SlideShare a Scribd company logo
1 of 60
Download to read offline
Modeling Speckles for Ultrasound Imaging 
by 
Huma Yusuf    
 
 
Submitted to the Department of Physics in partial fulfillment of 
the requirements for the degree of  
Bachelor of Arts in Physics 
at 
Mount Holyoke College 
May 2015 
 
© MOUNT HOLYOKE COLLEGE 2015. All rights reserved. 
 
Certified by…………………………………………………….. 
                                                                          Maria Teresa Herd 
                                            Laboratory Director, Department of Physics 
                                                                                     Thesis Supervisor   
Accepted by…………………………………………………… 
                                                                               Mark Peterson 
                 Co­Chair of Physics, Professor of Mathematics and Physics 
   
       
  
 
 
1 
Abstract 
Ultrasound has been used in a variety of clinical settings, including obstetrics and                         
gynecology, cardiology and cancer detection. The main advantage of ultrasound is                     
that, unlike x­ray imaging, it does not require ionizing radiation, which may                       
increase the risk of getting cancer. In addition, using ultrasound imaging as a                         
medical diagnostic tool is more cost effective compared to other radiation free                       
diagnostic techniques, such as the MRI. However, a difficult and crucial challenge                       
in ultrasound medical imaging is the necessity of reducing the appearance of                       
speckles in B­scan images. Speckles are caused by coherent interference of                     
reflected ultrasound waves by structures smaller than the Rayleigh scattering limit.                     
Appearance of speckles tend to reduce the perception of small structures, and                       
ultimately limit diagnostic accuracy of medical imaging system. Since it is difficult                       
to model speckles using architectural structure of tissues, modeling techniques                   
involve a more constructive approach: predicting the speckle pattern of an arbitrary                       
scattering medium for a given transducer geometry; such a model was developed                       
by Foster et al. (1983).  
 
Investigation of the origin and nature of speckles involve creating a mathematical                       
model to predict the RF output signal waveform of a particular transducer                       
geometry, and then extending that model to incorporate the effects of varying sizes                         
of scatterers on the RF voltage trace. The complete model will then be verified by                             
comparing theoretical results with experimental results obtained from pulse­echo                 
experiments. For this project, we have focused more on the experimental aspect of                         
the investigation by producing B­mode images of scatterer phantoms. These                   
phantoms are standard samples made using the discrete scatterer model, where                     
scatterers of uniform shape and size are embedded in a homogeneous medium.                       
Before imaging scatterers smaller than the Rayleigh scattering limit, it is important                       
to ensure that scatterers much larger than the Rayleigh limit (discrete scatterers)                       
can be successfully imaged. We conducted pulse­echo experiments using a 5 MHz                       
single­element focused transducer, and imaged phantoms containing discrete               
scatterers of the following sizes: 4.76 mm metal beads, 2.85 mm ­ 3.45mm glass                           
beads and 1.00 mm ­ 1.03 mm glass beads. The RF voltage traces from the                             
pulse­echo experiments were then converted to B­mode images using a Matlab                     
program.   
 
 
 
 
2 
Acknowledgements
First of all, I would like to thank my advisor Dr. Maria Teresa Herd 
for giving me the opportunity to do research in her lab. Without her continuous 
guidance and support, this thesis would not have been possible. I would also like to 
thank her for encouraging and helping me at times when I lost faith in my abilities 
over the past year. She has truly been a great role model and mentor.  
I would also like to thank the Physics Department for fostering my love for                           
physics and passion for research. I would like to thank all the professors: Mark                           
Peterson, Juan Burciaga, Neal Abraham, Spencer Smith, Alexi Arango and Kathy                     
Aidala for their enthusiasm and inspirational way of teaching.  
I am grateful to my lab mates Madeleina, Audrey and Colbie. Thank you,                         
Audrey and Madeleina for allowing me to observe your experiments, and helping                       
me learn the fundamentals of handling lab equipments, acquiring and analyzing                     
data. Thank you Colbie for your assistance during the first half of my project,                           
when we spent several hours at a stretch pouring over books looking for missing                           
functions for the RF voltage trace model. 
Next, I would like to thank my brother for always pushing me to become a                             
better physicist than him. I would also like to thank my boyfriend for his constant                             
encouragement, and for always being available to listen to me whine, complain                       
and cry over the phone whether he is at home or at an observing run at Mauna Kea                                   
or Coonabarbaran.  
Finally, I have to thank my mom for giving me full freedom to explore my                             
passion and pursue my dreams, no matter how crazy they are. Without her                         
unconditional support, guidance, and love, I would not have come this far.  
   
           Once again, a big thank you to each and everyone! 
 
3 
       Contents  
1 Introduction to Pulse­Echo Ultrasound Imaging                       7 
                         1.1 Making Ultrasound B­mode Images …………….. 7 
                         1.2 Speckles in Ultrasound Images ….………………. 9   
                         1.3 Importance of Modeling Speckle ………………... 10   
                         1.4 Current Theoretical Models for Speckles ………... 11 
                                       1.4.1 Discrete Scatterer Model ……………... 12 
                                       1.4.2 Inhomogeneous Continuum 
                                                Model …………………………………. 12 
                         1.5 Modeling Backscattered RF Trace  
                               of Human Tissue ………………………………… 13 
                         1.6 Complex Plane Representation of 
                                Ultrasound Point Spread Function ………………. 14 
2 Theory of Wave Propagation: Physics of Ultrasound                      15 
                         2.1 Classical Theory of Linear Acoustics …………….. 15 
                                       2.1.1 Equation of State ………………………. 16 
                                       2.1.2 Equation of Continuity ………………… 18 
                                       2.1.3 Euler’s Equation ……………………….. 18 
                         2.2 The Wave Equation ……………………………….. 20 
                         2.3 Wave Equation in the Frequency 
                                Domain ……………………………………………. 21 
                                      2.3.1 The Incident Pressure Field ……………... 23  
                                       2.3.2 The Scattered Pressure Field …………… 25 
                                       2.3.3 The Force on the Receiving Crystal ……. 26 
                                       2.3.4 The RF Voltage Trace ………………….. 28 
4 
3 Pulse­Echo Experiments Using a Single Element Transducer       31 
                         3.1 Making Scatterer Phantoms ……………………… 32 
                         3.2 Experimental Setup ………………………………. 34 
                         3.3 Electronic Setup ………………………………….. 35 
                         3.4 Aligning the Transducer …………………………. 36  
                          3.5 Collecting RF Pulse­Echo Signal  
                                 Data ……………………………………………... 39 
  4 Obtaining B­mode Images                                                                 42 
                         4.1 B­mode Image of 4.76 mm  
                               Metal Beads ………………………………………. 42 
                         4.2 B­mode Image of 2.85 mm ­ 3.45 mm 
                              Glass Beads …………………………………………44 
                         4.3 B­mode Image of 1.0 mm ­ 1.03 mm 
                              Glass Beads ……………………………………….. 45 
 
  5 Conclusion and Future Work                                                            47 
 
     Appendix                                                                                             49 
            A Proof of Lemma 1 (Chapter 2, 2.3.3)........................................ 49 
           B Proof of Lemma 2 (Chapter 2, 2.3.3)........................................ 50 
            C MATLAB Program for B­mode Images.................................... 53   
           D Preliminary Modeling: Incident Pressure Field......................... 56 
           E Preliminary Modeling: Scattered Pressure Field........................ 59 
   
5 
 
 
 
 
 
 
 
 
 
 
 
 
6 
Chapter 1 
Introduction to Pulse­Echo Ultrasound 
Imaging 
          1.1 Making Ultrasound B­mode Images  
Ultrasound has been used in a variety of clinical settings, including obstetrics and                         
gynecology, cardiology and cancer detection. The main advantage of ultrasound is                     
that structures can be observed without using radiation. Using ultrasound medical                     
imaging as a diagnostic tool is also more cost effective compared to other radiation                           
free diagnostic techniques, such as the MRI. Ultrasound can also be done much                         
faster than X­ray and other radiographic techniques.Ultrasound also provides                 
multiple modes available for versatile imaging. For instance, M­mode ultrasound                   
is extremely valuable for accurate evaluation of rapid movements due to its                       
excellent temporal resolution, and is used  for tracking the motion of heart                       
structures over time.  
For the purposes of modeling speckles, we are interested in producing                     
B­mode images. This is because B­mode images can be easily produced using the                         
amplitude of reflected ultrasound signals to reflect the intensity or brightness on a                         
two­dimensional gray scale. Owing to a wide gray scale available for ultrasound                   
imaging, very small differences in echogenicity are possible to visualize. 
The essential component of an ultrasound medical imaging system is a                     
transducer. Transducers are devices that convert electrical signals to mechanical                   
energy and vice versa. When electrical signals reach a transducer in the form of an                             
excitation pulse, these signals are converted to sound waves that travel to the                         
structure being imaged. All of the sound waves are then reflected by the structure,                           
some of which eventually find their way back to the transducer and the sound                           
waves are then converted to electrical signals. These electrical signals are collected                       
7 
in the form of radio frequency (RF) voltage traces that are processed to produce                           
B­mode images of the structure.  
The basic components of a typical spherically focused single­element                 
transducer are summarized in Figure 1. The active piezoelectric element has a                       
spherically curved shape to facilitate the focusing of the beam, and its thickness is                           
usually half the wavelength of the ultrasound produced. It is important to note that                           
the transducer generates a range of frequencies around a central frequency. This is                         
in part due to the fact that the piezoelectric element continues to produce                         
ultrasound for sometime even after the electric signal has been terminated.  
There are a number of parameters, which are useful for understanding the                       
general components of a single­element transducer. The range of frequencies                   
generated by the transducer is known as the bandwidth. The quality factor of a                           
transducer is a dimensionless parameter, which is calculated as the ratio of the                         
central frequency to the bandwidth; the higher the quality factor, the lower the rate                           
of energy loss compared to stored energy. Finally, the sensitivity of the transducer                         
is defined as the ability to detect reflected ultrasound and generate electrical signal.                         
The sensitivity and bandwidth of the transducer are dictated by the parameters of                         
the piezoelectric material, transducer backing and electrical and acoustic                 
impedance matching devices, which are needed to achieve the specific imaging                     
requirements.  
8 
 
                      Figure 1 : Single Element Ultrasonic Transducer   
 
There is a fundamental relationship between the properties of the generated                     
ultrasound beam and the quality of the resulting image. The ability to predict beam                           
characteristics is crucial for medical imaging. There are many factors, which affect                       
the beam profiles of transmitted and received signals. Most important are the                       
source geometry, source excitation (sinusoidal or pulsed), target response and the                     
type of electronic processing applied to signals. The theoretical treatment of beam                       
profiles is based on Huygen’s Principle, where each radiating source is divided                       
into infinitesimal elements, each radiating hemispherical wavelets. The resulting                 
acoustic fields obtained from the superposition of these wavelets obey the classical                       
theory of linear acoustics described in Chapter 2. 
A central aspect of quantitatively characterizing the beam profile is by                     
calculating the instantaneous pressure field distribution. The function for                 
calculating the instantaneous pressure field is obtained using the wave equation for                       
the velocity potential (discussed in detail in Chapter 2). Determining the                     
instantaneous pressure for incident and scattering field paves the  way for the                       
calculation of pulse­echo sensitivity profiles based on the spatio­temporal impulse                   
response of the transducer (discussed in detail in Chapter 2). 
 
          1.2 Speckles in Ultrasound Images 
               A difficult and crucial challenge in ultrasound medical imaging is the necessity of  
reducing the annoying “mottled” or “speckled” appearance of B­scan images.                   
Speckles are an interference phenomena caused by the interaction of coherent                     
waves produced by the transducer with multiple scattering structure of the tissue.                       
Such interactions result in the constructive and destructive coherent summation of                     
9 
ultrasound echoes. Speckles tend to reduce the perception of small structures, and                       
ultimately limit diagnostic accuracy of medical imaging systems.  
Generally there is no common enhancement approach for speckle noise                   
reduction. Different filtering techniques based on statistical methods have been                   
implemented in an attempt to reduce speckles, such as Signal­to­Noise Ratio                     
(SNR), Peak Signal­to­Noise Ratio (PSNR) and Root Mean Square Error (RMSE).                     
Another important averaging technique used for speckle noise reduction is                   
angular­weighted compounding, which stems from ultrasound elastography.             
Elastography is an imaging technique in which local strain in human tissues are                         
used to measure the axial shifts in tissue due to quasi static compression applied                           
using ultrasonic transducers; all the measurements are made in terms of RF echo                         
arrival times. Using elastography measurements, angular­weighted factors are               
derived from the relationship between axial and lateral strain components of strain                       
estimated along angular insonification directions. Experimental results using a                 
uniformly elastic tissue mimicking phantoms demonstrated the improvement in                 
signal­to­noise ratio obtained using angular­weighted compounding (Techavipoo             
et. al. 2004). However, removing speckle noise from original ultrasound images                     
still remains a challenge in image processing.  
 
1.3 Importance of Modeling Speckles 
Ultrasound imaging is a useful diagnostic tool, especially for biological tissues and                       
subcellular structures. However, ultrasound medical imaging is limited by two                   
important factors: spatial resolution and fluctuations in echo amplitude, known as                     
speckle noise. Speckles are considered one of the major obstacles in image                       
analysis, which include detection, classification and segmentation problems in                 
coherent imaging systems.  
Since speckle appears in all conventional ultrasound images, numerous                 
investigations have been carried out to understand the nature of speckles and what                         
10 
causes them. Establishing a model for ultrasound imaging would require                   
developing a mathematical description of the acoustical scattering properties of                   
tissues; however, this is often hindered due to the lack of knowledge concerning                         
cellular structures. Kossoff et al. (1976), attempted to relate speckles to the                       
architectural structure of tissues, while Linzer et al. (1979) had tried to relate                         
speckle patterns to the different diseased states of soft tissue. But in both studies it                             
was found that the speckle pattern bore little resemblance to the actual acoustical                         
tissue microstructure. Rather than directly using speckle patterns to decipher                   
details of complex cellular structures, a more constructive approach would be to                       
first predict the speckle pattern of an arbitrary scattering medium for a given                         
transducer geometry. Therefore developing theoretical models to describe the                 
scattering medium will help us understand the relationship between the imaging                     
system and the speckles that arise in the final image.   
Another reason for modelling speckles is that they are the primary                     
mechanism for elastography imaging, an emerging and exciting use of ultrasound                     
imaging technique.  In elastography, it is possible to trace localized elastic                     
characteristics because local movement is estimated using speckle tracking (Kim et                     
al. 2011). The application of speckles in elastography confirms that speckles are                       
not always undesirable. Today, ultrasound elastography has developed into an                   
image modality suitable for detection and diagnosis of cancers in the breast,                       
kidneys and thyroid. 
 
1.4 Current Theoretical Models for Speckles 
Theoretical models used to describe the scattering medium can be essentially                     
divided into two categories: the discrete scatterer model and the inhomogeneous                     
continuum model. For the discrete scatterer model, the scattering medium is                     
depicted as a collection of points, spheres or cylinders embedded in a                       
homogeneous medium. On the other hand, the inhomogeneous scattering model                   
11 
describes the scattering medium to be one where acoustical inhomogeneities, such                     
as density, compressibility and velocity, vary in a continuous manner throughout                     
the medium. Such models of scattering medium are then used in conjunction with                         
simulation models to investigate both theoretically and experimentally the effect of                     
speckles on ultrasound images.  
 
1.4.1 Discrete Scatterer Model 
Foster et al. (1983), in their study to investigate the effect of transducer geometry                           
and the position of the point spread function on speckle formation, developed a                         
mathematical model to simulate ultrasound B­scan images showing speckles. Prior                   
to the experimental approach adopted by Foster et al., simulation models were                       
based on two dimensional scattering medium. The dimension perpendicular to the                     
image plane was ignored for the two dimensional models; thus the number of                         
scatterers in a given distance off­axis remained constant in those models. Previous                       
models also excluded transducer geometry and the position dependence of the                     
point spread function. Foster et al. were able to recognize the limitations of the                           
two­dimensional models, and thus incorporated the three­dimensional             
characteristics of both the transducer and the scattering medium. The scatterer                     
phantoms were developed using the discrete scatterer model, and their studies                     
showed good agreement between theoretical and experimental results.  
 
1.4.2 Inhomogeneous Continuum Model 
In recent years, other approaches were implemented for modeling speckles. One                     
such approach relies on the surface roughness and the number of elemental                       
scatterers in the tissue surface. In their study to derive stochastic models for                         
speckle noise, Daba et al. (2009) outlined two distinct categories for describing                       
speckle distribution in conjunction with the roughness of scattering surface: fully                     
12 
developed, where the surface is comprised of infinite number of scatterers making                       
the surface very rough, and partially developed, where the surface has finite                       
number of scatterers. Daba et al. derived the statistical distribution for the partially                         
developed model using Poisson point process, where each scatterer is assumed to                       
have statistically independent random variables, which for the purposes of the                     
theory were amplitude and phase of the backscattered field. 
In order to obtain ultrasound images of a particular organ or tissue structure,                         
a pulse is emitted into the body and is scattered and reflected by density and                             
propagation velocity perturbations. Therefore developing a model for the received                   
pulse­echo pressure field of a transducer can also help to predict the speckle                         
pattern from a scattering medium. Using this approach, Jensen (1990) derived an                       
inhomogeneous wave equation to describe the propagation and scattering of                   
ultrasound by a collection of point scatterers in a homogeneous medium. The                       
solution to the wave equation was then combined with the field of a typical                           
transducer to develop a model for the pulse­echo pressure field. This approach                       
proved to be versatile because it enabled the derivation of scattered field for                         
different transducer geometries, whose incident pressure fields are known. 
1.5 Modeling Backscattered RF Trace of Human 
Tissue  
 
Instead of modeling arbitrary scattering medium, like those described in Sections                     
1.4, Gore and Leeman (1976) developed a realistic approximation of human tissue                       
inhomogeneities. Their approach involved describing wave propagation in               
inhomogeneous media, and then showing that the waves are scattered by                     
fluctuations in density and compressibility. The model is then used to calculate the                         
precise form of the backscattered field from tissues, which coincides with actual                       
signals recorded by diagnostic pulse­echo equipment.   
From their studies it was found that the usefulness of characterizing human                       
tissues is dependent on the nature of the structure of the tissue itself. They                           
13 
concluded that although two different tissues may have different compressibility                   
coefficient and density, the backscattered echo sequences are rotationally invariant                   
to first order. For less restricted systems, without an isotropic distribution of                       
scatterers, the rotational invariance does not hold; this indicated that there is an                         
angular dependence of the scattered field—this further confirms why                 
angular­weighted compounding is effective for speckle reduction. Therefore, the                 
studies concluded that in order to understand the diagnostic capability of grey scale                         
B­scan images, a knowledge of the scattering processes within the tissues and                       
quantitative image analysis in conjunction with pattern recognition is important.  
  
1.6 Complex Plane Representation of Ultrasound           
Point Spread Function 
 
Conventional ultrasound imaging probes a medium with high frequency                 
band­limited acoustic waves, and detects echoes scattered by inhomogeneities                 
within the medium. A spherical single element transducer placed in the medium is                         
used for both generation of pulses and reception of these echoes. The site of the                             
transducer responsible for both transmission and reception is known as the active                       
aperture, which is a set of piezoelectric crystals. The scattered echoes are summed                         
up coherently to yield a single RF voltage trace. 
In their study, Ng et al. (2005) solve the equation that governs wave                         
propagation in an inhomogeneous medium to show that the RF ultrasound signal                       
can be expressed as the result of filtering the tissue reflectivity by a point spread                             
function and the total pressure field. Ng et al. (2005) further extended their                         
analysis by establishing a link between the RF signal and the representation of                         
point scatterers as vectors with random phase in the complex plane. Their analysis                         
led to an insight into the useful techniques for simulation and analysis of speckle                           
statistics in the complex plane, in addition to developing a normalized covariance                       
of the RF signal in terms of the complex envelope of the point spread function. 
14 
The mathematical techniques implemented by Ng et. al. (2005) is relevant to                       
the modeling speckles because we in the future we aim to carry out impulse                           
response and instantaneous pressure field calculations in the frequency domain to                     
obtain the RF output voltage response of the transducer. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15 
Chapter 2 
Theory of Wave Propagation:  
Physics of Ultrasound  
 
Acoustic refers to the generation, transmission and reception of energy in the form                         
of vibrational waves in matter. Acoustic waves are the organized vibrations of the                         
molecules or atoms of the medium that supports the propagation of these waves.                         
Acoustic plane waves possess a particular frequency depending on the source of                       
the waves, and waves whose frequency is beyond the audible range of 20 KHz are                             
known as ultrasound waves.   
 
2.1 Classical Theory of Linear Acoustics 
In general plane acoustic waves through an acoustic medium is a combination of                         
transverse and longitudinal motion, although longitudinal motion seems to                 
dominate. The characteristic properties of plane waves is that each acoustic                     
variable, such as particle displacement, density and pressure, has constant                   
amplitude in any given plane perpendicular to the direction of propagation. An                       
expression describing the motion of a particle in a medium in the presence of                           
acoustic plane waves is given by:  
  
where is the equilibrium displacement of the particle, is the particle                         
displacement from the equilibrium position, is the angular frequency and φ is                         
the arbitrary phase factor.  
The above equation is one of the many solution of the wave equation. In the                             
following sections we will derive the wave equation to relate the changes in                         
16 
acoustic variables to one another. To derive a wave equation, we utilize the                         
following: 
● Equation of State, which relates pressure to density. 
● Equation of Continuity, which incorporates conservation of mass. 
● Euler’s Equation, which is derived from Newton’s Second Law. 
 
2.1.1 Equation of State: 
The relationships between pressure and density changes in a gas can also be                         
applied to sound traveling through gases, such as air. In acoustics the equation of                           
state for an ideal gas is given by:  
   
 
where γ is a gas­dependent constant, and T k is the temperature in Kelvin. However,                           
for most fluids and tissues, the adiabatic equation of state is either unknown or                           
highly complex. Thus, this relationship is determined experimentally, and verified                   
mathematically using the following Taylor Series expansion: 
   
where the partial derivatives are constants determined for adiabatic compression                   
and expansion of the medium about its equilibrium density,                  ,and  . 
 
2.1.2 Equation of Continuity 
In order to relate the motion of the fluid to its compression or dilation, we require a                                 
functional relation between the particle velocity and the instantaneous density. As                     
the sound waves propagate, mass has to be conserved. Thus, the extra mass that is                             
carried into a volume due to the propagating wave represents an increase in                         
17 
density. We will consider the rate of increase of mass in a fixed (imaginary)                           
volume V, with the net amount of mass that flows into , per unit time. In other                                 
words, if mass goes into then, because mass is conserved, the amount of mass                           
inside must be going up. The total mass in is given by the volume                               
integral  . The rate of change of mass in   can therefore be written as: 
 
The surface bounding is . The net flow of mass out through a small part of the                                 
surface at point is the dot product where is the outward unit                           
perpendicular to   . The net mass flowing in to  is therefore: 
 
To understand better where the come from, consider Figure 2. When the                         
angle between the beam and the normal to the aperture (call it ) is zero, all the                                 
beam gets through, but as α increases, less and less gets through until, when they                             
are at right­angles none gets through. In general, the amount getting through is  
.  
                                                     Figure 2:  Vector dot product     
 
18 
The divergence theorem, sometimes called Gauss’ divergence theorem, equates the                   
outward flux of a vector field over a closed surface with the volume integral of the                               
divergence of the vector field over the region enclosed by the surface:  
 
where is a vector field. This can be used to convert Equation (6) to the                               
following form: 
 
Equating Equations (7) and (4), we get: 
 
As this equation is zero for any arbitrary volume , the integrand must also be                             
zero hence:  
 
 
2.1.3 Euler’s Equation: 
Let us consider a medium of density and a particle with an incremental volume,                             
whose size is smaller than the wavelength of the ultrasound. The length of this                           
small volume in the longitudinal direction is , where is the area of the cross                               
section perpendicular to the longitudinal axis and the mass of the incremental                       
volume is given by        . Along the longitudinal direction, there exists an               
excess pressure denoted by        . In conjunction with the pressure field, and               
coupled to its longitudinal velocity field            represents the motion of the particle. 
19 
If the particle moves with the velocity              in the presence of an acoustic field, a               
minute force acts on the particle due to the change in pressure. Newton’s force                           
equation can be applied to the volume element, where the net force on the volume                             
can be written in partial differential form: 
 
where  . Notice the extra term in the force equation; this term takes                         
into account the fact that acceleration is not constant within the fluid for the                           
infinitesimal displacement traveled in a time . Thus acceleration is a                       
function of both time and displacement. 
 
Now  , Equation (11) becomes: 
 
Since pressure is force per unit area and                : 
 
Taking the limit as        , Equation (12): 
 
Rearranging Equation (13), we get Euler’s Equation in one dimension: 
 
 
Following from Equation (14), Euler’s Equation in three dimensions is: 
20 
 
2.2 The Wave Equation 
 
The wave equation in an ideal fluid can be derived from hydrodynamics and the                           
adiabatic relation between pressure and density. The equation for conservation of                     
mass, Euler’s equation (Newton’s second law), and the adiabatic equation of state                       
are respectively: 
  
 
and for convenience, we will define the quantity: 
 
where  c  will turn out to be the speed of sound in an ideal fluid. In the above                                   
equations, is the density, the particle velocity, the pressure. The ambient                           
quantities of the quiescent (time independent) medium are identified by the subscript                       
0. We use small perturbations for the pressure and density, and note that is also a                                 
small quantity; that is, the particle velocity which results from density and pressure                         
perturbations is much smaller than the speed of sound.  
Retaining higher­order terms in Equation ( 3 ) yields a nonlinear wave equation. The                       
nonlinear effects we include are contained in the quadratic density term in the                         
equation of state, ( 3 ). We first multiply Equation ( 15 ) by and take its divergence;                             
21 
next, we take the partial derivative of ( 9) with respect to time. Substituting one into                             
the other yields  : 
 
 
Here, the indices  i, j = 1, 2, 3   indicate  x, y  and  z ­components, respectively. Tensor 
notation is used; repeated indices signify a summation (e.g.,  ). 
The first term on the right­hand side of ( 16)  can be rewritten using ( 3)  and (15) as:  
 
2.3 Wave Equation in the Frequency Domain 
 
Conventional ultrasound imaging interrogates a medium with high frequency, band                   
limited waves and detects echoes scattered by inhomogeneities within the medium,                     
where scatterers represent such inhomogeneities. A single­element transducer               
immersed in the medium both generates ultrasound waves and receives the echoes.                       
The single element transducer consists of two types of piezoelectric crystals, the                       
transmitting and the receiving crystals. The transmitting crystal is excited coherently                     
to produce a focused beam, whereas the receiving crystal detects the scattered echoes,                         
which are then summed up coherently to yield an RF voltage trace. At each                           
transmission, the emitted wave propagating through the medium gives rise to an                       
incident pressure field, and the scattered wave gives rise to a scattered pressure field.                           
Multiple RF traces are formed by moving the centre of the crystals and repeating the                             
process; by lining up these individual traces next to each other in an image space, an                               
RF image is formed by summing the RF traces resulting from the scattered pressure                           
field over the piezoelectric surface and filtering the sum by the electromechanical                       
impulse response of the crystals. 
22 
With the physical description of the transducer mechanism in mind, we will consider                         
the mathematical expressions for the incident and scattered pressure fields in the                       
following sections. The wave equation (17) that we derived in the time domain can                           
also be expressed in the frequency domain as follows: 
 
where   
 
 
  and are density and compressibility terms respectively. The presence of                     
scatterers in the medium may be modelled by adding spatially dependent variables                       
and to the density and compressibility terms. The term is the scattering                           
operator defined as follows:   
 
and the scattering terms  and  are: 
 
 
Since Equation (18) is linear partial differential equation (PDE) we can write its                         
general solution as the sum of the solution to the corresponding homogeneous                       
equation (i.e. with the RHS set to zero) and any particular solution. Denoting                         
as the solution to the homogeneous equation as and                  the particular solution     
as , we can therefore write the total field as: 
 
23 
 
In order to assign a physical interpretation to                when we set RHS of Equation           
(18) to zero we have effectively set              . Thus we see that         
then is the pressure field that develops in the absence of any scatterers, which by                             
definition is the incident pressure field. We also know that the scattered pressure field                           
must obey Equation (18), and so we assign our particular solution to be the scattered                             
pressure field    . With these physical interpretations in mind, we see that the                     
total pressure field is indeed the sum of the incident and scattered pressure fields. 
 
2.3.1 The Incident Pressure Field 
In order to obtain an expression for the incident pressure field, we need to consider the                               
mechanics and geometry of the transducer. The generalized three­dimensional                 
coordinate system is shown in Figure 3. 
 
   Figure 3:  Coordinate system for describing scattering in an inhomogeneous medium 
   
where represents the area over the transmitter crystal, is the location of the                             
center of , is an arbitrary point on , is the volume within which the                               
24 
scatterers being considered are contained and is an arbitrary point in . The                         
surface may be considered to consist of infinitesimally small area element,                       
each of which behaves as a simple point source. The Huygen­Fresnel principle states                         
that each area element contributes a spherically expanding wave to the incident                       
pressure field. The incident pressure field ca then be obtained by summing the                         
spherical wave contribution from each area element. 
If we assume that the radius of curvature of is large enough, then is considered                                 
to be effectively flat. As a result, the incident pressure field can be expressed in terms                               
of the Rayleigh integral: 
 
where  is the temporal Fourier transform of the normal velocity on the                     
transmitter crystal’s surface; this normal velocity is not constant and varies from point                         
to point on . We note that term                corresponds to normal acceleration in         
the time domain, since the factor            corresponds to time differentiation. We have           
written  on the left­hand side of Equation (24) to explicitly indicate the dependence                       
of the incident pressure field on the location of the center of  . 
Although we have not shown the proof for Equation (24), we can intuitively see that it                               
is indeed the Huygen­Fresnel principle expressed mathematically: the integral on RHS                     
describes the summation of complex­valued spherically expanding waves (represented                 
by the complex exponential term), each weighted by the normal acceleration at its                         
source and decaying in amplitude with increasing distance from its source. 
For a typical transducer, we can consider the nominal normal velocity weighted by a                           
spatially variable term      to account for apodization, a filtering technique that               
removes discontinuities at the beginning and end of the sampled time record. We can                           
also model focusing by considering the normal velocity at each to be delayed by                             
.  
Substituting  into Equation (24), we obtain:   
25 
 
For convenience we define a new quantity that we define to be the spatial                             
transfer function: 
 
Equation (25) can then expressed more compactly as: 
 
This compact expression allows us to view the incident pressure field as the result of                             
temporally filtering the nominal normal velocity            by the spatial transfer       
function  . The spatial transfer function          incorporates 
the effects of apodization, and accounts for the spatial distribution of the incident                         
pressure field.  
 
2.3.2 The Scattered Pressure Field 
 
In order to find an expression for the scattered pressure field, we will consider the                             
Green’s function method. In our case, we consider spherical waves scattered from a                         
volume to be propagating into an effectively unbounded medium, in which case the                         
Green’s function takes the following form : 
 
The particular solution to Equation (23) is then the product of the RHS and the                               
Green’s function integrated over the volume. Furthermore if we define and to be                             
26 
zero outside of , then we can perform the integration over all of the three dimensional                               
space, and the scattered pressure field can be expressed as the convolution integral: 
   
Since we consider weak scattering, we assume that .                 
in Equation (23) becomes negligible, and            . 
Rewriting Equation (29) with        substituted by    yields: 
 
This approximation is referred to as the first Born approximation, and the equation (30)                           
states that the scattered pressure field is, to a first approximation, the spherically                         
expanding wave convolved with the scattering term. If we regard scatterers to be                         
idealised points in , then it is the equivalent of saying that each point scatterers                             
contribute spherically expanding wave independent of each other. Thus in making the                       
Born approximation, it is assumed that multiple scattering (i.e. waves scattered off a                         
particle that are then scattered off other particles ) is negligible. 
By substituting the expression for          , we can rewrite the scattering           
pressure field entirely in terms of the transducer’s characteristics and the scattering                       
operator: 
 
2.3.3 The Force on the Receiving Crystal 
Recall that the received RF voltage trace is obtained by summing the scattered pressure                           
field over the area of the receiving crystal, and filtering the sum using the                           
electromechanical impulse response of the transducer. In this section we will derive the                         
27 
summation formula for the scattered pressure field over the receiving crystal, and the                         
formula will be denoted by          , which is essentially the force on the receiving                 
crystal. 
Before we develop an expression for            , we will introduce two lemmas. These             
lemmas are essential in the derivation of the formula for the force on the receiving                             
crystal. 
Lemma 1 (Refer to Appendix A for proof): For any vector valued function A(x) and                             
scalar function b(x), if A(x) is zero outside some volume V’ then: 
 
 Lemma 2 (Refer to Appendix B for proof): At locations that are far away from the                                 
receiving crystal: 
 
Returning now to      , if we assume the same apodization and focusing at                   
reception as at transmission, then: 
 
where,  
 
We recognize that the integral is equal to                  , 
and so: 
 
From the definition of the scattering operator in Equation (20), we can rewrite Equation                           
(34) as follows: 
28 
 
To simplify Equation (35), we note that has finite limit, which allows us to use                               
Lemma 1  to simplify the integral on RHS of Equation (35): 
At sufficiently large distances away from the receiving crystal, we can use  Lemma 2 to                             
further simplify the force equation: 
 
 
2.3.4 The RF Voltage Trace 
 
In this section, we will discuss the conversion of the force on the receiving crystal into                               
a voltage RF trace. If we model the electromechanical transfer function of the                         
transducer to be      and the voltage trace to be            , we have: 
 
 
Substituting in the expression for          , we have: 
 
For convenience, we can group together the characteristics of the medium together, and                         
the electromechanical properties of the transducer together. We adopt definitions similar                     
to those in Equation (27): 
29 
 
 
 
where  represents convolution in the space domain. 
In keeping with the terminology introduced in Equation (27), we refer to the quantities                           
and  respectively as the pulse­echo wavelet and the tissue reflectivity or                   
scatterer field. 
We can also express the voltage trace in the time domain as follows: 
 
 
 
where  and  represent convolution in the space and time domain respectively. 
 
If we regard the quantity as the input signal and the quantity as output                               
signal, then Equations (40) and (43) indicate that the imaging system has a                         
spatiotemporal transfer function and the impulse response function                 
. Thus, the definition of the transfer function this way neatly                     
distinguishes between the electromechanical characteristics of the transducer (represented                 
30 
by the pulse­echo wavelet) and the geometry of the transducer, represented by                       
. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
31 
Chapter 3 
Pulse­Echo Experiments Using a Single 
Element Transducer 
 
The main goal of this project is to obtain B­mode images of scatterers small enough                             
to cause speckles, and then compare experimental results with theoretical                   
predictions from a mathematical model that would simulate B­mode images of                     
those scatterers showing speckles. Developing this model will rely on much of the                         
theoretical background described in section 2.3 of Chapter 2, which incorporates                     
transducer geometry and its corresponding pressure­field calculations for predicting                 
the RF voltage trace for that transducer geometry. However before an experimental                       
prototype and its corresponding theoretical model can be developed for describing                     
speckles, the first step would be to model and gather experimental data for discrete                           
scatterers, which are larger than speckles and the scatterers that cause speckles. This                         
chapter describes the experimental procedures for obtaining B­mode images of                   
discrete scatterers.  
The process of obtaining B­mode images for scatterers within a                   
homogeneous medium is very challenging. This is because measurements and data                     
collection are very sensitive to external factors, for example transducer alignment                     
and sample material. For instance, a comparison of backscatter coefficient                   
measurements performed at eight different ultrasound physics laboratories found                 
variations in the final results of almost two orders of magnitude (Wear, et. al, 2005).                             
However, some of this variation also arises due to differences in experimental                       
techniques.  
Techniques for measuring acoustic properties of phantoms may vary in                   
several ways, for instance bandwidth of the technique and the type of transducer                         
used. For our experiments, we have used a broadband, single­element focused 5                       
MHz transducer. The advantage of using a broadband transducer lies in the fact that                           
32 
it allows for simultaneous RF data acquisition over a range of frequencies; this also                           
means that the transducer can be excited with a short pulse rather than a                           
quasi­continuous wave excitation. Also using a single­element transducer instead of                   
an array of transducers reduces the complexity of data analysis of the signals                         
received.  
 
 
3.1 Making Scatterer Phantoms: 
 
Scatterer phantoms are a uniform distribution of symmetric structures in a                     
homogeneous medium, based on the discrete scatterer model mentioned in Chapter                     
1. Scatterer phantoms were made by embedding beads in an agarose base, enclosed                         
by plastic petri dishes of about 3.6 cm in diameter and 0.8 cm in height.  
Using the speed of sound in agarose at 24 °C, 1500 m/s, and the peak                             
frequency of the transducer, which for the purposes of our experiment was 5 MHz,                           
the minimum size of the scatterers were calculated: 
 
The minimum size of scatterers were calculated assuming Rayleigh scattering takes                     
place, where scatterers are assumed to be smaller than the wavelength of ultrasound                         
waves. This value is significant for the following reasons: first, it provides an                         
estimate of the size of scatterers, which may cause speckles (these scatterers are                         
expected to be of the order of several hundred microns); further more it enables us                             
to determine how large discrete scatterers should be compared to clusters of                       
indistinguishable scatterers that cause speckles.   
Beads of the following size ranges (from largest to smallest) were selected:                       
4.76 mm, 2.85 mm ­ 3.45 mm and 1.0 mm ­ 1.3mm. Each petri dish was first filled                                   
with a relatively thin layer of liquid agarose. This layer was allowed to congeal, and                             
then a second layer of liquid agarose was poured on top of the first layer. As soon                                 
33 
as the second layer was poured, the scatterers were embedded into this layer. The                           
reason why scatterers were not embedded in the first agarose layer is because when                           
scatterers are offset from the base of the petri dish, it is much easier to differentiate                               
RF signals of the base from those of the scatterers when the RF data is converted to                                 
B­mode images (since the scatterers are quite small, RF signals from the base have                           
a much higher intensity, and appear “brighter” in the grey scale B­mode image). 
Figure 4 shows the scatterer phantoms. The beads were placed in a                       
symmetric pattern in order to obtain an approximate value for the spacing between                         
each bead, and knowing this spacing in conjunction with scatterer size enables                       
better interpretation of B­mode images. The 4.76 mm beads and the 2.85 mm ­ 3.45                             
mm beads were arranged in a rectangular array using a tweezer, with an average                           
spacing of about 6 mm and 3mm respectively. Since the 1.0 mm ­ 1.03 mm beads                               
were too small to be arranged linearly, they were arranged radially outward from                         
the center using the tip of a thin metal strip, and arranged radially with respect to a                                 
2.85 mm ­ 3.45 mm glass bead placed at the center of the agar base; the average                                 
spacing was about 2 mm.  
 
 
                   (a)                                                (b)                                          (c) 
 
Figure 4:  Scatterer Phantoms (a) Metal beads: 4.76 mm, spacing between beads about  
6 mm, (b) 2.85 mm ­ 3.45 mm glass beads, spacing between beads about 3 mm,  
(c) 1.0 mm ­ 1.03 mm glass beads, spacing between beads about 2 mm. 
 
   
 
 
 
34 
3.2 Experimental Setup: 
 
As with all acoustic property measurements, the results of the pulse­echo                     
measurement can depend on temperature. Therefore, every attempt should be made                     
to achieve a uniform temperature in your sample, which is typically room                       
temperature (22.0 °C). For typical size phantom samples, the samples should be in                         
the water tank at this temperature for 1 to 2 hours prior to starting measurements.                             
This should allow enough time for the sample to uniformly be at 24 °C. For our                               
experiments, we used a 5 MHz focused transducer with a usable bandwidth of                         
about 60% of its central frequency (from about 3.5 MHz to 6.5 MHz). 
The initial setup of the experiments are illustrated in Figure 5: 
 
 
                         Figure 5:  Experimental setup for pulse­echo experiment 
 
The stage of the x­y­z translator is held in position with screws, such that the                             
transducer will face downwards when mounted onto the stage. After the stage is                         
securely stationed, the transducer is carefully mounted on the stage and held firmly                         
in position by tightening a pair of screws.   
For each phantom, the top of the petri dish is taken off and then the bottom                               
surface is securely attached to the base of a large, glass container using double                           
35 
sided tape. The phantoms should be fixed rather firmly to prevent them from being                           
displaced as they are being submerged in water. The container is then filled with                           
deionized water until the meniscus of the water level above the phantoms is about                           
5 cm, which is approximately equal to the focal length of the transducer (52.2                           
mm). The phantoms should be roughly at the focal length so that the distance                           
between each phantom and the transducer can be later adjusted to observe                       
scatterers at the correct time delay, specifically at the focus. 
 
 
3.3 Electronic Setup:  
The pulse­echo measurements require the use of an oscilloscope (Tektronix TDS                     
3014C Digital Phosphor Oscilloscope with 1.25 GS/s digitization), translator                 
motor for positioning the transducer, and pulse­receiver (Olympus               
Panametrics­NDT 5800). A pulse­receiver is a device that does just as its name                         
suggest: it sends out a voltage spike or pulse, and then turns itself into a broadband                               
receiver amplifier. The transducer is connected to the pulse­receiver using a BNC                       
cable, where the transducer is attached to the mounting bracket of the male UHF                           
adaptor.   
In order to provide a trigger to the oscilloscope, a cable is connected from the                             
"Ext Trig/+Sync" jack of the pulse­receiver to the "Ext" jack of the oscilloscope to                           
provide external trigger to the oscilloscope. Then "Gated RF" output of the                       
pulse­receiver is connected to channel 1 of the oscilloscope. In order to provide                         
some low­pass filtering, a low­pass filter is then inserted between the Gated RF                         
jack and the oscilloscope in order to ensure that signals associated with the central                           
frequency of the transducer, and its bandwidth are detected. Note that the filters                         
have an input and an output side. A schematic of the setup is shown in Figure 6. 
36 
   
     Figure 6:  Transducer, Oscilloscope and Pulse­Receiver connections 
 
The mode of the pulse­receiver is set to "Pulse­Echo". The settings for                       
"Gain", "Attenuation", and "Energy" are determined, and for our experiments they                     
are usually set to 20 dB, 0, 50 J respectively. After the pulse/receiver is turned                             
on, the gate settings and oscilloscope time­voltage divisions are adjusted to locate                       
the high amplitude echo signals from the front and back surfaces of the sample.  
 
3.4 Aligning the Transducer:  
The front surface of the sample defines one geometric plane in space. The                         
combination of the x and y­axes of the transducer translation system define another                         
geometric plane. It is important that the surface of the sample be as close as                             
possible to parallel to the plane of motion of the transducer. If this is not the case,                                 
37 
then the distance the beam travels in water will vary as the transducer moves to                             
obtain independent power spectrum estimates. Unfortunately, there is no automatic                   
alignment system with this setup. However, you can use the computer to determine                         
if the two planes are parallel.   
The motor of the x­y­z translator is controlled using a program called                       
COSMOS 3.6.1. The program has two different settings, which control the motors:                       
the “Quick Move­Single Axis” and “Virtual Jog”. The “Quick Move” setting is                       
most useful for carrying out raster scans because it moves the motor a specific                           
distance from a point of reference, defined by the step­size associated with each                         
click. However, the “Virtual Jog” setting is used for fine tuning the position of the                             
transducer; it is useful for centering the transducer, and locating the edge of the                           
sample.  
The translator motors are labelled 1 through 3, where 1 corresponds to the                         
y­axis, 2 corresponds to the x­axis and 3 corresponds to the z­axis. In order to                             
lower the transducer into the glass container, the “Quick Move” setting is used to                           
control motor 3. The transducer is vertically lowered until it is sufficiently below                         
the surface of the water at an appropriate distance from the phantom, and this is                             
determined using the time delay. For the time delay calculation the variables (in                         
s)  and   are considered, where: 
 
is the speed of sound in the agar base of the phantom and water,.                             
For our calculations we set the offset to zero, assuming that sound waves are not                             
significantly reflected from the top surface of the sample. Since all the RF data is                             
collected in a window of 10 s, the “window” term in the expression for is set to                                   
10 and multiplied by 0.000001 to convert into microseconds. After obtaining the                         
38 
values of and , and given that the sample is roughly located at the focal length                                 
of the transducer, the time delay  is calculated as follows: 
 
where  , is the speed of sound in water. For our experiments,                     
time delay was calculated to be 64.8 s. On the oscilloscope, the signal appears as                             
an envelope, which outlines the variation in amplitude of the scattered signal                       
against a time scale. Usually, the first high amplitude signals correspond to the                         
interface between the surrounding water and top surface of the phantom. The next                         
set of signals are recognized as scatterers distributed in the agar phantom because                         
the signal amplitude is much smaller than that of the interface, and each signal                           
corresponds to a scatterer. However, the amplitude of the scatterer signals is                       
largely dependent on the size of the scatterers. The last set of signals correspond to                             
the base of the phantom petri dish, and look very similar to the initial set of signals                                 
due to the interface. Next, the voltage time scale is adjusted to locate the signal                             
from the top surface of the phantom. After the top surface is located, the time delay                               
of the oscilloscope is set to 64.8 s. Next, in order to find the correct distance                               
between the transducer and phantom, motor 3 is used to further adjust the vertical                           
height of the transducer until the signal from the top surface is at the origin of the                                 
voltage­time axis. Once the appropriate distance is determined, the time delay is                       
offset by 5 s to 69.8 s so that the signal inside the phantom fills the oscilloscope                                 
screen, which we use to window our RF signal. 
Before centering the transducer, the axis of the stage is manually adjusted                       
using positioning knobs on the translation mount; this ensures that the transducer is                         
parallel to the surface of the sample. Once the transducer’s vertical orientation is                         
adjusted, the “Quick Move” setting of the program is used to move motors 1 and 2                               
to align the transducer to the center of the phantom.  
39 
Once we have centered the transducer, we then located the edge of the                         
sample in the x and y direction by observing changes in the signal amplitude. The                             
“Quick Move” setting is used to move motors 1 and 2 slowly along the x and y                                 
direction. The edges usually appear as a sub­envelope of signals, whose amplitude                       
increases and then decreases rapidly. This sub­envelope noticeably diminishes as                   
we move just off the edge of the sample until there is no signal at all. Figure 7                                   
shows an RF data at time delay for 4.76 mm scatterers. 
                      Figure 7:  RF data at time delay for 4.76 mm scatterers 
 
3.5 Collecting RF Pulse­Echo Signal Data: 
After the edge is located, the next step would be to figure out exactly how many                               
steps away from the center, and this is done using the “Virtual Jog” setting.                           
“Virtual Jog” uses step sizes, which are defined as the smallest move of the                           
motorized x­y­z stage. For instance, when the edge is located in the x direction,                           
motor 2 is moved away from the edge using a step size of 330 (any step size less                                   
than or equal to 400 should be reasonable). As the transducer is moved towards the                             
40 
edge of the phantom, it is important to keep track of the number of clicks it takes                                 
for the edge signals to appear and eventually diminish again. The same procedure                         
is repeated in the y direction using motor 1. 
In order to obtain RF voltage trace data, a raster scan is performed. But                           
before the scan can be performed it is important to know the beam width. The                             
beam width of the sound waves is calculated, assuming limited diffraction through                       
a circular aperture for the concave transducer surface: 
 
 
where is the beamwidth, is the focal length (0.0522 m), is the diameter                               
of the transducer element (0.5 inches) is the speed of sound in the phantom and                               
 is the frequency of the transducer (5 MHz). 
For the 5 MHz focused transducer, the beam width is 0.015 mm. The beam                             
width is converted to motor steps, and the corresponding value (238 motor steps) is                           
set as the step size for the raster scan in order to obtain independent beam lines.                               
Figure 8 shows the rectangular pattern of signal reception, where the transducer is                         
moved in a square with five steps in the x and y direction. When the raster scan is                                   
complete, all the data acquired from the oscilloscope is saved in a USB and loaded                             
into a MATLAB program, the raw data is the RF voltage trace. The program is                             
also designed to produce B­mode images of the horizontal or vertical                     
cross­sections of the phantoms by mapping the vector arrays (associated with the                       
RF data) on two­dimensional space domain, that is the x­y plane in our case. Refer                             
to Appendix C for a detailed description of how the program works.  
41 
   
                                Figure 8:  Raster scan pattern. The circle represents the top plane of 
                                 the sample. The raster scans were done in a square pattern where each   
                                 side was 1190 motor steps (0.075 mm) in length.   domain.  
 
Thus, the x and y axis of the final grayscale B­mode image indicate distances                           
within the phantom, where the y­axis shows how deep within the phantom the                         
scatterer is located relative to the top surface of the phantom. 
 
 
 
 
 
 
 
 
 
 
 
 
42 
Chapter 4 
Obtaining B­mode Images 
 
Before we add complexity to our experiments and make phantoms consisting of                       
scatterers in the Rayleigh scattering limit, it is important to verify phantom                       
preparation methods and reasonably image discrete scatterers. In this chapter, we                     
will look at B­mode images of the discrete scatterer phantoms discussed in Chapter                         
3. Having B­mode images of discrete scatterers will not only test accuracy of                         
preparation methods, but it will also enable comparison of experimental results with                       
the theoretical results once we establish our own mathematical model for obtaining                       
RF voltage trace and B­mode images for a single­element transducer geometry. 
 
4.1 B­mode Image of 4.76 mm Metal Beads: 
                                          Figure 9:  RF data for 4.76 mm scatterers. 
   
43 
                                  Figure 10 : B­mode image of 4.76 mm scatterers 
 
For the 4.76 mm metal beads, the brightest bands indicated in Figure 10 represent                           
the top and bottom surface of a single scatterer. The distance between the respective                           
bright bands is roughly the size of a scatterer, thus confirming that the bright bands                             
arise due to sound waves striking the scatterer head on. Looking at the RF data for                               
these scatterers, we are also able to identify the top and bottom surface of the                             
scatterer at about 70 s and 73.5 s. Thus, B­mode imaging of 4.76 mm scatterers,                             
was successful. 
 
 
 
 
44 
4.2 B­mode Image of 2.85 mm ­ 3.45 mm Glass 
      Beads: 
                        Figure 11:  RF data for 2.85 mm ­ 3.45 mm glass beads 
45 
                   Figure 12:  B­mode image of 2.85 mm ­ 3.45 mm glass beads 
 
Since the spacing between theses scatterers (about 3 mm) was approximately the                       
same as the average size of the scatterers (about 3.15 mm), we observed a series of                               
bright bands across the plane of the phantom. This is because as the scatterers were                             
relatively clustered, sound waves struck the edges of most of the scatterers giving                         
rise to a linear array of bright bands. However, some of the sound waves managed  
to strike a single scatterer head on, and again we observe a much brighter set of                               
bands as indicated in Figure 12. The distance between the brightest bands is about                           
2.75 mm, which is roughly equal to the average diameter of a single scatterer.                           
From Figure 11 we observe the front and back surface of the scatterer to be around                               
71 s and 72.8 s respectively, and this further confirms that the 2.85 mm ­ 3.45                               
mm scatterers were successfully imaged. 
 
4.3 B­mode Image of 1.0 mm ­ 1.03 mm Glass 
      Beads: 
                               Figure 13:  RF data for 1.0 mm ­ 1.03 mm glass beads 
46 
 
                          Figure 14:  B­mode image of 1.0 mm ­ 1.03 mm glass beads 
 
Spacing between the 1.0 mm ­ 1.03 mm scatterers (about 2mm) was not too large                             
compared to the average size of the scatterers (about 1.02 mm), and we again                           
observe a series of relatively bright bands like those in the B­mode image for the                             
2.85 mm ­ 3.45 mm scatterers. Among the series of bright bands, we observe                           
brighter bands corresponding to the front and back surface of a single scatterer as                           
seen in Figure 14. We estimate the distance between the bright bands to be around                             
1mm, which again is roughly the same size as the diameter of a single scatterer.                             
Figure 13 also indicates the front and back surface of the scatterer as seen on the RF                                 
data; the front surface is located at around 71.8 s, whereas the back surface is                             
around the 72.4  s mark.  
 
47 
Chapter 5 
Conclusion and Future Work 
 
From the results of our experiments, we can conclude that we have been able to                             
successfully produce B­mode images of discrete scatterers, which are larger than                     
scatterers that cause speckles. The results also indicate the phantom preparation                     
method complies with the discrete scatterer model, where the scatterer phantom is                       
depicted as a collection of points, spheres or cylinders embedded in a homogeneous                         
medium. However, the results presented in this thesis are the first steps towards                         
achieving the main goal of the experiment: producing B­mode images of scatterers                       
in the Rayleigh scattering limit.  
Now that we have a mathematical model for predicting the RF voltage trace,                         
as discussed in Chapter 2, the next step would be to extend that model to describe                               
scatterers of various sizes (higher or lower than the Rayleigh scattering limit), and                         
simulate ultrasound B­scan images of scatterer phantoms. We started some                   
preliminary modeling of the incident and scattered pressure fields, and both these                       
models are discussed in detail in Appendix D and E. However, these models are                           
still in the early stages of development and require addition of further complexity,                         
such as developing sub­functions for apodization and scattering terms, in order for                       
them to be more realistic.  
Thus in order to effectively model speckles, the process of writing up a                         
complete program model will involve pressure­field calculations similar to those                   
discussed in Chapter 2. These calculations will help model the RF voltage trace of a                             
given transducer geometry (a single element focused transducer in our case), and                       
then convert the RF data to B­mode images for comparison between experimental                       
and theoretical results. After a working model is established, keeping in mind the                         
discrete scatterer model for scatterer phantoms, the theoretical B­mode images will                     
be compared with experimental results (including the ones described in Chapter 4)                       
48 
to verify the accuracy of the mathematical model for describing speckles.                     
Achieving a good agreement between experimental and theoretical results will                   
enable a better understanding of the nature of speckles and what causes them.                         
Developing a model that accurately describes speckles will help us develop an                       
ultrasonic microscope to map scatterers at high frequency locations within                   
biological systems, which we are unable to detect using current ultrasound imaging                       
techniques. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
49 
 
Appendix: 
 
A Proof of Lemma 1 (Chapter 2, 2.3.3) 
 
We begin with the identity:  
 
which can be verified by expanding the RHS and simplifying. Integrating both                       
sides over  : 
 
By the divergence theorem, the second integral on the RHS is equal to the surface                             
integral: 
 
where is some reference surface enclosing and            is a unit vector normal         
to . Since      is zero outside , the surface integral reduces to zero and                     
Equation  follows.  Q.E.D. 
 
 
 
 
 
50 
 
B Proof of Lemma 2 (Chapter 2, 2.3.3) 
 
We define the wave vector          , where    , in other words         
is a unit vector parallel to and                is a vector also parallel to             
but with a magnitude        . We can then rewrite Equation for the spatial transfer                   
function as, 
 
Taking the gradient, 
 
We apply the condition        , which is equivalent to         
since  . The above equation then becomes: 
 
 
We also assume that the direction of does not vary very much over the receiving                             
crystal, which allows the   term on the RHS to be factored out of the integral, 
51 
Lemma 2 follows from this immediately.  Q.E.D. 
In practice,    is satisfied for virtually the entire imaged region               
since the wavelength from medical ultrasonic transducers are usually very short. In                       
the context of our experiments described in Chapter 3, the single­element focused                       
transducer is transmitting at 5 MHz with an average speed of sound of 1490 m/s in                               
agar and water. For this transducer            , which is small enough to           
be considered negligible. 
The condition that the direction of not vary very much over the receiving                           
crystal is much stricter and is only satisfied in regions far away from the receiving                             
crystal. To quantify, what is exactly meant by “far away” consider the following                         
figure: 
  Figure A:  Coordinate system for calculating an approximation to   
 
For a typical focused transducer, we expect the majority of the transmitted                       
acoustic energy to be concentrated in a small region enclosing the axial axis, and                           
we have chosen the scenario where lies along this axis. We can restate our                             
requirement that can be uniform over the surface of the receiving crystal by                           
requiring    to be approximately parallel to   . 
This requirement is most difficult to satisfy when we consider the point on                         
farthest from the center of the receiving crystal, i.e. when                    .  
52 
Our requirement that be approximately parallel to can be stated in terms of                             
the dot product      ; in the case where          (as shown in     
the figure above), we require . If we allow a 10% error in the                           
approximation of , we effectively impose the constraint where                   
. From the figure        , and so our requirement that           
translates to requiring      . 
What we have demonstrated in this brief discussion is that the approximation                       
considered as a part of lemma 2 is satisfied well at axial depths that are greater                               
than the diameter of the receiving crystal. For non­circular apertures, this diameter                       
is the smallest circle within which the receiving crystal can be ascribed. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
53 
C Matlab Program for Making B­mode   
          Images 
 
 
   
54 
Figure C1:  Matlab program for converting RF data to B­mode images 
 
Figure C1 shows the Matlab program that was written to convert the RF data                           
to B­mode images. The RF data is loaded on to the program using the defined                             
variables for file path and name of the designated folder, where all the data is                             
saved. Since the RF data file is essentially a three­dimensional array vector,                       
variables describing the RF vector array and the B­mode array are defined. The                         
vector array can be visualized as shown in Figure C2: 
 
                           Figure C2:  Three­dimensional vector array 
(Source:http://www.mathworks.com/help/matlab/math/multidimensional­arrays.html) 
 
 
55 
 
In the context of our pulse­echo experiments, ‘row’ is RF amplitude the along the                           
x­axis of the sample plane, ‘column’ is the RF amplitude along the sample in the y                               
direction and ‘page’ is the time interval between between transmitted and received                       
signals. Both the x and y components of the vector array have two sets of signal                               
associated with them: the input RF signal from the transmitting crystal, and the                         
output RF signal from the receiving crystal.  
The main job of the program is to convert the three­dimensional vector array                         
into a 2­by­2 vector matrix, which is then converted to a two­dimensional B­mode                         
image. This process is initiated by the iteration loop spanning lines 26 through 32                           
Currently this for­loop has a limitation that it can do a maximum of nine iterations,                             
which means that it can not handle more than nine sets of data from the raster scan;                                 
we use five sets of data to produce the B­mode image. The final B­mode image in                               
our case is a horizontal cross section along the sample plane, therefore only the                           
output RF signal is relevant along the x axis. Thus in line 29 of Figure C1                               
“sampleData (:, 2, :)” indicates that we are only considering the output RF signal                           
in the x direction. The operation “squeeze” then removes singleton dimensions and                       
converts the three­dimensional vector array into a 2­by­2 matrix in the space                       
domain. 
The two­dimensional vector array is then converted to a B­mode array by                       
taking the natural logarithm of the RF amplitude data vector array, which                       
graphically corresponds to the intensity or brightness level that we see in the final                           
B­mode image. Next, the x and y distances along the plane of the sample are                             
defined (line 43­54) using step­size (the smallest distance covered by the translator                       
motor), window size in microseconds and the number of raster scan data sets. Then                           
the image layout is specified as shown in Figure C1 between lines 57 to 67. 
 
56 
Thesis MHC 2015_Huma Yusuf - Google Docs
Thesis MHC 2015_Huma Yusuf - Google Docs
Thesis MHC 2015_Huma Yusuf - Google Docs
Thesis MHC 2015_Huma Yusuf - Google Docs

More Related Content

What's hot

NANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENT
NANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENTNANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENT
NANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENTKeshav Das Sahu
 
Phototoxicity in live cell imaging workshop CYTO2018
Phototoxicity in live cell imaging workshop CYTO2018Phototoxicity in live cell imaging workshop CYTO2018
Phototoxicity in live cell imaging workshop CYTO2018Jaroslav Icha
 
IRJET- Survey on Detection of Cervical Cancer
IRJET- Survey on Detection of Cervical CancerIRJET- Survey on Detection of Cervical Cancer
IRJET- Survey on Detection of Cervical CancerIRJET Journal
 
Proton vda brochure
Proton vda brochureProton vda brochure
Proton vda brochureMiha Ulčar
 
IRJET - Classification of Cancer Images using Deep Learning
IRJET -  	  Classification of Cancer Images using Deep LearningIRJET -  	  Classification of Cancer Images using Deep Learning
IRJET - Classification of Cancer Images using Deep LearningIRJET Journal
 
S Van Haver Optimized 2009 Enz Zernike
S Van Haver Optimized 2009 Enz ZernikeS Van Haver Optimized 2009 Enz Zernike
S Van Haver Optimized 2009 Enz ZernikeKnowledge_Broker
 
Comparitive study of brain tumor detection using morphological operators
Comparitive study of brain tumor detection using morphological operatorsComparitive study of brain tumor detection using morphological operators
Comparitive study of brain tumor detection using morphological operatorseSAT Journals
 
IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...
IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...
IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...IRJET Journal
 
2008-05-13 Optical Imaging NIH Presentation
2008-05-13 Optical Imaging NIH Presentation2008-05-13 Optical Imaging NIH Presentation
2008-05-13 Optical Imaging NIH PresentationLawrence Greenfield
 
SophieZhangXeroxPosterFinal2015
SophieZhangXeroxPosterFinal2015SophieZhangXeroxPosterFinal2015
SophieZhangXeroxPosterFinal2015Sophie Zhang
 
Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...
Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...
Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...Editor IJCATR
 
PPT on BRAIN TUMOR detection in MRI images based on IMAGE SEGMENTATION
PPT on BRAIN TUMOR detection in MRI images based on  IMAGE SEGMENTATION PPT on BRAIN TUMOR detection in MRI images based on  IMAGE SEGMENTATION
PPT on BRAIN TUMOR detection in MRI images based on IMAGE SEGMENTATION khanam22
 
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...yudhveersingh18
 
A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...
A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...
A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...ejhukkanen
 

What's hot (20)

NANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENT
NANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENTNANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENT
NANOPARTICLES IN CANCER DIAGNOSIS AND TREATMENT
 
Brain
BrainBrain
Brain
 
Phototoxicity in live cell imaging workshop CYTO2018
Phototoxicity in live cell imaging workshop CYTO2018Phototoxicity in live cell imaging workshop CYTO2018
Phototoxicity in live cell imaging workshop CYTO2018
 
Magnetic Tweezer
Magnetic TweezerMagnetic Tweezer
Magnetic Tweezer
 
Characterization techniques of nanoparticles
Characterization techniques of nanoparticlesCharacterization techniques of nanoparticles
Characterization techniques of nanoparticles
 
IRJET- Survey on Detection of Cervical Cancer
IRJET- Survey on Detection of Cervical CancerIRJET- Survey on Detection of Cervical Cancer
IRJET- Survey on Detection of Cervical Cancer
 
Proton vda brochure
Proton vda brochureProton vda brochure
Proton vda brochure
 
Viva201393(1).pptxbaru
Viva201393(1).pptxbaruViva201393(1).pptxbaru
Viva201393(1).pptxbaru
 
Middle pages
Middle pagesMiddle pages
Middle pages
 
IRJET - Classification of Cancer Images using Deep Learning
IRJET -  	  Classification of Cancer Images using Deep LearningIRJET -  	  Classification of Cancer Images using Deep Learning
IRJET - Classification of Cancer Images using Deep Learning
 
Imaging concepts group research overview
Imaging concepts group research overviewImaging concepts group research overview
Imaging concepts group research overview
 
S Van Haver Optimized 2009 Enz Zernike
S Van Haver Optimized 2009 Enz ZernikeS Van Haver Optimized 2009 Enz Zernike
S Van Haver Optimized 2009 Enz Zernike
 
Comparitive study of brain tumor detection using morphological operators
Comparitive study of brain tumor detection using morphological operatorsComparitive study of brain tumor detection using morphological operators
Comparitive study of brain tumor detection using morphological operators
 
IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...
IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...
IRJET- Segmentation of Nucleus and Cytoplasm from Unit Papanicolaou Smear Ima...
 
2008-05-13 Optical Imaging NIH Presentation
2008-05-13 Optical Imaging NIH Presentation2008-05-13 Optical Imaging NIH Presentation
2008-05-13 Optical Imaging NIH Presentation
 
SophieZhangXeroxPosterFinal2015
SophieZhangXeroxPosterFinal2015SophieZhangXeroxPosterFinal2015
SophieZhangXeroxPosterFinal2015
 
Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...
Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...
Brain Tumor Detection Using Artificial Neural Network Fuzzy Inference System ...
 
PPT on BRAIN TUMOR detection in MRI images based on IMAGE SEGMENTATION
PPT on BRAIN TUMOR detection in MRI images based on  IMAGE SEGMENTATION PPT on BRAIN TUMOR detection in MRI images based on  IMAGE SEGMENTATION
PPT on BRAIN TUMOR detection in MRI images based on IMAGE SEGMENTATION
 
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
Design of Modified Bio-Inspired Algorithm for Identification and Segmentation...
 
A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...
A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...
A Systems Approach to the Modeling and Control of Molecular, Microparticle, a...
 

Similar to Thesis MHC 2015_Huma Yusuf - Google Docs

SPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONS
SPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONSSPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONS
SPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONSZhuo Wang
 
Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...
Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...
Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...sela shefy
 
YamengCaoPhDThesis
YamengCaoPhDThesisYamengCaoPhDThesis
YamengCaoPhDThesisYameng Cao
 
Handbook of coherent domain optical methods
Handbook of coherent domain optical methodsHandbook of coherent domain optical methods
Handbook of coherent domain optical methodsSpringer
 
Energy-dispersive x-ray diffraction for on-stream monitoring of m
Energy-dispersive x-ray diffraction for on-stream monitoring of mEnergy-dispersive x-ray diffraction for on-stream monitoring of m
Energy-dispersive x-ray diffraction for on-stream monitoring of mJoel O'Dwyer
 
Optimisation of X-Ray CT within SPECTCT Studies
Optimisation of X-Ray CT within SPECTCT StudiesOptimisation of X-Ray CT within SPECTCT Studies
Optimisation of X-Ray CT within SPECTCT StudiesLayal Jambi
 
NNIN Convocation 2008
NNIN Convocation 2008NNIN Convocation 2008
NNIN Convocation 2008Jose Guevarra
 
Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...
Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...
Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...Ben Adekunle
 
davidldummerThesis2003_20070708secure
davidldummerThesis2003_20070708securedavidldummerThesis2003_20070708secure
davidldummerThesis2003_20070708secureDavid Dummer MS, PE
 

Similar to Thesis MHC 2015_Huma Yusuf - Google Docs (20)

SPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONS
SPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONSSPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONS
SPATIAL LIGHT INTERFERENCE MICROSCOPY AND APPLICATIONS
 
Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...
Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...
Quantification of Fluorophore Concentrations in Turbid Media by Spatial Frequ...
 
YamengCaoPhDThesis
YamengCaoPhDThesisYamengCaoPhDThesis
YamengCaoPhDThesis
 
Handbook of coherent domain optical methods
Handbook of coherent domain optical methodsHandbook of coherent domain optical methods
Handbook of coherent domain optical methods
 
Kenney16PhD1
Kenney16PhD1Kenney16PhD1
Kenney16PhD1
 
X ray crystallography analysis
X ray crystallography analysis X ray crystallography analysis
X ray crystallography analysis
 
Energy-dispersive x-ray diffraction for on-stream monitoring of m
Energy-dispersive x-ray diffraction for on-stream monitoring of mEnergy-dispersive x-ray diffraction for on-stream monitoring of m
Energy-dispersive x-ray diffraction for on-stream monitoring of m
 
Optimisation of X-Ray CT within SPECTCT Studies
Optimisation of X-Ray CT within SPECTCT StudiesOptimisation of X-Ray CT within SPECTCT Studies
Optimisation of X-Ray CT within SPECTCT Studies
 
NNIN Convocation 2008
NNIN Convocation 2008NNIN Convocation 2008
NNIN Convocation 2008
 
CV-Detailed-Pub
CV-Detailed-PubCV-Detailed-Pub
CV-Detailed-Pub
 
MPEF_Microscopy
MPEF_MicroscopyMPEF_Microscopy
MPEF_Microscopy
 
Basics of light microscopy and imaging sonderheft_olympus
Basics of light microscopy and imaging sonderheft_olympusBasics of light microscopy and imaging sonderheft_olympus
Basics of light microscopy and imaging sonderheft_olympus
 
Biophotonics
BiophotonicsBiophotonics
Biophotonics
 
Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...
Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...
Application of Medical Image Fusion models, to OCT and Fundus Photographic Im...
 
Cv Dillmann
Cv DillmannCv Dillmann
Cv Dillmann
 
davidldummerThesis2003_20070708secure
davidldummerThesis2003_20070708securedavidldummerThesis2003_20070708secure
davidldummerThesis2003_20070708secure
 
P3 Medical Physics
P3 Medical PhysicsP3 Medical Physics
P3 Medical Physics
 
X ray diffraction
X ray diffractionX ray diffraction
X ray diffraction
 
X ray diffraction
X ray diffractionX ray diffraction
X ray diffraction
 
Oxford_15-03-22.pptx
Oxford_15-03-22.pptxOxford_15-03-22.pptx
Oxford_15-03-22.pptx
 

Thesis MHC 2015_Huma Yusuf - Google Docs

  • 1. Modeling Speckles for Ultrasound Imaging  by  Huma Yusuf         Submitted to the Department of Physics in partial fulfillment of  the requirements for the degree of   Bachelor of Arts in Physics  at  Mount Holyoke College  May 2015    © MOUNT HOLYOKE COLLEGE 2015. All rights reserved.    Certified by……………………………………………………..                                                                            Maria Teresa Herd                                              Laboratory Director, Department of Physics                                                                                       Thesis Supervisor    Accepted by……………………………………………………                                                                                 Mark Peterson                   Co­Chair of Physics, Professor of Mathematics and Physics                     1 
  • 2. Abstract  Ultrasound has been used in a variety of clinical settings, including obstetrics and                          gynecology, cardiology and cancer detection. The main advantage of ultrasound is                      that, unlike x­ray imaging, it does not require ionizing radiation, which may                        increase the risk of getting cancer. In addition, using ultrasound imaging as a                          medical diagnostic tool is more cost effective compared to other radiation free                        diagnostic techniques, such as the MRI. However, a difficult and crucial challenge                        in ultrasound medical imaging is the necessity of reducing the appearance of                        speckles in B­scan images. Speckles are caused by coherent interference of                      reflected ultrasound waves by structures smaller than the Rayleigh scattering limit.                      Appearance of speckles tend to reduce the perception of small structures, and                        ultimately limit diagnostic accuracy of medical imaging system. Since it is difficult                        to model speckles using architectural structure of tissues, modeling techniques                    involve a more constructive approach: predicting the speckle pattern of an arbitrary                        scattering medium for a given transducer geometry; such a model was developed                        by Foster et al. (1983).     Investigation of the origin and nature of speckles involve creating a mathematical                        model to predict the RF output signal waveform of a particular transducer                        geometry, and then extending that model to incorporate the effects of varying sizes                          of scatterers on the RF voltage trace. The complete model will then be verified by                              comparing theoretical results with experimental results obtained from pulse­echo                  experiments. For this project, we have focused more on the experimental aspect of                          the investigation by producing B­mode images of scatterer phantoms. These                    phantoms are standard samples made using the discrete scatterer model, where                      scatterers of uniform shape and size are embedded in a homogeneous medium.                        Before imaging scatterers smaller than the Rayleigh scattering limit, it is important                        to ensure that scatterers much larger than the Rayleigh limit (discrete scatterers)                        can be successfully imaged. We conducted pulse­echo experiments using a 5 MHz                        single­element focused transducer, and imaged phantoms containing discrete                scatterers of the following sizes: 4.76 mm metal beads, 2.85 mm ­ 3.45mm glass                            beads and 1.00 mm ­ 1.03 mm glass beads. The RF voltage traces from the                              pulse­echo experiments were then converted to B­mode images using a Matlab                      program.            2 
  • 3. Acknowledgements First of all, I would like to thank my advisor Dr. Maria Teresa Herd  for giving me the opportunity to do research in her lab. Without her continuous  guidance and support, this thesis would not have been possible. I would also like to  thank her for encouraging and helping me at times when I lost faith in my abilities  over the past year. She has truly been a great role model and mentor.   I would also like to thank the Physics Department for fostering my love for                            physics and passion for research. I would like to thank all the professors: Mark                            Peterson, Juan Burciaga, Neal Abraham, Spencer Smith, Alexi Arango and Kathy                      Aidala for their enthusiasm and inspirational way of teaching.   I am grateful to my lab mates Madeleina, Audrey and Colbie. Thank you,                          Audrey and Madeleina for allowing me to observe your experiments, and helping                        me learn the fundamentals of handling lab equipments, acquiring and analyzing                      data. Thank you Colbie for your assistance during the first half of my project,                            when we spent several hours at a stretch pouring over books looking for missing                            functions for the RF voltage trace model.  Next, I would like to thank my brother for always pushing me to become a                              better physicist than him. I would also like to thank my boyfriend for his constant                              encouragement, and for always being available to listen to me whine, complain                        and cry over the phone whether he is at home or at an observing run at Mauna Kea                                    or Coonabarbaran.   Finally, I have to thank my mom for giving me full freedom to explore my                              passion and pursue my dreams, no matter how crazy they are. Without her                          unconditional support, guidance, and love, I would not have come this far.                  Once again, a big thank you to each and everyone!    3 
  • 4.        Contents   1 Introduction to Pulse­Echo Ultrasound Imaging                       7                           1.1 Making Ultrasound B­mode Images …………….. 7                           1.2 Speckles in Ultrasound Images ….………………. 9                             1.3 Importance of Modeling Speckle ………………... 10                             1.4 Current Theoretical Models for Speckles ………... 11                                         1.4.1 Discrete Scatterer Model ……………... 12                                         1.4.2 Inhomogeneous Continuum                                                  Model …………………………………. 12                           1.5 Modeling Backscattered RF Trace                                  of Human Tissue ………………………………… 13                           1.6 Complex Plane Representation of                                  Ultrasound Point Spread Function ………………. 14  2 Theory of Wave Propagation: Physics of Ultrasound                      15                           2.1 Classical Theory of Linear Acoustics …………….. 15                                         2.1.1 Equation of State ………………………. 16                                         2.1.2 Equation of Continuity ………………… 18                                         2.1.3 Euler’s Equation ……………………….. 18                           2.2 The Wave Equation ……………………………….. 20                           2.3 Wave Equation in the Frequency                                  Domain ……………………………………………. 21                                        2.3.1 The Incident Pressure Field ……………... 23                                          2.3.2 The Scattered Pressure Field …………… 25                                         2.3.3 The Force on the Receiving Crystal ……. 26                                         2.3.4 The RF Voltage Trace ………………….. 28  4 
  • 5. 3 Pulse­Echo Experiments Using a Single Element Transducer       31                           3.1 Making Scatterer Phantoms ……………………… 32                           3.2 Experimental Setup ………………………………. 34                           3.3 Electronic Setup ………………………………….. 35                           3.4 Aligning the Transducer …………………………. 36                             3.5 Collecting RF Pulse­Echo Signal                                    Data ……………………………………………... 39    4 Obtaining B­mode Images                                                                 42                           4.1 B­mode Image of 4.76 mm                                  Metal Beads ………………………………………. 42                           4.2 B­mode Image of 2.85 mm ­ 3.45 mm                                Glass Beads …………………………………………44                           4.3 B­mode Image of 1.0 mm ­ 1.03 mm                                Glass Beads ……………………………………….. 45      5 Conclusion and Future Work                                                            47         Appendix                                                                                             49              A Proof of Lemma 1 (Chapter 2, 2.3.3)........................................ 49             B Proof of Lemma 2 (Chapter 2, 2.3.3)........................................ 50              C MATLAB Program for B­mode Images.................................... 53               D Preliminary Modeling: Incident Pressure Field......................... 56             E Preliminary Modeling: Scattered Pressure Field........................ 59      5 
  • 7. Chapter 1  Introduction to Pulse­Echo Ultrasound  Imaging            1.1 Making Ultrasound B­mode Images   Ultrasound has been used in a variety of clinical settings, including obstetrics and                          gynecology, cardiology and cancer detection. The main advantage of ultrasound is                      that structures can be observed without using radiation. Using ultrasound medical                      imaging as a diagnostic tool is also more cost effective compared to other radiation                            free diagnostic techniques, such as the MRI. Ultrasound can also be done much                          faster than X­ray and other radiographic techniques.Ultrasound also provides                  multiple modes available for versatile imaging. For instance, M­mode ultrasound                    is extremely valuable for accurate evaluation of rapid movements due to its                        excellent temporal resolution, and is used  for tracking the motion of heart                        structures over time.   For the purposes of modeling speckles, we are interested in producing                      B­mode images. This is because B­mode images can be easily produced using the                          amplitude of reflected ultrasound signals to reflect the intensity or brightness on a                          two­dimensional gray scale. Owing to a wide gray scale available for ultrasound                    imaging, very small differences in echogenicity are possible to visualize.  The essential component of an ultrasound medical imaging system is a                      transducer. Transducers are devices that convert electrical signals to mechanical                    energy and vice versa. When electrical signals reach a transducer in the form of an                              excitation pulse, these signals are converted to sound waves that travel to the                          structure being imaged. All of the sound waves are then reflected by the structure,                            some of which eventually find their way back to the transducer and the sound                            waves are then converted to electrical signals. These electrical signals are collected                        7 
  • 8. in the form of radio frequency (RF) voltage traces that are processed to produce                            B­mode images of the structure.   The basic components of a typical spherically focused single­element                  transducer are summarized in Figure 1. The active piezoelectric element has a                        spherically curved shape to facilitate the focusing of the beam, and its thickness is                            usually half the wavelength of the ultrasound produced. It is important to note that                            the transducer generates a range of frequencies around a central frequency. This is                          in part due to the fact that the piezoelectric element continues to produce                          ultrasound for sometime even after the electric signal has been terminated.   There are a number of parameters, which are useful for understanding the                        general components of a single­element transducer. The range of frequencies                    generated by the transducer is known as the bandwidth. The quality factor of a                            transducer is a dimensionless parameter, which is calculated as the ratio of the                          central frequency to the bandwidth; the higher the quality factor, the lower the rate                            of energy loss compared to stored energy. Finally, the sensitivity of the transducer                          is defined as the ability to detect reflected ultrasound and generate electrical signal.                          The sensitivity and bandwidth of the transducer are dictated by the parameters of                          the piezoelectric material, transducer backing and electrical and acoustic                  impedance matching devices, which are needed to achieve the specific imaging                      requirements.   8 
  • 9.                         Figure 1 : Single Element Ultrasonic Transducer      There is a fundamental relationship between the properties of the generated                      ultrasound beam and the quality of the resulting image. The ability to predict beam                            characteristics is crucial for medical imaging. There are many factors, which affect                        the beam profiles of transmitted and received signals. Most important are the                        source geometry, source excitation (sinusoidal or pulsed), target response and the                      type of electronic processing applied to signals. The theoretical treatment of beam                        profiles is based on Huygen’s Principle, where each radiating source is divided                        into infinitesimal elements, each radiating hemispherical wavelets. The resulting                  acoustic fields obtained from the superposition of these wavelets obey the classical                        theory of linear acoustics described in Chapter 2.  A central aspect of quantitatively characterizing the beam profile is by                      calculating the instantaneous pressure field distribution. The function for                  calculating the instantaneous pressure field is obtained using the wave equation for                        the velocity potential (discussed in detail in Chapter 2). Determining the                      instantaneous pressure for incident and scattering field paves the  way for the                        calculation of pulse­echo sensitivity profiles based on the spatio­temporal impulse                    response of the transducer (discussed in detail in Chapter 2).              1.2 Speckles in Ultrasound Images                 A difficult and crucial challenge in ultrasound medical imaging is the necessity of   reducing the annoying “mottled” or “speckled” appearance of B­scan images.                    Speckles are an interference phenomena caused by the interaction of coherent                      waves produced by the transducer with multiple scattering structure of the tissue.                        Such interactions result in the constructive and destructive coherent summation of                      9 
  • 10. ultrasound echoes. Speckles tend to reduce the perception of small structures, and                        ultimately limit diagnostic accuracy of medical imaging systems.   Generally there is no common enhancement approach for speckle noise                    reduction. Different filtering techniques based on statistical methods have been                    implemented in an attempt to reduce speckles, such as Signal­to­Noise Ratio                      (SNR), Peak Signal­to­Noise Ratio (PSNR) and Root Mean Square Error (RMSE).                      Another important averaging technique used for speckle noise reduction is                    angular­weighted compounding, which stems from ultrasound elastography.              Elastography is an imaging technique in which local strain in human tissues are                          used to measure the axial shifts in tissue due to quasi static compression applied                            using ultrasonic transducers; all the measurements are made in terms of RF echo                          arrival times. Using elastography measurements, angular­weighted factors are                derived from the relationship between axial and lateral strain components of strain                        estimated along angular insonification directions. Experimental results using a                  uniformly elastic tissue mimicking phantoms demonstrated the improvement in                  signal­to­noise ratio obtained using angular­weighted compounding (Techavipoo              et. al. 2004). However, removing speckle noise from original ultrasound images                      still remains a challenge in image processing.     1.3 Importance of Modeling Speckles  Ultrasound imaging is a useful diagnostic tool, especially for biological tissues and                        subcellular structures. However, ultrasound medical imaging is limited by two                    important factors: spatial resolution and fluctuations in echo amplitude, known as                      speckle noise. Speckles are considered one of the major obstacles in image                        analysis, which include detection, classification and segmentation problems in                  coherent imaging systems.   Since speckle appears in all conventional ultrasound images, numerous                  investigations have been carried out to understand the nature of speckles and what                          10 
  • 11. causes them. Establishing a model for ultrasound imaging would require                    developing a mathematical description of the acoustical scattering properties of                    tissues; however, this is often hindered due to the lack of knowledge concerning                          cellular structures. Kossoff et al. (1976), attempted to relate speckles to the                        architectural structure of tissues, while Linzer et al. (1979) had tried to relate                          speckle patterns to the different diseased states of soft tissue. But in both studies it                              was found that the speckle pattern bore little resemblance to the actual acoustical                          tissue microstructure. Rather than directly using speckle patterns to decipher                    details of complex cellular structures, a more constructive approach would be to                        first predict the speckle pattern of an arbitrary scattering medium for a given                          transducer geometry. Therefore developing theoretical models to describe the                  scattering medium will help us understand the relationship between the imaging                      system and the speckles that arise in the final image.    Another reason for modelling speckles is that they are the primary                      mechanism for elastography imaging, an emerging and exciting use of ultrasound                      imaging technique.  In elastography, it is possible to trace localized elastic                      characteristics because local movement is estimated using speckle tracking (Kim et                      al. 2011). The application of speckles in elastography confirms that speckles are                        not always undesirable. Today, ultrasound elastography has developed into an                    image modality suitable for detection and diagnosis of cancers in the breast,                        kidneys and thyroid.    1.4 Current Theoretical Models for Speckles  Theoretical models used to describe the scattering medium can be essentially                      divided into two categories: the discrete scatterer model and the inhomogeneous                      continuum model. For the discrete scatterer model, the scattering medium is                      depicted as a collection of points, spheres or cylinders embedded in a                        homogeneous medium. On the other hand, the inhomogeneous scattering model                    11 
  • 12. describes the scattering medium to be one where acoustical inhomogeneities, such                      as density, compressibility and velocity, vary in a continuous manner throughout                      the medium. Such models of scattering medium are then used in conjunction with                          simulation models to investigate both theoretically and experimentally the effect of                      speckles on ultrasound images.     1.4.1 Discrete Scatterer Model  Foster et al. (1983), in their study to investigate the effect of transducer geometry                            and the position of the point spread function on speckle formation, developed a                          mathematical model to simulate ultrasound B­scan images showing speckles. Prior                    to the experimental approach adopted by Foster et al., simulation models were                        based on two dimensional scattering medium. The dimension perpendicular to the                      image plane was ignored for the two dimensional models; thus the number of                          scatterers in a given distance off­axis remained constant in those models. Previous                        models also excluded transducer geometry and the position dependence of the                      point spread function. Foster et al. were able to recognize the limitations of the                            two­dimensional models, and thus incorporated the three­dimensional              characteristics of both the transducer and the scattering medium. The scatterer                      phantoms were developed using the discrete scatterer model, and their studies                      showed good agreement between theoretical and experimental results.     1.4.2 Inhomogeneous Continuum Model  In recent years, other approaches were implemented for modeling speckles. One                      such approach relies on the surface roughness and the number of elemental                        scatterers in the tissue surface. In their study to derive stochastic models for                          speckle noise, Daba et al. (2009) outlined two distinct categories for describing                        speckle distribution in conjunction with the roughness of scattering surface: fully                      12 
  • 13. developed, where the surface is comprised of infinite number of scatterers making                        the surface very rough, and partially developed, where the surface has finite                        number of scatterers. Daba et al. derived the statistical distribution for the partially                          developed model using Poisson point process, where each scatterer is assumed to                        have statistically independent random variables, which for the purposes of the                      theory were amplitude and phase of the backscattered field.  In order to obtain ultrasound images of a particular organ or tissue structure,                          a pulse is emitted into the body and is scattered and reflected by density and                              propagation velocity perturbations. Therefore developing a model for the received                    pulse­echo pressure field of a transducer can also help to predict the speckle                          pattern from a scattering medium. Using this approach, Jensen (1990) derived an                        inhomogeneous wave equation to describe the propagation and scattering of                    ultrasound by a collection of point scatterers in a homogeneous medium. The                        solution to the wave equation was then combined with the field of a typical                            transducer to develop a model for the pulse­echo pressure field. This approach                        proved to be versatile because it enabled the derivation of scattered field for                          different transducer geometries, whose incident pressure fields are known.  1.5 Modeling Backscattered RF Trace of Human  Tissue     Instead of modeling arbitrary scattering medium, like those described in Sections                      1.4, Gore and Leeman (1976) developed a realistic approximation of human tissue                        inhomogeneities. Their approach involved describing wave propagation in                inhomogeneous media, and then showing that the waves are scattered by                      fluctuations in density and compressibility. The model is then used to calculate the                          precise form of the backscattered field from tissues, which coincides with actual                        signals recorded by diagnostic pulse­echo equipment.    From their studies it was found that the usefulness of characterizing human                        tissues is dependent on the nature of the structure of the tissue itself. They                            13 
  • 14. concluded that although two different tissues may have different compressibility                    coefficient and density, the backscattered echo sequences are rotationally invariant                    to first order. For less restricted systems, without an isotropic distribution of                        scatterers, the rotational invariance does not hold; this indicated that there is an                          angular dependence of the scattered field—this further confirms why                  angular­weighted compounding is effective for speckle reduction. Therefore, the                  studies concluded that in order to understand the diagnostic capability of grey scale                          B­scan images, a knowledge of the scattering processes within the tissues and                        quantitative image analysis in conjunction with pattern recognition is important.      1.6 Complex Plane Representation of Ultrasound            Point Spread Function    Conventional ultrasound imaging probes a medium with high frequency                  band­limited acoustic waves, and detects echoes scattered by inhomogeneities                  within the medium. A spherical single element transducer placed in the medium is                          used for both generation of pulses and reception of these echoes. The site of the                              transducer responsible for both transmission and reception is known as the active                        aperture, which is a set of piezoelectric crystals. The scattered echoes are summed                          up coherently to yield a single RF voltage trace.  In their study, Ng et al. (2005) solve the equation that governs wave                          propagation in an inhomogeneous medium to show that the RF ultrasound signal                        can be expressed as the result of filtering the tissue reflectivity by a point spread                              function and the total pressure field. Ng et al. (2005) further extended their                          analysis by establishing a link between the RF signal and the representation of                          point scatterers as vectors with random phase in the complex plane. Their analysis                          led to an insight into the useful techniques for simulation and analysis of speckle                            statistics in the complex plane, in addition to developing a normalized covariance                        of the RF signal in terms of the complex envelope of the point spread function.  14 
  • 15. The mathematical techniques implemented by Ng et. al. (2005) is relevant to                        the modeling speckles because we in the future we aim to carry out impulse                            response and instantaneous pressure field calculations in the frequency domain to                      obtain the RF output voltage response of the transducer.                                          15 
  • 16. Chapter 2  Theory of Wave Propagation:   Physics of Ultrasound     Acoustic refers to the generation, transmission and reception of energy in the form                          of vibrational waves in matter. Acoustic waves are the organized vibrations of the                          molecules or atoms of the medium that supports the propagation of these waves.                          Acoustic plane waves possess a particular frequency depending on the source of                        the waves, and waves whose frequency is beyond the audible range of 20 KHz are                              known as ultrasound waves.      2.1 Classical Theory of Linear Acoustics  In general plane acoustic waves through an acoustic medium is a combination of                          transverse and longitudinal motion, although longitudinal motion seems to                  dominate. The characteristic properties of plane waves is that each acoustic                      variable, such as particle displacement, density and pressure, has constant                    amplitude in any given plane perpendicular to the direction of propagation. An                        expression describing the motion of a particle in a medium in the presence of                            acoustic plane waves is given by:      where is the equilibrium displacement of the particle, is the particle                          displacement from the equilibrium position, is the angular frequency and φ is                          the arbitrary phase factor.   The above equation is one of the many solution of the wave equation. In the                              following sections we will derive the wave equation to relate the changes in                          16 
  • 17. acoustic variables to one another. To derive a wave equation, we utilize the                          following:  ● Equation of State, which relates pressure to density.  ● Equation of Continuity, which incorporates conservation of mass.  ● Euler’s Equation, which is derived from Newton’s Second Law.    2.1.1 Equation of State:  The relationships between pressure and density changes in a gas can also be                          applied to sound traveling through gases, such as air. In acoustics the equation of                            state for an ideal gas is given by:         where γ is a gas­dependent constant, and T k is the temperature in Kelvin. However,                            for most fluids and tissues, the adiabatic equation of state is either unknown or                            highly complex. Thus, this relationship is determined experimentally, and verified                    mathematically using the following Taylor Series expansion:      where the partial derivatives are constants determined for adiabatic compression                    and expansion of the medium about its equilibrium density,                  ,and  .    2.1.2 Equation of Continuity  In order to relate the motion of the fluid to its compression or dilation, we require a                                  functional relation between the particle velocity and the instantaneous density. As                      the sound waves propagate, mass has to be conserved. Thus, the extra mass that is                              carried into a volume due to the propagating wave represents an increase in                          17 
  • 18. density. We will consider the rate of increase of mass in a fixed (imaginary)                            volume V, with the net amount of mass that flows into , per unit time. In other                                  words, if mass goes into then, because mass is conserved, the amount of mass                            inside must be going up. The total mass in is given by the volume                                integral  . The rate of change of mass in   can therefore be written as:    The surface bounding is . The net flow of mass out through a small part of the                                  surface at point is the dot product where is the outward unit                            perpendicular to   . The net mass flowing in to  is therefore:    To understand better where the come from, consider Figure 2. When the                          angle between the beam and the normal to the aperture (call it ) is zero, all the                                  beam gets through, but as α increases, less and less gets through until, when they                              are at right­angles none gets through. In general, the amount getting through is   .                                                        Figure 2:  Vector dot product        18 
  • 19. The divergence theorem, sometimes called Gauss’ divergence theorem, equates the                    outward flux of a vector field over a closed surface with the volume integral of the                                divergence of the vector field over the region enclosed by the surface:     where is a vector field. This can be used to convert Equation (6) to the                                following form:    Equating Equations (7) and (4), we get:    As this equation is zero for any arbitrary volume , the integrand must also be                              zero hence:       2.1.3 Euler’s Equation:  Let us consider a medium of density and a particle with an incremental volume,                              whose size is smaller than the wavelength of the ultrasound. The length of this                            small volume in the longitudinal direction is , where is the area of the cross                                section perpendicular to the longitudinal axis and the mass of the incremental                        volume is given by        . Along the longitudinal direction, there exists an                excess pressure denoted by        . In conjunction with the pressure field, and                coupled to its longitudinal velocity field            represents the motion of the particle.  19 
  • 20. If the particle moves with the velocity              in the presence of an acoustic field, a                minute force acts on the particle due to the change in pressure. Newton’s force                            equation can be applied to the volume element, where the net force on the volume                              can be written in partial differential form:    where  . Notice the extra term in the force equation; this term takes                          into account the fact that acceleration is not constant within the fluid for the                            infinitesimal displacement traveled in a time . Thus acceleration is a                        function of both time and displacement.    Now  , Equation (11) becomes:    Since pressure is force per unit area and                :    Taking the limit as        , Equation (12):    Rearranging Equation (13), we get Euler’s Equation in one dimension:      Following from Equation (14), Euler’s Equation in three dimensions is:  20 
  • 21.   2.2 The Wave Equation    The wave equation in an ideal fluid can be derived from hydrodynamics and the                            adiabatic relation between pressure and density. The equation for conservation of                      mass, Euler’s equation (Newton’s second law), and the adiabatic equation of state                        are respectively:       and for convenience, we will define the quantity:    where  c  will turn out to be the speed of sound in an ideal fluid. In the above                                    equations, is the density, the particle velocity, the pressure. The ambient                            quantities of the quiescent (time independent) medium are identified by the subscript                        0. We use small perturbations for the pressure and density, and note that is also a                                  small quantity; that is, the particle velocity which results from density and pressure                          perturbations is much smaller than the speed of sound.   Retaining higher­order terms in Equation ( 3 ) yields a nonlinear wave equation. The                        nonlinear effects we include are contained in the quadratic density term in the                          equation of state, ( 3 ). We first multiply Equation ( 15 ) by and take its divergence;                              21 
  • 22. next, we take the partial derivative of ( 9) with respect to time. Substituting one into                              the other yields  :      Here, the indices  i, j = 1, 2, 3   indicate  x, y  and  z ­components, respectively. Tensor  notation is used; repeated indices signify a summation (e.g.,  ).  The first term on the right­hand side of ( 16)  can be rewritten using ( 3)  and (15) as:     2.3 Wave Equation in the Frequency Domain    Conventional ultrasound imaging interrogates a medium with high frequency, band                    limited waves and detects echoes scattered by inhomogeneities within the medium,                      where scatterers represent such inhomogeneities. A single­element transducer                immersed in the medium both generates ultrasound waves and receives the echoes.                        The single element transducer consists of two types of piezoelectric crystals, the                        transmitting and the receiving crystals. The transmitting crystal is excited coherently                      to produce a focused beam, whereas the receiving crystal detects the scattered echoes,                          which are then summed up coherently to yield an RF voltage trace. At each                            transmission, the emitted wave propagating through the medium gives rise to an                        incident pressure field, and the scattered wave gives rise to a scattered pressure field.                            Multiple RF traces are formed by moving the centre of the crystals and repeating the                              process; by lining up these individual traces next to each other in an image space, an                                RF image is formed by summing the RF traces resulting from the scattered pressure                            field over the piezoelectric surface and filtering the sum by the electromechanical                        impulse response of the crystals.  22 
  • 23. With the physical description of the transducer mechanism in mind, we will consider                          the mathematical expressions for the incident and scattered pressure fields in the                        following sections. The wave equation (17) that we derived in the time domain can                            also be expressed in the frequency domain as follows:    where          and are density and compressibility terms respectively. The presence of                      scatterers in the medium may be modelled by adding spatially dependent variables                        and to the density and compressibility terms. The term is the scattering                            operator defined as follows:      and the scattering terms  and  are:      Since Equation (18) is linear partial differential equation (PDE) we can write its                          general solution as the sum of the solution to the corresponding homogeneous                        equation (i.e. with the RHS set to zero) and any particular solution. Denoting                          as the solution to the homogeneous equation as and                  the particular solution      as , we can therefore write the total field as:    23 
  • 24.   In order to assign a physical interpretation to                when we set RHS of Equation            (18) to zero we have effectively set              . Thus we see that          then is the pressure field that develops in the absence of any scatterers, which by                              definition is the incident pressure field. We also know that the scattered pressure field                            must obey Equation (18), and so we assign our particular solution to be the scattered                              pressure field    . With these physical interpretations in mind, we see that the                      total pressure field is indeed the sum of the incident and scattered pressure fields.    2.3.1 The Incident Pressure Field  In order to obtain an expression for the incident pressure field, we need to consider the                                mechanics and geometry of the transducer. The generalized three­dimensional                  coordinate system is shown in Figure 3.       Figure 3:  Coordinate system for describing scattering in an inhomogeneous medium      where represents the area over the transmitter crystal, is the location of the                              center of , is an arbitrary point on , is the volume within which the                                24 
  • 25. scatterers being considered are contained and is an arbitrary point in . The                          surface may be considered to consist of infinitesimally small area element,                        each of which behaves as a simple point source. The Huygen­Fresnel principle states                          that each area element contributes a spherically expanding wave to the incident                        pressure field. The incident pressure field ca then be obtained by summing the                          spherical wave contribution from each area element.  If we assume that the radius of curvature of is large enough, then is considered                                  to be effectively flat. As a result, the incident pressure field can be expressed in terms                                of the Rayleigh integral:    where  is the temporal Fourier transform of the normal velocity on the                      transmitter crystal’s surface; this normal velocity is not constant and varies from point                          to point on . We note that term                corresponds to normal acceleration in          the time domain, since the factor            corresponds to time differentiation. We have            written  on the left­hand side of Equation (24) to explicitly indicate the dependence                        of the incident pressure field on the location of the center of  .  Although we have not shown the proof for Equation (24), we can intuitively see that it                                is indeed the Huygen­Fresnel principle expressed mathematically: the integral on RHS                      describes the summation of complex­valued spherically expanding waves (represented                  by the complex exponential term), each weighted by the normal acceleration at its                          source and decaying in amplitude with increasing distance from its source.  For a typical transducer, we can consider the nominal normal velocity weighted by a                            spatially variable term      to account for apodization, a filtering technique that                removes discontinuities at the beginning and end of the sampled time record. We can                            also model focusing by considering the normal velocity at each to be delayed by                              .   Substituting  into Equation (24), we obtain:    25 
  • 26.   For convenience we define a new quantity that we define to be the spatial                              transfer function:    Equation (25) can then expressed more compactly as:    This compact expression allows us to view the incident pressure field as the result of                              temporally filtering the nominal normal velocity            by the spatial transfer        function  . The spatial transfer function          incorporates  the effects of apodization, and accounts for the spatial distribution of the incident                          pressure field.     2.3.2 The Scattered Pressure Field    In order to find an expression for the scattered pressure field, we will consider the                              Green’s function method. In our case, we consider spherical waves scattered from a                          volume to be propagating into an effectively unbounded medium, in which case the                          Green’s function takes the following form :    The particular solution to Equation (23) is then the product of the RHS and the                                Green’s function integrated over the volume. Furthermore if we define and to be                              26 
  • 27. zero outside of , then we can perform the integration over all of the three dimensional                                space, and the scattered pressure field can be expressed as the convolution integral:      Since we consider weak scattering, we assume that .                  in Equation (23) becomes negligible, and            .  Rewriting Equation (29) with        substituted by    yields:    This approximation is referred to as the first Born approximation, and the equation (30)                            states that the scattered pressure field is, to a first approximation, the spherically                          expanding wave convolved with the scattering term. If we regard scatterers to be                          idealised points in , then it is the equivalent of saying that each point scatterers                              contribute spherically expanding wave independent of each other. Thus in making the                        Born approximation, it is assumed that multiple scattering (i.e. waves scattered off a                          particle that are then scattered off other particles ) is negligible.  By substituting the expression for          , we can rewrite the scattering            pressure field entirely in terms of the transducer’s characteristics and the scattering                        operator:    2.3.3 The Force on the Receiving Crystal  Recall that the received RF voltage trace is obtained by summing the scattered pressure                            field over the area of the receiving crystal, and filtering the sum using the                            electromechanical impulse response of the transducer. In this section we will derive the                          27 
  • 28. summation formula for the scattered pressure field over the receiving crystal, and the                          formula will be denoted by          , which is essentially the force on the receiving                  crystal.  Before we develop an expression for            , we will introduce two lemmas. These              lemmas are essential in the derivation of the formula for the force on the receiving                              crystal.  Lemma 1 (Refer to Appendix A for proof): For any vector valued function A(x) and                              scalar function b(x), if A(x) is zero outside some volume V’ then:     Lemma 2 (Refer to Appendix B for proof): At locations that are far away from the                                  receiving crystal:    Returning now to      , if we assume the same apodization and focusing at                    reception as at transmission, then:    where,     We recognize that the integral is equal to                  ,  and so:    From the definition of the scattering operator in Equation (20), we can rewrite Equation                            (34) as follows:  28 
  • 29.   To simplify Equation (35), we note that has finite limit, which allows us to use                                Lemma 1  to simplify the integral on RHS of Equation (35):  At sufficiently large distances away from the receiving crystal, we can use  Lemma 2 to                              further simplify the force equation:      2.3.4 The RF Voltage Trace    In this section, we will discuss the conversion of the force on the receiving crystal into                                a voltage RF trace. If we model the electromechanical transfer function of the                          transducer to be      and the voltage trace to be            , we have:      Substituting in the expression for          , we have:    For convenience, we can group together the characteristics of the medium together, and                          the electromechanical properties of the transducer together. We adopt definitions similar                      to those in Equation (27):  29 
  • 30.       where  represents convolution in the space domain.  In keeping with the terminology introduced in Equation (27), we refer to the quantities                            and  respectively as the pulse­echo wavelet and the tissue reflectivity or                    scatterer field.  We can also express the voltage trace in the time domain as follows:        where  and  represent convolution in the space and time domain respectively.    If we regard the quantity as the input signal and the quantity as output                                signal, then Equations (40) and (43) indicate that the imaging system has a                          spatiotemporal transfer function and the impulse response function                  . Thus, the definition of the transfer function this way neatly                      distinguishes between the electromechanical characteristics of the transducer (represented                  30 
  • 31. by the pulse­echo wavelet) and the geometry of the transducer, represented by                        .                                                    31 
  • 32. Chapter 3  Pulse­Echo Experiments Using a Single  Element Transducer    The main goal of this project is to obtain B­mode images of scatterers small enough                              to cause speckles, and then compare experimental results with theoretical                    predictions from a mathematical model that would simulate B­mode images of                      those scatterers showing speckles. Developing this model will rely on much of the                          theoretical background described in section 2.3 of Chapter 2, which incorporates                      transducer geometry and its corresponding pressure­field calculations for predicting                  the RF voltage trace for that transducer geometry. However before an experimental                        prototype and its corresponding theoretical model can be developed for describing                      speckles, the first step would be to model and gather experimental data for discrete                            scatterers, which are larger than speckles and the scatterers that cause speckles. This                          chapter describes the experimental procedures for obtaining B­mode images of                    discrete scatterers.   The process of obtaining B­mode images for scatterers within a                    homogeneous medium is very challenging. This is because measurements and data                      collection are very sensitive to external factors, for example transducer alignment                      and sample material. For instance, a comparison of backscatter coefficient                    measurements performed at eight different ultrasound physics laboratories found                  variations in the final results of almost two orders of magnitude (Wear, et. al, 2005).                              However, some of this variation also arises due to differences in experimental                        techniques.   Techniques for measuring acoustic properties of phantoms may vary in                    several ways, for instance bandwidth of the technique and the type of transducer                          used. For our experiments, we have used a broadband, single­element focused 5                        MHz transducer. The advantage of using a broadband transducer lies in the fact that                            32 
  • 33. it allows for simultaneous RF data acquisition over a range of frequencies; this also                            means that the transducer can be excited with a short pulse rather than a                            quasi­continuous wave excitation. Also using a single­element transducer instead of                    an array of transducers reduces the complexity of data analysis of the signals                          received.       3.1 Making Scatterer Phantoms:    Scatterer phantoms are a uniform distribution of symmetric structures in a                      homogeneous medium, based on the discrete scatterer model mentioned in Chapter                      1. Scatterer phantoms were made by embedding beads in an agarose base, enclosed                          by plastic petri dishes of about 3.6 cm in diameter and 0.8 cm in height.   Using the speed of sound in agarose at 24 °C, 1500 m/s, and the peak                              frequency of the transducer, which for the purposes of our experiment was 5 MHz,                            the minimum size of the scatterers were calculated:    The minimum size of scatterers were calculated assuming Rayleigh scattering takes                      place, where scatterers are assumed to be smaller than the wavelength of ultrasound                          waves. This value is significant for the following reasons: first, it provides an                          estimate of the size of scatterers, which may cause speckles (these scatterers are                          expected to be of the order of several hundred microns); further more it enables us                              to determine how large discrete scatterers should be compared to clusters of                        indistinguishable scatterers that cause speckles.    Beads of the following size ranges (from largest to smallest) were selected:                        4.76 mm, 2.85 mm ­ 3.45 mm and 1.0 mm ­ 1.3mm. Each petri dish was first filled                                    with a relatively thin layer of liquid agarose. This layer was allowed to congeal, and                              then a second layer of liquid agarose was poured on top of the first layer. As soon                                  33 
  • 34. as the second layer was poured, the scatterers were embedded into this layer. The                            reason why scatterers were not embedded in the first agarose layer is because when                            scatterers are offset from the base of the petri dish, it is much easier to differentiate                                RF signals of the base from those of the scatterers when the RF data is converted to                                  B­mode images (since the scatterers are quite small, RF signals from the base have                            a much higher intensity, and appear “brighter” in the grey scale B­mode image).  Figure 4 shows the scatterer phantoms. The beads were placed in a                        symmetric pattern in order to obtain an approximate value for the spacing between                          each bead, and knowing this spacing in conjunction with scatterer size enables                        better interpretation of B­mode images. The 4.76 mm beads and the 2.85 mm ­ 3.45                              mm beads were arranged in a rectangular array using a tweezer, with an average                            spacing of about 6 mm and 3mm respectively. Since the 1.0 mm ­ 1.03 mm beads                                were too small to be arranged linearly, they were arranged radially outward from                          the center using the tip of a thin metal strip, and arranged radially with respect to a                                  2.85 mm ­ 3.45 mm glass bead placed at the center of the agar base; the average                                  spacing was about 2 mm.                          (a)                                                (b)                                          (c)    Figure 4:  Scatterer Phantoms (a) Metal beads: 4.76 mm, spacing between beads about   6 mm, (b) 2.85 mm ­ 3.45 mm glass beads, spacing between beads about 3 mm,   (c) 1.0 mm ­ 1.03 mm glass beads, spacing between beads about 2 mm.              34 
  • 35. 3.2 Experimental Setup:    As with all acoustic property measurements, the results of the pulse­echo                      measurement can depend on temperature. Therefore, every attempt should be made                      to achieve a uniform temperature in your sample, which is typically room                        temperature (22.0 °C). For typical size phantom samples, the samples should be in                          the water tank at this temperature for 1 to 2 hours prior to starting measurements.                              This should allow enough time for the sample to uniformly be at 24 °C. For our                                experiments, we used a 5 MHz focused transducer with a usable bandwidth of                          about 60% of its central frequency (from about 3.5 MHz to 6.5 MHz).  The initial setup of the experiments are illustrated in Figure 5:                               Figure 5:  Experimental setup for pulse­echo experiment    The stage of the x­y­z translator is held in position with screws, such that the                              transducer will face downwards when mounted onto the stage. After the stage is                          securely stationed, the transducer is carefully mounted on the stage and held firmly                          in position by tightening a pair of screws.    For each phantom, the top of the petri dish is taken off and then the bottom                                surface is securely attached to the base of a large, glass container using double                            35 
  • 36. sided tape. The phantoms should be fixed rather firmly to prevent them from being                            displaced as they are being submerged in water. The container is then filled with                            deionized water until the meniscus of the water level above the phantoms is about                            5 cm, which is approximately equal to the focal length of the transducer (52.2                            mm). The phantoms should be roughly at the focal length so that the distance                            between each phantom and the transducer can be later adjusted to observe                        scatterers at the correct time delay, specifically at the focus.      3.3 Electronic Setup:   The pulse­echo measurements require the use of an oscilloscope (Tektronix TDS                      3014C Digital Phosphor Oscilloscope with 1.25 GS/s digitization), translator                  motor for positioning the transducer, and pulse­receiver (Olympus                Panametrics­NDT 5800). A pulse­receiver is a device that does just as its name                          suggest: it sends out a voltage spike or pulse, and then turns itself into a broadband                                receiver amplifier. The transducer is connected to the pulse­receiver using a BNC                        cable, where the transducer is attached to the mounting bracket of the male UHF                            adaptor.    In order to provide a trigger to the oscilloscope, a cable is connected from the                              "Ext Trig/+Sync" jack of the pulse­receiver to the "Ext" jack of the oscilloscope to                            provide external trigger to the oscilloscope. Then "Gated RF" output of the                        pulse­receiver is connected to channel 1 of the oscilloscope. In order to provide                          some low­pass filtering, a low­pass filter is then inserted between the Gated RF                          jack and the oscilloscope in order to ensure that signals associated with the central                            frequency of the transducer, and its bandwidth are detected. Note that the filters                          have an input and an output side. A schematic of the setup is shown in Figure 6.  36 
  • 37.          Figure 6:  Transducer, Oscilloscope and Pulse­Receiver connections    The mode of the pulse­receiver is set to "Pulse­Echo". The settings for                        "Gain", "Attenuation", and "Energy" are determined, and for our experiments they                      are usually set to 20 dB, 0, 50 J respectively. After the pulse/receiver is turned                              on, the gate settings and oscilloscope time­voltage divisions are adjusted to locate                        the high amplitude echo signals from the front and back surfaces of the sample.     3.4 Aligning the Transducer:   The front surface of the sample defines one geometric plane in space. The                          combination of the x and y­axes of the transducer translation system define another                          geometric plane. It is important that the surface of the sample be as close as                              possible to parallel to the plane of motion of the transducer. If this is not the case,                                  37 
  • 38. then the distance the beam travels in water will vary as the transducer moves to                              obtain independent power spectrum estimates. Unfortunately, there is no automatic                    alignment system with this setup. However, you can use the computer to determine                          if the two planes are parallel.    The motor of the x­y­z translator is controlled using a program called                        COSMOS 3.6.1. The program has two different settings, which control the motors:                        the “Quick Move­Single Axis” and “Virtual Jog”. The “Quick Move” setting is                        most useful for carrying out raster scans because it moves the motor a specific                            distance from a point of reference, defined by the step­size associated with each                          click. However, the “Virtual Jog” setting is used for fine tuning the position of the                              transducer; it is useful for centering the transducer, and locating the edge of the                            sample.   The translator motors are labelled 1 through 3, where 1 corresponds to the                          y­axis, 2 corresponds to the x­axis and 3 corresponds to the z­axis. In order to                              lower the transducer into the glass container, the “Quick Move” setting is used to                            control motor 3. The transducer is vertically lowered until it is sufficiently below                          the surface of the water at an appropriate distance from the phantom, and this is                              determined using the time delay. For the time delay calculation the variables (in                          s)  and   are considered, where:    is the speed of sound in the agar base of the phantom and water,.                              For our calculations we set the offset to zero, assuming that sound waves are not                              significantly reflected from the top surface of the sample. Since all the RF data is                              collected in a window of 10 s, the “window” term in the expression for is set to                                    10 and multiplied by 0.000001 to convert into microseconds. After obtaining the                          38 
  • 39. values of and , and given that the sample is roughly located at the focal length                                  of the transducer, the time delay  is calculated as follows:    where  , is the speed of sound in water. For our experiments,                      time delay was calculated to be 64.8 s. On the oscilloscope, the signal appears as                              an envelope, which outlines the variation in amplitude of the scattered signal                        against a time scale. Usually, the first high amplitude signals correspond to the                          interface between the surrounding water and top surface of the phantom. The next                          set of signals are recognized as scatterers distributed in the agar phantom because                          the signal amplitude is much smaller than that of the interface, and each signal                            corresponds to a scatterer. However, the amplitude of the scatterer signals is                        largely dependent on the size of the scatterers. The last set of signals correspond to                              the base of the phantom petri dish, and look very similar to the initial set of signals                                  due to the interface. Next, the voltage time scale is adjusted to locate the signal                              from the top surface of the phantom. After the top surface is located, the time delay                                of the oscilloscope is set to 64.8 s. Next, in order to find the correct distance                                between the transducer and phantom, motor 3 is used to further adjust the vertical                            height of the transducer until the signal from the top surface is at the origin of the                                  voltage­time axis. Once the appropriate distance is determined, the time delay is                        offset by 5 s to 69.8 s so that the signal inside the phantom fills the oscilloscope                                  screen, which we use to window our RF signal.  Before centering the transducer, the axis of the stage is manually adjusted                        using positioning knobs on the translation mount; this ensures that the transducer is                          parallel to the surface of the sample. Once the transducer’s vertical orientation is                          adjusted, the “Quick Move” setting of the program is used to move motors 1 and 2                                to align the transducer to the center of the phantom.   39 
  • 40. Once we have centered the transducer, we then located the edge of the                          sample in the x and y direction by observing changes in the signal amplitude. The                              “Quick Move” setting is used to move motors 1 and 2 slowly along the x and y                                  direction. The edges usually appear as a sub­envelope of signals, whose amplitude                        increases and then decreases rapidly. This sub­envelope noticeably diminishes as                    we move just off the edge of the sample until there is no signal at all. Figure 7                                    shows an RF data at time delay for 4.76 mm scatterers.                        Figure 7:  RF data at time delay for 4.76 mm scatterers    3.5 Collecting RF Pulse­Echo Signal Data:  After the edge is located, the next step would be to figure out exactly how many                                steps away from the center, and this is done using the “Virtual Jog” setting.                            “Virtual Jog” uses step sizes, which are defined as the smallest move of the                            motorized x­y­z stage. For instance, when the edge is located in the x direction,                            motor 2 is moved away from the edge using a step size of 330 (any step size less                                    than or equal to 400 should be reasonable). As the transducer is moved towards the                              40 
  • 41. edge of the phantom, it is important to keep track of the number of clicks it takes                                  for the edge signals to appear and eventually diminish again. The same procedure                          is repeated in the y direction using motor 1.  In order to obtain RF voltage trace data, a raster scan is performed. But                            before the scan can be performed it is important to know the beam width. The                              beam width of the sound waves is calculated, assuming limited diffraction through                        a circular aperture for the concave transducer surface:      where is the beamwidth, is the focal length (0.0522 m), is the diameter                                of the transducer element (0.5 inches) is the speed of sound in the phantom and                                 is the frequency of the transducer (5 MHz).  For the 5 MHz focused transducer, the beam width is 0.015 mm. The beam                              width is converted to motor steps, and the corresponding value (238 motor steps) is                            set as the step size for the raster scan in order to obtain independent beam lines.                                Figure 8 shows the rectangular pattern of signal reception, where the transducer is                          moved in a square with five steps in the x and y direction. When the raster scan is                                    complete, all the data acquired from the oscilloscope is saved in a USB and loaded                              into a MATLAB program, the raw data is the RF voltage trace. The program is                              also designed to produce B­mode images of the horizontal or vertical                      cross­sections of the phantoms by mapping the vector arrays (associated with the                        RF data) on two­dimensional space domain, that is the x­y plane in our case. Refer                              to Appendix C for a detailed description of how the program works.   41 
  • 42.                                     Figure 8:  Raster scan pattern. The circle represents the top plane of                                   the sample. The raster scans were done in a square pattern where each                                     side was 1190 motor steps (0.075 mm) in length.   domain.     Thus, the x and y axis of the final grayscale B­mode image indicate distances                            within the phantom, where the y­axis shows how deep within the phantom the                          scatterer is located relative to the top surface of the phantom.                          42 
  • 43. Chapter 4  Obtaining B­mode Images    Before we add complexity to our experiments and make phantoms consisting of                        scatterers in the Rayleigh scattering limit, it is important to verify phantom                        preparation methods and reasonably image discrete scatterers. In this chapter, we                      will look at B­mode images of the discrete scatterer phantoms discussed in Chapter                          3. Having B­mode images of discrete scatterers will not only test accuracy of                          preparation methods, but it will also enable comparison of experimental results with                        the theoretical results once we establish our own mathematical model for obtaining                        RF voltage trace and B­mode images for a single­element transducer geometry.    4.1 B­mode Image of 4.76 mm Metal Beads:                                            Figure 9:  RF data for 4.76 mm scatterers.      43 
  • 44.                                   Figure 10 : B­mode image of 4.76 mm scatterers    For the 4.76 mm metal beads, the brightest bands indicated in Figure 10 represent                            the top and bottom surface of a single scatterer. The distance between the respective                            bright bands is roughly the size of a scatterer, thus confirming that the bright bands                              arise due to sound waves striking the scatterer head on. Looking at the RF data for                                these scatterers, we are also able to identify the top and bottom surface of the                              scatterer at about 70 s and 73.5 s. Thus, B­mode imaging of 4.76 mm scatterers,                              was successful.          44 
  • 46.                    Figure 12:  B­mode image of 2.85 mm ­ 3.45 mm glass beads    Since the spacing between theses scatterers (about 3 mm) was approximately the                        same as the average size of the scatterers (about 3.15 mm), we observed a series of                                bright bands across the plane of the phantom. This is because as the scatterers were                              relatively clustered, sound waves struck the edges of most of the scatterers giving                          rise to a linear array of bright bands. However, some of the sound waves managed   to strike a single scatterer head on, and again we observe a much brighter set of                                bands as indicated in Figure 12. The distance between the brightest bands is about                            2.75 mm, which is roughly equal to the average diameter of a single scatterer.                            From Figure 11 we observe the front and back surface of the scatterer to be around                                71 s and 72.8 s respectively, and this further confirms that the 2.85 mm ­ 3.45                                mm scatterers were successfully imaged.    4.3 B­mode Image of 1.0 mm ­ 1.03 mm Glass        Beads:                                 Figure 13:  RF data for 1.0 mm ­ 1.03 mm glass beads  46 
  • 47.                             Figure 14:  B­mode image of 1.0 mm ­ 1.03 mm glass beads    Spacing between the 1.0 mm ­ 1.03 mm scatterers (about 2mm) was not too large                              compared to the average size of the scatterers (about 1.02 mm), and we again                            observe a series of relatively bright bands like those in the B­mode image for the                              2.85 mm ­ 3.45 mm scatterers. Among the series of bright bands, we observe                            brighter bands corresponding to the front and back surface of a single scatterer as                            seen in Figure 14. We estimate the distance between the bright bands to be around                              1mm, which again is roughly the same size as the diameter of a single scatterer.                              Figure 13 also indicates the front and back surface of the scatterer as seen on the RF                                  data; the front surface is located at around 71.8 s, whereas the back surface is                              around the 72.4  s mark.     47 
  • 48. Chapter 5  Conclusion and Future Work    From the results of our experiments, we can conclude that we have been able to                              successfully produce B­mode images of discrete scatterers, which are larger than                      scatterers that cause speckles. The results also indicate the phantom preparation                      method complies with the discrete scatterer model, where the scatterer phantom is                        depicted as a collection of points, spheres or cylinders embedded in a homogeneous                          medium. However, the results presented in this thesis are the first steps towards                          achieving the main goal of the experiment: producing B­mode images of scatterers                        in the Rayleigh scattering limit.   Now that we have a mathematical model for predicting the RF voltage trace,                          as discussed in Chapter 2, the next step would be to extend that model to describe                                scatterers of various sizes (higher or lower than the Rayleigh scattering limit), and                          simulate ultrasound B­scan images of scatterer phantoms. We started some                    preliminary modeling of the incident and scattered pressure fields, and both these                        models are discussed in detail in Appendix D and E. However, these models are                            still in the early stages of development and require addition of further complexity,                          such as developing sub­functions for apodization and scattering terms, in order for                        them to be more realistic.   Thus in order to effectively model speckles, the process of writing up a                          complete program model will involve pressure­field calculations similar to those                    discussed in Chapter 2. These calculations will help model the RF voltage trace of a                              given transducer geometry (a single element focused transducer in our case), and                        then convert the RF data to B­mode images for comparison between experimental                        and theoretical results. After a working model is established, keeping in mind the                          discrete scatterer model for scatterer phantoms, the theoretical B­mode images will                      be compared with experimental results (including the ones described in Chapter 4)                        48 
  • 49. to verify the accuracy of the mathematical model for describing speckles.                      Achieving a good agreement between experimental and theoretical results will                    enable a better understanding of the nature of speckles and what causes them.                          Developing a model that accurately describes speckles will help us develop an                        ultrasonic microscope to map scatterers at high frequency locations within                    biological systems, which we are unable to detect using current ultrasound imaging                        techniques.                                              49 
  • 50.   Appendix:    A Proof of Lemma 1 (Chapter 2, 2.3.3)    We begin with the identity:     which can be verified by expanding the RHS and simplifying. Integrating both                        sides over  :    By the divergence theorem, the second integral on the RHS is equal to the surface                              integral:    where is some reference surface enclosing and            is a unit vector normal          to . Since      is zero outside , the surface integral reduces to zero and                      Equation  follows.  Q.E.D.            50 
  • 51.   B Proof of Lemma 2 (Chapter 2, 2.3.3)    We define the wave vector          , where    , in other words          is a unit vector parallel to and                is a vector also parallel to              but with a magnitude        . We can then rewrite Equation for the spatial transfer                    function as,    Taking the gradient,    We apply the condition        , which is equivalent to          since  . The above equation then becomes:      We also assume that the direction of does not vary very much over the receiving                              crystal, which allows the   term on the RHS to be factored out of the integral,  51 
  • 52. Lemma 2 follows from this immediately.  Q.E.D.  In practice,    is satisfied for virtually the entire imaged region                since the wavelength from medical ultrasonic transducers are usually very short. In                        the context of our experiments described in Chapter 3, the single­element focused                        transducer is transmitting at 5 MHz with an average speed of sound of 1490 m/s in                                agar and water. For this transducer            , which is small enough to            be considered negligible.  The condition that the direction of not vary very much over the receiving                            crystal is much stricter and is only satisfied in regions far away from the receiving                              crystal. To quantify, what is exactly meant by “far away” consider the following                          figure:    Figure A:  Coordinate system for calculating an approximation to      For a typical focused transducer, we expect the majority of the transmitted                        acoustic energy to be concentrated in a small region enclosing the axial axis, and                            we have chosen the scenario where lies along this axis. We can restate our                              requirement that can be uniform over the surface of the receiving crystal by                            requiring    to be approximately parallel to   .  This requirement is most difficult to satisfy when we consider the point on                          farthest from the center of the receiving crystal, i.e. when                    .   52 
  • 53. Our requirement that be approximately parallel to can be stated in terms of                              the dot product      ; in the case where          (as shown in      the figure above), we require . If we allow a 10% error in the                            approximation of , we effectively impose the constraint where                    . From the figure        , and so our requirement that            translates to requiring      .  What we have demonstrated in this brief discussion is that the approximation                        considered as a part of lemma 2 is satisfied well at axial depths that are greater                                than the diameter of the receiving crystal. For non­circular apertures, this diameter                        is the smallest circle within which the receiving crystal can be ascribed.                                53 
  • 55. Figure C1:  Matlab program for converting RF data to B­mode images    Figure C1 shows the Matlab program that was written to convert the RF data                            to B­mode images. The RF data is loaded on to the program using the defined                              variables for file path and name of the designated folder, where all the data is                              saved. Since the RF data file is essentially a three­dimensional array vector,                        variables describing the RF vector array and the B­mode array are defined. The                          vector array can be visualized as shown in Figure C2:                               Figure C2:  Three­dimensional vector array  (Source:http://www.mathworks.com/help/matlab/math/multidimensional­arrays.html)      55 
  • 56.   In the context of our pulse­echo experiments, ‘row’ is RF amplitude the along the                            x­axis of the sample plane, ‘column’ is the RF amplitude along the sample in the y                                direction and ‘page’ is the time interval between between transmitted and received                        signals. Both the x and y components of the vector array have two sets of signal                                associated with them: the input RF signal from the transmitting crystal, and the                          output RF signal from the receiving crystal.   The main job of the program is to convert the three­dimensional vector array                          into a 2­by­2 vector matrix, which is then converted to a two­dimensional B­mode                          image. This process is initiated by the iteration loop spanning lines 26 through 32                            Currently this for­loop has a limitation that it can do a maximum of nine iterations,                              which means that it can not handle more than nine sets of data from the raster scan;                                  we use five sets of data to produce the B­mode image. The final B­mode image in                                our case is a horizontal cross section along the sample plane, therefore only the                            output RF signal is relevant along the x axis. Thus in line 29 of Figure C1                                “sampleData (:, 2, :)” indicates that we are only considering the output RF signal                            in the x direction. The operation “squeeze” then removes singleton dimensions and                        converts the three­dimensional vector array into a 2­by­2 matrix in the space                        domain.  The two­dimensional vector array is then converted to a B­mode array by                        taking the natural logarithm of the RF amplitude data vector array, which                        graphically corresponds to the intensity or brightness level that we see in the final                            B­mode image. Next, the x and y distances along the plane of the sample are                              defined (line 43­54) using step­size (the smallest distance covered by the translator                        motor), window size in microseconds and the number of raster scan data sets. Then                            the image layout is specified as shown in Figure C1 between lines 57 to 67.    56