieve much higher frame rate with electronic translation (Cannata,
Williams, Zhou, Ritter, & Shung, 2006) but their view width may be too narrow to image the
whole eye in one imaging plane.
20 MHz convex array transducers have been designed, fabricated, and characterized. The
primary application is ophthalmic imaging with the emphasis on the posterior segment.
However, these arrays should be useful for imaging any small organs close to the skin surface or
accessible through a cavity. This convex array’s 52° view angle with 23 mm radius of curvature
should be able to cover the entire eye in a single image plane, and its 20 MHz center frequency
and 30 mm focal depth will be suitable for posterior segment imaging. Electronic translation
enables higher frame rate. It can create multiple focal zones with dynamic aperture and
implement Doppler or color flow mapping. 2D/Doppler duplex mode which can show the
Doppler signal and 2D image simultaneously is also possible. Advantages of convex array
39

transducers over currently commercially available single element transducers are summarized in
Table 3.1.
TABLE 3.1. FEATURES COMPARISON OF SINGLE ELEMENT TRANSDUCERS AND CONVEX ARRAYS
Features Single element transducers Convex array transducers
Translation Mechanical Electrical
Vibration Yes No
Max. frame rate ~ 30 Hz > 100 Hz
Multiple focus No Yes
2D/Doppler duplex mode No Yes
Color No Yes
3D Yes Yes

3.2 Design of Convex Arrays
3.2.1 Array Geometry
In the design, human eyes were assumed a sphere of 1 inch diameter. The array geometry
and acoustic parameters were carefully chosen to cover the whole eye in one image plane.
Covering the organ in one image plane is advantageous since the user can easily formulate the 3D
rendered image by translating the array along the elevation direction only. It becomes harder
where the view window is limited and the user must translate the array in both azimuthal and
elevation directions. The convex array can easily be positioned over the sclera to image the
posterior segment so as to avoid the strong reflection from the lens as shown in Figure 3.1 (a)
(Wikipedia, 2010). The array length, i.e. the pitch and the number of total elements, the radius of
curvature and its created field of view is shown in Figure. 3.1 (b). A 7.0 mm elevation width and
30 mm geometric focus was chosen as a compromise between depth of field and resolution. The
40

center frequency of 20 MHz was chosen to achieve required resolution and penetration for the
posterior segment imaging.

Figure 3.1. Convex arrays for ophthalmic imaging: (a) a convex array placed over sclera of the
human eye (Wikipedia, 2010), (b) a trapezoidal imaging plane with the focal zone at the posterior
segment of the eye
41

For adequate sampling of the image resolution and to avoid aliasing in the image, the
number of acoustic lines per expected resolution cell in the lateral dimension must be 2 to 4
(McKeighen, 1998). Therefore, the pitch, P should satisfy:

) / ( 6 . 0 N R P λ ≤ (3.1)
where R is the radius of curvature, λ is the wavelength, and N is the number of channels. From
(3.1), the pitch should be lower than 1.69 λ and therefore 111 μm (=1.5 λ) was chosen. The kerf
width of 14 μm is acceptable considering current dicing capability. The array footprint of 21.3
mm × 7.0 mm is a proper size to be easily positioned over the sclera.
The two-way sound field was simulated by Field II (Jensen, 1996) using the curved
aperture of 192 array elements as shown in Figure 3.2. The two-way sound field in the azimuth
direction with the transmit geometric focus at 30 mm and receive dynamic focusing generated
grating lobes of –38 dB at 20º at a range of 30 mm. Its planar sound field at the focal depth in x-z
plane is shown in Figure 3.3. The lateral beam profile at the elevation focal point of 30 mm was
obtained (Figure 3.4) and the –6 dB lateral beamwidth was 318 μm. The simulated echo showed a
–6 dB axial beamwidth 50 μm, which is acceptable for imaging with 20 MHz transducers. The
axial beam profile (Figure 3.5) gives a depth of field of 8.6 mm. The sound field simulation
demonstrates that the array geometry summarized in Table 3.2 meets our design goal for 20 MHz
transducers.

42

Figure 3.2. Two-way curved aperture of 20 MHz 192 element convex array transducers created
for Field II simulation

Figure 3.3. 64 channel, 2-way planar beam profile of 20 MHz convex array transducers by Field
II simulation
43

Figure 3.4. 64 channel, 2-way lateral beam profile of 20 MHz convex array transducers by Field
II simulation

Figure 3.5. 64 channel, 2-way axial beam profile of 20 MHz convex array transducers by Field II
simulation
44

TABLE 3.2. ARRAY GEOMETRY AND ACOUSTIC PARAMETERS OF 20 MHZ CONVEX ARRAY
TRANSDUCERS
Parameters Values
Center frequency 20 MHz
Total number of elements 192
Number of channels 64
Pitch 111 μm
Kerf width 14 μm
Array length 21.3 mm
Elevation width 7.0 mm
Geometric focus 30.0 mm
Radius of curvature 23.6 mm
Maximum view angle 51.7 °

The images of a wire phantom and a dog eye with the linear array and the convex array
were simulated and compared as shown in Figure 3.6 and Figure 3.8 respectively. The H&E
(hemotoxylin and eosin) stain image of the dog eye (Figure 3.7) was processed to the gray scale
one with the corresponding scattering value at each pixel. It was assumed to image with the linear
array and the convex array positioned obliquely to avoid the lens as shown in Figure 3.1. The
linear array has the same number of elements (=192) and the same pitch (=111 μm) as the convex
array. The convex array gives the wider view area than the linear array as shown in Figure 3.6
and Figure 3.8. More wires can be displayed by convex arrays than by linear arrays. It can image
the whole eye in one plane. Both linear and convex arrays can depict layers of the posterior
segment well.

45

Figure 3.6. Wire phantom images simulated by Field II: (a) 20 MHz 192 element linear array, (b)
20 MHz 192 element convex array
46

Figure 3.7. H&E stain image of a dog eye processed to the gray scale with the corresponding
scattering value for Field II simulation
47

Lateral distance [mm]
Axial distance [mm]
-5 0 5
5
10
15
20
25
30
(a)

Lateral Distance [mm]
Axial Distance [mm]
-15 -10 -5 0 5 10 15
5
10
15
20
25
30
(b)

Figure 3.8. Simulated images of the dog eye by Field II with (a) the linear array, (b) the convex
array
48

3.2.2 Elevation Focusing
A lens is generally used for off-axis, or elevation, focusing for array transducers. Lens
can be either a convex or a concave shape in the elevation direction depending upon the sound
velocity of the lens material. If the material of a lower velocity than water were used, it should be
a convex shape. RTV silicone family can be used for a convex shape lens and typically has an
acoustic impedance of approximately 1.5 MRayl, a longitudinal velocity of 1100 m/s and an
acoustic loss of 7.0 dB/cm/MHz (McKeighen, 1998). For the elevation focus at 30 mm, the lens
thickness at the center of the array should be approximately 0.4 mm and results in the attenuation
of 5.6 dB at 20 MHz, which is undesirable for high frequency array transducers.
If its velocity is higher than water, a concave shape lens should be used. Epoxies with
very low viscosity and attenuation (Epo-Tek 301, Epoxy Technologies, Billerica, MA; TPX,
Westlake Plastics Company, Lenni, PA) have higher velocity than water. Acoustic properties of
Epo-Tek 301 were summarized by Cannata et al. (Cannata, Williams, Zhou, Ritter, & Shung,
2006) and those of TPX by Erickson et al. (Erickson, Kruse, & Ferrara, 2001). The epoxy lens
casting process for linear arrays as described by Cannata et al. (Cannata, Williams, Zhou, Ritter,
& Shung, 2006) requires a cylindrical rod as a mold to generate a concave shape lens. The main
problem anticipated in this lens casting process for convex arrays is how to create a correct radius
of curvature of the lens on the curved surface of the array in the other direction. High precision
tooling and positioning jigs are required to create a concave lens without errors.
3.2.3 1-3 Composite
For this 20 MHz convex array, physical conformation in the elevation direction was tried
in order to achieve higher sensitivity by eliminating lens attenuation and also to avoid fabrication
49

error in the lens casting process. Therefore the array requires a concave curvature in the elevation
direction and a convex curvature in the azimuth direction. As a result, in order to implement the
‘saddle’ shape aperture, 1-3 piezoelectric composite was chosen for piezoelectric material.
Piezoelectric composites not only have better piezoelectric characteristic than bulk ceramics but
also can be easily conformed.
For piezoelectric material, two different high dielectric constant piezo ceramics (TRS-
HK1-HD, TRS Technologies, State College, PA; TFT-L-155N, TFR Technologies, Tokyo Japan)
were tested. Their material properties are summarized in Table 3.3.
TABLE 3.3. PROPERTIES OF PIEZOELECTRIC MATERIALS
Parameters TRS HK1-HD
*
TFT L-155N
**

] N/m 10 [
2 10
33
×
E
c
15.7 15.2
] N/m 10 [
2 10
11
×
E
c
9.97 15.6
] N/m 10 [
2 10
12
×
E
c
10.5 10.5
] N/m 10 [
2 10
13
×
E
c
13.7 11.4
] N/m 10 [
2 10
44
×
E
c
2.21 2.1
0 33
ε ε
S

2900 2320
0 11
ε ε
S

2730 2569
0 33
ε ε
T

6690 5700
0 11
ε ε
T

5260 5500
] [C/m
2
33
e
30.9 31.16
] [C/m
2
31
e
-20.8 -9.93
] [C/m
2
15
e
23.2 23.5
density] , [kg/m
3
ρ
8000 7900
[m/s]
'
33
V
4000 3947
'
33
k
0.68 0.72
* Measured by Ritter et al. (Ritter, 2000)
** Datasheet provided by TFT Technologies, Tokyo, Japan

For the kerf filler, an epoxy with a very low viscosity (Epo-Tek 301, Epoxy
Technologies, Billerica, MA) was used to allow them to be infused easily in narrow kerfs. Its
properties have been summarized by Cannata et al. (Cannata, Ritter, Chen, Silverman, & Shung,
50

2003). This epoxy is also used for a matching layer and a backing block. Since it is nontoxic
complying with USP Class VI biocompatibility standards, it can be used at the patient contact
point.
Composite kerf widths should be narrower than certain limit to minimize the coupling
between the thickness mode and lateral mode resonances (Ritter, 2000). The kerf width and the
ceramic width in 1-3 composites should be kept within the limits determined by:

c
S
f
V
width Kerf
× ×
≤
2 4
_ (3.2)

2
_
_
height Ceramic
width Ceramic ≤ (3.3)

where V
s
is the shear wave velocity in the polymer and f
c
is the device center frequency.
Although the 14 µm kerfs fabricated using the double-index-dicing technique (Cannata,
Williams, Zhou, Ritter, & Shung, 2006) do not meet the requirement for 20 MHz composites,
they were evaluated because they currently represent the narrowest kerf possible using
mechanical dicing. The 9 µm polymer kerfs fabricated using interdigital pair bonding (Liu,
Harasiewicz, & Foster, 2001) did meet the width requirements for our array composite. Lateral
resonance characteristics depending on composite kerf width were evaluated by finite element
modeling (PZFlex, Weidlinger Associates, Mountain View, CA) and confirmed by the
measurement of test pieces. Impedance characteristics of composite test pieces with the 14 µm
kerfs and the 9 µm kerfs were measured and compared. As shown in Figure 3.9, the coupling
between the thickness mode and lateral mode resonances has been reduced by using a narrower
kerf satisfying requirements (Ritter, 2000).
51

Figure 3.9. Measured impedance of 1-3 composite test pieces of (a) TRS HK1-HD (dimension:
2.0 mm × 1.0 mm × 0.075 mm) with 14 µm kerf, (b) TFT L-155N (dimension: 2.0 mm × 1.0 mm
× 0.075 mm) with 7 µm kerf (both resonating in the air, magnitude: solid line, phase: dashed line)
The characteristics of composite were calculated by the effective medium theory (Smith
& Auld, 1991). The ceramic volume fraction has been chosen to optimize the electro-mechanical
coupling coefficient given the clamped dielectric constant, density, and longitudinal velocity in
52

the acceptable range and also upon available mechanical dicing options. Calculated parameters
are summarized in Table 3.4 and composite fabrication parameters are displayed in Figure 3.10.
TABLE 3.4. PROPERTIES OF COMPOSITES
Parameters
Option 1
TRS HK1-HD
Option 2
TFT L-155N
Option 3
TFT L-155N
Ceramic post width [µm] 36 28 30.1
Polymer kerf width [µm] 14 9 7
Element pitch [µm] 100 111 111.3
Ceramic volume fraction [%] 52 57 66
Number of ceramic posts in
an element
2 3 3
density] , [kg/m
3
ρ
4712 4998 5605
0 33
ε ε
S

1317 1370 1584
t
k
0.74 0.71 0.70
[m/s]
l
V
4326 4044 4105
[MRayl] Z
20.4 20.2 23.0

53

Figure 3.10. Design of 1-3 composites: (a) Option 1 for double index dicing, (b) Option 2 for
interdigital pair bonding, (c) Option 3 for interdigital pair bonding with a wider dicing blade
(length in µm)
54

3.2.4 Array Structure
Since the 1-3 composite gives much lower impedance than the bulk ceramic,
approximately 50% in this case, only one matching layer was needed. Two acoustic matching
layers were not necessary to achieve a desired –6 dB bandwidth over 50%. A custom made
flexible printed circuit board is bonded on the backing side of 1-3 composite, which is scratch-
diced for element separation and it is supported by a backing block. A matching layer is bonded
on the front surface of 1-3 composite and provides the ground of electrical path. Micro-
connectors (BTH-150-01-L-D-A, Samtec, Inc., New Albany, IN) mounted on a flex circuit are
connected by a custom made cable assembly (40AWG/75 Ω/256 core coaxial Comfort Cable,
Tyco Electronics, Berwyn, PA).
3.2.5 KLM Modeling
KLM modeling (PiezoCAD, Sonic Concepts, Woodinville, WA) was used to predict the
performance of a single array element. Optimized KLM modeling parameters and results are
summarized in Table 3.5. The pulse-echo signal and its spectrum for Option 2 are shown in
Figure 3.11. The –6 dB fractional bandwidth of 65%, the –6 dB center frequency of 20.0 MHz,
the –20 dB pulse width of 0.116 µs, and the loop sensitivity of –45 dB were predicted.

55

TABLE 3.5. DESIGN PARAMETERS OF 20 MHZ CONVEX ARRAY TRANSDUCERS
Parameters Option 1 Option 2 Option 3
Input
1-3 composite
Ceramic material TRS HK1-HD TFT L-155N TFT L-155N
Polymer material Epo-Tek 301 Epo-Tek 301 Epo-Tek 301
Thickness [µm] 75 71 75
Ceramic volume fraction [%] 52 57 66
Matching layer
Material Epo-Tek 301 Epo-Tek 301 Epo-Tek 301
Thickness [µm] 30 30 30
Backplates
Material Copper/Polyimide Copper/Polyimide Copper/Polyimide
Thickness [µm] 5/50 5/25 5/25
Backing
Material Epo-Tek 301 Epo-Tek 301 Epo-Tek 301
Flex circuit thickness [µm] 50 25 25
Output
Loop sensitivity [dB] -47.2 -45.2 -45.8
-6 dB center frequency [MHz] 19.3 20.0 20.6
-6 dB fractional bandwidth [%] 69.6 65.3 61.6
-20 dB pulse width [µs] 0.118 0.116 0.130

Figure 3.11. The KLM pulse-echo response (solid line) and its spectrum (dashed line) for a single
20 MHz array element.
56

3.2.6 Flexible Circuits
The flexible printed circuit board (flex circuit) provides the electrical interconnection
from the cable assembly to this array transducer. The flex circuit should be designed to meet
acoustical, electrical and mechanical specifications. Since it is bonded between the piezoelectric
composite and the backing block, which is in the acoustic path, the flex circuit should be
acoustically transparent. The material for the flex circuit was chosen polyimide (Kapton, Dupont,
Circleville, OH) and its thickness was 25 µm. Copper layer of 5 µm was bonded on a polyimide
film and a flash of gold was electroplated on top of copper layers. The flex circuit layers of
polyimide and copper were incorporated in KLM modeling as listed in Table 3.5.
The alignment between signal traces on the flex circuit and the diced pattern on the
composite is crucial since the misalignment will result in open or short elements of the array and
it is very hard to repair when that happens. Its trace width was 38 µm, narrower than the
composite element width of 97 µm, to be placed with enough margin to help avoid registration
errors.
Electrical crosstalk can be generated from all of the components delivering electrical
signal, namely flex circuit, micro-coax cables, micro connectors or connectors to the system
front-end. In the flex circuit, to minimize this electrical crosstalk, intermediate grounds were
placed in every other elements and they were all connected. Ground pads surrounding the flex
circuit were created and unused pins were also connected to the ground.
Another main design criterion in the flex circuit was how to reduce the tension at the
edges when it was folded perpendicular to the curved aperture. A series of 12 elements formed a
group of traces and those groups were divided half and routed to the left and right sides of the
57

circuit. It could create spaces in between traces and after laser cutting those spaces, let them serve
as strain relief slits. Figure 3.12 shows above mentioned design considerations.

Figure 3.12. A group of traces to generate laser slits in the flexible circuit to serve as a strain
relief and ground patterns in the final assembly for 20MHz convex arrays.

58

3.3 Fabrication of Convex Arrays
3.3.1 Fabrication Options
Given the array structure of a single matching layer, a composite, a flex circuit and a
backing block, four different fabrication options have been considered. The ultimate goal in
searching for fabrication options is to create a two-way curved aperture upon the desired radius of
curvature without creating any open or short elements of the array.
First, the “bending and bonding” option was considered. Conformed matching layer and
composite as shown in Figure 3.13 (a) are bonded onto the flex circuit, which was pre-bonded on
the backing block. Composite layers are conformed by a conforming fixture as shown in Figure
3.13 (b). This conforming fixture was designed to have a radius of curvature of 23.6 mm on the
front surface of composite and the gap between the upper press and the bottom hold was 75 µm to
host a composite of that thickness. The radius of curvature in the elevation direction was 30.0 mm,
which is the focal depth of the array. A conformed composite by this fixture is shown in Figure
3.13 (c). The alignment between curved layers is more difficult than flat ones since the curved
edges should be aligned and heat shrinkage of a composite may generate uneven contours that do
not match well with those of flex circuits.
59

Figure 3.13. “Bending and bonding” fabrication option: (a) conformed layers to be bonded, (b) a
composite conforming structure, (c) a conformed composite.
60

The second option is a “backbone” approach. A 200 µm thick backplate is used to
support layers. The flex circuit and the composite are bonded on the backplate and diced from the
composite surface to cut into the halfway through the back plate as shown in Figure 3.14. The
conformation is easier since bonded layers are front-diced. The element-to-element acoustic
separation is better since each element is completely separated by front-dicing and kerf can be
filled by softer material to compensate unwanted vibration. The misalignment that may be caused
in the bonding and conforming between the flex circuit and the composite does not happen since
it is already bonded and cut.
The third option is a “side attachment” way. In this option, the backplate is sputtered and
pre-diced. The flex circuit is cut in half as shown in Figure 3.15. Two half-cut flex circuits are
bonded at the sides of the backplate. Conformed composite is bonded on the backplate and a
matching layer is added on top of that. This option may show better acoustic characteristic since
the polyimide and copper/gold sputtered flex circuit is removed from the acoustic path.
The last option that actually used in the fabrication of prototypes is the simplest way,
“bonding and bending” as shown in Figure 3.16. All the layers were bonded first and conformed
later. The misalignment was minimal for the desired radii of curvature, 23.6 mm in a convex way
along the azimuth direction and 30.0 mm in a concave way along the elevation direction.

61

Figure 3.14. “Backbone” fabrication option: Pre-bonded and diced composite / flex circuit /
backplate layers to be conformed by a conforming fixture.
62

Figure 3.15. “Side attachment” fabrication option: Half-cut flex circuit to be bonded on the
sputtered, pre-diced and conformed backplate.
63

Figure 3.16. “Bonding and bending” fabrication option: Pre-bonded matching / composite / flex
circuit layers to be conformed.

64

3.3.2 1-3 Composites by Interdigital Pair Bonding
Three different 1-3 composites were fabricated as specified in Table 3.4, one by a double-
index-dicing (Cannata, Williams, Zhou, Ritter, & Shung, 2006) (Option 1) and others by an
interdigital pair bonding technique (Liu, Harasiewicz, & Foster, 2001) (Options 2 & 3).
For the interdigital pair bonding fabrication of 1-3 composites, four identical ceramic
plates (25.4 mm × 25.4 mm × 0.5 mm) were initially prepared. The ceramic dicing width, W
c
and
the polymer dicing width, W
p
were determined to create designed kerf width, W
k
=(W
p
-W
c
)/2 as
specified in Options 2 & 3 in Table 3.4. First, all four ceramic plates were diced as shown in
Figure 3.17 (a) and then interdigitated as in (b). The space generated by the difference between
the ceramic dicing width, W
c
and the polymer dicing width, W
p
, was then filled and cured by
polymer. Cured ceramic/polymer plates were lapped to expose ceramic and polymer strips as in
(c). Then, those two ceramic/polymer plates were diced along the other direction with same
dicing widths as in (d). The final ceramic/polymer plate was then filled and cured by polymer in
the same fashion as in (e). Both sides were lapped down to reach the final thickness of composite.
Figure 3.18 shows the final shape of finished 1-3 composite fabricated by double-index
dicing (a) and by interdigital pair bonding (b). The electro-mechanical coupling coefficients were
measured for finished 1-3 composites and measured values were approximately 15% lower than
theoretically calculated values, i.e. 0.60 to 0.71. One of contributing factors for this loss of
coupling may be micro-cracks at grain boundaries previously reported by Nix et al. (Nix, Corbett,
Sweet, & Ponting, 2005) as shown in Figure 3.19 (a). An SEM image of a fabricated composite
(Figure 3.19 (b)) by Option 1 does not show severe cracks as in (a) but the possibility of micro-
cracks in the area that the dicing blade passed still exists and could adversely affect to the
65

composite performance. Both sides of the composites were sputtered with a 500 Å chrome and
2000 Å gold electrode and one side was scratch- diced to electrically separate elements.

Figure 3.17. Interdigital pair bonding technique to fabricate 1-3 composites (W
c
: ceramic dicing
width, W
p
: polymer dicing width, W
k
=(W
p
-W
c
)/2: kerf width) from Liu et al. (Liu, Harasiewicz, &
Foster, 2001)
66

Figure 3.18. Pictures of finished 1-3 composites: (a) SEM image of Option 1: double-index dicing,
(b) microscope image of Option 3: interdigital pair bonding
67

Figure 3.19. Micro-cracks at grain boundaries: (a) from (Nix, Corbett, Sweet, & Ponting, 2005),
(b) Option 1
68

3.3.3 Matching Layers and Backing Blocks
A matching layer was cast in a mold, lapped down to the desired thickness, and sputtered
with a 500 Å chrome and 2000 Å gold electrode on one side. The electroplated matching layer
would serve as a ground plane for the array. A backing block was made from a silicone RTV
(RTV664, Sylpak, Inc., Pomona, CA) mold as described in Figure 3.20. An aluminum backing
block positive were glued on the glass plate and surrounded by glass plates to make dams as in
(a). Silicone RTV was filled and cured as in (b) and then an epoxy backing block (d) was made
from a silicone RTV (c).

Figure 3.20. Process to make a backing block: (a) an aluminum positive surrounded by dams, (b)
degassed RTV silicone rubber poured in the space, (c) a finished RTV mold, (d) a finished
backing block

69

3.3.4 Bonding and Conforming Layers
All the layers, a composite, a matching layer, a flex circuit, and a backing block, have
been cleaned by the plasma etching system (PX-250, March Instruments, Inc., Concord, CA)
before bonding with Argon gas under 25W power for 3 minutes to enhance bonding strength.
First, the composite was bonded onto the flex circuit. The signal traces on the flex circuit
were aligned on the scratch-diced pattern on the composite. Depending upon the amount of heat
exposed to the composite during fabrication, the actual pitch of composite may be decreased, and
therefore care should be taken not to generate any open connections. The amount of epoxy used
for bonding layers should be controlled as minimal not to be squeezed out excessively between
layers. Squeezed epoxy, if they were cured on the surface of flex circuit, will create an uneven
surface, which may hinder bonding next layers. Figure 3.21 shows a bonded composite and flex
circuit. A zoomed image shows the pitch of composite is smaller than that of signal traces of flex
circuit. This composite, Option 3 has a wider pitch, 111.3 µm than Option 2, 111.0 µm by 0.3 µm
but it turned out the heat shrinkage caused its actual pitch is even smaller than the pitch of signal
traces of flex circuit. Figure 3.21 demonstrates that the enlarged pitch compensates well the heat
shrinkage and as a result, the composite and flex circuit were well aligned without causing any
short or open elements.
The bonded composite and flex circuit was conformed by a saddle-shaped fixture at 90°C
and then bonded on the backing block as shown in Figure 3.22. The matching layer with ground
connection was later bonded to the array and flexible circuit. For Option 3, even matching layer
was pre-bonded before conforming as described in Figure 3.16. The epoxy used for all bonding
steps was Epo-Tek 301.
70

Figure 3.21. Picture of a composite bonded on the flex circuit, signal traces of the flex circuit
aligned with an array of elements of the composite
71

Figure 3.22. Conforming and bonding of matching layer/composite/flex circuit onto a backing
block
3.3.5 Finished Arrays
Figure 3.23 is the photograph of a finished 20 MHz convex array transducer (Option 2).
Composite, flex circuit and matching layer were all conformed and bonded to a backing block
with the saddle-shaped surface. Laser cut slits were used to give relief to flexible circuits when
they were bent perpendicular to the array transducer surface during the packaging process. The
finished array was poled by the high voltage power supply at 3V
DC
/µm for 5 minutes and its final
characteristics were measured.
72

Figure 3.23. A picture of the finished 20 MHz convex array successfully built to create two-way
curved aperture

73

3.4 Characterization of Convex Arrays
3.4.1 Quantitative Testing of Convex Arrays
Array transducers will be ultimately characterized by the optimized clinical images with
the scanning system. That involves quantitative measurements of the array transducer and the
system front-end circuit and the acoustical and electrical matching of those system front-end and
transducer. Electrical impedance, pulse-echo response and its spectrum, insertion loss, and
crosstalk are major quantitative parameters often provided. Images of wire phantom by synthetic
aperture technique are also preferred to evaluate initial quality of images when the system front-
end is not yet available. Array transducers are usually fully characterized, element-by-element in
the commercial production stage. Loop sensitivity, center frequency, bandwidth, fractional
bandwidth, pulse length and time-of-flight from the transducer surface to the geometrical focal
point obtained by the pulse-echo test are major parameters controlled by target specifications and
preset ranges.
Pulse-echo tests for convex arrays require a curved reflector target unlike linear arrays or
phased arrays which need a flat reflector only. The radius of curvature of such target should be
that of convex array plus the elevation focal length. If, either the radius of curvature of a convex
array was not correct due to fabrication error or the radius of curvature of a curved target was
wrong, then the evaluation of maximum echoes from the geometric focal point would not be
possible as intended. A point target mechanically translated along the arc at the geometrical focal
point will be suitable for the pulse-echo test of the convex array and it requires significant amount
of time for full characterization if they are not automated. A stainless steel curved target were
74

fabricated as in Figure 3.24. Pulse-echo responses of all 192 elements have been obtained and
analyzed.

Figure 3.24. Stainless steel curved target for pulse-echo test of convex arrays with the radius of
curvature of 53.6 mm to place the reflector at the focal depth
3.4.2 Electrical Impedance
The electrical impedance was measured using an impedance analyzer (Agilent 4294A,
Agilent Technologies, Santa Clara, CA). Figure 3.25 shows the electrical impedance magnitude
and phase of an element of the finished array fabricated by Option 2. The main resonance was at
a higher frequency than expected and led to the higher pulse-echo center frequency for this array
transducer.
75

Figure 3.25. Measured impedance magnitude and phase of an element of a finished 20 MHz
convex array (Option 2)
3.4.3 Pulse-Echo Response
The pulse-echo responses were acquired for typical array elements after connecting the
finished acoustic array module to the custom made cable assembly (40AWG/75 Ω/256 core
coaxial Comfort Cable, Tyco Electronics, Berwyn, PA). The ultrasonic analyzer (5900PR,
Panametrics Inc., Waltham, MA) was used to excite the transducers, and a digital oscilloscope
(LC534, LeCroy Corp., Chestnut Ridge, NY) was used to acquire the echo reflected from a quartz
target placed at the focal point. Pulse-echo tests for an array built by Option 2 showed 15 open
elements and 0 shorted elements. The time-domain waveform and frequency-domain spectrum of
an element of a full size array (Option 2) are shown in Figure 3.26. By implementing narrower
composite kerfs with interdigital pair bonding, we could achieve a much shorter ring down time,
76

0.106 µs compared to 0.240 µs of Option 1. The higher –6 dB center frequency of 24.4 MHz for
the full size array was due to lapping the composite slightly thinner than desired. The –6 dB
fractional bandwidth for the Option 2 array was 64% and the spectrum shape was close to
Gaussian. The axial peak was found at 28.0 mm for Option 2 and was within an acceptable range
of the designed focus at 30.0 mm.
Pulse-echo test for all 192 elements was done for an array built by Option 3. A curved
target shown in Figure 3.24 was used as a reflector. It was aligned at the focal depth to generate a
maximum echo from each element. Acquired waveforms from individual elements were stored
and analyzed to calculate loop sensitivity (LS), –6 dB center frequency (Fc
-6
), –6 dB bandwidth
(BW
-6
), –6 dB fractional bandwidth (FBW
-6
), –20 dB pulse width (PW
-20
), and time of flight
(TOF). The typical end-of-load voltage level of 100V from the ultrasonic analyzer (5900PR,
Panametrics Inc., Waltham, MA) was assumed in the loop sensitivity calculation and also the
attenuation and gain of the analyzer were considered. Plots of pulse-echo responses in Figure 3.27
and the summary in Table 3.6 show an acceptable uniformity in element-to-element performance
variation.
77

Figure 3.26. Pulse-echo test results of an element of a finished 20 MHz convex array (Option 2):
(a) an echo waveform at the focus, (b) its spectrum
TABLE 3.6. SUMMARY OF PULSE-ECHO TEST RESULTS OF A 20 MHZ CONVEX ARRAY
TRANSDUCER
Parameters LS [dB] Fc
-6
[MHz] BW
-6
[MHz] FBW
-6
[%] PW [µs] TOF [µs]
Average -63.7 19.6 13.6 69.2 0.18 39.5
Maximum -61.7 20.5 15.9 78.8 0.27 39.5
Minimum -68.4 18.8 12.3 62.2 0.11 39.5
Std. Dev. 1.3 0.4 0.7 3.5 0.02 0.0

78

Figure 3.27. Pulse-echo test results for all 192 elements of 20 MHz convex array transducers:
Loop sensitivity [dB], -6 dB center frequency [MHz], -6 dB bandwidth [MHz], -6 dB fractional
bandwidth [%], -20 dB pulse width [µs], time of flight [µs] vs. element number
79

3.4.4 Crosstalk
The interference from adjacent elements makes its effective element size bigger than
designed and results in narrower angular response of array elements. Therefore, the crosstalk,
element-to-element separation is one of main issues of convex array transducers as well as phased
arrays using beam steering since convex arrays have wider field of view than linear arrays and so
the wider acceptance angle is needed. For this measurement the array surface was covered by a
coupling gel to protect and a function generator (AFG3251, Tektronix, Inc., Anaheim, CA), set in
burst mode, was used to excite a representative element with the applied voltage measured as a
reference. Voltages across the nearest and the next nearest elements were measured and compared
to this reference voltage. Crosstalk was measured for the finished array (Option 3) in the range of
10-50 MHz. The maximum crosstalk was –21 dB for the nearest element. Generally acceptable
crosstalk level for imaging situations is approximately –30 dB (McKeighen, 1998). One of
contributing factors why this prototype array shows a higher crosstalk level may be insufficient
grounding of the flex circuit. Ground pads were placed surrounding the flex circuit and unused
connector pins were connected together and routed to the ground pads but still the electrical path
from each element to the ground was not short enough and generated negative effects to element-
to-element separation. The measured crosstalk was therefore obtained after adding another
ground plane on top of the flex circuit by sputtering 5000 Å gold layer. However, in the next
version of flex circuit, an added 5µm thick copper ground plane connected from each element
through vias will reduce crosstalk level significantly. A “backbone” approach described in Figure
3.14 may help reduce the crosstalk between elements since the front dicing separating elements
physically can reduce acoustic crosstalk considerably.

80

3.4.5 Insertion Loss
The measured two-way insertion loss for a representative array was 45.0 dB. After
compensating for attenuation in the water path, reflective loss in the quartz target, and a
diffraction loss assuming doubled effective element width due to a high level crosstalk (Selfridge,
1983), an estimated insertion loss was 18.8 dB at 24 MHz for the full size array (Option 2). This
crosstalk level is comparable to those of linear array transducers previously built by Ritter et al.
(Ritter, Shrout, Turwiler, & Shung, 2002) and Cannata et al. (Cannata, Williams, Zhou, Ritter, &
Shung, 2006).
3.4.6 Images
Images of a custom-made wire target phantom drawn in Figure 3.28, composed of five
20- μm-diameter tungsten wires (California Fine Wire Company, Grover Beach, CA) were
acquired with the finished array. When the 64-channel beamforming system is available, the
expected lateral and axial resolutions predicted by Field II simulation are 310 µm and 67 µm
respectively. Due to absence of an imaging system for this array, a synthetic aperture imaging
technique (Karaman, Li, & O'Donnell, 1995) was used to ascertain imaging capability of the
array. For the imaging, one center element of a 16-element subgroup was excited and
subsequently one element in the subgroup was used to receive echoes. This procedure was
repeated until all elements in the subgroup participated in receiving echoes produced by the
center element. Since all the data acquisition has been done manually, i.e. one element in transmit
and one element in receive, only 64 elements at the center, from element number 65 to 128, were
used for this synthetic imaging experiment. Therefore, there were 64 subgroups available and
each subgroup had 16 radio frequency (RF) data sets; approximately, total 1000 RF data have
81

been recorded. For this, two ultrasonic analyzers (5900PR, Panametrics Inc., Waltham, MA) were
used for the transmit/receive procedures. One scanline was formed from adjacent 32 subgroup
data by applying time delays corresponding to receive and synthetic focusing for the scanline.
Figure 3.29(a) is the acquired image with a 40 dB dynamic range. It is displayed by a linear gray
scale. The image of five wires clearly demonstrates that our convex array is capable of providing
a wider field of view. If it were a linear array with the same number of elements, 64, we would be
able to see the center wire only and a shadow of the second wire. When the full size image is
implemented, the lateral distance covered by a single image plane at the geometric focus, 30 mm
will be 46 mm. A linear array with the same pitch and number of elements can cover the lateral
distance of 21 mm, which is less than half of the distance covered by a convex array. Figure 3.30
shows plots of the lateral and axial line spread functions for the center wire. The measured –6 dB
beamwidths acquired after interpolating those line spread functions were 175 μm and 60 μm for
the lateral and axial directions, respectively. These measurements match well with the resolutions
of 173 μm in the lateral direction and 60 μm in the axial direction simulated by Field II shown in
Figure 3.31. Higher sidelobe level showed in the lateral beam spread function in Figure 3.29 (a) is
inevitable from the synthetic aperture imaging technique used in this experiment since one
subgroup consisting of 16-receiving elements was used for receive focusing and 32 subgroup
were synthesized for one scanline. Multiplication of wide receiving and narrow synthesizing
beam spread functions brings about higher sidelobes although the mainlobe width becomes
narrower.
82

Figure 3.28. A simplified drawing of a wire phantom used for synthetic imaging experiments

83

Figure 3.29. Synthetic aperture images of a wire phantom by (a) measured RF data and (b) Field
II simulation. The dynamic range of the image is 40 dB and the display uses a linear gray scale
for mapping.
84

Figure 3.30. Line spread functions for the center wire of the image, Figure 3.29(a) synthetic
aperture image from measured RF data: (a) lateral and (b) axial.
85

Figure 3.31. Line spread functions for the center wire of the image, Figure 3.29(b) synthetic
aperture image from Field II simulation: (a) lateral and (b) axial.
86

Different synthetic aperture imaging techniques have been tried to search the way to
improve spatial resolution. Different wire phantom with a smaller interval between wires were
used for imaging. For the imaging, one element was excited and the echo RF data from that
element was recorded. To limit the effect of grating lobes, array elements were used for
reconstruction only if they are within an acceptance angle ±9° for a point in the formed image.
Backprojection to each pixel was accomplished by the following formula (Ritter, Shrout,
Turwiler, & Shung, 2002):

∑
=
⎥
⎦
⎤
⎢
⎣
⎡
+ − − =
N
e
o o e e e i i
z x x
c
t R w z x P
1
2 2
) (
2
) , ( (3.5)

where , is the location of the pixel in the image plane, e is the index of the element over the
range of 1 to N (number of elements), is the apodization function, is the time domain
response,
i
x
i
z
e
w
e
R
t is the time, c is the propagation velocity, , is the location of the point in object
space, and is the position of the array element. Since it is using only 64 RF data sets, it is
useful for checking the imaging capability without extensive data collection. However, it is not
using a receive focusing and therefore the spatial resolution of Figure 3.32 is not as good as
Figure 3.29. The measured –6 dB beamwidths acquired after interpolating those line spread
functions were 382 μm and 71 μm for the lateral and axial directions, respectively. The dynamic
range of the image is 40 dB and the display uses a linear gray scale for mapping. No apodization
or thresholding was implemented during reconstruction.
o
x
o
z
e
x
Images of ex vivo porcine eye were also obtained. The lens of the porcine eye was cut out
to remove the attenuation from the lens. It was particularly necessary for this imaging experiment
since only one element can be used for transmit. For the imaging, 80 out of 192 elements were
87

chosen. One element was excited and the echo was received from that excited element. Scanlines
were synthesized by adjacent ones with the maximum of 64 lines. Fundamental imaging with a 19
MHz, 142V peak-to-peak, two-cycle sinusoid was tried. Pulse inversion tissue harmonic imaging
was also tried (Averkiou, 2000). For harmonic imaging, a 11 MHz, 142V peak-to-peak, three-
cycle sinusoid was used for excitation and for each element, 180° phase shifted signal followed to
cancel out the main frequency component later in post-processing. A 10 kHz – 250 MHz power
amplifier (75A250A, Amplifier Research, Souderton, PA) was used for amplification of transmit
signals. The pulse inversion tissue harmonic image shown in Figure 3.33 (b) provides better
lateral resolution and less artifacts than the fundamental image in Figure 3.33 (a). When the
imaging system is available with transmit beam focusing, more clinically useful images with
better spatial resolution would be obtained.

Figure 3.32. Synthetic aperture image of a wire phantom reconstructed using a half-angle of
9°and no apodization or thresholding. The dynamic range of the image is 40 dB and the display
uses a linear gray scale for mapping.

88

Figure 3.33. Synthetic aperture images of a porcine eye: (a) fundamental imaging with a 19 MHz
tone burst, (b) pulse inversion tissue harmonic imaging with a 11 MHz tone burst
89

90

3.5 Advantages of Convex Arrays
High frequency 20 MHz convex array transducers have been designed, fabricated and
characterized. The flexibility of 1-3 composite made it possible to conform the array elements in
two different directions into a saddle shape without cracking or distortion. Strain relief slits in the
flexible circuits helped to remove excessive tension at the edges and enabled a compact
packaging of the array. The narrower composite kerf width achieved by interdigital pair bonding
significantly improved both time-domain and frequency-domain characteristics of the array’s
two-way response and we could observe that the echo after-ringing was reduced substantially. It
was found that composite heat shrinkage could be compensated by implementing a larger dicing
pitch of the composite. It worked successfully to prevent misalignment between signal traces of
flexible circuits and the scratch-diced composite. It is believed that high frequency convex array
transducers may lead us to new clinical imaging opportunities. Imaging experiments including
clinical evaluation will follow when supporting imaging system becomes available.
CHAPTER 4 SUMMARY AND FUTURE WORK
4.1 Summary
Two different types of transducers were proposed for imaging ocular tissues of the
human eye. First, 20 MHz/40 MHz dual element transducers for harmonic imaging were designed,
fabricated, and tested. The concentric type dual elements packaged in a single housing worked
well as a dedicated harmonic imaging transducer. Although the ring shape of the transmit element
resulted in increased sidelobe levels, the advantage of harmonic imaging in suppressing sidelobes
well compensated its shortcoming to achieve an improved lateral resolution. It also proved to be
useful for frequency compounding due to its wideband combined from two different frequency
spectra. Harmonic imaging with 20MHz transmit and 40 MHz receive demonstrated superior
capability than fundamental imaging at 20 MHz to diagnose the retinal diseases at the posterior
segment.
High frequency convex array transducers have been designed, fabricated, and
characterized. It was demonstrated that convex array transducers of 20 MHz with an aperture
curved both in azimuth and elevation direction could be fabricated with an acceptable uniformity
in element-by-element performance. Different fabrication options have been examined and a
simple bonding and conforming approach worked successfully to fabricate prototype arrays. All
192 elements of the array were fully characterized by the pulse-echo test, crosstalk measurement,
insertion loss test and synthetic aperture imaging. Average loop sensitivity was -63.7 dB and
average –6 dB fractional bandwidth 69.2%, which are acceptable for imaging purpose. Ringing of
echoes was substantially reduced by using a lower composite kerf by the interdigital pair bonding
91

for fabricating 1-3 composites. Created images of ex vivo tissues with a trapezoidal field of view
provided wider view angle than linear arrays.
4.2 Future Work
The ultimate goal of 20 MHz convex arrays in ophthalmic imaging is to differentiate
retina/choroid/sclera layers, image the optic nerve, and get the color flow in retinal blood vessels.
Its higher frequency version, 30 – 40 MHz will be investigated to achieve better spatial resolution
and also a wider view angle version with a smaller radius of curvature will also be studied. The
flexibility of 1-3 composite has a great potential to be conformed well enough for a tighter
curvature. More advanced fabrication techniques should follow to meet such needs. A thinner
flex circuit made by 12.5 µm thick polyimide will be tried and also listed fabrication options in
this research. Although the primary application is at ophthalmic imaging, high frequency convex
arrays can be also used for other imaging applications. Different frequencies and different
geometries will be investigated to meet each clinical need. As convex arrays replaced linear
arrays in the low frequency range, 2.0 to 5.0 MHz in early 1980s, convex arrays from 20 to 40
MHz can change the high frequency imaging market by providing clinical advantages over
currently available linear arrays.
Photoacoustic imaging which combines optics and ultrasound has been recently emerged
as a complementary imaging technique. 20 MHz ring shape ultrasonic transducers were used as a
receiver in a compact monolithic device, which allows focused laser and ultrasonic beams to be
launched collinearly. This concentric structure enabled a far extended working distance and a
significantly improved lateral resolution which made imaging choroid/retina layers of the
posterior segment of the eye.
92

Dual element transducers and convex arrays above mentioned will be the next steps for
this type of photoacoustic imaging. The same laser source can be placed at the center of dual
element transducers and these dual frequency receivers can implement harmonic imaging or
frequency compounding. As the concentric laser/ultrasound transducer could identify layers
which were not obtainable with the pulse-echo 20 MHz ultrasound, 20 MHz/40 MHz harmonic
imaging could provide even higher resolution to delineate layers. Convex arrays also can provide
wide view images with multiple focus capability.

93

BIBLIOGRAPHY
Averkiou, M. A. (2000). Tissue harmonic imaging. Proc. IEEE Ultrasonics Symposium 2000, 2,
pp. 1563-1672.
Baddour, R. E., Sherar, M. D., Hunt, J. W., Czarnota, G. J., & Kolios, M. C. (2005). High-
frequency ultrasound scattering from microspheres and single cells. J. Acoust. Soc. Am. , 117 (2),
934-943.
Brown, J. A., Foster, F. S., Needles, A., Cherin, E., & Lockwood, G. R. (2007). Fabrication and
performance of a 40-MHz linear array based on a 1-3 composite with geometric elevation
focusing. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 54 (9), 1888-1894.
Cannata, J. M., Ritter, T. A., Chen, W. -H., Silverman, R. H., & Shung, K. K. (2003). Design of
efficient, broadband single-element (20-80 MHz) ultrasonic transducers for medical imaging
applications. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 50 (11), 1548-1557.
Cannata, J. M., Williams, J. A., Zhou, Q. F., Ritter, T. A., & Shung, K. K. (2006). Development
of a 35-MHz piezo-composite ultrasound array for medical imaging. IEEE Trans. Ultrason.
Ferroelect. Freq. Contr. , 53 (1), 224-236.
Cannata, J. M., Williams, J. A., Zhou, Q. F., Sun, L., Shung, K. K., Yu, H., et al. (2008). Self-
focused ZnO transducers for ultrasonic biomicroscopy. Journal of Applied Physics , 103 (8), art.
084109.
Chang, J. H., Kim, H. H., Lee, J., & Shung, K. K. (2010). Frequency compounded imaging with a
high-frequency dual element transducer. Ultrasonics , 50, 453-457.
Coleman, D. J., Silverman, R. H., Chabi, A., Rondeau, M. J., Shung, K. K., Cannata, J. M., et al.
(2004). High-resolution ultrasonic imaging of the posterior segment. Ophthalmology , 111 (7),
1344-1351.
Duck, F. A. (2002). Nonlinear acoustics in diagnostic ultrasound. Ultrasound in Med. & Biol. ,
28, 1-18.
94

Erickson, S., Kruse, D., & Ferrara, K. (2001). A hand-held, high frequency ultrasound scanner.
Proc. IEEE Ultrasonics Symposium, (pp. 1465-1468).
Food and Drug Administration. (1997). Information for manufacturers seeking marketing
clearance of diagnostic ultrasound systems and transducers. Center for Devices and Radiological
Health. Food and Drug Administration.
Foster, F. S., Mehi, J., Lukacs, M., Hirson, D., White, C., Chaggares, C., et al. (2009). A new 15-
50 MHz array-based micro-ultrasound scanner for preclinical imaging. Ultrasound in Med. &
Biol. , 35 (10), 1700-1708.
Foster, F. S., Pavlin, C. J., Lockwood, G. R., Ryan, L. K., Harasiewicz, K. A., Berube, L., et al.
(1993). Principles and applications of ultrasound backscatter microscopy. IEEE Trans. Ultrason.
Ferroelect. Freq. Ctrl. , 40 (5), 608-616.
Foster, F. S., Zhang, M. Y., Zhou, Y. Q., Liu, G., Mehi, J., Cherin, E., et al. (2002). A new
ultrasound instrument for in vivo microimaging of mice. Ultrasound in Med. & Biol. , 28 (9),
1165-1172.
Gawande, A. (2004). Casualty of war - military care for the wounded from Iraq and Afghanistan.
New England Journal of Medicine , 351, 2471-2475.
GE Healthcare. (2010). GE Vscan Datasheet. Retrieved from http://www.gehealthcare.com/
Huang, D., Swanson, E., Lin, C., Schuman, J., Stinson, W., Chang, W., et al. (1991). Optical
Coherence Tomography. Science , 254, 1178-1181.
International Electrotechnical Commission. (2007). IEC 60601-2-37, Medical electrical
equipment - Part 2-37: Particular requirements for the basic safety and essential performance of
ultrasonic medical diagnostic and monitoring equipment.
Ito, Y., Kushida, K., Sugawara, K., & Takeuchi, H. (1995). A 100-MHz ultrasonic transducer
array using ZnO thin films. IEEE Trans. Ultrason. Ferroelect. Freq. Ctrl. , 42 (2), 316-324.
95

Jensen, J. A. (1996). Field: A program for simulating ultrasound systems. Med. Biol. Eng.
Comput. , 34 (1), 351-353.
Karaman, M., Li, P. -C., & O'Donnell, M. (1995). Synthetic aperture imaging for small scale
systems. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 42 (3), 429-442.
Kettering, J. A., Aristizabal, O., Turnbull, D. H., & Lizzi, F. L. (2005). Design and fabrication of
a 40-MHz annular array transducer. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 52 (4), 672-
681.
Kettering, J. A., Ramachandran, S., & Aristizabal, O. (2006). Operational verification of a 40-
MHz annular array transducer. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 53 (3), 623-630.
Kim, H. H., Cannata, J. M., Liu, R., Chang, J. H., Silverman, R. H., & Shung, K. K. (2008). 20
MHz/40 MHz dual element transducers for high frequency harmonic imaging. IEEE Trans.
Ultrason. Ferroelect. Freq. Contr. , 55 (12), 2683-2691.
Klein Biomedical Consultants, Inc. (2007). The medical diagnostic ultrasound market in the
USA: Challenges & opportunities in the new millennium, 2006 report.
Kong, F., Chen, Y. C., Lloyd, H. O., Silverman, R. H., Kim, H. H., Cannata, J. M., et al. (2009).
High-resolution photoacoustic imaging with focused laser and ultrasonic beams. Appl. Phys.
Letters , 94, 033902.
Liang, H. -D., & Blomley, M. J. (2003). The role of ultrasound in molecular imaging. Br. J.
Radiol. , 76, S140-S150.
Liu, J. B., & Goldberg, B. B. (1999). 2-D and 3-D endoluminal ultrasound: vascular and
nonvascular applications. Ultrasound in Med. & Biol. , 25, 159-174.
Liu, R., Harasiewicz, K. A., & Foster, F. S. (2001). Interdigital pair bonding for high frequency
(20-50MHz) ultrasonic composite transducers. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. ,
48 (1), 299-306.
96

Lockwood, G. R., Turnbull, D. H., & Foster, F. S. (1995). Fabrication of high frequency
spherically shaped ceramic transducers. IEEE Trans. Ultrason., Ferroelect., Freq. Contr. , 41 (2),
231-235.
McKeighen, R. E. (1998). Design guidelines for medical ultrasonic arrays. SPIE Intern. Symp.
Med. Imaging.
Merks, E. J., Bouakaz, A., Bom, N., Lancee, C. T., van der Steen, A. F., & de Jong, N. (2006).
Design of a multilayer transducer for acoustic bladder volume assessment. IEEE Trans. Ultrason.
Ferroelect. Freq. Contr. , 53 (10), 1730-1738.
Nakamura, K., Fukazawa, K., Yamada, K., & Saito, S. (2006). An ultrasonic transducer for
second harmonic imaging using a LiNbO3 plate with a local ferroelectric inversion layer. IEEE
Trans. Ultrason. Ferroelect. Freq. Contr. , 53 (3), 651-655.
National Electrical Manufacturers Association. (2004). Standard for real-time display of thermal
and mechanical acoustic output indices on diagnostic ultrasound equipment, Revision 2.
Nix, E. L., Corbett, J., Sweet, J. H., & Ponting, M. (2005). Dicing and grinding of
electroceramics. CARTS Europe 2005. Prague.
Passmann, C., & Ermert, H. (1996). A 100-MHz ultrasound imaging system for dermatologic and
ophthalmologic diagnostics. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 43 (4), 545-552.
Reinstein, D. Z., Silverman, R. H., Raevsky, T., Simoni, G. J., Lloyd, H. O., Najafi, D. J., et al.
(2000). Arc-scanning very high-frequency ultrasound for 3-D pachymetric mapping of the
corneal epithelium and stroma in laser in situ keratomileusis. J. Refract. Surg. , 16, 414-430.
Ritter, T. A. (2000). Design, fabrication, and testing of high frequency (>20MHz) composite
ultrasound imaging arrays. State College, PA: The Pennsylvania State University.
Ritter, T. A., Shrout, T. R., Turwiler, R., & Shung, K. K. (2002). A 30-MHz piezo-composite
ultrasound array for medical imaging applications. IEEE Trans. Ultrason. Ferroelect. Freq.
Contr. , 49 (2), 217-230.
97

Selfridge, A. (1983). The design and fabrication of ultrasonic transducers and transducer arrays.
Stanford, CA: Stanford University.
Shung, K. K. (2006). Diagnostic Ultrasound: Imaging and Blood Flow Measurements. Boca
Raton, FL: CRC Press.
Silverman, R. H., Coleman, D. J., Ketterling, J. A., & Lizzi, F. L. (2005). High-frequency
harmonic imaging of the eye. Proc. Medical Imaging: Ultrasonic Imaging and Signal Processing,
5750, pp. 16-25.
Silverman, R. H., Kong, F., Chen, Y. -C., Lloyd, H. O., Kim, H. H., Cannata, J. M., et al. (2010).
High-resolution photoacoustic imaging of ocular tissues. Ultrasound in Med. & Biol. , in press.
Smith, W. A., & Auld, B. A. (1991). Modeling 1-3 composite piezoelectrics: thickness mode
oscillations. IEEE Trans. Ultrason. Ferroelect. Freq. Contr. , 38 (1), 40-47.
Sonomed. (2006). Ultrasound biomicroscopy system VuMax-II brochure. Retrieved from
http://www.sonomedinc.com/VuMax.pdf
SonoSite. (2010). Mountable Diagnostic Ultrasound Machines. (SonoSite) Retrieved from
Mountable Diagnostic Ultrasound Machines: http://www.sonosite.com/products/mountable-
ultrasound/
Turnbull, D. H. (1999). In utero ultrasound backscatter microscopy of early stage mouse
embryos. Comput. Med. Imaging Graph. , 23 (1), 25-31.
Turnbull, D. H., Starkoski, B. G., Harasiewicz, K. A., Semple, J. L., From, L., Gupta, A. K., et al.
(1995). A 40-100 MHz B-scan ultrasound backscatter microscope for skin imaging. Ultrasound
in Med. & Biol. , 21 (1), 79-88.
VisualSonics. (2010). VisualSonics Vevo2100. Retrieved from
http://www.visualsonics.com/vevo2100/index.htm
98

99

Vos, H. J., Frijlink, M. E., Droof, E., Goertz, D. E., Blacquiere, G., Gisolf, A., et al. (2005).
Transducer for harmonic intravascular ultrasound imaging. IEEE Trans. Ultrason. Ferroelect.
Freq. Contr. , 52 (12), 2418-2422.
Wikipedia. (2010). Schematic diagram of the human eye with English annotations. Retrieved
from
http://en.wikipedia.org/wiki/Image:Schematic_diagram_of_the_human_eye_with_English_annot
ations.svg
Woo, J. (1998). A short history of the development of ultrasound in obstetrics and gynecology.
Retrieved from http://www.ob-ultrasound.net/history2.html
Zanelli, C., & Howard, S. (2006). Schlieren metrology for high frequency medical ultrasound.
Ultrasonics , 44 (supplement 1), e105-e107.

Abstract (if available)

Abstract Ultrasound transducer solutions were proposed for imaging the posterior segment of the human eye. In one approach, concentric annular type dual element transducers for second harmonic imaging at 20 MHz / 40 MHz were designed and fabricated for imaging the posterior segment of the eye. The outer ring element was designed to transmit the 20 MHz signal and the inner circular element was designed to receive the 40 MHz second harmonic signal. Multiple prototype transducers were fabricated and characterized quantitatively. Images of a posterior segment of an excised pig eye and a choroidal nevus of human eye were obtained and the advantages of dual element harmonic imaging were demonstrated. In another approach, 20 MHz 192 element convex array transducers have been designed, fabricated, and characterized. It was demonstrated that convex array transducers of 20 MHz with an aperture curved both in azimuth and elevation direction could be fabricated with an acceptable uniformity in element-by-element performance. All 192 elements of the array were fully characterized by the pulse-echo test, crosstalk measurement, insertion loss test and synthetic aperture imaging. Average pulse-echo loop sensitivity was -63.7 dB and average -6 dB fractional bandwidth 69.2%, which are acceptable for imaging purpose. Ringing of echoes was substantially reduced by using a lower composite kerf by the interdigital pair bonding for fabricating 1-3 composites. Created images of ex vivo porcine eye tissues with a trapezoidal field of view were shown to a wider view angle than linear arrays.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A high frequency array- based photoacoustic microscopy imaging system

PDF

Development of high-frequency (~100mhz) PZT thick-film ultrasound transducers and arrays

PDF

High-frequency ultrasound array-based imaging system for biomedical applications

PDF

Development of novel 1-3 piezocomposites for low-crosstalk high frequency ultrasound array transducers

PDF

High frequency ultrasonic phased array system and its applications

PDF

Novel beamforming with dual-layer array transducers for 3-D ultrasound imaging

PDF

High-frequency ultrasound imaging system with Doppler features for biomedical applications using 30~35 mHz linear arrays and an analog beamformer

PDF

Configurable imaging platform for super-harmonic contrast-enhanced ultrasound imaging

PDF

Development of front-end circuits for high frequency ultrasound system

PDF

Improved contrast in ultrasound imaging using dual apodization with cross-correlation

PDF

Highly integrated 2D ultrasonic arrays and electronics for modular large apertures

PDF

High frequency ultrasound array for ultrasound-guided breast biopsy

PDF

Development of high frequency focused transducers for single beam acoustic tweezers

PDF

Development of back-end processing system for high frequency ultrasound b-mode imaging

PDF

High frequency ultrasonic imaging for the development of breast biopsy needle with miniature ultrasonic transducer array

PDF

Development of high frequency annular array ultrasound transducers

PDF

Transducers and signal processing techniques for simultaneous ultrasonic imaging and therapy

PDF

Miniature phased-array transducer for colorectal tissue characterization during TEM robotic surgery; and, Forward-looking phased-array transducer for intravascular imaging

PDF

Audio and ultrasound MEMS built on PZT substrate for liquid and airborne applications

PDF

High-frequency ultrasonic transducers for photoacoustic applications

Asset Metadata

Creator Kim, Hyung Ham (author)

Core Title Array transducers for high frequency ultrasound imaging

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 04/27/2010

Defense Date 03/09/2010

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag convex array,curved linear array,high frequency ultrasound,OAI-PMH Harvest,ophthalmic imaging,PZT composite,ultrasound imaging

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Shung, K. Kirk (committee chair), Cannata, Jonathan Matthew (committee member), Kim, Eun Sok (committee member), Meng, Ellis F. (committee member), Yen, Jesse T. (committee member)

Creator Email hhkim@ieee.org,hyungham.kim@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2956

Unique identifier UC1178387

Identifier etd-Kim-3636 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-310531 (legacy record id),usctheses-m2956 (legacy record id)

Legacy Identifier etd-Kim-3636.pdf

Dmrecord 310531

Document Type Dissertation

Rights Kim, Hyung Ham

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu