-
Cleveland State UniversityEngagedScholarship@CSU
ETD Archive
2011
Detecting Vulnerable Plaques with MultiresolutionAnalysisSushma
SrinivasCleveland State University
How does access to this work benefit you? Let us know!Follow
this and additional works at:
http://engagedscholarship.csuohio.edu/etdarchive
Part of the Biomedical Engineering and Bioengineering
Commons
This Dissertation is brought to you for free and open access by
EngagedScholarship@CSU. It has been accepted for inclusion in ETD
Archive by anauthorized administrator of EngagedScholarship@CSU.
For more information, please contact [email protected].
Recommended CitationSrinivas, Sushma, "Detecting Vulnerable
Plaques with Multiresolution Analysis" (2011). ETD Archive. Paper
279.
http://engagedscholarship.csuohio.edu?utm_source=engagedscholarship.csuohio.edu%2Fetdarchive%2F279&utm_medium=PDF&utm_campaign=PDFCoverPageshttp://engagedscholarship.csuohio.edu/etdarchive?utm_source=engagedscholarship.csuohio.edu%2Fetdarchive%2F279&utm_medium=PDF&utm_campaign=PDFCoverPageshttp://library.csuohio.edu/engaged/http://engagedscholarship.csuohio.edu/etdarchive?utm_source=engagedscholarship.csuohio.edu%2Fetdarchive%2F279&utm_medium=PDF&utm_campaign=PDFCoverPageshttp://network.bepress.com/hgg/discipline/229?utm_source=engagedscholarship.csuohio.edu%2Fetdarchive%2F279&utm_medium=PDF&utm_campaign=PDFCoverPageshttp://engagedscholarship.csuohio.edu/etdarchive/279?utm_source=engagedscholarship.csuohio.edu%2Fetdarchive%2F279&utm_medium=PDF&utm_campaign=PDFCoverPagesmailto:[email protected]
-
DETECTING VULNERABLE PLAQUES WITH
MULTIRESOLUTION ANALYSIS
SUSHMA SRINIVAS
Bachelor of Engineering Electronics and Communications
University of Mysore
September, 1997
Master of Science - Physics
Cleveland State University
May, 2007
Submitted in partial fulfillment of requirements for the
degree
DOCTOR OF ENGINEERING
in
APPLIED BIOMEDICAL ENGINEERING
at the
CLEVELAND STATE UNIVERSITY
November, 2011
-
Copyright by SUSHMA SRINIVAS 2011
-
This dissertation has been approved
for the Department of Chemical and Biomedical Engineering
and the College of Graduate Studies by
________________________________________________
________________________________
Dissertation Committee Chairperson,
Aaron J. Fleischman Ph.D.
Biomedical Engineering, Cleveland Clinic
________________________________________________
________________________________
Academic Advisor, George P. Chatzimavroudis Ph.D.
Cleveland State University
________________________________________________
________________________________
Advisor, Miron Kaufman Ph.D.
Dept. of Physics, Cleveland State University
________________________________________________
________________________________
Advisor, Randolph M. Setser Ph.D.
Manager, Research Collaborations, Angiography & X-Ray
Siemens Healthcare
________________________________________________
________________________________
Clinical Advisor, Stephen Nicholls M.D, Ph.D.
Heart and Vascular Institute, Cleveland Clinic
________________________________________________
________________________________
Advisor, William Davros Ph.D.
Diagnostic Radiology, Cleveland Clinic
-
Dedicated to:
My sound children two inexhaustible acoustic sources
You will NEVER get your P etch D!
Jahnavi (age 7)
I am happy with you on this planet, why do you want me to become
an astronaut?
Chandni (age 4)
and
The few souls whose arteries were imaged for this study
-
ACKNOWLEDGEMENTS
First and foremost, I wish to express gratitude to my advisor,
Dr. Aaron
Fleischman who encouraged and challenged me through my
dissertation years. His
patience in listening to my viewpoints and reasoning, and
strategies for my ideas
are to be admired. I take it as a responsibility to be
successful and surpass his
expectations of me, as it is more rewarding to my advisor than
words can thank him
for the rich experience in his laboratory.
It is a pleasure to thank my ever accommodating committee. The
valuable
advice from Dr. George Chatzimavroudis, there is life beyond PhD
helped me start
every day with a positive outlook. I thank him for all his
advice on fulfilling academic
requirements and also teaching me medical imaging and signal
processing; his
lessons on fluid dynamics were most enjoyable. Words cannot
adequately thank Dr.
Miron Kaufman for his advice on choosing projects, mentors and
making university
and career choices. I regard highly, his valuable advice of
choosing CSU over Case
Western/Univ of Pittsburgh for the sake of my family. I
appreciate his efforts and
involvement in the development and training of his students. I
must thank Dr.
Randolph Setser for his mentoring during my Masters project as
well as my doctoral
studies. I thank him for introducing me to the most beautiful
imaging modality
MRI through his clear and comprehensive instructions. I respect
his professionalism
and discipline with which he helps students in completing
projects. I thank Dr.
Steven Nicholls for his support and for serving as a
dissertation committee member.
-
I also thank Dr. William Davros for his enthusiastic teachings
on medical physics and
for serving as a committee member.
I must also thank Dr. Peter Lewin at Drexel University. It was
his enthusiasm
for physics and medical applications of ultrasound that brought
me into the world of
ultrasonic imaging.
I extend my thanks to Dr. Nicholas Ferrel for culturing MDCK
cells and also
providing pancreatic and breast tumor cells; Ken Gorski and Bill
Magyar from IVUS
lab core for acquiring OCT images; Lindsey and Paul Bishop for
providing peripheral
arteries; Dr. Ofer Reizes for providing fat tissue samples; Dr.
Xuemui Gao, from the
laboratory of Dr. Linda Graham for providing rabbit aortic
grafts; Dr. Sanjay Anand,
from the laboratory of Dr. Edward Maytin for providing
adenocarcinoma samples
and helping me with mice experiments; Vivek from the laboratory
of Dr. George
Muschler for providing tissue scaffolds, and personnel from the
laboratory of Dr.
Ronald Midura for sharing osteoporotic bone samples. I would
like to thank CHTN
for shipping carotid arteries. I must thank Dr. Cheri Deng and
her student Yi-Sing
Hsiao, from University of Michigan, for allowing access to their
laboratory and take
measurements with their hydrophone.
I acknowledge Dr. Judith Drazba and her joyful team, Dr. John
Peterson and
Diane Mahovic for their efforts on sectioning and staining of
difficult samples.
It is an honor to thank Dr. Joanne Belovich, the program
director of Applied
Biomedical Engineering at CSU, for her support and timely advice
during difficult
-
times. It is an honor to thank Drs. Linda Graham and Marcia
Jarrett for their timely
advice.
Special thanks to all the secretaries for assisting me in many
different ways.
Ms. Rebecca Laird, who, even during her vacation days reminds us
of our deadlines,
secretly cares like a mother although she finds amusing to say I
am not your
mother. I cannot thank her enough for her time and efforts for
providing more than
administrative support throughout the years. Many thanks to Ms.
Darlene
Montgomery, who keeps her cool even when the masses annoy her
greatly, for her
support in many remarkable ways. Thanks to Jill Rusticelli and
Sandi Zelewensky for
handling my many requests for appointments with Drs. Nicholls
and Davros.
I would like to thank my friends and seniors Drs. Powrnima
Joshi, Srividya
Sunderaraman, Eun Jung Kim and Nicholas Ferrel for helping me
get through the
difficult times, and for all the emotional support, comradeship,
entertainment, and
caring they provided. Dr. Joshi was very instrumental in having
me complete my
thesis writing along with reminding me that sanity and happiness
are worth more,
when I lost my composure during chaotic discontinuities in the
laboratory. I would
also like to thank Marianne for her kindness and giving me
company when
experiments ran late into dark. Thanks to Dr. Judd Gardner for
encouraging me to
stay focused on my goals of completing the thesis during the
last few months. I
would also like to thank experienced wise individuals at
Cleveland Clinic, who wish
to remain anonymous, for offering guidance at variable
times.
-
I would like to acknowledge the funding sources for financial
support of my
studies: the American Heart Association, for the pre-doctoral
fellowship and the
Doctoral Dissertation Research Expense Award from CSU for
funding all my
materials, without which this thesis would not have been
possible.
I am indebted to the Physics and Chemical & Biomedical Eng.
departments at
CSU for granting me admission to the respective programs; I
enjoyed the memorable
lectures and every class kept me captivated by the wealth of
knowledge of the
professors. I also thank the CSU library and OhioLink, without
which I would not
have access to tremendous source of information and
textbooks.
I would not have been able to spend time in the laboratory
without the help
of sittercity.com. I would like to thank Dr. Sandra Halliburton
for recommending the
website. I extend my deepest thanks to all of the nannies, from
the special ones who
assumed the role of a grandmother, to the ones who burnt down
the kitchen.
Special thanks to my adorable children who went through
vulnerable periods
during my doctoral studies. I offer my apologies and infinite
thanks to them for
weathering difficult times and being resilient during the years.
I also thank my
husband, parents, sister, brother-in-law and extended family for
their support.
-
ix
DETECTING VULNERABLE PLAQUES WITH
MULTIRESOLUTION ANALYSIS
SUSHMA SRINIVAS
ABSTRACT
This thesis seeks to address the unmet need of identifying
vulnerable
plaques, which result in 75% of the acute coronary episodes.
With the limited
resolution of conventional IVUS transducers, the thin cap of the
fibroatheromas
cannot be identified before they rupture. This dissertation
evaluated the application
of harmonic imaging in characterizing lipid cores based on
nonlinear propagation.
The hypothesis is that the multiresolution analysis of IVUS
radiofrequency signals
with a focused broadband polymer transducer will result in
additional diagnostic
information. The rationale is that tissue nonlinearity has a
structural dependency
and the detection of this property can better resolve and
differentiate plaque
components.
As part of this study, the system linearity, essential for
harmonic imaging,
was established for a polymer micro-machined ultrasound
transducer (PMUT)
imaging device. Pressure profiles of PMUTs were measured with a
wideband
hydrophone. Nonlinear parameters of various fluids and fat from
biological
-
x
specimen were estimated. New methods using wavelets were
developed to
accurately measure the thin caps of fibroatheromas, to identify
lipids and to
estimate stent apposition. An algorithm based on velocity
inhomogeneity was
developed to differentiate lipids from necrotic regions. A
real-time synchronized
pullback system was developed.
Measurements from multiresolution analysis of thin caps in
excised human
coronary and carotid arteries (n = 5) ranged from 26 8 m to 73
28m. The
harmonic signals were better able to identify thin caps and
micro-calcifications than
in fundamental mode. Lipid accumulations, as thin as 200 m to
1.5 mm thick were
identified signifying the early detection of plaque formation
with wavelet analysis of
fundamental signals. However, the harmonic signals from lipid
regions in fresh
tissue were significantly weaker than harmonics from fixed
tissue. The specificity
and sensitivity of the new methods developed in this study need
to be evaluated
with more ex vivo coronary arteries. The successful adaptation
of these methods in
clinical imaging may enhance diagnostic capabilities and reduce
the incidence of
acute coronary syndrome.
-
xi
TABLE OF CONTENTS
Page
NOMENCLATURE
..........................................................................................................XX
LIST OF TABLES
.......................................................................................................XXIII
LIST OF
FIGURES......................................................................................................
XXIV
I
INTRODUCTION.............................................................................................................
1
1.1 Disease
.......................................................................................................
6
1.1.1 Morphology of coronary arteries
................................................. 7
1.1.2 Pathophysiology of atherosclerotic
plaque................................. 7
1.1.3 Remodeling
..................................................................................
10
1.1.4 Vulnerable Plaque
.......................................................................
12
1.1.5 Mechanisms of Plaque
Rupture.................................................. 13
1.1.6 Restenosis
....................................................................................
16
1.1.7 Risk factors
..................................................................................
16
1.1.8 Therapies
.....................................................................................
17
1.1.9 Reversal of CAD
...........................................................................
17
1.2
Diagnosis...................................................................................................
18
1.2.1 Biomarkers of vulnerable
plaque............................................... 19
1.2.2 Non-invasive imaging
.................................................................
20
-
xii
1.2.2.1 Magnetic Resonance Imaging
.................................................. 20
1.2.2.2 Computed Tomography Imaging
............................................ 21
1.2.2.3 Nuclear Imaging
.......................................................................
22
1.2.2.4 Hybrid Imaging PET/MR, PET/CT,
SPECT/CT.................... 23
1.2.3 Invasive
imaging..........................................................................
24
1.2.3.1
Angiography..............................................................................
24
1.2.3.2
Angioscopy................................................................................
25
1.2.3.3 Elastography
.............................................................................
25
1.2.3.4 Thermography
..........................................................................
26
1.2.3.5 Near infrared spectroscopy
..................................................... 27
1.2.3.6 OCT
............................................................................................
28
1.2.3.7 IVUS
...........................................................................................
29
1.3 Overview of limitations of Imaging Modalities
..................................... 32
II PROBLEM FORMULATION
.......................................................................................
34
2.1 Specific Aims
............................................................................................
37
2.2 Significance of this study
........................................................................
39
III MATERIALS AND METHODS
..................................................................................
40
3.1 Materials
..................................................................................................
40
3.1.1 PVDF-TrFE
..................................................................................
40
-
xiii
3.1.2 Reflectors
.....................................................................................
42
3.1.3
Amplifiers.....................................................................................
42
3.1.4 SMA cables
...................................................................................
43
3.1.4 Tissue specimen
..........................................................................
44
3.2 Making of the Device
...............................................................................
45
3.2.1 Fabrication of Transducer
.......................................................... 45
3.2.2 Preamplifier Circuit
.....................................................................
46
3.2.3 External Amplifier
.......................................................................
48
3.2.3 Testing of Transducers
...............................................................
49
3.3 Data Acquisition
......................................................................................
50
3.3.1 Synchronized pull back
...............................................................
50
3.3.2 Data acquisition system
..............................................................
51
3.3.3 Acquisition of IVUS RF harmonic signals
................................. 51
3.3.4 Processing of harmonic signals
.................................................. 54
3.3.5 Multi resolution analysis of harmonic signals
.......................... 54
3.3.6 Histological Correlation
..............................................................
55
3.3.7 Estimation of nonlinear
parameters.......................................... 56
3.3.8 Enhancement of spectral
parameters........................................ 58
3.3.9 Estimation of extent of neointimal
hyperplasia........................ 58
-
xiv
3.4 Imaging of various biological specimen
................................................ 59
3.4.1 Imaging of Carotid
arteries.........................................................
59
3.4.2 Imaging of Peripheral arteries
................................................... 60
3.4.3 Imaging of adenocarcinoma
....................................................... 60
3.4.4 Imaging of MDCK cells
................................................................
61
3.4.5 Imaging of scaffolds for tissue
engineering............................... 61
IV HARMONIC IMAGING
...............................................................................................
62
4.1 Development of Harmonics
....................................................................
62
4.2 Advantages of Harmonics
.......................................................................
65
4.3 Methods of Harmonic Imaging
...............................................................
66
4.3.1 Filters Approach
..........................................................................
66
4.3.2 Pulse Inversion Imaging
.............................................................
67
4.4 Harmonic Signal Processing
...................................................................
71
V MULTIRESOLUTION ANALYSIS
...............................................................................
72
5.1 Methods of analyzing a signal
................................................................
72
5.1.1 Fourier frequency analysis
......................................................... 73
5.1.2 Windowed Fourier Transform
................................................... 74
5.1.3 Wavelet Transform
.....................................................................
75
5.2 The uncertainty principle
.......................................................................
76
-
xv
5.3 Multiresolution
Analysis.........................................................................
76
5.4 Application in characterization of plaque
............................................. 78
VI RESULTS I
...............................................................................................................
80
6.1 PMUT characterization
...........................................................................
80
6.2 Device components characterization
.................................................... 82
6.2.1 Quarter Matching
.....................................................................
82
6.2.2 Minimum Gain Required on the
Preamplifier........................... 83
6.2.3 Operating range of Miteq Amplifier
........................................... 84
6.3 System linearity Aim 1(a)
....................................................................
85
6.3.1 Harmonic contribution from D/A
card...................................... 86
6.3.2 Harmonic contribution from the preamplifier
......................... 87
6.3.3 Harmonic contribution from other
amplifiers.......................... 87
6.3.4 Harmonic transduction from PVDF-TrFE film
.......................... 88
6.3.5 Optimal BW for transmit waveforms
........................................ 90
VII RESULTS II
............................................................................................................
92
7.1 Axial radiation
profiles...........................................................................
93
7.2 Lateral radiation profiles
........................................................................
95
7.3 2D radiation profiles
...............................................................................
97
7.4 Variability of Axial
Resolution................................................................
99
-
xvi
VIII RESULTS III
.......................................................................................................
100
8.1 Fluid nonlinearity Aim 1(a)
...............................................................
100
8.1.1 Distinct attenuation curves for
harmonics.............................. 100
8.1.2 Harmonic generation in fatty fluids
......................................... 102
8.1.3 Egg Yolk and Egg White
............................................................
103
8.2 Tissue
nonlinearity....................................................................................
104
8.2.1 Harmonic generation in diseased aorta
.................................. 104
8.2.2 Lipid nonlinearity
......................................................................
104
8.2.3 Nonlinearity of blood
................................................................
105
IX RESULTS IV
...........................................................................................................
107
9.1 Analysis with
wavelets..........................................................................
107
9.1.2 Uncovering nonlinearity
........................................................... 107
9.1.3 Seeing with
wavelets.................................................................
109
9.1.4 Precise measurements with MRA
............................................ 110
9.1.5 Pathological differences with harmonics
................................ 111
X RESULTS V
..............................................................................................................
112
10.1 Aim 1(b)
.................................................................................................
112
10.1.2 Fundamental and harmonic images of coronary artery ......
112
10.1.3 Fundamental and harmonic images from a porcine model.
114
-
xvii
10.1.4 Harmonic signal strength from healthy tissue
..................... 114
10.1.4 Utility of low signal strength harmonics
............................... 116
10.1.5 MRA identification of thin
cap................................................ 118
10.1.6 MRA identification of lipids
.................................................... 118
10.1.7 MRA identification of borders
................................................ 119
10.1.8 Characterization by velocity
differences............................... 121
XI RESULTS VI
...........................................................................................................
122
11.1 Aim
1(c)..................................................................................................
122
11.1.1 Extension of spectral parameters
.......................................... 122
11.1.2 Estimation of nonlinear parameters
..................................... 123
XII RESULTS VII
........................................................................................................
125
12.1 Aim 2(a-c)
..............................................................................................
125
12.1.1 Bare-metal stent in a silicone tubing
.................................. 126
12.1.2 Imaging of aortic grafts
........................................................ 126
12.1.3 Importance of focal
region................................................... 129
12.1.4 Harmonic imaging of stents
................................................. 130
12.1.5 MRA of harmonics and fundamental
..................................... 131
12.1.6 Identification of necrotic regions
........................................... 131
12.1.7 Stent
apposition.......................................................................
133
-
xviii
XIII RESULTS VIII
.....................................................................................................
135
13.1 Carotid arteries Aim 3(a)
...................................................................
135
XIV RESULTS IX
........................................................................................................
138
14.1 Cell clusters Aim
3(b).........................................................................
138
14.1.1 Ultrasound
bio-microscopy....................................................
139
14.1.2 Aim
...........................................................................................
140
14.1.3 Processing of echoes from cell
clusters................................. 140
14.1.4 Cell Culture
..............................................................................
142
14.1.4 Detection of inflection points
................................................. 143
14.1.5 Wavelet coefficient
reconstruction........................................ 144
14.1.6 3D reconstruction of cell clusters
.......................................... 145
XV RESULTS X
...........................................................................................................
147
15.1 Scaffolds for tissue engineering Aim 3(b) continued
...................... 147
15.1.1 Scaffolds
...................................................................................
148
15.1.2 2-dimensional
scaffold............................................................
148
15.1.3 3-dimensional
scaffold............................................................
149
XVI DISCUSSION
..........................................................................................................
151
XVII
CONCLUSION.......................................................................................................
163
REFERENCES.................................................................................................................
165
-
xix
APPENDICES
.................................................................................................................
189
APPENDIX A
.......................................................................................................
189
APPENDIX
A1.........................................................................................
190
APPENDIX
A2.........................................................................................
191
APPENDIX
A3.........................................................................................
194
APPENDIX
A4.........................................................................................
195
APPENDIX
A5.........................................................................................
196
APPENDIX
A6.........................................................................................
197
APPENDIX B
.......................................................................................................
198
APPENDIX
B1.........................................................................................
199
APPENDIX
B2.........................................................................................
200
APPENDIX
B3.........................................................................................
201
APPENDIX
B4.........................................................................................
202
APPENDIX
B5.........................................................................................
203
-
xx
NOMENCLATURE
ACS: Acute coronary syndrome
AHA: American Heart Association
ATCC: American Type Culture Collection
AMI: Acute myocardial infarction
CAD: Coronary artery disease
CHD: Coronary heart disease
CRP: C-reactive protein
CT: Computed tomography
CWT: Continuous wavelet transform
Db2, db4: Daubechies wavelets
DI: Deionized
F20: Fundamental 20 MHz
F40: Fundamental 40 MHz
18F: Flourine 18
18F-FDG: Flourine 18 Fludeoxyglucose
FT: Fourier Transform
EBCT: Electron beam CT
-
xxi
EC: Endothelial Cell
FHS: Framingham Heart Study
FIR: Finite impulse response
H40: Harmonic 40 MHz
H80: Harmonic 80 MHz
HPF: High pass filter
hs-CRP: High sensitivity C-reactive protein
HU: Hounsfield units
IL2: Interleukin 2
IVUS: Intravascular ultrasound
LAD: Left anterior descending
LDL: Low density lipoprotein
LPF: Low pass filter
MDCK:Madin Darby Canine Kidney cells
MDCT:Multi detector CT
MI: Myocardial infarction
MMP: Matrix metalloproteinase
MRA: MR Angiography / Multiresolution analysis
MRI: Magnetic resonance imaging
-
xxii
OCT: Optical coherence tomography
PE: pulse echo
PET: Positron emission tomography
PI: Pulse inversion
PMUT:Polymer micromachined ultrasound transducer
PSD: Power spectral density
PVDF-TrFE: Polyvinylidene fluoride trifluoroethylene
PZT: Lead Zirconate Titanate
SCD: Sudden cardiac death
SES: Sirolumis eluting stent
SMC: Smooth muscle cell
SNR: Signal to noise ratio
SPECT: Single photon emission computed tomography
99mTc: Metastable Technicium
TCFA: Thin cap fibroatheromas
THI: Tissue harmonic imaging
TIMP: Tissue inhibitor of metalloproteinase
UBM: Ultrasound biomicroscopy
WFT: Windowed Fourier Transform
-
xxiii
LIST OF TABLES
Table Page
Table 1: Classification by Committee on Vascular Lesions of the
Council on
Atherosclerosis of AHA 11
Table 2: Seven Category Classification by Virmani et.
al.,.12
Table 3: Imaging capabilities of various modalities w.r.t.
vulnerable plaque 33
Table 4: Range of Transducer Characteristic Parameters 81
Table 5: BW for different lengths of cable 83
-
xxiv
LIST OF FIGURES
Figure Page
Figure 1: Plaque rupture leading to death of heart muscle
........................................... 2
Figure 2: Illustration of normal and diseased human coronary
artery ........................ 8
Figure 3: Classification of atherosclerosis by Virmani et. al.,
...................................... 11
Figure 4: Different morphologies of vulnerable plaques
............................................. 13
Figure 5: Mechanism of plaque rupture
........................................................................
14
Figure 6: Illustration of IVUS catheter
...........................................................................
30
Figure 7: Various diagnostic methods for the detection of
vulnerable plaque .......... 33
Figure 8: 40 MHz PMUT transducer
..............................................................................
46
Figure 9: Preamplifier circuit for a PMUT
.....................................................................
47
Figure 10 : Experimental setup for tissue
imaging.......................................................
51
Figure 11: Excitation pulses for harmonic
imaging......................................................
52
Figure 12: Development of harmonics
..........................................................................
64
Figure 13: Pulse inversion technique
............................................................................
69
Figure 14: Decomposition with
MRA.............................................................................
78
Figure 15: PE and PSD of a high resolution transducer
............................................... 81
Figure 16: Demonstration of broad bandwidth of the PMT
transducer..................... 82
-
xxv
Figure 17: Operating Range of Miteq
Amplifier............................................................
85
Figure 18: Harmonic contribution from the source
..................................................... 86
Figure 19: Harmonic contribution from the preamplifier
........................................... 88
Figure 20: Frequency transduction of PVDF-TrFE and optimal
BW........................... 89
Figure 21: Axial radiation patterns of fundamental and harmonics
at 50 V .............. 93
Figure 22: Axial radiation patterns of fundamental and harmonics
at 100 V............ 94
Figure 23: Lateral radiation
profiles..............................................................................
96
Figure 24: 2D radiation profiles for 20 MHz
.................................................................
97
Figure 25: 2D radiation profiles for 40 MHz
.................................................................
98
Figure 26: Variability of axial
resolution.......................................................................
99
Figure 27: Distinct attenuation curves for fundamental and
harmonics ................. 101
Figure 28: Harmonics development in fatty fluids
..................................................... 103
Figure 29: Harmonic generation in diseased aorta
.................................................... 105
Figure 30: Nonlinearity parameter values of egg and mice fat
................................. 106
Figure 31: Egg dual bilayer membranes imaged with
harmonics............................. 108
Figure 32: Better Resolution and contrast with MRA
................................................ 109
Figure 33: Precise measurement of egg membranes with MRA
............................... 110
Figure 34: Pathological sections on different scales
.................................................. 111
-
xxvi
Figure 35: Fundamental and harmonic images of a fresh coronary
arterial section
.........................................................................................................................................
113
Figure 36: Fundamental and harmonic images from a control void
of lipids .......... 115
Figure 37: Harmonic signal strength from healthy tissue
......................................... 116
Figure 38: Significance of harmonics in imaging thin cap
......................................... 117
Figure 39: MRA of thin cap of
fibroatheromas............................................................
119
Figure 40: Lipid identification by MRA
.......................................................................
120
Figure 41: Characterization by measuring the change in velocity
............................ 121
Figure 42: Extension of spectral parameters from nonlinear
imaging..................... 123
Figure 43: Image generation based on differences between
fundamental and
harmonics
......................................................................................................................
124
Figure 44: Self-expanding stent imaged with IVUS, OCT and
PMUT......................... 127
Figure 45: Harmonic characterization of neointimal growth over a
graft ............... 128
Figure 46: Degradation of lateral resolution
..............................................................
129
Figure 47: Minimal harmonics from
restenosis..........................................................
130
Figure 48: MRA of fundamental and harmonics
......................................................... 132
Figure 49: Differentiating low echogenic regions
...................................................... 133
Figure 50: MRA evaluation of stent apposition
.......................................................... 134
Figure 51: Harmonic detection of thin cap of a carotid plaque
................................. 136
-
xxvii
Figure 52: Thin cap, lipid region and intimal thickening in
carotid arteries ........... 137
Figure 53: Setup for imaging cell clusters
...................................................................
143
Figure 54: Detection of inflection points
.....................................................................
144
Figure 55: Wavelet coefficient reconstruction
........................................................... 145
Figure 56: Reconstructed images of cells on membrane
........................................... 146
Figure 57: 3D reconstruction of cell clusters
..............................................................
146
Figure 58: Wavelet reconstruction of a 2D scaffold image
........................................ 149
Figure 59: Wavelet reconstruction of a 3D scaffold image
........................................ 150
Figure 60: PMUT& OCT image comparison of a stented artery
................................ 199
Figure 61: PMUT, OCT, Revo, HE of healthy artery
.................................................... 200
Figure 62: PMUT, OCT, Revo & HE of artery with intimal
thickening....................... 201
Figure 63: PMUT, OCT, Revo & HE of artery with thin cap
........................................ 202
Figure 64: 0.8 mm PMUT images of stent apposition
................................................ 203
Figure 65: 0.6 mm PMUT images of stent apposition
................................................ 204
-
1
CHAPTER I
INTRODUCTION
June 13th 2008 Tim Russert died at the age of 58 after
collapsing at work.
The untimely death of the NBC host had many of us have the
alarming thought of
could it happen to me? Mr. Russerts autopsy confirmed the
rupture of a
cholesterol plaque in a branch of the LAD, causing sudden
cardiac death. Sudden
death is ancient to humans and the earliest record of sudden
death possibly due to
atherosclerotic coronary occlusion is suggested in an Egyptian
relief sculpture from
the tomb of a noble of the Sixth Dynasty ( 2625- 2475 B.C.) [1].
Although FHS data
from 1950 to 1999 suggests 49% decline in sudden deaths, SCD
claims 300,000 lives
in the US every year [2]. Unfortunately, the difficulty with
diagnosing the risk for
SCD is that, in many people, SCD is the first and last
manifestation. 50% of men and
64% of women who die of sudden CHD have no symptoms prior to the
acute event
[2].
Mr. Russert had passed the exercise stress test just 2 months
prior to his
death but autopsy showed significant blockages in several
arteries [3]. The severity
and the anatomical status of CAD remain undetected without an
appropriate
-
2
diagnostic test. Plaque rupture can be silent and the lack of
symptoms would not
suggest an invasive test needed to make a definitive diagnosis.
An illustration of the
blockage in the artery due to plaque rupture is shown in Figure
1.
Figure 1: Plaque rupture leading to death of heart muscle
There are several non-invasive and invasive diagnostics tests
for the
estimation of extent of CAD. Several noninvasive methods have
been demonstrated
to be of clinical value, but serious difficulties due to the
small size of the coronary
arteries, cardiac and respiratory motion, flow disturbances,
challenging anatomy
-
3
and mainly the limited spatial resolution need to be overcome.
If noninvasive
diagnostic modalities were to be routine examinations and
tomographic view of the
arterial system could be obtained, noninvasive methods still
lack the resolution
needed to diagnose early stage disease as well as the culprit
lesions smaller than the
imaging device limit. Due to the limited resolution, noninvasive
modalities tend to
focus on managing the disease by the estimation of stenosis that
is
hemodynamically significant. In 85% of the ACS, the culprit
lesions were less than
70% stenotic prior to rupture. This might explain why managing
hemodynamically
significant stenoses have not proven effective in predicting SCD
[4, 5].
Among the invasive diagnostic tests, X-ray angiography has been
considered
the gold standard for defining the degree of stenosis. Other
main clinically available
modalities are OCT and IVUS. Several studies have dispelled the
skepticism towards
the accuracy and reliability of both IVUS and OCT. The use of
OCT as an
intracoronary imaging modality has been growing and has shown
significance in
successful outcomes [6, 7]. IVUS offers tomographic
visualization of the arteries but
with limited resolution compared to OCT, with the current
clinical IVUS catheters.
The advances in IVUS have resulted in automated plaque
characterization and 3D
visualization but the efficacy of these methods in identifying a
vulnerable plaque is
yet to be proven. These invasive methods are not called for
unless the patient
presents with symptoms and is first diagnosed by a noninvasive
modality. This is
mainly due to the lack of detection capability of the current
invasive techniques in
identifying the early stage disease and also the cost of an
additional procedure. The
-
4
goal is to identify late stage disease to prevent acute events
and also the early
diagnosis of the disease with accuracy and reliability.
This dissertation describes my attempts at imaging the human
coronary
arteries in an effort to detect mainly the lipid pools and thin
caps of vulnerable
plaques, not possible at this time. Multiresolution analysis
with wavelets is the
approach employed for my hypothesis.
Section 2 of this chapter describes the atherosclerotic disease
manifestations,
causes, prevention and treatment. Section 3 describes the
current methods of
diagnosing atherosclerotic plaques. Both non-invasive and
invasive methods, their
merits and limitations are discussed.
Chapter 2 formulates the medical problem, states the hypothesis
and lists the
specific aims of this thesis which test the hypothesis, that
multiresolution analysis of
IVUS signals lead to better classification of plaques.
Chapter 3 describes the materials and the methods that are
common to most
of the experiments conducted during my research. Transducer
materials and
various components used are explained. The synchronized data
acquisition system
is described. Experimental protocols of imaging tissue specimen
and signal analysis
are also detailed.
Chapter 4 connects harmonic imaging to the hypothesis and
describes
development of harmonics by nonlinear propagation in biological
tissue.
-
5
Chapter 5 describes various methods of signal analysis, the
Heisenberg
uncertainty principle and application of multiresolution
analysis for the
characterization of plaques.
Chapter 6 presents the transducer characteristics that are
fundamental to
acquiring signals of good quality. The transducer and the
various electronic
components are tested for linearity and any nonlinear modes of
operation are
discussed.
Chapter 7 presents the acoustic pressures radiated by the PMUT
as measured
by a hydrophone.
Chapter 8 presents results from experiments demonstrating
nonlinearity of
fluids and tissue specimen.
Chapter 9 shows how multiresolution can be applied for
plaque
characterization and identification of nonlinear components.
Chapters 10 through 12 present the results of specific aims
using coronary
arteries.
Chapters 13 through 15 present results of imaging various other
biological
specimens like the carotid arteries, cell clusters and tissue
scaffolds.
In the Discussion, Chapter 16, the results are examined; the
conclusions and
future research are provided in Chapter 17.
-
6
1.1 Disease
Hurry, Worry & Curry Recipe for Heart Disease.
-Teachings of Sathya Sai Baba on health by Srikanth Sola,
M.D
Atherosclerosis, the primary cause of heart attack, stroke and
other
conditions of the extremities remains a major contributor to
morbidity and
mortality. Atherosclerosis originates from Greek words atheros
meaning gruel, a
soft pasty material corresponding to the necrotic core in the
arterial wall and
sclerosis meaning hardening or indurations matching the thin cap
of the plaque.
With increasing age, arterial walls thicken leading to focal
atherosclerotic lesions
that eventually advance to complex plaques that could block the
lumen limiting
blood flow or rupture generating a thrombus leading to total
occlusion. Several risk
factors like high cholesterol diet, smoking, metabolic-syndrome,
diabetes, obesity,
psychological stress along with predisposition to genetic
background induce
atherosclerosis [4, 5]. Atherosclerosis is a progressive
systemic disease. However,
the plaque pathology differs depending on the vascular bed [8].
Although sections
from other sites like renal, peripheral and carotids were also
imaged in this study
due to lack of availability of coronary arteries, the plaque
characteristics described
in this section refer to the coronary plaques as the number of
studies reporting the
differences in vascular beds are very few.
-
7
1.1.1 Morphology of coronary arteries
Coronary arteries are muscular and comprise three layers:
intima, media and
the adventitia. The internal and external laminae separate the
intima-media and the
media-adventitia layers respectively. Intima can vary in
thickness. The thinnest
segments of the intima comprise the endothelium, basement
membrane and
subendothelial layer, which consist of elastin, collage,
proteoglycans, and scattered
smooth muscle cells. Thicker segments express a layer of
longitudinally aligned
SMCs that originate in the medial layer and internal elastic
lamina. Adventitial layer
is comprised of elastic fibers, collagen and fibroblasts. Vasa
vasorum, the
microvasculature that nourish the arteries and nerve fibers are
found in the
adventitia. Healthy arteries do not exhibit advanced lesions in
the arterial wall.
Atherosclerotic lesions occur more frequently in certain sites
on the
coronary tree. The left coronary artery has a higher incidence
where the trunk
bifurcates, proximal to the LAD and circumflex. Lesions are seen
more in the
proximal and middle segments [9].
1.1.2 Pathophysiology of atherosclerotic plaque
Pathological states can be reached by different mechanisms.
Based on new
insights, due to progress in cell and molecular approaches,
these mechanisms can be
summarized in to three main hypotheses response to injury,
oxidized LDL and
inflammation [10-12]. Response to injury due to mechanical
stress from variation
-
8
in the flow, wall tension and maturity often manifest as the
variation in the intimal
thickness. This is more pronounced at the bifurcations or side
branches, which are
predisposed to atherosclerotic lesions [9]. Oxidized LDL
hypothesizes that LDL in
the blood oxidized by macrophages and SMCs that form cholesterol
clefts within the
arterial wall contribute to atherosclerosis [11]. Inflammation
hypothesis postulates
that immune cells interact with various metabolic risk factors
to progress the
disease from initiation to terminal thrombogenic state [12].
These mechanisms
result in activation and alteration of the intima, media and
adventitial layers leading
to the formation of atherosclerotic plaques that further
progress to advanced
lesions. Figure 2 illustrates normal and diseased human
arteries.
Figure 2: Illustration of normal and diseased human coronary
artery
-
9
In the diseased state, intima thickening may be eccentric,
diffuse or
circumferential. An eccentric bell shaped thickening is commonly
seen [13]. Intima
to media thickness varies from normal ratio of 0.1-1 to 4.1 in
the age-related disease
[14]. Activated ECs in the intima lead to degradation of the
ECM, proliferate and
migrate to initiate angiogenesis, a process which has been shown
to partake in many
pathological conditions. Proliferation and migration of ECs
through ECM is
facilitated by the integrin 3 and integrin 3 stimulated MMP-2
degradation of
ECM [15]. ECs maintain the vascular tone and hence blood
pressure by, the
controlled release of vasodilators like, NO, prostacyclin, and
PGI2, and
vasoconstrictors like endothelins and PAFs. In a normal state,
NO inhibits platelet
adhesion, leucocyte adhesion and injury induced neointimal
proliferation. Shear
stress alters the production of NO and thus affects various
regulatory mechanisms
[16]. An activated endothelium due to inflammation expresses
adhesion molecules
resulting in binding and extravasation of leucocytes [17].
ECs in an inactivated state prevent the proliferation of SMCs
and when
activated, have mitogenic effect on SMCs by the secretion of
PDGF along with other
growth factors [18]. The media in a healthy artery is about 100
m [14]. The
function of SMCs is to contract and serve as an elastic
reservoir from the pulse of the
blood flow. The main pathologies of SMCs are vasoconstriction
and hypertension. In
response to vascular injury, SMCs proliferate into the intima
and stabilize a
developing plaque by forming a fibrous cap [16].
The onset of plaque formation occurs in early childhood leading
to fatty
streaks or xanthomas [19]. Fatty streaks are fat-laden
macrophages in the intima.
-
10
One or many mechanisms of disturbance of the endothelium result
in the immune
cell adhesion to ECs and migration through ECs to capture LDL to
form foam cells. In
case of pathological intimal thickening, extracellular lipids
accumulate and appear
slightly raised and yellowish in color to naked eye. SMCs may
also contain lipids.
Secretion of MMPs result in degradation of the ECM and apoptosis
of macrophages
and denudation of the ECs resulting in a lipid core separated
from the lumen by a
fibrous cap/capsule. Lipid core is made up of necrotic remains,
cholesteryl esters,
lipoproteins and phospholipids. The size of the lipid core
depends on the number of
macrophages in the lesion [20]. The lipid core and the thickness
of the fibrous cap
are inversely related [21]. Thin capsules have less collagen,
abundant macrophages
and other inflammatory cells and loss of SMCs due to MMPs [22].
Such fragile spots
are found in the regions where the plaque meets the unaffected
part of the artery.
Such a region is termed shoulder of the plaque, Plaques with a
large lipid core with
a thin cap infiltrated by macrophages are termed thin cap
fibroatheroma (TCFA).
Different classifications of atherosclerotic lesions based on
lipid content and the
fibrous cap have been proposed and are as shown in Figure 3 and
Table 1 and Table
2 [19, 23].
1.1.3 Remodeling
The process of increasing the lumen size in order to accommodate
the blood
flow and wall tension is called remodeling [24]. The vessel wall
reorganizes its
cellular and extracellular components in early stage disease,
prior to significant
-
11
luminal stenosis [25]. Remodeling is bidirectional. Plaques
responsible for ACS often
show outward remodeling preserving the lumen size [26]. Plaques
causing stable
angina usually present inward growth resulting in lumen
constriction.
Figure 3: Classification of atherosclerosis by Virmani et.
al.,
Table 1: Classification by Committee on Vascular Lesions of the
Council on
Atherosclerosis of AHA
Type I Fat-laden macrophages
Type II Fatty streak. Lipids remain intracellular
Type III Pre-atheromatous lesion. Extracellular lipids
Type IV Fibrolipid. Soft plaque defined capsule and lipid
core
Type V Hard plaque collagen and SMCs
Type VI Complicated lesion
-
12
Table 2: Seven Category Classification by Virmani et. al.,
Non-atherosclerotic lesions Intimal thickening, intimal
xanthoma
Progressive atherosclerotic lesions
Pathological intimal thickening, fibrous
capsule, thin cap fibrous atheroma (TCFA),
calcified nodule, fibrocalcific plaque
1.1.4 Vulnerable Plaque
Some of the other terms for vulnerable plaque are high risk
plaque,
thrombosis-prone plaque, unstable plaque and TCFA. The following
types are
considered vulnerable: TCFA, sites of erosion, some plaques with
calcified nodules.
Although the plaques with large lipid cores and thin caps
(inflamed TCFA) are
strongly suspected to be vulnerable, there appear to be plaques
without these
features to be thrombogenic that also lead to ACS [27]. In a
study involving SCDs,
thrombosis was seen at eroded sites, sites other than thin cap
and lipid pool which
are considered vulnerable [28]. Such plaques at sites with
erosion expressed
increased proteoglycans. Another study identified a calcified
nodule to be
potentially vulnerable [29, 30]. Different morphologies of
plaques that are
considered vulnerable are shown in Figure 4. It is also known
that TCFAs can be
found at autopsy suggesting the low specificity of TCFA as
vulnerable [30]. There is
still not a prospective definition or a prospective method of
identifying vulnerable
plaques.
-
13
Figure 4: Different morphologies of vulnerable plaques
1.1.5 Mechanisms of Plaque Rupture
A number of intrinsic and extrinsic factors contribute to plaque
vulnerability
size of lipid core, thickness and collagen content of the
fibrous cap and
-
14
inflammation within the plaque. Factors like hemodynamic stress
may cause cap
disruption. An illustration of plaque rupture is shown in Figure
5.
Figure 5: Mechanism of plaque rupture
-
15
Endothelial cells are exposed to hydrostatic forces by the
blood,
circumferential stress caused by the vessel wall and the shear
stress caused by
blood flow. According to Laplaces law, the wall tension
developed is directly
proportional to the pressure on the wall and the luminal
diameter. This
phenomenon may lead to unbearable stress on the thin cap and at
the shoulder of
the plaque [31]. In case of fibrous caps, a moderately stenosed
plaque may be at
higher risk for rupture than a severely stenosed plaque due to
higher wall tension in
the former type [32-34].
Lipid core size and consistency are also factors that contribute
to plaque
rupture. It has been shown that a large proportion of disrupted
plaques were
occupied by lipid rich core than intact plaques causing < 70%
stenosis [35].
Most vulnerable area of the plaque is the shoulder region where
the cap is
the thinnest [36]. Reduced collagen content in the cap also
increases the risk of
rupture. Also a reduction in the SMCs in the fibrous cap would
destabilize the plaque
[37].
Neovascularizations are seen in plaques and may be involved in
plaque
disruption. The postulation is that the newly formed vessels are
fragile and thus
promote intra-plaque hemorrhage increasing the lipid volume
further leading to
unbearable stress on the thin cap [38].
-
16
1.1.6 Restenosis
Restenosis is the re-narrowing of the arterial lumen after an
intervention to
such as endarterectomy, bypass grafting and intraluminal
approaches (angioplasty,
atherectomy, stent angioplasty) to enlarge the stenosed lumen.
Greater than 20% of
interventions fail due to restenosis. Failures occur 12 months,
failure occurs due to underlying atherosclerosis [39].
Restenosis can result due to elastic recoil of the artery within
minutes of angioplasty
intimal hyperplasia in case of stenting, reorganization of
thrombus, and remodeling.
Remodeling seemed to show greater loss of luminal area than
intimal hyperplasia
[40]. In case of restenosis, a neointimal response to injury (by
stenting, surgery or
angioplasty) is seen where the VSMCs proliferate creating a
thickened intima. The
rates of restenosis at 20% 40% is similar in all vessels. In 30%
of the cases,
restenosis leads to significant luminal stenosis [41]. Efforts
to limit restenosis may
involve targeted drug delivery, genetic therapies and improving
the resistance of
vascular beds.
1.1.7 Risk factors
Some of the risk factors for CHD are family history, smoking,
hypertension,
dyslipidemia (elevated LDL, low levels of HDL, elevated
triglycerides), metabolic
syndrome, diabetes, obesity, reduced fitness, and psychological
risk factors
(depression, hostility, anxiety, stress) [3].
-
17
1.1.8 Therapies
Attempts to stabilize vulnerable plaques have been made by
targeting
different pathways leading to plaque rupture. Some of them are
endothelium
passivation by increasing the antioxidant NO by physical
exertion, by reducing LDL
deposition by statins, MMP inhibition by TIMPS or doxycycline,
and by increasing
collagen deposition [42, 43]. High levels of HDL show marked
positive influence on
endothelial function and also the reversal of lipid accumulation
in the arterial wall
[44].
1.1.9 Reversal of CAD
Making healthy dietary and lifestyle changes can delay and, even
reverse
heart disease after one year. These lifestyle changes include
whole foods, plant-
based diet, smoking cessation, routine physical activity and
stress management.
This was scientifically demonstrated by the Lifestyle Heart
Trial and prior studies
[45, 46] . Regression of the disease was seen to be more in 5
years than 1 year in the
experimental group, whereas, the disease progressed and more
cardiac events
occurred in the control group.
The next section gives a review of latest diagnostic methods of
identifying a
vulnerable plaque.
-
18
1.2 Diagnosis
A new scientific truth does not triumph by convincing its
opponents and making
them see the light, but rather because its opponents eventually
die, and a new
generation grows up that is familiar with it.
Max Planck
During the evolution of CAD to MI, atherosclerotic plaques
undergo
progression and cause ischemic events either by direct luminal
stenosis or by an
occlusive thrombus. Estimates show that 13 million individuals
suffer from
coronary artery disease (CAD), 75% of acute coronary episodes
are due to plaque
rupture and 87% of all strokes are ischemic [47]. Detection of
atherosclerosis at an
early stage may recognize vulnerable patients at an early stage
of CAD and help
undertake preventive measures. Several diagnostic imaging and
physiology based
detection modalities have attempted to identify the vulnerable
plaque. Each
modality offers unique diagnostic information which in the
future may be combined
to help make integrated clinical decision in identifying a
vulnerable patient. The
characteristics of vulnerable plaque are: size of lipid core
(40% of entire plaque),
thickness of fibrous cap (23 19 m to 150 m), presence of
inflammatory cells,
amount of remodeling and plaque-free vessel and 3D morphology
[23, 48, 49].
-
19
1.2.1 Biomarkers of vulnerable plaque
Markers are molecules that leave the site of plaque and enter
the
bloodstream for detection peripherally. There may be unique cell
types expressed in
the blood due to CAD as well. Cholesterol and lipid content
estimation are poor
markers of sudden events as fewer than 50% of the patients with
ACS have elevated
lipid levels. Five inflammation-sensitive plasma proteins when
elevated along with
hypercholesterolemia have been associated with high risk for
stroke and MI,
whereas without elevation, proteins did predict high risk [50].
Studies with specific
immunoassay detection of oxLDL in the plasma show elevated oxLDL
in CAD
patients [51]. Studies show that CRP is directly associated with
plaque formation
[52, 53]. CRP stimulates additional inflammatory molecules and
its opsonization of
LDL mediates uptake by macrophages [53, 54]. Although hs-CRP
elevations
correlate with ACS, correlation with histopathology is poor [55,
56]. Soluble and
membrane bound CD40 ligand levels have been shown to be elevated
in unstable
angina patients [57, 58]. MMPs are extracellular enzymes and are
found in plaques and
ingest lipids. High blood levels of MMP-2 and MMP-9 were found
in patients with
ACS compared with stable angina patients [58]. The successful
identification of a
biomarker of vulnerable plaque could lead to non-invasive tests
for ACS.
-
20
1.2.2 Non-invasive imaging
The desirable goal in order to manage patients with ACS is the
non-invasive
identification of vulnerable plaque.
1.2.2.1 Magnetic Resonance Imaging
MR differentiates plaque components based on the biophysical
and
biochemical properties. In vivo MR plaque imaging is achieved
with high resolution
sequences like FSE and black blood spin echo [59, 60]. Bright
blood imaging is
employed to image the fibrous cap thickness [60].
Characterization is usually based
on the signal intensities and plaque appearance on T1-weighted
proton density-
weighted and T2-weighted images. Calcifications, due to their
low mobile proton
density, can be identified by signal loss [61]. Fibrocellular
regions provide high
signal intensities in all weightings, and lipids present with
low signal on T2w and
hyperintense on T1w [62]. High resolution black blood MRI of
normal and
atherosclerotic human coronary arteries showed statistically
significant differences
in the wall thickness and no change in lumen area due to outward
remodeling [63].
This study required breath holding and this was eliminated by
employing
respiratory gating and slice position correction [64, 65].
Respiratory gating
provided a quick way to image a long segment of the coronary
artery.
Dynamic contrast enhanced MRI with gadolinium as the signal
enhancing
contrast has been used in preliminary studies to image
inflammation through
-
21
identifying neovascularization of atherosclerotic plaque in
human carotid arteries
[66]. The low molecular weight of the contrast agent diffuses
rapidly aiding the
early detection of binding after injection. Human studies with
SPIO contrast agents
that result in signal loss on T2*-weighted sequence, showed the
accumulation of
iron oxide particles in the macrophages within carotid plaques
[61, 67]. Further
development on T2*-effects should allow for better detection of
iron oxide
accumulation within the plaque [68, 69].
1.2.2.2 Computed Tomography Imaging
Due to its high sensitivity to calcifications, CT has become the
established
method for calcium scoring [70]. However, sensitivity for
earlier stage disease is
lower due to lack of in-plane spatial resolution. Complex
plaques in the vicinity of
high calcifications may be difficult to assess due to the same
reasons [71]. MDCT and
EBCT allow faster acquisition than spiral CT [72]. EBCT showed
good correlation
with non EBCT systems in assessing the volume of calcium [73,
74]. 16CDT provides
voxels with improved spatial resolution on the order of
sub-millimeter. Beam
hardening artifacts of calcium are thus reduced due to reduced
partial volume effect
[75].In vivo study using contrast enhance MDCT showed good
correlation in
differentiating soft, intermediate and calcified plaques as
compared to IVUS [76].
Intravascular thrombi appear with low attenuation of 20 -30 HU
[74]. Non-calcified
plaques and blood have similar attenuation (50 70 HU).
Significant enhancement
-
22
of the vessel over the non-calcified plaques is achieved by a
contrast medium (200
HU) [76].
Contrast enhanced CTA for plaque characterization is although
challenging, it
has been demonstrated that CTA can assess plaque area, density
and volume with a
good correlation with IVUS examinations [77, 78]. A study
examining 10, 037
coronary arterial segments from 1059 patients suspected of CAD
reported the use of
contrast enhanced CTA in identifying vulnerable plaques before
an acute event [79]!
The same study also had the findings of more frequent spotty
calcification and
extensive remodeling in patients who had an ACS in the follow up
duration of 27
months.
With improved spatial resolution from 320 and 256 DCT and better
temporal
resolution from the dual source CT, better characterization and
identification of
vulnerable plaques can be achieved [80-82].
1.2.2.3 Nuclear Imaging
PET and SPECT benefit from the ability to detect low
concentrations of
radiotracers but lack resolution compared to other imaging
modalities.
Radioisotopes are labeled with molecules that localize to
certain regions and can be
imaged with non-invasive tomographic scintigraphy. PET (3-4 mm)
has a superior
resolution than SPECT (10-15 mm). Capability of SPECT to image
MMP activation
and degradation of the fibrous cap was demonstrated by the
accumulation of the
labeled radiotracer 3 times greater in the affected plaque
compared to unaffected
-
23
regions [83]. Higher resolution images of the same can be
obtained with the new
MMP inhibitor labeled 18F for PET imaging [84]. Since
macrophages and leukocytes
demonstrate increased oxidative metabolism and glucose use, 18F
FDG is used to
predict plaque rupture and clinical events [85]. Although higher
uptake of FDG is
seen in plaques that progress to rupture and thrombosis, FDG can
also accumulate
in the ECs and lymphocytes, reducing specificity [86-89].
Tracers more specific than
FDG are being developed. Coronary artery imaging has the issues
of respiratory
movement, myocardial FDG uptake and the small size of the
coronary arteries.
1.2.2.4 Hybrid Imaging PET/MR, PET/CT, SPECT/CT
The high sensitivity of nuclear imaging methods when combined
with higher
resolution modalities like CT and MR provide better
understanding of the disease
characterization along with better anatomical information. A
study using SPECT/CT
tracked indium-labeled monocytes to the plaque regions [90].
Another study
tracked T lymphocytes to culprit lesions in case of patients
awaiting carotid
endarterectomy using 99Tc labeled IL2 and a significant
reduction of the tracer
uptake was seen after statin therapy [91]. The limitation of
partial volume effect
with PET is now being overcome with the PET/MR coupling where
the exact volume
can be identified with MR [92].
-
24
1.2.3 Invasive imaging
Noninvasive identification of vulnerable plaque must be the
ultimate goal in
order to arrive at a cost-effective solution with minimal risk.
Most noninvasive
modalities face the drawbacks of coronary artery motion, small
size and the
location. With several competing invasive techniques, the
initial prospective
identification of vulnerable plaques may be achieved by an
intracoronary modality.
1.2.3.1 Angiography
Coronary angiography has been the gold standard for estimating
luminal
narrowing. Angiography can assess lumen borders, but not the
plaque morphology,
components and the severity of the disease. Remodeling
phenomenon affects most
coronary lesions and preserves the luminal area and hence is not
detected by
angiography [93-96]. Diffuse nature of atherosclerosis results
in underestimation of
the stenosis. Concentric and symmetrical disease may give the
appearance of a
completely normal artery under angiography [93-95]. The
interobserver and
intraobserver variability is high when the stenosis is 30-80% of
the diameter [97].
The predictive power of angiography is low since 70% of the
acute events occur
despite normal angiograms [98]. Also, studies show that in
48-78% of the MI
patients, stenosis is
-
25
has a low discriminating power to identify vulnerable plaques,
it provides
information on the entire coronary tree and serves a guide for
invasive imaging and
therapy.
1.2.3.2 Angioscopy
Thrombi, plaque surface and ruptures can be directly visualized
with
intracoronary angioscopy. Extent of the disease is diagnosed by
the color of the
plaque. Multiple yellow plaques indicating higher plaque
instability were seen in all
three coronary arteries in patients with MI [102]. ACS occurred
more frequently in
patients with yellow plaques than in patients with white plaques
[103]. Angioscopy
requires the total occlusion of the artery and blood flushed out
with saline which
may induce ischemia. Angioscopy can be performed in a limited
part of the vessel.
1.2.3.3 Elastography
Elastography is based on the principle that deformation or the
strain of a
tissue is related to its mechanical properties. Ultrasound is
used as a stressor and
the strain per angle is plotted as a color-coded contour of the
lumen. Increased
circumferential stress leads to increased radial deformation of
the plaque
components. Typically, for pressure differences of 5 mmHg, the
strain induced is 1%
which requires sub-micron estimation of the deformation. Speckle
tracking in video
signals is the main method of using elastography. For
intravascular purposes a
correlation based elastography is employed. The displacement of
the vessel wall and
-
26
the region in the plaque are found by cross-correlation. The
strain of the tissue is
then found using the differential displacement between the two.
This method is
suited for strain values
-
27
between patients with stable angina, unstable angina and acute
MI [112]. The
thermistor of the catheter has a temperature accuracy of 0.05 C,
time constant of
300 ms and a resolution of 0.5mm. It was also seen that patients
with higher
temperature gradient have a significantly worse outcome than
patients with a low
gradient [113].
1.2.3.5 Near infrared spectroscopy
Molecular vibrational trasnsitions measured in the near infrared
region
(750-2500 nm) gives the chemical composition, qualitative and
quantitative
information about the plaque components. When a molecule is
exposed to infrared
radiation, the atoms absorb a portion of the light at
frequencies that induce physical
changes in the molecule. A spectrometer measures the frequencies
of the radiation
absorbed by the molecule as a function of energy. The magnitude
of absorption is
related to the concentration of species within the material.
Combinations of carbon-
hydrogen and carbon-oxygen functional groups, water and other
components in
tissue result in characteristic absorbance patterns. The
presence or absence of
particular frequencies is the basis for tissue characterization.
Photons in the NIR
region penetrate the tissue well and no preparation of the
sample is necessary. Also,
the hemoglobin has relatively low absorbance making diffuse NIR
spectroscopy an
attractive technique [114]. Algorithms have been developed to
identify lipid pools
like the partial least squares discriminate analysis [115].
PLS-DA model was able to
distinguish lipid pool and other tissue samples through up to
3mm of blood with at
-
28
least 86% sensitivity and 72% specificity [116]. The issue of
probe illumination area
of 1cm in diameter that may result in misclassification needs to
be resolved. A 3.2 Fr
NIR catheter has been developed for in vivo validation.
1.2.3.6 OCT
OCT measures the intensity of the back-reflected light with a
Michelson
interferometer technique. Wavelength of 1300 nm is used since it
minimizes the
energy absorption by vessel wall components. The light is split
into two signals. One
is sent into the tissue while the other to a reference arm with
a mirror. Both signals
are reflected and cross-correlated by interference of the light
beams. The mirror is
dynamically translated to achieve incremental cross-correlation
with penetration
depths in the tissue. High resolution images ranging from 4 m to
20 m can be
achieved with a penetration depth of up to 2 mm [117]. The frame
rate is ~15
frames/sec. Lipid pools generate decreased signal intensity
compared to fibrous
regions [118]. Compared to IVUS, OCT demonstrates superior
delineation of the thin
caps or tissue proliferation [119]. OCT can also be used in
pharmacologic or catheter
based interventions like stenting. This high resolution
technique has shown to
detect few cell layers of neointimal growth after an
intervention [120]. In vitro
characterization of plaques with OCT demonstrated high
sensitivity of 79%, 95%,
90% and specificity of 97%, 97%, 92% for fibrous, fibrocalcific
and lipid-rich
regions respectively [121]. In vivo studies show that OCT can
identify intimal
hyperplasia and lipid pools more frequently than IVUS [122]. A
study at 6-month
-
29
follow-up after drug eluting stent placement, OCT identified
neointimal coverage of
SES that could not be detected with IVUS [6]. A recent study
with AMI patients, the
incidence of plaque rupture was 73% with OCT compared to 47% and
40% with
angioscopy and IVUS respectively [123]. In the same study, the
thin cap was
estimated as 49 21 m. Limitations are low penetration depth and
light
absorbance and scattering by blood which requires saline
infusion.
1.2.3.7 IVUS
Conventional IVUS is based on the intensity of the backscattered
echoes.
Lumen and the vessel wall can be visualized in real time and
with high resolution.
Current IVUS catheters for coronary imaging have a center
frequency of 25- 40 MHz
with theoretical resolutions of 31-19 m respectively. The axial
resolution is ~80
m and the lateral resolution about 300 m. Frame rate is
30frames/sec [95]. An
illustration of the IVUS catheter is shown in Figure 6.
Studies comparing IVUS and histology show that the plaque
calcification can
be detected with a sensitivity of 86-97% [124, 125]. Sensitivity
for
microcalcification is ~60% [126]. Lipid pools are detected with
sensitivity of 78-
95% and a low specificity of 30% due to misclassification of
echolucent areas by
necrotic tissue [127, 128]. Positive remodeling associated with
unstable plaques
may be classified as high risk with IVUS [129]. In a follow-up
study of 114 patients,
patients who experienced ACS were found to have eccentric
plaques at the time of
previous IVUS imaging [130]. A study reported that IVUS guidance
during DES
-
30
implantation has the potential to influence treatment strategy
and reduce both DES
thrombosis and the need for repeat revascularization [131].
Figure 6: Illustration of IVUS catheter
3D IVUS has led to important observations regarding the
longitudinal extent
of plaque and restenosis after coronary interventions [132].
Bi-plane angiography is
used along with IVUS that produce more accurate 3D images [133].
Three-
dimensional IVUS (3D-IB-IVUS) allows volumetric reconstruction
of sequential
circumferential scans 1mm apart. RF Integrated backscatter
obtained with a
conventional 40 MHz IVUS catheter is color coded for better
plaque
-
31
characterization. The applicability of 3D-IB-IVUS in detecting
reduction in lipid
volume after 6 months of statin therapy and also quantification
of the increase in
fibrous region of the plaque volume was reported [134, 135]. In
this study, changes
were seen without any significant change in the lumen area and
hence suggest that
this technique is able to identify early changes in plaque
characteristics.
Spectral analysis of IVUS backscatter has led to classifying
lesions as calcified,
fibrofatty, calcified-necrotic core, and lipid-rich areas [136].
This study assessed
various spectral algorithms like the classic Fourier transform
(CPSD), Welch power
spectrum (WPSD) and autoregressive models (MPSD) and found that
the
autoregressive classification tree provided the best correlation
with histology. The
algorithm accepts two borders luminal and media-adventitial
border. For each
window of 480 m within a scanline, eight frequency domain
features are estimated
and each combination of these parameters was mapped to one of
four histologically
derived categories. The predictive accuracy was ~80% for all
four tissue types.
Limitation of VH is that calcification from necrotic core cannot
be distinguished.
Also there is a 480 m window over which the parameters are
estimated. It is
questionable when parameters over a smaller region are estimated
will show any
significance to characterization. A recent study evaluated the
feasibility of
combined use of VH IVUS and OCT for detecting TCFA [137]. The
study concluded
that neither modality alone is sufficient for detecting TCFA,
suggesting a combined
use of OCT and IVUS in the future.
A recent study examined the feasibility of wavelet analysis of
IVUS signals in
detecting lipid-laden plaques in vitro as well as in vivo [138].
RF signals from lipid
-
32
regions showed different pattern than fibrous regions on a
certain scale that
signified smaller wavelengths and thus higher resolution. Fatty
plaques could be
detected from the clinical samples with a sensitivity of 81% and
a specificity of 85%.
Limitation is that all the plaques analyzed had a thickness
>0.5 mm, and any lipid
core had a thickness >0.3 mm. Therefore, it is not known
whether it is possible to
analyze thinner plaques or to identify very thin lipid cores
with this method.
Although IVUS characterization of plaques has been very
promising, no one
has yet produced a technique with sufficient spatial and
parametric resolution to
identify a lipid pool with a thin cap.
1.3 Overview of limitations of Imaging Modalities
An overview of different diagnostic methods for detecting
vulnerable plaques
is shown in Figure 7. New methods may identify additional
characteristics of the
plaque enabling physicians to plan diverse treatments. Although
a multifocal disease
requiring systemic therapies, detecting vulnerable plaques may
still prevent MI and
strokes, reducing the effort and cost of managing a systemic
disease.
Limitations, requirements w.r.t. imaging vulnerable plaque and
image
resolution of different imaging modalities and the specific
tissue the modality best
identifies is given in Table 3. Each imaging technique has its
insufficiencies that
need to be resolved. From a clinical diagnosis perspective, a
combination of many of
these imaging modalities may be a requisite to identify a
vulnerable patient.
-
33
Figure 7: Various diagnostic methods for the detection of
vulnerable plaque
Table 3: Imaging capabilities of various modalities w.r.t.
vulnerable plaque
OCT IVUS MRI CTA Angiography
Spatial
Resolution
(m)
5-20 80-120 80-300 400-800 100-200
Probe Size
(m)140 700 N/A N/A N/A
Thin Cap Yes No No No No
Best suited
for
Thin caps of
atheromasFibroatheromas
Inflammation and
Characterization
Calcium
scoring
Lumen
variations
-
34
CHAPTER II
PROBLEM FORMULATION
The most serious mistakes are not being made as a result of
wrong answers.
The truly dangerous thing is asking the wrong questions
- Peter Drucker
In vivo identification of vulnerable plaque by imaging
techniques is essential
to prevent acute events. Several non-invasive and invasive
imaging techniques
discussed in previous chapter, Diagnosis, are currently under
development and
validation. None of these techniques can identify a vulnerable
plaque alone or
predict its further development. Of all the vascular imaging
modalities, the ability of
IVUS to directly image the vascular wall with high resolution
unlike angiography has
enabled its use in assisting physicians to detect plaques and
evaluate therapeutic
interventions [139, 140].The high sensitivity of IVUS in
detecting atherosclerosis
and quantifying plaques has been clinically accepted [141-144].
Miniaturization of
the IVUS transducers permits tomographic visualization of a
cross-sectional arterial
anatomy [145]. Although several studies have reported plaque
imaging abilities of
IVUS, Narula et al. identify that clinical identification of
culprit plaques has still not
-
35
been achieved [146]. DeMaria et al. state that none of these
methods are definitive
because the morphology descriptors are based on retrospective
studies and
vulnerable plaque characteristics vary across studies [147].
Also, non-culprit
plaques exhibit similar characteristics as culprit plaques
[148]. These shortfalls of
IVUS arise due to the imaging device limitations and lack of
appropriate tissue
characterization methods.
Foremost, the resolution of conventional transducers is not
adequate to
image the thin cap of the plaque, thickness ranging 2319 m [23].
Limitations of
conventional IVUS transducers based on PZT include narrow
bandwidth of 43%
(lower axial resolution, best around 62 m), inability to focus
(lower lateral
resolution, around 300 m) and the extended ring-down of the PZT
transducers
[149, 150]. Furthermore, clinically available systems rely on
the standard Fourier
transform for processing of the RF backscattered signals and
tissue characterization.
Better delineation of the plaque is possible by improved
transducer design and new
methods of analyzing RF backscatter signals.
In order to address the need for identifying vulnerable plaques,
the
combination of a high resolution focused polymer transducer and
the multi
resolution analysis of RF signals from tissue harmonic imaging
of the atherosclerotic
plaque was proposed.
A high resolution focused transducer fabricated using PVDF-TrFE,
termed
PMUT was developed in the BioMEMS laboratory at the Lerner
Research Institute
[151]. In comparison with the conventional transducers, the
focused PMUTs exhibit
-
36
broad bandwidth (~120% at -6dB). With the appropriate assembly
of the polymer
film, backing, and electrical and acoustical impedance matching,
Near-theoretical
axial resolution of ~19 m and diffraction limited lateral
resolution of 80 100 m,
were demonstrated [152]. The broad bandwidth of the transducer
facilitates
harmonic imaging. The polymer transducer allows focusing of
harmonic content to
within the narrow coronary geometry [152].
Tissue harmonic imaging, considered a recent breakthrough in
diagnostic
ultrasound, as important as Doppler, offers substantial
advantages such as
nonlinear information, improved lateral resolution, higher
contrast resolution, low
near field spatial variation and decreased side lobes [153-156].
These studies were
based on frequencies below 10 MHz. The results of the harmonic
imaging
experiments showed the feasibility of intravascular THI with a
conventional IVUS
catheter both in a phantom and in vivo rabbit aorta [157, 158].
The harmonic
acquisitions also showed the potential of THI to reduce image
artifacts compared to
fundamental imaging. The harmonic imaging of human coronary
arteries at 20 MHz,
30 MHz and 40 MHz using the pulse inversion technique was
reported by the
BioMEMS laboratory [159]. This study was limited to the
feasibility of pulse
inversion technique with PMUTs and further analysis of harmonic
signals for tissue
characterization was not suggested. The RF harmonic signals were
further analyzed
using multiresolution analysis (discussed in forthcoming
chapter) instead of Fourier
analysis of the signals for various reasons explained later on
and showed that each
frequency offers unique vessel wall information [160].
Consequently it was
hypothesized that there may have always been much anticipated
information about
-
37
lipids and thin caps in the fundamental and harmonic RF signals
and if processed
with the appropriate methods may result in better tissue
characterization, leading
to additional diagnostic information.
2.1 Specific Aims
The hypothesis is that multi resolution analysis of IVUS RF
signals from tissue
harmonic imaging of the vulnerable plaque with a focused
broadband polymer
transducer will result in additional diagnostic information.
This hypothesis is tested by undertaking the following specific
aims:
Aim 1
(a) Establish system linearity and fluid/lipid nonlinearity with
a PMUT.
(b) Multi resolution analysis of RF backscattered fundamental
and harmonic signals
thereby identifying plaque morphology, comp