University of Southern California Dissertations and Theses

High frequency ultrasonic phased array system and its applications

(USC Thesis Other)

High frequency ultrasonic phased array system and its applications

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

HIGH FREQUENCY ULTRASONIC PHASED ARRAY SYSTEM AND ITS
APPLICATIONS

by

Fan Zheng

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)

December 2012

Copyright 2012 Fan Zheng
ii
Acknowledgements
First, I would like to thank my mentor and advisor, Dr. K. Kirk Shung, for giving
me the opportunity to do exciting works in National Institutes of Health (NIH) Resource
Center for Ultrasonic Transducer Technology at University of Southern California. His
patience and understanding allowed me to develop individuality and independence. His
advice and support helped me overcome difficulties.
I would also like to thank the exemplary members of my dissertation committee:
Dr. Qifa Zhou, Dr. Jesse T. Yen, and Dr. Eun Sok Kim. Their indispensable inputs helped
me refine my dissertation and broaden my horizon.
Much of the research work for this dissertation would not have been possible
without the cooperation with all the current and former members in our center. I am
grateful to Dr. Changhong Hu, Dr. Lequan Zhang, Dr. Hsiu-Sheng Hsu, Dr. Ying Li, Dr.
Kwok Ho Lam, Dr. Hao-Chung Yang, Dr. Jonathan Cannata, Dr. Hyung Ham Kim, Dr.
Xiang Li, Dr. Hojong Choi, Chi Tat Chiu, Ruimin Chen, Bong Jin Kang, Changyang Lee,
Yang Li, Teng Ma, Jay Williams, and Thomas Cummins.
I would like to thank all of my friends, particularly Dr. Changgeng Liu, Dr.
Xiangyang Zhang and Dr. Ping Sun for their assistance and encouragement during my
study in USC.
Finally, this dissertation is dedicated to my parents and my wife for their endless
love and unlimited support, getting me through hilltop and valley of my life. Thank you
for always being there for me.
iii
Table of Contents
Acknowledgements ii
List of Tables v
List of Figures vi
Abstract xi
Chapter 1 Introduction 1
1.1 Background of this work 1
1.2 Motivations and objectives of this study 3
1.3 Outline of this dissertation 4
Chapter 2 High Frequency Ultrasound Background 6
2.1 High frequency ultrasound 6
2.2 Ultrasonic transducers 8
2.2.1 Single element transducer 9
2.2.2 Array transducer 10
2.3 Scanning in ultrasonic imaging 11
2.3.1 Mechanical scanning on single element transducer 11
2.3.2 Electronic scanning on array transducer 12
2.4 Geometric approach on time delay calculation for a phased array 14
2.4.1 Linear phased array transmit beamforming 15
2.4.2 Linear phased array receive beamforming 16
2.5 B-mode imaging 18
2.6 Scan conversion 18
2.7 Grating lobe when steering the beam 20
Chapter 3 Development of a 64-Channel High Frequency Ultrasonic Phased
Array System 24
3.1 Introduction 24
3.2 System design 25
3.3 Transmit subsystem 27
3.3.1 FPGA based transmit beamformer 28
3.3.2 Bipolar pulse waveform generator 30
3.3.3 High voltage pulser 31
3.4 Receive subsystem 34
3.4.1 Low noise amplifier 34
3.4.2 Passive anti-aliasing filter 34
3.4.3 Analog-to-digital converter (ADC) 36
iv
3.4.4 FPGA based receive beamformer 37
3.5 DSP based data transceiver 45
3.6 Gigabit Ethernet 47
3.7 Data post-processing and visualization on PC 49
3.8 System evaluation 51
3.8.1 Signal-to-noise ratio of the system 51
3.8.2 26 MHz linear phased array 52
3.8.3 Wire phantom image 53
3.8.4 Anechoic sphere phantom image 55
3.8.5 In vitro rabbit eyeball experiment 58
3.9 Discussion and summary 59
Chapter 4 Non-destructive Testing with a High Frequency Ultrasonic Array
System 61
4.1 Introduction 61
4.2 Methods 63
4.2.1 Array transducer and scanning method 63
4.2.2 Reflection 64
4.2.3 Refraction receive beamforming 65
4.2.4 C-mode scan and motorized stage system 67
4.3 Imaging results 71
4.3.1 Aluminum sample imaging 71
4.3.2 Silicon carbide sample imaging 74
4.4 Discussion and summary 76
Chapter 5 Acoustic Trapping with a High Frequency Linear Phased Array 78
5.1 Introduction 78
5.2 Methods 80
5.3 Results 84
5.4 Discussion and summary 85
Chapter 6 Conclusion and Future Work 87
6.1 Conclusion 87
6.2 Future work 88
6.2.1 Improve the current system 88
6.2.2 System function expansion: color flow Doppler 90
6.2.3 Design transmit subsystem for 1D/2D higher frequency array
trapping 90
Bibliography 92

v
List of Tables
Table 3.1 Specifications of the 26.3 MHz prototype linear phased array 52
Table 4.1 Properties of silicon carbide at 20
o
C 62
Table 4.2 Specifications of the 30 MHz linear array transducer 64
Table 4.3 Properties of aluminum at 20
o
C 71

vi
List of Figures
Figure 2.1 Approximate frequency ranges: corresponding to sound,
ultrasound, and high frequency ultrasound 6
Figure 2.2 Diagram of an ultrasound transducer in reflection ultrasonic
imaging 8
Figure 2.3 Typical structure of a single element ultrasound transducer 10
Figure 2.4 Simple diagram for the structure of an ultrasonic linear array
transducer (Cannata et al., 2011) 11
Figure 2.5 Mechanically translate a single element transducer to form
images. (A) Linear scan: a transducer is moved along a straight-
line path; (B) Sector scan: a transducer is rotated around a
rotation center. 12
Figure 2.6 In linear array system, each active subgroup forms one beam
(scan line). Transmit delays help the beam to be focused at
specific depth. 13
Figure 2.7 All phased array elements are excited at almost the same time to
form the scan line. With different delay profiles, a phased array
can steer the beams in different directions and focus the beams
in different depth. 14
Figure 2.8 Transmit beamforming on a linear phased array 16
Figure 2.9 Receive beamforming on a linear phased array 17
Figure 2.10 Schematic of scan conversion using bilinear interpolation 19
Figure 2.11 Field II simulations of B-mode wire phantom images with ± 45º
steering angles on linear phased arrays. (Up) A 64-element
linear phased array with the pitch of half-wavelength, no grating
lobe appears in B-mode image; (Down) A 64-element linear
phased array with the pitch of two-wavelength; artifacts due to
grating lobes can be easily found when the beams are steered. 20
Figure 2.12 First grating lobe positions vs. steering angles; the pitch of an
array is 0.5, 0.75, 1, and 2 wavelengths, respectively 21
Figure 2.13 Steering angles vs. array pitch (without grating lobes) 22
Figure 3.1 Overall high frequency ultrasonic array system architecture 26

vii
Figure 3.2 Photograph of the developed ultrasonic phased array system
including analog front-end boards (front two boards), digitizer &
receive beamformer boards (four boards behind), transmit
control board (top one) and DSP board (up-right corner). 27
Figure 3.3 Block diagram of the 64-channel transmit subsystem. 28
Figure 3.4 Block diagram of the FPGA based transmit beamformer 29
Figure 3.5 Photograph of the transmit control board 30
Figure 3.6 The waveform of the pulse generated by FPGA based transmit
beamformer. The time duration of the waveform is four clocks. 31
Figure 3.7 Block diagram of the 64-channel FPGA based pulse waveform
generator 31
Figure 3.8 (Left) Normalized waveform of a single pulse generated in
channel #1; (Right) Spectrum of the pulse generated in channel
#1. 32
Figure 3.9 Center frequency and -6 dB frequency ranges for the pulses
generated in all 64 channels 33
Figure 3.10 Photograph of one of the two 32-channel transceiver boards 33
Figure 3.11 Schematic of the anti-aliasing 4
th
order band-pass Butterworth
filter 35
Figure 3.12 Simulated vs. measured frequency response of the anti-aliasing
4
th
order band-pass filter 36
Figure 3.13 Photograph of one of four digitizer & receive beamformer
boards. Each board can handle 16 high speed ADCs. 37
Figure 3.14 The FPGA on one of four receive beamformer boards can handle
16 high speed ADCs. Each ADC connects to FPGA via 12-bit
(24-pin actually used) LVDS databus. All ADCs and FPGA are
synchronized through a global clock “G_CLK”. 38
Figure 3.15 Block diagram of the FPGA based receive beamformer, which
consists of ADC pre-processing, delay computing, RAM I, Sum
(& others), and RAM II modules. 39
Figure 3.16 Block diagram of the ADC pre-processing module in receive
beamforming FPGA 40
Figure 3.17 Block diagram of the delay computing module in receive
beamforming FPGA 41
Figure 3.18 Block diagram of RAM I module in receive beamforming FPGA
for temporally storing the processed data from ADCs. 42
viii
Figure 3.19 Aperture selection sub-module: Pin configurations and function
descriptions. 43
Figure 3.20 Apodization sub-module: Pin configurations and function
descriptions. 43
Figure 3.21 Adder sub-module: Pin configurations and function descriptions. 44
Figure 3.22 RAM II module: Pin configurations and function descriptions. 45
Figure 3.23 Photograph of the DSP board 46
Figure 3.24 DSP and four receive beamforming FPGA are connected
through EMIF interface 46
Figure 3.25 Block diagram of the gigabit Ethernet connection in current
array system 48
Figure 3.26 Screenshot of simple continuous connection throughput test
between array system with gigabit Ethernet and PC with a
gigabit NIC 49
Figure 3.27 Screenshot of the graphical user interface (GUI) used to
visualize the B-mode image. An image of a tissue-mimicking
phantom was shown in this developed GUI. 49
Figure 3.28 Block diagram of the multi-thread application for the data post-
processing and visualization on PC. Front thread and
background thread should be carefully synchronization to avoid
unpredictable data corruption. 50
Figure 3.29 Signal-to-noise ratio (SNR) for all 64 channels 51
Figure 3.30 Photograph of the prototype 26.3 MHz linear phased array 52
Figure 3.31 A wire phantom used in the experiments. (Left) The photograph
of the wire phantom, multiple steps are used to hold fine wires.
(Right) Schematic of the cross section of the wire phantom with
five tungsten wires (20 μm diameter) 53
Figure 3.32 (Left) Measured wire phantom B-mode image obtained with the
26.3 MHz custom-designed phased array and the current
imaging system. (Right) Field II simulated wire phantom B-
mode image. Both are in 50 dB dynamic range. 55
Figure 3.33 Measured lateral and axial spread obtained with the 26.3 MHz
customized phased array and the current imaging system. The -6
dB lateral and axial resolutions were found to be 209 μm and
104 μm, respectively. 55

ix
Figure 3.34 Tissue-mimicking phantom. (Top) A photograph of a tissue-
mimicking phantom. (Bottom) The diagram of the tissue-
mimicking phantom. Anechoic spheres with different sizes are
embedded in nine small blocks with echogenic background
(Madsen et al., 2010) 56
Figure 3.35 Images of an anechoic sphere phantom with 1090-, 825-, 400-,
300- μm spheres using the 26.3 MHz custom-designed phased
array with the current imaging system (all in 50 dB dynamic
range). One transmit (5 mm) and dynamic receive foci were used
to form the images. The CNR of the 1090-, 825-, 400-, 300- μm
sphere phantom images are 4.14, 4.03, 4.04, and 3.27,
respectively. 57
Figure 3.36 B-mode image of an excised rabbit eye obtained with the 26.3
MHz custom-designed phased array and imaging system. One
transmit (5.5 mm) and dynamic receive foci were used to form
the image. The cornea, iris, ciliary muscle and lens are clearly
visible in the image with 50 dB dynamic range. 59
Figure 3.37 B-mode scan image of an anechoic sphere phantom with 530 μm
sphere using the 26.3 MHz custom-designed linear phased array
with the current imaging system. The steering angle is ± 45
degree. 60
Figure 4.1 Ultrasound energy distributions (left) in water-aluminum and
(right) in water-SiC interfaces. Only 6.2% ultrasound energy in
water-aluminum and 1.9% ultrasound energy in water-SiC
interface can return to the transducer. 65
Figure 4.2 Simplified diagram showing the refraction of ultrasound beam at
the SiC-water or aluminum-water boundary 67
Figure 4.3 1D array transducer moves in a 2D (X-Y) plane to form a C-scan
image. Typically, in X direction, the array moves with discrete
steps. In each step, electronic linear scanning is performed. 68
Figure 4.4 Photograph of the motorized stage system. (Left) SHOT-204MS
motorized stage controller. (Right) SGSP33-200(XY) 2D linear
stage with array transducer fixed on it 69
Figure 4.5 Block diagram of the connection between the motorized stage
system and the high frequency ultrasonic array system. 70
Figure 4.6 Flow chart for controlling the motorized stage, shaded blocks
run repeatedly. 70
Figure 4.7 Aluminum sample. (A) Photograph of an aluminum sample; (B)
Schematic of the side view of the aluminum sample; (C)
Schematic of the top view of the aluminum sample 72
x
Figure 4.8 B-mode image of the aluminum sample (45 dB dynamic range) 72
Figure 4.9 C-mode image of the aluminum sample (bottom surface) 73
Figure 4.10 Maximum amplitude projection (MAP) image of the aluminum
sample. The image was constructed by first removing the top
and bottom surface signals and then projecting the remaining
maximum signal amplitude along each A-line onto the
corresponding length-width plane. 74
Figure 4.11 Photograph of a silicon carbide sample block 74
Figure 4.12 B-mode image of the silicon carbide sample with conventional
time-delay beamforming in 35 dB dynamic range 75
Figure 4.13 B-mode image of the silicon carbide sample with modified time-
delay beamforming in 35 dB dynamic range 76
Figure 5.1 Phased array geometry and focal point 81
Figure 5.2 Three transmit time delay patterns were loaded into the transmit
beamformer before the experiment. 82
Figure 5.3 Simulation of the ultrasound intensity when the phased array
steers the beam and focuses at three different azimuth positions
where depth/axial positions are all 5 mm. The color represents
the normalized intensity. (a) 0 μm in azimuth direction. (b) 350
μm in azimuth direction. (c) 450 μm in azimuth direction. 83
Figure 5.4 Block diagram of the linear phased array acoustic trapping and
moving experiment. 84
Figure 5.5 Micro-particles (45 μm mean diameter) were trapped and moved
with transmit different time delay patterns. E: Elevation
direction; A: Azimuth direction. (a) No ultrasound was
transmitted. (b) Time delay pattern #1. (c) Time delay pattern #2
(d) Time delay pattern #3. (e) Back to time delay pattern #2. (f)
Back to time delay pattern #1. 85
Figure 6.1 The waveform of the new pulse generated by FPGA, the time
duration of the pulse waveform is two clocks instead of four
clocks for 30 MHz arrays. 91
Figure 6.2 New pulses with 140 MHz system clock. (Left) Spectrum of the
pulse generated in channel #1; (Right) Center frequency and -6
dB frequency range for the new high voltage pulses generated in
all 64-channels 91

xi
Abstract
While similar to a linear-switched array in structure, an ultrasound linear phased
array (simply called phased array) is quite different in operation. It is capable of beam
steering to form fan-shaped sector images without inducing grating lobes. Moreover, the
relative smaller array footprint is useful where only very limited contact surface is
permitted. High frequency phased arrays combine the advantages of the phased array and
high frequency ultrasound, which offer both small footprint and high spatial resolution.
This dissertation reports the design and development of a digital ultrasonic imaging
platform with raw RF data acquisition capability, which can be paired with a prototype
64-element 26 MHz phased array transducer. A wire phantom image showed that -6 dB
lateral and axial resolutions were 209 and 104 μm, respectively, which were in good
agreement with the Field II simulation. Anechoic cyst tissue-mimicking phantom images
demonstrated its capability to detect cysts of 300 μm in diameter. An image of a rabbit
eyeball in vitro was also acquired. This imaging platform is designed for the purpose of
testing high frequency phased arrays under development as well as facilitating the
development in novel array signal processing algorithms.
Without additional change in the system hardware, the current high frequency
array system can be also applied in the industrial non-destructive testing (NDT)
applications. A motorized XY stage allows the array system to acquire C-mode scan as
well as normal B-mode scan with bigger scanning area. Electronic scanning in array
system could dramatically reduce the total scanning time compare to single element
xii
system. An aluminum sample and a silicon carbide sample were examined to test the
system in non-destructive testing.
Besides the applications in biomedical imaging and non-destructive testing, the
high frequency ultrasonic phased array and system are shown to be capable of trapping
and translating microparticles precisely and efficiently, made possible due to the fact that
the acoustic beam produced by a phased array can be both focused and steered. Acoustic
manipulation of microparticles by a phased array is advantageous over a single element
transducer since there is no mechanical movement required for the array. Experimental
results show that 45 μm diameter polystyrene microspheres can be easily and accurately
trapped and moved to desired positions by a 64-element 26 MHz phased array.

1
Chapter 1 Introduction
1.1 Background of this work
Ultrasound is a sound wave with a frequency above the range of human hearing.
As early as 1880s, Curie brothers discovered piezo-electric effect in that mechanical
pressure would be produced when an electric potential was exerted on a quartz crystal
and vice versa. This was the humble beginning of ultrasound today. Further research on
piezo-electricity soon followed.
The use of ultrasound for medical diagnoses may date back to more than 50 years
ago, i.e. 1950s. Nowadays ultrasonic imaging has become one of the most utilized
diagnostic imaging modalities. The widespread usage of ultrasound should thank unique
characteristics of ultrasound and ultrasonic imaging methods, which include safety (non-
ionizing radiation), real-time (direct/non-reconstructive imaging), high resolution
(millimeter range for commercial ultrasound modalities), portable and less expensive
(Shung, K. Kirk, 2006).
During the first couple of decades on diagnostic ultrasound, most of the ultrasonic
imaging systems relied on single element transducers. To generate an image, transducers
had to be mechanically translated. It is not desirable in terms of reliability. Fixed focus is
another drawback of single element transducers. To overcome these problems, from
1960s, researchers began to develop ultrasonic linear-switched array systems as well as
phased array systems, which utilized Thomas Young’s principle of constructive and
destructive interaction of waves: waves that combine in phase reinforce each other, while
2
out-of-phase waves will cancel each other. Electrical beam steering and dynamic
focusing are then possible in phased array systems.
In the early 1970s, commercial phased array systems for medical diagnosis first
appeared. The relative smaller array footprint is quite useful in trans-thoracic cardiac
ultrasound (Kisslo et al., 1976; Vonramm et al., 1976; Martin et al., 1978) and intra-
cardiac imaging (Packer et al., 2002; Marrouche et al., 2003; Ren et al., 2004). However,
due to the electronics limitations, the original phased array systems were very complex,
cumbersome and required much power in limited space to deal with multi-channel beam
steering and focusing. With the help of rapid transition from analog world to digital
world, the new generation of digital phased array systems begin to appear (Kozak et al.,
2001).
Another trend in ultrasonic imaging systems is to increase the transducer’s
frequency. The main advantage that can be achieved by using the higher frequency is the
improvement of spatial resolution. High frequency ultrasound has been applied to
dermatology (Turnbull et al., 1995; Vogt et al., 2007), ophthalmology (Foster et al., 1993;
Coleman et al., 2004; Silverman et al., 2007) and small animal cardiac imaging (Zhou,
Y.-Q. et al., 2004; Sun et al., 2008). Also commercially available high frequency single
element transducer system (Vevo 770, VisualSonics, Inc., Toronto, Canada) has been
found useful in quite a lot of applications with the high spatial resolution required.
While most of the high frequency ultrasonic systems use single element
transducers attached to a mechanical translation system to form ultrasonic images, a
number of high frequency linear-switched array transducer based ultrasonic systems have
3
been reported (Hu et al., 2006; Zhang et al., 2010; Hu et al., 2011). These linear-switched
array systems used electronic scanning and focusing to increase the scanning speed and
the depth of filed. Quite recently, a high frequency linear-switched array system (Vevo
2100, VisualSonics, Inc., Toronto, Canada) has become commercially available (Foster et
al., 2009).
1.2 Motivations and objectives of this study
Although many works have been done on the low frequency ultrasonic phased
array systems, high frequency single element transducer systems and high frequency
linear-switched array systems, currently there is no platform available for high frequency
phased array to perform fan-shaped sector scanning by beam steering and focusing.
Inspired by this, to combine the benefits of the phased array systems and high frequency
ultrasound, it is necessary to design a high frequency ultrasonic phased array imaging
platform. This dissertation presents an initial effort in developing such a digital high
frequency phased array system. Preliminary experiments and applications performed on
this system will also be introduced.
The overall objective of this dissertation is to develop a high frequency ultrasonic
phased array system to work with a 64-element phased array with around 30 MHz center
frequency.
The specific aims of this study are:
1. To design and develop a 64-channel high frequency ultrasonic array system,
which is capable of performing beam steering and dynamic focusing;
4
2. To apply the high frequency ultrasonic array system in non-destructive testing
for industrial applications;
3. To demonstrate the capabilities of trapping and moving microparticles with
high frequency phased array and array system.
1.3 Outline of this dissertation
The rest of this dissertation is organized as follows:
In Chapter 2, some basics on the high frequency ultrasound, ultrasonic transducer,
and imaging are introduced.
The main theme of Chapter 3 is the development of a 64-channel high frequency
phased array system, which is capable of handling a 64-element linear phased array with
26 MHz center frequency. B-mode sector scanning imaging is provided by implementing
both beam steering and dynamic focusing. Tissue-mimicking phantom imaging and in
vitro rabbit eyeball imaging are presented to demonstrate the capabilities of the array
system in biomedical imaging applications.
Chapter 4 proposes the high frequency array system can work along with a
motorized XY stage for the industrial non-destructive testing applications. B-mode and
C-mode imaging are supported in the system. Both silicon carbide sample imaging and
aluminum sample imaging are presented.
Chapter 5 demonstrates that, similar to the single-beam trapping mechanism with
single element transducers, the current 26 MHz 64-element ultrasonic phased array and
5
array system are capable of trapping and translating 45 μm diameter polystyrene
microspheres by using electronic beam steering and focusing.
Chapter 6 summarizes the works of this dissertation. Suggestions for the future
research are also mentioned.
6
Chapter 2 High Frequency Ultrasound Background
2.1 High frequency ultrasound
Since the physical nature of ultrasound is acoustic wave with pressure fluctuation,
the propagating ultrasound waves can be described by the wave equation

∇
−
1

=0,
(2.1)
where ∇
is the Laplace operator, is the ultrasound pressure, and is the speed of
ultrasound (Kinsler et al., 2000). The speed of ultrasound varies in different materials,
such as 1480 meter/second in water, 1570 meter/second in liver and 6420 meter/second in
aluminum (Shung, K. Kirk, 2006).
The frequency of ultrasound lies beyond the limits of the human hearing (20 Hz ~
20 kHz). Conventional low frequency ultrasound imaging uses 2 to 15 MHz, whereas
high frequency ultrasound uses the frequency range above 15 MHz (Figure 2.1).

Figure 2.1 Approximate frequency ranges: corresponding to sound, ultrasound, and high frequency
ultrasound
7
The benefit of high frequency is high spatial resolution, which describes the
ability of an imaging system to distinguish small details of an object. In ultrasound
system, the lateral and axial resolutions are the spatial resolutions in lateral and axial
direction, respectively. The lateral resolution is determined by the ultrasound beam width,
and the axial resolution is determined by the bandwidth of the transducer. The lateral and
axial resolutions can be estimated using Equation (2.2) and (2.3):

=
#
, (2.2)
=
2∙ , (2.3)
where is the speed of sound, is the center frequency of the transducer, #
(F number)
is defined as the ratio of focal length to the aperture of the transducer, and

represents the -6 dB bandwidth of the transducer (Foster et al., 2000). It appears that
increasing the center frequency or aperture size of the transducer will improve the lateral
resolution and increasing the bandwidth of the transducer will improve the axial
resolution.
Although better spatial resolution can be achieved by increasing the frequency,
the penetration depth will be reduced since the attenuation of an ultrasound wave is
proportional to frequency (Shung, K. Kirk, 2006). Attenuation here refers to the decrease
of the intensity of ultrasound wave. Spreading of the ultrasound beam, scattering, and
absorption may result in attenuation. Equation (2.4) shows the relationship of pressure vs.
penetration depth,
8
( ) =
∙
, (2.4)
where ( ) is the wave pressure at position in z-direction (depth direction), is
attenuation coefficient. The trade-off of high frequency and attenuation should be
considered in system design and scope of the applications. Moreover, the sensitivity and
bandwidth of the transducer should be optimized to compensate the attenuation with the
increasing frequency.
2.2 Ultrasonic transducers
An ultrasonic transducer is a device that converts electrical energy into
mechanical vibration energy and vice versa. Normally, in reflection mode ultrasonic
imaging (Figure 2.2), the pulse generator excites the transducer, generating ultrasound
waves to propagate through the tested object. When the wave fronts hit discontinuities
throughout the wave-propagating path, reflected ultrasound waves are generated and
propagate in the opposite direction. The same transducer receives the reflected ultrasound
waves and converts into electrical signals. With the help of Transmit/Receive (T/R)
switch, the resultant echo signals can be processed and displayed.

Figure 2.2 Diagram of an ultrasound transducer in reflection ultrasonic imaging
9
Piezoelectric materials are used in transducers to convert energy (Shung, K. K. et
al., 2007). The most commonly used piezoelectric materials are lead zirconate titanate
(PZT) ceramics (Foster et al., 1991; Zipparo et al., 1997), single crystals such as lithium
niobate (LiNbO
3
), and piezoelectric polymers such as polyvinylidene fluoride (PVDF)
and its copolymer. The thickness of the piezoelectric materials determines the resonant
frequency.
There are varieties of transducers ranging from single element transducers to
array transducers.
2.2.1 Single element transducer
Single element transducers are the most common transducers for high frequency
ultrasonic imaging. A typical structure of a single element transducer is shown in Figure
2.3. The core part in a single element ultrasound transducer is a layer of piezoelectric
material for energy conversion. Multiple matching layers are added in front of
transducers to allow the energy to be transmitted into the loading medium maximally.
Backing layers are also added in the back of transducers to provide a rigid support for the
fragile piezoelectric element and adjust the bandwidth and sensitivity of transducers.
10

Figure 2.3 Typical structure of a single element ultrasound transducer

2.2.2 Array transducer
Ultrasonic array transducers just like put multiple small single-element
transducers together and therefor have the same overall structure as the single-element
transducers. These elements can be arranged concentrically (called annular array) or in a
line (called linear array), or in rows and columns (called tow-dimensional array).
A number of high frequency linear array transducers have been developed
recently. Cannata developed a 2-2 piezo-composite 35 MHz (Cannata et al., 2006) and an
interdigitally bonded 2-2 piezo-composite 30 MHz linear array (Cannata et al., 2011).
Brown developed a 1-3 composite 40 MHz linear array (Brown et al., 2007). With the
help of electronic dynamic focusing, the depth of field can be increased. Moreover, using
electronic scanning instead of mechanical scanning in single element transducer imaging,
11
the high imaging speed can be achieved and the reliability of the whole system can be
improved. The pitch (distance between the centers of two adjacent elements) of a linear
array is typically around one wavelength. Figure 2.4 shows a typical structure of a linear
array.

Figure 2.4 Simple diagram for the structure of an ultrasonic linear array transducer (Cannata et al., 2011)

A linear phased array transducer, often simply called phased array, is constructed
in a similar way to a linear array transducer. However, a phased array has pitch smaller
than half wavelength compared to a linear array normally with one wavelength pitch.
With smaller pitch, a phased array can steer beams without inducing grating lobes.
Moreover, the relative smaller array footprint is useful where only very limited contact
surface is permitted.
2.3 Scanning in ultrasonic imaging
2.3.1 Mechanical scanning on single element transducer
Most high frequency single element transducers use mechanical scanning to form
images. Although quite a lot scanning schemes can be used to form images, linear scan
12
and sector scan are the two most commonly used. In linear scan, the transducer is moved
along a straight-line path with a constant speed, where ultrasound beam is orthogonal to
linear moving path (Figure 2.5(A)). While in sector scan, the transducer (and ultrasound
beam) is rotated around a rotation center to form a fan-shaped sector image (Figure
2.5(B)).

Figure 2.5 Mechanically translate a single element transducer to form images. (A) Linear scan: a
transducer is moved along a straight-line path; (B) Sector scan: a transducer is rotated around a rotation
center.

2.3.2 Electronic scanning on array transducer
It is relatively easy to build and less costly for a single element system. However,
single element transducer does not have the capability to perform dynamic focusing, and
the reliability of mechanical arrangement could be a problem. To solve these problems, a
linear array transducer can be used.
With linear arrays, images are forming by first activating and exciting a group of
elements at left end of the linear array to create first scan line. Then with the help of
Mechanically linear scan
(A) (B)
Mechanically sector scan
13
multiplexers, deactivate one element at most left and add one new element from right to
the active group to form a new scan line. The new scan line is parallel to the one before
with one pitch distance. In each active group of elements, slight time delays on
excitations are applied to focus the beam at specific distance in depth (Figure 2.6).
Changeable focal zone is the advantage over single element transducer. However, the
lengths of the linear array transducers limit their rectangular field of views. Trying to
steer the beams will cause grating lobes appearing in ±90° field-of-view.

Figure 2.6 In linear array system, each active subgroup forms one beam (scan line). Transmit delays help
the beam to be focused at specific depth.

1st Active Group
2nd Active Group
3rd Active Group
n
th
Active Group
N
th
Active Group
focus
Transmit delay
14
A phased array is quite different in operation. All elements are excited almost at
the same time to form one scan line. By using proper time delay, phased arrays can steer
and focus the beam to create a fan-shaped sector scan with a wide field of view (Figure
2.7). Phased arrays produce a sector scan format in which the scan lines emanate from the
center of the phased array transducer. Scan conversion should be applied to form the
image shown on the screen. There will be no grating lobes appearing in the field of view
for phased array with pitch shorter than half wavelength.

Figure 2.7 All phased array elements are excited at almost the same time to form the scan line. With
different delay profiles, a phased array can steer the beams in different directions and focus the beams in
different depth.

2.4 Geometric approach on time delay calculation for a phased array
To steer and focus the beam, time delays have to be applied in transmission and
reception. Normally the time delays are calculated off-line, and store in some memory
space for looking up.
15
2.4.1 Linear phased array transmit beamforming
To transmit ultrasound, steer the ultrasound beam into the suggested direction and
focus at specific point, a transmit beamforming should be used.
Consider a 1-D elements linear phased array with pitch (Figure 2.8). Assume
array elements are point sources. The delay required for the
element can be expressed
as

=
∆
+

=
− +

=
− +
−2∙
∙ ∙cos +

(2.5)
where is the distance from the center of
element to the center of the array and
=
− , ∈ [1,] ; is a constant time delay needed to ensure that all transmit
time delays are positive. is the distance between focus and the
element. is focal
length. It is a constant for one focal zone in transmission or several constants for multiple
transmit focal zones (multiple transmissions in one steering angle). By selecting proper
transmit time delays, the ultrasound wave from all array elements will arrive at picked
focal point nearly at the same time.

16

Figure 2.8 Transmit beamforming on a linear phased array

2.4.2 Linear phased array receive beamforming
After the beam is transmitted, part of the beam energy will be reflected by inner
discontinuities and return to the phased array (Figure 2.9). Similar to the situation in
transmit delay, the receive delay for the
th
n element can be expressed as

=
∆
+

=
−

=
−
+ +
−2∙
∙
∙cos
(2.6)
In receiving part, delay profiles for dynamic receive beamforming are normally quite
large and cannot be put into the system’s inner memory. To solve this problem, a
r
X
Z
n
x
q
p
Focal point
Transducer element
Origin
(0,0)
123 N
17
simplified equation is used to partially calculate the delays on-line. By using Taylor
series to expand above equation and only keep up to 2
nd
derivative items, it is easy to
have time delays needed for a phased array with even number of elements as

,
,
=
+ 1
2
−∙ ∙ cos ( ) ∙

−

+ 1
2
−∙ ∙ sin( )∙
∙
(2.7)
where is n
th
element, is j
th
sample in one scan line, is k
th
steer angle, and is
sampling frequency of an analog-to-digital converter. Equation (2.7) can be directly used
in calculating dynamic receiving delays for each sample in DSP. After applying delays on
echo signals received, all channel are summed together to form one scan line, this is
usually called delay-sum beamforming.

Figure 2.9 Receive beamforming on a linear phased array
θ
r
X
Z
n
x
q
Point object
p
Transducer element
Origin
(0,0)
18
2.5 B-mode imaging
B-mode (brightness mode) imaging is the most commonly used scheme in
ultrasound imaging. Normally B-mode imaging consists of envelope detection, dynamic
range compression, and gray-scale mapping.
The envelope detection is used to estimate the envelope of the signal from a point
scatter, which has the same envelope as the transmit pulse. Normally, lowpass filtering
and Hilbert transform are the most common used methods to do the envelope detection.
The dynamic range of the echo signals after envelop detection is still too large to
show the weak singles which represents the details of biomedical structures. Generally, a
log compression is used to map the amplitude of the signals into logarithmic domain for
efficient detail visualization.
After log compression, the signals then map to the gray-scale to visualize on
monitor.
2.6 Scan conversion
In the linear phased array sector scanning, the coordinate system of the echo
signal plane is different to that of the monitor. In a conventional sector scanning, an
ultrasound B-mode image is formed by using transmit time delays to steer the beam in a
sequence of angular directions. In addition, the receive time delays are applied to
estimate the reflectivity as a function of range along the particular angular direction ,
which is in polar coordinates. While a monitor, such as an LCD screen uses Cartesian
coordinates to display the image. The purpose of scan conversion in a digital ultrasound
19
application is to translate the data captured in polar coordinates into Cartesian coordinates
for display (Robinson et al., 1982; Park et al., 1984; Lee, M. H. et al., 1986; Sikdar et al.,
2001).
One simple and normal way to perform scan conversion is bilinear interpolation.
Figure 2.10 shows the diagram of using bilinear interpolation in scan conversion. The
coordinate conversion can be performed by the following equation:
, =
,
=
(− ) (− ), + ( − ), +
( − ) (− ) , + ( − ) , ∆ ∙ ∆
(2.8)
where = √ +
, = tan

, ∆ = −
and ∆ = −
.

Figure 2.10 Schematic of scan conversion using bilinear interpolation

1
2
3
20
2.7 Grating lobe when steering the beam
Grating lobe is a special side lobe. In ultrasonic linear array, when the pitch of an
array is longer than half wavelength, the aliasing effect will make the amplitude of some
side lobes become larger approaching the level of the main lobe. Grating lobe artifacts
are unwanted in ultrasound imaging. Figure 2.11 shows a simulation example of B-mode
images with and without grating lobes.

Figure 2.11 Field II simulations of B-mode wire phantom images with ± 45º steering angles on linear
phased arrays. (Up) A 64-element linear phased array with the pitch of half-wavelength, no grating lobe
appears in B-mode image; (Down) A 64-element linear phased array with the pitch of two-wavelength;
artifacts due to grating lobes can be easily found when the beams are steered.

Depth (mm)

-6 -4 -2 0 2 4 6
2
4
6
8
10
10
20
30
40
50
Depth (mm)
Lateral distance (mm)

-6 -4 -2 0 2 4 6
2
4
6
8
10
10
20
30
40
50
21
The position of the first grating lobe can be found by Equation:
=sin

sin ±

(2.9)
where is the wavelength of the ultrasound in the surrounding medium, is the pitch of
the array (the distance between the center of two adjacent array elements) and is the
steering angle (Ensminger et al., 2008).
As an example, consider a linear array with pitch = when the steering angle
=0°, first grating lobe will happen at =sin

( sin 0 ° ± 1 ) = ± 90°. When
= 30°, first grating lobe happens at =sin

( sin 30° − 1 ) = − 30°. And when the
steering angle = − 20°, first grating lobe will appear at =sin

( sin ( −20° ) +1 ) ≈
−41°. Figure 2.12 shows cases on the position of the first grating lobe vs. steering angles
when pitch is 05, 0.75, 1, and 2 wavelengths, respectively.

Figure 2.12 First grating lobe positions vs. steering angles; the pitch of an array is 0.5, 0.75, 1, and 2
wavelengths, respectively
-90 -45 0 45 90
-90
-45
0
45
90
First grating lobe position (degree)
p = 0.5λ

-90 -45 0 45 90
-90
-45
0
45
90
Steering angle (degree)
p = 0.75λ
-90 -45 0 45 90
-90
-45
0
45
90
p = λ
-90 -45 0 45 90
-90
-45
0
45
90
p = 2λ
22
Typically, to steer the beam on a linear phased array, it is required that the pitch
of the array should be shorter than 0.5 wavelength. However, if the steering angles are
limited in some range, it is possible to have no grating lobes even if the pitch of the array
is greater than 0.5 wavelength. Figure 2.13 shows if the pitch is less than 0.5 wavelength,
the steering angles can be as large as ±90°. When the pitch is 0.75 wavelength, the
steering angels are limited to around ±20° without inducing grating lobes. When the
pitch is longer than one wavelength, steering will always generate grating lobes.

Figure 2.13 Steering angles vs. array pitch (without grating lobes)

0.2 0.4 0.6 0.8 1 1.2 1.4
0
10
20
30
40
50
60
70
80
90
Pitch of an array (# of wavelength λ)
β (degree)
Pitch vs. steering angles without grating lobes ( ± β )

pitch vs. β
pitch = 0.5 λ
pitch = 0.75λ
pitch = λ
23
Some real situations should be considered beside above discussion. Since the
wavelength is calculated by using the center frequency of the array, the bandwidth of the
center frequency will low the steering angles. In addition, the sound speed will also affect
the calculation of wavelength. Therefore, the higher sound speed will raise the steering
angles range and vice versa.
24
Chapter 3 Development of a 64-Channel High Frequency
Ultrasonic Phased Array System
3.1 Introduction
High frequency (higher than 20 MHz) ultrasound provides a non-invasive
imaging method for many pre-clinical, clinical, and research applications requiring
superior spatial resolution. While most of the high frequency ultrasonic systems use
single element transducers attached to a mechanical translation system to form ultrasound
images, a number of high frequency linear array transducer based ultrasonic systems have
been reported (Hu et al., 2006; Zhang et al., 2010; Hu et al., 2011). These linear array
systems used electronic scanning and focusing to increase the scanning speed and the
depth of filed. Quite recently, a high frequency linear array system (Vevo 2100,
VisualSonics, Inc., Toronto, Canada) has become commercially available (Foster et al.,
2009).
Although similar to a linear array in structure, an ultrasound linear phased array
(simply called phased array) is quite different in operation, which is capable of beam
steering to form fan-shaped sector images without inducing grating lobes. Moreover, the
relative smaller array footprint is useful where only very limited contact surface is
permitted. High frequency phased arrays combine the advantages of phased arrays and
high frequency ultrasound, which offer both small footprint and high spatial resolution.
This chapter reports the design and development of a digital ultrasonic imaging
platform with raw RF data acquisition capability, which can be paired with a prototype
25
64-element 26 MHz phased array transducer. Imaging is the ultimate test for evaluating
the performance of an ultrasonic imaging system, although the final image quality is also
dependent on the array transducer’s performance. The images acquired by the developed
ultrasonic array system will be presented to demonstrate the capabilities of the system in
biomedical research applications.
3.2 System design
To a multi-channel high frequency ultrasonic imaging system, it will be more
challenging for the system hardware to provide reliable operations. Several requirements
are considered in the system-level design, which include:
1. High speed, wide bandwidth, and low noise electrical components selection for
high frequency imaging;
2. High computational power and large data throughput for real-time imaging;
3. High noise-suppression capabilities for multi-channel system (differential
connections massively being used);
4. Other requirements including appropriate decoupling capacitor selection, equal
length traces among all channels, analog and digital ground and power supply separation,
good stack-up in PCB layout design and so on.
The design of this 64-channel high frequency ultrasonic phased array system
follows the guidelines mentioned above.
The overall system architecture of the developed high frequency ultrasonic phased
array system is demonstrated in Figure 3.1. The whole system consists of transmit
26
subsystem, receive subsystem, and PC subsystem. Due to the need for full electronic
control of a 64-element phased array transducer, 64 independent electronic channels are
required.

Figure 3.1 Overall high frequency ultrasonic array system architecture

In the transmit subsystem, a field programmable gate array (FPGA) based
transmit beamformer determines the time delays (normally implemented by lookup table
(LUT)), pulse waveform and sequence, which sets the desired transmit focal point. A 64-
channel high voltage pulser generates high voltage bipolar pulses to excite the phased
array transducer.
In the receive subsystem, the 64-channel Transmit/Receive (T/R) switch blocks
the high voltage transmit pulses, and only allow the low voltage echo singles to get into
low voltage receiver circuits. A 64-channel low-noise amplifier (LNA) then follows the
T/R switch to amplify the week echo signals. After amplification, the analog echo signals
are converted to digital ones by analog digital converters (ADCs). In each channel, a
passive bandpass filter sits at the input path of an ADC to avoid sampling aliasing.
27
Followed by ADCs is a digital receive beamformer, which is implemented in FPGA and
DSP. All the data then move to a PC through a gigabit Ethernet connection.
In the PC subsystem, the gigabit network interface card (NIC) receives the data
from the array system. A developed software application then performs the B-mode
processing (envelope detection and log compression) and scan conversion. The final
images are also shown in the application.
Figure 3.2 shows a photograph of the developed high frequency phased array
system.

Figure 3.2 Photograph of the developed ultrasonic phased array system including analog front-end boards
(front two boards), digitizer & receive beamformer boards (four boards behind), transmit control board (top
one) and DSP board (up-right corner).

3.3 Transmit subsystem
The transmit subsystem in an ultrasonic array system typically consists of a
transmit beamformer and a pulser (pulse generator). The transmit beamformer generates
28
trigger signals with proper phase shift/time delay to allow the ultrasound from all array
elements to arrive the specific focal point at the same time. The pulser usually generates
high voltage pulses to excite the array elements. The block diagram of the transmit side in
the developed array system is shown in Figure 3.3. When the system boots up, the
transmit beamformer program is downloaded from the flash chip into the FPGA. Then
the transmit beamformer starts to run on the FPGA. A 64-channel diode bridge based
expander was also included in the transmit side to block the reflected signals from the
array transducer (Lockwood et al., 1991).

Figure 3.3 Block diagram of the 64-channel transmit subsystem.

3.3.1 FPGA based transmit beamformer
A high-speed FPGA XC3S1200E-4FG320 (Xilinx Inc., 2009a) was used as the
transmit beamformer. It has 1,200,000 gates and a total of 516,096 bit of block random
access memory (RAM) space, which can be used to store the transmit time delay table.
29
Since total memory space needed for 64-element array with 256 scan lines and 10-bit
delay resolution with one transmit focus is
# ∗ ∗ #
= 256 ∗ 1 0 ∗ 64 = 163,840,
this transmit delay table can be easily loaded into the memory space in this FPGA.
The architecture of the transmit beamformer in FPGA is shown in Figure 3.4. All
the transmit delay profiles are stored in a ROM generated by the FPGA block RAM. The
line trigger goes to a line counter to generate the address for the ROM. The frame trigger
goes to the “reset pin” of the ROM to reset the line number to “zero”. The digital clock
manager embedded in FPGA converts the system clock to form four clocks with 0°, 90°,
180°, and 270° phase shift. Use these four clocks, the delay generator can achieve finer
delays. Therefore, a delay resolution as fine as 1.79 ns (140 MHz system clock) could be
achieved.

Figure 3.4 Block diagram of the FPGA based transmit beamformer

30
A photograph of the top transmit control board (Figure 3.5) shows “Frame trigger” and
“Line trigger” SMA connectors, the system clock generator and the transmit
beamforming FPGA. All 64-channel delayed waveforms are sent to pulsers within two
transceiver boards to generate high voltage pulses.

Figure 3.5 Photograph of the transmit control board

3.3.2 Bipolar pulse waveform generator
Bipolar pulse scheme demonstrates good energy efficiency than unipolar pulse
one. Xu et al. has been reported a low-cost bipolar pulse generator for high-frequency
ultrasound applications (Xu et al., 2007), which used discrete components to implement
the bipolar pulser and pulse cycle counter. In our design, instead of using discrete
components, the transmit FPGA not only implement the transmit beamforming, but also
generate the waveforms (logic “1” and “0” in sequence) for 64-channel bipolar pulser
(Figure 3.6).
31

Figure 3.6 The waveform of the pulse generated by FPGA based transmit beamformer. The time duration
of the waveform is four clocks.

A total of 194 I/O pins and 50 input-only pins are available on this FPGA, which
is sufficient for 128 (64 pairs) connections to the MOSFET drivers and MOSFET to
excite 64-element ultrasound array transducer (Figure 3.7).

Figure 3.7 Block diagram of the 64-channel FPGA based pulse waveform generator

3.3.3 High voltage pulser
Instead of using two level-shifters in Xu’s approach, a dedicated MOSFET driver
ISL55110 (Intersil Americas Inc., 2011) is used to charge the MOSFET as quickly as
64
64-channel
delayed trigger
FPGA
Waveform
generator
P
N
Ch 1
P
N
Ch 64
32
possible and translates TTL logical signals (3.3V from FPGA) into MOSFET turn-on
voltage (normally 10~30V, 12V was used here). The maximum operating frequency of
this MOSFET driver is 70 MHz, which is sufficient for our applications.
Just after the MOSFET driver, a high speed, high voltage, gate-clamped N-
channel and P-channel MOSFET pair TC6320 (Supertex Inc., 2008) is used to produce
high-voltage pulses for ultrasound applications. It offers 400V peak-to-peak breakdown
voltage (200V for N-channel MOSFET and -200V for P-Channel MOSFET), which is
suitable to produce a high-voltage bipolar pulse to excite transducers.
With the system clock is set to 140 MHz, Figure 3.8 show the measured
normalized waveform and its spectrum of the bipolar pulse generated in the channel #1.
Figure 3.9 gives the center frequency and -6 dB frequency range of the 64-channel pulser,
which is good for 30 MHz array transducer (normally has 50% ~100% -6 dB bandwidth).

Figure 3.8 (Left) Normalized waveform of a single pulse generated in channel #1; (Right) Spectrum of
the pulse generated in channel #1.

-100 -50 0 50 100 150 200 250
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Time (ns)
Normalized amplitude
0 10 20 30 40 50 60 70
-25
-20
-15
-10
-5
0
Frequency (MHz)
Normalized amplitude (dB)
33

Figure 3.9 Center frequency and -6 dB frequency ranges for the pulses generated in all 64 channels

Figure 3.10 shows the position of pulsers on a transceiver board (there are two
transceiver boards, and each one can handle 32-channel transceivers).

Figure 3.10 Photograph of one of the two 32-channel transceiver boards
0 10 20 30 40 50 60
0
10
20
30
40
50
60
70
Channel index
Frequency (MHz)

Center frequency
34
3.4 Receive subsystem
Receiver analog electronics have significant influence on the quality of high
frequency ultrasonic imaging system. Careful electronic components selection and circuit
design are required for the weak echo signal acquisition.
3.4.1 Low noise amplifier
A quad-channel 3-stage amplifier AD8334 (Analog Devices Inc., 2010) was used
as the first stage ultralow noise pre-amplifier (LNA), the second stage variable gain
amplifier (VGA) and the third stage post-amplifier. The first stage LNA has 19 dB fixed
gain with a single-ended input and differential output. By adjusting a single external
resistor, the input impedance of the LNA can match the impedance of an array transducer
without compromising noise performance. The second stage VGA has -27~+21 dB gain
range, which is useful for a variety of applications. The third stage post-amplifier
supports 3.5 dB (Low gain mode) or 15.5 dB (High gain mode) gains with a pin clamping
level selection, which can optimize the output noise level for a subsequent analog-to-
digital converter (ADC). The total gain of the amplifier ranges from -4.5 dB to +43.5 dB
in low gain mode and from 7.5 dB to 55.5 dB in High gain mode. The 100 MHz -3 dB
bandwidth of AD8334 is suitable for ultrasonic array transducers with 30 MHz center
frequency.
3.4.2 Passive anti-aliasing filter
An anti-aliasing differential filter was used to restrict the bandwidth of the input
signal. Typically, an anti-aliasing filter is a low-pass filter. However, for bandwidth
35
limited signals, a band-pass filter can be used as an anti-aliasing filter. In the system, a 4
th

order Butterworth band-pass filter with 30 MHz center frequency and 100% bandwidth
(Figure 3.11) was designed. The measured frequency response results matches with the
simulation (Figure 3.12).

Figure 3.11 Schematic of the anti-aliasing 4
th
order band-pass Butterworth filter
36

Figure 3.12 Simulated vs. measured frequency response of the anti-aliasing 4
th
order band-pass filter

3.4.3 Analog-to-digital converter (ADC)
Based on Nyquist sampling theorem, to give satisfactory performance on digital
receive beamforming, the rule of thumb for the sampling rate should be 4 to 10 times of
the transducer’s center frequency (Steinberg, 1992). More realistically, around 4 times of
the transducer’s center frequency, i.e. 140 MSPS sampling rate for the transducer with 30
MHz center frequency and 100% -6 dB bandwidth, is used in ADC, and extra zero
padding interpolation is used in receive beamforming to increase the accuracy. Following
the requirement mentioned above, a high performance analog-to-digital converter (ADC)
AD9230-210 (Analog Devices Inc., 2007) was used to offer maximum 210 mega samples
per second (MSPS) sampling rate with 12-bit sampling resolution. After digitization, the
output digital signals were transferred to the receive beamforming FPGA through a 12-bit
0 20 40 60 80
-70
-60
-50
-40
-30
-20
-10
0
Frequency (MHz)
Magnitude (dB)

Simulated
Measured
37
(24-pin) low voltage differential signal (LVDS) databus. LVDS standard can suppress the
common-mode noise on high frequency data transmission and can easily interface to
current FPGA I/O ports.
3.4.4 FPGA based receive beamformer
In this system design, the whole FPGA based receive beamformer involves four
high performance FPGAs (XC5VFX70T-FF1136, Xilinx Inc., CA), which were used to
support fast transfer of the data from 64 high speed ADCs and perform high speed
receive beamforming signal processing. Each FPGA has 640 user I/Os (Xilinx Inc.,
2009b), which can easily handle 16 ADCs with 12-bit (24-pin) LVDS databus. The
photograph of one of the four receive beamforming boards with one FPGA and 16 ADCs
is shown in Figure 3.13.

Figure 3.13 Photograph of one of four digitizer & receive beamformer boards. Each board can handle 16
high speed ADCs.

38
All ADCs and FPGA are synchronized via a buffered global clock, shown in
Figure 3.14.

Figure 3.14 The FPGA on one of four receive beamformer boards can handle 16 high speed ADCs. Each
ADC connects to FPGA via 12-bit (24-pin actually used) LVDS databus. All ADCs and FPGA are
synchronized through a global clock “G_CLK”.

The FPGA based receive beamformer consists of five modules, which are ADC
pre-processing, delay computing, RAM I, Sum (& others), and RAM II modules. The
block diagram of this receive beamformer is shown in Figure 3.15. The system can
bypass the receive beamformer and works in “Raw RF data” mode to acquire pre-
beamformed RF data.
39

Figure 3.15 Block diagram of the FPGA based receive beamformer, which consists of ADC pre-processing,
delay computing, RAM I, Sum (& others), and RAM II modules.

(1) ADC pre-processing module
The ADC input pre-processing module prepares the input data from ADCs for
later beamforming. Since the digital output of AD9230 is 12-bit width LVDS signal with
offset binary encoding, it has to be converted to single-ended signal with two’s
complement binary for easily computing in FPGA or any further computing cores (Figure
3.16).

40

Figure 3.16 Block diagram of the ADC pre-processing module in receive beamforming FPGA

There are three steps involved in this module. First using “IBUFDS” (differential
signaling input buffer) primitive in Xilinx FPGA converts differential input signals into
single-ended signals. Then using “IODELAY” primitive in Xilinx FPGA adjusts the
input signal delays. This makes sure all the signals from different ADCs to arrive FPGA
at the same time. Finally convert offset binary coding into two’s complement binary by
“exclusive or” the input with hexadecimal number “0x800”.

(2) Delay computing module
One of the challenges of the phased array system is that much more range is
required in the delay profiles. Consider the time delay calculation for the array with pitch
when steering the beam to angle

=
∙ ∙ ∙sin ( )
−
(∙ ∙ ∙cos ( ))

where is the sampling rate; is the sound speed; is the element index; the sample
index and ∈ [1,2048] for dynamic receiving focusing on all 2048 samples (updated
41
with every sample); and is the scan line index. The size of delay table for one FPGA is
16 ∗ 2048 ∗ 256 (for 16 array elements, 2048 foci, 256 scan lines), and it is too large to
be placed into the current FPGA’s internal memory store space. However, with

=

∙∙ ∙ ( )
and

=−
( ∙∙ ∙ ( ))
, then the time delay can be expressed as:

=
+

and

are tables with much more small size (both are 16 ∗ 256) and can be easily
stored in the FPGA. While this method involves a division operation, by using Xilinx
fully pipelined LogiCORE IP divider with “Clocks per Division” setting as one, the
output of one division operation will appear per clock cycle after an initial latency
(dividend-width plus 4) and will not incur extra time cost for computing. The whole
delay-computing module is shown in Figure 3.17.

Figure 3.17 Block diagram of the delay computing module in receive beamforming FPGA

42
(3) RAM I module
Random Access Memory (RAM) I module is used to temporally store the
processed data from ADCs. By using the calculated delay value (output of the delay-
computing module) as the read address of this dual port RAM (Port A is for writing and
Port B is for reading operation), the delayed echo signals can be generated. This RAM
module consisted of 16 small RAMs with 12-bit width and 2048 depth. Each small RAM
is generated by Block Memory Generator (Ver 6.1) of Xilinx ISE in order to handle one
channel data. “CLKA” and “CLKB” are clocks both stemmed from global clock of the
system (Figure 3.18).

Figure 3.18 Block diagram of RAM I module in receive beamforming FPGA for temporally storing the
processed data from ADCs.

(4) Sum (& others) module
This module is used to perform aperture selection (Figure 3.19), apodization
(Figure 3.20) and sum 16-channel delayed data (Figure 3.21).
43

Figure 3.19 Aperture selection sub-module: Pin configurations and function descriptions.

Figure 3.20 Apodization sub-module: Pin configurations and function descriptions.
44

Figure 3.21 Adder sub-module: Pin configurations and function descriptions.

(5) RAM II module
RAM II module (Figure 3.22) is used to store the sum of 16-channel post-
beamforming data and transfer the data to the EMIF interface in DSP. This module
consists of one two-port 12-bit width and 2048-depth RAM, which is an asynchronous
RAM (“CLKA” is stemmed from the global clock of the system and “CLKB” is stemmed
from the clock of EMIF interface).
45

Figure 3.22 RAM II module: Pin configurations and function descriptions.

3.5 DSP based data transceiver
Four 16-channel FPGA based receive beamformer boards are connected to one
DSP board (Figure 3.23) through EMIF (External Memory InterFace) interface (Figure
3.24). The high-performance fixed-point DSP (TMS320C6455, Texas Instruments Inc.,
TX) responds the commands from PC, sends the “frame trigger” and “line trigger” to
transmit beamformer and receive beamformer, gets the data from digitizer & receive
beamformer boards.
46

Figure 3.23 Photograph of the DSP board

Figure 3.24 DSP and four receive beamforming FPGA are connected through EMIF interface

This DSP can finish 9600 million instructions per second (MIPS) at 1.2 GHz
clock rate. Directed memory access (DMA) can be used to low the burden of DSP. The
EMIF_Clk
47
DSP can work in “FPGA based receive beamforming” mode, “DSP based receive
beamforming” mode and “RF data” mode.
In “FPGA based receive beamforming” mode, the DSP sums the post-
beamforming data from the four receive beamforming FPGAs. This is used to help
complete the FPGA based receive beamforming.
In “DSP based receive beamforming” mode, the FPGA based receive beamformer
works in “RF data” mode. The DSP acquires the pre-beamforming data from four
digitizer & receive beamformer boards and performs delay-sum beamforming in DSP.
Since it is easier to program in DSP (c language), this mode can be used to test the
performance on new receive beamforming and signal processing approaches. A DDR2
memory is used to hold temporary data in beamforming process.
And in “RF data” mode, the DSP acquires the pre-beamforming data from four
digitizer & receive beamformer boards and directly send the data to PC for data storing.
This mode can be used for future off-line beamforming test.
3.6 Gigabit Ethernet
All digital data (RF data or bemformed data) are sent to computer for data storing,
signal processing and image visualization via a gigabit Ethernet connection. Gigabit
Ethernet is one of the most popular connection schemes. It can move data at a rate of one
gigabit per second theoretically, faster than another common connection scheme, USB
2.0 with 480 megabit per second throughput.
48
A gigabit Ethernet solution typically consists of an Ethernet Media Access
Control module (EMAC), an Ethernet physical layer transceiver and a RJ45 Jack (Figure
3.25).

Figure 3.25 Block diagram of the gigabit Ethernet connection in current array system

The EMAC module, integrated in TMS320C6455 DSP, is used to control and
move the flow of the packet data between the DSP processor and another host. It supports
full duplex gigabit (1000BaseT: 1000Mbits/sec) operation with standard Gigabit Media
Independent Interface (GMII) to the outside world with IEEE 802.3 standard complied.
Internally, it acts as DMA master to either internal or external memory (connected to
DSP) for fast data transfer.
Gigabit physical layer transceiver DP83865 (Texas Instruments, Inc., TX) directly
interfaces the MAC layer through IEEE 802.3z GMII to twisted pair media (standard
RJ45 jack with magnetics). It is designed for easy implementation of gigabit Ethernet
with ultralow power requirement. A category 5e cable is typically used to connect the
network interface card (NIC) in host PC, which supports gigabit Ethernet networking by
4-pair twisted wires design for noise rejection.
EMAC
Ethernet physical layer
transceiver
RJ45 Jack
with magnetics
Gigabit Network
Interface Card
CAT5e cable
PC
DSP
49
The simple continuous connection test shows that this Ethernet subsystem can
support at least 50 MegaByte/sec (around 400 MegaBit/sec) throughput (Figure 3.26).

Figure 3.26 Screenshot of simple continuous connection throughput test between array system with gigabit
Ethernet and PC with a gigabit NIC

3.7 Data post-processing and visualization on PC
To perform system control, data store, data post-process and image visualization
on a computer, an application software in Microsoft Windows environment has been
developed by using C# language and Microsoft Windows Form technology with
Microsoft .Net 4.0 Framework support. The Graphic User Interface (GUI) of the
developed application software is shown in Figure 3.27.

Figure 3.27 Screenshot of the graphical user interface (GUI) used to visualize the B-mode image. An
image of a tissue-mimicking phantom was shown in this developed GUI.
50
Multi-thread programming is widely utilized to optimize the program’s
performance. One of the benefits of using multi-thread programming in an application is
that each thread can be executes asynchronously. This allows time-consuming tasks to be
performed in the background thread, such as gigabit Ethernet data flow control, B-mode
signal processing (filtering, envelope detection and log compression), scan conversion,
data saving, and stage controlling in C-mode scan. And at the same time, the system
initial setup, system control and image visualization in front GUI thread remain
responsive. To avoid unpredictable data corruption when different threads attempt to
access the same resources, thread synchronization should be carefully coordinated. Figure
3.28 demonstrates the relationship of the multi-thread programming scheme.

Figure 3.28 Block diagram of the multi-thread application for the data post-processing and visualization on
PC. Front thread and background thread should be carefully synchronization to avoid unpredictable data
corruption.

51
3.8 System evaluation
3.8.1 Signal-to-noise ratio of the system
Signal-to-noise ratio (SNR) is a measure to quantify how much a signal has been
corrupted by noise. It is defined as:

=20 log

, (3.1)
where and
are amplitude of signal and noise, respectively.
The noise was acquired by powering on the system but without sending the
triggers to transmit beamformer. The signal amplitude is the full scale of ADC. Figure
3.29 shows that the SNRs for all 64 channels are around 45 3 ± dB.

Figure 3.29 Signal-to-noise ratio (SNR) for all 64 channels

0 10 20 30 40 50 60
40
45
50
Channel index
SNR (dB)
52
3.8.2 26 MHz linear phased array
To evaluate the developed ultrasonic array system, the system was paired with a
64-element 2-2 piezo-composite linear phased array (Figure 3.30), which was developed
recently in our lab.

Figure 3.30 Photograph of the prototype 26.3 MHz linear phased array

The center frequency of this linear phased array is 26.3 MHz. The pitch of this
array is 30 μm, which is around half wavelength of 26.3 MHz ultrasound in water. More
specifications are shown in Table 3.1.
Table 3.1 Specifications of the 26.3 MHz prototype linear phased array
Specifications Values
Center frequency (MHz) 26.3
Bandwidth (%) 58
Number of elements 64
F number 2.6
Element width in azimuth direction ( μm) 24
Kerf width in azimuth direction ( μm) 6
Pitch in azimuth direction ( μm) 30
Element length in elevation direction (mm) 2
Lens with elevation focus at (mm) 5.5
3
to
im
w
tu
S
ax
Fi
st
tu

.8.3 Wire p
The u
o evaluate th
mage plane
wire phantom
ungsten wire
hown in Fig
xial distance
igure 3.31 A w
teps are used to
ungsten wires (
phantom im
usual way to
he cross-sec
and immers
m used in th
es of 20 μm i
gure 3.31, fiv
e, respectivel
wire phantom u
o hold fine wire
20 μm diamete
mage
assess the s
ctional image
sed in an ec
e present ex
in diameter
ve wires are
ly (Lei et al.
used in the exp
es. (Right) Sch
er)
spatial resolu
e of a wire
cho-free me
xperiment co
(California F
separated by
, 2007).

periments. (Lef
hematic of the c
ution of an
phantom, w
dium (norm
onsisted of f
Fine Wire C
y 0.65 mm a
ft) The photogr
cross section o
ultrasonic im
which is perp
mally in disti
five diagona
Company, Gr
and 1.5 mm
raph of the wir
of the wire phan
maging syste
pendicular t
illed water)
ally arranged
rover Beach,
in the latera

re phantom, mu
ntom with five
53
em is
to the
. The
d thin
, CA).
al and
ultiple
e
54
In the process of imaging the wire phantom, the transmit focus was set to 5.5 mm
in the system, and the total number of scan lines was set to 200 within 60 (±30) degrees
range. The dynamic range of logarithmic compression was 50 dB. The wire phantom
images obtained by the phased array and Field II (Jensen et al., 1992; Jensen, 1996b)
simulation are shown in Figure 3.32. The lateral and axial beam profiles obtained from
the wire located closest to the transmit focus are shown in Figure 3.33, from which the
lateral and axial resolutions were determined to be from the -6 dB envelope width, which
were 209 μm and 104 μm, respectively. These measured results were in reasonable
agreement with the Field II simulation of the phased array system (195 μm lateral and 76
μm axial resolutions, respectively). The backing of the transducer may cause the
deterioration of the axial resolution, compared to the result from the Field II simulation.
Since if the sound energy cannot be fully absorbed by the backing, some of the energy
will be reflected forward into the target. This will lead to a longer waveform duration and
lower in resolution. The lateral beam distribution shows asymmetry. One possible reason
is that some of the array elements may have low sensitivity compare to others.

Fi
ph
ar

Fi
th
re

3
fr
igure 3.32 (Le
hased array and
re in 50 dB dyn
igure 3.33 Me
he current imag
espectively.
.8.4 Anech
EL M
requency ult
eft) Measured w
d the current im
namic range.
easured lateral
ging system. Th
hoic sphere p
Madsen et al.
trasound ima
wire phantom B
maging system
and axial sprea
he -6 dB latera
phantom im
developed a
aging system
B-mode image
m. (Right) Field
ad obtained wit
al and axial reso
mage
an alternativ
m (Madsen et
e obtained with
d II simulated w
th the 26.3 MH
olutions were f
ve method to
t al., 2010).
h the 26.3 MHz
wire phantom B
Hz customized
found to be 20
o assess perfo
With the ph
z custom-desig
B-mode image

phased array a
09 μm and 104
formance of
hantom consi
55

gned
. Both
and
μm,
high-
isting
o
p
st
p
Fi
Th
sm

T
ob
F
d
f anechoic
erformance
tudy, such a
latform.
igure 3.34 Tis
he diagram of
mall blocks wit
The si
Totally 200 s
btained from
igure 3.35,
iameter was
spheres of
of a high f
phantom (F
sue-mimicking
the tissue-mim
th echogenic b
ingle transm
scan lines w
m the sectio
which are
300 μm and
different si
frequency ul
Figure 3.34)
g phantom. (To
micking phanto
ackground (M
mit focus was
were used to
ons with 10
all in 50 dB
d the system
ize embedde
ltrasonic im
is used to ev
op) A photogra
m. Anechoic s
adsen et al., 20
s used to foc
o scan beam
090-, 825-, 4
B dynamic
could not re
ed in a tiss
maging syste
valuate the c
aph of a tissue-
spheres with di
010)
cus the ultras
ms within 60
400-, 300- μ
range. The
esolve 200 μ
sue-mimicki
em can be e
current phas
-mimicking ph
ifferent sizes ar
sound beam
0 degrees ra
μm spheres
minimum d
μm anechoic
ing material
evaluated. In
sed array ima

hantom. (Bottom
re embedded in
in 5.5 mm d
ange. The im
are displaye
detectable sp
spheres.
56
l, the
n this
aging
m)
n nine
depth.
mages
ed in
phere
Fi
M
tr
30

ev

w
re
igure 3.35 Ima
MHz custom-de
ansmit (5 mm)
00- μm sphere p
The c
valuate the q
where ,
espectively.
ages of an anec
esigned phased
) and dynamic
phantom image
contrast-to-n
quality of ac
are the mean
is the sta
choic sphere ph
d array with the
receive foci w
es are 4.14, 4.0
noise-ratio (
quired imag
n of the sign
andard devia
hantom with 10
e current imagin
were used to for
03, 4.04, and 3
(CNR) is a
es. It is defin
=
| −
nal intensitie
ation of the b
090-, 825-, 400
ing system (all
rm the images.
.27, respective
measure u
ned as
− |

es for the ta
background
0-, 300- μm sph
in 50 dB dyna
The CNR of t
ely.
used in med
arget and for
(Seo et al.,
heres using the
amic range). On
the 1090-, 825-
dical imagin
(4.
r the backgro
2008). The
57

e 26.3
ne
-, 400-,
ng to
1)
ound,
CNR
58
of the image of 1090-, 825-, 400-, 300- μm spheres were estimated to be 4.14, 4.03, 4.04
and 3.27 at the range close to the transmit focus.
3.8.5 In vitro rabbit eyeball experiment
Rabbit eyes are one of the ideal and frequently used animal models in ophthalmic
research since the anatomy of rabbit eyes is similar to that of the human eyes (Tsonis,
2008). High frequency ultrasound has played an important role in the imaging of the
anterior chamber angle of eyes (Ishikawa, 2007), which is crucial in the assessment of
glaucoma patients. To evaluate the performance of the phased array system in vitro
environment, an excised eye from a New Zealand white rabbit (Sierra for Medical
Science, Whittier, CA) was imaged. In this experiment, the transmit focus was also set to
5.5 mm in the system, and the total number of scan lines was set to 200 within 60 (±30)
degrees sector scanning range. 50 dB dynamic range was used to show the image. An
image of a section of the anterior portion of the excised rabbit eye is shown in Figure
3.36. The anatomical details, such as cornea, iris, ciliary muscle, and lens, are clearly
visible, which can be used to measure the anterior chamber angle and diagnose the
potential disease in the iris.
Fi
ar
Th

3
w
µ
im
tr
w
igure 3.36 B-m
rray and imagin
he cornea, iris,
.9 Discussi
In the
within ±45 de
µm sphere wi
mage becom
ransmitting a
will decrease
mode image of
ng system. One
, ciliary muscle
ion and sum
e sector scan
egree. Figure
ith ±45 degr
mes darker.
and receivin
when increa
f an excised rab
e transmit (5.5
e and lens are c
mmary
nning of line
e 3.37 show
ree steering a
This is bec
ng in the dire
asing the ste
bbit eye obtain
mm) and dyna
clearly visible
ar phased ar
s the image
angle. When
cause the in
ection norma
ering angles
ned with the 26
amic receive fo
in the image w
rrays, the ult
of an anech
n the beam s
dividual ele
al to the tran
s (Hoskins et
.3 MHz custom
oci were used t
with 50 dB dyn
trasound bea
oic sphere p
steers beyond
ements are m
nsducer’s su
t al., 2010).
m-designed pha
to form the ima
namic range.
am usually s
phantom with
d ±30 degree
most efficie
urface. Sensi
59

ased
age.
steers
h 530
e, the
ent in
itivity
Fi
cu

el
sh
re
ph
im
p
th
fa
igure 3.37 B-m
ustom-designed
In con
lement 26 M
hows the -6
easonable a
hantom ima
mage of a
erformance
he purpose
acilitating th
mode scan ima
d linear phased
nclusion, the
MHz center
dB spatial r
greement w
ages demons
rabbit eyeb
in biomedic
of testing h
he developme
age of an anech
d array with the
e developed
frequency
resolution of
with the Fie
strated its ca
ball in vitro
cal imaging
high freque
ent in novel
hoic sphere pha
e current imagi
d phased arr
phased arra
f 209 μm (la
eld II simul
apability to
o was also
applications
ncy phased
array signal
antom with 530
ing system. Th
ray system i
ay transduce
ateral) and 10
lation. Anec
detect cysts
acquired to
s. This imag
d arrays und
l processing
0 μm sphere us
he steering ang
is capable o
er. The wire
04 μm (axia
choic cyst t
s of 300 μm
o demonstra
ging platform
der developm
algorithms.
sing the 26.3 M
gle is ± 45 degr
of handling
e phantom im
al), which we
tissue-mimic
m in diameter
ate the syst
m is designe
ment as we
60

MHz
ree.
a 64-
mage
ere in
cking
r. An
tem’s
ed for
ell as

61
Chapter 4 Non-destructive Testing with a High Frequency
Ultrasonic Array System
4.1 Introduction
Besides applications in biomedical imaging, high frequency ultrasonic array and
system can be also applied in industrial non-destructive testing (NDT), which refers to a
wide group of techniques to evaluate/test the structure of a component or properties of a
material without causing damages. While using ultrasonic waves in NDT has a long
history and many different approaches, one of the most common and extensively used
methods is pulse-echo method, where a transducer acts first as an emitter of ultrasonic
pulses and then as a receiver to detect echoes from defects in the test object (Figure 2.2).
This is because the defects within the materials are likely to cause regions with acoustic
impedance mismatch, which could be detected by using pulse-echo ultrasound.
By increasing the frequency of the ultrasound being used, smaller scale defects
can be visualized. In addition, by using arrays instead of single element transducers, the
scanning speed could be improved. Therefore, high frequency ultrasonic array technique
is well suited to this non-destructive testing application for micro defect detection.
While the conventional ultrasonic NDT using contact method to have good
imaging results, for some test samples that cannot be scratched, non-contact mode is
preferred. So instead of placing the transducer directly against the samples, water
coupling should be used in this special ultrasonic evaluation. However, the huge
differences of sound speed between in water and in solid samples will cause refraction
62
and introduce delay errors in beamforming. To address, a modified beamforming scheme
for the delay error correction will be presented in this chapter.
This study will focus on using high frequency ultrasonic array with pulse-echo
method to evaluate ceramic materials, such as silicon carbide, immersed in water.
Silicon carbide (a compound of silicon and carbon with chemical formula SiC) is
ceramic material, which has good ultrasonic transmission characteristics. It is an
excellent material for high performance optical application due to its high hardness, high
thermal conductivity and low thermal expansion coefficient (Johnson et al., 2002). For
instance, silicon carbide material was selected for the production of the world largest
telescope mirror ever (with 3.5 meter aperture), which was used on the Herschel Space
Observatory launched by European Space Agency in 2009 (Doyle et al., 2009). In Table
4.1, some properties of SiC material are listed for reference (Munro, 1997; Johnson et al.,
2002).

Table 4.1 Properties of silicon carbide at 20
o
C
Properties (unit) Values
Density 3.21 ( /
)
Elastic modulus (E) 466 (GPa)
Thermal expansion ( ) 2.2 (x10
-6
/K)
Thermal conductivity ( ) 280 (W/mK)
Longitudinal sound velocity 11,820 ( / )
Transverse sound velocity 7,520 ( / )
Characteristic acoustic impedance = ∙ 37.4 ( )

While SiC is an excellent material for high performance optical application, the
mechanical and thermal properties of SiC are highly dependent on the material
63
microstructure. The performance of the SiC mirror will be degraded if there exists any
defects inside. The introductions of defects into the SiC mirror are often during the
manufacturing or machining of the mirror (Hurst et al., 1987). One example is the
presence of unreacted additive within the material. Depending of the amount of residue, it
can result in increasing internal stress, affecting the mechanical properties of the mirror.
If the stressed region is close to the mirror surface, the optical properties would be also
affected. Due to the high cost of SiC mirror fabrication, it would hence be beneficial if
the internal conditions of the SiC mirrors can be evaluated in a non-destructive manner
before integrating the mirrors to the optical systems. Ultrasound scanning would be one
of the techniques that have the potential in testing the defects in the SiC mirror. This is
because the defects within the mirror are likely to cause regions with acoustic impedance
mismatch, which could be detected by using pulse-echo ultrasound. By increasing the
frequency of the ultrasound used, small-scale defects can be visualized.
4.2 Methods
4.2.1 Array transducer and scanning method
To perform non-destructive testing, the ultrasonic array system was paired with an
experimental 30 MHz 1D array transducer (Blatek Inc., State College, PA). It is a
Piezoelectric Composite based Micro-machined Ultrasound Transducer (PC-MUT).
Some specifications are shown in Table 4.2.

64

Table 4.2 Specifications of the 30 MHz linear array transducer
Specifications Values
Center frequency (MHz) 31.4 ± 0.5
Bandwidth (%) 42 ± 4
Sensitivity (mV) 400 ±61
Crosstalk (dB) -28 to -33
Total number of elements 64
Pitch ( μm) 132
Element width ( μm) 116
Array elevation (mm) 4

Firmware of the imaging system was modified to perform linear scan. Scan lines
are forming by first activating and exciting a group of 32 elements at left end of the array,
then deactivate one element at most left and add one new element from right to the active
group to form a new scan line. The new scan line is parallel to the one before with one
pitch distance. Using electronic scanning instead of mechanical scanning, the higher
imaging speed could be achieved.
4.2.2 Reflection
Reflection is the phenomenon of the change of propagation direction of a
wavefront at the boundary of two different materials. Reflection coefficient is the ratio of
the intensity of the reflected wave to the incident wave. It decides the amount of energy
reflected as a percentage of the original energy. Reflection coefficient can be calculated
by:

= −
+

65
where and are the acoustic impedance of two materials. The acoustic impedances
of water, aluminum, and SiC are around 1.5, 17.1, and 37.5 MRayl, respectively. The
calculation results show only 6.2% energy can come back from the bottom of aluminum
sample, and 1.9% energy will come back from the bottom of SiC sample (Figure 4.1).
This energy percentage is very small for transducer reception. Increasing the transmit
voltage might be one of the solutions. In the experiment, ± 30V transmit voltage was
used to make sure the bottom of sample being seen.

Figure 4.1 Ultrasound energy distributions (left) in water-aluminum and (right) in water-SiC interfaces.
Only 6.2% ultrasound energy in water-aluminum and 1.9% ultrasound energy in water-SiC interface can
return to the transducer.

4.2.3 Refraction receive beamforming
One of the main challenges in visualizing the internal structure of the solid
materials is the high speed of sound within the target. The sound speeds within aluminum
and SiC are approximately 4 times and 8 times of the speed in water, which can cause
large errors in time delay calculation for beamforming. Furthermore, the refraction of
66
ultrasound wave at the water-solid boundary would also complicate the problem.
Previous studies didn’t address this problem due to the limited thickness of test sample
(Portune et al., 2010). Using the conventional beamforming scheme would result serious
artifacts.
In view of this, a modified beamforming scheme is suggested for the delay error
correction. The basic idea is first to estimate the location of the water-solid interface, and
then calculate the travelling distance of the ultrasound with the two materials separately.
Snell’s law describes the relationship between the angles of incidence and refraction. It
states that
sin sin =

where and are the incident and refracted angles; and are the sound speed in
water and in solid, respectively. From the geometry (Figure 4.2), we have
= tan

−
= tan

then
sin tan

−
sin tan

=

Using Newton-Raphson method in numerical analysis, for each , we can have numerical
solution of =() . So the receive time delay will be
= −

67
where
=
+

=
( − )
+
+
+

Figure 4.2 Simplified diagram showing the refraction of ultrasound beam at the SiC-water or aluminum-
water boundary

4.2.4 C-mode scan and motorized stage system
C-mode scan (simply called C-scan) is a scan to form images parallel to the
surface of the scanned object. To form a C-scan image, a motorized stage system is
usually used to move the transducer or test object in an X-Y plane. The positions of any
inner discontinuities are synchronized with the moving transducer along X-Y coordinates.
α
β
x
r
Silicon Carbide
(or Aluminum)
Water
Array
Point object
68
C-scan images in various depths give a convenient way to show a more complete picture
of the defect positions.
While C-scan in an ultrasonic array system is similar to the one from the
conventional single element system and the array moves along X-axis and Y-axis like
single element transducer does, in each mechanical moving step, an electronic linear
scanning is performed in array system (Figure 4.3). Using electronic scanning to replace
part of mechanical scanning, the total scanning time could be dramatically reduced.

Figure 4.3 1D array transducer moves in a 2D (X-Y) plane to form a C-scan image. Typically, in X
direction, the array moves with discrete steps. In each step, electronic linear scanning is performed.

The stage system used in this C-scan imaging consists of a motorized stage
controller (SHOT-204MS, Sigma Koki Co., Ltd, Japan), and an X-Y motorized linear
stage (SGSP33-200(XY), Sigma Koki Co., Ltd, Japan), shown in Figure 4.4. It has 200
mm by 200 mm X-Y travel range and 0.006 mm positional repeatability with the help of
5-phase stepping motor and high-precision guide actuator. The stage was fastened on a
69
600 mm by 450 mm by 14 mm heavy base plate in order to attenuate the vibration of the
system when the stage moves.

Figure 4.4 Photograph of the motorized stage system. (Left) SHOT-204MS motorized stage controller.
(Right) SGSP33-200(XY) 2D linear stage with array transducer fixed on it

The SGSP33-200(XY) motorized linear stage is powered and controlled by the
SHOT-204MS stage controller. The stage controller connects a PC via RS232 serial
interface (a RS232-to-USB adaptor can be used for the new computer without RS232
interface) to obtain the commands from the computer and send back the real-time stage
information (Figure 4.5). The programming flow chart to control the stage is shown in
Figure 4.6.
70

Figure 4.5 Block diagram of the connection between the motorized stage system and the high frequency
ultrasonic array system.

Figure 4.6 Flow chart for controlling the motorized stage, shaded blocks run repeatedly.
71
4.3 Imaging results
4.3.1 Aluminum sample imaging
Since aluminum is a common material and easy to cut compared to SiC, a pre-
drilled aluminum sample was imaged first to evaluate the array and the system in non-
destructive testing. Some properties of aluminum are shown in Table 4.3.

Table 4.3 Properties of aluminum at 20
o
C
Properties (unit) Values
Density 2.7 ( /
)
Longitudinal sound velocity 6,320 ( / )
Transverse sound velocity 3,150 ( / )
Characteristic acoustic impedance = ∙ 17.1 ( )

The pre-drilled aluminum sample block is shown in Figure 4.7(A). Eight holes
were drilled from the side of the sample. The diameters of eight holes (#1 to #8) were
around 300 µm, 500 µm, 600 µm, 700 µm, 800 µm, 900 µm, 1000 µm, and 2400 µm,
respectively (Figure 4.7(B)). The depth of these holes were different (Figure 4.7(C)) and
the hole #8 was drilled all through the sample. The aluminum sample was immersed in a
small water tank filled with deionized water when imaging.
72

Figure 4.7 Aluminum sample. (A) Photograph of an aluminum sample; (B) Schematic of the side view of
the aluminum sample; (C) Schematic of the top view of the aluminum sample

Using the modified refraction time-delay beamforming scheme, the obtained B-
mode image from the aluminum sample is given in Figure 4.8. All eight holes and bottom
surface of the aluminum sample can be found in the B-mode image.

Figure 4.8 B-mode image of the aluminum sample (45 dB dynamic range)

73
C-mode image of the bottom surface of the aluminum sample is shown in Figure
4.9. The region of holes is clearly defined in the C-scan image.

Figure 4.9 C-mode image of the aluminum sample (bottom surface)

Maximum amplitude projection (MAP) image is imaging reconstruction method
normally used in photoacoustic imaging. The MAP image of aluminum sample was
constructed by first removing the top and bottom surface signals and then projecting the
remaining maximum signal amplitude along each A-line onto the corresponding length-
width plane. Through the MAP image, the holes can be more easily differentiated in
Figure 4.10.
74

Figure 4.10 Maximum amplitude projection (MAP) image of the aluminum sample. The image was
constructed by first removing the top and bottom surface signals and then projecting the remaining
maximum signal amplitude along each A-line onto the corresponding length-width plane.

4.3.2 Silicon carbide sample imaging
A SiC sample block was pre-diced, shown in Figure 4.11. The cuts are located at
the bottom side (with respect to the transducer position) of the SiC block. The width of
each cut is around 350 µm, and the spacing between the cuts range from 500 µm to 4,000
µm, increasing from left to right. The SiC sample was immersed in a small water tank
filled with deionized water.

Figure 4.11 Photograph of a silicon carbide sample block
75
The obtained B-mode images of the SiC block are given in Figure 4.12 and Figure
4.13, using the conventional time-delay beamforming scheme and the modified refraction
beamforming scheme, respectively. When comparing these two images, an improvement
can be seen by employing the refraction delay-error correction. The bottom surface is not
clearly visible without using the error correction. Moreover, the cut regions are much less
clearly defined in Figure 4.12 than in Figure 4.13. These both indicate an improvement in
defect detection ability by using the modified beamforming scheme.

Figure 4.12 B-mode image of the silicon carbide sample with conventional time-delay beamforming in 35
dB dynamic range

Time of flight ( μ sec )
Lateral distance (mm)
5 10 15 20 25 30 35 40
5
5.5
6
6.5
7
7.5
8
8.5
9
9.5
10
76

Figure 4.13 B-mode image of the silicon carbide sample with modified time-delay beamforming in 35 dB
dynamic range

4.4 Discussion and summary
One point to note is that the lateral resolution within the aluminum and SiC blocks
are worse than that in the water. This is due to the increased wavelength following by the
increased sound speed. Possible solutions to this include the use of advanced
beamforming techniques like synthetic aperture. Increasing the ultrasound frequency is
another option, since the attenuation within metal materials and ceramic materials is very
low and should not be a problem in higher frequency ultrasound imaging.
In conclusion, without additional change in the system hardware, the current high
frequency array system can be paired with a linear array and has the potential to be used
as a non-destructive testing approach for evaluating the quality of metal materials (such
Time of flight ( μ sec )
Lateral distance (mm)
5 10 15 20 25 30 35 40
5
5.5
6
6.5
7
7.5
8
8.5
9
9.5
10
77
as aluminum) and ceramic materials (such as SiC). A motorized XY stage allows the
array system to acquire C-mode scan as well as normal B-mode scan with bigger
scanning area. C-mode image can rapidly locate the inner defects in transverse plane, and
B-mode image can tell more details on the damages in depth direction. The combination
of these two modes is quite useful in industrial non-destructive testing. Electronic
scanning in array system could dramatically reduce the total scanning time compare to
single element system. This non-contact non-destructive testing approach has also been
demonstrated successfully by utilizing the modified refraction beamforming methodology.

78
Chapter 5 Acoustic Trapping with a High Frequency Linear
Phased Array
5.1 Introduction
Pioneered by Arthur Ashkin and his colleagues at Bell labs, optical tweezers
technique (also known as the single beam gradient force trap) introduces a method to
manipulate microparticles using a tightly focused laser beam (Ashkin, 1970; Ashkin et al.,
1986). This technique relies on a force created by the high gradient in the electric field
near the beam waist of the focused laser beam, which is able to trap micro dielectric
particles in three dimensions. And optical tweezers have become a powerful tool for
research in the fields of biological science, such as trapping and manipulation of single
live motile bacteria and Escherichia coli bacteria (Ashkin et al., 1987), sorting
microscopic particles (Macdonald et al., 2003) and even stretching Deoxyribonucleic
Acid (DNA) (Wang et al., 1997).
While the highly focused laser beam may manipulate the microparticles very
precisely, it is can only be used in light transparent media. Moreover, the local heating
and photodamage induced by highly focused laser beam may cause biological damages in
cells or living tissues (Neuman et al., 1999).
Similar to the trapping mechanism of optical tweezers (Svoboda et al., 1994;
Moffitt et al., 2008), when acoustic gradient force (from refraction) exceeds scattering
force (from reflection), an object can be attracted and trapped by a tightly focused
ultrasound beam (Lee, J. et al., 2005; Lee, J., Lee, et al., 2010). Acoustic trapping could
79
be more useful in the light opaque media. It was also demonstrated that the thermal and
mechanical effects in acoustic trapping are negligible when the energy is maintained in
the diagnostic range (Lee, J. et al., 2006).
Recently, high frequency single element ultrasonic transducers have been
successfully used to carry out single beam acoustic trapping (Lee, J. et al., 2009; Lee, J.,
Teh, et al., 2010; Hsu et al., 2012). In these approaches, in order to move a trapped
microparticle, a mechanical scanning stage has to be utilized to move the transducer and
its focus. To avoid mechanical movement of transducers, ultrasonic phased arrays could
be an optimal option. An ultrasonic linear phased array transducer is a transducer
consisting of multiple small transducer elements, which usually are rectangular in shape
and arranged on a straight line. The advantages of phased array transducers over
conventional single element transducers are their capabilities of steering the ultrasound
beam into different directions and/or changing the focus at different depths, not by
mechanically moving transducers, but by applying electronic phase shift/time delays on
the transmitting pulses to the elements of the phased array. Eliminating the mechanical
movement of the transducer increases the system reliability and the speed of the
experiment.
This chapter presents experiments showing that it is possible to trap and move
microparticles with a high frequency ultrasonic linear phased array without mechanical
movement of the transducer.
80
5.2 Methods
The 26.3 MHz linear phased array transducer, which was previously mentioned in
imaging system evaluation, was also used in this experiment. Part of the developed high
frequency ultrasonic system, more precisely, the field programmable gate array (FPGA)
based 64-channel transmit beamformer and a 64-channel bipolar pulser (50 Vpp) were
used to drive the phased array. The different transmit time delay patterns could be loaded
into the transmit beamformer to steer and focus the ultrasound beam.
From phased array geometry shown in Figure 5.1, the transmit time delay of any
element of a phased array for a specific focal point can be calculated from the equation
below (Thomenius, 1996):

=

−

+

−

(6.1)
which allows the ultrasound wave from all elements to arrive at the transmit focal point at
the same time. In the equation above, is the transmit time delay for array element , and
is the distance from the center of array element to the origin. The symbols
and

are the coordinates of the focal point while
is the distance from the origin to the
focal point and is the sound speed in the medium surrounding the array.
81

Figure 5.1 Phased array geometry and focal point

In the experiment, three parabolic transmit time delay patterns (Figure 5.2) were
loaded into the FPGA based transmit beamformer, which would focus the propagating
ultrasound waves to three specific points. Time delay pattern #1 generated a focus at 5
mm (depth/axial direction) and 0 μm (azimuth direction) away from the center of the
phased array. Time delay pattern #2 and #3 generated two foci at the same axial distance
(5 mm) as the delay pattern #1, but at 350 μm and 450 μm (azimuth direction) away from
the center of the phased array, respectively.
i
x
82

Figure 5.2 Three transmit time delay patterns were loaded into the transmit beamformer before the
experiment.

Using Field II ultrasound simulation (Jensen et al., 1992; Jensen, 1996b), the
normalized ultrasound intensity (in dB) is plotted in Figure 5.3, which demonstrates by
implementing three transmit time delay patterns, the phased array can steer the beam and
focus at 0 μm, 350 μm and 450 μm away from the center of the phased array in the
azimuth direction, respectively. The highly focused ultrasound could generate a sharp
intensity variation in the azimuth direction and allow particles to be trapped at desired
positions.
83

Figure 5.3 Simulation of the ultrasound intensity when the phased array steers the beam and focuses at
three different azimuth positions where depth/axial positions are all 5 mm. The color represents the
normalized intensity. (a) 0 μm in azimuth direction. (b) 350 μm in azimuth direction. (c) 450 μm in azimuth
direction.

The experimental arrangement for acoustic trapping and translation of
microparticles with a phased array is shown in Figure 5.4. The high frequency phased
array was mounted on a transducer holder. The array and the holder remained motionless
during the experiment. The phased array was immersed in deionized (DI) water in a
designed chamber. There was a transparent mylar film on the bottom of the chamber. The
motion of the microparticles could be observed through the mylar film by an inverted
microscope and recorded by a CMOS camera attached to the microscope. Polystyrene
Depth [mm]
(a)
-1 0 1
1
2
3
4
5
6
7
8
Azimuth [mm]
(b)
-1 0 1
1
2
3
4
5
6
7
8
(c)

-1 0 1
1
2
3
4
5
6
7
8
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
dB
84
microspheres of 45 μm mean diameter were added into the chamber as targeted particles
to be trapped and moved.

Figure 5.4 Block diagram of the linear phased array acoustic trapping and moving experiment.

5.3 Results
In the experiment, the position of the phased array was fixed. As the experiment
began, the micro-particles were at rest, shown in Figure 5.5(a). After applying transmit
time delay pattern #1 and exciting the phased array elements, the ultrasound beam was
transmitted and focused at 0 μm in the azimuth direction and the micro-particles were
being trapped, shown in Figure 5.5(b). Next, in Figure 5.5(c), applying delay pattern #2,
the micro-particles travelled to the position 350 μm away from the original trapping
location in the azimuth direction. Again, with the time delay pattern #3, the micro-
FPGA based 64-channel
transmit beamformer
64-channel pulser
DI Water
Microscope & CMOS
camera
64-element phased array
Microparticles
Mylar film
Depth
Azimuth
Elevation
85
particles were further moved to the position 450 μm away from the original trapping
location, shown in Figure 5.5(d). To move the particles back to the original trapping
location, we then applied delay pattern #2 to move the particles back to the “350 μm”
position, shown in Figure 5.5(e). Continued applying delay pattern # 1, the particles
eventually moved back to the original trapping position, shown in Figure 5.5(f).

Figure 5.5 Micro-particles (45 μm mean diameter) were trapped and moved with transmit different time
delay patterns. E: Elevation direction; A: Azimuth direction. (a) No ultrasound was transmitted. (b) Time
delay pattern #1. (c) Time delay pattern #2 (d) Time delay pattern #3. (e) Back to time delay pattern #2. (f)
Back to time delay pattern #1.

5.4 Discussion and summary
A high frequency ultrasonic phased array is shown to be capable of trapping and
translating microparticles precisely and efficiently, made possible due to the fact that the
acoustic beam produced by a phased array can be both focused and steered. Acoustic
manipulation of microparticles by a phased array is advantageous over a single element
86
transducer since there is no mechanical movement required for the array. Experimental
results show that 45 μm diameter polystyrene microspheres can be easily and accurately
trapped and moved to desired positions by a 64-element 26 MHz phased array.
The beam width in azimuth direction of the current phased array was around 200
μm, it could not yet trap and move a single particle of several-micrometer diameter.
Work is underway to develop a phased array with higher frequency and more elements
(bigger aperture size) to form a more tightly focused beam, which should allow single
particle trapping and translation. These results also suggest that microparticle trapping
and translation in both azimuth and elevation directions may be realized with a 2D high
frequency ultrasound array.
87
Chapter 6 Conclusion and Future Work
6.1 Conclusion
This dissertation reports the development of a high frequency ultrasonic phased
array imaging system and its potential applications.
The developed ultrasonic imaging platform can be paired with a prototype 64-
element 26 MHz phased array transducer. It provides B-mode imaging and has the
capability of raw RF data acquisition. A wire phantom image showed that -6 dB lateral
and axial resolutions were 209 μm and 104 μm, respectively, which were in good
agreement with the Field II simulation, 195- μm lateral resolution, and 76-μm axial
resolution. Anechoic cyst tissue-mimicking phantom images demonstrated its capability
to detect cysts of 300 μm in diameter. An image of a rabbit eyeball in vitro was also
acquired to demonstrate possible applications in biomedical imaging. This imaging
platform is designed for the purpose of testing high frequency phased arrays under
development as well as facilitating the development in novel array signal processing
algorithms.
Without additional change in the hardware, the current high frequency array
system can be also applied in the industrial non-destructive testing. A motorized XY
stage allows the array system to acquire C-mode scan as well as normal B-mode scan
with bigger scanning area. Electronic scanning in array system can dramatically reduce
the total scanning time compare to single element system. Aluminum and silicon carbide
88
targets were imaged to demonstrate the ability of the system in finding micro defect
inside the targets.
Besides the applications in biomedical imaging and non-destructive testing, the
high frequency ultrasonic phased array and system are shown to be capable of trapping
and translating microparticles precisely and efficiently, made possible due to the fact that
the acoustic beam produced by a phased array can be both focused and steered. Acoustic
manipulation of microparticles by a phased array is advantageous over a single element
transducer since there is no mechanical movement required for the array. Experimental
results showed that 45 μm diameter polystyrene microspheres could be easily and
accurately trapped and moved to desired positions by a 64-element 26 MHz phased array.
6.2 Future work
6.2.1 Improve the current system
Phased array transducers have fewer elements, smaller aperture size than normal
linear array transducers. One advantage is that the smaller footprint allows phased array
transducers to image between the ribs and it is good for echocardiography. High
frequency transducers have much smaller size, and this makes it possible to image heart
movement even in small animals, such as mice. However, the heart of a mouse can beat
with a very fast heart rate (400 ~ 800 beats/minute or 6 ~ 14 beats/second). To image the
movement of the mice heart, the array system requires not only superior spatial resolution
but also a higher temporal resolution/imaging frame rate (Zhang et al., 2010). Current
platform has the frame rate at 12 frames per second with 200 scan lines. The relative low
89
frame rate might be attributed to two bottlenecks. One is the low data transfer speed on
DSP with the gigabit Ethernet connection. This can be resolved with a high-speed digital
I/O boards (for example, PCIe-6537, National Instrument Inc., Texas), which can provide
1600 Mbits/sec throughput and is sufficient for a frame rate of over 100 frames/sec (with
256 scan lines/frame, 2048 samples/scan line, 12-bit signal). With multiple (up to five)
PCIe-6537 working together, real-time (20 frames/sec on 64-element array) pre-
beamforming data transfer is possible, which is also desired in software beamforming to
reduce the system complexity and lower hardware cost (Li et al., 2011). Another
bottleneck is the speed of scan-conversion on the computer. This can be improved
dramatically by implementing OpenGL (with GPU computing) protocol (Steelman et al.,
2011). Some experiments on hearts of small animals can be implemented after the
improvement of the frame rate.
For a phased array system, focusing to a specific depth is achieved by both beam
steering and transmit/receive focusing to reduce the effective beam width and improve
lateral resolution. In prototype phased array system, the dynamic receive focusing on
every sample point was implemented. However, there was only one transmit focusing in
the system. To maintain lateral resolution as a function of depth, apply multiple transmit
focal zones is a good choice. Each focal zone requires separate pulse echo sequences to
acquire data. One way to do is to transmit multiple times along one beam for different
focal zone but only accept the echoes in each focal zone. This may require extra time and
reduce the frame rate.
90
6.2.2 System function expansion: color flow Doppler
The system can be further developed to exploit its potential in measuring flow
velocities in small tissue. This can give valuable information in applications like eye
imaging (Silverman et al., 1999) and tumor blood flow assessment (Jain, 1988; Jain et al.,
1997). In order to obtain the flow information, the conventional color flow imaging
algorithms (Jensen, 1996a; Cobbold, 2007) can be readily implemented in the current
system to provide a robust yet efficient way to estimate the flow velocities within the
region of interest.
6.2.3 Design transmit subsystem for 1D/2D higher frequency array trapping
The current 26 MHz linear phased array can be used to trap and manipulate 45 μm
microparticles. It could not yet trap a single microparticle of several-micrometer diameter.
Higher operational frequency will result in higher spatial resolution and narrower beam
width. Recently, a number of much higher center frequency (higher than 50 MHz or even
beyond 100 MHz) annular array (Liu, Djuth, et al., 2012) and linear arrays (Wu et al.,
2009; Zhou, Q. et al., 2010; Liu, Zhou, et al., 2012) have been reported. These arrays
may allow forming a more tightly focused beam and may be possible to trap small
particles in several-micrometer diameter, which is close to human cells in size. While
currently transmit subsystem can support around 30 MHz center frequency with 150% -6
dB bandwidth (Figure 3.9), high voltage pulses with higher frequency range can be
achieved in the current system by shortening pulse waveforms (Figure 6.1) in the FPGA
91
of transmit subsystem. The new pulses have around 50 MHz center frequency with 120%
-6 dB bandwidth (Figure 6.2).

Figure 6.1 The waveform of the new pulse generated by FPGA, the time duration of the pulse waveform
is two clocks instead of four clocks for 30 MHz arrays.

Figure 6.2 New pulses with 140 MHz system clock. (Left) Spectrum of the pulse generated in channel #1;
(Right) Center frequency and -6 dB frequency range for the new high voltage pulses generated in all 64-
channels

Moreover, the trapping and translation in both elevation and azimuth directions
may be realized with a 2D high frequency ultrasonic array. In this case, more channels
may also be needed in the transmit subsystem to support a 2D array with at least 256 (16
x 16) elements normally.
Clk
Pulse
waveform
0 20 40 60 80 100
-30
-25
-20
-15
-10
-5
0
Frequency (MHz)
Normalized amplitude (dB)
0 10 20 30 40 50 60
0
10
20
30
40
50
60
70
80
90
Channel index
Frequency (MHz)

Center frequency
92
Bibliography
Analog Devices Inc., "12-bit, 170 MSPS/210 MSPS/250 MSPS, 1.8V analog-to-digital
converter," AD9230 Datasheet, Rev. 0, 2007.

Analog Devices Inc., "Ultralow noise VGAs with preamplifier and programmable Rin,"
AD8331/AD8332/AD8334 Datasheet, Rev. G, 2010.

Ashkin, A., "Acceleration and trapping of particles by radiation pressure," Physical
Review Letters, vol. 24, pp. 156-159, 1970.

Ashkin, A., Dziedzic, J. M., Bjorkholm, J. E., and Chu, S., "Observation of a single-beam
gradient force optical trap for dielectric particles," Optics Letters, vol. 11, pp. 288-290,
1986.

Ashkin, A., and Dziedzic, J. M., "Optical trapping and manipulation of viruses and
bacteria," Science, vol. 235, pp. 1517-1520, 1987.

Brown, J. A., Foster, F. S., Needles, A., Cherin, E., and Lockwood, G. R., "Fabrication
and performance of a 40-MHz linear array based on a 1-3 composite with geometric
elevation focusing," IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency
Control, vol. 54, pp. 1888-1894, 2007.

Cannata, J. M., Williams, J. A., Zhou, Q., Ritter, T. A., and Shung, K. K., "Development
of a 35-MHz piezo-composite ultrasound array for medical imaging," IEEE Transactions
on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 53, pp. 224-236, 2006.

Cannata, J. M., Williams, J. A., Zhang, L., Hu, C.-H., and Shung, K. K., "A high-
frequency linear ultrasonic array utilizing an interdigitally bonded 2-2 piezo-composite,"
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 58, pp.
2202-2212, 2011.

Cobbold, R. S. C., "Foundations of biomedical ultrasound," Oxford University Press,
2007.

Coleman, D. J., Silverman, R. H., Chabi, A., Rondeau, M. J., Shung, K. K., Cannata, J.,
and Lincoff, H., "High-resolution ultrasonic imaging of the posterior segment,"
Ophthalmology, vol. 111, pp. 1344-1351, 2004.

Doyle, D., Pilbratt, G., and Tauber, J., "The Herschel and Planck space telescopes,"
Proceedings of the IEEE, vol. 97, pp. 1403-1411, 2009.

93
Ensminger, D., and Stulen, F. B., "ULtrasonics: Data, Equations, and Their Practical
Uses," CRC Press, 2008.

Foster, F. S., Ryan, L. K., and Turnbull, D. H., "Characterization of lead zirconate
titanate ceramics for use in miniature high-frequency (20-80 MHz) transducers," IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 38, pp. 446-453,
1991.

Foster, F. S., Pavlin, C. J., Lockwood, G. R., Ryan, L. K., Harasiewicz, K. A., Berube, L.,
and Rauth, A. M., "Principles and applications of ultrasound backscatter microscopy,"
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 40, pp.
608-617, 1993.

Foster, F. S., Pavlin, C. J., Harasiewicz, K. A., Christopher, D. A., and Turnbull, D. H.,
"Advances in ultrasound biomicroscopy," Ultrasound in Medicine & Biology, vol. 26, pp.
1-27, 2000.

Foster, F. S., Mehi, J., Lukacs, M., Hirson, D., White, C., Chaggares, C., and Needles, A.,
"A new 15–50 MHz array-based micro-ultrasound scanner for preclinical imaging,"
Ultrasound in Medicine & Biology, vol. 35, pp. 1700-1708, 2009.

Hoskins, P. R., Martin, K., and Thrush, A., "Diagnostic ultrasound : Physics and
equipment." Cambridge medicine. 2nd ed: Cambridge University Press, 2010.

Hsu, H.-S., Zheng, F., Li, Y., Lee, C., Zhou, Q., and Shung, K. K., "Focused high
frequency needle transducer for ultrasonic imaging and trapping," Applied Physics
Letters, vol. 101, pp. 024105, 2012.

Hu, C., Xu, X.-C., Cannata, J. M., Yen, J. T., and Shung, K. K., "Development of a real-
time, high-frequency ultrasound digital beamformer for high-frequency linear array
transducers," IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control,
vol. 53, pp. 317-323, 2006.

Hu, C., Zhang, L., Cannata, J. M., Yen, J., and Shung, K. K., "Development of a 64
channel ultrasonic high frequency linear array imaging system," Ultrasonics, vol. 51, pp.
953-959, 2011.

Hurst, J. B., and Dutta, S., "Simple Processing Method for High-Strength Silicon
Carbide," Journal of the American Ceramic Society, vol. 70, pp. 303-308, 1987.

Intersil Americas Inc., "ISL55110, ISL55111," ISL55110 Datasheet, FN6228.4, 2011.

Ishikawa, H., "Anterior segment imaging for glaucoma: OCT or UBM?," The British
Journal of Ophthalmology, vol. 91, pp. 1420-1421, 2007.
94
Jain, R. K., "Determinants of Tumor Blood Flow: A Review," Cancer Research, vol. 48,
pp. 2641-2658, 1988.

Jain, R. K., Schlenger, K., Hockel, M., and Yuan, F., "Quantitative angiogenesis assays:
Progress and problems," Nature Medicine, vol. 3, pp. 1203-1208, 1997.

Jensen, J. A., and Svendsen, N. B., "Calculation of pressure fields from arbitrarily shaped,
apodized, and excited ultrasound transducers," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 39, pp. 262-267, 1992.

Jensen, J. A., "Estimation of blood velocities using ultrasound : a signal processing
approach," Cambridge University Press, 1996a.

Jensen, J. A., "FIELD: A program for simulating ultrasound systems," Medical &
Biological Engineering & Computing, vol. 34, pp. 351-353, 1996b.

Johnson, J. S., Grobsky, K., and Bray, D. J., "Rapid fabrication of lightweight silicon
carbide mirrors," Proceedings of SPIE, vol. 4771, pp. 243-253, 2002.

Kinsler, L. E., Frey, A. R., Coppens, A. B., and Sanders, J. V., "Fundamentals of
acoustics." 4th ed. New York: Wiley, 2000.

Kisslo, J., Vonramm, O. T., and Thurstone, F. L., "Cardiac imaging using a phased array
ultrasound system II. Clinical technique and application," Circulation, vol. 53, pp. 262-
267, 1976.

Kozak, M., and Karaman, M., "Digital phased array beamforming using single-bit delta-
sigma conversion with non-uniform oversampling," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 48, pp. 922-931, 2001.

Lee, J., Ha, K., and Shung, K. K., "A theoretical study of the feasibility of acoustical
tweezers: Ray acoustics approach," The Journal of the Acoustical Society of America,
vol. 117, pp. 3273-3280, 2005.

Lee, J., and Shung, K. K., "Effect of ultrasonic attenuation on the feasibility of acoustic
tweezers," Ultrasound in Medicine & Biology, vol. 32, pp. 1575-1583, 2006.

Lee, J., Teh, S.-Y., Lee, A., Kim, H. H., Lee, C., and Shung, K. K., "Single beam
acoustic trapping," Applied Physics Letters, vol. 95, pp. 073701, 2009.

Lee, J., Lee, C., and Shung, K. K., "Calibration of sound forces in acoustic traps," IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 57, pp. 2305-
2310, 2010.

95
Lee, J., Teh, S.-Y., Lee, A., Kim, H. H., Lee, C., and Shung, K. K., "Transverse acoustic
trapping using a Gaussian focused ultrasound," Ultrasound in Medicine & Biology, vol.
36, pp. 350-355, 2010.

Lee, M. H., Kim, J. H., and Park, S. B., "Analysis of a scan conversion algorithm for a
real-time sector scanner," IEEE Transactions on Medical Imaging, vol. 5, pp. 96-105,
1986.

Lei, S., Richard, W. D., Cannata, J. M., Feng, C. C., Johnson, J. A., Yen, J. T., and Shung,
K. K., "A high-frame rate high-frequency ultrasonic system for cardiac imaging in mice,"
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 54, pp.
1648-1655, 2007.

Li, Y.-F., and Li, P.-C., "Software beamforming: comparison between a phased array and
synthetic transmit aperture," Ultrason Imaging, vol. 33, pp. 109-118, 2011.

Liu, C., Djuth, F., Li, X., Chen, R., Zhou, Q., and Kirk Shung, K., "Micromachined high
frequency PMN-PT/epoxy 1–3 composite ultrasonic annular array," Ultrasonics, vol. 52,
pp. 497-502, 2012.

Liu, C., Zhou, Q., Djuth, F. T., and Shung, K. K., "High-frequency (>50 MHz) medical
ultrasound linear arrays fabricated from micromachined bulk PZT materials," IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 59, pp. 315-318,
2012.

Lockwood, G. R., Hunt, J. W., and Foster, F. S., "The design of protection circuitry for
high-frequency ultrasound imaging systems," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 38, pp. 48-55, 1991.

Macdonald, M. P., Spalding, G. C., and Dholakia, K., "Microfluidic sorting in an optical
lattice," Nature, vol. 426, pp. 421-424, 2003.

Madsen, E. L., Frank, G. R., Mccormick, M. M., Deaner, M. E., and Stiles, T. A.,
"Anechoic sphere phantoms for estimating 3-D resolution of very-high-frequency
ultrasound scanners," IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency
Control, vol. 57, pp. 2284-2292, 2010.

Marrouche, N. F., Martin, D. O., Wazni, O., Gillinov, A. M., Klein, A., Bhargava, M.,
Saad, E., Bash, D., Yamada, H., Jaber, W., Schweikert, R., Tchou, P., Abdul-Karim, A.,
Saliba, W., and Natale, A., "Phased-array intracardiac echocardiography monitoring
during pulmonary vein isolation in patients with atrial fibrillation," Circulation, vol. 107,
pp. 2710-2716, 2003.

96
Martin, R. P., Rakowski, H., French, J., and Popp, R. L., "Localization of pericardial
effusion with wide angle phased array echocardiography," The American Journal of
Cardiology, vol. 42, pp. 904-912, 1978.

Moffitt, J. R., Chemla, Y. R., Smith, S. B., and Bustamante, C., "Recent advances in
optical tweezers," Annual Review of Biochemistry, vol. 77, pp. 205-228, 2008.

Munro, R. G., "Material properties of a sintered alpha-SiC," Journal of Physical and
Chemical Reference Data, vol. 26, pp. 1195-1203, 1997.

Neuman, K. C., Chadd, E. H., Liou, G. F., Bergman, K., and Block, S. M.,
"Characterization of photodamage to Escherichia coli in optical traps," Biophysical
Journal, vol. 77, pp. 2856-2863, 1999.

Packer, D. L., Stevens, C. L., Curley, M. G., Bruce, C. J., Miller, F. A., Khandheria, B.
K., Oh, J. K., Sinak, L. J., and Seward, J. B., "Intracardiac phased-array imaging:
methods and initial clinical experience with high resolution, under blood visualization:
Initial experience with intracardiac phased-array ultrasound," Journal of the American
College of Cardiology, vol. 39, pp. 509-516, 2002.

Park, S. B., and Lee, M. H., "A new scan conversion algorithm for real time sector
scanner," IEEE Ultrasonics Symposium, pp. 723-727, 1984.

Portune, A. R., and Haber, R. A., "Quantitative ultrasound characterization of silicon
carbide mirrors," AIP Conference Proceedings, vol. 1211, pp. 1257-1264, 2010.

Ren, J.-F., Marchlinski, F. E., and Callans, D. J., "Left atrial thrombus associated with
ablation for atrial fibrillation: Identification with intracardiac echocardiography," Journal
of the American College of Cardiology, vol. 43, pp. 1861-1867, 2004.

Robinson, D. E., and Knight, P. C., "Interpolation scan conversion in pulse-echo
ultrasound," Ultrasonic Imaging, vol. 4, pp. 297-310, 1982.

Seo, C. H., and Yen, J. T., "Sidelobe suppression in ultrasound imaging using dual
apodization with cross-correlation," IEEE Transactions on Ultrasonics, Ferroelectrics,
and Frequency Control, vol. 55, pp. 2198-2210, 2008.

Shung, K. K., "Diagnostic ultrasound: Imaging and blood flow measurements," Taylor &
Francis, 2006.

Shung, K. K., Cannata, J. M., and Zhou, Q. F., "Piezoelectric materials for high
frequency medical imaging applications: A review," Journal of Electroceramics, vol. 19,
pp. 141-147, 2007.

97
Sikdar, S., Managuli, R., Mitake, T., Hayashi, T., and Kim, Y., "Programmable
ultrasound scan conversion on a mediaprocessor-based system," Proceedings of SPIE, vol.
4319, pp. 699-711, 2001.

Silverman, R. H., Kruse, D. E., Coleman, D. J., and Ferrara, K. W., "High-resolution
ultrasonic imaging of blood flow in the anterior segment of the eye," Investigative
Ophthalmology & Visual Science, vol. 40, pp. 1373-1381, 1999.

Silverman, R. H., Ketterling, J. A., and Coleman, D. J., "High-frequency ultrasonic
imaging of the anterior segment using an annular array transducer," Ophthalmology, vol.
114, pp. 816-822, 2007.

Steelman, W. A., and Richard, W. D., "Performance of new GPU-based scan-conversion
algorithm implemented using OpenGL," Ultrason Imaging, vol. 33, pp. 143-150, 2011.

Steinberg, B. D., "Digital beamforming in ultrasound," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 39, pp. 716-721, 1992.

Sun, L., Lien, C.-L., Xu, X., and Shung, K. K., "In vivo cardiac imaging of adult
zebrafish using high frequency ultrasound (45-75 MHz)," Ultrasound in Medicine &
Biology, vol. 34, pp. 31-39, 2008.

Supertex Inc., "TC6320: N- and P-Channel enhancement-mode MOSFET pair," TC6320
Datasheet, 2008.

Svoboda, K., and Block, S. M., "Biological applications of optical forces," Annual
Review of Biophysics and Biomolecular Structure, vol. 23, pp. 247-285, 1994.

Thomenius, K. E., "Evolution of ultrasound beamformers," Proceedings of IEEE
Ultrasonics Symposium, vol. 2, pp. 1615-1622, 1996.

Tsonis, P. A., "Animal models in eye research," Academic Press, 2008.

Turnbull, D. H., Starkoski, B. G., Harasiewicz, K. A., Semple, J. L., From, L., Gupta, A.
K., Sauder, D. N., and Foster, F. S., "A 40–100 MHz B-scan ultrasound backscatter
microscope for skin imaging," Ultrasound in Medicine & Biology, vol. 21, pp. 79-88,
1995.

Vogt, M., and Ermert, H., "In vivo ultrasound biomicroscopy of skin: Spectral system
characteristics and inverse filtering optimization," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 54, pp. 1551-1559, 2007.

Vonramm, O. T., and Thurstone, F. L., "Cardiac imaging using a phased array ultrasound
system I. System design," Circulation, vol. 53, pp. 258-262, 1976.
98
Wang, M. D., Yin, H., Landick, R., Gelles, J., and Block, S. M., "Stretching DNA with
optical tweezers," Biophysical Journal, vol. 72, pp. 1335-1346, 1997.

Wu, D.-W., Zhou, Q., Geng, X., Liu, C.-G., Djuth, F., and Shung, K. K., "Very high
frequency (beyond 100 MHz) PZT kerfless linear arrays," IEEE Transactions on
Ultrasonics, Ferroelectrics, and Frequency Control, vol. 56, pp. 2304-2310, 2009.

Xilinx Inc., "Spartan-3E FPGA Family: Data Sheet," Spartan-3E Datasheet, DS312
(V3.8), 2009a.

Xilinx Inc., "Virtex-5 Family Overview," Virtex-5 Datasheet, DS100 (V5.0), 2009b.

Xu, X., Yen, J. T., and Shung, K. K., "A low-cost bipolar pulse generator for high-
frequency ultrasound applications," IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control, vol. 54, pp. 443-447, 2007.

Zhang, L., Xu, X., Hu, C., Sun, L., Yen, J. T., Cannata, J. M., and Shung, K. K., "A high-
frequency, high frame rate duplex ultrasound linear array imaging system for small
animal imaging," IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency
Control, vol. 57, pp. 1548-1557, 2010.

Zhou, Q., Wu, D., Liu, C., Zhu, B., Djuth, F., and Shung, K. K., "Micro-machined high-
frequency (80 MHz) PZT thick film linear arrays," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 57, pp. 2213-2220, 2010.

Zhou, Y.-Q., Foster, F. S., Nieman, B. J., Davidson, L., Chen, X. J., and Henkelman, R.
M., "Comprehensive transthoracic cardiac imaging in mice using ultrasound
biomicroscopy with anatomical confirmation by magnetic resonance imaging," Physiol
Genomics, vol. 18, pp. 232-244, 2004.

Zipparo, M. J., Shung, K. K., and Shrout, T. R., "Piezoceramics for high-frequency (20 to
100 MHz) single-element imaging transducers," IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, vol. 44, pp. 1038-1048, 1997.

Abstract (if available)

Abstract While similar to a linear-switched array in structure, an ultrasound linear phased array (simply called phased array) is quite different in operation. It is capable of beam steering to form fan-shaped sector images without inducing grating lobes. Moreover, the relative smaller array footprint is useful where only very limited contact surface is permitted. High frequency phased arrays combine the advantages of the phased array and high frequency ultrasound, which offer both small footprint and high spatial resolution. This dissertation reports the design and development of a digital ultrasonic imaging platform with raw RF data acquisition capability, which can be paired with a prototype 64-element 26 MHz phased array transducer. A wire phantom image showed that -6 dB lateral and axial resolutions were 209 and 104 μm, respectively, which were in good agreement with the Field II simulation. Anechoic cyst tissue-mimicking phantom images demonstrated its capability to detect cysts of 300 μm in diameter. An image of a rabbit eyeball in vitro was also acquired. This imaging platform is designed for the purpose of testing high frequency phased arrays under development as well as facilitating the development in novel array signal processing algorithms. ❧ Without additional change in the system hardware, the current high frequency array system can be also applied in the industrial non-destructive testing (NDT) applications. A motorized XY stage allows the array system to acquire C-mode scan as well as normal B-mode scan with bigger scanning area. Electronic scanning in array system could dramatically reduce the total scanning time compare to single element system. An aluminum sample and a silicon carbide sample were examined to test the system in non-destructive testing. ❧ Besides the applications in biomedical imaging and non-destructive testing, the high frequency ultrasonic phased array and system are shown to be capable of trapping and translating microparticles precisely and efficiently, made possible due to the fact that the acoustic beam produced by a phased array can be both focused and steered. Acoustic manipulation of microparticles by a phased array is advantageous over a single element transducer since there is no mechanical movement required for the array. Experimental results show that 45 μm diameter polystyrene microspheres can be easily and accurately trapped and moved to desired positions by a 64-element 26 MHz phased array.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

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

PDF

Array transducers for high frequency ultrasound imaging

PDF

High-frequency ultrasonic transducers for photoacoustic applications

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

PDF

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

PDF

Development of high frequency focused transducers for single beam acoustic tweezers

PDF

Intravascular imaging on high-frequency ultrasound combined with optical modalities

PDF

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

PDF

Development of front-end circuits for high frequency ultrasound system

PDF

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

PDF

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

PDF

High frequency ultrasound array for ultrasound-guided breast biopsy

PDF

Single-cell analysis with high frequency ultrasound

PDF

Multi-modality intravascular imaging by combined use of ultrasonic and opticial techniques

PDF

High frequency ultrasound elastography and its biomedical applications

PDF

Design and development of ultrasonic array transducers for specialized applications

PDF

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

Asset Metadata

Creator Zheng, Fan (author)

Core Title High frequency ultrasonic phased array system and its applications

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 11/12/2012

Defense Date 09/14/2012

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

Tag high frequency,imaging system,OAI-PMH Harvest,phased array,ultrasound

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Shung, Kirk Koping (committee chair), Kim, Eun Sok (committee member), Yen, Jesse T. (committee member), Zhou, Qifa (committee member)

Creator Email fanzheng@gmail.com,fzheng@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-109547

Unique identifier UC11290195

Identifier usctheses-c3-109547 (legacy record id)

Legacy Identifier etd-ZhengFan-1284.pdf

Dmrecord 109547

Document Type Dissertation

Rights Zheng, Fan

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA