Upload
others
View
24
Download
0
Embed Size (px)
Citation preview
by
Carlton F. O. Hoy
A thesis submitted in conformity with the requirements for the degree of Master of Applied Science
Graduate Department of Mechanical and Industrial Engineering University of Toronto
© Copyright by Carlton Hoy 2015
Cardiovascular Computed Tomography Phantom Fabrication and Characterization through the
Tailored Properties of Polymeric Composites and Cellular Foams
ii
Abstract
Cardiovascular Computed Tomography Phantom Fabrication and Characterization through
the Tailored Properties of Polymeric Composites and Cellular Foams
Carlton F. O. Hoy
Master of Applied Science and Engineering
Graduate Department of Mechanical & Industrial Engineering
University of Toronto
2015
The overall objective of this thesis was to control the fabrication technique and relevant
material properties for phantom devices designated for computed tomography (CT) scanning.
Fabrication techniques using polymeric composites and foams were detailed together with
parametric studies outlining the fundamentals behind the changes in material properties
which affect the characteristic CT number. The composites fabricated used polyvinylidene
fluoride (PVDF), thermoplastic polyurethane (TPU) and polyethylene (PE) with
hydroxylapatite (hA) as additive with different composites made by means of different
weight percentages of additive. Polymeric foams were fabricated through a batch foaming
technique with the heating time controlled to create different levels of foams. Finally, the
effect of fabricated phantoms under varied scanning media was assessed to determine
whether self-made phantoms can be scanned accurately under non-water or rigid
environments allowing for the future development of complex shaped or fragile material
types.
iii
Acknowledgments
I would like to thank both Professor Hani Naguib and Dr. Narinder Paul for being attentive,
encouraging and knowledgeable supervisors throughout my MASc research. Their guidance
and support has helped me in countless ways through my studies and has been illustrated in
the following thesis.
I would also like to thank all of my fellow lab mates and colleagues in SAPL for aiding me
through their technical knowledge, equipment experience, and insight through ideas and
input on my research. I am thankful for the help that Reza Rizvi, Shahrzad Ghaffari, Farooq
Al Jahwari, Eunji In, Janice Song, Terence Lee, Mohamad Kshad, Kyle Eastwood, Shahriar
Ghaffari, Gary Sun, Nazanin Khalili, Muhammad Anwer, and Harvey Shi had provided the
past 2 years. I would further like to thank my undergraduate students Pranav Kadhiresan,
Kate Lonergan, Yuyang Pang, and Sherif Ramadan for their tireless help, incredible work
ethic and enthusiasm.
For their encouragement, belief and love and I would like to thank my family, Vance,
Hiroko, Vanessa and Jon. Without your support I would not be where I am now.
Last but not least I would like to dedicate this research to Ashton Hong. Your love and
endless support through countless nights of worry has lead me to be able to complete this
research thesis.
iv
ABSTRACT .......................................................................................................................................... II
ACKNOWLEDGMENTS ................................................................................................................. III
TABLE OF CONTENTS ................................................................................................................... IV
LIST OF FIGURES .......................................................................................................................... VII
LIST OF TABLES ............................................................................................................................. IX
CHAPTER 1 INTRODUCTION ......................................................................................................... 1 1.1 PREAMBLE ................................................................................................................................. 1 1.2 THESIS OBJECTIVES ................................................................................................................... 2
1.3 THESIS ORGANIZATION ............................................................................................................. 5
CHAPTER 2 BACKGROUND AND LITERATURE SURVEY ..................................................... 6
2.1 COMPUTED TOMOGRAPHY MEDICAL IMAGING PHANTOMS ..................................................... 7 2.2 COMPUTED TOMOGRAPHY MATHEMATICAL MODELING ......................................................... 8
2.2.1 Computed Tomography Numbers - Hounsfield Units ....................................................... 9 2.2.2 Attenuation Coefficient & Electron Density ...................................................................... 9
2.2.3 Effective Photon Energy & Atomic Number ................................................................... 12
2.2.4 Bilinear Scaling ................................................................................................................ 13 2.3 MATERIAL FABRICATION ........................................................................................................ 15
2.3.1 Polymeric Composite Materials ....................................................................................... 16
2.3.2 Microcellular Foam Materials .......................................................................................... 17 2.4 COMPUTED TOMOGRAPHY PHANTOM STUDIES ...................................................................... 19
2.4.1 Gammex 467 – Commercial CT Phantom ....................................................................... 19 2.4.2 Polymeric Phantom Design .............................................................................................. 19
2.4.3 Coronary Artery Phantom Design .................................................................................... 21 2.5 SUMMARY ................................................................................................................................ 23
Table of Contents
v
CHAPTER 3 POLYMERIC COMPOSITES FOR CARDIAC CT PHANTOM
APPLICATIONS ................................................................................................................................. 25 3.1 MOTIVATION ............................................................................................................................ 26
3.2 MATERIALS AND METHODS ..................................................................................................... 28 3.2.1 Polymeric Matrices & Filler ............................................................................................. 28
3.2.2 Phantom Material Processing ........................................................................................... 28
3.2.3 Computed Tomography Number ..................................................................................... 29 3.2.4 Physical Properties ........................................................................................................... 30
3.3 RESULTS .................................................................................................................................. 30
3.3.1 Effect of Hydroxyapatite Content on CT Number ........................................................... 30 3.3.2 Density & Elemental Composition Values ...................................................................... 32
3.3.3 Effect of Linear Attenuation Coefficient on CT Number ................................................ 33 3.4 DISCUSSION ............................................................................................................................. 36
3.5 SUMMARY ................................................................................................................................ 37
CHAPTER 4 POLYMERIC CELLULAR FOAMS FOR LOW DENSITY COMPUTED
TOMOGRAPHY PHANTOM APPLICATIONS ............................................................................ 39
4.1 MOTIVATION ............................................................................................................................ 40 4.2 MATERIALS AND METHODS ..................................................................................................... 43
4.2.1 Foaming Procedure .......................................................................................................... 43 4.2.2 Material Fabrication ......................................................................................................... 44
4.2.3 Computed Tomography Number ..................................................................................... 46 4.2.4 Foam Characterization Methods ...................................................................................... 47
4.3 RESULTS .................................................................................................................................. 48
4.3.1 Effect of Cell Morphology on CT Number ...................................................................... 50 4.3.2 Effect of Cell Size on CT Number ................................................................................... 52
4.3.3 Effect of Density on CT Number ..................................................................................... 52
4.4 DISCUSSION ............................................................................................................................. 53
4.5 SUMMARY ................................................................................................................................ 56
CHAPTER 5 EFFECT OF SCANNING MEDIUM ON THE IN-HOUSE FABRICATED
POLYMERIC COMPOSITE CT PHANTOM DEVICES ............................................................. 58 5.1 INTRODUCTION ........................................................................................................................ 58
5.2 MATERIALS AND METHODS ..................................................................................................... 61 5.2.1 Material Selection & Processing ...................................................................................... 61
vi
5.2.2 Computed Tomography Scanning Conditions ................................................................. 62
5.2.3 Physical Properties ........................................................................................................... 63 5.2.4 Data Analysis ................................................................................................................... 63
5.3 RESULTS .................................................................................................................................. 64 5.3.1 CT Number and Material Property Measurements .......................................................... 64
5.3.2 CT Number Comparison .................................................................................................. 66
5.3.3 Material Property Comparisons ....................................................................................... 66 5.3.4 Linear Attenuation Coefficient to CT number Comparisons ........................................... 68
5.4 DISCUSSION ............................................................................................................................. 71
5.4.1 CT Number Comparison .................................................................................................. 71 5.4.2 Material Property Comparisons – Linear Attenuation Coefficient .................................. 72
5.5 SUMMARY ................................................................................................................................ 74
CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS .................................................... 75
6.1 CONCLUSION ............................................................................................................................ 75 6.2 LIMITATIONS ............................................................................................................................ 77 6.3 RECOMMENDATIONS ............................................................................................................... 78
BIBLIOGRAPHY ............................................................................................................................... 81
vii
FIGURE 1–1. RANGES OF CT NUMBER DESIRED FOR CARDIAC CT PHANTOM FABRICATION. ....................... 4
FIGURE 2–1. FLOWCHART ILLUSTRATING THE EFFECT OF MATERIAL PROPERTIES ON CT NUMBER
THROUGH ELECTRON DENSITY AND ATTENUATION COEFFICIENT. EXPLANATION DETAILING THE
METHOD OF CALCULATION AND SUBSCRIPTS ARE PRESENTED IN SECTION 2.2.2. ............................... 8
FIGURE 3–1. SCHEMATIC REPRESENTATION OF THE COMPOUNDING PROCESSING TECHNIQUE UTILIZED
TO FABRICATE THE POLYMERIC COMPOSITES. .................................................................................... 27
FIGURE 3–2. RELATIONSHIP BETWEEN CT NUMBER AND HA% FOR COMPOSITE SAMPLES AT 100KVP.
HA WAS ADDED AT 0, 2.5, 5, 10, 15, 20, AND 25 WEIGHT PERCENT FOR PE AND TPU AND 0, 2.5,
5, AND 20 WEIGHT PERCENT FOR PVDF. ............................................................................................ 31
FIGURE 3–3. COMPARISON OF TPU/HA AND PE/HA COMPOSITES IN TERMS OF MEASURED CT
NUMBER AND LINEAR ATTENUATION COEFFICIENT COMPARED TO THE BILINEAR TREND EXHIBITED
BY THE GAMMEX 467 PHANTOM WITH VALUES OBTAINED AT 100KVP. 35
FIGURE 4–1. FOAMING TECHNIQUE SETUP WITH REFERENCE TO JACOBS ET AL. [112] BUT
CONTROLLED FOR OUR PURPOSES OF FOAMING TPU. ........................................................................ 43
FIGURE 4–2. CT SCANNING ARRANGEMENT OF TPU FOAM SETS A THROUGH E. INCLUDED ARE
TPU100 SAMPLES ACTING AS PURE TPU FOR REFERENCE DURING SCAN. LABELING FROM A
THROUGH E REPRESENT TPU SETS A THROUGH E RESPECTIVELY. ................................................... 46
FIGURE 4–3. COMPARISON OF TPU FOAM SETS FOR CHANGING KVP VALUES TO CT NUMBER. ................ 49
FIGURE 4–4. RELATIONSHIP EXHIBITED BETWEEN CT NUMBER (HU) AND POROSITY (%), CELL
DENSITY (CELLS/µM2), AND AVERAGE CELL SIZE. AN INCREASE IN POROSITY, CELL DENSITY AND
AVERAGE CELL SIZE ALL SAW GENERAL TRENDS OF DECREASING CT NUMBER. THE POINTS
REPRESENT THE MEAN WHERE AS THE ERROR BARS REPRESENT THE MAXIMUM AND MINIMUM
MEASURED CT VALUES ...................................................................................................................... 51
FIGURE 5–1. TREND ANALYSIS AND COMPARISON BETWEEN CT NUMBER AND ITS CORRESPONDING
CALCULATED LINEAR ATTENUATION COEFFICIENT OF SCANS COMPLETED IN A WATER BATH
(YW), SALINE SOLUTION (YS), AND AIR ENVIRONMENTS (YA) AT 80KVP. ........................................ 69
List of Figures
viii
FIGURE 5–2. TREND ANALYSIS AND COMPARISON BETWEEN CT NUMBER AND ITS CORRESPONDING
CALCULATED LINEAR ATTENUATION COEFFICIENT OF SCANS COMPLETED IN A WATER BATH
(YW), SALINE SOLUTION (YS), AND AIR ENVIRONMENTS (YA) AT 100 KVP ...................................... 69
FIGURE 5–3. TREND ANALYSIS AND COMPARISON BETWEEN CT NUMBER AND ITS CORRESPONDING
CALCULATED LINEAR ATTENUATION COEFFICIENT OF SCANS COMPLETED IN A WATER BATH
(YW), SALINE SOLUTION (YS), AND AIR ENVIRONMENTS (YA) AT 80KVP. ........................................ 70
FIGURE 5–4. TREND ANALYSIS AND COMPARISON BETWEEN CT NUMBER AND ITS CORRESPONDING
CALCULATED LINEAR ATTENUATION COEFFICIENT OF SCANS COMPLETED IN A WATER BATH
(YW), SALINE SOLUTION (YS), AND AIR ENVIRONMENTS (YA) AT 80KVP. ........................................ 70
ix
TABLE 2-1. LIST OF ATTENUATION COEFFICIENT CONSTANT VALUES FOR A, B, M, N, K, AND L AS
NOTED BY WATANABE ET AL. [10] ..................................................................................................... 11
TABLE 2-2. SUMMARY OF MEASURED CT NUMBERS UTILIZING 80, 100, 120 AND 135KVPS.
FURTHERMORE, ILLUSTRATED ARE KEY MATERIAL PROPERTIES: MASS DENSITY, ZT AND ZR. ......... 19
TABLE 3-1. CT NUMBER VALUES OF PE, TPU, & PVDF COMPOSITES AT 80, 100, 120 AND 135KVP. ....... 30
TABLE 3-2. MATERIAL PROPERTIES OF FABRICATED PE, TPU & PVDF COMPOSITES OF EFFECTIVE
ATOMIC NUMBERS CALCULATED BY EQN. 3 WITH M AND N VALUES AT 3.4 AND 1.5 FOR ZT AND
ZR RESPECTIVELY. DENSITY IS CALCULATED BY MEANS OF EQN. 6. .................................................. 32
TABLE 3-3. ELEMENTAL COMPOSITION OF VARIED PE, TPU AND PVDF COMPOSITES ORDERED BY
ADDED HA WEIGHT PERCENT. ............................................................................................................. 33
TABLE 3-4. COMPARISON OF LINEAR ATTENUATION COEFFICIENT VALUES (µ) THROUGH A MATERIALS
AND CT SCAN APPROACH AS CALCULATED BY EQUATIONS (1) AND (2). MEASUREMENTS WERE
TAKEN AT 100KVP. ............................................................................................................................. 34
TABLE 4-1. PURE TPU MATERIAL PROPERTIES ILLUSTRATED AS ATOMIC COMPOSITION AND PHYSICAL
PROPERTIES. ........................................................................................................................................ 44
TABLE 4-2. OUTLINE OF PROCESSING TECHNIQUE UTILIZED FOR TPU FOAM SETS A THROUGH E. THE
VARIATION IN PROCESSING CONDITION IS BASED UPON THE LENGTH OF TIME UPON WHICH THE
TPU SAMPLES ARE INTRODUCED INTO THE HEATED WATER BATH. ................................................... 47
TABLE 4-3. CT NUMBER VALUES MEASURED IN HOUNSFIELD UNITS (HU) FOR SOLID TPU AND
SUBSEQUENT TPU FOAM SAMPLES. ADDITIONAL DATA OF MEASURED DENSITY THROUGH THE
ASTM STANDARD .............................................................................................................................. 50
TABLE 5-1. WEIGHT PERCENTAGES OF H, C, N, O, F, P, AND CA WITHIN POLYMERS AND POLYMERIC
COMPOSITES FABRICATED FOR CT PHANTOM STUDIES. MEASURED DENSITIES ARE ILLUSTRATED
FOR EACH POLYMERIC COMPOSITES. EFFECTIVE ATOMIC NUMBERS (ZƮ AND ZR) AS WELL AS
RELATIVE ΡE WERE CALCULATED BY MEANS OF EQUATIONS (2) THROUGH (5). ................................. 65
List of Tables
x
TABLE 5-2. COMPARISON OF MEASURED CT NUMBERS FOR THE SAME POLYMERIC COMPOSITES
MEASURED UNDER SEPARATE CONDITIONS OF AIR, SALINE SOLUTION AND WATER BATH
ENVIRONMENTS. ................................................................................................................................. 65
TABLE 5-3. COMPARISON OF DIFFERENCES IN CT NUMBER BETWEEN THE REFERENCED WATER BATH
MEASUREMENTS AND SALINE SOLUTION OR AIR. DIFFERENCES IN CT NUMBER ARE
ILLUSTRATED FOR ALL POLYMERIC COMPOSITE CT PHANTOMS FABRICATED AT 80, 100, 120
AND 135KVPS. .................................................................................................................................... 66
TABLE 5-4. COMPARISON OF ASSESSED EEFF VALUES THROUGH WATER BATH, SALINE SOLUTION AND
AIR ENVIRONMENTS AT 80, 100, 120, AND 135 KVPS. ........................................................................ 67
TABLE 5-5. COMPARISON OF U VALUES AS CALCULATED THROUGH EQUATION (2). COMPARISONS ARE
MADE AS A % DEVIATION FROM WATER BATH ENVIRONMENT SCANS AND IS CONDUCTED FOR
BOTH SALINE SOLUTION AND AIR ENVIRONMENTS AT 80, 100, 120 AND 135KVPS. .......................... 68
1
Cardiovascular disease is one of the leading causes for mortality within Canada, most
prominently with respect to coronary artery disease (CAD) [1]. The current gold standards
for CAD diagnostic, including catheter coronary angiography and intra-vascular ultrasound
are limited in imaging capability, invasive, and time intensive [2], [3].
Recently, Computed Tomography Coronary Angiography (CTCA) has been noted as
an effective imaging technique to confirm CAD in patients with minimal to moderate disease
[4], [5]. However, CTCA is still limited in such that there is still the need for a more accurate
evaluation of well established and advanced coronary disease evaluation [4], [6]. To improve
detection of CAD through CTCA in those patients, CT limitations such as resolution, post
processing, and calcified plaque characterization must be addressed through the optimization
of image acquisition parameters and post-processing techniques [4]–[7].
Chapter 1 Introduction
1.1 Preamble
2
To overcome these limitations, CT medical phantom devices are leveraged. A CT
phantom acts as an anthropomorphic, or life-like, model that can be used as a consistent
testing method to introduce new imaging techniques and algorithms to improve CT imaging
techniques. This eliminates the need for human tissue and with the control and understanding
of material properties; the CT properties can be tailored to match that of the desired tissue or
system [8].
Current commercial CT medical phantoms, including the Gammex 467 phantom – a
common utilized phantom product, are listed per tissue, however are limited in tissue variety
and specificity. Many systems, particularly that of the cardiac system, require a wide range of
tissue properties and are exhibited as such under the CT scanner. Coronary plaque is a key
example of variation that, under the CT scanner, requires a high level of accuracy and could
potentially have a wide range of CT property variation due to various types of plaque and
levels of plaque calcification.
The main points of motivation are listed as follows:
1. To overcome the mortality of CTCA, improved scanning techniques are necessitated,
in which CT medical phantoms act as a synthetic means for modeling purposes
2. Current commercial CT phantoms are costly, do not illustrate a consistent means of
fabrication method, and lack tissue variety and specificity
3. The focus can be aimed towards the cardiac system, as its complexity and variation in
CT properties necessitates CT phantoms with these features
1.2 Thesis Objectives
3
The overall objective of this research is to detail polymeric fabrication techniques for
the development of CT phantom devices. The novelty and main contribution of this work
towards the field of CT imaging and materials engineering is describing the fine-tuning of
CT number by means of selected polymeric fabrication techniques. We aim to create a bridge
between materials and clinical engineering specifically to improve CT imaging techniques by
means of CT phantom fabrication. Current CT phantoms are limited to commercial means or
have described studies to which materials are roughly selected to match the need of the
specified study.
From this, we aim to have CT phantoms can be fabricated with controlled CT
properties thus accurately mimicking any tissue under CT scanning. The specific aim is to
mimic the various constituents around the coronary artery under CT imaging techniques. To
do so, the material must attenuate identically to that of the desired constituent. Attenuation is
measured, through the CT scanner, by means of a loss of x-ray energy due to electron scatter
or photoelectric absorption and is represented most commonly quantitatively as its CT
number in Hounsfield units (HU) [9], [10]. This value is a function of primarily the electron
density, or mass density, and effective atomic numbers of the material being scanned [9]–
[11]. Therefore, for the phantom, these material properties must be controlled to attenuate
with likeness to that of the native coronary artery. As such synthetic polymers are ideal
materials for phantom device fabrication due to their varying electron densities with diverse
compositions of molecular weights, mass densities, and effective atomic numbers [12].
Polymers can also undergo highly controlled material processing conditions, allowing for the
fine-tuning of attenuation values as well as creating consistent conditions for product
fabrication [12]. Studies have continued to be developed based upon polymer composites for
4
CT phantom purposes particularly the use of hydroxyapatite (hA) [4], [13], [14].
Additionally, foamed CT phantom are also often considered for low density tissue mimics
and is, thus, also considered as a key fabrication process potential for CT phantom use [15].
The three key objectives for the research involve the following:
1. To fabricate polymeric composites through controlled twin screw compounding
technique thus creating ranged but controlled properties capable of mimicking mid to
high density cardiac and plaque components
2. To fabricate polymeric cellular foams through a batch foaming process thus creating
ranged but controlled properties capable of mimicking low density cardiac and
adipose tissue environments
3. To investigate ideal scanning methods and environments for said fabricated
polymeric CT phantom devices through the modeling of material properties to the CT
scanner characteristic CT number
With reference to CT numbers, the desired goal is ranged to mimic: adipose / fatty plaque (-
300 to -50HU), soft tissues (0 to 100HU), and finally ranges and variations of coronary
plaque (0 to 1000HU).
Figure 1–1. Ranges of CT number desired for cardiac CT phantom fabrication.
5
The body of this thesis is comprised of four chapters followed by concluding remarks and
suggestions for further research. A background a literature review is presented in Chapter 2
which illustrates: the physical and mathematical concepts behind CT imaging, the numerical
goals desired for cardiac tissue imaging, compilation of CT phantom studies, and the
fabrication concepts and techniques behind both polymeric composites and cellular foams.
Chapter 3 and 4 then outline the fabrication and characterization of polymeric composite and
polymeric cellular foam CT phantoms respectively. Chapter 3 describes specifically the
effect of hydroxyapatite (hA) as an additive to select polymers as a parametric study
considering mass density, atomic composition, and CT numbers. Chapter 4 then describes the
effect of select parameters controlling polymeric cellular foaming with a comparison
between mass density, cell properties and CT numbers. Chapter 5 then presents studies
concerning the scanning conditions and environments ideal for fabricated CT phantoms. A
comparison between scans completed in the typically used water bath is contrasted against
enclosed saline solution and air environments. The studies are compared such that Chapter 3
and 4 relate to the CT control to range between the CT numbers necessary for a cardiac
system. Chapter 5 then studies the samples fabricated within studies 3 and 4 to determine an
alternative scanning media that can accurately scan these fabricated phantoms, necessary for
cardiac CT phantom scanning. Lastly, the conclusion presents an overview of complete
research and current studies, as well as suggestions for continuation in this field of study.
1.3 Thesis Organization
6
As the purpose and motivation of this research is aimed towards the fabrication,
characterization and analysis of CT medical phantoms, it is pertinent to define concepts
involved with such. This section, first, introduces a definition of CT phantoms and its
mechanism with respect to the CT scanner. Continuing from this, a numerical methodology
is introduced where properties expected to affect the CT number are defined. This includes
defining key concepts involved in CT scanning including Hounsfield Units (HU),
attenuation, electron density, effective atomic number and effective energy. Continuing from
definitions and key concepts, the selected materials and fabrication technique and
mechanisms behind such are illustrated. Polymeric bases suitable for CT phantom processing
are introduced together with the additive necessary for composite processing. Both polymer
composite processing through twin screw compounding and foam processing are introduced
as key methods to which the CT properties may be controlled. Lastly, included is a section
illustrating current studies within the field to which an extended motivation to this research
can apply to.
Chapter 2 Background and Literature Survey
7
As such, this review aims to review the topics of CT scanning, medical imaging
phantoms and its implementation with respect to the coronary artery. Furthermore, this
review will present and assess recent studies relating to mathematical modeling techniques
for CT scanning and polymer processing techniques, both of which will be integral in
justifying the approach for the present CT coronary artery phantom study.
In terms of medical imaging, a CT scanner is an x-ray based radiological device which will
take specified cross sections of a given element and will produce 2D slice images which can
further be stacked to create the 3D element volume. Each slice is represented as an image
based upon measured x-ray attenuation by the CT scanner per pixel. This value, in turn, can
be represented then as a CT number that is roughly the average relative linear attenuation of a
voxel, or a volume element pixel. This CT number scale, which will be explained further in
the following sections, is a relative value measured in Hounsfield units (HU) and is based
upon water having a CT number of 0 HU [16], [17].
CT phantoms are an ideal anthropomorphic device that will mimic the desired tissue
based upon its CT properties which is mainly that of its CT number. They can be leveraged
in place of live or cadaveric tissue for calibration and imaging studies. Since any native
human tissue is limited in resource and, depending on the tissue, can carry inconsistencies
between individuals, having an imaging phantom can be beneficial for a variety of purposes.
Device calibration, theoretical testing and staff training may require phantom devices with a
high level of accuracy compared to that of native tissue. With this mind, the phantom must
then be identified as anthropomorphic to that of the native tissue, with respect to the imaging
2.1 Computed Tomography Medical Imaging Phantoms
8
process used. In terms of CT scanning, the phantom should represent the native tissue by
means of primarily electron density and mass density, both of which relating to the
attenuation coefficient. This is further detailed in the following sections [18]–[20].
The overall objective of this research is to fabricate phantom devices with the ability to
mimic the various constituents around the coronary artery under CT imaging techniques. To
do so, the material must attenuate identically to that of the desired constituent. Attenuation is
measured, through the CT scanner, by means of a loss of x-ray energy due to electron scatter
or photoelectric absorption and is represented most commonly quantitatively as its CT
number in Hounsfield units (HU) [9], [10]. This value is a function of primarily the mass
density and electron densities of the material [11]. As such the following flow chart
illustrates the modeling process in which the materials properties can be correlated to CT
number through a number of formulaic interpretations.
Figure 2–1. Flowchart illustrating the effect of material properties on CT number through electron density and attenuation coefficient. Explanation detailing the method of calculation and subscripts are presented in Section 2.2.2.
2.2 Computed Tomography Mathematical Modeling
9
Figure 2-1 is a flow chart describing how to ascertain a prediction of the CT number
from basic materials properties. Formulae were used in its simplest forms based primarily on
related studies [21], [22]. Here, within Ng, NA refers to Avogadro’s number, MWTot the total
molecular weight, ni the number of ith atoms/molecule and Zi being its corresponding atomic
number. Within the electron density formulae, ρ is the mass density and Ng,w being the
electrons/gram of water. Within the attenuation coefficient formula, µw refers to the relative
linear attenuation coefficient with respect to water while µw refers to the linear attenuation
coefficient of water. As will be described in the subsequent sections, added complexity will
be incorporated based upon scanner conditions and material choice.
Given the desired coronary artery and its environment, an appropriate range of CT numbers
would be between -300 to +700 HU which would encompass fully that of adipose tissue,
fatty plaque, vulnerable plaque and various levels of calcified plaque [23], [24]. Included is
the control of CT numbers at discrete CT tube voltages. Varying the tube voltage, or photon
energy, can cause the phantom to attenuate at different values [25]. As such discrete tube
voltages at which CT imaging is typically done are considered [10]. With select polymeric
systems, this range of appropriate CT numbers for the coronary artery environment can be
obtained. Given the control of relevant material properties, most importantly that of mass
density and effective atomic numbers, desired CT numbers can be obtained.
The CT number is primarily based upon the average relative linear attenuation of a voxel at a
specified x-ray energy level. Essentially, a CT number takes a relative value, based upon the
2.2.1 Computed Tomography Numbers - Hounsfield Units
2.2.2 Attenuation Coefficient & Electron Density
10
attenuation of water. Attenuation is a value based upon its attenuation coefficient which,
through specific material properties, will change. The material properties we are concerned
with are primarily electron density and mass density, both of which can affect how the
material attenuates from the exposed x-ray [10], [16].
In theory CT numbers can be derived through attenuation coefficient numerically
from a set of equations as originally described by McCullough et al. [21] and also by
Rutherford et al. [26]. Firstly, CT numbers can be determined through a ratio between linear
attenuation coefficients of the scanned material and water:
𝐶𝑇 𝑁𝑢𝑚𝑏𝑒𝑟 = 1000 𝜇 − 𝜇!𝜇!
(Eqn. 1)
where the CT number is measured in Hounsfield Units (HU) and µ and µw are the linear
attenuation coefficients of the material being scanned and water respectively. Watanabe et al.
[10] described how the linear attenuation coefficient (µ) can be described in terms of the sum
of photon effects of photoelectric absorption, coherent (Rayleigh) scattering and incoherent
(Compton) scattering terms. Attenuation coefficients are typically described by both mass
and linear attenuation coefficients with the latter having material mass density incorporated.
In its simplest form, linear attenuation coefficients can be almost proportional to the electron
density, which therefore allows many to consider using relative electron densities to calculate
the CT numbers. However, as described in equation 2, the linear attenuation coefficient
theoretically considers small values based upon x-ray – material interactions which can affect
the linear attenuation coefficient [10], [27]:
11
𝜇(𝐸) = 𝜌! 𝑎𝑍!!
𝐸! + 𝑏𝑍!!
𝐸! + 𝑐(𝐸) (Eqn. 2)
E refers to the energy in keV. To calculate the linear attenuation coefficient as such,
the energy value in keV relative to the tube voltage applied is needed. Therefore, based on
the tube voltage applied (80, 100, 120 and 135kVp), a non-linear regression analysis was
completed through the SigmaPlot program (SPSS Inc., Chicago, IL, USA) between CT
number and the material properties to find the effective energy value in keV. Then a, b, m, n,
k, and l are fitting constants defined which give the best fit to the actual attenuation. Initial
estimates of constants and further iterations through non-linear regression modeling will
determine the values and can vary by means of the number of iterations done and initial
estimates chosen. ZT and ZR are the effective atomic numbers for photoelectric absorption
and coherent scattering respectively and are defined as:
𝑍!,! = 𝑛!𝑍!!,!! !,!
(Eqn. 3)
where ni and Zi are the electron fraction and atomic number respectively of the ith element
being described. A study conducted by Watanabe et al.[10] compares various studies
including that of McCullough et al. [21] and Rutherford et al. [26] that outline the most
accurate approach to determining the fitting constants through an in-depth phantom analysis.
Table 2-1. List of attenuation coefficient constant values for a, b, m, n, k, and l as noted by Watanabe et al. [10]
Variable Value a 2.30E-23 b 1.70E-24 m 3.40 n 1.50 k 3.20 l 1.60
12
The c(E) term described in equation (2) is the Compton scattering effect and can be
described mathematically by means of the Klein-Nishina formula:
𝑐(𝐸) =8𝜋3 𝑟!! 1− 2
𝐸𝑚!𝑐!
+ 5.2𝐸
𝑚!𝑐!!
− 13.3𝐸
𝑚!𝑐!!
(Eqn. 4)
where me and c are the electron mass and velocity of light respectively and ro = 2.818*10-13
cm. Lastly, ρe as described in equation (2) is the electron density of the material. This can be
numerically determined by equation (5):
𝜌! = 𝜌 𝑁! 𝑛!𝑍!𝑀𝑊!"#
(Eqn. 5)
where ρ is the mass density, NA is Avogadro’s number at 6.02*10-23, ni and Zi are the
electron fraction and atomic number of the ith element respectively, and MWTot being the
total molecular weight. These formulae encompass the relationship between key material
properties and CT number.
The database used which has compiled the photon effects or ‘photon cross-sections’
is known as the XCOM: NIST database [28] which has considered both theoretical and
experimental work to create a wide energy and atomic number ranged database of mass
attenuation coefficients [29]–[32]. The database has also considered mixtures and
compounds at which all of the effects are taken into consideration.
The measured HU of a given element can vary significantly with changing x-ray energies. By
revisiting equation 2, it can be seen that altering the x-ray energy will affect the various
photon cross-sections but in varied ways. However, also referenced to equation 2, the extent
2.2.3 Effective Photon Energy & Atomic Number
13
to which x-ray energy will affect the element is highly dependent on the effective atomic
number of the material as well.
An example of a study related to this was conducted by Ebert et al. [33] where
experiments were conducted on a CIRS phantom containing various electron density plugs.
CT scans were taken at discrete x-ray energies of 80, 100, 120, 140 kV. With increasing
electron density we will see an increase in atomic number and, thus, have taken this into
consideration for the following trend.
The notable trend is that with as energy increases, the CT numbers tend to decrease at
high electron densities. As well, this decrease in CT number with increasing energy becomes
more prominent as electron density increases. However, there is a notable trend occurrence at
lower electron densities.
Below a relative electron density of 1.0, the CT number values tend to marginally
increase with increasing energies. However at relative electron densities above 1.0, the trend
alters to that of decreasing CT number with increasing energies. This phenomenon can
further be described by means of NIST compiled data.
Multiple studies have described a phenomenon in modeling the linear attenuation coefficient
against the CT number in which a bilinear relationship is illustrated. Typically, an analysis
can be made to calculate the linear attenuation coefficient (µ) by experimentally finding the
CT number. Equation 1 illustrates the formula to do so, however, by collected data, it has
been studied that this equation is only representative for a CT number roughly less than 0
HU.
2.2.4 Bilinear Scaling
14
In a study conducted by Bai et al. [34] and further illustrated by Saw et al. [35] and Brown et
al. [36], materials can be described as belonging in two distinct regions. Below an HU of 0, a
‘water-air assumption’ is made in which the trend between linear attenuation coefficient and
CT number is as in equation 1. Above an HU of 0, a ‘water-bone assumption’ is made in
which the reference taken is not between that of air and water but now water and that of
highly attenuating bone. The basis of this work, conducted by Blankespoor et al. [37]
considered a piecewise bilinear fitting technique was utilized to fit the attenuation against CT
number to a best fit. All of these studies have experimentally illustrated the bilinear
relationship exhibited between linear attenuation coefficient and CT number. This
phenomenon can be described through a number of different effects.
The first potential effect may be described through the act of beam hardening of the
incident polychromatic CT x-ray beam. Beam hardening is the effect of high-density
materials absorbing lower energy beams.
Another possible effect has noted that CT images can accurately estimate the
attenuation coefficients for muscle and soft tissue, however as the tissue density increases
towards that of bone, larger margins of error are noticed. As such, with higher density
tissues, there is an inaccurate measure of attenuation coefficients from CT scans [38]. To
further emphasize this point, Kinahan et al. [39]–[41] have described how the mass
attenuation coefficients of water and most soft tissues are relatively similar.
This can be attributed to equation 2 with density removed. The basis of mass
attenuation coefficients is the atomic number related to its photon effect and thus, we can
stipulate that the effective atomic number is similar between water and most soft tissues. On
15
the contrary, it can be said other tissues, most importantly bone, will have a noticeably high
mass attenuation coefficient. Since soft tissue is organized primarily of organic elements and
bone is, alternatively, composed primarily of calcium based molecules the effective atomic
number of bone will be significantly higher, thus having a much higher mass attenuation
coefficient than soft tissue or water [42].
Also to be noted is the effect of increasing photon energy on the bilinear trend. At
high photon energies, as described above, the mass attenuation coefficient of bone reaches
the same values as that of muscle and air. This can also be referenced back to equation 2 and
it can be noted that the photon effects of photoelectric absorption and coherent scattering
become negligible, thus the mass attenuation coefficient becoming dominated by the
incoherent scattering effect.
Further studies conducted have begun to describe a tri-linear scaling in which CT
numbers higher than that of 1000 HU belong to a different scaling relation between linear
attenuation coefficient and CT number [43], [44]. Again, a similar principle applies to the
bilinear relationship; however, an added region is necessary due to the CT numbers
extending higher than the previously attenuating bone tissue.
The materials proposed for the phantom device must be controlled to attenuate with likeness
to that of the native coronary artery. As such synthetic polymers are ideal materials for
phantom device fabrication due to their varying electron densities with diverse compositions
of molecular weights, mass densities, and effective atomic numbers [12]. Polymers can also
undergo highly controlled material processing conditions, allowing for the fine-tuning of
2.3 Material Fabrication
16
attenuation values as well as creating consistent conditions for product fabrication [12].
Furthermore, the use of polymeric materials allows for a high degree of property consistency
and long life stability of material thus allowing for the samples to undergo repeated
consistent scans. A limitation may involve degradation at high temperatures or ultraviolet
(UV) environments, however in a clinical setting we expect this to be minimal at best.
To adequately mimic the entire range of CT numbers necessary for the coronary artery and
its environments, a wide array of phantoms must be fabricated. Further, the phantoms must
be controlled by its elemental composition with respect to the previous section. Polymer
composites serve as the ideal classification of materials due to its fabrication process control,
composition control, and a number of different choices of polymer and composite
combinations. The latter point requires a look at current phantom devices and its material
composition, thus providing a basis on which composites may be most beneficial.
Depending on the desired tissue to be mimicked, the phantom material chosen can
vary. Bone, other mineralized tissues and calcified plaque tend to be composed of calcium
carbonate (CaCO3) or other calcium based products such as hA [10], [18], [42]. These
materials will tend to attenuate with a CT number roughly around 500 to 1500 HU depending
on its density and photon energy used. Soft tissues at higher densities, such as the artery wall
have been mimicked by a variety of polymeric materials due to its low attenuation. The CT
number rests between 0 and 200 depending again on density and photon energy. Typical
polymers used and studied as phantoms have been poly(methyl methacrylate) (PMMA),
polyurethane (PU), acrylonitrile butadiene styrene (ABS) or polyethylene (PE) [42], [45].
2.3.1 Polymeric Composite Materials
17
Lastly, to mimic the various adipose low density soft tissue surrounding the arteries,
polymethysilane (PMS), polyethylene (PE), and nylon, as examples, have been studied [42],
[46], [47].
To further fine tune the attenuation to a desired HU value together with a consistent
polymer base, a polymer composite can be fabricated. Polymer composites for biomedical
applications, or biocomposites, have become common place and are often sought out for
studies for varied applications such as implants and tissue scaffolds [48]. As such it is
pertinent to understand how to properly fabricate polymer composites for its given purpose.
One of the main concerns of polymer-composites is its characteristic consistency. The
additives, when blended with the polymer, based on the application will likely benefit most if
the additives are well or uniformly dispersed within the polymer matrix. A number of studies
have been conducted on the use of a twin screw compounder to uniformly disperse nano-
particles within a polymeric matrix [49]–[52]. Pertaining to our research, it has been
demonstrated that polymer-clay nanocomposites can adequately compound with certain
polymers such as PMMA, PU, and PE [53]–[55], as well as carbon nanotubes with
poly(lactic acid) (PLA) [49].
To further control the CT number of the element by material properties, microcellular
processing can prove to be highly beneficial. Microcellular processing essentially can create
a foam from a polymeric based material. This condition allows for a porous structure to be
fabricated, thus decreasing the overall density of the material. By this, a polymer can be
2.3.2 Microcellular Foam Materials
18
controlled to have a lower CT number by means of its given density depending on the
porosity of the sample.
There are a variety of methods to fabricate a microcellular network from a polymer or
polymer composite. Some methods include additives such as chemical blowing agents which
are compounded with the polymer and further degraded so that they may enter a volatile state
thus escaping the matrix and leaving pores. However, most beneficial to our studies, we
considered batch foaming processes with a physical blowing agent [56]–[59].
The use of physical blowing agents, as opposed to chemical blowing agents, typically
leads to a greater degree of porosity and potentially more consistency with controlled
foaming conditions [60]. The process behind the use of physical blowing agents, typically
that of nitrogen (N2) or oxygen (O2) gas is exposing the polymeric sample to a supercritical
pressure of that gas. At this stage, the gas interacts with the polymer as a liquid and thus can
diffuse into the polymer. Depending on the polymer, the amount of time it takes for the gas
to saturate within the sample will vary. Additionally, depending also on the supercritical gas
used, the saturation point will vary. An ideal amount of time must be assessed based on the
polymer and gas used which is derived experimentally through material diffusivity.
The other main factor to consider for batch foaming with physical blowing agents is
the pressure drop threshold. The rate at which the pressure drop occurs from its supercritical
state to standard pressure will significantly affect the cell structure that is produced. What we
aim to create is a thermodynamic instability in which cell nucleation will occur by means of
this rapid pressure drop [61].
19
To better define the standard of current commercial CT phantoms, a commercial and mature
CT phantom available is the Gammex 467. This phantom mimics a variety of different
tissues through cylindrical inserts interchangeable within a PMMA block. Listed below are
the measured CT numbers of the inserts along with material properties: mass density, ZƮ and
ZR.
Table 2-2. Summary of measured CT numbers utilizing 80, 100, 120 and 135kVps. Furthermore, illustrated are key material properties: mass density, ZT and ZR.
Sample CT Number (HU) Density
Measured (g/cm3)
ZT ZR 80kVp 100kVp 120kVp 135kVp
LN-300 -717.3 -712.3 -710.0 -708.5 0.294 7.60 6.61 LN-450 -571.6 -567.4 -565.6 -562.8 0.454 7.60 6.61 Adipose -104.9 -92.2 -85.8 -83.3 0.956 6.23 5.74 Breast -47.8 -41.0 -37.9 -36.3 0.998 7.00 5.99 Solid Water 2.7 0.3 -1.6 -2.8 1.025 7.83 6.32 Brain 4.4 12.3 18.4 21.6 1.060 6.11 5.59 True Water 0.3 0.9 1.3 0.6 1.000 7.49 6.95 Liver 89.3 82.5 80.3 78.7 1.114 7.84 6.32 Inner Bone 359.0 261.2 215.5 194.1 1.160 10.55 7.94 B-200 379.5 283.0 233.8 210.2 1.172 10.56 7.94 CB2 - 30% 771.7 624.7 535.4 485.4 1.342 11.04 8.19 CB2 - 50% 1473.6 1185.9 1014.7 913.7 1.565 12.67 9.71 Cortical Bone 2169.7 1759.3 1506.8 1360.5 1.829 13.76 10.92 Polypropylene -115.2 -100.2 -92.1 -88.1 0.905 5.52 5.03
2.4 Computed Tomography Phantom Studies
2.4.1 Gammex 467 – Commercial CT Phantom
2.4.2 Polymeric Phantom Design
20
It can be noted that with the processing techniques for fabrication both microcellular and
composite materials, it is possible to create CT phantoms highly anthropomorphic to the
desired tissue. As a reference, previous analyses have illustrated that PE and TPU will have
CT numbers of approximately -100 and 100 HU respectively. Furthermore, the calcium
based hA tends to mimic bone at around 1000 HU depending on the density and
concentration. As such, it can be stipulated that with discrete additions of hA to the given
base polymers, we can fabricate a phantom of finely tuned CT number between the CT
numbers of the base polymer and hA. Furthermore, by microcellular processing, the material
can decrease its density while also introducing voids not discernible to the CT scanner. As
such the CT number can be decreased to that close to air at -1000 HU while still seemingly
having a homogenous cross sectional area, as seen by the CT scanner.
There have been studies that focus on varied concentrations of hA for the assessment
of beam hardening. Since increasing hA will lead to a higher atomic number element and,
thus, will likely affect the photon interactions that occur. Deurerling et al. [13] considered the
fabrication of HDPE-hA composite phantoms with varied concentrations of hA. The
measurements of x-ray attenuation based on micro-CT measurements were analyzed. It was
evident that from the polymer-HA phantom, that increasing amounts of hA revealed a non-
linear trend with linear attenuation coefficient. As mentioned, the reason behind the non-
linearity was like beam hardening and scattering artifacts [62]–[65]. Also noted was that
following the scan a beam hardening correction algorithm was applied from the
manufacturer, however the extent to which beam hardening seems to occur at high atomic
number elements can overtake this correction.
21
Polymers and various polymeric composites have been commonly utilized as CT phantoms
due to their manufacturability and potential for wide ranged material properties [9], [66]–
[69]. However, current phantom devices are specific to the desired tissue whereas some
studies aiming to improve CT medical imaging necessitate the control of the phantom CT
characteristics to a fine degree [10], [70]. Certain tissues and structures within the human
body may differ and require a fine degree of CT number control to be anthropomorphic.
Coronary plaque is an example of such, which has ranging CT number values based upon its
composition and degree of calcification. Furthermore, the geometrical shape of these
phantoms would benefit by mimicking the native structures[4], [6], [66], [71].
In terms of arterial phantom design for CT imaging, several obstacles must be
overcome, leading to multiple studies on such. Typically CT imaging can be conducted on
coronary arteries to detect various types of plaque to determine the patient risk when
considered at low to medium risk [72], [73]. However, this procedure requires that the
processed image has high clinical accuracy, thus certain limitations of the pairing between
CT scanning and the coronary artery must be overcome [74], [75]. Considering that the
lumen opening size can be significantly small, the opening may be as large as a few pixels on
the CT image. Furthermore, with small lumen sizes and highly calcified plaque, image
artifacts may become prominent, enough to exaggerate the size of the plaque. Lastly, the
environment of the artery may have CT numbers which are very close to one another. To this
point, when imaging the arterial fluid is typically dyed with iodine and can cause the CT
numbers of both the arterial plaque and fluid to be equal, thus making any discrepancy
between the two difficult [71], [76].
2.4.3 Coronary Artery Phantom Design
22
To further our understanding on the relationship between CT imaging and the
coronary artery and its environment, coronary artery plaque phantoms and its environment
have been analyzed in multiple studies [4], [67], [77]. An analysis of the various types of
plaque that may exist within the coronary artery under CT imaging is necessary. Continuing
from the study conducted by Toepker et al. [67] with regards to varying percentages of artery
stenosis, research has also been documented for varied plaque types. Firstly, a slight
deviation of this was done with coronary artery stent imaging by Yang et al. [66] by means of
an acrylonitrile-butadiene-styrene resin (ABS) based phantom. The vessels were filled with a
contrast agent attenuating at about 300 HU at 100 and 120 kVp tube voltages. Lastly,
commercially available stents were inserted into the phantom and imaged under CT
scanning. This study provides a basis on image clarity and CT number accuracy in an arterial
environment. Similarly, we can see how a highly attenuating stent (metal alloy or steel based
materials) will affect the arterial CT imaging. Prior studies were also conducted in a similar
manner however with different materials or live patients used. Regardless of the imaging
method, the results trended towards imaging capability with limitations in possible artifacts
caused by the highly attenuating stents [78]–[80].
From this, we can consider a less attenuating, but potentially larger sized coronary
plaque with varied levels of calcification. Typical studies consider both the variation of
plaque calcification and lumen size. One study, conducted by Kristanto et al., however,
considered only non-calcified plaque with varying lumen sizes [5]. As such, varied levels of
lipid content were considered for the plaque. Alternatively, studies have been conducted to
determine the accuracy of varied levels of calcified plaque under CT imaging. One
complication of measuring different levels of calcified plaque is the classification of soft,
23
intermediate and fully calcified plaque. Analyses have been made for elemental compositions
of plaque in different patients, thus giving an estimate of a range of CT numbers to classify
the various types of plaque [24]. Studies were made in relation to this by first classifying the
type of plaque through an intracoronary ultrasound by means of its echoic behaviour. The
stronger the signal, the more dense and calcified the plaque was considered. Following this
CT scan was taken, thus measuring the CT numbers for each given classification [23], [81],
[82].
CT imaging has been considered as an effective method to determine the risk of CAD within
a patient. However, current techniques are limited in a number of ways, including resolution
and radiation dosage. To overcome these obstacles, new post-processing techniques and
analysis tools are necessary. Since cadaveric and live tissue are limited in resource, the use of
an anthropomorphic device, a CT imaging phantom device, is ideal for consistent and
repeated use. As such, it is pertinent that the imaging phantom be as closely mimicking as
possible to the desired tissue.
The targeted region of interest for this study considers the coronary artery, coronary
plaque, and its surroundings. Thus the developed phantom must be able to accurately
illustrate this area through the phantom device. The main property necessary for control to be
anthropomorphic under CT imaging is the CT number measured in HU. Given the photon
energy supplied, the CT number of the part of the device must mimic that of the desired
element. Following this, various artery designs can be constructed based upon varied lumen
sizes and plaque calcification content. Conclusively, with a successful design from a highly
2.5 Summary
24
accurate coronary artery mimicking phantom, countless tools and post-processing techniques
can be developed to better CT coronary imaging techniques.
25
This chapter investigated the fabrication and characterization of polymeric composites for
use in cardiac CT phantom applications. Biomedical phantoms are commonly used for
various medical imaging modalities to improve imaging quality and procedures. Current
biomedical phantoms fabricated commercially are high in cost and limited in its specificity of
human environments and structures that can be mimicked. This study aimed to control the
measurable computed tomography (CT) number in Hounsfield units (HU) through polymeric
biomedical phantom materials using controlled amounts of hydroxyapatite (hA). The purpose
was to fabricate CT phantoms capable of mimicking various coronary plaque types. The CT
number is tunable based on the controlled material properties of electron density and atomic
numbers. Three different polymeric matrices of polyethylene (PE), thermoplastic
polyurethane (TPU) and polyvinylidene fluoride (PVDF) were used together with additions
of hA in mass percentages of 2.5, 5, 10, and 20% hA as well as a 0% hA as a control for each
polymeric material. By adding hA to PE, TPU, and PVDF an increasing trend was exhibited
between CT number and weight percent of hA added in a controlled manner.
Chapter 3 Polymeric Composites for Cardiac CT Phantom Applications
26
Polymers and various polymeric composites have been commonly utilized as CT phantoms
due to their manufacturability and potential for wide ranged material properties [9], [66]–
[69]. However, current phantom devices are specific to the desired tissue whereas some
studies aiming to improve CT medical imaging necessitate the control of the phantom CT
characteristics to a fine degree [10], [70]. Certain tissues and structures within the human
body may differ and require a fine degree of CT number control to be anthropomorphic.
Coronary plaque is an example of such, which has ranging CT number values based upon its
composition and degree of calcification. Furthermore, the geometrical shape of these
phantoms would benefit by mimicking the native structures [4], [6], [66], [71].
This study focuses on the control of CT number through phantom fabrication with a
range of CT numbers necessary for studies for CTCA. Our aim was to mimic the CT
numbers noted by various types and ranges of coronary plaque including soft (non-calcified),
intermediate and calcified plaque in which respective ranges are between -50 to 50 HU, 50 to
150 HU, and any plaque greater than 150 HU up to 800 HU to fully encompass high density
calcified plaque exhibited clinically [4], [6], [20], [67]. Further, the accuracy of these values
between the expected and measured CT numbers have been noted in various studies varying
at most by 40 HU [4], [23], [66].
The three polymers of interest were polyethylene (PE), thermoplastic polyurethane
(TPU), and polyvinylidene fluoride (PVDF) as their density values are varied allowing for
wide ranged CT phantom applications by means of controlled additives. Three polymer types
were selected due to the theoretical density of each polymer, which is expected to directly
relate to its given CT number. PE, TPU, and PVDF each attain a low, medium and high
3.1 Motivation
27
density value as base polymers with theoretical densities at 0.922, 1.2, and 1.78 g/cm3
respectively. Furthermore, all three polymers selected are thermoplastic polymers which are
able to be processed and shaped with ease under heat without degrading the polymeric
structure, however each polymer necessitates different processing temperatures due to higher
glass transition and melting temperatures. The additive used was nano-powdered
hydroxyapatite (hA), which is a commonly used product for phantom products and as well is
comprised of elements highly resembling that of bone tissue under x-ray exposure [13].
Various tissue engineering applications have utilized systems containing a combination of
polymer and hA successfully and thus is the platform for this study [83]–[87].
The method of control is also based upon the fabrication technique of the polymeric
composites. Nano-powdered hA will be introduced to the polymer through twin screw
compounding. This technique allows for the fabrication of a polymeric composite with
uniformly dispersed hA, indiscernible under CT medical imaging due to the particle size
[88]. This method for specific polymer/hA systems has been both examined and utilized for
various biomedical applications [89]–[91].
Figure 3–1. Schematic representation of the compounding processing technique utilized to fabricate the polymeric composites.
28
For the purposes of this study, three polymeric matrices – low density polyethylene (LDPE),
thermoplastic polyurethane (TPU), and polyvinylidene fluouride (PVDF) were selected with
hA as the filler, or additive. The polymers selected through a screening process in which
mass density was the main property through which the selection was made. LDPE and PVDF
acted as both the low and higher end of the mass density spectrum of commonly available
polymers. TPU was selected as an average mass density polymer and was selected ahead of
other average mass density polymers through common use of polyurethane as a material for
CT phantoms in commercial and selected study use. Theoretical densities were noted as
0.922, 1.2, and 1.67 g/cm3 for PE, TPU, an PVDF respectively. hA was then selected due to
its high density as an additive, its commonplace documentation as an additive to polymeric
matrices, as well as it being commonly utilized in biomedical studies as bone and calcium
based tissue substitutes.
Low density polyethylene (LDPE; Nova Chemicals Novapol, LC-0522-A), the polyester-
type thermoplastic polyurethane (TPU; Desmopan 385E), and polyvinylidene fluoride
(PVDF; ARKEMA, Kynar 720) were used as the base polymers. Polymers were processed as
received in pellet form. The additive, hydroxyapatite (hA; 677418, Sigma-Aldrich) was a
nanopowder with controlled particle sizing of less than 200nm.
PE and TPU polymers were combined individually with hA at 0, 2.5, 5, 10, 15, 20,
and 25 weight percent of hA. PVDF was combined with hA at 2.5, 5, 10 and 20 weight
3.2 Materials and Methods
3.2.1 Polymeric Matrices & Filler
3.2.2 Phantom Material Processing
29
percent hA. Composite materials underwent microfabrication by means of a 15 mL benchtop
twin screw microcompounder (DSM 15, DSM Xplore). The polymer and hA were added
simultaneously to the rotating twin screw compounder at 130oC, 200oC, 230oC for PE TPU
and PVDF respectively. Combinations were processed at 50 rpm for 15 minutes (100 rpm
and 10 minutes for PE). The extruded polymeric composites were quenched immediately
from the outlet in room temperature water.
This means of mixing creates a homogenous like mixture between two materials;
namely that of the polymer and hA in the case of our phantoms Compounded samples were
then molded into identical disk shaped samples for CT measurement by means of extruding
the polymer, cutting the strand into pellets, and then placing the pellets into a compression
mold (12 Ton Carver Hot Press) by a disc shaped mold (12.8-cm height by 0.65-cm
diameter). The Gammex 467 phantom, used as a comparative measure, was also scanned
under identical acquisition parameters.
The phantoms were measured in a wide volume CT (320 MDCT AquilionONE; Toshiba
Medical Systems, Japan) with CT numbers measured using ImageJ on post-processed
images. The scanning conditions were set at a 0.5mm detector collimation, 0.5 second gantry
rotation time, 50mA nominal tube current and a tube potentials of 80, 100, 120 and 135kVp.
The images were reconstructed using a single filter kernel with 1mm sections at 0.5mm
intervals. Furthermore, the shape and orientation of scans were conducted under a similar
condition to the Gammex 467 phantom to which cylindrical shaped inserts were oriented in a
circular fashion with a surrounding water-equivalent media.
3.2.3 Computed Tomography Number
30
Density measurements were carried out by ASTM Standard D792 – 13 by the measurement
of plastic density through water displacement. The elemental composition was determined
based upon available data sheet information pertaining to each material.
The attenuation of each sample was averaged between 3 measurements per image with 3
images measured per sample and 3 samples per composite measured. As such, per composite
measure, per kVp, 27 measurements we made and averaged as included per CT number. CT
number measurements were made at 4 different kVp values.
Table 3-1. CT number values of PE, TPU, & PVDF composites at 80, 100, 120 and 135kVp.
Polymer hA% CT Number (HU) 80kVp 100kVp 120kVp 135kVp
PE
0.0 -147.4 -124.5 -111.7 -103.5 2.5 -109.8 -91.0 -83.6 -78.8 5.0 -69.6 -61.7 -58.0 -51.4 10.0 34.3 23.7 18.7 9.4 15.0 137.2 109.6 91.2 87.9 20.0 257.6 203.0 171.8 160.3 25.0 376.5 298.9 253.4 240.7
TPU
0.0 80.9 95.1 104.1 110.4 2.5 141.3 145.1 149.1 145.0 5.0 186.4 185.2 182.4 180.9 10.0 328.7 289.9 266.8 253.6 15.0 383.1 328.4 297.2 280.1 20.0 466.4 392.5 346.8 320.3 25.0 537.2 444.0 387.1 352.9
PVDF
0.0 634.0 591.4 551.7 509.6 2.5 669.0 612.1 564.4 541.2 5.0 723.1 642.0 582.0 546.6 10.0 889.1 765.4 689.2 641.0 20.0 1195.7 995.3 869.2 794.1
3.2.4 Physical Properties
3.3 Results
3.3.1 Effect of Hydroxyapatite Content on CT Number
31
The subsequent focus of research was aimed towards 100kVp as a commonly utilized
acquisition tube voltage for a more in-depth analysis. The following trends were exhibited by
the composites for added hA to the polymers. Through data analysis and processing, within
the range of added weight percent of hA to PE, TPU and PVDF, general trends between
weight percent hA and CT number were exhibited. Increasing the amount of hA within each
polymer described an increasing relationship of CT number as described in Figure 3-2.
Figure 3–2. Relationship between CT number and hA% for composite samples at 100kVp. hA was added at 0, 2.5, 5, 10, 15, 20, and 25 weight percent for PE and TPU and 0, 2.5, 5, and 20 weight percent for PVDF.
Under the given scan conditions, a representative range of -124.5 to 995.3 HU is
obtained. The overall trend of increasing CT number with increasing hA is exhibited with
slight differences in trend between the PE/hA, TPU/hA and PVDF/hA composites in terms of
the relationship between hA weight% and CT number. It can be noted that adding hA to a
given polymeric base in discrete amounts will have similar increases in CT number.
PVDF; y = 2101.3x + 563.65 R² = 0.9848
TPU; y = 1384.9x + 115.26 R² = 0.9798
PE; y = 1703.7x - 137.48 R² = 0.9965
-200
0
200
400
600
800
1000
1200
0% 5% 10% 15% 20% 25% 30%
CT
Num
ber (
HU
)
hA%
PVDF TPU PE
32
The effective atomic numbers were calculated as indicated in equation (3). Table 3-2
illustrates the mass density, photoelectric effect and incoherent scattering effective atomic
number of pure PE, TPU, PVDF and hA. These values will largely affect the linear
attenuation coefficient. Table 3-3 illustrates the elemental composition of the composites
fabricated, which dictate the effective atomic number values.
Table 3-2. Material properties of fabricated PE, TPU & PVDF composites of effective atomic numbers calculated by eqn. 3 with m and n values at 3.4 and 1.5 for ZT and ZR respectively. Density is calculated by
means of eqn. 6.
Polymer hA% ZT ZR Density (g/cm3)
PE
0.0 5.51 5.03 0.862 2.5 6.59 5.31 0.918 5.0 7.35 5.59 0.942 10.0 8.49 6.14 0.974 15.0 9.36 6.66 1.019 20.0 10.08 7.18 1.054 25.0 10.70 7.68 1.093
TPU
0.0 6.31 5.85 1.167
2.5 7.16 6.11 1.176
5.0 7.83 6.36 1.205 10.0 8.86 6.86 1.249
15.0 9.67 7.35 1.273
20.0 10.35 7.82 1.295 25.0 10.94 8.28 1.309
PVDF
0.0 7.95 7.55 1.743
2.5 8.50 7.76 1.733
5.0 8.97 7.97 1.772 10.0 9.77 8.38 1.770
20.0 11.02 9.16 1.865
3.3.2 Density & Elemental Composition Values
33
Table 3-3. Elemental composition of varied PE, TPU and PVDF composites ordered by added hA weight percent.
Polymer hA% Elemental Composition Weight% H C N O F P Ca
PE
0.0 14.37 85.63 0.00 0.00 0.00 0.00 0.00 2.5 14.02 83.49 0.00 1.04 0.00 0.46 1.00 5.0 13.66 81.35 0.00 2.07 0.00 0.92 1.99 10.0 12.95 77.07 0.00 4.14 0.00 1.85 3.99 15.0 12.25 72.78 0.00 6.21 0.00 2.77 5.98 20.0 11.54 68.50 0.00 8.28 0.00 3.70 7.98 25.0 11.54 68.50 0.00 8.28 0.00 3.70 7.98
TPU
0.0 9.07 64.35 6.00 20.57 0.00 0.00 0.00 2.5 8.85 62.74 5.85 21.09 0.00 0.46 1.00 5.0 8.63 61.13 5.70 21.61 0.00 0.92 2.00 10.0 8.19 57.91 5.40 22.66 0.00 1.85 3.99 15.0 7.74 54.70 5.10 23.70 0.00 2.77 5.98 20.0 7.30 51.48 4.80 24.74 0.00 3.70 7.98 25.0 6.85 48.26 4.50 25.78 0.00 4.62 9.97
PVDF
0.0 3.15 37.51 0.00 0.00 59.34 0.00 0.00
2.5 3.07 36.58 0.00 1.04 57.85 0.46 1.00 5.0 3.00 35.64 0.00 2.07 56.37 0.92 1.99 10.0 2.85 33.76 0.00 4.14 53.40 1.85 3.99 20.0 2.56 30.01 0.00 8.28 47.47 3.70 7.98
Linear attenuation coefficient is a measure that combines the effect of density and elemental
composition. As shown in equation (1), the linear attenuation coefficient has a direct
relationship with the CT number. Furthermore, as illustrated in equation (2), the value of the
linear attenuation coefficient is based highly upon both the density and effective atomic
numbers of the material scanned. The numerical model was verified by means of two
methods.
Firstly, the linear attenuation coefficient was calculated by means of measured
material properties, in which we will call the materials approach. This approach uses
equation (2) with measured material properties, including the mass density and effective
3.3.3 Effect of Linear Attenuation Coefficient on CT Number
34
atomic number values to determine the linear attenuation coefficient. This approach is then
compared to a theoretical approach. This method uses the measured CT number values
obtained through ImageJ analysis and equation (1) to determine the linear attenuation
coefficient. Table 3-4 then compares these values to one another to determine the level of
accuracy to which this theoretical model can be leveraged to fabricate accurate and fine tuned
CT number phantoms.
Table 3-4. Comparison of linear attenuation coefficient values (µ) through a materials and CT scan approach as calculated by equations (1) and (2). Measurements were taken at 100kVp.
Polymer hA% Materials Approach µ (cm-1)
Theoretical Approach µ (cm-1)
PE
0.0 0.164 0.171 2.5 0.177 0.177 5.0 0.184 0.183 10.0 0.196 0.200 15.0 0.211 0.217 20.0 0.223 0.235 25.0 0.237 0.253
TPU
0.0 0.216 0.214 2.5 0.221 0.223 5.0 0.230 0.231 10.0 0.246 0.252 15.0 0.257 0.259 20.0 0.269 0.272 25.0 0.280 0.282
PVDF
0.0 0.313 0.311 2.5 0.316 0.315 5.0 0.328 0.320 10.0 0.339 0.345 20.0 0.379 0.389
We then aim to compare the accuracy of a commonly used CT phantom device to that of our
fabricated phantom devices as well as our theoretical approach. Figure 3-3 illustrates the
linear attenuation coefficient values exhibited by the Gammex 467 phantom and various
tissue mimicking plugs. The phantom was scanned and measured based on its CT numbers to
35
be compared against the fabricated composite phantoms of this study. Figure 4 then also
compiles the Gammex phantom data and the fabricated PE, TPU and PVDF phantom data to
illustrate a comparison between the CT number – linear attenuation coefficient trends
between the fabricated composites (PE, TPU, & PVDF) and Gammex 467 phantom. We can
see that, in general, the material approach, through our fabricated composites, quite closely
aligns to the theoretical approach at lower CT number values. However at increasing CT
numbers, the materials approach tends to deviate from the theoretical – that of increasing hA
on both TPU and PVDF composites. However, it can also be noted that the Gammex 467
also depicts this variation to a higher degree than that of the fabricated phantoms.
Figure 3–3. Comparison of TPU/hA and PE/hA composites in terms of measured CT number and linear attenuation coefficient compared to the bilinear trend exhibited by the Gammex 467 phantom with values
obtained at 100kVp.
-1000
-500
0
500
1000
1500
2000
0 0.1 0.2 0.3 0.4 0.5 0.6
CT
Num
ber (
HU
)
Linear Attenuation Coefficient (cm-1)
Materials Approach - Gammex 467
Materials Approach - Composites
CT Scan Approach
36
The fabricated composites of PE, TPU and PVDF with hA demonstrated an inclusive range
of -124.5 to 995.3 HU which is representative of the range necessary to replicate various
coronary plaque types. Although, between fabricated TPU and PVDF polymeric composite,
CT numbers between about 400 to 600 HU are not illustrated, it is expected with increasing
hA amounts to TPU or, alternatively, a polymeric density between TPU and PVDF that range
can be achieved. By determining the relationship between added hA to a polymeric base, a
high degree of CT number control can be attained. By adding hA to a polymeric base in
discrete amounts, CT number values between the given range can be fabricated.
To predict how the addition of hA can affect a given polymer, a numerical approach,
given by equations (1) through (5) is illustrated. The addition of hA to a polymer contributes
both to an increase in mass density as well as an increase in the effective atomic number
values. When predicting the linear attenuation, and further the CT number through measured
material properties, the expected CT number compared to the measured CT number are
closely aligned.
Alternatively, when compared to the Gammex 467 phantom, it can be seen there is
some deviation of the CT scan approach relationship to that of the materials approach
relationship. Notably, a simple relationship between linear attenuation coefficient and CT
number typically illustrates a bilinear phenomenon. This phenomenon has been studied and
is noted as an effect of beam hardening [37]. This beam hardening effect is a result of x-ray
variation throughout the specimen due to sample size, external environment and intrinsic
attenuation coefficients. X-ray beams are typically hardening through CT scanner filters
however, due to the polychromatic nature of the CT scanner, the resulting scan will produce
3.4 Discussion
37
an x-ray spectrum. With samples at higher effective atomic number values, this beam
hardening effect is more prominent thus resulting in an increase effective energy. This
variation of effective energy thus changes our numerical approach by means of altering our
theoretical linear attenuation coefficient values. Though within this study, we utilize our
numerical approach through a single consistent effective energy value, in actuality this value
changes with changing effective atomic number.
Through this, it can be seen in figure 3-3 that there is a notable bilinear trend, though
only applicable to high photoelectric effect effective atomic numbers. As such, the fabricated
polymeric composites did not demonstrate the bilinear trend to a large degree however, in
comparison to that of the Gammex phantom, the samples which demonstrated the bilinear
trend had photoelectric effect effective atomic numbers greater than those of the polymeric
composite.
Through this study, we have demonstrated a method in which polymeric composites using
PE, TPU, and PVDF with hA in discrete weight percentages can be viewed as
anthropomorphic to various types of coronary plaque under CT scanning. By means of a
range of -50 to 800 HU, this can encompass most plaque types from fatty plaque to ranges of
calcified plaque. Through our polymeric composites we have managed to fabricate samples
which can attenuate within this range through our control of hA additions.
The numerical method illustrated in equations (1) through (5) also has a high level of
accuracy for predicting the CT number of the fabricated polymeric composites through
material properties and linear attenuation coefficients. However, when compared against a
3.5 Summary
38
Gammex 467 standard phantom, the phantom is known to deviate from the numerical method
by means of a bilinear relationship between CT number and linear attenuation coefficient.
This effect was not prominent for the fabricated composites, however, due to relatively lower
overall values of effective atomic number values.
For future considerations, we hope to encourage studies involving the fabrication of
polymeric composites for use in coronary plaque CT phantom applications. An important
aspect to study would involve developing both geometrically and CT property accurate
phantoms specifically for coronary plaque. Due to the use of thermoplastic polymers, the
phantoms developed are able to be molded to the shape desired with ease and further set in a
solid manner. To this matter, future studies would benefit by describing the ease to which
these composites can be fabricated and integrated into a full artery phantom. This would also
include developing phantoms specifically for the desired calcification amount, including that
between 400 to 600 HU, not included in this study.
39
Tissue mimicking CT phantoms have utilized various polymers and can be fabricated and
altered to have controlled elemental compositions with increasing densities depending on
mixture compositions. Their mass density currently limits polymers, as CT phantoms.
Polyethylene, a low-density polymer, will minimally attenuate at ~ -80 HU whereas the
clinical need is to simulate adipose tissues and fatty plaques for which the attenuation can
reach -200 HU or less with lung structures as low as -500HU and the lung itself nearing -
1000HU [10], [92], [93]. Our aim is to leverage studies which have been conducted on
cellular foam polymers for low density CT phantoms such as adipose and lung tissue mimics
[15], [94], [95]. The utility of mimicking low attenuating tissues under CT imaging has been
demonstrated and low-density polymeric foams are commonly used however most foams
considered are commercially made and not specified to CT phantom applications [69], [96],
[97].
Chapter 4 Polymeric Cellular Foams for Low Density Computed Tomography Phantom Applications
40
A few select techniques have demonstrated procedures in which polymeric foams can
be fabricated with controlled material properties. Notably, in these cellular foaming
techniques, the morphological (porosity, cell size, cell density) and material (density)
characteristics can be controlled and measured [98]–[101].
Polymers can be fabricated through cellular processing to create a porous network
and decrease the overall density of the polymer through void propagation [98]. Cellular
processing can be controlled to a degree in which the pores are at a micron or nano level,
potentially imperceptible to the CT scanner. This creates a seemingly homogenous phantom
that can be manipulated to present a wide range of material attenuations to mimic a diverse
array of tissue environments; from densely calcified tissue to adipose or lung tissue, without
being perceived as a porous network.
By detailing a fabrication technique in which CT phantoms can be custom-made
specific to the desired tissue, studies to improve CT imaging diagnostics can proliferate. As
well, a consistent fabrication technique will create phantoms with reliable and expected CT
values that may not be attained by native tissues due to tissue variance between patients.
The last two decades have seen a rapid increase in utilization of CT for a wider number of
clinical applications; and these have increased the need for accurate tissue characterization
and for a deeper understanding of tissue attenuation response to multispectral CT [6], [102]–
[105]. However, during diagnosis, it has been reported that a CT machine dependent variance
of 5% in attenuation measurements can occur [96]. Furthermore, the effect of reconstruction
algorithms can vary the CT attenuation by up to 9.4% [96]. As such, there is a real need to
4.1 Motivation
41
more fully understand the influence of various CT parameters on accurate tissue
characterization and this, in turn has driven the demand for anthropomorphic CT phantoms
with the ability to mimic various tissues to a high degree of accuracy. Through this, various
post-processing tools and imaging techniques can be implemented without the need for
cadaveric or live tissue specimens. Clinical applications and studies have demonstrated the
need for accurate and specific CT parameters. Examples include characteristic studies of
adrenal adenomas [102], coronary artery disease [6], and development of CT tools [106].
Furthermore, a number of studies have detailed the use of commercial phantoms or material
substitutes for CT electron density calibration and treatment planning systems. However
there are a limited number and CT characteristic ranges of phantoms available which has thus
seen an increase in studies involving low density tissue-equivalent substitutes [21], [22],
[70], [94], [95], [15] .
Accurate and specific CT numbers are essential in diagnostic imaging for tissue
characterization for example; differentiation of adrenal and renal nodules or masses[102],
characterization of non-calcified coronary artery plaque[6], and determination of the
vascularity of a non-calcified mass through measurement of tissue enhancement following
injection of intravascular iodinated contrast. Accurate CT numbers are also critical to
development of advanced CT visualization tools used in 3D applications such as virtual
applications[106] including colonoscopy, bronchoscopy and 3d angiography.
Furthermore, several studies have detailed the use of commercial phantoms or
material substitutes for CT electron density calibration and for use in treatment planning
systems. However, detailed evaluation demonstrates that there are a limited number of low-
density phantom devices available and these have a limited range of values compared to the
42
larger range in the materials manufactured in this manuscript.[22], [70], [94], [95], [15],
[107] For example, Landry et al.[108] and Bazalova et al.[109] detail the various properties
of Gammex phantom inserts. It is capable of mimicking a wide range of various mid to high
density tissues and, thus, CT numbers, however it is limited in the availability of low density
CT phantoms. As such it can be noted that specific tissues can be mimicked by means of the
Gammex phantom, however, values less than -100 HU and greater than -500 HU are not
available. This range can be important, for example, when attempting to mimic a highly
accurate adipose CT phantom. As illustrated in Kim et al.[110], the most prominent range of
adipose tissue is between -190 and -30 HU, and further values found as low as -250 HU. In
addition, lung structures have measured CT attenuation as low as -500HU and lung
parenchyma measures ~ -1000HU.[10], [92], [93]
CT imaging phantom devices have proven to be beneficial in improving CT
diagnostic techniques. Though commercial phantoms are available with tissue mimicking
properties, there is a lack of low density tissue specificity and variety. This study proposes a
method for the fabrication of various low density tissue mimicking CT imaging phantoms.
By illustrating the fabrication technique, material properties are controlled and assessed
against characteristic CT imaging properties, most particularly, the CT number in Hounsfield
Units (HU). Thus, this study can allow for highly specific CT imaging phantoms to be
fabricated and tailored matching the properties of the desired low density tissue.
The aim of this study is to utilize a batch cellular foaming technique to fabricate
polymeric CT phantoms to create specific foam material properties of porosity, average cell
size, cell density, and mass density. CT number values were assessed against these
characteristic foam material properties to determine the relationship between the two and the
43
level of consistency and accuracy. Material characteristics were examined by means of a
morphological assessment under scanning electron microscopy (SEM) and ASTM standards
for density measurements. The material characteristics of each CT phantom material were
then compared to the measured CT number using varying kVp.
The foaming technique utilized for our fabrication method was a batch foaming technique
using supercritical CO2. As described in Jacobs et al [112]. This phenomenon involves the
following setup which we have leveraged for our experimentation:
Figure 4–1. Foaming technique setup with reference to Jacobs et al. [112] but controlled for our purposes of foaming TPU.
Firstly, polymeric samples are placed in a high pressure environment with CO2
introduced. The high pressures allow the polymer to become saturated with CO2 as well as
decreasing the glass transition temperature (Tg). This temperature determines whether the
polymer is in a glassy, or brittle state at temperatures below Tg or in a rubbery, or soft state
4.2 Materials and Methods
4.2.1 Foaming Procedure
44
above Tg. Decreasing the Tg gives the polymer a soft state allowing CO2 to effectively diffuse
within the throughout the polymer. Following this is an induced thermodynamic instability
through rapid pressure release. This leads to an instantaneous oversaturation of CO2 within
the polymer. Further, a heated environment is often necessary to nucleate and grow cells,
then allowing CO2 to diffuse out of the polymer, if the Tg has not been reached. Once the
polymer has returned to a value below its Tg, nucleation and growth will stop.
The heated water bath had a selected temperature of 80oC based upon an optimal
temperature; low enough such that the degree of foaming can be controlled but high enough
to induce desired foam characteristics at reasonable lengths of time (0.5 min to 10 min).
The polyester-type pellet form thermoplastic polyurethane (TPU) (Desmopan 385E; Bayer
AG, Germany) was used as the base polymer for fabrication. For the foaming procedure,
TPU was selected due to the potential for a wide range of foams produced through various
studies for TPU and its ease and known conditions for batch foam processing [60], [99],
[100], [111].
Table 4-1. Pure TPU material properties illustrated as atomic composition and physical properties.
Atomic Composition (Weight %) Physical Properties H (Z = 1) C (Z = 6) N (Z = 7) O (Z = 8) ρ (g/cm3) ρe Relative ρe
9.07 64.35 6.00 20.57 1.200 3.82E+23 1.145
If the average cell size is greater than that of the limiting spatial resolution of the CT
image, the phantom will appear heterogeneous, with distinct regions of pores and solid
samples. Thus, the aim is to fabricate phantoms that are viewed as homogenous under CT
4.2.2 Material Fabrication
45
scans, with the largest cell sizes falling below the limiting spatial resolution of the CT
imaging chain defined by the defined field of view (DFOV) of each scan.
An aluminum mold with a capacity of 16 samples was used together with a Carver
Bench Top Standard Heat Press to mold the TPU to a desired sample shape. The heated
plates were set to 200oC and the TPU was exposed for 15 minutes. The heated TPU was then
compressed to 7 metric tons for another 5 minutes with the plate remaining heated.
Following compression, the samples were immediately cooled in a room temperature water
bath for 1 minute and then removed carefully from the mold. Sixteen samples were made per
batch, each cylindrical sample measured 12.5mm in diameter by 6.5mm in height.
To create the cellular foam from the molded TPU, the samples were introduced, in
sets of 4, to an aluminum pressure chamber. The batch foaming processing technique was
then completed on all the samples under different conditions. The batch foaming process
consists of: sealing the sample in an aluminum chamber under high pressures, rapidly
releasing the pressure, thus introducing a thermodynamic instability, and then immediately
placing the samples into a heated water bath. TPU samples were inserted into the pressure
chamber at 800psi for 24 hours. The samples were then exposed to a heated water bath of
80oC for either 0.5, 1, 1.5, 2, 5, or 10 minutes and then immediately placed in a room
temperature water bath to cool. The completed samples were then placed in a room
temperature vacuum chamber for 24 hours to allow for adequate drying and removal of
internal moisture. Following the foaming procedure, sample dimensions were left
unrestricted allowing for slight increases in size.
46
The phantoms were scanned in a wide volume CT (320 MDCT AquilionONE; Toshiba
Medical Systems, Japan) after calibration of the CT unit and CT numbers were measured
using ImageJ (ROI Manager, NIH, USA) on post-processed images. The scanning conditions
were set at 160 x 0.5mm detector confirmation, a gantry rotation of 0.5 second, fixed tube
current of 50mA and tube potentials of 80, 100, 120 and 135kVp. The display field of view
(DFOV) used was 30cm. The images were reconstructed using a single filter kernel (FC08,
standard for beam hardening) [113] as 1mm sections at 0.5mm intervals. Additionally, figure
4-2 presents the arrangement in which the phantoms were scanned. All phantoms were
scanned simultaneously under each of the mentioned tube potentials. Scans were completed
with samples oriented between enclosed saline solution environments.
Figure 4–2. CT scanning arrangement of TPU foam sets A through E. Included are TPU100 samples acting as pure TPU for reference during scan. Labeling from A through E represent TPU sets A through E respectively.
4.2.3 Computed Tomography Number
47
Table 4-2. Outline of processing technique utilized for TPU foam sets A through E. The variation in processing condition is based upon the length of time upon which the TPU samples are introduced into the heated water
bath.
Sample Foaming Chamber Water Bath Pressure (psi) Time (hrs) Temperature (oC) Time (min)
TPU Set A 800 24 80 0.5 TPU Set B 800 24 80 1 TPU Set C 800 24 80 2 TPU Set D 800 24 80 5 TPU Set E 800 24 80 10
Our aim was to correlate the measured CT number to distinct material properties which both
characterize cellular foams and affect the CT number. To direct our study, we focused on
commonly measured foam characteristics of physical density, porosity, average cell size and
cell density.
Density measurements were carried out according to the water-displacement method
(ASTM Standard D792 – 13) to characterize mass density and porosity.
To characterize the TPU foams, the samples underwent scanning electron microscopy (SEM)
(JSM 6060; JEOL, Japan) with image analysis using a common morphological characterizing
tool, ImageJ [100], [101], [114]. From this, cell density and average cell sizes were
measured.
Firstly, the porosity can then be described by Eqn. 7 [99], [101], [115]:
𝜙! = 1− 𝜌!"#$% 𝜌!"#$
∗ 100% (Eqn. 7)
where 𝜙v is the porosity, or void fraction, measured in percentage, 𝜌foam is the foamed
density measured and 𝜌solid is the solid un-foamed density of the sample.
4.2.4 Foam Characterization Methods
48
Following this, the cell density and cell sizes were interrogated through SEM imaging
and further image analysis by means of the ImageJ software (ROI Manager; NIH, USA). The
ImageJ software is able to measure the number and average size of cells within a given
image, thus giving a representation of the sample as a whole. Once analyzed, the average cell
size of the sample could be shown while the cell density of the sample was calculated as
detailed in Eqn. 8 [101], [115], [116]:
𝐶𝑒𝑙𝑙 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 =𝑁𝑜. 𝑜𝑓 𝐶𝑒𝑙𝑙𝑠
𝐴𝑟𝑒𝑎
!! 𝑆𝑜𝑙𝑖𝑑 𝐷𝑒𝑛𝑠𝑖𝑡𝑦𝐹𝑜𝑎𝑚 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 (Eqn. 8)
The attenuation of each TPU sample was averaged from 3 measurements per image with 3
image measured per sample. Each TPU sample set consisted of 6 samples, thus 54
measurements were averaged per CT number value and the values at different kVp. All error
bars are indicative of the maximum and minimum values measured from the data set. A
comparison was made between kVp and CT number to determine if any change in trends
were discernible with changing kVp. As illustrated in figure 4-3, the fabricated TPU foams
do not illustrate a large dependence on kVp.
4.3 Results
49
Sample CT Number (HU)
80kVp 100kVp 120kVp 135kVp TPU Set A 58.76 73.12 82.13 83.81 TPU Set B -82.73 -65.38 -57.50 -52.39 TPU Set C -171.20 -157.74 -151.15 -145.20 TPU Set D -199.78 -189.09 -182.84 -178.02 TPU Set E -225.36 -212.34 -205.44 -200.96
Figure 4–3. Comparison of TPU foam sets for changing kVp values to CT number.
The effect of increasing kVp and, thus, keV upon CT scanned specimen typically sees
an increase in CT number. However, in reference to decreasing electron and physical
densities, this variation of CT number with increasing kVp becomes less prominent. In fact,
at relative electron densities to water below 1, a reverse trend of increasing CT number with
increasing kVp is noted [33]. For our study, due to the similarity in trend, our analysis
focused on a CT number analysis at 100kVp.
-250
-200
-150
-100
-50
0
50
100
70 80 90 100 110 120 130 140
CT
Num
ber
(HU
)
Tube Voltage (kVp)
TPU Set A TPU Set B TPU Set C TPU Set D TPU Set E
50
Table 4-3. CT number values measured in Hounsfield Units (HU) for solid TPU and subsequent TPU foam samples. Additional data of measured density through the ASTM standard
TPU Set Sample CT Number 100kVp (HU)
TPU Set A 1 84.67 2 84.77 3 49.91
TPU Set B 4 -47.62 5 -97.15 6 -51.37
TPU Set C 7 -142.69 8 -145.94 9 -184.60
TPU Set D 10 -186.03 11 -194.58 12 -186.65
TPU Set E 13 -203.53 14 -226.25 15 -207.24
Due to a similarity in trend, our analysis focused on a CT number analysis at 100kVp.
The cell morphology of the various fabricated TPU foam sets was measured and calculated
by means of eqn. 1 and 2. Cell density measurements also required a morphological analysis
of the foam using the JEOL SEM to image the samples, whereas the cells were counted and
measured with the ImageJ software. An area of about 0.04mm2 was examined per image to
attain both the number of cells and average cell size, representing that of the entire sample.
4.3.1 Effect of Cell Morphology on CT Number
51
The relationships between porosity, cell density and average cell size against CT number at
100kVp were evaluated and is shown in figure 4. It can be noted that there is a negative
influence of all foam characteristics to measured CT number.
Figure 4–4. Relationship exhibited between CT number (HU) and porosity (%), cell density (cells/µm2), and average cell size. An increase in porosity, cell density and average cell size all saw general trends of decreasing
CT number. The points represent the mean where as the error bars represent the maximum and minimum measured CT values
.
-300
-250
-200
-150
-100
-50
0
50
100
150
0% 10% 20% 30% 40%
CT
Num
ber
(HU
)
Porosity (%) -300
-250
-200
-150
-100
-50
0
50
100
150
0.E+00 1.E-04 2.E-04 3.E-04 4.E-04
CT
Num
ber
(HU
)
Cell Density (cells/µm2)
-300 -250 -200 -150 -100 -50
0 50
100 150
4 5 6 7 8 9 10 11
CT
Num
ber
(HU
)
Average Cell Size (µm)
52
Figure 4-5 presents SEM images of the fabricated foam sets illustrating both the average cell
size The fabricated foams demonstrated both a maximum cell size that was smaller than the
highest spatial resolution of the CT imaging chain of 400 x 400µm.
Figure 4–5. SEM Images illustrate the morphological characteristics of the controlled cellular foam sets of (1) Set 1, (b) Set 2, (c) Set 3, (d) Set 4, (e) Set 5. Differences in foam characteristics are introduced by means of the
processing parameters illustrated in table 4-2.
Values of measured density are plotted against measured CT number with a positive
relationship noted between material density and CT number. A linear regression trend line
4.3.2 Effect of Cell Size on CT Number
4.3.3 Effect of Density on CT Number
53
was applied to determine the numerical linear relationship between CT number and physical
density as illustrated in figure 4-6.
Figure 4–6. Relationship exhibited between CT number (HU) and mass density where an increase in CT number results in an increase in CT number. The points represent the mean where as the error bars represent the maximum and minimum measured CT values. Linear regression analysis was also completed and illustrated as
the noted trend line.
The results of this study have demonstrated the ability of TPU foams to conform to a wide
range of CT number values to mimic various low density human tissues. The morphological
key metrics, which include porosity, cell density and average cell size, show general trends in
which an increase in foamed characteristics demonstrate a decrease in CT number. These
characteristics are then limited by the resolution of the CT scanner in which we aim to have a
homogenous sample. Hence, the cell sizes of the samples should be less than the minimum
y = 1155.7x - 1157.6 R² = 0.97832
-300
-250
-200
-150
-100
-50
0
50
100
150
0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10
CT
Num
ber
(HU
)
Density (g/cm3)
4.4 Discussion
54
pixel size of a CT image, which was also demonstrated through a morphological analysis
through SEM imaging. These images are thus seen as homogenous under CT scanning.
The error bars associated with figures 4-4 and 4-6 are indicative of the range of CT
numbers measured from each sample likely introduced due to noise. The foam cells, as noted
by the SEM images, are generally of the same size, however the range of cell sizes and
distribution due to the fabrication technique can exhibit a possible range in CT number
within the sample.
It was found that the prominent trend was decreasing CT number with increasing
characteristics of foamed properties. This can be described by increasing porosity, cell
density and average cell size, while additionally we find a decrease in physical density. These
properties can be attributed to two concurrent factors of an increase in void space acting as
air and a decrease in overall physical density.
The effect of void space can first be attributed to the porosity, cell density and the
average cell size. Thus we are likely to find a greater volume of the sample that has a CT
number of -1000 HU contributing as air. In addition, a greater amount of void space results in
a lower density due to an increased amount of polymeric expansion. The combined effect of
the CT scanner defining void space as air and an overall decrease in the density of the sample
contributes to a decrease in measured CT number.
However, when considering average cell size, competing factors can be considered.
Though an increase in average cell size would likely contribute to a greater amount of void
space, similar average cell sizes can have different cell densities thus varying the cell wall
thickness. High average cell sizes and cell densities can result in overall thinner cell walls
compared to that of the same average cell size but with low cell densities [59], [110], [115].
55
Our results generally showed increasing cell sizes in correspondence with increasing cell
densities though there was some variation in data, as demonstrated in figure 4-4.
Furthermore, when analyzing equations 7 and 8, a relationship can be noted between a ratio
of densities between solid and foamed density. This can then be attributed to the general
trend similarities seen of decreasing CT number with increasing foam properties. The
decrease in polymer density contributes to both cell density and porosity thus decreasing CT
number. Additionally, some of the discrepancy then seen with the relationship between
average cell size and CT number can be noted due to the competing effects of cell size and
cell density. With increasing cell size one can expect a decrease in cell density again as seen
through equation 8.
Ideally, the cell size we would aim for would be less than the pixel size. A typical CT
scanner sees the pixel size at no less than 0.4mm x 0.4mm. Thus, the average cell size would
be optimally produced to be less than this to introduce the CT scan as homogenous if desired
as such. With regards to cell density and porosity, we would select these properties based
upon the desired CT number. As mentioned, the lower the CT number aimed towards, the
higher the cell density and porosity necessary. Additionally, having a homogenous structure
through CT scans depends on the tissue and need for such, for the given applications of fatty
plaque and adipose, it would be desired that the scan illustrate the phantom as homogenous.
With reference to table III, there is some variation in the CT numbers of samples
within the same foam set. The discrepancy can be attributed to the wait period between
retrieving the sample from the heated water bath to submerging the sample within the cold
water bath. Since samples are retrieved one at a time from the heated water bath, certain
samples are left in the water bath for a slightly longer period. At longer heated water bath
56
submerging times, the foaming properties are not as prominently affected in compared to
short heated water bath periods such as 30 - 60 seconds.
Furthermore, a current limitation is the fact that a single polymer system was
analyzed. Though the polymer is expected to have wide ranging foam properties, we expect
that further research will include and desire the use of various polymers utilizing foam
processing. To do so, water bath temperature and foaming time will necessitate optimization
based on the polymer used essentially based upon the Tg of the polymer system.
Lastly, one of the main formulaic concepts behind this research is based upon the
relationship between CT number and mass density. Illustrated by a study conducted by
Watanabe et al.[10], mass density is directly proportional to attenuation coefficient, which, in
turn is also proportional to CT number. Porosity and cell density as expressed through
equations (1) and (2), are both inversely proportional to the foam density (ρfoam). As
mentioned earlier and supported by the effect of density, an increase in porosity and cell
density are a result of a decrease in foam density. This is because our processing results in a
decrease in mass density due to the addition of voids within the sample. Any of these effects,
in turn, result in a decrease in CT number, relative to the relationship exhibited between
physical density and CT number. Though various constituents also affect the CT number,
most notably elemental composition, the fabricated foam phantoms are constructed of the
same material though at different mass densities.
This proof of concept study demonstrates that custom design and the manufacturing of
polymeric foam CT phantoms to meet rigorous imaging tasks is feasible and controllable
4.5 Summary
57
with a high degree of precision using the described fabrication technique. Low density CT
imaging phantoms were created with the ability to have tailored CT number through the
controlled fabrication technique illustrated.
The concept of using commercially available foam phantoms has been frequently
studied to determine whether different foams can be differentiated using CT imaging, The
current study demonstrates a method in which foams can be fabricated and controlled to a
fine degree to achieve a desired CT number. This approach is important when aiming to
mimic various types of adipose and low-density tissues. The materials analysis approach in
which foam characteristics are measured against CT number illustrates how porosity, cell
density, and mass density affect the HU value. The presented fabrication technique can be
used to control the necessary material properties and tailor the CT number to the desired
characteristics for the CT phantom. It is suggested that further studies include the
advancements in CT scanning with multispectral analysis and dual energy decomposition
utilizing varied foam fabrication techniques for low-density CT phantoms. It is further
suggested that follow up studies conduct large scale fabrication processes utilizing similar
fabrication techniques on larger equipment to create anthropomorphic phantom devices such
as the lungs.
58
The fabrication of CT phantom devices allows researchers to have the accessibility to
customized size, shape and tissue mimicking properties. With the flexibility and choice of
material for a phantom to mimic specific human tissues and systems, it is important that the
fabricated phantom have a high degree of CT consistency and predictability based upon its
measured material properties. Due to CT calibration, a phantom device typically requires a
water bath or water-equivalent environment which often is a rigid poly(methyl methacrylate)
(PMMA) surrounding. Both scanning environments, though standard for CT phantom
scanning, can limit the potential complex, tortuous, and non-rigid phantom designs.
CT phantom devices are also limited to which they cannot be scanned accurately in
only air surrounding environments. Though scanning without any surrounding water bath or
Chapter 5 Effect of Scanning Medium on the In-house Fabricated Polymeric Composite CT Phantom Devices
5.1 Introduction
59
water-equivalent media would allow for an entirely flexible design of CT phantom, it is
expected that when scanning these phantoms in air that significant inaccuracies will occur. It
would be expected that within air environments, the characteristic CT number values in HU
should vary from water bath environment scans due to the polychromatic nature of CT
scanning.
Any deviations that result are often described through the probabilistic nature of the
CT scanner by the processes of photon production, interaction, and detection [9], [36], [118].
This variation is reflected upon the attenuation coefficient (µ) and, thus, the measured CT
number of the phantom device. This is due to the variation upon the CT scanning conditions
and the CT machine used [22], [33], [79], [105], [119], [120]. Since CT scanning is of
polychromatic nature, depending on the path of the beam, lower energy x-rays can be
removed from the spectrum prior to interacting with the phantom. Under CT scanning, this
can occur under two methods. The first is the filters applied immediately to the beam for
modern CT scanners, thus removing lower energy x-rays. This filter will vary by the CT
scanner and protocol used [103]. The second method is a beam hardening effect. This effect
is well studied and is understood to change the attenuation of the phantom [64], [121], [122].
This beam hardening effect is introduced by applying a surrounding media. The scanning
environment will then further remove low energy x-rays which is often a water or water-
equivalent surrounding environment [11], [120], [123]. Since CT scanners are clinically
calibrated to water or a water-equivalent material, a similar surrounding media must be
introduced around the phantom for CT number accuracy and consistency. The lack of
scanning media may result in a heterogeneous scan to which the outer edge of the sample
60
may have a higher level of x-ray interaction while middle sections may result in less
attenuation due to beam hardening throughout the solid.
Fabricated CT phantoms scanned under a water bath environment are compared to
scans of an enclosed saline solution surrounding and open air environments. The comparison
involves differences in measured CT numbers between scans in the different media. These
differences are then further assessed through the likeness of calculated linear attenuation
coefficient (µ) to measured CT numbers scanned in the different media. The definition of µ
includes both measured material properties and effective energy calculated through a non-
linear regression analysis of CT numbers and electron density values. The differences in
measured CT numbers between water bath environment scans and saline solution shows the
least variation at higher applied voltages (120 and 135kVps) as well as having effective
atomic numbers (ZT and ZR) closer to that of water. As such it is possible that utilizing an
enclosed saline bag surrounding environment alternatively to a water bath or rigid water-
equivalent will allow for accurate scans of fabricated CT phantoms.
For this study, it is proposed that accurate phantom scans can be conducted on
fabricated CT phantoms by means of a surrounding environment of enclosed bags of saline
solution. As opposed to the rigid environments of the water bath and water-equivalent
polymer surrounding environments, the saline solution does not submerge the sample in
solution, nor does it limit the shape and conformation of the fabricated CT phantoms.
CT phantoms can be fabricated such that it conforms precisely to the study at hand.
The comprehension of CT numbers in relation to material properties can allow for a high
level of anthropomorphic behavior, thus allowing studies to mimic various human
61
environments accurately. However, CT phantoms typically are placed under a water or
water-equivalent environment with reference to CT calibration matching that of the human
body. Fabricated CT phantoms on the other hand may perform ideally under non-aqueous or
non-rigid surrounding media due to porosity, fragile nature, or complex shape. Thus, ideally
scans of fabricated CT phantoms would prefer to avoid a water bath or any rigid water-
equivalent surrounding medium. This study proposes the use of saline solution in enclosed
bags allowing for placement of the surrounding media in any conformation without
submerging the phantoms in a wet environment.
In-house fabricated CT phantoms have been proposed and scanned under a
conventional submerged water bath environment and, in contrast, both saline solution and
open air environments. The in-house fabricated CT phantoms are polymeric composites with
ranging material properties with expected CT number values ranging between -100 and 1200
HU, ideally to mimic widely varying biological tissue. This study will contrast the
effectiveness of fabricated CT phantoms under a surrounding saline solution and open air
environments as compared against scanning under a submerged water bath. We evaluate a
comparison between the CT numbers measured as well as the accuracy of measured material
properties to the measured CT numbers.
Three polymeric base materials were selected in combination with a single additive in order
to fabricate CT phantoms with the ability to mimic a wide range of CT numbers encountered
in clinical applications. These applications include various adipose, soft tissue and hard
5.2 Materials and Methods
5.2.1 Material Selection & Processing
62
tissues with CT numbers ranging between that of -100 and 1100. Polyethylene (PE),
thermoplastic polyurethane (TPU), and polyvinylidene fluoride (PVDF) were all selected
based on their differing densities and material properties influential to evaluating CT
numbers. Furthermore, the addition of hA in discrete amounts allowed for a fine control of
mass and electron density and, thus, the CT number.
Polymeric composite CT phantoms were fabricated by means of the three polymers: low
density polyethylene (LDPE; Nova Chemicals Novapol, LC-0522-A), polyester-type
thermoplastic polyurethane (TPU; Desmopan 385E), and polyvinylidene fluoride (PVDF;
ARKEMA, Kynar 720). Polymers were processed as received in pellet form. The additive,
hydroxyapatite (hA; 677418, Sigma-Aldrich) nanopowder with controlled particle sizing of
less than 200nm was used in combination with the polymeric bases.
Polymers were fabricated with discrete amounts of hA at 2.5, 5, 10 and 20 weight percentage
with all sets containing a pure polymer type with no hA added. Polymer and hA additions
were melt compounded by means of a 15 mL benchtop twin screw microcompounder (DSM
15, DSM Xplore). The polymer and hA were added simultaneously to the rotating twin screw
compounder at 130oC, 200oC, 230oC for PE TPU and PVDF respectively. Combinations
were processed at 100 rpm for 15 minutes and then extruded to be quenched immediately
from the outlet in room temperature water.
Extruded composites were processed into identical cylindrical shaped samples for CT
measurement by means of compression molding (12 Ton Carver Hot Press) into a disc
shaped mold (12.8-cm height by 0.65-cm diameter).
5.2.2 Computed Tomography Scanning Conditions
63
The phantoms were measured in a wide volume CT (320 MDCT AquilionONE; Toshiba
Medical Systems, Japan) with CT numbers measured using ImageJ on post-processed
images. The scanning conditions were set at 160 x 0.5mm detector confirmation, a gantry
rotation of 0.5 seconds, fixed tube current of 50mA a tube potential of 100kVp. The images
were reconstructed using a single filter kernel with 1mm sections at 0.5mm intervals.
The fabricated phantoms were introduced into the three environments of a water bath,
saline solution and air. For the water bath environment, the phantoms were submerged into
the centre of a 10cm length water bath. For the saline solution environment, the saline was
enclosed in 5cm bags in which these bags were placed above and below the phantoms. For
the air environment, the samples were placed with no surrounding environment.
Density measurements were carried out by ASTM Standard D792 – 13 by the measurement
of plastic density through displacement. The elemental composition was assessed as weight
percents per element for each polymeric composite..
Spherical regions of interest (ROI) measuring 10mm diameter were prescribed within the
centre of the sample, so that the outer perimeter was excluded to remove any CT variation
due to beam hardening at the perimeter of the phantom. The CT number in Hounsfield Units
(HU) was noted. Each CT number dataset was constructed by an average of 27 data points, 3
samples representing each polymer composite type with 3 measurements made per slice with
5.2.3 Physical Properties
5.2.4 Data Analysis
64
3 slices per sample. Concurrently, samples were measured under 4 different x-ray tube
voltages of 80, 100, 120, and 135kVps. Density value measurements were taken by 3
samples with 3 continuous measurements
Each CT number dataset was constructed by an average of 27 data points, 3 samples
representing each polymer composite type with 3 measurements made per slice with 3 slices
per sample. The same sample sets were measured in air, water bath, and saline solution
environments. Concurrently, samples were measured under 4 different tube voltages of 80,
100, 120, and 135kVps. Density value measurements were taken by 3 samples with 3
continuous measurements made each. CT numbers were measured with a centered field of
view (FOV).
Table 5-1 illustrates the weight percent of all polymeric composite phantoms fabricated.
Further, the material properties of measured mass density, calculated effective atomic
numbers and relative electron density are also presented. CT number measurements are
illustrated in table 5-2 and are compared by means of material composition and CT number
measurements made in air and water environments.
5.3 Results
5.3.1 CT Number and Material Property Measurements
65
Table 5-1. Weight percentages of H, C, N, O, F, P, and Ca within polymers and polymeric composites fabricated for CT phantom studies. Measured densities are illustrated for each polymeric composites. Effective
atomic numbers (ZƮ and ZR) as well as relative ρe were calculated by means of equations (2) through (5).
H C N O F P Ca
PE100% 14.37 85.63 0.00 0.00 0.00 0.00 0.00 0.862 5.54 5.17 0.886PE95% / hA5% 13.66 81.35 0.00 2.07 0.00 0.92 1.99 0.942 7.61 5.89 0.961PE90% / hA10% 12.95 77.07 0.00 4.14 0.00 1.85 3.99 0.974 8.79 6.54 0.989PE80% / hA20% 11.54 68.50 0.00 8.28 0.00 3.70 7.98 1.054 10.40 7.73 1.055TPU100% 9.07 64.35 6.00 20.57 0.00 0.00 0.00 1.167 6.34 5.98 1.145TPU95% / hA5% 8.63 61.13 5.70 21.61 0.00 0.92 1.99 1.205 8.04 6.61 1.177TPU90% / hA10% 8.19 57.92 5.40 22.66 0.00 1.85 3.99 1.249 9.12 7.20 1.215TPU80% / hA20% 7.30 51.48 4.80 24.74 0.00 3.70 7.98 1.295 10.64 8.29 1.249PVDF100% 3.15 37.51 0.00 0.00 59.34 0.00 0.00 1.743 7.99 7.65 1.570PVDF95% / hA5% 3.00 35.64 0.00 2.07 56.37 0.92 1.99 1.772 9.11 8.15 1.596PVDF90% / hA10% 2.85 33.76 0.00 4.14 53.40 1.85 3.99 1.770 9.96 8.62 1.594PVDF80% / hA20% 2.56 30.01 0.00 8.28 47.47 3.70 7.98 1.865 11.24 9.49 1.678
ZRRelative ρe
ZƮρ (g/cm3)
SampleWeight Percent per Element (% )
Table 5-2. Comparison of measured CT numbers for the same polymeric composites measured under separate conditions of air, saline solution and water bath environments.
CT Number (HU)
Sample Water Bath Saline Solution Air Medium
80kVp 100kVp 120kVp 135kVp 80kVp 100kVp 120kVp 135kVp 80kVp 100kVp 120kVp 135kVp
PE100% -125.0 -108.8 -95.0 -91.7 -147.4 -124.5 -111.7 -103.5 -173.4 -150.7 -139.3 -130.5 PE95% / hA5% -62.8 -52.3 -44.5 -48.1 -69.6 -61.7 -58.0 -51.4 -78.5 -68.4 -62.7 -65.7 PE90% / hA10% 32.6 19.0 19.2 14.7 34.3 23.7 18.7 9.4 41.4 32.7 27.8 15.6 PE80% / hA20% 250.1 202.5 174.5 160.1 257.6 203.0 171.8 160.3 297.0 236.4 207.5 188.6 TPU100% 103.5 114.1 120.7 123.0 80.9 95.1 104.1 110.4 60.1 81.8 92.1 97.3 TPU95% / hA5% 200.2 196.2 191.9 191.7 186.4 185.2 182.4 180.9 203.0 196.1 191.5 186.2 TPU90% / hA10% 333.6 294.3 269.9 258.6 328.7 289.9 266.8 253.6 348.2 321.3 282.6 268.0 TPU80% / hA20% 434.6 370.9 329.3 306.5 466.4 392.5 346.8 320.3 510.6 435.4 383.0 350.6 PVDF100% 622.4 581.1 543.1 512.8 634.0 591.4 551.7 509.6 652.0 599.4 555.3 521.9 PVDF95% / hA5% 797.6 708.1 646.7 601.0 723.1 642.0 582.0 546.6 841.3 743.3 672.2 619.1 PVDF90% / hA10% 879.0 765.2 690.4 638.7 889.1 765.4 689.2 641.0 991.4 850.8 761.5 697.4 PVDF80% / hA20% 1183.0 983.9 867.9 791.0 1195.7 995.3 869.2 794.1 1372.8 1154.3 987.1 896.8
66
5.3.2 CT Number Comparison
The CT number values measured from the acquired CT scans performed using the water
bath, saline solution and air environment were compared. The difference in CT number
(ΔCT) varied between compositions however were more closely aligned between scans taken
in the water bath and saline solution environments as compared with the water bath and air
environment scans. The increase in added hA to each polymer results in increased atomic
number and consequently in increased ΔCT.
Table 5-3. Comparison of differences in CT number between the referenced water bath measurements and saline solution or air. Differences in CT number are illustrated for all polymeric composite CT phantoms
fabricated at 80, 100, 120 and 135kVps.
80kVp 100kVp 120kVp 135kVp 80kVp 100kVp 120kVp 135kVp
PE100% -22.41 -15.75 -16.72 -11.75 -48.47 -41.87 -44.30 -38.78PE95% / hA5% -6.84 -9.39 -13.54 -3.30 -15.77 -16.10 -18.23 -17.55PE90% / hA10% 1.71 4.71 -0.52 -5.32 8.77 13.71 8.63 0.83PE80% / hA20% 7.52 0.54 -2.69 0.19 46.95 33.91 33.03 28.52TPU100% -22.64 -18.98 -16.56 -12.60 -43.47 -32.31 -28.54 -25.68TPU95% / hA5% -13.86 -10.95 -9.49 -10.77 2.77 -0.08 -0.41 -5.45TPU90% / hA10% -4.89 -4.47 -3.09 -5.00 14.58 26.97 12.74 9.43TPU80% / hA20% 31.82 21.52 17.53 13.80 76.04 64.49 53.77 44.12PVDF100% 11.53 10.26 8.52 -3.12 29.53 18.21 12.12 9.16PVDF95% / hA5% -74.44 -66.10 -64.65 -54.35 43.78 35.21 25.56 18.13PVDF90% / hA10% 10.07 0.18 -1.22 2.28 112.35 85.58 71.07 58.63PVDF80% / hA20% 12.77 11.38 1.28 3.17 189.80 170.38 119.22 105.89
ΔCT (HU) = CTWater - CTSaline ΔCT (HU) = CTWater - CTAirSample
To quantify a comparison between the measured material properties prevalent to CT
scanning within water, saline solution and air environments, eqn. (1) through (5) can be
5.3.3 Material Property Comparisons
67
leveraged. Material properties illustrated in table 5-2 can represent a means to calculate the
linear attenuation coefficient.
Linear attenuation coefficient is a property derived upon both material properties and
effective energy (Eeff) values. Eeff values were determined through a non-linear regression
analysis conducted through the SigmaPlot program (SPSS Inc., Chicago, IL, USA). Eqn. (1)
through (5) were utilized and re-arranged in the following manner:
(𝐶𝑇 𝑁𝑢𝑚𝑏𝑒𝑟 + 1000)1000
= 𝜌! 𝑎 𝑍!
!
𝐸! + 𝑏 𝑍!!
𝐸! + 𝑐(𝐸)
𝜌!,! 𝑎𝑍!,!!
𝐸! + 𝑏𝑍!,!!
𝐸! + 𝑐(𝐸) (Eqn. 9)
where a ratio of µ over µw was conducted which then followed a combining of eqn. (1) and
(2). By applying the values illustrated in table 5-2 as well as the measured CT numbers in
table 5-3, the remaining E constant can be determined through the non-linear regression
model. The results are illustrated in table 5-4.
Table 5-4. Comparison of assessed Eeff values through water bath, saline solution and air environments at 80, 100, 120, and 135 kVps.
80kVp 100kVp 120kVp 135kVpWater Bath 61.1 71.4 82.2 91.5Saline Solution 58.6 69.4 82.9 95.8Air 53.0 61.0 69.8 78.7
MediumEeff (keV)
By then utilizing the Eeff values determined in table 5-4, we can then calculate the µ
values within water bath, saline solution and air environments. Ultimately, a comparison can
be made between the referenced water bath scans and the saline solution or air environment
scans. The differences are illustrated as a percent deviation from the µ values calculated from
the water bath environment.
68
Table 5-5. Comparison of u values as calculated through equation (2). Comparisons are made as a % deviation from water bath environment scans and is conducted for both saline solution and air environments at 80, 100,
120 and 135kVps.
80kVp 100kVp 120kVp 135kVp 80kVp 100kVp 120kVp 135kVp
PE100% 1.4% 0.9% -0.3% -1.9% 3.6% 4.4% 6.1% 7.8%
PE95% / hA5% 1.9% 1.2% -0.4% -2.1% 5.2% 5.8% 7.4% 8.9%
PE90% / hA10% 2.4% 1.4% -0.4% -2.2% 6.7% 7.2% 8.7% 9.9%PE80% / hA20% 3.3% 1.9% -0.5% -2.6% 9.4% 9.7% 11.1% 11.9%TPU100% 1.6% 1.1% -0.3% -2.0% 4.3% 5.0% 6.7% 8.4%TPU95% / hA5% 2.2% 1.3% -0.4% -2.2% 5.9% 6.5% 8.0% 9.4%TPU90% / hA10% 2.7% 1.5% -0.4% -2.3% 7.4% 7.8% 9.3% 10.4%TPU80% / hA20% 3.5% 2.0% -0.5% -2.7% 9.9% 10.3% 11.6% 12.4%PVDF100% 2.3% 1.4% -0.4% -2.2% 6.3% 6.9% 8.4% 9.8%PVDF95% / hA5% 2.8% 1.6% -0.4% -2.4% 7.7% 8.2% 9.7% 10.8%PVDF90% / hA10% 3.2% 1.8% -0.5% -2.6% 9.0% 9.4% 10.9% 11.8%PVDF80% / hA20% 4.0% 2.2% -0.6% -2.9% 11.3% 11.7% 13.1% 13.7%
% deviation of µAir from µWaterSample
% deviation of µSaline from µWater
With both the linear attenuation coefficients and CT number values assessed, a comparison
can be determined as trend lines between these two values. The following figure illustrates a
comparison between trend lines developed between µ and CT number for water bath, saline
solution and air environment scans.
5.3.4 Linear Attenuation Coefficient to CT number Comparisons
69
Figure 5–1. Trend analysis and comparison between CT number and its corresponding calculated linear attenuation coefficient of scans completed in a water bath (Yw), saline solution (Ys), and air environments (Ya)
at 80kVp.
Figure 5–2. Trend analysis and comparison between CT number and its corresponding calculated linear attenuation coefficient of scans completed in a water bath (Yw), saline solution (Ys), and air environments (Ya)
at 100 kVp
Yw = 5169.7x - 1060.1
Ys = 4968.9x - 1048.2
Ya = 4964.8x - 1116.4
-300
-100
100
300
500
700
900
1100
1300
1500
0.1 0.2 0.3 0.4 0.5 0.6
CT
Num
ber (
HU
)
Linear Attenuation Coefficient (cm-1)
80kVp Water Bath Saline Air Linear (Water Bath) Linear (Saline) Linear (Air)
Yw = 5191x - 985.05
Ys = 5069.3x - 980.94
Ya= 5055.8x - 1047.6
-300
-100
100
300
500
700
900
1100
1300
1500
0.1 0.2 0.3 0.4 0.5 0.6
CT
Num
ber (
HU
)
Linear Attenuation Coefficient (cm-1)
100kVp Water Bath Saline Air Linear (Water Bath) Linear (Saline) Linear (Air)
70
Figure 5–3. Trend analysis and comparison between CT number and its corresponding calculated linear attenuation coefficient of scans completed in a water bath (Yw), saline solution (Ys), and air environments (Ya)
at 80kVp.
Figure 5–4. Trend analysis and comparison between CT number and its corresponding calculated linear attenuation coefficient of scans completed in a water bath (Yw), saline solution (Ys), and air environments (Ya)
at 80kVp.
Yw = 5204.1x - 923.76
Ys = 5225.8x - 931.88
Ya = 5086.9x - 985.32
-300
-100
100
300
500
700
900
1100
1300
1500
0.1 0.2 0.3 0.4 0.5 0.6
CT
Num
ber (
HU
)
Linear Attenuation Coefficient (cm-1)
120kVp Water Bath Saline Air Linear (Water Bath) Linear (Saline) Linear (Air)
Yw = 5201.1x - 878
Ys = 5337.6x - 887.32
Ya = 5172.6x - 949.62
-300
-100
100
300
500
700
900
1100
1300
1500
0.1 0.2 0.3 0.4 0.5 0.6
CT
Num
ber (
HU
)
Linear Attenuation Coefficient (cm-1)
135kVp Water Bath Saline Air Linear (Water Bath) Linear (Saline) Linear (Air)
71
Illustrated in figures 5-1 to 5-4, over the spectrum of calculated linear attenuation
coefficients, the corresponding CT numbers are closely aligned between water bath and
saline solution scans and largely varying for that of air scans.
The measured CT numbers obtained from the water bath and saline solution environments
are closely aligned. However, the ΔCT fluctuates depending on the material composition.
This fluctuation has a notable trend of increasing ΔCT with increasing hA% added to each
polymeric base. The increase in hA increases the values of two key material properties
related to the CT number, ρ and elemental composition that pertains to effective atomic
numbers, ZƮ and ZR. With the exception of outlier data at PVDF95% / hA5%, the ranges of
ΔCT are notably higher in scans performed at 80kVp; these can be noted as variations
between -22.64 to 31.82 HU between water bath and saline solution scans respectively.
However, as the applied x-ray voltage is increased, the effective range of ΔCT is significantly
decreased, notably at 135kVp to a range of -12.60 to 13.80 HU. In contrast, a comparison of
CT number values between water bath and air scans demonstrates large ΔCT values even at
135kVp with a range between -38.78 and 105.89 HU.
From this comparison, two specific deductions can be made. Firstly, both ρ and atomic
number ZƮ and ZR are directly related to the range in ΔCT values. Addition of hA in discrete
amounts to separate polymers seems to have varying effects on ΔCT, however, the general
trend is an increase in ΔCT with increasing hA%. Previous studies that evaluated the effect of
5.4 Discussion
5.4.1 CT Number Comparison
72
increasing hA% on CT scans have noted that due to the polychromatic nature of CT there is a
resulting variation in beam hardening effect, that requires beam hardening correction
algorithms to compensate[13], [62], [63], [124]. The beam hardening correction algorithm,
however, is calibrated using a water bath, or water-equivalent environment which can cause
variations in measured CT number when using saline solution and air environments as
demonstrated.
Secondly, the ΔCT variation with change in applied voltage, ρ, and elemental composition all
pertain to the characteristic linear attenuation coefficient, µ. Though the ρ and elemental
composition remain consistent between scans in different environments, the resulting
effective energy within each environment will change, thus altering the expected µ values.
To analyze the effect of material properties on CT numbers in different scanning media, it is
pertinent to describe how material properties and CT numbers are related through µ; µ is a
combination of material properties and resulting effective energy following beam hardening
of the incident polychromatic beam. The effective energy is expected to be similar for saline
solution scans whereas the air scans should have an effective energy significantly lower due
to the absence of a water or water-equivalent material to remove lower energy x-ray photons
from the x-ray spectrum. Table 5 demonstrates this phenomenon with comparative calculated
keV in which the values differ by no more than 5keV between the water bath and saline
solution environment scans. In contrast, the difference can be as large as 13keV between
water bath and air environment scans.
5.4.2 Material Property Comparisons – Linear Attenuation Coefficient
73
As mentioned, the effective energy can be further characterized in combination with the
material properties of ρ, ZƮ and ZR through calculated µ values. The µ values show a
relatively high level of consistency between water bath and saline solution environment
scans. A maximum difference of 5% deviation of µsaline to µwater was exhibited among all
applied voltages with most of the deviation nearing 0% at applied voltages aside from
80kVp; µair however had large percentage deviations from µwater at all applied voltages,
reaching as high as 14.4%. This phenomenon can be greatly attributed to the calculated
effective energies. As indicated by equations (2) and (4), the effective energy affects the
attenuation coefficient both in a proportional manner due to incoherent scattering and in an
inversely proportional manner due to photoelectric absorption and coherent scattering. The
contribution of incoherent scattering to the attenuation coefficient, as noted in equation (4), is
not a property of effective atomic number and thus will be constant, however photoelectric
absorption and coherent scattering tend to decrease the attenuation coefficient with
increasing effective atomic number. This variation of increasing attenuation coefficient with
increasing effective atomic number is more apparent in the percentage deviation of µair to
µwater as opposed to µsaline to µwater due to the significantly lower effective energy values from
air environment scans.
The least amount of variation in CT number is observed at 120kVp between µsaline to µwater
percentage deviation. As noted in table 5-4, the effective energies between saline and water
are most closely aligned and thus will show the least amount of variation in linear attenuation
coefficient, regardless of change in effective atomic number. This, in turn, should result in
closely aligned CT number values between saline solution and water bath environment scans.
74
Through our study we have proposed the use of enclosed bags of saline solution as a
surrounding environment for CT phantoms when a water bath or rigid water-equivalent
environment is suboptimal. This can allow for accurate CT scans to be conducted under a
dry, shape conforming environment. Ideally, the use of saline solution bags will have the
least variation from water bath scans at all applied voltages (80, 100, 120 and 135kVp) and
demonstrate densities most closely aligned to the effective atomic numbers of water (ZT and
ZR). As compared to the use of merely a water solution, a saline solution is readily available
in a clinical setting while also resembling the scanning conditions introduced to native
vessels prior to the addition of contrasting agents. By noticing that the saline solution
medium allows for similar results to that of the water bath medium, we can suggest that
saline solution can act as an alternative scanning environment, which can conform, to any
shape complexities and phantom fragility.
This study aimed to encourage the fabrication of CT phantoms to a higher degree of
anthropomorphic behavior in order to more closely mimic native human tissues by CT
numbers ranging between that of -100 to 1100 HU. Further studies necessitate the validation
of fragile, complex and anthropomorphic phantoms under controlled enclosed saline solution
environments. Varied thicknesses and conformations of saline solutions may also be assessed
to further this research for clinical studies. Ultimately, the aim is to improve CT scanning
phantom studies in such a way that the CT phantoms used may wholly resemble its desired
tissues and environments.
5.5 Summary
75
From our research, we have noted the following key findings:
1. CT Number Fine-Tuning: The control of CT number can be completed and fine-tuned
through both polymeric composite and foam fabrication techniques. The ranges to
which we have successfully done so are necessitated towards the cardiac system. As
such the range of control includes: -225 to 0HU for adipose and fatty plaque, 0 to
100HU for soft tissue, and 0 to 1200HU for range of coronary plaque.
2. Specified CT Property Control: CT numbers have been noted as controlled through
the processing techniques of polymeric compounding and batch foaming. Both are
noted through the control of linear attenuation coefficient however more specifically
this involves mass density for select foam systems and mass density together with
effective atomic number for varied composites.
Chapter 6 Conclusions and Recommendations
6.1 Conclusion
76
3. Alterative Scanning Media: Due to the potential shape complexity and fragility of a
cardiac phantom by its complex system, the alternative scanning media of enclosed
saline solution can be alternatively used in place of a water bath or solid water bath
equivalent. This environment is closely aligned to that of a water bath medium while
also allowing for an open space environment for the placement of any type of CT
phantom.
CT phantoms are capable of enhancing the potential for CT scanning in terms of diagnostic
ability, imaging quality and calibration techniques. The use of polymers is often considered
due to its flexibility in material properties when characterizing and fabricating CT phantoms.
By being able to design and tailor key material properties, the attenuation and thus,
characteristic CT number can be calibrated to match any desired tissue. This capability is
particularly important for studies involving the cardiac tissue and its coronary artery
environment. The intricate and detailed cardiac system requires that various be mimicked.
Further, when detailing the coronary artery itself, to define coronary plaque, a wide range of
CT number values are required to mimic both the environment and the varied levels of
coronary plaque.
This research thesis investigated polymeric fabrication techniques that can both
increase and decrease the CT number in a controlled manner of a base polymer to mimic the
desired tissue. Firstly, the fabrication of CT phantoms through polymeric composite twin
screw compounding assessed the potential of increasing control of CT number. This study
considered polymer matrices such as PE, TPU, and PVDF in combination with discrete
amounts of micro-sized particle hA through twin screw compounding to attain uniform
mixing of hA particles within the polymeric matrix. This includes the soft tissue of the heart
77
and arterial wall, surrounding adipose tissue and various types of coronary plaque including
increasing levels of calcification. Secondly, the fabrication of polymeric foams was studied
for its capability in mimicking low density tissues, particularly within the cardiac
environment such as adipose tissue. This fabrication technique involved TPU and a batch
polymeric foaming technique utilizing a physical blowing agent of CO2. Following the
fabrication techniques, polymeric CT phantoms then necessitated an environment in which
they could be scanned without the use of water or a rigid water equivalent due to potential
sample fragility. Since, most commonly, CT phantoms are scanned either in a water bath or a
solid PMMA surrounding environment, any fabricated samples are limited in size, shape and
design.
The current research has demonstrated the ability of polymeric fabrication techniques and
specified scanning environments to be utilized for cardiac CT phantom studies. However,
specific to each study, limitations of each have been listed.
CT phantom fabrication has shown a closely aligned predictive nature at effective
atomic numbers and densities closer to that of water. With deviation from that of water at a
Zτ of 7.49 and a ZR of 6.95 with a mass density of 1.0 g/cm3 we find that the theoretical
predictive model seems to deviate from our results causing, also for more variation in the
measured data. Furthermore, the use of varied kVps illustrates that the values, which are
closely aligned to the theoretical, are of 100 or 120kVp. The use of 80kVp and 135kVp tend
to show a greater variation.
6.2 Limitations
78
Similarly, in terms of sample consistency, the closer the samples were to
characteristics of water, the less variation in CT properties. When referenced to chapters 3, 4,
and 5, it can be noted that there are smaller error bars associated with phantoms with
characteristics closer to water. However, large and small effective atomic numbers from
added hA illustrated greater variation in CT number. Additionally, chapter 4 illustrates
cellular foams with potential variation in processing technique. As can be seen, there is some
variation with the same processing technique for CT number. Samples processed at 0.5 and 1
min in the heated water bath have varied densities and thus CT number, however at longer
processing times we find that there is more consistency.
Additionally, our samples were fabricated as cylindrical shapes, similar to
commercial phantom inserts such as the Gammex 467 phantom. CT scans are not
dimensionally critical as it is a polychromatic, gantry-loaded technique. Thus slight
variations in length, width or thickness due to processing technique should not affect the CT
properties. However, large variations in shape and thickness of material can affect the
accuracy of the scans thus potentially causing variations from the theoretical model
illustrated.
The fabrication and scanning techniques for both polymeric composite and foam phantoms
act as the starting point for creating a cardiac system phantom. The control of CT number
allows for the components of the cardiac system to be mimicked to the desire of the
physician or researcher. It is suggested that following this thesis study, first a geometrically
6.3 Recommendations
79
accurate cardiac system CT phantom be fabricated. This system can then be CT scanned and
compared against patient scans to determine its likeness.
The geometrically accurate phantom can be considered to be fabricated both under
static and dynamic conditions. Depending on the research direction, a static cardiac system
phantom would likely be aimed at studies involving coronary disease, particularly that of
lumen size identification under varied plaque levels of calcification. The goal within the
static cardiac phantom would be to determine scanning methods, post processing algorithms
and image diagnostics, under which the lumen size, or patency, can be identified accurately.
It would be suggested that by utilizing the fabrication techniques of polymeric composites,
particularly the polymer / hA system, a wide range of varied coronary plaque can be
fabricated and utilized.
Under dynamic conditions, various material and mechanical properties must be
considered. The use of both the compounding and foaming techniques can be used in
combination with a polymer material selection process in which ideal mechanical properties
are selected. The fabrication of a dynamic heart phantom can aid in studies improving motion
scans of the heart. By creating a system which can be consistently scanned under various
motion conditions, mimicking that of the heart, these scans can be tested and improved upon,
again by means of controlled scanning conditions, post processing algorithms and image
diagnostics.
Lastly, as this study has identified a method in which a polymeric system can both be
increased and decreased in CT number by fabrication techniques, it can be suggested that
various intricate and complex CT phantoms be fabricated. A first means of fabrication may
have various composite of foam processing techniques within a single material. Potentially,
80
the various structures such as the cardiac wall and coronary artery tissues may require
different levels of CT number within the same structure. Though multiple materials may be
stacked upon one another, it is possible to use a single product with functionally graded
properties.
By means of 3D printing technology and the implementation of in-house polymer
composite and foam fabrication, a number of different low and high density tissues, accurate
in both geometry and CT number, be fabricated. Potentially, low density phantoms with
uniformly dispersed cellular structure can be fabricated based on the desired foam size and
3D printer resolution. The use of 3D printing can also accurately fabricate delicate and small
dimensional objects including the coronary artery and degrees of various plaque.
Additionally, due to the ease of fabrication, mold-ability and machinability of a wide variety
of polymeric systems, phantoms can be fabricated in a number of desired shapes and
conformations. Lastly, bone studies can involve both foam and composite fabrication due to
the porous nature of the bone as well as the high density values.
81
[1] CANSIM, “Deaths, by cause, Diseases of the circulatory system (I00 to I99), age
group and sex, Canada, annual (number),” 2014.
[2] K. Pohle, S. Achenbach, B. Macneill, D. Ropers, M. Ferencik, F. Moselewski, U.
Hoffmann, T. J. Brady, I.-K. K. Jang, and W. G. Daniel, “Characterization of non-
calcified coronary atherosclerotic plaque by multi-detector row CT: Comparison to
IVUS,” Atherosclerosis, vol. 190, no. 1, pp. 174–180, Jan. 2007.
[3] S. E. Nissen and P. Yock, “Intravascular ultrasound: novel pathophysiological insights
and current clinical applications.,” Circulation, vol. 103, no. 4, pp. 604–616, 2001.
[4] N. S. Paul, J. Blobel, H. Kashani, M. Rice, and A. Ursani, “Quantification of arterial
plaque and lumen density with MDCT.,” Med. Phys., vol. 37, no. 8, pp. 4227–4237,
2010.
[5] W. Kristanto, P. M. a Van Ooijen, M. J. W. Greuter, J. M. Groen, R. Vliegenthart, and
M. Oudkerk, “Non-calcified coronary atherosclerotic plaque visualization on CT:
Effects of contrast-enhancement and lipid-content fractions,” Int. J. Cardiovasc.
Imaging, vol. 29, no. 5, pp. 1137–1148, 2013.
Bibliography
82
[6] J.-H. H. Choi, J. K. Min, T. M. Labounty, F. Y. Lin, D. D. Mendoza, D. H. Shin, N. S.
Ariaratnam, S. Koduru, J. F. Granada, T. C. Gerber, J. K. Oh, H.-C. C. Gwon, and Y.
H. Choe, “Intracoronary transluminal attenuation gradient in coronary CT angiography
for determining coronary artery stenosis,” JACC Cardiovasc. Imaging, vol. 4, no. 11,
pp. 1149–1157, Nov. 2011.
[7] Y. Funama, S. Oda, D. Utsunomiya, and K. Taguchi, “Coronary Artery Stent
Evaluation by Combining Iterative Reconstruction and High-resolution Kernel
at Coronary CT Angiography,” Acad. Radiol., vol. 19, no. 11, pp. 1324–1331,
2012.
[8] T. Boltz, W. Pavlicek, R. Paden, M. Renno, A. Jensen, and M. Akay, “An
anthropomorphic beating heart phantom for cardiac X-ray CT imaging evaluation,” J.
Appl. Clin. Med. Phys., vol. 11, no. 1, pp. 191–199, 2010.
[9] P. M. Shikhaliev, “Dedicated phantom materials for spectral radiography and CT.,”
Phys. Med. Biol., vol. 57, no. 6, pp. 1575–93, Mar. 2012.
[10] Y. Watanabe, “Derivation of linear attenuation coefficients from CT numbers for low-
energy photons.,” Phys. Med. Biol., vol. 44, no. 9, pp. 2201–2211, 1999.
[11] D.-H. Kim, H.-J. Kim, C.-L. Lee, H.-M. Cho, H.-S. Park, S.-W. Lee, Y.-N. Choi, Y.-
S. Kim, and S.-J. Park, “Absolute measurement of effective atomic number and
electron density using dual-energy computed tomography images,” in SPIE Medical
Imaging, 2012, vol. 8313, p. 83132X.
[12] L. H. Sperling, Introduction to physical polymer science. 2006.
[13] J. M. Deuerling, D. J. Rudy, G. L. Niebur, and R. K. Roeder, “Improved accuracy of
cortical bone mineralization measured by polychromatic microcomputed tomography
using a novel high mineral density composite calibration phantom.,” Med. Phys., vol.
37, no. 9, pp. 5138–5145, 2010.
83
[14] P. Homolka, a Gahleitner, M. Prokop, and R. Nowotny, “Optimization of the
composition of phantom materials for computed tomography.,” Phys. Med. Biol., vol.
47, no. 16, pp. 2907–2916, 2002.
[15] G. J. Kemerink, H. H. Kruize, R. J. Lamers, and J. M. van Engelshoven, “Density
resolution in quantitative computed tomography of foam and lung.,” Medical physics,
vol. 23, no. 10. pp. 1697–1708, 1996.
[16] G. T. Herman and H. T. Gabor, Fundamentals of computerized tomography: image
reconstruction from projections, vol. 325, no. 5946. Laboratory of Atomic and Solid-
State Physics, Cornell University, Ithaca, NY 14853, USA. [email protected] DOI -
10.1126/science.1176112 SRC - Pubmed ID2 - 19745146 FG - 0, 2009.
[17] J. R. Mikolich, “Cardiac computed tomographic angiography and the primary care
physician.,” J. Am. Osteopath. Assoc., vol. 112, no. 5, pp. 267–75, 2012.
[18] H. Spitz, M. Jenkins, J. Lodwick, and R. Bornschein, “A new anthropometric phantom
for calibrating in vivo measurements of stable lead in the human leg using x-ray
fluorescence.,” Health Phys., vol. 78, no. 2, pp. 159–169, 2000.
[19] J. H. Hubbell, “Review of photon interaction cross section data in the medical and
biological context.,” Phys. Med. Biol., vol. 44, no. 1, pp. R1–R22, Jan. 1999.
[20] S. B. Kaginelli, T. Rajeshwari, Sharanabasappa, B. R. Kerur, and A. S. Kumar,
“Effective atomic numbers and electron density of dosimetric material.,” J. Med.
Phys., vol. 34, no. 3, pp. 176–179, 2009.
[21] E. C. McCullough and T. W. Holmes, “Acceptance testing computerized radiation
therapy treatment planning systems: direct utilization of CT scan data.,” Med. Phys.,
vol. 12, no. 2, pp. 237–242, 1985.
[22] S. J. Thomas, “Relative electron density calibration of CT scanners for radiotherapy
treatment planning,” Br. J. Radiol., vol. 72, no. AUG., pp. 781–786, 1999.
84
[23] S. Schroeder, a F. Kopp, a Baumbach, C. Meisner, a Kuettner, C. Georg, B.
Ohnesorge, C. Herdeg, C. D. Claussen, and K. R. Karsch, “Noninvasive detection and
evaluation of atherosclerotic coronary plaques with multislice computed
tomography.,” J. Am. Coll. Cardiol., vol. 37, no. 5, pp. 1430–1435, 2001.
[24] J. I. Murungi, S. Thiam, R. E. Tracy, J. W. Robinson, and I. M. Warner, “Elemental
analysis of soft plaque and calcified plaque deposits from human coronary arteries and
aorta.,” J. Environ. Sci. Health. A. Tox. Hazard. Subst. Environ. Eng., vol. 39, no. 6,
pp. 1487–1496, 2004.
[25] R. Molteni, “From to Hounsfield Units in Cone Beam Volumetric Imaging: the effect
of artifacts,” 2011.
[26] R. A. Rutherford, B. R. Pullan, and I. Isherwood, “Measurement of effective atomic
number and electron density using an EMI scanner.,” Neuroradiology, vol. 11, no. 1,
pp. 15–21, 1976.
[27] T. P. Szczykutowicz, Z. Qi, and G.-H. Chen, “A simple image based method for
obtaining electron density and atomic number in dual energy CT,” Cleve. Clin. J.
Med., vol. 7961, no. 1, p. 79613A–79613A–6, 2011.
[28] M. J. Berger, J. H. Hubbell, S. M. Seltzer, J. Chang, J. S. Coursey, R. Sukumar, and D.
S. Zucker, “XCOM: photon cross sections database,” NIST Stand. Ref. Database, vol.
8, pp. 87–3597, 1998.
[29] J. H. Hubbell, W. J. Veigele, E. a. Briggs, R. T. Brown, D. T. Cromer, and R. J.
Howerton, “Atomic form factors, incoherent scattering functions, and photon
scattering cross sections,” J. Phys. Chem. Ref. Data, vol. 4, no. 3, pp. 471–538, 1975.
[30] J. Hubbell and I. Øverbø, “Relativistic atomic form factors and photon coherent
scattering cross sections,” J. Phys. Chem. Ref. Data, vol. 8, no. 1, p. 69, 1979.
[31] J. H. Scofield, “Theoretical photoionization cross sections from 1 to 1500 keV.,”
Lawrence Livermore Lab. Rep. No. UCRL-51326, 1973.
85
[32] J. H. Hubbell, H. a. Gimm, and I. O ̸verbo̸, “Pair, Triplet, and Total Atomic Cross
Sections (and Mass Attenuation Coefficients) for 1 MeV-100 GeV Photons in
Elements Z=1 to 100,” J. Phys. Chem. Ref. Data, vol. 9, no. 4, p. 1023, 1980.
[33] M. a Ebert, J. Lambert, and P. B. Greer, CT-ED conversion on a GE Lightspeed-RT
scanner: influence of scanner settings., vol. 31, no. 2. 2008.
[34] C. Bai, L. Shao, A. J. Da Silva, and Z. Zhao, “A Generalized Model for the
Conversion from CT Numbers to Linear Attenuation Coefficients,” IEEE Trans. Nucl.
Sci., vol. 50, no. 5 II, pp. 1510–1515, 2003.
[35] C. B. Saw, A. Loper, K. Komanduri, T. Combine, S. Huq, and C. Scicutella,
“Determination of CT-to-density conversion relationship for image-based treatment
planning systems,” Med. Dosim., vol. 30, no. 3, pp. 145–148, 2005.
[36] S. Brown, D. L. Bailey, K. Willowson, and C. Baldock, “Investigation of the
relationship between linear attenuation coefficients and CT Hounsfield units using
radionuclides for SPECT,” Appl. Radiat. Isot., vol. 66, no. 9, pp. 1206–1212, Sep.
2008.
[37] S. Blankespoor, X. Wu, K. Kalki, J. K. Brown, C. E. Cann, and B. H. Hasegawa,
“Attenuation correction of SPECT using x-ray CT on an emission-transmission CT
system,” IEEE Nucl. Sci. Symp. Med. Imaging Conf., vol. 2, pp. 1126–1130, 1995.
[38] K. J. LaCroix, B. M. W. Tsui, B. H. Hasegawa, and J. K. Brown, “Investigation of the
use of X-ray CT images for attenuation compensation in SPECT,” IEEE Trans. Nucl.
Sci., vol. 41, no. 6, pp. 2793–2799, 1994.
[39] P. Kinahan, “PET / CT Issues : CT-based attenuation correction ( CTAC ), Artifacts ,
and Motion Correction,” 2008.
[40] P. E. Kinahan, B. H. Hasegawa, and T. Beyer, “X-ray-based attenuation correction for
positron emission tomography/computed tomography scanners,” Semin Nucl Med, vol.
33, no. 3, pp. 166–179, 2003.
86
[41] P. E. Kinahan, D. W. Townsend, T. Beyer, and D. Sashin, “Attenuation correction for
a combined 3D PET/CT scanner,” Med. Phys., vol. 25, no. 10, pp. 2046–2053, 1998.
[42] D. R. White, R. J. Martin, and R. Darlison, “Epoxy resin based tissue substitutes,” Br.
J. Radiol., vol. 50, no. 599, pp. 814–821, 1977.
[43] M. Abella, A. M. Alessio, D. a Mankoff, L. R. MacDonald, J. J. Vaquero, M. Desco,
and P. E. Kinahan, “Accuracy of CT-based attenuation correction in PET/CT bone
imaging,” Phys. Med. Biol., vol. 57, no. 9, pp. 2477–2490, May 2012.
[44] a. H. R. Lonn, “Evaluation of method to minimize the effect of X-ray contrast in PET-
CT attenuation correction,” 2003 IEEE Nucl. Sci. Symp. Conf. Rec. (IEEE Cat.
No.03CH37515), vol. 3, 2003.
[45] P. R. Costa, “Computer Tomography Phantom Applications,” in The Phantoms of
Medical and Health Physics, 2014, pp. 123 – 142.
[46] M. E. Poletti, D. Gonçalves, and I. Mazzaro, “X-ray scattering from human breast
tissues and breast-equivalent materials.,” Phys. Med. Biol., vol. 47, no. 1, pp. 47–63,
2002.
[47] J. W. Byng, J. G. Mainprize, and M. J. Yaffe, “X-ray characterization of breast
phantom materials.,” Phys. Med. Biol., vol. 43, no. 5, pp. 1367–1377, 1998.
[48] S. Ramakrishna, J. Mayer, E. Wintermantel, and K. W. Leong, “Biomedical
application of polymer-composite materials: a review,” Compos. Sci. Technol., vol.
61, pp. 1189–1224, 2001.
[49] T. Villmow, P. Pötschke, S. Pegel, L. Häussler, and B. Kretzschmar, “Influence of
twin-screw extrusion conditions on the dispersion of multi-walled carbon nanotubes in
a poly(lactic acid) matrix,” Polymer (Guildf)., vol. 49, no. 16, pp. 3500–3509, 2008.
[50] J. . Cho and D. . Paul, “Nylon 6 nanocomposites by melt compounding,” Polymer
(Guildf)., vol. 42, no. 3, pp. 1083–1094, 2001.
87
[51] Aroon V. Shenoy, Rheology of Filled Polymer Systems. 1999.
[52] F. Hussain, “Review article: Polymer-matrix Nanocomposites, Processing,
Manufacturing, and Application: An Overview,” J. Compos. Mater., vol. 40, no. 17,
pp. 1511–1575, 2006.
[53] L. Biasci, M. Aglietto, G. Ruggeri, and A. D’Alessio, “Functionalization of
montmorillonite by acrylamide polymers containing side-chain ammonium cations,”
Polym. Adv. Technol., vol. 6, no. 10, pp. 662–670, 1995.
[54] L. Fang, Y. Leng, and P. Gao, “Processing and mechanical properties of
HA/UHMWPE nanocomposites,” Biomaterials, vol. 27, no. 20, pp. 3701–3707, 2006.
[55] Z. Wang and T. J. Pinnavaia, “Nanolayer Reinforcement of Elastomeric
Polyurethane,” Chem. Mater., vol. 10, no. 12, pp. 3769–3771, 1998.
[56] V. Sendijarevic, “Daniel, Klempner, Handbook of Polymeric Foams and Foam
Technology.,” Munich Hanser, 1991.
[57] K. T. Okamoto, “Microcellular Processing,” Radiol. Clin. North Am., vol. 41, no. 2, p.
186, 2003.
[58] G. Liu, C. B. Park, and J. a Lefas, “Production of low-density LLDPE foams in
rotational molding,” Polym. Eng. Sci., vol. 38, no. 12, pp. 1997–2009, 1998.
[59] G. Guo, K. H. Wang, C. B. Park, Y. S. Kim, and G. Li, “Effects of nanoparticles on
the density reduction and cell morphology of extruded metallocene polyethylene/wood
fiber nanocomposites,” J. Appl. Polym. Sci., vol. 104, no. 2, pp. 1058–1063, 2007.
[60] W. Michaeli and S. Habibi-Naini, “New Technology for Foam Injection Moulding
with Physical Blowing Agents,” in Blowing Agents and Foaming Processes, 2001, p.
Paper 9, Page 1 – 8.
88
[61] S. N. Leung, C. B. Park, D. Xu, H. Li, and R. G. Fenton, “Computer simulation of
bubble-growth phenomena in foaming,” Ind. Eng. Chem. Res., vol. 45, no. 23, pp.
7823–7831, 2006.
[62] G. J. Kazakia, a J. Burghardt, S. Cheung, and S. Majumdar, “Assessment of bone
tissue mineralization by conventional x-ray microcomputed tomography: comparison
with synchrotron radiation microcomputed tomography and ash measurements.,” Med.
Phys., vol. 35, no. 7, pp. 3170–3179, 2008.
[63] A. J. Burghardt, G. J. Kazakia, A. Laib, and S. Majumdar, “Quantitative assessment of
bone tissue mineralization with polychromatic micro-computed tomography,” Calcif.
Tissue Int., vol. 83, no. 2, pp. 129–138, 2008.
[64] L. Mulder, J. H. Koolstra, and T. M. G. J. Van Eijden, “Accuracy of microCT in the
quantitative determination of the degree and distribution of mineralization in
developing bone.,” Acta radiol., vol. 45, no. 7, pp. 769–777, 2004.
[65] J. a. Meganck, K. M. Kozloff, M. M. Thornton, S. M. Broski, and S. a. Goldstein,
“Beam hardening artifacts in micro-computed tomography scanning can be reduced by
X-ray beam filtration and the resulting images can be used to accurately measure
BMD,” Bone, vol. 45, no. 6, pp. 1104–1116, 2009.
[66] W. J. Yang, K. M. Chen, L. F. Pang, Y. Guo, J. Y. Li, H. Zhang, and Z. L. Pan, “High-
definition computed tomography for coronary artery stent imaging: a phantom study,”
Korean J Radiol, vol. 13, no. 1, pp. 20–26, 2012.
[67] M. Toepker, G. Euller, E. Unger, M. Weber, D. Kienzl, C. J. Herold, and H. Ringl,
“Stenosis quantification of coronary arteries in coronary vessel phantoms with second-
generation dual-source CT: Influence of measurement parameters and limitations,”
Am. J. Roentgenol., vol. 201, no. 2, pp. W227–34, Aug. 2013.
[68] H. Spitz and R. Bornsched, “A New Anthropometric Phantom for Calibrating in vivo
Measurements of Stable Lead in the Human Leg Using X-ray Fluorescence,” Health
Phys., vol. 78, no. 2, pp. 159–169, 2000.
89
[69] B. D. R. White, M. Sc, and D. Ph, “Epoxy resin based tissue substitutes,” Br. J.
Radiol., vol. 50, no. 599, pp. 814–821, 1977.
[70] C. Constantinou, “An electron density calibration phantom for CT-based treatment
planning computers,” Med. Phys., vol. 19, no. 2, p. 325, 1992.
[71] F. Cademartiri, N. R. Mollet, G. Runza, N. Bruining, R. Hamers, P. Somers, M.
Knaapen, S. Verheye, M. Midiri, G. P. Krestin, and P. J. de Feyter, “Influence of
intracoronary attenuation on coronary plaque measurements using multislice computed
tomography: observations in an ex vivo model of coronary computed tomography
angiography.,” Eur. Radiol., vol. 15, no. 7, pp. 1426–31, Jul. 2005.
[72] C. Robert, “ACCF/ACR/SCCT/SCMR/ASNC/NASCI/SCAI/SIR 2006
Appropriateness Criteria for Cardiac Computed Tomography and Cardiac Magnetic
Resonance Imaging,” J. Am. Coll. Radiol., vol. 3, no. 10, pp. 751–771, 2006.
[73] D. B. Mark, D. S. Berman, M. J. Budoff, J. J. Carr, T. C. Gerber, H. S. Hecht, M. a
Hlatky, J. M. Hodgson, M. S. Lauer, J. M. Miller, R. L. Morin, D. Mukherjee, M.
Poon, G. D. Rubin, and R. S. Schwartz,
“ACCF/ACR/AHA/NASCI/SAIP/SCAI/SCCT 2010 expert consensus document on
coronary computed tomographic angiography: a report of the American College of
Cardiology Foundation Task Force on Expert Consensus Documents.,” Catheter.
Cardiovasc. Interv., vol. 76, no. 2, pp. E1–E42, 2010.
[74] K. A. Eagle, “ACC/AHA 2004 guideline update for coronary artery bypass graft
surgery: a report of the American College of Cardiology/American Heart Association
Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for
Coronary Artery Bypass Graft),” Circulation, vol. 110, pp. e340 – e437, 2005.
[75] S. C. Smith, J. T. Dove, A. K. Jacobs, J. Ward Kennedy, D. Kereiakes, M. J. Kern, R.
E. Kuntz, J. J. Popma, H. V Schaff, D. O. Williams, R. J. Gibbons, J. P. Alpert, K. a
Eagle, D. P. Faxon, V. Fuster, T. J. Gardner, G. Gregoratos, R. O. Russell, and S. C.
Smith, “ACC/AHA guidelines for percutaneous coronary intervention (revision of the
90
1993 PTCA guidelines)—executive summary,” J. Am. Coll. Cardiol., vol. 37, no. 8,
pp. 2215–2238, 2001.
[76] J. Sun, Z. Zhang, B. Lu, W. Yu, Y. Yang, Y. Zhou, Y. Wang, and Z. Fan,
“Identification and quantification of coronary atherosclerotic plaques: A comparison
of 64-MDCT and intravascular ultrasound,” Am. J. Roentgenol., vol. 190, no. 3, pp.
748–754, Mar. 2008.
[77] K. Jansen, M. Wu, A. F. W. van der Steen, and G. van Soest, “Lipid detection in
atherosclerotic human coronaries by spectroscopic intravascular photoacoustic
imaging,” Opt. Express, vol. 21, no. 18, p. 21472, 2013.
[78] H. Seifarth, M. Ozgün, R. Raupach, T. Flohr, W. Heindel, R. Fischbach, and D.
Maintz, “64- Versus 16-slice CT angiography for coronary artery stent assessment: in
vitro experience.,” Invest. Radiol., vol. 41, no. 1, pp. 22–27, 2006.
[79] C. Cheng, L. Zhao, M. Wolanski, Q. Zhao, J. James, K. Dikeman, M. Mills, M. Li, S.
P. Srivastava, X. Q. Lu, and I. J. Das, “Comparison of tissue characterization curves
for different CT scanners: implication in proton therapy treatment planning,” Transl.
Cancer Res., vol. 1, no. 4, pp. 236–246, 2013.
[80] F. Cademartiri, J. D. Schuijf, F. Pugliese, N. R. Mollet, J. W. Jukema, E. Maffei, L. J.
Kroft, a Palumbo, D. Ardissino, P. W. Serruys, G. P. Krestin, E. E. van der Wall, P. J.
de Feyter, and J. J. Bax, “Usefulness of 64-slice multislice computed tomography
coronary angiography to assess in-stent restenosis,” J.Am.Coll.Cardiol., vol. 49, no.
22, pp. 2204–2210, 2007.
[81] a. F. Kopp, S. Schroeder, a. Baumbach, a. Kuettner, C. Georg, B. Ohnesorge, M.
Heuschmid, R. Kuzo, and C. D. Claussen, “Non-invasive characterisation of coronary
lesion morphology and composition by multislice CT: First results in comparison with
intracoronary ultrasound,” Eur. Radiol., vol. 11, no. 9, pp. 1607–1611, 2001.
[82] A. W. Leber, A. Knez, A. Becker, C. Becker, F. Von Ziegler, K. Nikolaou, C. Rist, M.
Reiser, C. White, G. Steinbeck, and P. Boekstegers, “Accuracy of multidetector spiral
91
computed tomography in identifying and differentiating the composition of coronary
atherosclerotic plaques: A comparative study with intracoronary ultrasound,” J. Am.
Coll. Cardiol., vol. 43, no. 7, pp. e340 – 437, 2004.
[83] H.-Y. Mi, X. Jing, M. R. Salick, T. M. Cordie, X.-F. Peng, and L.-S. Turng,
“Morphology, mechanical properties, and mineralization of rigid thermoplastic
polyurethane/hydroxyapatite scaffolds for bone tissue applications: effects of
fabrication approaches and hydroxyapatite size,” J. Mater. Sci., vol. 49, no. 5, pp.
2324–2337, Dec. 2013.
[84] M. Selvakumar, S. K. Jaganathan, G. B. Nando, and S. Chattopadhyay, “Synthesis and
Characterization of Novel Polycarbonate Based Polyurethane/Polymer Wrapped
Hydroxyapatite Nanocomposites: Mechanical Properties, Osteoconductivity and
Biocompatibility,” J. Biomed. Nanotechnol., vol. 10, no. xx, pp. 1–15, 2014.
[85] W. Bonfield, “Polyethylene as an Analogous Material for Bone Replacement,” Ann. N.
Y. Acad. Sci., vol. 523, no. 1, pp. 173–177, 1988.
[86] W. Suchanek and M. Yoshimura, “Processing and properties of hydroxyapatite-based
biomaterials for use as hard tissue replacement implants,” J. Mater. Res., vol. 13, no.
01, pp. 94–117, Jan. 2011.
[87] L. Di Silvio, M. Dalby, and W. Bonfield, “In vitro response of osteoblasts to
hydroxyapatite-reinforced polyethylene composites,” J. Mater. Sci. Mater. Med., vol.
9, pp. 845–849, 1998.
[88] A. V. Shenoy, Rheology of filled polymer systems. Norwell: Kluwer Academic, 1999.
[89] M. Wang, D. Porter, and W. Bonfield, “Processing, characterisation, and evaluation of
hydryoxyapatite reinforced polyethylene composites,” in British Ceramic
Transactions, 1994, pp. 91–95.
[90] E. Bilotti, R. Zhang, H. Deng, M. Baxendale, and T. Peijs, “Fabrication and property
prediction of conductive and strain sensing TPU/CNT nanocomposite fibres,” J.
Mater. Chem., vol. 20, no. 42, p. 9449, 2010.
92
[91] R. Sousa, R. Reis, A. Cunha, and M. Bevis, “Coupling of HDPE/hydroxyapatite
composites by silane-based methodologies,” J. Mater. Sci. …, vol. 4, pp. 475–487,
2003.
[92] T. Yoshizumi and T. Nakamura, “Abdominal Fat: Standardized Technique for
Measurement at CT,” Radiology, vol. 211, pp. 283–286, 1999.
[93] M. van der Glas, Principles of computerized tomographic imaging. 2000.
[94] A. K. Jones, D. E. Hintenlang, and W. E. Bolch, “Tissue-equivalent materials for
construction of tomographic dosimetry phantoms in pediatric radiology.,” Med. Phys.,
vol. 30, no. 8, pp. 2072–2081, 2003.
[95] S. Kinase, M. Kimura, H. Noguchi, and S. Yokoyama, “Development of lung and soft
tissue substitutes for photons,” Radiat. Prot. Dosimetry, vol. 115, no. 1–4, pp. 284–
288, 2005.
[96] Z. H. Levine, M. Li, A. P. Reeves, D. F. Yankelevitz, J. J. Chen, E. L. Siegel, A.
Peskin, and D. N. Zeiger, “A low-cost density reference phantom for computed
tomography,” Med. Phys., vol. 36, no. 2, p. 286, 2009.
[97] K.-P. Chang, S.-H. Hung, Y.-H. Chie, A.-C. Shiau, and R.-J. Huang, “A comparison
of physical and dosimetric properties of lung substitute materials.,” Med. Phys., vol.
39, no. 4, pp. 2013–20, Apr. 2012.
[98] D. L. Tomasko, H. Li, D. Liu, X. Han, M. J. Wingert, L. J. Lee, and K. W. Koelling,
“A Review of CO 2 Applications in the Processing of Polymers,” Ind. Eng. Chem.
Res., vol. 42, pp. 6431–6456, 2003.
[99] R. Rizvi and H. Naguib, “Porosity and composition dependence on electrical and
piezoresistive properties of thermoplastic polyurethane nanocomposites,” J. Mater.
Res., vol. 28, no. 17, pp. 2415–2425, Aug. 2013.
93
[100] N. Hossieny, M. Barzegari, and M. Nofar, “Crystallization of hard segment domains
with the presence of butane for microcellular thermoplastic polyurethane foams,”
Polymer (Guildf)., vol. 55, pp. 651–662, 2014.
[101] S. G. Mosanenzadeh, H. E. Naguib, C. B. Park, and N. Atalla, “Development of
polylactide open-cell foams with bimodal structure for high-acoustic absorption,” J.
Appl. Polym. Sci., vol. 131, no. 7, pp. 1–11, Apr. 2014.
[102] J. W. Shi, H. Z. Dai, L. Shen, and D. F. Xu, “Dual-energy CT: clinical application in
differentiating an adrenal adenoma from a metastasis.,” Acta radiol., Aug. 2013.
[103] J. A. a. Bryant, N. A. a. Drage, and S. Richmond, “CT number definition,” Radiat.
Phys. Chem., vol. 81, no. 4, pp. 358–361, Apr. 2012.
[104] W. A. Kalender, “X-ray computed tomography.,” Phys. Med. Biol., vol. 51, no. 13, pp.
29–43, Jul. 2006.
[105] P. Mah, T. E. Reeves, and W. D. McDavid, “Deriving Hounsfield units using grey
levels in cone beam computed tomography.,” Dentomaxillofac. Radiol., vol. 39, no. 6,
pp. 323–35, Sep. 2010.
[106] D. S. Sharma, S. D. Sharma, K. K. Sanu, S. Saju, D. D. Deshpande, and S. Kannan,
“Performance evaluation of a dedicated computed tomography scanner used for virtual
simulation using in-house fabricated CT phantoms.,” J. Med. Phys., vol. 31, no. 1, pp.
28–35, 2006.
[107] E. C. McCullough, “Acceptance testing computerized radiation therapy treatment
planning systems: Direct utilization of CT scan data,” Med. Phys., vol. 12, no. 2, p.
237, 1985.
[108] G. Landry, J. Seco, M. Gaudreault, and F. Verhaegen, “Deriving effective atomic
numbers from DECT based on a parameterization of the ratio of high and low linear
attenuation coefficients.,” Phys. Med. Biol., vol. 58, no. 19, pp. 6851–66, 2013.
94
[109] M. Bazalova, J.-F. Carrier, L. Beaulieu, and F. Verhaegen, “Dual-energy CT-based
material extraction for tissue segmentation in Monte Carlo dose calculations,” Phys.
Med. Biol., vol. 53, pp. 2439–2456, 2008.
[110] W. H. Kim, C. G. Kim, and D. W. Kim, “Optimal CT number range for adipose tissue
when determining lean body mass in whole-body F-18 FDG PET/CT studies,” Nucl.
Med. Mol. Imaging (2010)., vol. 46, no. 4, pp. 294–299, 2012.
[111] W. Michaeli and R. Heinz, “Foam extrusion of thermoplastic polyurethanes (TPU)
using CO2 as a blowing agent,” Macromol. Mater. Eng., vol. 284–285, pp. 35–39,
2000.
[112] L. J. M. Jacobs, M. F. Kemmere, and J. T. F. Keurentjes, “Sustainable polymer
foaming using high pressure carbon dioxide: a review on fundamentals, processes and
applications,” Green Chem., vol. 10, no. 7, p. 731, 2008.
[113] D. Mozejko, “Image Texture , Uniformity , Homogenity and Radiation Dose
Properties in CT,” 2013.
[114] L. Chen, R. Ozisik, and L. S. Schadler, “The influence of carbon nanotube aspect ratio
on the foam morphology of MWNT/PMMA nanocomposite foams,” Polymer
(Guildf)., vol. 51, no. 11, pp. 2368–2375, 2010.
[115] D. Jahani, A. Ameli, M. Saniei, W. Ding, C. B. Park, and H. E. Naguib,
“Characterization of the Structure, Acoustic Property, Thermal Conductivity, and
Mechanical Property of Highly Expanded Open-Cell Polycarbonate Foams,”
Macromol. Mater. Eng., vol. 300, no. 1, pp. 48–56, Jan. 2015.
[116] S. Ghaffari Mosanenzadeh, H. E. Naguib, C. B. Park, and N. Atalla, “Effect of
biopolymer blends on physical and Acoustical properties of biocomposite foams,” J.
Polym. Sci. Part B Polym. Phys., vol. 52, no. 15, pp. 1002–1013, 2014.
[117] S. N. S. Leung, “Mechanism of Cell Nucleation, Growth and Coarsening in Plastic
Foaming: Theory, Simulation and Experiments,” p. 239, 2009.
95
[118] G. T. Herman, Fundamentals of Computerized Tomography: Image Reconstructions
from Projections, vol. 2. 2009.
[119] E. P. S. Sande, A. C. T. Martinsen, E. O. Hole, and H. M. Olerud, “Interphantom and
interscanner variations for Hounsfield units--establishment of reference values for HU
in a commercial QA phantom.,” Phys. Med. Biol., vol. 55, no. 17, pp. 5123–5135,
2010.
[120] U. Schneider, E. Pedroni, and a Lomax, “The calibration of CT Hounsfield units for
radiotherapy treatment planning.,” Phys. Med. Biol., vol. 41, no. 1, pp. 111–124, 1996.
[121] D. D. Robertson and H. K. Huang, “Quantitative bone measurements using x-ray
computed tomography with second-order correction,” Med. Phys., vol. 13, no. 4, pp.
474–479, 2014.
[122] G. Van Gompel, K. Van Slambrouck, M. Defrise, K. J. Batenburg, J. de Mey, J.
Sijbers, and J. Nuyts, “Iterative correction of beam hardening artifacts in CT.,” Med.
Phys., vol. 38 Suppl 1, no. 2011, p. S36, Jul. 2011.
[123] E. C. McCullough, “Photon attenuation in computed tomography,” Med. Phys., vol. 2,
no. 6, pp. 307–320, 1975.
[124] R. J. Fajardo, E. Cory, N. D. Patel, A. Nazarian, A. Laib, R. K. Manoharan, J. E.
Schmitz, J. M. DeSilva, L. M. MacLatchy, B. D. Snyder, and M. L. Bouxsein,
“Specimen size and porosity can introduce error into microCT-based tissue mineral
density measurements.,” Bone, vol. 44, no. 1, pp. 176–184, 2009.