mustafa karamanoglu. simulation measurement and analysis of the propagating pressure pulse in the...
DESCRIPTION
A few chapters of this PhD work into pressure pulse propagation in the human arterial system was published (see below). Other chapters which remain in manuscript form will be submitted shortly.Karamanoglu M, O'Rourke MF, Avolio AP and Kelly RP (1993). An analysis of the relationship between central aortic and peripheral upper limb pressure waves in man. European Heart J 14: 160-167Karamanoglu M, Gallagher DE, Avolio AP and O'Rourke MF (1994). The functional origin of reflected pressure waves in a multi-branched model of the human arterial system. Am. J. Physiol 267:H1681-1688. Karamanoglu M, Gallagher DE, Avolio AP and O'Rourke MF (1995). Pressure wave propagation in a multi-branched model of the human upper limb. Am. J. Physiol 269:H1363-1369.Karamanoglu M (1997). A system for analysis of arterial blood pressure waveforms in humans. Comput Biomed Res 30:244-255 Karamanoglu M (2002). A Reduced Model of the Human Arterial System. Proc. of IEEE-EMBS 3:1234-1235Karamanoglu M (2004). A Method for Estimation of Intensity of Wave Reflection in an Elastic Tube. Cardiovasc Eng 4 (3): 229-236TRANSCRIPT
i
omne ignotum pro magnifico
from Vesalius, 1545 from Arterial Model ,1992
i
Recent developments in non-invasive techniques of measurement of arterial pressure waveforms have increased the need for better understanding of the human arterial pressure pulse. This thesis examines the information carried by the propagating arterial pressure pulse in human systemic arteries, particularly in the peripheral upper limb arteries, in relation to properties and transmission characteristics of the arterial vasculature.
Simulations were performed in a mathematical model of the whole arterial tree where arterial segments are represented as uniform transmission lines and arranged as a binary tree structure. Experiments involved measurement and analysis of propagating pressure waves (a) in a single elastic tube (b) in aortic, carotid, brachial, radial, femoral and dorsalis pedis arteries of human subjects during: (i) altered heart rates, (ii) external compression of the vascular beds of the hand, (iii) vasodilatation with nitroglycerine and nitroprusside, (iv) ingestion of glucose.
Findings from model simulations and studies in human subjects are: (i) pressure waves in the central and peripheral upper limb arteries are different in the time domain and the relationship in the frequency domain (transfer function) is relatively constant; (ii) local reflections in the peripheral terminations modulate the initial oscillations seen in the peripheral arterial pressure pulse in early systole; (iii) reflected waves originating from the lower aortic trunk modulate the secondary oscillations seen in the arterial pressure pulse; (iv) under normal conditions wave reflection in the hand is high and relatively constant; (v) vasodilatation does not produce substantial reductions in reflected waves from the hand; (vi) ingestion of glucose significantly reduces the intensity of reflected waves from the vasculature of the lower trunk segment but does not alter reflections from the upper and lower limb terminations explaining the paradoxical disparity of wave shapes in central and peripheral upper limb arteries after vasodilatation.
The above findings resulted in the design and development of a computer based clinical tool to reconstruct the calibrated central aortic pressure waveform from pressure wave recordings in the radial artery. This application complements the current methods of non-invasive measurement of blood pressure by providing a comprehensive analysis of the arterial pressure waveform.
i
DECLARATION
I hereby declare that this submission is my own work and that, to the best of my
knowledge and belief, it contains no material previously published or written by another
person nor material which to a substantial extent has been accepted for the award of any
other degree or diploma of a university or other institute of higher learning, except
where due acknowledgment is made in the text.
Mustafa Karamanoglu
December, 1992
Sydney
i
SIMULATION, MEASUREMENT AND ANALYSIS OF THE PROPAGATING PRESSURE PULSE IN THE HUMAN
ARTERIAL SYSTEM
Mustafa Karamanoglu
(BE, MBiomedE)
A Thesis Presented for the Degree of
DOCTOR OF PHILOSOPHY
at the
University of New South Wales
December, 1992
Sydney
i
ACKNOWLEDGMENTS
I would like to thank my supervisor Dr. Albert Avolio for his patience in shaping this
work. During the entire investigation, he was at enough distance to let free thought
flourish, yet at times he was very close to scrutinise each and every statement that I
made. I will also remember Prof. Michael O'Rourke with his always youthful
enthusiasm to research. This work would never have started, let alone be finalised,
without his knowledge, his active participation and the research environment that he
created and maintained. I thank Dr. Raymond Kelly for his help in during my initial
acquaintance with the field. Thanks to my colleague and friend Dr. David Gallagher,
who over many years shared the joy and the stress of exploring the unknown. We
performed most of the experiments together. His calm nature and frankness will always
be with me. Sincere thanks to Dr. Gary Dobson who has contributed to theoretical
discussions. Special thanks to Miss Lina L. Lee, who spent many long hours in
recruiting the subjects and collecting the data presented in this thesis. Her contribution
to the analysis and presentation is most appreciated. I would also like to thank Ms
Georgina McPhee who typed the thesis and patiently dealt with the confusions that I
created. I also thank Mr. Jason Young on his help in correcting the manuscript. Finally
I would like to express my gratitude to many who helped me to develop the necessary
curiosity and persistence in realising this work.
My greatest thanks to my always understanding wife Sahure who recognised the
consequences of undertaking such a task.
ii
ABSTRACT
Recent developments in devices for non-invasive recording of arterial pressure contour such as arterial tonometry and photoplethysmography have highlighted the inadequacies of conventional methods of arterial pressure measurement and increased the need for better understanding of the human arterial pressure pulse. The arterial pressure waveform is a complex quantity containing much more information than the two singular points - systolic and diastolic pressures - provided by the widely used sphygmomanometric techniques. The arterial pulse contour undergoes marked alterations during its travel in the arterial system under normal conditions which is modified by aging, level of mean blood pressure and vasoactive state. This thesis examines the information carried by the propagating arterial pressure waveform in human systemic arteries in relation to vascular properties and transmission characteristics of the arterial tree. Since the arterial pulse is often measured in the upper limb, the thesis concentrates particularly on the relationship between the central aortic and the peripheral upper limb pulse.
Propagated pressure waves were simulated in a mathematical model of the whole arterial tree consisting of 142 uniform transmission line segments arranged in a binary tree structure. This model allowed detailed investigation of the effect of reflected waves originating from different regions in the arterial system on the central and peripheral pulse. Wave transmission in the upper limb during simulated changes in upper limb vascular properties were investigated and transmission parameters were estimated from experimental results. Experiments involved: (a) pulse propagation in a single elastic tube to test the model implementation (b) measurement of the aortic, carotid, brachial, radial, femoral and dorsalis pedis arterial waveforms and their relationship during: (i) altered heart rate in subjects instrumented with implantable pacemakers, (ii) alterations in regional wave reflection in the hand caused by graded compression of the palm, (iii) vasodilatation with nitroglycerine and nitroprusside, (iv) altered lower aortic trunk wave reflection at the region of splanchnic circulation caused by glucose ingestion.
Results from experimental investigations and model simulations have produced the following findings in the human arterial system: (i) pressure waves in the central and peripheral upper limb arteries are different in the time domain and the relationship in the frequency domain (transfer function) is relatively constant; (ii) reflected waves from upper and lower limbs are mainly confined within respective segments and they form
iii
the initial peak of the pressure waveforms in these segments; (iii) the vasoactive state of arterial beds of the trunk segment modulates global reflected waves which form the secondary oscillations seen in the central aortic and upper limb pulses; (iv) under normal conditions the intensity of reflected waves originating from the vasculature in the hand is high and close to that obtained by total occlusion of this bed. This finding explains the relative stability of the upper limb transfer function under these conditions; (v) Vasodilating agents such as nitroglycerine and sodium nitroprusside do not alter reflections from the palm vasculature significantly but decrease reflections from the trunk vasculature. This explains the paradoxical disparity of wave shapes in central and peripheral upper limb arteries after the administration of these drugs; (vi) ingestion of glucose causes significant reduction in the intensity of wave reflection from the vasculature of the lower trunk segment as evidenced by the arterial pressure wave contours and by analysis of transfer functions of the upper limb, trunk and lower limb segments. These changes are similar to those observed after administration of vasodilators.
The above findings resulted in the design and development of a computer based clinical tool where sphygmomanometric measurement of arterial pressure are augmented by recording of the radial artery waveform from which the calibrated central aortic pressure waveform is reconstructed. This application complements the current methods of non-invasive measurement of blood pressure by providing a comprehensive analysis of the arterial pressure waveform.
iv
TABLE OF CONTENTS
ACKNOWLEDGEMENTS i ABSTRACT ii TABLE OF CONTENTS iv LIST OF SYMBOLS x
INTRODUCTION 1
SECTION I BACKGROUND
CHAPTER 1 HISTORY 7
CHAPTER 2 THEORETICAL FOUNDATIONS 13
2.1 MATHEMATICAL MODELS OF THE ARTERIAL SYSTEM 14 2.1.1 Steady Flow 15 2.1.2 Oscillatory Flow 17
Windkessel Approach 17 One Element Windkessels 17 Modified Windkessels 19
Womersley's Approach 20 Discussion of the Validity of the Womersley Approach 27 Validation of Womersley's Approach 30
Other Approaches 31 Linear Models 31 Non-linear Models 31
(i) Method of Characteristics 31 (ii) Wavelet Analysis 36
2.2 ANALOGUES OF THE ARTERIAL SYSTEM 37 2.2.1 Lumped Analogues 38
Zero Order Analogues 38 First Order Analogues 38 Second Order Analogues 39 High Order Analogues 40
2.2.2 Uniform Transmission Line Analogues 41
v
2.2.3 Non-Uniform Transmission Line Analogues 44 2.2.4 Implications of Transmission Line Concepts 47
Transmission Direction 47 Impedance 47
Longitudinal Impedance 47 Transverse Impedance 48 Characteristic Impedance 49 Input Impedance 51
Propagation Coefficient 51 Transmission Ratio 52
2.2.5 Application of Transmission Line Concepts in Haemodynamics 53 Interpretation of Impedance 53 Interpretation of the Transfer Function 58
2.3 APPLICATION OF TUBULAR MODELS OF THE ARTERIAL SYSTEM 64
2.3.1 Single Tube Models 64 2.3.2 Two Tube Models 65
Tubes in Parallel 65 Tubes in Series 66
2.3.3 Three or More Tube Models 66 Random Branching Models 67 Anatomically Arranged Branches 67
Tree Structure 67 Mesh Structure 68
CHAPTER 3 MEASUREMENT OF ARTERIAL BLOOD PRESSURE 70
3.1 INVASIVE MEASUREMENTS 71 3.2 NON-INVASIVE MEASUREMENTS 73 3.2.1 Peak Systolic Value Measurement 74
Auscultatory Method 74 3.2.2 Contour Measurement 75
Sphygmography 75 Plethysmography 76 Applanation Tonometry 77
vi
SECTION II MODELLING AND SIMULATION
CHAPTER 4 MODEL CONSTRUCTION 83
4.1 REPRESENTATION OF THE ARTERIAL TREE 84 4.1.1 Parent Daughter Relations 85
Zero Order Relation 85 First Order Relation 88 Second Order Relation 88
4.1.2 Computational Algorithm 89 4.2 ANATOMICAL DATA 91
CHAPTER 5 A BINARY TREE MODEL OF THE HUMAN ARTERIAL SYSTEM 96
5.1 PRESSURE WAVE PROPAGATION IN THE HUMAN ARTERIAL MODEL 98
5.1.1 Pressure Contour Maps 98 5.1.2 Introduction of Reflected Waves 101 5.1.3 Tracking Of Reflected Waves 102 5.1.4 Wave Velocities and Reflection Coefficients 103 5.1.5 Input to the Model 104 5.1.6 Input Impedance 104 5.1.7 Pressure Wave Transmission 106 5.1.8 Reflections From Compartments 107 5.2. PRESSURE WAVE PROPAGATION IN THE UPPER LIMB 110 5.2.1 Effect of Change in Wall Elastance (E) 112 5.2.2 Effect of Change in Reflection Coefficient (Γ) 113 5.2.3 Effect of Changes in Time Constant (τ) 114 5.2.4 Effect of Change in Wall Viscosity (Θo) 116 5.3 DISCUSSION 117
vii
CHAPTER 6 MODEL VALIDATION AND PARAMETER ESTIMATION IN A SINGLE ELASTIC TUBE 125
6.1 SINGLE ELASTIC TUBE EXPERIMENT 126 6.1.1 Estimation of Tube Elastance 128
Calculation of Elastance from Stress-Strain Relation by Longitudinal Stretch 128
Calculation of Elastance from Pressure-Volume Relations 129 Calculation of Elastance from Wave Velocity: 129
6.2 ESTIMATION OF MODEL PARAMETERS 130 6.3 RESULTS 132 6.3.1 Tube Elastance 132 6.3.2 Convergence of the Model 133 6.3.3 Predicted Pressure Waves 134 6.3.4 Transfer Function 136
Modulus 136 Phase Velocity 139
6.3.5 Reflection Coefficient 139 6.4 DISCUSSION 140
SECTION III EXPERIMENTAL STUDIES
CHAPTER 7 DETERMINATION OF WAVE PROPAGATION CHARACTERISTICS IN THE HUMAN UPPER LIMB 145
7.1 METHODS 147 7.1.1 Subjects 147 7.1.2 Measurements 147
Calibration of Waves 148 7.1.3 Data Analysis 148
Pressure Wave Amplification 148 Pressure Wave Shapes 148 Transfer Functions 149 Parameters Obtained from the Transfer Function 149
Resonant frequency 150 Reflection coefficient 150
Reduced Model Of the Upper limb Arterial System 151 7.1.4 Statistics 152 7.2 RESULTS 152 7.3 DISCUSSION 158
viii
CHAPTER 8 QUANTIFICATION OF THE INTENSITY OF REFLECTED WAVES IN THE UPPER LIMB 161
8.1 METHODS 162 8.1.1 Subjects 162 8.1.2 Procedure and Data Acquisition 163 8.1.3 Estimation of Intensity of Reflected Waves 166
(i) Estimation from Phase Velocities 167 (ii) Estimation from Pulse Pressures 167 (iii) Estimation from the Model 167
8.1.4 Statistics 168 8.2 RESULTS 168 8.3 DISCUSSION 177
CHAPTER 9 EFFECTS OF VASODILATORS ON UPPER LIMB WAVE PROPAGATION PROPERTIES 185
9.1 METHODS 186 9.1.1 Nitroglycerine (NTG) 186 9.1.2 Sodium Nitroprusside (SNP) 188 9.1.3 Statistics 188 9.2 RESULTS 189 9.2.1 Nitroglycerine 189 9.2.1 Sodium Nitroprusside 191 9.3 DISCUSSION 194
CHAPTER 10 EFFECT OF THE SPLANCHNIC CIRCULATION ON THE FORMATION OF THE ARTERIAL PULSE 196
10.1 METHODS 197 10.1.1 Subjects 197 10.1.2 Protocol 198 10.1.3 Measurements and Data Analysis 198 10.1.4 Statistics 200 10.2 RESULTS 200 10.3 DISCUSSION 205 10.3.1 Reduced Model of the Arterial System 209
ix
SECTION IV APPLICATIONS
CHAPTER 11 THE SYNTHESIS OF THE CENTRAL PRESSURE WAVEFORM FROM THE PERIPHERAL PULSE IN THE UPPER LIMB 217
11.1 METHODS 219 11.1.1 Synthesis in the Frequency Domain 219 11.1.2 Synthesis in the Time Domain 221 11.2 RESULTS 222 11.2.1 Frequency Domain 222 11.2.2 Time Domain 226 11.3 DISCUSSION 228
CHAPTER 12 A SYSTEM FOR ON-LINE ANALYSIS OF BLOOD PRESSURE WAVEFORMS 233
12.1 IMPLEMENTATION 235 12.1.1 Data Acquisition 235 12.1.2 Signal Conditioning 237 12.1.3 Feature extraction 238 12.1.4 Calculated Parameters 241 12.1.5 Database Engine 244 12.2 THE TRIAL OF THE SYSTEM IN A CLINICAL ENVIRONMENT 247
SECTION V SUMMARY AND CONCLUSIONS
REFERENCES 256 APPENDIX I 283
x
LIST OF SYMBOLS
<, j Phase angle
α Womersley's non-dimensional number
C Capacitance
c, co Wave velocity, Inviscid wave velocity
D Diameter
E, Ec Elastance, Complex Elastance
ε Strain
F10 the expression ( )
( )230
23
2312
jJjjJαα
α
f Frequency
Γ Reflection coefficient
γ Propagation constant
H Weighted volume of the wall substance taking into account external inertial loading of pe
H(ω) Frequency (ω) dependent transfer function
h(t) Impulse response function of H(ω)
h Vessel wall thickness
η Viscosity coefficient of the vessel wall
I Current
J0, J1 Zeroth and first order Bessel functions of the first kind with complex arguments
K Spring coefficient of the tethered tissue
L Inductance
r, l Radial and axial distance
M Mass of the vessel wall element
M Modulus of complex quantity
µ Viscosity of the blood
N Vessel's tethering coefficient
xi
P Pressure
Q Flow, Charge
θ Phase angle of Ec
R Internal radius
R Resistance
ρ, ρ0 Vessel wall and blood density
r, z, θ Radial, axial and tangential coordinates in cylindrical axes
S Vessel wall area
σ, σc Poisson's ratio, Complex Poisson's ratio
τ Time constant of the Windkessel
t Time
u, v, w Radial, tangential and axial component of the instantaneous flow velocity
V Voltage, Volume
ω, ωn Angular, Resonant frequency
ξ, ζ Radial and Axial displacement of the vessel wall
Z Impedance
Z0, Zin Characteristic and Input Impedance
Zl, Zr, ZT Longitudinal, Transverse and Terminal Impedance
Ψ Efflux of fluid through the walls per unit length
1
INTRODUCTION
Although pulsatile haemodynamics forms a relatively small branch of cardiovascular
physiology, investigators from many different disciplines have confronted the many
challenges and complexities of the field. From disciplines of mathematics, physics and
engineering investigators studied the principles of the field of haemodynamics, with the
aim to gain basic understanding leading to improved clinical diagnosis and treatment of
cardiovascular disease. Although many important discoveries were made, clinical
applications resulting from basic investigations in pulsatile haemodynamics have been
scarce. Practicing clinicians have found them either too complex to understand or too
difficult to apply. Inevitably, it has proved difficult for experimental and theoretical
concepts to emerge from the confines of the laboratories.
Recent advances in technology and results of large scale epidemiological studies are
producing considerable changes in many traditional and largely conservative attitudes.
It is now possible to obtain important basic quantities non-invasively by echo doppler
imaging (Nichols and O'Rourke 1990) and tonometry (Kelly et al 1989b). Due to the
developments made in computer technology it is also possible to perform complex
calculations in real time. Recent studies also highlighted the importance of the arterial
pulse wave contour in cardiovascular medicine and increased the awareness of the effect
of pulsatile load on the heart and on the arteries. Systolic pressure is now considered a
major risk factor for cardiovascular disease (Kannel et al 1981) and the shape of the
central aortic pulse is an important indicator of left ventricular-vascular coupling
(Takazawa 1987). A current trend in clinical cardiology is to develop techniques to
measure and establish background work to interpret the arterial pressure pulse.
2
Not surprisingly, the pulse in the human arterial system has been investigated
extensively in the past. In these early studies, specific attention was directed towards
the pressure wave in the central aorta and the lower limbs while the pressure was often
recorded in the upper limb. The pressure pulse contour in the upper limb is not only
considerably different from the central pulse but also undergoes different directional
changes under certain circumstances. Compared to the central pulse, the pulse in the
upper limb is affected differently by aging, vasodilatation and physical manoeuvres
leading to difficulties in interpretation. Although much has been published in
documenting the morphology of the upper limb pulses under these circumstances (Kelly
et al 1989a, Simkus and Fitchett 1990), there has been little systematic investigation on
the characteristics of pressure wave propagation in the upper limb.
This thesis aims to extend the understanding of what arterial parameters are relevant in
the formation of arterial pressure pulse. It attempts to understand the information
content of propagating arterial pressure in human systemic arteries by examining the
vascular properties and transmission characteristics of the arterial tree. It also aims to
characterise the properties of the arterial system by extraction of specific relevant
features from the pressure waveform obtained from different arteries. It particularly
concentrates on the pressure pulses in the central aorta and in the peripheral upper limb
arteries.
In Section I, the background work to the broad field of haemodynamics and wave
propagation phenomena will be discussed. The historical account of events leading to
present knowledge is followed by a theoretical survey. In this section, various concepts
of pulsatile haemodynamics will be discussed. Since these concepts will form the
foundations of the modelling section, the underlying assumptions will be studied in
detail. In the remainder of this section the current methods on measuring the pressure
waveform will be reviewed.
3
Section II is devoted to the simulation of pressure wave propagation by computer
implementation of theoretical modelling concepts mentioned in Section I.
The first chapter of this section, Chapter 4, describes the construction of a mathematical
model of the pressure/flow relations in a viscoelastic tube. Using methods originally
developed by Taylor (1966a, b) and Avolio (1976) a binary tree model of the arterial
system is constructed which finds extensive use in Section III. Experiments performed
in Section III are simulated by the model to estimate the properties of the arterial
vasculature.
Chapter 5 describes the application of the binary tree model to the human arterial system
where a new approach in exploring pressure wave propagation is developed. The
pressure contour maps developed in this chapter provide new insights into the origins of
reflected waves. The important physical parameters in the upper limb wave propagation
are also described in this chapter. The sensitivity of transfer functions of the upper limb
model on propagation parameters are explored by altering physical quantities.
The third of these chapters, Chapter 6, is devoted to the verification of the model and the
development of parameter estimation techniques based on this model. Using a bench
top experiment, the theory and the integrity of model algorithm are tested.
Section III presents chapters describing the clinical studies. In Chapter 7 the presence of
upper limb transfer function and its clinical implications were investigated by
conducting experiments on eighteen human subjects. Results of this study confirmed
predictions made by the model.
Chapter 8 describes experiments directed towards investigating the effect of reflected
waves on the formation of pressure waves in the upper limb. The effect of reflected
waves originating from the palm were investigated by graded compression of the hand
4
in thirty three human subjects. It was found that the reflection indices correlate well
with the systolic pressure and could even be reduced by increasing the compliance of the
peripheral hand vessels.
Chapter 9 describes findings of the analysis of changes in central and upper limb pulses
in response to vasodilatation as reported by other investigators using techniques
developed in earlier chapters. It was found that the effect of vasodilator agents on the
propagation properties of the upper limb vasculature is negligible and does not explain
the reported differences.
The outcome of experiments in Chapter 9 and model predictions in Chapter 5 suggested
studies to explore vasodilatation of the splanchnic circulation. In Chapter 10, the effects
of glucose ingestion on the pressure waveforms along the arterial tree is explored in ten
human subjects. Results explained the reported discrepancies of pressure waveforms in
different arteries in response to vasodilators. These findings were confirmed by results
from the arterial model.
The general findings described above established that the propagation properties of the
upper limb arterial system are relatively stable and do not account for observed changes
in a wide variety of conditions. Section IV investigates the possibility of synthesis of
the central pressure wave from the peripheral pulse in the upper limb using the finding
of a relatively constant transfer function. In Chapter 11, methods to synthesise central
pressure waveforms from peripheral recordings are described. These methods are then
tested using published records of central and peripheral pulses. Chapter 12 describes a
practical application resulting from this investigation. In this chapter, an on-line system
is described to synthesise the central aortic pulse from peripheral pulses in the upper
limb. It also describes the inner workings of the system together with algorithms
developed to extract features and analyse the pulses from different recording sites in the
5
upper limb. The system described in this chapter is designed and implemented as a
clinical tool to be used in the clinical environment.
This investigation has highlighted the importance of quantification of wave propagation
characteristics in the arterial system especially in the upper limb vasculature. Systems
that measure the various features of the arterial pulse do not take wave propagation into
account. Information obtained using these techniques can be limiting in both content
and interpretation. Results of the present study suggest means of overcoming these
inadequacies leading to improved clinical applications through measurement and
analysis of the arterial pressure waveform.
6
SECTION I
BACKGROUND
7
HISTORY The beating of the heart, the presence of the arterial pulse and the movement of the chest
are the three vital signs that have been traditionally associated with the presence of life.
A large part of the history of medicine covers the attempts to reconcile these signals of
life and volumes of monographs summarise this historical treatment (McDonald 1960,
Fishman and Richards 1964; Schwartz et al 1981; Milnor 1989; Nichols and O'Rourke
1990; O'Rourke et al 1992). In this chapter, the arterial pressure wave and its
distribution along the arterial tree will be addressed with respect to historical
development.
Ancient Greeks had an inaccurate notion of the arterial pressure pulse. They believed
that the arteries (air ducts) actively contracted and dilated, like bellows, to pump vital
spirits to and from the heart. Prominent amongst these thinkers was Praxagoras who
Chapter 1
8
lived in 400 BC (Harris 1981). The first opposing view came in 300 BC by Erasistratus
of Cos, who on pure mechanical grounds, argued the role of the left ventricle and
proposed the propagation of the wave from the heart (Schechter et al 1969). This point
was ignored by Galen (129-199 AD) who by his works, influenced medicine for the next
fourteen centuries. The only improvement of the Galenic theory was to allow blood to
run in the arteries rather than just the spirits. Jean Furnel (1497-1558) and Realdo
Columbus (1510-1559) both argued that the pulse travelled away from the heart
(Webster 1981). This was taken up by William Harvey (1578-1637) in his influential
work "De motu cordis" in 1628. Harvey, regarded as the father of modern
cardiovascular medicine for his work in establishing the modern views on circulation,
failed to point out the propagation delay in wave transmission (Franklin 1979). John
Floyer (1649-1734) was first to demonstrate the effect of the elastic wall on removing
the intermittency of the pump on his cow gut experiments. Twenty-five years earlier
than Hales, he quoted "The force of water injected (by a pump) protruded the gut, and
the annular fibres by their natural restitution promoted the motion of the water, and
kept the stream from any interruption though the injection was made by intervals"
(Schechter et al 1969; Taylor 1991). Stephen Hales (1677-1761) best known for
measuring blood pressure in a horse, arrived at the same conclusion through a different
route. He reasoned that although there were pulsations in the arteries, none was
observable in the veins. In his acclaimed book "Haemostatics' (Hales 1773) he also
mentioned the sites of resistance on the arterial system, undoubtedly capillaries as
described by Malpighi in 1620 and the estimates of cardiac output by post mortem
measurements of chambers of the heart. This book was translated into German and
formed the important foundation for German physiologists in the late 19th and early
20th century. Cardiac output was measured in vivo by the German physiologist Fick in
1870, a hundred years later, while the amount of resistance was measured by J.P.
Poiseuille (1797-1869) by assuming a laminar viscous flow. Poiseuille was the first to
introduce mercury as a fluid to measure the height of the blood pressure column. This
mercury manometer was improved upon by Carl Ludwig in 1842 whose pupils
9
dominated physiology throughout the latter part of the 19th century and middle of 20th
century (Milnor 1989).
The introduction of calculus to physics by Isaac Newton led to the proliferation of
scientists applying this tool to different natural phenomena. Leonhard Euler (1707-
1783), not only contributed to expansion of techniques of calculus, but also was the first
to develop the equation of motion in liquids. His theory was hailed as the complete
theory of fluid flow by others but did not include the viscosity effect "defectus lubricus"
as first mentioned by Newton. These equations remained insoluble until 1958, when
Lambert tackled it with the method of characteristics (Skalak 1972). The effect of
viscosity is taken into account by Navier in 1822. This formulation was later extended
and corrected by Stokes to be known as the Navier-Stokes equation (Milnor 1989).
The arterial wall properties were first investigated by Spallanzani in 1773, who fitted
rings of fixed diameters around the aorta to detect dilation of the vessel by pressure
waves. He was the first to show that the pressure wave is propagated through the wall
rather than the blood (Fishman and Richards 1964). The same idea was also taken up by
Thomas Young (1773-1829) (also known for the theory of colour-vision) when he
investigated the propagation speed of pressure waves in an artery. Young (1808)
concluded the ongoing debate of contraction of arterial wall muscle and the formation of
the pressure pulse. He examined the speed of a free falling body in a compressible tube-
fluid combination (Noordergraaf 1969). The resulting equations related the pressure-
diameter relationship (elasticity) to the wave speed - later rediscovered by Bramwell-
Hill and Otto Frank (McDonald 1974). Wave equations and their implications became
very topical in the early 19th Century, thus when the Weber brothers published the
monograph "Wellenlehre" in 1825 they laid the fundamental principles of wave
propagation and wave reflection. The later work done independently by Moens in 1878
and its analysis by Korteweg and Resall (McDonald 1974) led to the well-known wave
speed equation.
10
While these advances in wave travel were being made, the Chinese views of the pulse
were spreading (Osler 1921). These views which first arrived in Europe in the 17th
Century stipulated that the disturbances in the function of any organ could be detected
by changes in the pulse waves in the radial artery (Fung 1984). The detection of the
pulse from superficial arteries (specifically radial and carotid) and interpretation of the
pulse reached its peak in the 19th Century with the continuous addition of numerous
versions of sphygmographs (sphygmo - Greek: beat, throb) to the arsenal of the
practitioner who only possessed the pulse watch, thermometer and stethoscope at the
time (O'Rourke et al 1992). Etienne Marey (1830-1904) advanced the kymography
technique introduced by Ludwig by incorporating it into instruments of pressure
measurement (Snellen 1980). He also developed the first intra-cardiac pressure
measuring catheter (Geddes 1970) which he used in a horse.
In the late 19th Century, the tide of research was turning more towards cardiac
properties. Mackenzie in the preface to his monograph 'The Study of the Pulse' stated
that "It is now fashion to decry the value of the sphygmograph" (Mackenzie 1902). It is
therefore not surprising that the works done by Mahomed (1874, 1892), Broadbent
(1890) and others were forgotten for almost one hundred years (O'Rourke et al 1992).
Instead blood pressure began to be assessed by non-invasive occlusion methods under
clinical conditions. Works of Riva-Rocci (1896) and Korotkoff (Korotkoff 1905;
Geddes 1966) laid the foundations of the sphygmomanometer. Although the
consequences of wave propagation were not properly addressed, this technique gained
wide acceptance and still dominates clinical practice (O'Rourke et al 1992).
Since the study of the heart involves invasive techniques, researchers needed better
equipment to measure pressures inside the cavities. The development of fluid-filled
manometers by Rollesten and Hurtle (Geddes 1970) and many others attempted to fulfil
this demand. However, there were problems in the fidelity of the manometers and there
was no standardisation. Otto Frank (1865-1944), the last pupil of Ludwig, called for
11
physical principles to be followed in the design of these manometers. He solved the
differential equation of motion of a single degree of freedom system (Frank 1903). The
solutions are quite easy to follow and still in use today (Nichols and O'Rourke 1990).
However, preoccupied with manometer theory he went on to apply the same equations
to the arterial system while borrowing from Hales' concept of an elastic reservoir (Frank
1905). The "Windkessel" theory, as we know it today, views the arterial system as an
overdamped catheter to explain the aortic pressure waveform. It is important to note
that although he must have been aware of the wave equations and transmission lines (the
transatlantic telegraph cable was already laid down in 1866), he failed to employ them
in the basic equations. Nevertheless, Frank and his followers such as Wiggers (1928),
Hamilton (1944), Wetterer (1954) and Remington and Wood (1956) dominated the area
of haemodynamics for the next fifty years (O'Rourke 1965).
It was Aperia (1940), one of the followers of Frank, who pointed out the inconsistencies
in the entire assumptions made by Frank. The Windkessel theory could not explain the
different wave shapes in different arteries without resorting to the wave equations.
Aperia tried to reconcile these concerns but his premature death prevented him doing so
(McDonald 1974).
Although the first complete set of analyses of wave propagation was made by Witzig in
his PhD thesis in 1914, this was not widely circulated and thus remained obscure
(Milnor 1989). The treatise of transmission equations was starting to have an impact on
physiology when McDonald and Womersley collaborated in 1952 to publish their
elaborate one on arterial fluid dynamics. Almost simultaneously with Hodgkin and
Huxley, who used telegraph equations for description of action potential propagation in
nerve cells (Hodgkin and Huxley 1952), Womersley linearised and solved the Navier-
Stokes equations with well-documented assumptions (Womersley 1955a). The
mathematics of this theory required heavy computation that had hindered earlier
attempts. Having access to computer facilities that were scarce at the time, helped him
12
to generate tables necessary for the theory (Dinnar 1981). When the question of
reflected waves appeared in 1957 (Taylor 1957a) the solutions he proposed were easily
modified to include these effects (Womersley 1958). The results of this theory and its
predictions were published in textbook form by DA. McDonald in 1960. The book
'Blood Flow in Arteries' became the standard reference in the field of arterial
haemodynamics (McDonald 1960).
The success of this linearised theory and its applications dominated research in
haemodynamics in the 1960s and 1970s. The arterial impedance (Patel et al 1963,
O'Rourke and Taylor 1966, 1967a) and transmission of the pulse (Taylor 1966b), led to
a widespread continuation of research into wall properties (Learoyd and Taylor 1966;
Gow and Taylor 1968, Gow 1972) and wave reflection phenomena (O'Rourke 1967b,
Westerhof et al 1972). During this period several models of the arterial system were
proposed (de Pater and Van den Berg 1963; Taylor 1966a, 1966b; Westerhof 1969;
O'Rourke 1967b; Anliker et al 1978; Avolio 1980). By the 1980's, the influence of
reflected waves on arterial waveforms (Murgo et al 1980a, 1980b; Fuji et al 1987, Kelly
et al 1989a) and on aortic impedance (O'Rourke 1982; O'Rourke et al 1987) on left
ventricular function (Laskey et al 1985) was investigated. Arterial haemodynamics has
never experienced such consistent and fruitful collaboration between theory and
practice.
13
THEORETICAL FOUNDATIONS The conclusion drawn from the recorded pressure waveform depends on the observer's
conception of the arterial vasculature, that is, it will depend on the assumed circulatory
"model".
The "model" is as vague a term as the phenomenon it tries to describe. There appears to
be little agreement upon the definition, its use and its interpretation. It was previously
referred to as "something simple by scientists to help them understand something
complicated" (Fitzhugh 1969), or "an effective visualisation of a functional relation"
(Kenner 1978), or as "a set of mathematical relations between relevant quantities" (Arts
1978). In this respect there are different ways to approach modelling. One can define
important parameters for the system under investigation and observe only these
parameters to construct a set of rules. Alternatively, one can define sets of rules of
Chapter 2
14
importance and eliminate the rules according to the subsequent observations. A final
approach would be to permute all the available rules and all the available sets of
observations and determine the ones that fulfil a pre-defined optimum criterion. This
optimum criterion clearly dictates the rules and observations to be made, yet it can not
change the experimental data. It is equivalent to saying that interpretation of data in
different ways will not create more data points. Thus, a model is a tool, an abstraction
technique, to transform real world observations to a known domain of few rules, to
interpolate and/or extrapolate the real world in an abstract predefined optimum. The
model, therefore, is not meant to replace the real system. Rather it is to gain better
understanding of the real system given certain constraints.
2.1 MATHEMATICAL MODELS OF THE ARTERIAL SYSTEM
Previous investigators attempted to model the arterial system to understand the complex
interaction between the heart and the arterial tree (Yin 1987a), to understand the
interaction between the flow and pressure waves at each point in the tree (Avolio 1980)
and to explain differences in the pressure and flow shapes due to altered vascular
properties (Westerhof and Noordergraaf 1970). To assert the existence of a steady state
condition, the nervous control of the heart and the vasculature is either ignored or is
assumed negligible in most attempts. In these models, it has always been acknowledged
that the heart is a force generator in the circulation that keeps blood in motion. This
motion of blood ensures a steady concentration gradient of nutrients and metabolic by
products to exist at the cellular level. In gross terms, the general model of the
circulation includes a power generator, whose energy is both taken up by the vessel
walls as elastic energy and dissipated by the viscous forces in the wall and in the blood
during circulation. Thus it is not surprising that the first attempts to model the
15
circulation concentrated on this dissipative force as a first order approximation to the
circulation.
2.1.1 Steady Flow
Probably the first observation leading to this model was done by Hales. Using a time
clock and the gut of an animal he observed a reduction of flow velocity in the region
close to the capillaries, accounting for the presence of resistance at around the capillary
level (Schechter et al 1969). This work was taken up by Poiseuille who could not
document any difference in blood pressure along the large arteries but between the veins
and the arteries indicating the small arteries beyond 2 millimetres. When he studied the
water flow in small tubes of 30µm to 140µm internal diameter, he noted the following
relation (Milnor 1989).
( )
lPPKD
Q 214 −
= (2.1)
Where Q, P1-P2 and l are the volume flow, pressure drop and length of the tube
respectively. D is the tube internal diameter, and K is a constant. He found that K is
independent of everything else but temperature. Subsequent developments indicated
that K was related to viscosity. Independent theoretical studies by Wieden in 1856 and
Hagenbach in 1860, found (Nichols and O'Rourke 1990) that
K =
πµ128 (2.2)
Consider an axisymmetric flow of an incompressible fluid of viscosity µ across a
pressure gradient (P1-P2)/l in a tube of radius R and length l. Then, the expression for
velocity may be obtained using the Navier-Stokes equations in cylindrical coordinates
16
∂∂
∂∂
∂µ∂
∂µ∂
2
2w
rw
r rPz
r wt
+ + = (2.3)
Since ∂w/∂t=0 the equation reduces to
∂∂
∂∂
∂µ∂
2
2w
rw
r rPz
+ = − (2.4)
or
z
Pdr
rdwrdrd
µ∂∂
−=
(2.5)
Integrating twice yields
( ) 212 CrLnCr
zPw ++−=
µ∂∂ (2.6)
w should be finite at the axis and at the radius R. This yields C1 0= , and
C Pz
R22
4= −
∂µ∂
. Substituting for C1 and C2
( )22
4Rr
zPw −−=
µ∂∂
(2.7)
Volume flow then becomes
( )µ
π∂∂π
82
4
0
RzPdrrwQ
R
−== ∫ (2.8)
substituting ( )lPP 21 − for − ∂
∂Pz
yields equation 2.1
17
The assumptions and validity of these assumptions in the derivation of this formulation
are quite important and discussed elsewhere (Caro et al 1978; Milnor 1989; Nichols and
O'Rourke 1990). The term 8µl/πR4 can be thought of as the resistance term to bolus
flow by analogy with electrical circuits where the pressure drop along the tube is the
voltage and the input bolus flow becomes the current. Analogous to electrical networks,
any arterial bed in the systemic circulation could be modelled as a parallel and series
combination of resistances. Since the element lengths are fixed for anatomical reasons,
the resistance of the entire network becomes proportional to the diameter of individual
elements. The most useful and probably the most used rule of resistance is the total
peripheral resistance (TPR)derived as
TPR MeanArterial essure MeanVenous essureCardiacOutput
=−Pr Pr
(2.9)
2.1.2 Oscillatory Flow
Windkessel Approach
Apart from the slowing down of flow, Hales also observed the diminishing of pulsations
from the arterial to the venous side. Thus he likened the arterial system to a
contemporary fire engine with reservoirs, which operated like elastic chambers. These
ideas were later taken up by German physiologists, particularly by Otto Frank, from the
German translation of the book 'Haemostatics' by Hales. In German translation the
elastic reservoir was called Windkessel or "air kettle" and was attached to this
description by Otto Frank.
One Element Windkessels
According to Frank, the left ventricular ejection would be taken up during systole by the
arterial wall, which is to be passed to the blood to drive it against a resistance, namely
18
the peripheral resistance (Nichols and O'Rourke 1990; Milnor 1989; Skalak 1972).
Although Frank was aware of concepts of wave phenomena (Frank 1899) which was
evident from his work on manometers (Frank 1903), he totally ignored the forces
associated with blood inertia and blood viscosity as well as wall viscosity. Thus the
resultant differential equation contained only two terms. Assuming an elastic chamber
where rate of pressure change dP is related to volume change dV as
( )PfdVdP
= (2.10)
and a discharge from the resistance element, R, of the form
dVdt
PR
= − (2.11)
then the equation 2.10 and 2.11 lead to
( ) Rdt
PdP
Pf−=
1 (2.12)
In special case where f(P) is a constant of the form f(P)=1/C then
C dPP
dtR
= − (2.13)
integration and assuming P=Po at t=0 yields
P P e t RC= −0 (2.14)
This equation and its solution is basically an RC network in electrical engineering.
Overall, it is an improvement on a one element (resistive) model of the circulation. It
attempts to describe the systolic, diastolic and pulsatile pressures and the shape of the
19
central aortic pressure waveform. The model found great enthusiasm amongst the
German school (Wetterer 1954) and in the USA (Wiggers 1928; Remington et al 1948;
Remington and Wood 1956) and dominated the physiological research until the 1950s.
The major drawback of the model is that it does not allow for wave transmission since
the chamber is filled instantaneously, and by implication there is no wave reflection. It
also neglects the inertial forces. Noticing these flaws Aperia (1940) tried to reconcile
wave transmission with the elastic chamber theory.
Remington tried to explain the differences in peripheral arterial wave shapes due to
resonances and thus incorporated second order systems (Remington and O'Brien 1970;
Remington 1974). The system is described by two parameters, R and C, where one
should be calculated independently since the other could be determined from the
diastolic decay of the arterial pulse. This led to a major effort in calculating the
effective Windkessel capacitance. Although the model did not consider the propagation
of waves, there were attempts to relate wave velocity to the capacitance, C (O'Rourke et
al 1968). The model also had problems describing the pressure flow relations during
early part of the systole. Although it implied that the pressure rise was proportional to
the inflow, the experimental results suggested that pressure itself was proportional to
inflow. Furthermore, it not only directly related the diastolic pressure and the peripheral
resistance but also systolic pressure to the contractility of the heart and to the wall
stiffness (O'Rourke et al 1992). The model, although under close scrutiny, finds use in
predicting stroke volume from pressure waveform (Skalak 1978). Due to shortcomings
of two-element Windkessels, three and more element Windkessels were devised.
Modified Windkessels
Given the inherent limitations of the original Windkessel theory, it is often modified to
take into account blood inertia and the transmission delay in filling the chamber, leading
to incorporation of inductance and series resistance to the original RC parallel network
20
(Westerhof et al 1977; Milnor 1989). The influence of non-linear properties of the
arterial wall was also incorporated into a model by assuming non-linear pressure-
volume relations in quadratic or higher forms (Cope 1965). The assumption of
combined parabolic and linear forms yielded deviations of 5-10% of the actual values,
when stroke volume was estimated from pressure waveform (Cope 1965). This led to
investigation by McDonald (1974) of the estimation of stroke volume by pulse contour
methods. These Windkessel equations appear to give satisfactory results when the
effect of wave travel becomes small due to increased wave speed.
Versions of modified Windkessels and especially second order systems have been
modelled (Guier 1981). They are also used to explain the differences in wave contours
in different arteries (Wiggers 1928; Warner 1957; Remington 1974). These models all
suffer from infinite wave speed and lack of wave reflection (McDonald 1974).
Because modified Windkessels are easy to implement as hydraulic analogues, they enjoy
widespread use in testing cardiac valves, intra-aortic balloon pump (Niederer and Schilt
1988) and modelling of the function of the heart (Sunawaga et al 1983). However, they
ignore the effects of wave reflection therefore their use is limited. Modified
Windkessels are also used to lump arterial beds with few model parameters in modelling
studies of the systemic circulation (Liu et al 1989, Burattini et al 1989b).
Womersley's Approach
Previous models described above take into account neither the nature of blood flow in
the arteries, (although "the most prominent feature of blood flow in arteries is that it is
pulsatile" (McDonald 1960)), nor the effect of wave propagation. Amongst the many
previous attempts (Witzig 1914; Crandal 1927; Iberal 1950; Morgan and Kiely 1984),
Womersley's treatise of the subject is particularly significant (Womersley 1955a, 1955b;
1957a, 1957b, 1958). The model he described is the most realistic compromise due to
21
close collaboration with DA McDonald and MG Taylor who supplied the necessary
information about the validity of the assumption. In his work, he documented every
assumption and the final work was simplified for immediate testing. Thus, the model he
devised is still the best recognised, although there were many other attempts to
supersede it (Noordergraaf 1968; Cox 1969).
Womersley assumed a homogenous, Newtonian, incompressible, viscous fluid in a long
and straight cylindrical tube with thin, isotropic walls and longitudinally tethering while
externally constrained. The flow is considered to be laminar and axisymmetric with no
slip at the wall. The pressure gradient is assumed to be oscillatory.
The equations of motion in cylindrical coordinates for the type of fluid given above in
Navier-Stokes form are
++−=− 2
2
2
2
0 zw
rrw
rw
tw
zP
∂∂
∂∂
∂∂µ
∂∂ρ
∂∂ (2.15)
−++−=− 22
2
2
2
0 ru
zu
rru
ru
tu
zP
∂∂
∂∂
∂∂µ
∂∂ρ
∂∂ (2.16)
The tangential velocity component v is omitted. The convective non-linear terms
u∂w/∂r, w∂w/∂z, u∂u/∂r, and w∂u/∂z are assumed to be negligible and also dropped
from the equations. Continuity equation gives
∂∂
∂∂
ur
ur
wz
+ + = 0 (2.17)
After non-dimensionalising r by setting y r R= −1 and assuming a travelling pressure
wave of the form
22
( )cztjeAP −= ω1 (2.18)
where ω and c are the angular frequency and wave velocity respectively. The solutions
of equations 2.15 and 2.16 using 2.17 and 2.18 become
( )( )
( )cztjec
AjJ
yjJCw −
+= ω
ραα
0
123
0
230
1 (2.19)
( )( )
( )cztjec
yAjJjyjJC
crju −
+= ω
ραααω
0
123
023
231
12
2 (2.20)
where α is the non-dimensional parameter giving the ratio of inertial to viscous forces as
µωρα 022 R= (2.21)
J0 and J1 are zero and first order Bessel functions with complex arguments. C1 is
integration constant. If one defines
( )( ) 101023
023
231
102 ε
ααα M
jJjjJF == (2.22)
where M10 is the modulus and ε10 is the phase of F10, then the fluid motion at the
boundary becomes
( )cztjy e
cACw −
=
+= ω
ρ0
111 (2.23)
( )cztjy e
cyACF
cRju −
=
+= ω
ρω
0
11101 2
(2.24)
23
The equation of the wall motion is obtained by assuming
( )EjEEc ∆+= ω1 (2.25)
( )σωσσ ∆+= jc 1 (2.26)
for complex elastic modulus, Ec ,and complex Poisson ratio,σc , of the wall. Defining ρ
as the wall density, ξ and ζ as the radial and axial displacements of the wall which are
travelling sinusoidal waves as
( )cztjeD −= ωξ 1 (2.27)
( )cztjeE −= ωζ 1 (2.28)
The descriptive equations for the radial wall motion become
( ) 01 222
2
=
+−
+−RzRH
hEH
Pt
c
c
c ξ∂∂ζσ
σρρ∂ξ∂ (2.29)
where the first term is the inertial, the second is the transmural pressure and the third is
the viscoelastic component of the wall movement in radial direction. The axial
component is also
( ) 01 2
2
22
2
=
+
−−
++−
= zzRHhE
zu
rw
HHK
tc
c
c
Rr ∂ζ∂
∂∂ξσ
σρ∂∂
∂∂
ρµ
ρζ
∂ζ∂ (2.30)
The terms in the equation from left to right are inertial component, spring force of the
external constraint, the drag force from the fluid motion, and viscoelastic deformation of
the wall substance in axial direction. K represents the tethering tissues spring
24
coefficient and is related to undamped natural frequency of wall element in the z
direction as
2HMK ρ= (2.31)
The wall and the fluid velocity should couple at the boundary, ie
( )cztjeDjdtdu −== ωωξ
1 (2.32)
( )cztjeEjdtdw −== ωωζ
1 (2.33)
Substituting equations 2.32 and 2.33 to equations 2.23 and 2.24 at the boundary (y=1)
yields
c
ACEj0
111 ρ
ω += (2.34)
+=
cACF
cRD
0
11101 2 ρ
(2.35)
Equations 2.29, 2.30, 2.34 and 2.35 form a set of simultaneous equations with four
unknowns: A1, B1, C1 and D1 or if A1 is given C1, D1, E1 and c , the complex wave
velocity. Inspection of 2.29, 2.30, 2.34 and 2.35 indicates that the solution contains only
the known constraints plus the complex wave velocity. Solving for c gives
( )χσρ 20 11
cchE
Rc
−= (2.36)
or
25
( )χσ 2
0
12
11c
cEE
cc−= (2.37)
or written as real and imaginary parts
( ) ( )10
20
11
FjYX
cc c
−
−=−=
σ (2.38)
where c0 of the pulse wave velocity in an infinitesimally thick tube filled with inviscid
fluid and equal to the one given by Moens-Korteweg equation
c EhR0
02=
ρ (2.39)
then X cω 0 and Y cω 0 now become the phase shift and the attenuation coefficient
respectively. The χ is given as
( )
( )2
22
11
c
c HGGσ
σχ
−
′−−±= (2.40)
where
G
Fkc
c=−
−+
′+ −
5 41 2
1 410
σσ
(2.41)
′ =
+ ′−
−H kF
1 21
110 (2.42)
−= 2
0
' 1ωρρ
ρHK
RHk (2.43)
On the other hand, inserting equations 2.34, 2.35 into equation 2.27 gives
26
c
NAC
01
1
ρ= (2.44)
where N is defined as
( ) c
c
c FFN
σσ
σχ 221
22
1010 −−
−−
= (2.45)
The mean velocity across the lumen is obtained by integrating equation 2.19 from y=0
to y=1 which yields
( ) ( )cztjeNFc
Aw −+= ω
ρ 100
1 1 (2.46)
This final equation does not take into account the wave reflection, yet it can be included
if pressure, P, is given as superposition of forward and backward travelling pressure
waves, ie
( ) ( )cztjcztj eAeAP +− += ωω21 (2.47)
then using equation 2.47, equation 2.46 becomes
( ) ( )
( ) ( ) ( )cztjcztj eNFc
AeNFc
Aw +− +−
++= ωω
ρρ 100
210
0
1 11 (2.48)
or,
( ) ( ) ( )[ ]
−+= +− cztjcztj eAeA
cjNF
jw ωωω
ωρ 21100
11 (2.49)
the terms in curly bracket is in fact − ∂ ∂P z , which can be rewritten as
27
− = ′∂∂
ωPz
A e j t (2.50)
then equation 2.49 becomes
( ) tjeNFjAw ω
ωρ 100
1+′
= (2.51)
The instantaneous average blood velocity across the lumen is proportional to the
pressure gradient and the term ( )100 1 NFj +ωρ becomes the longitudinal impedance of
the tube, Zz. The characteristic impedance of the tube is related to longitudinal
impedance by wave velocity (McDonald 1974) as
zZj
cZω
=0 (2.52)
expansion gives
( )10
00 1 NF
cZ+
=ρ (2.53)
and inserting equation 2.37 in equation 2.53 and considering a strongly tethered tube, ie.
N= -1, gives
210
00 11
c
FcZ
σρ
−−
= (2.54)
Discussion of the Validity of the Womersley Approach
Womersley introduced several assumptions in derivation of the solution to fluid flow.
These have been discussed previously (McDonald 1960, Fry and Greenfield 1964;
Milnor 1989, Nichols and O'Rourke 1990). The important points to be related to these
assumptions are the following:
28
(i) Flow is laminar. This condition is fulfilled for most of the cardiac cycle.
However, small bursts of turbulence might occur in the ventricular outflow tracts. Non-
laminar flow increases the viscous dissipation, hence pressure gradient might be higher
for a given flow. The curvature of the ascending aorta introduces rotational flow which
might introduce some error.
(ii) Newtonian flow. In arteries larger than 500 microns the shear rate is high leading
to an asymptotic value of viscosity of blood. However, below this diameter because of
the Fahraeus-Lindquist effect, anomalous viscosity will be apparent (Nichols and
O'Rourke 1990).
(iii) Uniform cylindrical tube. Although arteries are uniform between branching,
there exists a tapering which might lead to convective acceleration (Fry and Greenfield
1964).
(iv) Entrance effects. The entrance effects are shown to exist in the ascending aorta
and major bifurcations leading to an unestablished flow.
(v) Reflected waves. The existence of reflected waves is shown not to affect the
longitudinal impedance provided that forward and backward propagation coefficients
are equal.
(vi) Linearisation of equations. During linearisation, Womersley assumed that the
terms given by u∂w/∂r, w∂w/∂z, u∂u/∂r, and w∂u/∂z are small with respect to the linear
terms, considering no convective acceleration in radial and axial direction. He also
assumed that axial flow velocity to pulse wave velocity ratio, w/c, and radial flow
velocity to pulse wave velocity, u/c, were small. During the cardiac cycle, c is about
500 cm/sec, while the peak flow is 100 cm/sec, which lasts only briefly. The non-linear
pressure-diameter relation of the vessel wall also introduces errors in the boundary. He
29
also neglected the terms ( ) αω ⟨⟨2cR implying that at all times the radius, R, is much
smaller than the wave length. Although the non-linearities described above impose
errors, experiments (Dick et al 1968) showed that they are small and negligible with
respect to the measurement errors. It is thus justifiable that a closed form solution
obtained by linearisation can be used in vivo without sacrifying the ease of calculation.
(vii) Thin tube. It is known that the wall thickness to diameter ratio increases towards
to the periphery. This assumption is used for the Moens-Korteweg equation of pulse
wave velocity for inviscid fluid. This increase, however, in the order of 10-18% causes
10% of errors in wave velocity estimations.
(viii) Isotropic tube. Although it has been assumed isotropic, the vessel wall is far
from being so. Although this point does not introduce major errors it has to be taken in
account in interpretation of wall stresses calculated using these equations.
(ix) Incompressible wall. The compressibility effects the Poisson's ratio, s, making it
different from 0.5. Yet, it is been reported that the wall is normally incompressible
(Carew et al 1968).
(x) Newtonian wall viscosity. Although the viscosity of the wall in arteries is far
from constant, the phase angles in the complex elastic moduli are in the order of 10°-
15°. Being small and not changing with pressure (Gow 1972), the errors introduced by
this assumption are negligible.
(xi) Wall and fluid densities are equal. Arterial wall density is 2% higher than that
of blood.
Although none of the above assumptions are correct in vivo, the errors that could be
introduced are normally less than the measurement errors. Yet, there are consistent
30
differences in the estimation of pressure gradient and derived longitudinal resistance,
which is consistently underestimated (Fry and Greenfield 1964) and differences in the
attenuation of harmonics are greater.
Validation of Womersley's Approach
During validation of any theory, the only and most valuable form of action is to compare
theoretical predictions with experimental findings. In an elaborate theory such as
Womersley's, one then has to measure all the boundary conditions (physical properties
of the wall), fluid properties (viscosity and density) and pressure and flow
simultaneously. One has to measure these variables as precisely as possible to be
confident about the accuracy of predictions. However, even with current
instrumentation the limits of measurement error are not less than about 5%, thus the
theory can only be confirmed within the observable limits.
In the last four decades it has been repeatedly shown (McDonald 1955; Bergel et al
1958; Greenfield 1966; Milnor 1978) that flow wave contour estimated from the
pressure gradient is satisfactory. However, there is growing evidence that the resistance
term and the attenuation predicted by the theory is different from the experimental
findings (Fry and Greenfield 1964, Milnor 1989). These differences have also shown to
exist in every linear theory (Cox 1969). Thus it can be speculated that this is the result
of linearisation of the Navier-Stokes equations, by eliminating the convective
acceleration terms. It normally does not cause gross errors under normal conditions.
However, when one includes the effect of tapering, this effect can cause discrepancies as
much as 200 % (Ling and Atabek 1972). The other effect that has to be taken into
account is the non-linear wall properties which causes steepening of the wave front as it
moves distally (Rockwell 1969).
31
Other Approaches
Linear Models
Womersley was not the only investigator who linearised the Navier-Stokes equations
(see Witzig 1914; Crandall 1927; Jager et al 1965; Whirlow and Rouleau 1965; Atabek
1968; and many others). Witzig (1914) was the first to linearise the Navier-Stokes
formulation for a thin elastic tube. In 1927, Crandall published the expressions for the
velocity profile, average velocity and fluid impedance for an oscillatory flow in a thin
rigid tube. In 1950, Iberal published the analysis of the same flow in a one-dimensional
radial-elasticity equation with restrained wall. After Womersley's solution, there has
been an explosion of linear theories for blood flow in arteries. The detailed analysis of
all these models with respect to Womersley solutions is described in a paper by Cox
(1969) and subsequently by Milnor (1989). It was shown that in the limiting case all
these models predict the same relations for fluid flow and wave propagation. The
individual differences constituted less than 5%. Addition of wall viscosity predicts an
increase in phase velocity and attenuation. The effects of wall compressibility make the
predictions of thin and thick walled models converge as Poisson's ratio approaches zero
and propagation constants in thick and thin walled models become similar as h/R falls
below 0.1.
Non-linear Models
The existence of tapering and non-linear wall elastance as seen above, causes
discrepancies between theory of linear models and experimental findings. Although
small and negligible, advances in computer capabilities often encouraged investigators
to explore the addition of these terms on the predicted flows.
(i) Method of Characteristics
32
It was known by Euler that the closed form solutions to non-linear differential equations
were not available using known analytical techniques, when he developed the first
theoretical analysis of inviscid flow in a tube in 1775. In developing this analysis he
became aware that the boundary was non-linear, ie wave velocity was dependent on
pressure. The problem was not tackled until 1958 when digital computers began
making an impact on mathematical solutions to non-linear equations. It was Lambert
who used this opportunity to suggest a method of characteristics which could be used to
solve the problem.
The major drawback of the approach is that flow is one-dimensional and requires
directional terms to be supplied by another theory or experimentation. One-
dimensionality leads to the characterisation of flow and pressure in the axial direction in
time but lacks the description of radial distribution of the velocities. In that sense, the
technique is cruder than using linearised solutions.
The basic equation of motion is in radial coordinates
fzP
zww
tw
=++∂∂
ρ∂∂
∂∂
0
1 (2.55)
where w is the axial velocity and f the net effect of shear stresses due to viscosity of
fluid at the boundary of the vessel. The continuity equation is (Rockwell 1969, Skalak
1972)
( ) 0=Ψ++z
SwtS
∂∂
∂∂ (2.56)
S, the vessel lumen area is considered to be a function of pressure and distance.
S S P z= ( , ) (2.57)
33
In equation 2.56, Ψ denotes the efflux of fluid through the walls per unit length. This is
a conventional way to include branching. The equation of motion and the equation of
continuity can be combined to the form
21 LLL λ+= (2.58)
where L1 is the continuity and L2 motion equations. After eliminating dependent
variable S by considering
zP
PS
tS
z ∂∂
∂∂
∂∂
= (2.59)
and
Pz z
SzP
PS
zS
+
=
∂∂
∂∂
∂∂
∂∂ (2.60)
equation 2.58 leads to
( ) 01
0
=Ψ+
+−
+
+
+
++= λ
∂∂λ
∂∂
∂∂
∂∂λρ∂
∂λ∂∂
∂∂λ
P
z
z PSwf
tP
zP
PS
wPS
tw
zwwL (2.61)
Taking derivatives of P and w with respect to t leads to
tP
dtdz
zP
dtdP
∂∂
∂∂
+= (2.62)
and
34
tw
dtdz
zw
dtdw
∂∂
∂∂
+= (2.63)
Inspecting equations inside the curly brackets in equation 2.61 it is evident that the first
bracket will be dw/dt if
λ+= wdtdz (2.64)
and the second brackets will be dP/dt if
zPS
wdtdz
+=
∂∂λρ0
1 (2.65)
then combining equation 2.64 and 2.65 one obtains
zPS
ww
+=+
∂∂λρ
λ
0
1 (2.64)
which is
cSP
z mm =
=0ρ
∂∂
λ (2.65)
where c is the pulse wave velocity. Then the four sets of equations obtained
dzdt
w c= m (2.66)
and
35
Ψ
=± mmm
tS
zSw
Scf
dtdP
cdtdw
∂∂
∂∂
ρ0
1 (2.67)
After writing the equations on a finite difference scheme using a mesh ratio of
Θ ∆ ∆= t z and satisfying the condition of stability and convergence by
( )cwz +⟩∆ (2.68)
At all times the intermediate values can be obtained with respect to the generated mesh
using boundary conditions and the known relations between P, z and S and functions
for f, Ψ (Corey et al 1975).
Porenta et al (1986) considered ∂ ∂P z = 0 at the branch level of an arterial bed which
was represented as a modified Windkessel. The branch flow becomes
012
1
22 =
+−+− ββ
RR
RP
dtdCR
dtdPC (2.69)
where C is the compliance, R1 and R2 are the series and parallel resistances to the
compliant element, β is the difference between the flow before and after the branching
point in the main arterial stem. The frictional force was written as f FV ni i i= −2 where
Fi, Vi and ni are the constants in i= +z, -z directions. The solution of these equations in
such a leaking tube displayed the progressive increase in pulse amplitude and the
steepening of the pressure wave. It also displayed a good agreement in calculated mean
pressure.
The models of this kind also employed viscoelastic wall (Anliker et al 1978), branches
(Porenta et al 1986) and obstructions (Anliker et al 1978; Porenta et al 1986). The
predicted wave shapes are close to the ones obtained in vivo.
36
Extensive use of computers in the field of biomedical research increases the potential
application of these models. Since the computing time was quite expensive and this
technique did not produce enough improvements to the modelling, it coined the term
"steam-hammer for crushing peanuts". The technique also needs measurement of
diameter change and/or viscous forces which are difficult to achieve.
(ii) Wavelet Analysis
An alternative approach has been suggested by Parker and Jones (1990) using wavelets
to describe the energy per unit area by constituting equations of
zz dPdPdP −+ += (2.70)
zz dwdwdw −+ += (2.71)
indicating forward, +z, and backward, -z, waves. It was assumed that if the viscous
losses are ignored in large arteries the integral
∫±=±
P
Pz cdPwR
0 0ρ (2.72)
is constant along the characteristic paths dzdt
w c= ± . Thus, the energy of the wavelet
becomes
220 zz dRdRcdPdw −+ −= ρ (2.73)
leading to the separation of expansion of wavelets from other factors. Real time
separation of reflected waves by this technique is its big promise.
37
2.2 ANALOGUES OF THE ARTERIAL SYSTEM
"An analogue implies a recognised relationship of consistent mutual similarity between
the equations and structures of two or more fields of knowledge." (Attinger 1964). To
facilitate the transfer of knowledge between two systems, the parameters and structural
elements of the similar kind are identified and related. For this purpose, often the
differential equations describing two different systems are inspected and related. The
most common of these systems is the electrical circuit theory, since it is easy to
implement and to confirm results with the other fields.
Therefore, the electrical analogies are of practical importance not only for didactic
purposes but also for the establishment of an elegant, clear and relatively simple
treatment. The analogues of fluid systems in electrical circuitry terms are well
established (Schonfield 1953), as given in Table 2.1.
Symbol Electrical
Symbol Mechanical
Q Charge x Displacement I Current dx
dt
Velocity
V Voltage F Force L Inductance M Mass R Resistance R Frictional Resistance C Capacitance K Spring Constant
Table 2.1 Analogies indicated by differential equations for the electrical system and the mechanical systems.
38
2.2.1 Lumped Analogues
Zero Order Analogues
The simplest analogy is given in terms of electricity is Ohm's law to the one of
peripheral resistance:
( )tIzC
zI
∂∂
∂∂
=− (2.74)
where V is the voltage, corresponding to pressure gradient and I is the current,
corresponding to flow. Since the differential equation describing this equation is zero
order, they are called the zero order approximation. A borrowed term from electrical
engineering called impedance relates the voltage to current. The impedance has a
frequency dependent nature in describing this relation. Basically it is written as
( ) ( )( )ωωω
IVZ = (2.75)
thus the impedance of zero order system is
( ) pRZ =ω (2.76)
First Order Analogues
This analogue represents the Rp and C elements in parallel, where Rp is the Poiseuille
resistance and C is the compliance of the bed, both given as per unit length
RRp =8
4µ
π (2.77)
39
+
+
=
hRE
hRR
C21
13 2ρ (2.78)
The impedance of the analogue is given as
( )CRj
RZ
p
p
ωω
+=
1 (2.79)
which is also called the impedance of the Windkessel (Nichols and O'Rourke 1990).
There are other variations of this circuitry, one called Westkessel (Nichols and O'Rourke
1990) described as
( )CRj
RRZ
p
pc ω
ω+
+=1
(2.80)
where Rc is the characteristic resistance, added in series, after Westerhof et al (1977).
These analogues as they imply, lump the behaviour of the arterial bed. Therefore, they
can describe the overall behaviour of this lumped system rather than its components
(Cevenini et al 1987).
Second Order Analogues
Another lumped analogue is the second order system where the inertia of the blood is
also taken into account. The inductance, L, thus is in series with the Windkessel.
( )CRj
RLjZ
p
p
ωωω
++=
1 (2.81)
Where inductance per unit length, L, is given as
40
LR
=ρ
π 2 (2.82)
The equation given above also describes the action of the fluid filled manometer. It is
not surprising therefore that the Windkessel as first described by Frank (1899), is also
used in conjunction with manometer theory which is also a derivation of Frank (1903).
The second order system given above also describes a resonance phenomenon-at
frequency LCn /1=ω . This notion is taken up by followers of Frank to explain the
different shapes of the pulse in different arterial beds (Warner 1957; Remington and
O'Brien 1970). This concept of resonance is still in fashion in different circles (Watt
and Burrus 1976; Wang et al 1991).
The models described above lack the description of flow shapes in different arteries.
However, they can be quite useful in lumping an arterial bed downstream without
having to know individual geometry of branching and vessel characteristics. This point,
when taken in gross terms, had been used in the past to calculate stroke volume from
pressure pulses (Frank 1930; Cope 1965; see McDonald (1974) for historical review).
This is useful in limited cases because it does not account for wave reflection and
transmission. Also, satisfactory results have been obtained in modelling the arterial
system as transmission lines and where distal arterial beds are terminated with complex
loads (Liu et al 1989; Burattini et al 1989b).
High Order Analogues
The arterial system could be better modelled as a composition of several viscoelastic
tubes representing vessel properties as a function of space. This type of dispersion of
physical properties along the arterial tree generates parameters of wave travel. It also
solves many of the problems associated with lumping such as infinite wave speeds, no
conception of reflections and no change in wave shape. The arterial segments can be
represented with various analogues. If arterial segments are chosen significantly small
41
enough, the model constructed with these analogues approximates the real system. In
early representation of arterial elements by electrical analogues (de Pater and Van den
Berg 1963, Jager et al 1965; Westerhof et al 1969), modified Windkessels with inertial
terms had been used to represent the artery under investigation. By taking data from a
wide variety of sources, early investigators were able to construct an entire arterial
system by adding elements representing individual arteries. The values that were used
were represented by equations 2.79 to 2.82 which converged to give encouraging results.
However, there were considerable discrepancies with the experimental data, probably
due to lumping of the individual segments and ignoring frequency dependence of
elements. The other problem with this approach was that the model elements could
hardly be changed; each element contained four passive elements and there were 121
elements in the entire representation.
2.2.2 Uniform Transmission Line Analogues
If it was possible to reduce the arterial segment size represented by each element to a
finite minimum, then a lumped analogue for that segment approaches a uniform
electrical transmission line. The arterial system represented by these elements then
would represent the continuum of varying physical properties reflected on the electrical
analogue. This approach coupled with the increased computing power led to the
exploitation of transmission line equations and other properties derived from it, as
analogues of fluid flow in arteries (Taylor 1957a, 1957b).
Linearisation of Navier-Stokes equations together with continuity equations reduces the
partial differential equations describing the fluid flow in a tube to the telegraphist
equation, with parameters Rz, Rr, L and C per unit length (Nichols and O'Rourke 1990)
42
tILIR
zV
z ∂∂
∂∂
+=− (2.83)
and
− = +∂∂
∂∂
Iz
VR
CItr
(2.84)
Where Rz is the longitudinal resistance, L is the fluid inertance, C is the arterial
compliance, Rr is the radial resistive component of the vessel wall and V(z, t) and I(z, t)
is the pressure and flow. Since steady state oscillations are assumed to be present in the
system, one can write current, I, and voltage, V, as
( ) ( ) tjezItzI ω=, (2.85)
( ) ( ) tjezVtzV ω=, (2.86)
Then from equation 2.83 and 2.84 one obtains
( ) ( )zIz
zI 22
2
γ∂
∂= (2.87)
where the propagation coefficient, γ , and characteristic impedance, Z0 , are defined as
( )( )CRj
LjRCjr
z
ωω
ωγ+
+=
12 (2.88)
( )( )( )Cj
CRjLjRZ rz
ωωω ++
=1
0 (2.89)
The solution of equation 2.87 then becomes
zz BeAezI γγ += −)( (2.90)
43
where A and B are arbitrary constants. If the boundary conditions are given as
0)0( II = , at z=0 (2.91)
V l I l ZT( ) ( )= , at z=l (2.92)
where ZT is the terminal impedance. If, reflection coefficient, Γ , is defined as
0
0
ZZZZ
T
T
+−
=Γ (2.93)
one can now write
V z I Z e ee
z z l
l( )( )
=−
+
− −
−0 0
2
21
γ γ
γΓ
Γ (2.94)
Using equations 2.88 and 2.89 together with 2.52 and defining
( )jYXcj
−=ωγ (2.95)
One obtains analogues that are also defined in Womersley equations(Taylor 1959b)
( )
+−==
1000 1
1ImReNF
ZRz ρωµγ (2.96)
( )
+==
1000 1
1ReIm1NF
ZL ργω (2.97)
( )( )( ) ( )
+−==
1020
20 1
1ImReNFjYX
cZRr ρωγ (2.98)
44
( ) ( )( ) ( )
+−=−=
1020
20 1
1ReIm1NFjYX
cZC
ργω (2.99)
Thus, there is a one-to-one correspondence between Womersley solutions and the
electrical uniform transmission lines. Thus under the assumptions of linearisation one
can utilise existing notions of transmission lines such as propagation constant,
definitions of impedances and reflection coefficient. These terms in turn, help to
understand the reasons why there are reflections and delays in the arterial system and the
implications of wave propagation in the arteries. Backed by solid theoretical and
experimental evidence this analogue is the most elaborate. It can be used as the
fundamental tool in investigation of the problems with pressure measurements. See
chapters (6.2 through 6.4).
2.2.3 Non-Uniform Transmission Line Analogues
It is well known that the arterial system is not composed of uniform tubes, it undergoes
both geometric and elastic taper with subsequent increase in wall stiffness and wave
velocity. Although it is possible to model this system as a series of tubes of increasing
stiffness, general analytic expressions using uniform tubes are not available. Taylor
(1965) investigated wave travel and reflection in non-uniform tubes, where losses are
ignored. Replacing the C and L in equations 2.85 and 2.86 by C(z) and L(z) and
ignoring radial and axial losses one can obtain
( )tIzC
zI
∂∂
∂∂
=− (2.100)
( )tIzL
zV
∂∂
∂∂
=− (2.101)
Writing I, and V in complex form
45
I I jI= +1 2 (2.102)
21 jVVV += (2.103)
where dependence on z is not indicated. With boundary conditions of
( ) ( ) TZlIlV 11 = (2.104)
( ) ( ) TZlIlV 22 = (2.105)
( ) ( ) ( )000 21 jVVV += (2.106)
where ZT is the terminal impedance. The reflection coefficient at distance z becomes
)()(
)()(
211
)()()(
)(2
11
000
000
lZZzZzlZ
lcj
lZZzZzlZ
lcj
Tlz
Tlz
+
+
−
−
=Γ
=
=
∂∂
ω
∂∂
ω (2.107)
where c(z) is the nominal wave velocity and Z0(z) is nominal characteristic impedance
defined as
)()(
1)(zCzL
zc = (2.108)
)()(
1)(0 zCzLzZ = (2.109)
The voltage at point z then becomes:
∂∂
∂∂
∂∂
ω2
2
2
21 0V
z L zL z
zVz c z
V− − + − =( )
( )( )
(2.110)
46
where the solution of equation 2.110 becomes
∫−
=
z
zcdzj
eZ
zZVzV 0
)(
0
0
)0()()0()(
ω
(2.111)
By considering a wave velocity of the form )(2)0()( zCosczc π−= and inductance of
the form L(z)=1, Taylor (1965) was able to show that non-uniformity enabled the load
to be decoupled from the input, with a smoother impedance spectrum than the one with
the uniform line. Furthermore, the wave amplitude was dictated by equation 2.111.
This kind of dependence of wave amplitude to characteristic impedance ratios partly
explains the progressive increase in pulse pressure from central to peripheral arteries. In
contrast to the uniform line, the transmission ratio was always higher than unity, even
for higher harmonics. This agrees with experimental findings.
A non-uniform tube was analysed by Fich et al (1966), who considered spatial
distribution of radius of the form 2)0()( kzeRzR −= . This was then converted for
expressions of C(z) and L(z), which was than subsequently employed in obtaining
closed form solutions for spatial distribution of pressure and flow along this tube.
Another tube where an exponential function was assigned to the inductance L(z) term is
also proposed (Einav et al 1988). The solutions were similar to the one proposed by
Fich et al (1966).
The non-uniform analogy does not provide a closed form solution, unless spatial
distribution of physical properties is given in analytical form. Thus, it does not give
expressions for spatial distribution of wave velocity and characteristic impedance.
47
2.2.4 Implications of Transmission Line Concepts
Transmission Direction
The concept of transmission, certainly involves wave travel in given directions, namely
upstream and downstream. All the concepts given below assume that the waves are
travelling downstream which can be affirmed by looking at the wave and/or its
components at two arbitrary sites. If the time delay between the two waves is negative,
ie. the wave at the second site arrives later, then the first one is called the downstream
direction, +z, indicating the sources of these waves are on the upstream direction, -z,
and vice versa.
Impedance
Impedance can be generally defined as the opposition to flow (or current) as dictated by
the downstream properties of the system. It can be subdivided into 4 categories:
longitudinal, transverse, characteristic and input impedances.
Longitudinal Impedance
This is the opposition of fluid in the direction of pressure gradient. It can be formulated
as
( )102
0
1 NFRj
zPZz +
=−=π
ωρ∂∂ (2.112)
Since ∂ ∂P z is a function of the spatial direction, it is a point description. For an
oscillatory flow this definition is not effected by the properties of vasculature at the
terminations that might introduce reflected waves. It can be described as electrical
analogue
48
zzz LjRZ ω+= (2.113)
where Rz is the longitudinal resistance describing the losses during the movement of
fluid along the z direction. Lz becomes the hydraulic inductance, related to the inertial
forces opposing to the fluid flow. It can be shown (Fry and Greenfield 1964) that those
terms can be confined with 2.112 after normalisation to the resistive term at zero
frequency
( )
+=
10
2
1Re
NFRz
α (2.114)
( )
+=
101Im
NFjLz (2.115)
It has been shown (Fry and Greenfield 1964, Ling and Atabek 1972) that although Lz
measured experimentally agrees well with the theory, Rz derived this way can be quite
different. This is probably due to the convective terms in the Navier-Stokes equations
which were ignored during the derivation of the equation 2.112.
Transverse Impedance
This form of impedance takes into account the vessel wall properties and continuity
equations in radial directions. In short form it is written as:
zQ
PZr ∂∂−= (2.116)
where zQ ∂∂− denotes the movement of the wall. If the wall element is thought of as
a series resistive element Rr with a capacitive one C the transverse impedance in
analogue form becomes
49
Z Rj Cr r= +
1ω
(2.117)
where Rr denotes the viscous losses due to movement of the wall, while C is the elastic
properties of the wall. It can be shown that the wall element's capacitance can be
written (McDonald 1974) as:
EhRC
22π= (2.118)
indicating a dependence on the wave velocity c0. The value of Rr cannot be deduced in
an analytical way, but it can be described by taking into account interaction between the
fluid and the wall (Taylor 1957b) as in equations 2.98 and 2.99.
Characteristic Impedance
If the transmission line is not terminated, the relation between pressure and flow
becomes the characteristic impedance.
0
00 Q
PZ = (2.119)
Where P0 and Q0 denotes pressure and flow when there is no load at the end of the line
respectively. It indicates the interaction between the wall properties and the fluid
motion and can also be written as
Z Z Zz r0 = (2.120)
The expression of the electrical analogue is given by equation 2.89. Noting its relation
to propagation coefficient and longitudinal impedance, characteristic impedance is
50
ωjcZZ z=0 (2.121)
Considering a viscoelastic wall the elastance can be rewritten in complex form (Bergel
1961a, 1961b; Gow 1972)
ωηjEEc += (2.122)
where η is the wall viscosity coefficient of the wall. The phase of Ec then becomes
= −
ETan ωηθ 1 (2.123)
The phase angle is shown to approach a constant with higher frequencies. The
exponential form which is used to model this behaviour is given as (Taylor 1966a)
( )ωθθ ke−−= 10 (2.124)
where k is found the be around 2 (Gow and Taylor 1968, Imura et al 1990).
Then using equation 2.54; where it was shown that characteristic impedance is directly
proportional to wave velocity and complex elastance, the characteristic impedance per
unit area for strongly tethered tube becomes
2
1022
000 1
11
θ
σπ
ρ jeFR
cZ −
−−= (2.125)
None of these equations take reflected waves into account, so they are only valid for
propagation of forward or backward waves.
51
Input Impedance
This is the impedance seen at the source (or input) of the transmission line, whether the
line is loaded (terminated with a load) or not. It is expressed as
( ) ( )( )ωωω
QPZin = (2.126)
Where P(ω) and Q(ω) represent the frequency components of pressure and flow
respectively. This equation, when displayed as modulus and phase, presents the
characteristics of a transmission line as a function of frequency. It expresses the load
presented to the source, lumping all the properties of the line at the point of source and
beyond. When there is no load at the termination, this equation reduces to the one of
characteristic impedance. Thus the relation between characteristic and input impedance
is given as
l
l
in eeZZ γ
γ
2
2
0 11
−
−
Γ−Γ+
= (2.127)
where Γ is the coefficient of reflection given in Equation 2.93. By definition Γ has a
value of 1 for closed and -1 for open end termination.
Propagation Coefficient
As waves propagate, their amplitude becomes attenuated. Also due to finite wave speed
the propagation occurs in a finite time (see equation 2.87). The two events can be
described by propagation coefficient in the form of
jba +=γ (2.128)
52
where a is the attenuation constant (nepers/cm), describing the losses in the line due to
wall and fluid viscosity. The phase constant, b, is the time delay introduced by the
interaction of the forces between the wall and the fluid and denoted by radians/cm.
Another form of writing this expression is
γω σ
=−
+c NF0
2
10
11
(2.129)
Transmission Ratio
Any measured wave any point along the line, Pp , can be represented as combination of
forward Pf and backward Pb travelling waves
bfp PPP += (2.130)
If at distance l from the origin O a reflecting site L exists with a complex reflection
coefficient Γ , then the forward pressure wave at this point becomes P efl−γ and the
reflected pressure wave will be ΓP efl−γ . Using equation 2.130, the measured pressure
wave at point L will be
lf
lfL ePePP γγ −− Γ+= (2.131)
If reflected wave travels upstream to site O, it will attenuate by e l−γ amount. Again
using 2.130 the pressure wave at O will be
llffo eePPP γγ −−Γ+= (2.132)
Then, the relation of PL with respect to Po will be
PP
e ee
L
o
l l
l=+
+
− −
−
γ γ
γΓ
Γ1 2 (2.133)
53
Which is called the transmission ratio or transfer function between sites O and L.
Obviously, for flow waves the Γ has to be denoted as -Γ for energy considerations.
2.2.5 Application of Transmission Line Concepts in Haemodynamics
The arterial system is a network of tubes, not a single chamber. Therefore, transmission
line theory on the properties of fluid flow is more appropriate. This approach was
extensively used by Taylor (1957a) who first considered the electrical transmission line
as an analogue of the arterial system. By assuming the arterial system to be in steady
state oscillation and linear, since the pulsations are small and wave lengths are long,
Taylor was able to apply telegraph equations to the problem of fluid flow in arteries.
Although Porjé (1946) was the first to use Fourier analysis in this system, it was Taylor,
being a close associate of McDonald and Womersley, who pointed out the physical
meaning of this analysis.
In the treatment of the problem, he used electrical analogues to represent the fluid flow
and considered concepts such as wave propagation and wave reflection, input
impedance, terminal impedance and characteristic impedance. Subsequent works by
Gabe (1965), Patel et al (1965), O'Rourke and Taylor (1966) and Milnor et al (1966)
highlighted these concepts as useful in describing the characteristics of certain arterial
beds.
Interpretation of Impedance
Input impedance of an arterial bed can be calculated by making measurements of
pressure and flow at a site proximal to an arterial bed. If the site is the ascending aorta
then the impedance will be the systemic impedance. After calculation of the frequency
54
components of the pressure, P , and of the flow, Q , by discrete Fourier Transform
(DFT) or by correlation techniques (Taylor 1966c), one can then obtain
( ) ( )( ) ( ) ( )ωϕωωωω jeM
QPZ == (2.134)
where modulus, ( )ωM , and the phase, ( )ωϕ , of impedance are expressed as
( ) ( )( )ωω
ωQP
M = (2.135)
( ) ( ) ( )ωωωϕ QP ∠−∠= (2.136)
The input impedance modulus when plotted against frequency displays the peripheral
resistance as the zeroth harmonic. It then descends steeply from this value to essentially
a minimum as dictated by the reflections present in the system. Modulus then oscillates
around a mean value, (characteristic impedance), to approximate it asymptotically. The
phase on the other hand starts from a negative value, slowly approaches zero, often
crosses it, then oscillates around zero mostly being positive (Nichols and O'Rourke
1990). The spectrum obtained in this way will obviously be an estimate of the actual
spectrum. Since it is sampled at the fundamental rate, heart rate, the continuous
spectrum can be estimated by interpolating the harmonics by linear analysis or by pacing
the heart at a range of frequencies (Taylor 1966c). It is often suggested (Nichols and
O'Rourke 1990) that the characteristic impedance, Zo, be estimated by averaging the
values after the first minimum or above 2 hz. The complex reflection coefficient can be
estimated by
0
0
ZZZZ
in
in
+−
=Γ (2.137)
55
Although there are no propagating waves, thus no wave reflections at zero frequency,
zero intercept of the reflection coefficient can be calculated as analogous to ZT in
equation 2.93. This takes into account the low frequency components of the impedance
spectrum but excludes the often used peripheral resistance (O'Rourke and Taylor 1966)
or the oscillations in the frequency spectrum (Nichols and O'Rourke 1990) both of
which can change independently by certain drugs (Yaginuma et al 1986).
The minima of the impedance spectrum can be used to estimate the distance to the
reflection site. When a single tube is being considered there will be one termination and
the distance can be estimated using quarter wavelength formula:
min
0
44 fc
l ==λ (2.138)
where co is the assumed wave velocity for this tube. This equation for a single tube
assumes that the second minimum will be at triple of this minimal frequency. If it is not
the case, ie a second minimum is found at frequency other than this, it suggests a second
tube parallel to the first one that is also been terminated by another load. The distance
to this reflection site can also be estimated using equation 2.138 (O'Rourke 1967b;
Nichols and O'Rourke 1990). Note that the wave speed remains the same in both cases.
This kind of impedance spectrum was mostly found in the ascending aorta of different
species, including man (Avolio 1976a, Nichols and O'Rourke 1990), leading to the
reduced model of systemic arteries as an eccentric T-Tube, where the long arm supplies
to the lower part of the body but the short is to the upper part (O'Rourke and Taylor
1967a). This view is later taken up by Liu et al (1989) and Burattini et al (1989b) who
extended the model and applied it to estimate vascular parameters. The influence of
wall viscosity, blood viscosity, reflection coefficient and wave speed has been studied
extensively using models (Taylor 1966a, Avolio 1980, Liu et al 1989) and in vivo
56
(O'Rourke et al 1968). It was shown that viscosity of the blood and of the wall
influences the oscillations in the impedance spectra.
The effects of various vasoconstrictor drugs on input impedance had been investigated
in the past using epinephrine, angiotensin or norepinephrine (O'Rourke et al 1992).
Data show little difference from the control, apart from the effects of increased wave
velocity brought by increased mean arterial pressure. This indicates the presence of
significant vasoconstrictive state under control conditions. The estimated reflection
coefficient obtained using peripheral resistances is 0.8 for control and 0.98 under
vasoconstrictor therapy.
The effects of vasodilator therapy are profound, however. Acetylcholine (O'Rourke and
Taylor 1966), nitroprusside (Pepine et al 1979; Gundel et al 1981; Yin et al 1983,
Merrillon et al 1984; Laskey et al 1987, Chang et al 1990) and nitroglycerine (Westling
et al 1984, Yaginuma et al 1986; Latson et al 1988, Fitchett et al 1988) profoundly
effects the lower harmonics of the aortic input impedance and thus the reflection
coefficient from 0.8 to 0.5 or lower.
The effect of various physical manoeuvres; exercise (Laskey and Kussmaul 1987;
Murgo et al 1980a) and Valsalva manoeuvres (Murgo et al 1981) have been investigated
and have shown to change the impedance spectrum. These alterations are brought by
changes in the magnitude of reflected waves or wave velocity or the peripheral
resistance.
The systemic input impedance has also been used as a unique way of expressing the left
ventricular load. When expressed as steady and pulsatile load, it can differentiate
between the efficiencies of different arterial systems. It is also related to the oxygen
demand of the left ventricle for a given supply (Weber and Janicki 1977). The external
power output of the heart, Wn , is defined through impedance as
57
nn
N
nnn CosZQQPW ϕ∑
=
+=1
200 2
1 (2.139)
where the first term is the zeroth and second term is the higher terms of the Fourier
components of the pressure and flow and impedance respectively and ϕn is the phase of
the impedance. This equation does not include power dissipation due to kinetic energy
of the blood since this component is relatively small. The term P Q0 0 thus represents
the steady power while the other term is the pulsatile component. The ratio of pulsatile
to total power is used as an index of "arterial efficiency" with respect to
ventricular/vascular coupling. This ratio is around 10% under normal conditions
(Nichols and O'Rourke 1990).
Since the basic characteristic shape of the flow pulse does not alter from individual to
individual, even from species to species (Nichols and O'Rourke 1990), it was suggested
that the alterations in the pressure waveform are mainly due to changes in the impedance
spectrum. Furthermore, since in experimental animals and in young healthy humans
pulse wave velocity is similar in corresponding arteries, pressure wave patterns are
largely determined by length of arterial segments and by body patterns (O'Rourke 1982).
Murgo et al (1980b) investigated the possibility of classifying the arterial pulse due to
ascending aortic input impedance spectrum. It was concluded that the second shoulder
appearing in central pressure waveform was due to an oscillatory impedance spectrum
brought by either earlier or increased wave reflection. This study was later supported
with the finding that, in older subjects with increased characteristic impedance (Avolio
et al 1983b, 1985a), the systolic pressure augmentation also increases (Kelly et al
1989a, 1989b, 1990a) due to early wave reflection (O'Rourke 1988).
The normalisation of impedance spectrum facilitates the comparison from different sets
of data from different arterial beds or from different species. For this purpose the units
used in describing the modulus is often normalised (Nichols and O'Rourke 1990). It
was found not surprisingly that when modulus is described as a proportion of the
58
characteristic impedance, it normalises the impedances obtained from different species.
Using equation 2.127 and expressing input impedance as a proportion of the
characteristic impedance leads to
l
lin
ee
ZZ
γ
γ
2
2
0 11
−
−
Γ−Γ+
= (2.140)
This normalisation allows one to express input impedance as a function of reflection
coefficient, propagation coefficient and the length. Further normalisation is possible if
one combines equation (2.129) with equation (2.138) to obtain
10
2
minmin
0
10
2
0 11
2411
NFff
fc
NFcl
+−
=+−
=σπσωγ (2.141)
leading γl to be proportional to f fmin . Thus, if the abscissa is normalised by fmin, the
impedance spectra for both modulus and phase can be expressed as a unique identity,
just a function of ( 101 NF+ ) and thus α and the complex reflection coefficient Γ . This
enables comparison of experiments as a function of α and Γ .
Interpretation of the Transfer Function
Transfer function (transmission ration) between two sites, proximal and distal, can be
calculated by taking two simultaneous pressure measurements from these sites and
subjecting them to Fourier analysis either by discrete Fourier transform (DFT) or
spectral techniques (Nichols and O'Rourke 1990). The obtained moduli, ( )ωdM and
( )ωpM , and phases, ( )ωϕd and ( )ωϕ p , of the proximal and distal pressures are then
related as
( ) ( )
( ) ( ) ( ) ( )ωϕωϕ
ωϕ
ωωω j
jp
jd
DP eMeMeMH
p
d
==− (2.142)
59
to obtain the transfer function HP D− . Here, while ( )ωM describes the modulus or the
amplification ( )ωϕ describes the phase delay between these two signals.
The modulus of the transfer function when plotted against frequency, displays
successive minima and maxima depending on the intensity of reflected waves (Figure
2.1, Top).
Figure 2.1 A typical transfer function between ascending aortic and the radial arterial pressure pulse. Modulus of the transfer function starts from unity at zero hertz and exceeds unity at higher frequencies to reach a peak due to presence of reflected waves. The phase delay presents itself as time difference between the ascending aortic and radial pressure waves. Phase of transfer function starts from zero radians at zero frequency indicating no delay. With increase in frequency, phase becomes negative indicating a progressive phase delay. Note the oscillations in the phase values which settle to a constant value dictated by wave speed (see text for more explanation).
The component at zero frequency displays the ratio of segment resistance to that of the
resistance distal to the measurement site and it is often very close to unity (Nichols and
60
O'Rourke 1990). Since there is an elastic tapering (Taylor 1964) as well as reflections
in the arterial system under normal circumstances, the modulus reaches a value above
unity with increase in frequency. The position of this peak on the abscissa can be
estimated using the quarter wave length formula (equation 2.138). The modulus then
gradually decreases to a value below unity.
Because of this prominent peak, it was often assumed that the segments of arteries
behave like a resonating second order systems (Warner 1957; Remington and O'Brien
1970) drawing an analogy between this and manometers. Utilising this similarity,
Warner used four pole networks to simulate the effects of pulse transmission in arteries
(1957). Although, this manometer analogy was proved successful to an extent, it is not
complete due to discrepancy in attenuation at higher frequencies. The second order
system analogues always attenuate at 40 dB per octave above "resonant frequency"
(Oppenheim et al 1983) while lesser attenuation was observed in experimental settings
(Lasance et al 1976; Karamanoglu et al 1992). The values obtained approached close to
the theoretical predictions (Taylor 1965) brought by presence of elastic tapering in the
arteries (Taylor 1969).
Phase of the transfer function starts from zero and remains negative, being linearly
dependent on the frequency (Figure 2.1, Bottom). This behaviour had been treated more
carefully than modulus and shown to contain information about reflected waves. Porjé
analysed the phase relation between the two pressure signals to obtain a reliable estimate
of the pulse wave velocity (Porjé 1946). For this purpose, he used the following
equation to calculate phase velocity
( )ωϕωlcph −= (2.143)
61
where cph ( )ω is expressed in cm/sec and ( )ωϕ is the phase of transfer function. To his
surprise, he found that the lower harmonics were travelling at speeds higher than the
high frequency components. Porjé correctly identified the cause of this phenomenon-as
wave reflection (Cox 1971). In solving the telegraphist equations, Taylor derived the
formula relating the wave speed to reflections and later confirmed his findings in
experiments in rubber tubes (Taylor 1957a, 1957b). The expression was of the form
( )[ ] ( )[ ]( )[ ] ( )[ ]βα
βα
−−−−−
−−−−−=
′
bzl2Sinbaazl2Sinh
bzl2Cosazl2Coshcc
0
(2.144)
where a, b are the real and imaginary parts of the complex propagation coefficient; α
and β are the elements of the reflection coefficients which is of the form
( )βα je +=Γ 2 (2.145)
The expression 2.144 is a point description and transforms to
( )[ ] ( )[ ]( )[ ] ( )[ ]βα
βα
−−−−−
−−−−−−
=′
bzl2Sinbaazl2Sinh
bzl2Cosazl2Coshln)za(z2
1cc
120
(2.146)
when actual measurements are done. In this form z1 and z2 are the proximal and distal
measurement sites respectively. These equations point out that wave velocity is
dependent on the distance to the reflecting site and to the magnitude and phase of the
reflection coefficient.
The phase velocity is also related to impedance by (McDonald 1974) by
10
20
MRcZin π
ρ= (2.147)
62
where c is the wave velocity derived using equation 2.143. Since radius, R , and M10
are relatively constant, α does not change considerably since it approaches its limit
value very quickly, there is a linear relationship between the modulus of the impedance
spectrum and the phase velocity (McDonald 1974). It was shown that the phase velocity
obtained this way was closely correlated with impedance under a wide variety of
conditions (Latham et al 1985; Milnor 1989), including valve disease.
Although viscosity of the wall and the blood as well as the cross-sectional area have
been shown to be very influential on the transmission of the pressure wave, these factors
have little influence on the input impedance (Taylor 1966b).
The relation between the modulus and phase of the transfer function is complementary
as they both contain similar information. For example, the peaks of moduli correspond
to the slower wave speed as estimated from the phase velocity. The positions of the
prominent peaks are also related to the asymptotic value obtained by the phase velocity
by equation 2.143.
The normalisation of the transfer function for comparative purposes can be obtained by
scaling the frequency axis by the quarter wave length formula using equation 2.143.
The phase velocity axis can also be scaled using the asymptotic value as proposed in the
impedance section. This leads to the direct comparison of the impedance spectrum with
the phase velocity spectrum. The reflection coefficient from phase velocity spectra can
be obtained by using formulation given by Taylor (1957a)
( )azlecccc
22
0
0 −
+′−′
≅Γ (2.148)
63
where the assumptions are that reflections are positive and that the point z2 is close to
the termination with respect to z1. A much easier formulation can be obtained by
assuming l z− ≅2 0
0
0
cccc
+′−′
≅Γ (2.149)
where c' is the phase velocity estimate at around zero hertz. If this value is not
available, it can be obtained by linear interpolation. This expression of reflection
coefficient and the equation 2.137 is obvious.
The wave travel phenomenon is most apparent when one considers the delay introduced
by transmission. Therefore, most of the early investigations of wave travel were
devoted to accurate measurement of wave velocity. In order to measure wave velocity, a
point is defined on the wave and time needed for this point to travel some distance was
measured. The most probable candidate for such an identifiable point was obviously the
foot of the wave, and it turned out to be the most valid. There are problems, however, in
the accurate description of the foot of the wave. It was defined as
(i) the maximum point of the first derivative,
(ii) the minimum pressure point before the upstroke,
(iii) the intersection of the tangents drawn from the diastolic point and from the
systolic point,
(iv) maximum point of second derivative etc.
Since the upstroke of the wave is less contaminated by the reflected waves that often
interfere with measurements (Nichols and O'Rourke 1990), the consistency of these
descriptions around the upstroke is not surprising. Detecting a point around the
64
upstroke involves often differentiation, or multiplication of the wave by ω, to enhance
the high frequency components of the wave. Thus, the ideal candidate of a pre-defined
point becomes the part of the wave where high frequency components are rich or where
phase velocities are constant. These predictions have been verified by measurements of
pulse wave velocities which are made by the techniques described above (i-iv) (Chiu et
al 1991).
2.3 APPLICATION OF TUBULAR MODELS OF THE ARTERIAL SYSTEM
2.3.1 Single Tube Models
The simplest application of transmission line concept not surprisingly involves a single
tube. This approximation was first used by Taylor to validate predictions of Womersley
theory (Taylor 1957a, b). It was later utilised in modelling of the arterial system. The
maxima and minima of harmonic components of the pressure waveform are already
studied in the arterial tree of the dog (McDonald 1974). These points are then used in
estimation of the reflection site by drawing analogies to nodes and antinodes present in a
single tube. Similarly, the minima and maxima of the impedance modulus at various
arterial segments and beds were also investigated through modelling (Taylor 1966b) and
experimentation (O'Rourke and Taylor 1967a). Although this analogue was initially
useful, the theoretical predictions of the input impedance spectrum for single tube
deviated from in vivo measurements. It was further shown that the model parameters
needed to fit for a given impedance spectrum might not have any physiological meaning
(Patel et al 1963, Sipkema and Westerhof 1975, Campbell et al 1989). However, this
model still finds use in assessing gross functional parameters from the impedance
65
spectra obtained in various ways, yet the attempts are directed to more sophisticated
models (Nichols and O'Rourke 1990).
2.3.2 Two Tube Models
The presence of two minima in the input impedance spectrum that are separated by
frequency doubling rather than tripling, led to models incorporating two tubes in parallel
(T-Tube) or in series.
Tubes in Parallel
The T-tube model introduced by O'Rourke and Taylor (1967a, b) attempts to reconcile
anatomical structure, body size and distribution of body mass with the ascending aortic
impedance spectra. According to this model, the heart, being eccentrically placed, is
connected to two parallel circulations, one to the head and the upper limbs and the other
perfusing the lower body and the lower limbs (O'Rourke 1967b). This hypothesis is
supported by the occlusion of the descending aorta and the brachiocephalic artery of the
dog. When the former is occluded the ascending aortic input impedance was found
similar to that of the brachiocephalic arteries. In contrast, when the brachiocephalic
artery was occluded it was similar to that of the descending aortas. Furthermore, flow
pulses in the brachiocephalic artery and descending aorta showed reciprocal oscillations,
indicating waves travelling to and from these arteries (Mills et al 1970). This kind of
interpretation coupled with aortic impedance data implies that reflections from these
two functionally separate circulations cancel each other and present a more uniform load
to the heart. However, the model had two weaknesses. Firstly, the aortic input
impedance spectra was very oscillatory . Although smaller than the single tube, this
oscillatory behaviour could not be dampened out with physiological viscous forces
unless complex loads at the terminations were introduced (Sipkema and Westerhof
66
1975). The second and probably the most important drawback is that the model
parameters cannot be estimated without recordings made in the ascending aorta (Liu et
al 1989; Burattini 1989a; Burattini et al 1989b). In this respect, this model can only be
thought of as a framework.
Tubes in Series
Two tubes in series were first proposed by Wetterer (1968) to explain the ascending
aortic impedance spectrum. According to this model, a tube representing the thoracic
aorta is terminated by another tube representing the femoral bed. The model was
revived in a study done by Latham et al (1985), where the model found use in
explaining the wave shapes along the aorta. In this model, the positions of the junction
of these two tubes are calculated to be at or below the renal arteries while the second
tube terminates at around the aortic bifurcation.
2.3.3 Three or More Tube Models
Increasing the number of tubes has been shown to yield better results even for a few
extra branches (McIlroy et al 1986). The models involving more branches obviously
exhibit an improved description of the behaviour of the arterial system, since they
acknowledge the presence of multiple reflection sites in the arterial system. In that
respect, the models with more elements explain different shapes of pressure waves in
different arteries, but they introduce extra difficulties associated with estimating the
physical parameters of individual tubes. These models are, however, superior in taking
into account the arrangement or morphology of the connections that might be important
in deciding the outcome.
67
Random Branching Models
These models are implemented in assessing the impedance and transmission
characteristics due to variation of various parameters. Taylor (1966a, b) was the first to
implement this kind of model in a digital computer, which later became a template for
subsequent investigations. In this model, it was shown that other than the non-
uniformity of the arteries, the scattered terminations reduced the reflected waves. As
Womersley (1955a) predicted, the terminations of random lengths behave like a single
unit because the wave length becomes comparable to the segment length, especially at
higher frequencies. The generated phase differences led to cancellation of reflected
waves at higher frequencies. The model was successful in predicting realistic input
impedance spectra, representative of an arterial bed. It was concluded that the
architecture of the arterial system has an important role in determining the characteristic
shape of aortic impedance.
Anatomically Arranged Branches
Tree Structure
When Taylor's random branching model (1966a, 1966b) was modified to allow for the
architecture of the arterial system, it gave representative values for ascending aortic
impedance (Avolio 1980) and for wave transmission (Salotto et al 1986). Both being
derivations of Taylor's original model, they viewed the arterial system as a continuously
branching tree of parallel elements. This simplified the solution, since the
implementation on a digital computer was relatively easy but tedious. It required the
input of many physical parameters for individual elements and a specification of their
interconnection in a matrix form.
In these models, parameters such as vessel diameters and lengths, wall properties and
input wave shapes, had to be given externally and were obtained from the literature
68
(Noordergraaf 1969; Westerhof et al 1969) and from Roentgenographic recordings
(Avolio 1976a). When applied to different species (Avolio et al 1976b, 1983, 1984a)
and for intra-aortic balloon pumps (Avolio 1976a), the models predicted the
experimentally observed impedance spectra. This does seem surprising because these
models do not take into account branching angles and inlet lengths. Peripheral
resistance values measured in vivo could not be estimated accurately probably due to
finite number of elements in the models. In an Avolio model there were 128 branches.
For simulation purposes peripheral resistance values were obtained from known values
of cardiac output and mean pressure. The models were then used to simulate ageing by
changes in pulse wave velocity values. Being multi-branched, the models predicted
realistic pressure shapes along the arterial tree. This was achieved for large arteries
(Avolio 1980) and in small arteries (Salotto et al 1986). It seems that the computational
procedure, as described by Taylor (1966a, b), is valid for estimating the vascular
parameters. This procedure, thus is useful in investigating the regional pressure and flow
patterns in different parts of the body.
Mesh Structure
The computational technique for tree structures cannot be directly applied to model
loops and grafts. To overcome this problem, an alternative computational method was
suggested involving 2-Port networks (Helal et al 1990). In this method the equations of
pressure and flow transfer functions are rewritten.
Considering the pressure, Pu, and flow, Qu, waves were input to the two-port, the
output Pressure, Pd, and flow, Qd, is related to input as
Γ+
Γ+=
−
1
ll
dueePP
γγ
(2.150)
and
69
d
dT Q
PZ = (2.151)
combining 2.154 and 2.93 yields
Γ =−+
P Z QP Z Q
d d
d d
0
0 (2.152)
Using 2.152 and 2.150 gives
( ) ( )lSinhZQlCoshPP ddu γγ 0+= (2.153)
and replacing Γ with -Γ for flow relations gives
( ) ( )lCoshQlSinhZPQ d
du γγ +=
0
(2.154)
Equations 2.153 and 2.154 represent a set of linear equations with unknowns Pd and
Qd. Then the solution of proximal waveforms can be calculated for the downstream
properties along a given path (Helal et al 1990). This kind of representation eliminates
one of the steps normally needed (backward iteration for impedance and forward for
transmission ratios) in Taylor's computational method. The model based on this
computational scheme predicted pressure patterns in the Circle of Willis in line with in
vivo recordings. The arterial models involving species that have a confluence (rather
than branching) of large vessels, such as reptiles seem best suited for this
scheme(Avolio et al 1982, 1983a).
70
MEASUREMENT OF ARTERIAL BLOOD PRESSURE Arterial blood pressure is one of the most fundamental physical quantities related to
haemodynamics. Because of the perceived need to quantify pressure levels throughout
the cardiovascular system, measurement techniques have been developed which, at
times have obscured the true nature of this pulsatile phenomenon. In quantifying blood
pressure oscillations, these techniques emphasised the mean value, peak values or both.
Blood pressure is, however, a composite wave that undergoes significant changes in its
contour while travelling within the arterial system. Thus the term "blood pressure",
when used alone, does not express this wave nature.
Chapter 3
71
The measurement techniques are divided into direct registration of pressure wave by
invasive means and non-invasive methods where the pressure wave is obtained by
methods employing external transducers.
3.1 INVASIVE MEASUREMENTS
The first invasive measurement of blood pressure was done by Hales in 1733. By
inserting a glass tube into the femoral artery of a horse he observed the blood column to
reach "eight feet three inches perpendicular above the level of the left ventricle". He
further observed that "when it was at its full height (the blood column), it would rise and
fall at and after each pulse two, three or four inches" (Geddes 1970). The former
observation then described as the mean pressure and the latter as the pulsatile pressure.
Since the proportion of the latter to the former was quite small, it was ignored. The
same phenomenon-was observed in Poiseuille's experiments in 1828, where he replaced
the blood column with a mercury column, hence the token of millimetres of mercury
(mmHg) as the unit of blood pressure. This observation led to the belief that the major
pressure in the arteries is the mean component and the pulsatile component can be
ignored, pressure oscillations were of course again attenuated by the inertia of the
mercury column. Quantification of the pulsatile component was first proposed by Carl
Ludwig, who attached a smoked drum to the mercury manometer using a cantilever
mechanism in 1847. Using this graphic recorder (kymograph), Ludwig was convinced
that the true systolic and diastolic pressure values were different from the recorded ones,
and he attributed this discrepancy to the mass of the system he employed. Later
attempts were directed to employing mechanical systems to transmit the pressure wave
to the kymograph rapidly. The most remarkable of these was the one developed by
Marey (1860), in which the instrument was so refined that it was possible to record the
cardiac pressures (Geddes 1970). The introduction of fluid to transmit pressures, since
72
fluid is incompressible and so it is relatively easy to build devices around it, was first
suggested by Rolleston in 1887. The theoretical problems related with this mode of
transmission were treated by Otto Frank in 1903 who proposed an index or figure of
merit (Güte) of the manometers by solving second order differential equations
describing the operation of manometers (Wiggers 1928). Figure of merit stated that the
manometers should be designed with high stiffness to mass ratio to accomplish a high
frequency response.
The implementation of Frank's recommendations on manometers increased the
resolution necessary to distinguish the features of the pressure waveform (Wiggers
1928). It further accelerated the investigation on the formation of the pressure pulse
(Hamilton 1944; Wood et al 1951; Remington and Wood 1956). The introduction of
mechano-electric transducers of capacitance type by Schutz in 1937, of inactive
transducer type Wetterer in 1943 and of strain-gauge type by Lambert in 1947, not only
increased the frequency characteristics but also improved the amplification and
recording stages of the measurement system (Geddes 1970).
Since all these systems use incompressible fluid as a medium for wave transmission, the
equations of transmission line apply to the description of operation of fluid filled
catheters. This is especially important when transmission time cannot be neglected,
otherwise it is quite possible to use the system with a single degree of freedom as first
described by Frank. In the latter the behaviour of the system is related to the
parameters: damping coefficient, β , and natural frequency, ωn , (McDonald 1974)
ERl
n 20
πρ
ω = (3.1)
73
El
R πρµβ 0
3
4= (3.2)
and the amplitude at frequency Ω = ω ωn to its true value is
( ) 2222 41
1
Ω+Ω−=
βA (3.3)
and the phase is
Ω−Ω
= −2
1
12βϕ Tan (3.4)
Expressions for resonant frequency 3.1 and damping coefficient 3.2 can be combined
with Moens-Korteweg equation to yield similar expressions
lc
n 42 0π
ω = (3.5)
20
16Rc
lµβ = (3.6)
Obviously, these equations are only valid when the termination of the measuring tube is
totally closed (Gabe 1972), otherwise the system contains more than one degree of
freedom and equations 3.1 through 3.6 do not hold.
3.2 NON-INVASIVE MEASUREMENTS
The most common non-invasive methods of blood pressure measurement give either
peak values and/or mean value of the pressure wave or contour of the pressure wave
74
(uncalibrated or with some form of indirect calibration) or both. All of these techniques
involve the arterial wall as a prime transmitting medium and thus prone to errors due to
wide variations in vascular, muscular and subcutaneous tissue properties of different
individuals.
3.2.1 Peak Systolic Value Measurement
In these measurement methods the information content of the pressure wave is assumed
to be contained only in the peak values and/or mean value of the waveform. They often
include compression of an artery and surrounding tissue with a cuff filled with air. The
pressure inside the cuff is then raised and subsequently decreased. During this cycle,
blood flow distal to the cuff or oscillations in the cuff or the sounds generated by the
non-linear oscillations of the wall and of the fluid are monitored. These signals are then
related to the pressure inside the cuff. The cuff pressure at the onset of these signals is
often used as the systolic, diastolic and mean pressures. The validation of the methods
is done by comparing the pressures obtained by the technique and the measurements
done invasively. Detailed histories of different methods can be found in Geddes (1970).
However, because of its importance in clinical medicine the auscultatory methods will
be briefly described here.
Auscultatory Method
The sounds generated by the non-linear behaviour of the collapsible tubes (Conrad
1969) were first described by Korotkoff in 1905. This was followed by extensive
research into the phenomenon-(Rodbard et al 1953; Ur and Gordon 1970) and its
validation (Van Bergen et al 1954, Borow et al 1982). Since its introduction,
sphygmomanometry has remained virtually unchanged. The American Heart
Association proposed five revisions for measurement methods in the last forty years.
75
These were concerns over the size of the cuff and whether the phase IV or phase V of
the Korotkoff sounds represented the diastolic pressure. Research has also been
performed on the establishing the relation between values obtained manually and by
microprocessor controlled devices. Yet, neither the pressure wave amplification in the
upper limb nor the generations of sounds that depends on arterial elasticity, heart rate,
distance to the central artery is addressed in these recommendations. Furthermore, the
explanation of the variation of the differences between invasive and non-invasive blood
pressure measurement in different individuals is still lacking.
3.2.2 Contour Measurement
Sphygmography
Since the arterial pulse had been described in many ways qualitatively, (in ancient
Sanskrit medicine there were 100 different descriptions), the graphic display of the
arterial pulse had created a large interest in the 19th Century. The devices developed in
that era were all mechanical and non-invasive. They were required to be applied
externally and be pressed firmly on an artery, often the radial. Etienne Jules Marey was
first to record arterial pulse accurately in man (Schechter 1969). The sphygmographs he
developed were taken up by Mahomed (1874), Broadbent (1890) and Mackenzie (1902).
They made changes in the original design leading to improvements in amplification and
conditioning of the obtained pressure pulses (Geddes 1970). However, because of the
inherent problems associated with these instruments such as high frequency attenuation
and high inertia brought by air or fluid filled capsules, they could not record the pressure
waveform with sufficient fidelity (O'Rourke et al 1992). Instead, these instruments find
use in detecting the presence of pulse and its rhythm.
76
Plethysmography
Plethysmography is another technique to register the pressure wave contour. The
change in vessel dimension caused by pressure pulse is detected by strain gauges, photo-
electric sensors or electrical impedance electrodes. Assuming a linear relation between
blood pressure and vessel dimension then yields a pressure contour. Although widely
used because of ease and relative low cost, the non-linear relation between the pressure
and diameter of the vessel often invalidates the technique. The pulse waves recorded
are not pressure waves.
A technique introduced by Penaz (1973) initially described by Wiederhelm (O'Rourke et
al 1992) tries to overcome this difficulty introduced by non-linearity of the stress strain
relation of the wall. It relies on the fact that, when the arterial wall is unloaded at zero
transmural pressure, the wall becomes totally compliant and thus exhibits the maximum
strain per unit stress. When the wall is clamped at this maximum strain level by
adjusting the external pressure at all times, it is possible to obtain the pressure in the
artery (Yamakoshi et al 1980). In effect, this procedure is not much different from
calculation of instantaneous pressure using a linear pressure volume relation. However,
in this technique, the non-linear pressure-volume curve of the vessel is estimated first
and subsequently used to calculate instantaneous pressure by measured instantaneous
volume.
The instrument first designed by Penaz and subsequently developed by Wesseling et al
(1982) is marketed by OHMEDA as a device known as "FINAPRES". This device first
"learns" the pressure-volume curve of the digital artery of the middle finger. It measures
the photoplethysmographic volume of the finger while increasing the pressure in steps
within a small cuff surrounding the finger. After detecting the maximum volume
oscillations in the photoplethysmographic recordings, it then uses a servo driven
pressure pump to clamp finger arterial volume at this maximum value. The pressure in
77
the cuff, needed to achieve this, is then considered to be the intra-arterial pressure in the
finger artery. The technique claims to yield calibrated pressure waveforms, but it
definitely fails when the compliance curve cannot be estimated or changes actively
during measurement (Wesseling 1985).
Applanation Tonometry
The applanation tonometry technique relies on the elimination of vessel-wall curvature
with a known force, thus transferring the normal stresses to the applanating element.
The theoretical basis for the technique has solid foundations (O'Rourke et al 1992)
based on a curved beam theory (Drzewiecki and Noordergraaf 1980). The intra-arterial
pressure waveform has been shown to be registered faithfully provided that curvature
changes introduced by the tonometer are small with respect to the radius and
axisymmetric (Drzewiecki et al 1984). A probe has been designed based on these
model predictions (Hayward 1989). The probe has a Millar micromanometer in its tip
that operates on the principle of piezo resistivity. Application of force on to the active
element on this tip, bends this cantilever element which changes its resistance. The
element forms one arm of a Wheatstone bridge, where this unbalanced resistance is
detected and generated voltage is amplified. The design of the tonometer as a pencil
probe, facilitates the applanation of an artery that can be readily palpated. The pressure
sensitive area on the probe is 0.5 mm2 and sits on an area of the probe tip of 133 mm2.
This construction allows most of the arteries being applanated and the force being
detected quite accurately which is then converted to pressure. The validation of this
tonometer (Kelly et al 1989b) displayed a very high accuracy and fidelity compared to
the intra-arterial pressure waveform. The frequency response of the system was more
than 1 kHz and had a sensitivity of 0.2 volts/mmHg when used with its own
preamplifier boxes (Millar TCB-500).
78
SECTION II
MODELLING AND SIMULATION
The in vivo estimation of important parameters in the arterial system is often hampered
by the type of instrumentation and methodology used in experiments. The noise present
in the experimental set up often blunts the responses, causing repeat of experimentation
and increases the cost. The introduction of digital computers into the laboratory
environment, however, changed this entirely. The simulation techniques now available
enable experimenters to perform the experiments with a reduction in time spent on real
experiments. Various modifications to the computer model can be done repeatedly in
considerably shorter time. The experimental findings can then be used to confirm the
predictions made by the model, enabling the adjustments on the model parameters. All
these assertions are valid, provided that the model is close to the simulated experimental
phenomena. However, the model fails in reproducing events if it does not use the
physical constraints and formulations of real life situations. Deviation from reality
causes models to converge to an arbitrary solution, if any, very slowly. The strict
79
adherence to the reality, on the other extreme, increases computation time significantly
with a concomitant requirement for exact duplication of boundary conditions. This
approach then approximates the system under investigation, nullifying the benefit
gained from modelling.
The properties of the arterial system are traditionally described by pressure-flow
relations in the frequency domain. However, pressure wave propagation parameters can
also be used to describe the same arterial properties. Quantification of pressure-pressure
relations in the frequency domain has an undisputed advantage over the pressure-flow
relations. Instead of being restricted by the technique of flow measurements, one can
use pressure measurements to quantify the gross properties of the arterial system. The
pressure wave recordings can also be made non-invasively. With the advent of non-
invasive pressure wave registration with high-fidelity tonometers (Kelly et al 1989b),
limitation of measurement technique is becoming less restricted. Tonometry is best
suited for arteries that can be applanated against a rigid structure such as the carotid
artery, radial artery, femoral artery and dorsalis pedis artery. These sites can also be
used to obtain Doppler-derived flow profiles, thus enabling determination of pressure-
flow relations. However, since the flow measurement techniques are not as accurate as
those of pressure (Nichols and O'Rourke 1990), this technique is used less frequently.
This fact imposes certain limitations in reality. Firstly, the number of measurements or
known variables often exceed the number of equations that wholly describe pressure-
flow relations. Secondly, the global nature of wave propagation phenomena in the
arterial system deviates from local properties. For example, any change in arterial
properties in the lower body can change the pressure waves recorded in an upper limb
artery (Karamanoglu et al 1990). The introduction of modelling to the system
overcomes these two problems, given that the model accurately represents the
underlying physical phenomena. Using this model, one can then estimate spatial and
temporal distribution of pressure and flow waves, solving the second problem. The first
80
problem is however more complex to solve. One needs to explore all the possible
solutions of the equations by successively assuming different values for one or more
equation. This technique is commonly known also as parameter estimation methods.
Using a heuristic approach, then the model can be optimised for some parameter sets,
which can then be assumed to be the system parameters set.
In this thesis, modelling of the arterial system is limited to the following investigations:
i) To reproduce documented changes in the arterial pulse due to ageing,
vasodilatation and vasoconstriction in humans.
ii) To explain the differences between central and peripheral, upper limb pressure
pulses under wide variety of conditions in humans.
iii) To suggest ways of analysing pressure waveforms under clinical conditions, for
diagnostic and therapeutic purposes in humans.
iv) To suggest ways to monitor central haemodynamic events from peripheral
recordings in humans.
The above limitations assert that the model implemented has to take into account the
distributed nature of the arterial system. The emphasis in this model has to be the
contribution of physical quantities on the formation of the pulse. It should also take into
account the changes induced by ageing, vasoactive drugs and the anatomical differences
between individuals. The effect of local as opposed to general properties of the vessels,
has to be treated separately.
The model that satisfies these criteria was implemented in a computer first by Taylor
(1966a, 1966b) as random branching networks. This work was subsequently extended
81
to the mammalian arterial system by Avolio (1980) who obtained reasonable agreement
with real measurements. Although both were satisfactory, these investigators did not
extend their simulations to the problems of pressure waveform genesis and propagation
in clinical situations. Taylor's model suffered from the fact that the parameters selected
bore no resemblance to the ones observed in the real human arterial system. For
example, although the progressive increase in the stiffness of the arteries was present,
the extent and variation were not taken into account; also no attempt was made to
simulate anatomical branching patterns. The Avolio model was considerably enhanced,
since it took into account the anatomical distribution of arterial segments and utilised
reasonable physical properties for the segments. However, it also suffered from the lack
of realistic alteration of these properties due to ageing and vasoactivation. These data
became available recently (Ho 1982; Avolio et al 1983b; Latham et al 1985). Both
models also suffered from their applicability to the clinical environment. Being
extensively dependent upon computer resources, they were unfortunately tied to
mainframe hardware. This meant that they could only be used in circumstances where
access to a mainframe was possible and this imposed severe limitations. For example,
the limited number of peripheral terminations due to computer storage limitations
imposed restrictions on how far along the arterial tree vascular impedance could be
faithfully determined. This however did not influence calculation of input impedance
and transmission along the aortic trunk.
Since computers are now regarded as another instrument in experimental physiology
thanks to recent developments in microprocessor design and implementation, they are in
abundance. This fact and the others described above, necessitated the implementation
of these modelling algorithms in a much more widely used computer platform while
taking into account the vital data exchange methods between the existing laboratory
equipment. It also required that the implementation be done in a higher level language
that is supported by contemporary hardware and software. Finally, the model has to be
designed and implemented in a user friendly way (difficult-to-use systems often
82
discourage use) and enabling the end user, most likely the physician, to be the ultimate
judge.
83
MODEL CONSTRUCTION Several models of the human circulation have been suggested previously. These models
were either linear or non-linear, lumped or distributed, implemented in hardware or in
software. The model of Avolio (1980) was the best to be chosen to represent the human
arterial system for this investigation. This model is chosen because it is not only a
distributed model, representing the architecture of the human arterial tree with its
morphology and physical properties of the arterial segments, but also because it is based
on linear theory, allowing one to use techniques such as spectral analysis and
transmission line theory on pressure and flow waves. This advantage brought by linear
approximation outweighs any disadvantage due to restrictions of this model, since not
only the non-linear behaviour of the elastic properties is small within the physiological
range of frequencies and pressures but also the tapering, which introduces convective
non-linear terms in the Navier-Stokes equations, is considered to be negligible due to
side branches (Chapter 2).
Chapter 4
84
The basic unit in the Avolio model is an arterial segment that was represented by a
viscoelastic, thin and tethered tube of radius r, length l and wall thickness h. The
solutions of the linearised Navier-Stokes equations for fluid flow in this tube are as
described by Womersley. Description of these equations, assumptions made during
derivation and limitations of the theory verified by experimentation are already
described in detail (see Chapter 2). Furthermore, the uniform lossless transmission line
analogue is assumed to hold for representation of this tube. The concepts and
implications that are derived from this analogue have also been discussed previously
(Chapter 2).
4.1 REPRESENTATION OF THE ARTERIAL TREE
Any vascular tree, excluding anastomoses and grafts, can be represented as a binary tree
(B-tree) since branches are likely to bifurcate (Zamir 1978). B-tree's are finite element
sets which empty or contain a single element called the root of the tree and whose
remaining elements are partitioned into two diagonal subsets, each of which is itself a
binary tree (Figure 4.1). These two subsets are called the left and right subtrees of the
original tree. Each element of a binary tree is called a node of the tree. The B-Tree in
Figure 4.1 consists of nine nodes with A as the root. Its left subtree is rooted at B and
right subtree is rooted at C. Here, node A is the parent of nodes B and C. B and C is in
turn daughters of node A and parents of D, E and F respectively. The nodes D, G, H
and I have no descendants and hence are called the leaves.
For the arterial bed displayed in Figure 4.1, the coding technique is such that, a unique
key is assigned to each element for the path it takes to reach the node. This is achieved
by assigning the root of the entire tree, A in this case, code M, short for Main. If one
85
E
I J
F
H
D
G
CB
A
Figure 4.1 A representative binary tree composed of nine nodes and four generations.
then travels to the Left a letter L is appended to the previous code and for Right a letter
R is appended. Thus, for example, node E in this scheme of coding becomes MLR,
which designates a departure from main, turn left then right. Obviously this coding
technique does not take into account the functionality of the branching but the
morphology. It is therefore different from the Strahler system where the coding starts
from the leaves upstream taking into account the diameters of the parent-daughter pairs
(Yen et al 1984). The advantage of the B-tree, however, is that it provides a simple
description of the generations and facilitates computations based on generations.
4.1.1 Parent Daughter Relations
Zero Order Relation
In the zeroth order relation a parent does not have any offspring, thus it is a leaf. The
algorithm for calculation of leaf impedance takes into account, for this special case a
reflection coefficient. Since it is assumed that for practical purposes the vascular bed
86
extending from this leaf cannot be described for its individual elements physical
properties and morphologies, it can be lumped as a modified Windkessel (Figure 4.2).
If the parent characteristic impedance is Zp0 then the impedance mismatch between the
two elements, proximal element and the modified Windkessel can be expressed as
Γww po
w po
Z ZZ Z
=−+
(4.1)
where Zw is the Windkessel impedance and can be described as
Z Z Rj R Cw d
d
d d= +
+0 1 ω (4.2)
Rd and Cd are defined as the Poiseuille resistance and capacitance per unit length
respectively and given by equations 2.77 and 2.78.
Distal ElementProximal Element
Zdo
Cd
Zpo
Rd
Figure 4.2 The modified windkessel representing the termination at the leaf level.
Although there are different approaches to the modelling of the vessel wall
(Langewouters et al 1984, Elad et al 1988), it is often considered that the distal bed
pressure, Pd, is related to bed volume, Vd, in a non-linear fashion with an exponential
law (Kawasaki et al 1987)
87
P AedBVd= (4.3)
Assuming fixed oxygen requirements of the distal bed, thus constant blood flow,
pressure, Pd, is related to peripheral resistance, Rd, as
P Q Rd d d= (4.4)
Where A, B and flow (Qd) are constants. Then from (4.3) and later by (4.4)
dVdP BP
d
d d=
1 (4.5)
dVdP BQ R
d
d d d=
1 (4.6)
and since compliance is also defined as Cd=dVd/dPd then
R CBQd d
d= =
1 Constant (4.7)
Thus the time constant , τ=RdCd, of the modified Windkessel will be fixed under :
(i) The existence of an exponential pressure-volume relationship of the vessel wall
and
(ii) The fixed oxygen requirements of the arterial bed downstream (ie constant blood
flow).
By combining equation 4.1 with equation 4.2 and assumingZ Zp d0 0= for reasons of
continuity, Γw could be expressed as a function of ω (Figure 4.3).
88
CdRdτ=
R (R +2Z )d d do
Figure 4.3. The frequency dependent behaviour of reflection coefficient when the terminal element is a Windkessel representing the distal arterial bed. The reflection coefficient in this arrangement is dependent on frequency of which is determined by the windkessel time constant τ.
In this new expression, Γw approaches ( )dodd ZRR 2/ + as ω approaches zero while as ω
approaches infinity Γw approaches zero. From equation 4.1 and equation 4.2 it is seen
that the frequency dependent behaviour of Γw is determined by the time constant RdCd
as given by equation 4.7. Unique parameters, Γw and τ, set this way, used in this
termination mode to describe the lumped vascular properties downstream.
First Order Relation
In the first order representation one of the offspring of the parent does not exist. The
terminal impedance then becomes the remaining daughter's input impedance.
Second Order Relation
Both of the daughters exist. In this case the terminal impedance becomes the parallel
addition of the offspring impedances.
89
4.1.2 Computational Algorithm
In an assembly of a binary tree, the computations are possible for impedances, for
description of pressure-flow relations, and for transmission ratios for pressure-pressure
relations. This can be achieved by utilising the three possible configurations of parent-
daughter relations as described in 4.1.1. The computational algorithm for calculation of
the transmission properties and impedances of the entire B-tree becomes a
straightforward implementation of the following recursive procedure in Pascal language
(Algorithm 4.1)
Algorithm 4.1 Pseudo code for calculation of the arterial tree as viewed from source.
The coding for this configuration is M, MR, ML, MLR, MLL where ML is the dummy
element of the length ε, where 0<ε<<1 cm, ie ε is a finite length close to zero. For
example, using this model a long tube can be expressed as elements of the tree of
MRRRRRR indicating no diatomic branching (a single tube). In this respect, the code
for a conventional T-Tube representation is M, MR and ML. Using this nomenclature
of expressing the nodes, a branching structure of three elements from a single root can
be specified by inserting a dummy branch into the sequence. An example of such a
configuration is shown in Figure 4.4.
Procedure Calculate Tube Downstream(Node) Begin If Node is not Leaf then Begin Calculate Tube (Left Subtree of the Node) Calculate Tube (Right Subtree of the Node) End Calculate Terminal Impedance (Sisters) Use 4.1.1, 4.1.2 and 4.1.3 Calculate Characteristic Impedance Use equation 2.125 Calculate Propagation Coefficient Use equation 2.129 Calculate Reflection Coefficient Use equation 2.93 Calculate Input Impedance Use equation 2.127 Calculate Transmission Ratio Use equation 2.133 END
90
DummyElement
M
M
MR
ML
MLRMLL
Figure 4.4 A parent with three daughters, a possible representation of renal arterial branching of the abdominal aorta (left). The equivalent B-Tree representation of the branching node after inserting a dummy element of length ε, where 0<ε<<1 cm (right).
This kind of representation does not take into account the bifurcation angles and inlet
lengths, since they are not relevant parameters in this type of simulation. Apart from
these concerns, arithmetic operations and algorithmic calculations are continuous.
Algorithm 4.2 Pseudo code for calculation of arterial tree as seen from an arbitrary node.
The above computational algorithm calculates the input impedance downstream. If,
however, the calculation of upstream impedances and transmission ratios are required,
the calculation order has to be modified, which is given in Algorithm 4.2. This
Procedure Calculate Tube Upstream (Node) Begin If Node is not Leaf Begin Calculate Tube Downstream (Sister of the Node) Use Algorithm 4.1 Calculate Tube Upstream (Parent) End Calculate Terminal Impedance (Sister and Parent) Use 4.1.1, 4.1.2 and 4.1.3 Calculate Characteristic Impedance Use equation 2.125 Calculate Propagation Coefficient Use equation 2.129 Calculate Reflection Coefficient Use equation 2.93 Calculate Input Impedance Use equation 2.127 Calculate Transmission Ratio Use equation 2.133 End
91
obviously necessitates the insertion of a reflection coefficient for the main root,
ascending aorta in this case, indicating the source impedance to the B-tree.
4.2 ANATOMICAL DATA
The anatomical data for the human arterial system is obtained from the literature
(Avolio 1976a, Haimovici 1981). The elastance data, however, is found to yield a
relatively slow pulse wave velocity for the various segments. The data for elastances
has been altered to obtain the reported values in humans as well as changes induced by
ageing for various segments (McDonald 1968; Avolio et al 1983b). Using interpolated
data for twenty years of age, the pulse wave velocity in the trunk was found to be 6.2
m/sec, in the upper limbs 8.4 and in the lower limbs 10.9 m/sec. For the age of 60
years, the respective values obtained were 10.2 m/sec, 10.6 m/sec and 14.9 m/sec.
Modifications have also been made to relate the wall viscosity to the smooth muscle
content of the vessel wall. This has been achieved by relating the phase of the wall
viscosity, θ, to the wall thickness to diameter ratio, h/2R, by a function of the form
θ θ= 0 2hR
(4.8)
where the parameter θ0 is assumed to be 10 degrees. Since the vessel wall becomes
thicker towards the periphery, this formula assumes an increase of the viscous losses per
unit diameter length.
The overall data of the constructed human arterial segments are given in Table 4.1 with
their B-tree codes. A representative arterial tree constructed from this model is given in
Figure 4.5.
92
30 Cm
Figure 4.5. The model of the human arterial system employed in the thesis (prepared by Miss Lina L. Lee from data given in Table 4.1). Arterial segments are drawn to scale both for length and for diameter. Total number of elements are 142 of which 14 are dummy elements of length ε (see text for more information).
93
B-Tree Code Name L
(cm)
D
(cm)
h
(cm)
E
dyne.cm-2
(x106)
Yng Old
θ
(deg)
MLLR SUBCLAVIAN ARTERY(L) 6.80 0.80 0.07 7.28 10.19 7.00 MLLRL (D) SUBCLAVIAN 3 (L) 0.10 0.80 0.07 8.17 11.44 7.00 MLLRR (D) SUBCLAVIAN 1 (L) 0.10 0.80 0.07 8.17 11.44 7.00 MLLRLL INTERNAL MAMMARY (L) 15.00 0.20 0.03 10.40 14.56 12.00 MLLRLR VERTEBRAL ARTERY (L) 14.80 0.19 0.05 10.40 14.56 21.05 MLLRRL COSTO-CERVICAL (L) 5.00 0.40 0.05 10.40 14.56 10.00 MLLRRR (D) SUBCLAVIAN 2 (L) 0.10 0.80 0.07 8.17 11.44 7.00 MLLRRLR SUPRASCAPULAR (L) 10.00 0.20 0.02 10.40 14.56 8.00 MLLRRRL THYROCERVICAL (L) 5.00 0.20 0.03 10.40 14.56 12.00 MLLRRRR AXILLARY ARTERY (L) 6.10 0.72 0.06 8.58 12.01 6.67 MLLRRRRL THOROCO-ACROMIAL(L) 3.00 0.30 0.04 20.80 29.12 10.67 MLLRRRRR AXILLARY ARTERY (L) 5.60 0.62 0.06 8.58 12.01 7.74 MLLRRRRRL SUBSCAPULAR (L) 8.00 0.30 0.04 20.80 29.12 10.67 MLLRRRRRR BRACHIAL ARTERY (L) 6.30 0.56 0.05 9.74 13.64 7.14 MLLRRRRRLL CIRCUMPLEX SCAPULAR (L) 5.00 0.20 0.03 20.80 29.12 12.00 MLLRRRRRRL PROFUNDA BRACHII (L) 15.00 0.30 0.04 10.40 14.56 10.67 MLLRRRRRRR BRACHIAL ARTERY (L) 6.30 0.52 0.05 9.55 13.37 7.69 MLLRRRRRRRL SUP. ULNAR COLATERAL(L) 5.00 0.14 0.02 20.80 29.12 11.43 MLLRRRRRRRR BRACHIAL ARTERY (L) 6.30 0.50 0.05 9.56 13.38 8.00 MLLRRRRRRRRL INF. ULNAR COLATERAL(L) 5.00 0.12 0.02 20.80 29.12 13.33 MLLRRRRRRRRR BRACHIAL ARTERY (L) 4.60 0.48 0.05 9.73 13.62 8.33 MLLRRRRRRRRRL ULNAR ARTERY (L) 6.70 0.42 0.05 10.40 14.56 9.52 MLLRRRRRRRRRR RADIAL ARTERY (L) 11.70 0.32 0.04 10.40 14.56 10.00 MLLRRRRRRRRRLL ULNAR ARTERY (L) 8.50 0.38 0.05 10.40 14.56 10.53 MLLRRRRRRRRRLR INTEROSSEA ARTERY (L) 7.90 0.18 0.03 20.80 29.12 13.33 MLLRRRRRRRRRRR RADIAL ARTERY (L) 11.70 0.32 0.04 11.55 16.17 10.00 MLLRRRRRRRRRLLL ULNAR ARTERY (L) 8.50 0.38 0.05 11.14 15.60 10.53 MRR SUBCLAVIAN ARTERY(R) 6.80 0.80 0.07 7.28 10.19 7.00 MRRL (D) SUBCLAVIAN 3 (R) 0.10 0.80 0.07 8.17 11.44 7.00 MRRR (D) SUBCLAVIAN 1 (R) 0.10 0.80 0.07 8.17 11.44 7.00 MRRLL INTERNAL MAMMARY (R) 15.00 0.20 0.03 10.40 14.56 12.00 MRRLR VERTEBRAL ARTERY (R) 14.80 0.19 0.05 10.40 14.56 21.05 MRRRL COSTO-CERVICAL (R) 5.00 0.40 0.05 10.40 14.56 10.00 MRRRR (D) SUBCLAVIAN 2 (R) 0.10 0.80 0.07 8.17 11.44 7.00 MRRRLR SUPRASCAPULAR (R) 10.00 0.20 0.02 10.40 14.56 8.00 MRRRRL THYROCERVICAL (R) 5.00 0.20 0.03 10.40 14.56 12.00 MRRRRR AXILLARY ARTERY (R) 6.10 0.72 0.06 8.58 12.01 6.67 MRRRRRL THOROCO-ACROMIAL(R) 3.00 0.30 0.04 20.80 29.12 10.67 MRRRRRR AXILLARY ARTERY (R) 5.60 0.62 0.06 8.58 12.01 7.74 MRRRRRRL SUBSCAPULAR (R) 8.00 0.30 0.04 20.80 29.12 10.67 MRRRRRRR BRACHIAL ARTERY (R) 6.30 0.56 0.05 9.74 13.64 7.14 MRRRRRRLL CIRCUMPLEX SCAPULAR (R) 5.00 0.20 0.03 20.80 29.12 12.00 MRRRRRRRL PROFUNDA BRACHII (R) 15.00 0.30 0.04 10.40 14.56 10.67 MRRRRRRRR BRACHIAL ARTERY (R) 6.30 0.52 0.05 9.55 13.37 7.69 MRRRRRRRRL SUP. ULNAR COLATERAL(R) 5.00 0.14 0.02 20.80 29.12 11.43 MRRRRRRRRR BRACHIAL ARTERY (R) 6.30 0.50 0.05 9.56 13.38 8.00 MRRRRRRRRRL INF. ULNAR COLATERAL(R) 5.00 0.12 0.02 20.80 29.12 13.33 MRRRRRRRRRR BRACHIAL ARTERY (R) 4.60 0.48 0.05 9.73 13.62 8.33 MRRRRRRRRRRL ULNAR ARTERY (R) 6.70 0.42 0.05 10.40 14.56 9.52 MRRRRRRRRRRR RADIAL ARTERY (R) 11.70 0.32 0.04 10.40 14.56 10.00 MRRRRRRRRRRLL ULNAR ARTERY (R) 8.50 0.38 0.05 10.40 14.56 10.53
94
B-Tree Code Name L
(cm)
D
(cm)
h
(cm)
E
dyne.cm-2
(x106)
Yng Old
θ
(deg)
MRRRRRRRRRRLR INTEROSSEA ARTERY (R) 7.90 0.18 0.03 20.80 29.12 13.33 MRRRRRRRRRRRR RADIAL ARTERY (R) 11.70 0.32 0.04 11.55 16.17 10.00 MRRRRRRRRRRLLL ULNAR ARTERY (R) 8.50 0.38 0.05 11.14 15.60 10.53 MRL COMMON CAROTID (R) 8.90 0.74 0.06 7.28 10.19 6.49 MRLL COMMON CAROTID (R) 8.90 0.74 0.06 7.28 10.19 6.49 MRLLL CAROTID EXTERNAL (R) 5.90 0.36 0.05 10.40 14.56 11.11 MRLLR CAROTID INTERNAL (R) 11.80 0.30 0.04 10.40 14.56 10.67 MRLLLL SUPERIOR THYROID (R) 4.00 0.14 0.02 10.40 14.56 11.43 MRLLLR EXTERNAL CAROTID (R) 5.90 0.26 0.04 10.40 14.56 12.31 MRLLRL MIDDLE CEREBRAL (R) 3.00 0.12 0.02 20.80 29.12 13.33 MRLLRR CEREBRAL ARTERY (R) 5.90 0.16 0.03 20.80 29.12 15.00 MRLLLRL FACIAL ARTERY (R) 4.00 0.20 0.03 20.80 29.12 12.00 MRLLLRR EXTERNAL ARTERY (R) 5.90 0.16 0.03 20.80 29.12 15.00 MRLLRRR OPTHALMIC ARTERY (R) 3.00 0.14 0.02 20.80 29.12 11.43 MRLLLRLL LINGUAL ARTERY (R) 3.00 0.20 0.03 10.40 14.56 12.00 MRLLLRRL SUPERFICIAL TEMPORAL(R) 4.00 0.12 0.02 20.80 29.12 13.33 MRLLLRRR MAXILLIARY ARTERY (R) 5.00 0.14 0.02 20.80 29.12 11.43 MLR COMMON CAROTID (L) 8.90 0.74 0.06 7.28 10.19 6.49 MLRR COMMON CAROTID (L) 8.90 0.74 0.06 7.28 10.19 6.49 MLRRR COMMON CAROTID (L) 3.10 0.74 0.06 7.28 10.19 6.49 MLRRRL CAROTID INTERNAL (L) 11.80 0.30 0.04 10.40 14.56 10.67 MLRRRR CAROTID EXTERNAL (L) 5.90 0.36 0.05 10.40 14.56 11.11 MLRRRLL CEREBRAL ARTERY (L) 5.90 0.16 0.03 20.80 29.12 15.00 MLRRRLR MIDDLE CEREBRAL (L) 3.00 0.12 0.02 20.80 29.12 13.33 MLRRRRL EXTERNAL CAROTID (L) 5.90 0.26 0.04 10.40 14.56 12.31 MLRRRRR SUPERIOR THYROID (L) 4.00 0.14 0.02 10.40 14.56 11.43 MLRRRLLL OPTHALMIC ARTERY (L) 3.00 0.14 0.02 20.80 29.12 11.43 MLRRRRLL EXTERNAL ARTERY (L) 5.90 0.16 0.03 20.80 29.12 15.00 MLRRRRLR FACIAL ARTERY (L) 4.00 0.20 0.03 20.80 29.12 12.00 MLRRRRLLL MAXILLIARY ARTERY (L) 5.00 0.14 0.02 20.80 29.12 11.43 MLRRRRLLR SUPERFICIAL TEMPORAL(L) 4.00 0.12 0.02 20.80 29.12 13.33 MLRRRRLRR LINGUAL ARTERY (L) 3.00 0.20 0.03 10.40 14.56 12.00 M A.AORTA 4.00 2.90 0.16 6.58 13.29 4.41 ML AORTIC ARCH 2.00 2.24 0.13 6.98 14.43 4.64 MR BRACHIOCEPHALIC 3.40 1.24 0.09 7.28 11.65 5.81 MLL AORTIC ARCH 1.90 2.14 0.13 7.41 15.85 4.86 MLLL DESCENDING AORTA 7.20 2.00 0.12 7.41 15.85 4.80 MLLLL THORACIC AORTA 5.20 1.90 0.12 7.54 16.13 5.05 MLLLLL THORACIC AORTA 5.20 1.90 0.12 7.54 16.79 5.05 MLLLLLL COELIAC ARTERY 1.00 0.78 0.06 7.54 16.13 6.15 MLLLLLR ABDOMINAL AORTA 5.30 1.74 0.11 7.54 17.47 5.06 MLLLLLLL GASTRIC ARTERY 7.10 0.36 0.05 7.54 16.13 11.11 MLLLLLLR (D) COELIAC 1 0.10 0.78 0.06 7.54 16.13 6.15 MLLLLLRL (D) ABDOMINAL 1 0.10 1.74 0.11 7.54 17.47 5.06 MLLLLLRR SUPERIOR MESENTRIC 5.90 0.86 0.07 7.54 16.13 6.51 MLLLLLLRL SPLENIC ARTERY 6.30 0.56 0.05 7.54 16.13 7.14 MLLLLLLRR (D) COMMON HEPATIC 0.10 0.44 0.06 7.54 16.13 10.91 MLLLLLRLL RENAL ARTERY (L) 3.20 0.52 0.05 7.54 16.13 7.69 MLLLLLRLR (D) ABDOMINAL 2 0.10 1.74 0.11 7.54 17.47 5.06 MLLLLLLRRL HEPATIC ARTERY 6.60 0.44 0.05 7.54 16.13 9.09 MLLLLLLRRR GASTRIC ARTERY 3.20 0.52 0.05 7.54 16.13 7.69
95
B-Tree Code Name L
(cm)
D
(cm)
h
(cm)
E
dyne.cm-2
(x106)
Yng Old
θ
(deg)
MLLLLLRLRL ABDOMINAL AORTA 5.30 1.14 0.08 8.32 17.80 5.61 MLLLLLRLRR RENAL ARTERY (R) 3.20 0.52 0.05 7.54 16.13 7.69 MLLLLLRLRLL ABDOMINAL AORTA 5.30 1.14 0.08 8.58 18.35 5.61 MLLLLLRLRLR INFERIOR MESENTRIC 5.00 0.32 0.04 8.58 18.35 10.00 MLLLLLRLRLLL COMMON ILIAC (L) 5.80 1.04 0.08 9.24 17.52 6.15 MLLLLLRLRLLLL EXTERNAL ILIAC (L) 8.30 0.58 0.05 8.51 15.67 6.90 MLLLLLRLRLLLR INTERNAL ILIAC (L) 5.00 0.40 0.04 13.28 20.93 8.00 MLLLLLRLRLLLLL EXTERNAL ILIAC (L) 6.10 0.54 0.05 8.10 14.59 7.41 MLLLLLRLRLLLLLL FEMORAL ARTERY (L) 12.70 0.48 0.05 7.55 13.02 8.33 MLLLLLRLRLLLLLR PROFUNDIS ARTERY(L) 12.60 0.46 0.05 13.28 12.60 8.70 MLLLLLRLRLLLLLLL FEMORAL ARTERY (L) 12.70 0.46 0.05 7.40 12.75 8.70 MLLLLLRLRLLLLLLLL POPLITEAL ARTERY (L) 9.40 0.40 0.05 6.54 11.07 10.00 MLLLLLRLRLLLLLLLLL POPLITEAL ARTERY (L) 9.40 0.40 0.05 6.97 10.98 10.00 MLLLLLRLRLLLLLLLLLL ANTERIOR TIBIAL (L) 2.50 0.26 0.04 5.63 8.87 12.31 MLLLLLRLRLLLLLLLLLR POSTERIOR TIBIAL (L) 16.10 0.36 0.05 6.24 9.83 11.11 MLLLLLRLRLLLLLLLLLLL ANTERIOR TIBIAL (L) 15.00 0.20 0.04 4.67 6.87 16.00 MLLLLLRLRLLLLLLLLLLR PERONEAL ARTERY (L) 15.90 0.16 0.02 7.70 12.14 10.00 MLLLLLRLRLLLLLLLLLRR POSTERIOR TIBIAL (L) 16.10 0.36 0.05 6.68 10.53 11.11 MLLLLLRLRLLLLLLLLLLLL ANTERIOR TIBIAL (L) 14.00 0.20 0.04 5.03 7.93 16.00 MLLLLLRLRLLLLLLLLLLRR PERONEAL ARTERY (L) 14.90 0.16 0.02 9.39 14.80 10.00 MLLLLLRLRLLLLLLLLLLLLL (D) ANTERIOR TIBIAL (L) 0.1 0.20 0.04 5.03 7.93 16.00 MLLLLLRLRLLLLLLLLLLRRR (D) PERONEAL ARTERY (L) 0.1 0.16 0.02 9.39 14.80 10.00 MLLLLLRLRLLR COMMON ILIAC (R) 5.80 1.04 0.08 9.24 14.56 6.15 MLLLLLRLRLLRL EXTERNAL ILIAC (R) 8.30 0.58 0.05 8.51 13.41 6.90 MLLLLLRLRLLRR INTERNAL ILIAC (R) 5.00 0.40 0.04 13.28 20.93 8.00 MLLLLLRLRLLRLL EXTERNAL ILIAC (R) 6.10 0.54 0.05 8.10 12.77 7.41 MLLLLLRLRLLRLLL FEMORAL ARTERY (R) 12.70 0.48 0.05 7.55 11.90 8.33 MLLLLLRLRLLRLLR PROFUNDIS ARTERY(R) 12.60 0.46 0.05 13.28 20.93 8.70 MLLLLLRLRLLRLLLL FEMORAL ARTERY (R) 12.70 0.46 0.05 7.40 11.66 8.70 MLLLLLRLRLLRLLLLL POPLITEAL ARTERY (R) 9.40 0.40 0.05 6.54 10.31 10.00 MLLLLLRLRLLRLLLLLL POPLITEAL ARTERY (R) 9.40 0.40 0.05 6.97 10.98 10.00 MLLLLLRLRLLRLLLLLLL ANTERIOR TIBIAL (R) 2.50 0.26 0.04 5.63 8.87 12.31 MLLLLLRLRLLRLLLLLLR POSTERIOR TIBIAL (R) 16.10 0.36 0.05 6.24 9.83 11.11 MLLLLLRLRLLRLLLLLLLL ANTERIOR TIBIAL (R) 15.00 0.20 0.04 4.67 7.36 16.00 MLLLLLRLRLLRLLLLLLLR PERONEAL ARTERY (R) 15.90 0.16 0.02 7.70 12.14 10.00 MLLLLLRLRLLRLLLLLLRR POSTERIOR TIBIAL (R) 16.10 0.36 0.05 6.68 10.53 11.11 MLLLLLRLRLLRLLLLLLLLL ANTERIOR TIBIAL (R) 14.00 0.20 0.04 5.03 7.93 16.00 MLLLLLRLRLLRLLLLLLLRR PERONEAL ARTERY (R) 14.90 0.16 0.02 9.39 14.80 10.00 MLLLLLRLRLLRLLLLLLLLLL (D) ANTERIOR TIBIAL (R) 0.1 0.20 0.04 5.03 7.93 16.00 MLLLLLRLRLLRLLLLLLLRRR (D) PERONEAL ARTERY (R) 0.1 0.16 0.02 9.39 14.80 10.00
TABLE 4.1 Physical data for human arterial system together with B-Tree codes for major arteries. The dummy elements are marked by (D) in the B-Tree code. The name field codes (L) and (R) indicate left and right extremities while L and D are the length and diameter of the elements. Wall properties, h (wall thickness), E (wall elastance) and θ(phase of wall viscosity) are also given for young (yng) and old.
96
A BINARY TREE MODEL OF THE HUMAN ARTERIAL SYSTEM The alteration of wave shapes in the human arterial system due to ageing, vasoactive
drugs and physical manoeuvres are well investigated (Nichols and O'Rourke 1990). The
consensus on these phenomena is that the presence of wave reflection can help explain
these observations (O'Rourke et al 1992). However, this concept was challenged by
investigators who question the location of the sites of reflection or the presence of re-
reflection (Papagiorgiou and Jones 1988; Ting et al 1990). There is also scepticism of
the explanations provided by this theory on differential action of drugs (Kelly et al
1990b; Simkus and Fitchett 1990).
These concerns cannot be answered readily. Firstly there are the experimental
difficulties in introducing a single reflected wave and in following its retrograde
CHAPTER 5
97
transmission along the arterial system. Secondly, even if this could be achieved, the
numerous branches of the arterial system present interferences to the propagated wave
that would hamper the interpretation of the results. Yet, modelling of the arterial system
can assist in addressing these issues. By selective alteration of the physical properties of
different branches reflected waves can be induced and spatial and temporal relationship
could be obtained. This information allows the specific origin of wave reflection from
in vivo recordings to be determined. It is even possible to devise experiments that
specifically target these origins to verify model predictions.
In this chapter, the multi-branching B-tree model (Chapter 4) is extended to simulate the
branching structure of the human arterial tree. By using this multi-branching network,
the transmission and reflection of pressure waves are investigated. To address the role
of wave reflection on the arterial pulse, the B-tree model is subdivided into anatomically
separable compartments (upper limb, trunk and lower limb). Influence of reflected
waves originating from these compartments is then studied. By mapping the pressure
waves along a path that extends through these compartments, the distribution of
reflected waves is visualised and interpreted.
Since the pressure pulse in the upper limb is significantly important in clinical settings,
the investigations are extended to the upper limb. In this simulation the relation
between the upper limb pulse and the central aortic pulse is studied in the frequency
domain by transfer function analysis. The properties of the upper limb arterial system
(elastance, reflection coefficient, time constant and wall viscosity) are altered and their
influence on the transfer function and the wave shapes are also investigated.
98
5.1 PRESSURE WAVE PROPAGATION IN THE HUMAN ARTERIAL MODEL
In the simulations that follow, the wave propagation path used for investigation extends
from right arm (radial- brachial- axillary-subclavian arteries) to the trunk (ascending-
descending- thoracic- abdominal aorta) and then to the right leg (iliac-femoral-popliteal-
anterior tibial arteries), Figure 5.1. All side branches that are supplied by these arteries
are considered to be the beds specific to these compartments.
30 Cm
Figure 5.1 The path in which the propagation of waves are simulated. It extends from right arm (radial- brachial- axillary- subclavian arteries) to trunk (ascending- descending- thoracic- abdominal aorta) and then to the right leg (iliac-femoral-popliteal-anterior tibial arteries).
5.1.1 Pressure Contour Maps
To facilitate the interpretation of the model findings, data maps of pressure waves were
generated. These maps are formed by stacking the pressure waves, which are also
functions of time, in their spatial distance from the ascending aorta or from the input
(Figure 5.2).
99
Figure 5.2 3-D Pressure wave propagation map showing the spatial and temporal distribution of pressures along the path. The generated pressure wave in ascending aorta travels to the upper limb, to the trunk and the lower limbs. The left most wave is that which is recorded in the radial artery while the right most that in the anterior tibial artery. The pressure wave peaks due to wave reflection during its travel along the path.
The maps obtained this way is called the pressure propagation maps in three
dimensions, P(P, z, t). When grided appropriately and interpolated, the surface
obtained yields the pressure value at a given location at a given time. Using this
method, one can obtain a visual expression of wave propagation and wave reflection.
Slicing the 3-D propagation maps with the pressure axis at different pressures, Pi, and
projecting these slices onto z-t plane, results in isobaric contour maps, similar to
pressure maps in weather forecasting or to topographical mapping of heights, Figure 5.3.
These maps are useful in estimation of wave propagation properties, characteristic
impedance and wave reflections. Pulse wave velocities, for example, can be estimated
along the constant Pi lines, by calculating dz/dt on these contour maps. Since the
reflected waves will travel in negative space but positive time, their presence will be
indicated by negative pulse wave velocities. Wave fronts in these maps will be
indicated by closely spaced Pi with respect to time. The strong impedance mismatches
will appear as closely spaced Pi lines along the constant z axis.
100
ISOBARIC MAP
RADIAL
A.AORTA
ILIACBIFURCATION
ANTERIORTIBIAL
TIME (milliseconds)
Figure 5.3 Isobaric map of Figure 5.2 showing the spatial and temporal distribution of pressure waves along the path. The pressure waves at the same level (3 mmHg) are joined by continuous lines. Closely spaced lines indicate a steep rise or fall of the pressure values (wavefronts). The generated pressure wave at the level of the aortic root travels along the path (arrows) both to upper limbs (down arrow) and to trunk and lower limbs (up arrow). The travel speed of the wave is determined by the pulse wave velocities of branches given by the Moens-Korteweg equation. The calculated values at different anatomical locations using this equation indicated by the heavy dark line. Deviations of the direction of wavefront from this line indicate the presence of reflected waves.
Figure 5.3, corresponding to Figure 5.2, can be deciphered precisely by inspection. The
left most contours are the lowest pressures in the system. As one moves right, with
increasing time, the contours become closely spaced at first, denoting the wave front,
then spaced approximately evenly apart. In the latter part of the map, the contours
gradually denote decreased pressure. In this figure, every contour is at steps of 3
mmHg. These are stacked together to form a hill on the 3-D map. At the anterior tibial
level, a summit occurs at around 260 milliseconds that can also be seen in the 3-D map.
The wave velocities here can be calculated by taking partial derivatives of the contours
with respect to time and space. This procedure produces similar values of wave
101
velocities as indicated by the similarity between the solid line in Figure 5.3 and the wave
front which is determined from Figure 5.4. Any wave front contour in this map which
matches with the wave velocity indicates a forward-travelling contour. When it
decreases in speed, it denotes a reflected wave is present in the waveform as seen in the
hump in the right of Figure 5.2.
5.1.2 Introduction of Reflected Waves
Wave reflection is a consequence of the impedance mismatch between the source and
the load. The alterations of the values of load impedances can be used to generate
reflected waves given that the source impedance remains constant. The alteration of
terminal impedances at the leaf levels of the B-tree can be used to generate reflected
waves originating from the terminations. If these waves are made to propagate
upstream, without altering the properties of the upstream branches, they will appear in
more proximal branches after being summed with waves from different origins. The
changes in the terminal impedances at the leaf levels can be thought of as changes in the
vascular lumen area (wall thickness to diameter ratio) or wall elastance (changes in wall
material properties) or both. Instead of specifying the type of parameter change, one can
define a reflection coefficient to lump all these factors into one. Previous investigations
have suggested that the reflections often originate from the arteriolar level and have
values of 0.85 under normal conditions (O'Rourke and Taylor 1966). Vasodilatation
decreases it to a lower value. However, since the model employed in this model could
not be extended to the arteriolar level branching, a lumped approach has been devised
which takes into account the numerous branches downstream. Presence of such a
configuration ensures a smooth decline in reflection coefficient modulus at higher
frequencies due to cancellation of reflected waves (Taylor 1966b). To simulate this
behaviour of the reflection coefficient, the leaves are assumed to be terminated with
modified Windkessels (see chapter 4).
102
5.1.3 Tracking Of Reflected Waves
Injection of a retrograde wave at each termination separately and following its upstream
travel is probably the most instructive method of all. Yet, for practical purposes,
extension of this to every branch in the model is tedious. However, one can lump
reflected waves originating from the terminal branches downstream and follow the
travel of this composite wave upstream. To obtain this composite reflected wave one
can make use of the superposition principle (the system is linear). It is known that when
the reflection coefficient is zero at the terminations, no reflections will be generated
from these leaves. However, when it is different from zero reflected waves are
generated. One can then take the difference of these two situations to obtain the net
effect of introduced positive reflection coefficient, hence the composite reflected wave.
Extending this logic to the entire arterial tree, a track of reflected waves can be obtained
by allowing certain branches to have zero and non-zero reflection coefficients at the
terminations and subtracting them. When displayed in a map form, the spatial and
temporal distribution will then be the track of reflected waves. The distribution of
reflected waves along a segment in this study is obtained by subtracting a surface where
terminations are altered from a control surface. The remaining surface can be expressed
as for a single tube
( )acfPP Γ−Γ=∆ (5.1)
Where Pf, forward wave, Γc and Γa are the control and altered reflection coefficients
respectively. If, for special case Γa =0 then
∆ ΓP Pf c= (5.2)
Thus one can estimate the intensity of reflected waves by inducing total vasodilatation
and subtracting it from the control conditions. Since this procedure requires a reflection
103
coefficient of absolute zero to be introduced at the terminations, it is only suitable in
modelling studies where this condition is fulfilled.
5.1.4 Wave Velocities and Reflection Coefficients
The wave velocities for these compartments were adjusted to simulate pulse wave
propagation in the arterial system of young and old human subjects. The values are
taken from published data (Ho 1982; Avolio et al 1983b, 1985) for age 20 and for 60
years for the mean values of the compartments (see chapter 4). The values from the
model for the spatial distribution are given in figure 5.4 for the young (Open Circles)
and for the old (Closed Circles). The terminal reflection coefficients of the
compartments were selected to be 0.0 for vasodilatation 0.85 for normal state and 0.95
for vasoconstriction. The time constant of the modified Windkessels (see chapter 4)
representing the leaves was chosen to be 250 milliseconds in each case (Liu et al 1989).
Figure 5.4 The pulse wave velocities of the model elements that are calculated using Moens-Korteweg equation. The simulation values for the young (Closed Circles) and for the old (Open Circles) show a relative stiffening of the peripheral branches. The mean values of upper limb, trunk and lower limb segments are: 10.9 m/sec, 8.0 m/sec and 9.0 m/sec respectively for the young simulation. Corresponding values for the old simulation are 11.7 m/sec, 12.9 m/sec and 11.3 m/sec.
104
5.1.5 Input to the Model
In generation of pressure waveforms a digitised flow wave from Nichols et al (1987)
was used. This wave is considered to have an ejection duration of 330 milliseconds at
heart rate of 72 beats/minute (Figure 5.5). The peak flow was scaled to yield a pulse
pressure of 40 mmHg for early systole (up to the first shoulder; Murgo et al 1980b).
Due to finite number of branches, the model could not reproduce the actual values of
peripheral resistances (see chapter 4). Therefore, a mean blood pressure of 100 mmHg
was assumed to exist at every branch of the arterial tree.
Figure 5.5. The flow waveform used as input at the ascending aorta in the model to generate pressure waveforms throughout the branching system. The peak value of the flow wave is normalised to yield a pulse pressure of 40 mmHg between end diastole and the first systolic shoulder. Flow waveform is digitised from Nichols et al (1987).
5.1.6 Input Impedance
The calculated input impedance of the arterial tree at the main root, ascending aorta,
simulating conditions of ageing (young, left; old, right), control (Γc =0.85),
vasodilatation (Γc =0.00) and vasoconstriction (Γc =0.95) are displayed in figure 5.6.
105
When compared to previous publications of the ascending aortic impedance (Murgo et
al 1980b; Nichols and O'Rourke 1990) there was a close agreement between experiment
and simulated impedance pattern. As suggested previously (O'Rourke and Avolio 1980)
ageing shifts the spectrum to the right (early wave reflection) with concomitant increase
in low frequency modulus (increased reflections). Increase in terminal reflection
coefficients (vasoconstriction) increases both the modulus and the phase delay at the
low frequency components. The opposite is observed with the decrease in terminal
reflection coefficients (vasodilatation).
ASCENDING AORTIC IMPEDANCE
YOUNG OLD
Figure 5.6 Ascending aortic input impedance of the arterial tree displaying simulations for the young (Left) and for the old (Right) model. Calculated values for control (Γc = 0.85, Circles), Vasodilatation (Γc = 0.0, Squares) and vasoconstriction (Γc = 0.95, Triangles) is also shown for each case.
106
5.1.7 Pressure Wave Transmission
The spatial and temporal distribution of the pressure waves is shown in figure 5.7. The
pressure pulse amplitude increases as waves travel peripherally due to increased wave
reflection. This increase in wave reflection is due to increases in wave velocity and
getting closer to the terminations. The presence of a second systolic hump that spans
the entire length of the path in the older subject's 3-D map can be made more visible
when plotted in isobaric map format.
YOUNG OLD
Figure 5.7 The 3-D pressure wave propagation map (Top) together with Isobaric contour map (Bottom) in the arterial tree for the case of control (Γc =0.85). Each map
is similar to that in Figures 5.2 and 5.3 and calculated for the young (Left) and for the old (Right).
107
5.1.8 Reflections From Compartments
Figure 5.8 displays the propagation of waves reflected from upper limb in both 3-D map
and isobaric map format. These maps are generated by assigning zero for terminal
reflection coefficients of the upper limb and subtracting the maps obtained from the
ones in Figure 5.7.
UPPER LIMBS
YOUNG OLD
Figure 5.8 The 3-D pressure wave propagation map (Top) together with Isobaric contour map (Bottom) in the arterial tree for the case of upper limb reflections only. Each map is similar to that in Figures 5.2 and 5.3 and calculated for the young (Left) and for the old (Right).
Compared to Figure 5.7, the terminations are the major source of reflections. Waves
travelling in the retrograde direction cross the ascending aorta to spread into the rest of
108
the circulation after being heavily attenuated. The impedance of the trunk and leg
compartments as seen from the upper limb are lower than the upper limb impedance,
thus the branch becomes negatively matched. This reverses of sign of reflection
coefficient at the junction of ascending aortic-brachiocephalic branch. Aging increases
the amount of reflections by apparent shortening of the upper limb vessels, thus
decreasing absolute attenuation in retrograde direction.
LOWER LIMBS
YOUNG OLD
Figure 5.9 The 3-D pressure wave propagation map (Top) together with Isobaric contour map (Bottom) in the arterial tree for the case of lower limb reflections only. Each map is similar to that in Figures 5.2 and 5.3 and calculated for the young (Left) and for the old (Right).
Reflections in the lower limbs have the same characteristics as those described for the
upper limb and shown in Figure 5.9. The phenomenon-of negative mismatch of the
109
limb-trunk junction also occurs here as seen from the figure. The waves appear not to
be able to propagate beyond the junction due to negative wave reflection. Aging
increase these reflected waves, due to increased wave velocity.
TRUNK
YOUNG OLD
Figure 5.10 The 3-D pressure wave propagation map (Top) together with Isobaric contour map (Bottom) in the arterial tree for the case of trunk reflections only. Each map is similar to that in Figures 5.2 and 5.3 and calculated for the young (Left) and for the old (Right). Note the increased amplitude seen in the upper limb and in ascending aorta when reflections are originating from the trunk compared to those when reflected waves are originated from periphery as seen in Figure 5.8 and 5.9.
The reflections from the trunk and their propagation are shown in Figure 5.10. The
reflections have a spatial origin around the abdominal aorta (30 cm from the ascending
aorta) and immediately after the wave front. They travel into both compartments (upper
limbs and lower limbs) without much attenuation. During this travel they also appear in
110
the ascending aortic waveform, boosting late systolic pressure. The same increase in
systolic pressure does not happen in the upper and lower limbs, due to blunting effect of
local reflections that are more prominent than these secondary waves. Aging does not
affect the origin of these waves but increase their intensity.
5.2. PRESSURE WAVE PROPAGATION IN THE UPPER LIMB
The modulus of the pressure transfer function between the ascending aorta and radial
artery displays a characteristic peaking (figure 5.11, top left). It has a DC (zero
frequency) value of unity rising to a peak value of 2.55 at 3.5 Hz. It then decreases
steadily with increasing frequency to attain a value of 0.65 at higher harmonics. The
negative phase of the transfer function (figure 5.11, bottom left) indicates the delay
between the frequency components of waves in the ascending aorta and the radial artery.
It starts from a DC value of zero to have an inflection point at around the peak of the
modulus. After subsequent oscillations it reaches an asymptotic value at higher
frequencies indicating a constant group delay.
Similar to second order system, this behaviour was described in relation to models of
the upper limb (Warner 1957; Gardner 1982). The upper limb often is likened to a
catheter system having a resonant frequency of 4 Hz and a damping coefficient of 0.3
(Gardner 1982). In these models however, wave propagation is not taken into account
although its presence is evident from the phase pattern of the transfer function. In a
typical second order system, modulus has a single peak with attenuation of 40 dB/octave
at higher frequencies. In the model constructed above, it is clear that there is more than
one peak and a logarithmic attenuation. The transfer function displays secondary
oscillations on the descending limb of the modulus and it never reaches zero value, even
111
asymptotically, at higher values. It rather approaches a value other than zero, around
0.5. On the other hand, phase of a second order system has an asymptotic value of -π at
higher frequencies which is not valid for this case. The phase never attains a value of -
π/2 at the peak modulus (break) frequency (Oppenheim et al 1983).
HAA RA−
Figure 5.11 The simulation results obtained in the arterial model of a "Young" subject showing the pressure waveform contours (right) and the transfer function (HAA RA− , left)
between ascending aorta () and the radial artery (•−•). The reflection coefficient, Γ, is 0.85 and time constant, τ, is 250 milliseconds at the terminations of the upper limb. The modulus of HAA RA− (top left ) peaks at 3.5 Hz exceeding unity to a value of 2.55
indicating amplification of the pressure wave. Note the oscillations in the phase values which settle to a constant value dictated by wave speed (bottom, left). The phase delay presents itself as time difference between the ascending aortic and radial pressure waves
The pressure waveform in the radial artery differs markedly from that in the central
aorta. Its peak is accentuated and the systolic pressure exceeds that in the central aortic
one due to reflected waves (Figure 5.11, Right). The foot of the wave is delayed due to
112
finite wave velocity. The radial diastolic pressure at the foot of the wave is very close to
that in the aorta.
5.2.1 Effect of Change in Wall Elastance (E)
The effect of increase wave velocity (to simulate the age-related increase in arterial
stiffness) obtained by increase in elastance is to move the peak of transfer function
modulus to the right (Figure 5.12, top left).
HAA RA−
Figure 5.12 The effects of stiffening of the upper limb arteries on the transfer function (HAA RA− , left) and on the radial pressure waveform contours (right). The average pulse
wave velocities in the upper limb is varied from 10.9 to 15.3 m/sec in increments of 10%. The reflection coefficient, Γ, is 0.85 and time constant, τ, is 250 milliseconds at the terminations of the upper limb. Increase in elastance shifts the transfer function to the right with concomitant decrease in the phase delay (arrows). The general pressure pulse shape, however, is little effected.
113
However, because of viscous attenuation due to wall and blood viscosity, the peaking
becomes less prominent. The phase becomes less negative with increasing frequency.
The pressure wave in the upper limb becomes steeper and the foot becomes less delayed
(figure 5.12, top right). The systolic pressure in the radial artery is not effected by these
changes.
5.2.2 Effect of Change in Reflection Coefficient (Γ)
When reflection coefficient decreases from 0.95 (vasoconstriction) to 0.85 (normal
vasoconstriction) then to 0.5 (vasodilatation) and to 0.0 (total vasodilatation), the
peaking of the transfer function modulus becomes less prominent (figure 5.13, top left).
Although pulse wave velocities were similar in each case, the frequency at which the
peak occurs gradually shifts left which can also be interpreted as decrease in pulse wave
velocity.(Figure 5.13, bottom left).
However, the asymptotic value of phase delay does not support this interpretation. The
zero reflection coefficient, that is a perfect impedance match at the terminations,
abolishes the peaking of transfer function modulus. The oscillations present in the
phase of the transfer function decreases with decreases in reflection coefficient.
Corresponding changes in the wave shape indicate that most of the energy in the
reflected waves is contained in the systolic peak of the pressure pulse (Figure 5.13, top
right). With the decrease in reflection coefficient, the pressure pulse becomes less and
less amplified approaching to the same value of ascending aorta at zero reflection
coefficient. Pulse pressure decreases and so does the systolic pressure. However, the
diastolic pressure does not change as much indicating that reflections are the major
cause of systolic peaking. Although it can be argued that presence of augmented
systolic pressure without simultaneous decrease in diastolic pressure should result in
increase in energy in the pulse, close inspection of the descending limb of the pressure
114
wave reveals equal but opposite fluctuation in the pressure wave, thus fulfilling the
criterion that wave reflection does not augment the net energy content of the pulse.
HAA RA−
Figure 5.13 The changes in transfer function between ascending aorta and the radial artery (Left) and radial pressure waveform contour (Right) caused by changes in reflection coefficient. Decrease in reflection coefficient, Γ, from 0.95 (), 0.85 (•••), 0.50 (−−−) and to 0.0 (•−•) is accompanied by reductions in the pressure pulse amplification evident from the initial peak and the systolic pressure. In each case asymptotic values of phase delay were similar yet the peak amplification frequency decreased. The superimposed oscillations on the phase also decrease, indicating reductions in reflections.
5.2.3 Effect of Changes in Time Constant (τ)
The Windkessel time constant, τ, for the terminal load is varied from 0.6 seconds to
zero in steps of 0.2 seconds (Figure 5.14). This was done to simulate the decreased
compliance of small arteries at the terminations without arteriolar vasoconstriction. As
expected, when the time constant is zero (no capacitance) the load becomes purely
115
resistive and thus introduces the highest amount in peaking (figure 5.14, top left). With
increase in the time constant however, the peak in the modulus becomes less and the
peak point shifts to the left. The phase becomes less oscillatory, without altering its
asymptotic behaviour. The pressure waves at the peripheral site, corresponding to this
simulation, display attainment of higher systolic pressures with decreases in time
constant (figure 5.14, top right). These findings also support the view that decreases in
compliance of the small arteries for a given vasoactive state actually increases the
reflected waves.
HAA RA−
Figure 5.14. The changes in transfer function between ascending aorta and the radial artery (Left) and radial pressure waveform contour (Right) due to changes in Windkessel time constant. Increase in time constant, τ, from 0.0 second () to 0.2 second (•••) and to 0.4 second (---) to 0.6 second (−•−) is accompanied by reductions in the amplification. In each case asymptotic values of phase delay were similar yet the peak amplification frequency decreases. The superimposed oscillation on the phase decreases indicating reductions in reflections.
116
5.2.4 Effect of Change in Wall Viscosity (Θo)
To simulate the changes in viscous losses in the arterial wall, wall viscosity Θo is varied
from 0o to 15o in steps of 5o (Figure 5.15). The resultant transfer function indicates
that viscosity effects are important in producing oscillations in the modulus as first
suggested by Taylor (1966b). Although there are little changes in the low frequency
components of the transfer function moduli and therefore on the radial pressure
waveform, higher frequencies are heavily attenuated. This indicates that under
physiological conditions, ie normal heart rates, the only important part of the transfer
function in determining the wave shape is to the left of the peak amplification.
HAA RA−
Figure 5.15. The changes in transfer function between ascending aorta and the radial artery (Left) and radial pressure waveform contour (Right) caused by changes in wall viscosity. Increase in time constant, Θ0, from 0.0 degrees () to 5 degrees (•••) and to 10 degrees (---) to 15 degrees (−•−) is accompanied by reductions in the amplification at higher frequencies. In each case value of phase delay were similar yet the peak amplification decreases without much alteration in the actual pulse pressures.
117
5.3 DISCUSSION
The human arterial system has been modelled in the past by various investigators
employing different techniques (see chapter 2). This study, however, is different from
those attempts, since it directs its attention to the propagation of the pressure waves
rather than pressure-flow relations. Although the mathematical and clinical implications
are the same, the former is readily applicable in clinical settings.
In this study, a technique of mapping the pressure wave propagation is also introduced.
Similar to the weather maps, this visualisation technique enhances the visual appeal of
otherwise complex data and does not suffer from lack of spatial information as present
in conventional graphs. Careful analysis of these maps provides information about the
wave velocities, reflection sites and the intensity of reflections. Coupled with
superposition principle, these maps can be useful in detection of reflected waves from
different compartments. Previously, the same information was obtainable by complex
analysis of the pressure-flow data (Nichols and O'Rourke 1990).
The findings of this study indicate that retrograde transmission of pressure waves from
the limbs is heavily attenuated. This supports the view that although reflected waves
become prominent when there is elastic non-uniformity, the retrograde transmission of
generated, reflected waves become heavily attenuated (Li et al 1984). Thus the arterial
system favours the forward transmission, tending to decouple the heart from the
periphery (Taylor 1964). The reflections from the limbs reach the ascending aorta often
heavily attenuated. However reflections originating the ones from the splanchnic
circulation are 30% or more of the forward travelling pulse. Thus, splanchnic
circulation being not only close to the ascending aorta, but also being favoured for
retrograde transmission plays an important role in determining ascending aortic
impedance. This assertion is supported by the blood flow data to those compartments.
It is well known that, under resting conditions, some 40% of the blood is used to perfuse
118
the organs around the abdominal aorta (McDonald 1974). One would expect, therefore,
a higher influence of the vasoactive state of this compartment on the overall
haemodynamics.
The limited effect of lower limb vasculature on human arterial pulse compared to
influence of the trunk seems surprising. A close inspection of the arterial model (Figure
4.4) reveals that there are not many branches in the lower limb circulation. This is
especially true for the proximal arteries of the lower limb circulation that perfuses the
muscles around the thigh. However, the method employed in the delineation of the
reflected waves does take this fact into account. The alteration of reflection coefficients
from 0.85 (control) to 0.00 (complete vasodilatation) eliminates any load presented to
the common iliac, thus simulating the total vasodilatation of the distal branches
irrespective of their numbers. Still, there was not enough changes in the ascending
aortic impedance spectrum (see below) to suggest reasonable effect of this manoeuvre.
It is possible that when values less than zero were used expected impedance changes
might be generated. However, there is evidence that the reflection coefficient at the
aorta iliac bifurcation never becomes less than zero even with maximal vasodilatation
(Li 1984).
This concept was further supported by changes in the aortic impedance spectrum caused
by alteration of reflection coefficient of the compartments. The influence of reflections
from splanchnic circulation is far greater than from both limbs (figure 5.16, Squares).
The effect of reflections from both limbs is minor compared to the splanchnic
circulation. This result of the simulation warrants an investigation of the splanchnic
circulation. Chapter 10 of this thesis describes a study in which this hypothesis is
verified. Vasodilatation of this circulation caused by ingestion of glucose, explained the
reduction in reflected waves in the central waveform, similar to simulation results.
119
ASCENDING AORTIC IMPEDANCE
YOUNG OLD
Figure 5.16 Ascending aortic input impedance of the arterial tree displaying simulations for the young (Left) and for the old (Right) model. Calculated values for control (Closed Circles), Vasodilatation of the upper limb (Open Circles), lower limbs (Triangles) and trunk (Squares) is also shown for each case.
These findings are further confirmed by superimposing the reflected waves originating
from each compartment (Figure 5.17). This technique enabled to isolate the path of
reflected waves which would otherwise be blunted by multiple reflections from different
compartments. Again, the ascending aortic and the upper limb pressure pulse is heavily
affected by the reflected waves from the trunk compartment.
When the pressure waves in the ascending aorta and in the radial artery is decomposed
into its components by separating the effects of reflections from different compartments,
it supports the view that the major reflecting site is in the trunk area (Figure 5.18, Top).
120
Time (milliSeconds)
0 200 400 600 800Radial
Ant.Tibial
Aortic Bifurcation
A.Aorta
Brachial
Renal Branches
Femoral
Figure 5.17 Isobaric map showing the spatial and temporal distribution of travelling pressure waves along the path given in Figure 5.1. Arrows indicate the direction of travel. After the initial wave (Heavy Dark Lines) which travels both to the upper limbs and the lower body it is reflected in four major reflection sites, radial termination, splanchnic branches, aortic termination, femoral bifurcation and the tibial termination, (Dark lines). These reflected waves are re-reflected during retrograde travel (Light Lines). Although the reflected waves from both limbs appear to be trapped by reciprocal reflections, reflected waves originating from the trunk travel in all directions. The influence of these waves on ascending aortic wave contour is more pronounced than those from the peripheral limbs.
Reflected waves originating from the trunk area travels into the ascending aorta and into
the upper limb compartment without much alteration in its amplitude. In these
simulations they formed the secondary oscillations present in both composite waves.
Although peak systolic pressure in the aortic pulse is augmented with these reflected
waves, the radial systolic pressure has not. In the radial waveform, reflections from the
upper limb compartment boosts the initial systolic upstroke blunting the reflected waves
121
ASC. AORTA RADIAL
TRUNK
LOWER LIMB
UPPER LIMB
COMPOSITE
Figure 5.18 Decomposition of pressure pulses in the ascending aorta (Left) and in the radial artery (Right) into reflected waves from different origins. (Top) The reflected waves originating from the trunk area augment the peak systolic pressure in the ascending aorta but appear in the radial artery pressure waveform as secondary oscillations. (Middle) Upper limb reflections are prominent and constitute the first peak in the radial pulse but have little influence on central aortic pressure pulse . Note the ripples in the reflected waves from the trunk and lower limb compartments indicating the presence of re-reflections. (Bottom) Reflections originating from the lower limb compartment have little effect on the wave shapes in both upper limb and in ascending aorta. (Dash dot lines indicate the instant of diastolic and peak systolic pressures in the composite wave while dotted lines indicate the level of mean pressure).
122
from the trunk (Figure 5.18, Middle). As expected, the effect of lower limb
compartment is minor and negligible in both pulses (Figure 5.18, Bottom). Peak
systolic pressure in ascending aorta and in radial artery is defined by reflections from
two different origins. While local reflections originating from the upper limb
terminations defines peak systolic pressure in the radial artery, reflections originating
from the trunk area are responsible for the same phenomenon in the ascending aorta. It
is therefore possible to alter the peak systolic pressures recorded from these two sites
independent from each other by changing the characteristics of these two origins.
Reflections from the trunk undergoes secondary reflections in the upper limb arteries as
evident from the oscillations in the composite waveforms. Similar superimposed
oscillations are also observed in the lower limb reflected waves.
A close inspection of the intensity and the timing of reflected waves originating from
the upper limb also reveals that the early systolic peak in the upper limb will often be
different from the one at central aorta (Figure 5.18). However, the diastolic and the
mean pressures in these locations are approximately equal. These phenomena, caused
by the timing of reflected waves can be exploited in calibrating waveforms in both
arteries. Since pressure waves obtained by arterial tonometry could not be calibrated
with the current techniques, use of pressure waves obtained by other means, such as
invasive radial artery waveforms, enable these waveforms to be calibrated. This
technique involves equating the diastolic and mean pressures in both waveforms,
invasive and tonometric, to obtain a linear scale which would then be used to calibrate
the tonometric waveforms. In Section III this technique is employed to calibrate carotid
and radial tonometric waveforms.
In the latter part of the simulation it was shown that the pressure wave is amplified
during its travel along the upper limb. Although it has been known for a long time
(Warner 1957), this study is first to investigate the relevant determinants of this
amplification. As opposed to earlier attempts, in this study it was shown that the upper
123
limb is different from second order system analogue, due to wave travel. The second
order system approximation can be justified if the wave velocity is high enough (Figure
5.12). The increase in wave velocity shifts the peak of the transfer function to the right
and decreases the amplification. Thus, second order approximation becomes feasible to
apply.
Amongst all the physical parameters investigated, the profound effect is observed when
the terminal reflection coefficient is varied. It generated the most radical changes in the
transfer function and thus the peripheral radial wave shapes. Yet the influence was not
linear and was little for changes from 0.5 to zero but substantial for changes from 0.95
to 0.80. The time constant, however, is linear in its effect to the transfer function thus to
the radial pressure wave. Since there is a direct relation between the time constant and
the reflection coefficient (chapter 4) for a given terminal compliance, this suggests a
stable transfer function under these conditions.
From a practical point of view, the data obtained for changes in wall viscosity indicate
that although there can be substantial changes in the high frequency components of the
transfer function, the peripheral pressure waveform is quite insensitive to these changes.
This is because the aortic pressure pulse is already band limited (Nichols and O'Rourke
1990).
In summary, the B-Tree model employed in this study predicted changes similar to that
observed experimentally in the human arterial system. Using this model and the
pressure maps, propagation of reflected waves from different compartments was
investigated. Amongst the compartments studied (upper limbs, lower limbs and the
trunk) reflected waves originating from the trunk influence the ascending aortic
impedance more than other compartments. Furthermore, the reflected waves from the
trunk travel into every other compartment and form the secondary oscillations in the
124
pressure waveforms. Similar phenomena were not observed in other compartments
indicating the possible role played by trunk vasculature on the overall haemodynamics.
When extended to the upper limb, this model predicted that there is a more complex
relation between central aortic and radial pressure waveforms than a simple second
order system. The transfer function which defines the relationship between central
aortic and peripheral radial pressure waveform is insensitive to alteration of arterial
parameters at lower frequencies. At higher frequencies, however, the transfer function
might vary considerably but due to limited bandwidth of the central aortic pressure
waveform the peripheral radial pressure waveform changes little.
125
MODEL VALIDATION AND PARAMETER ESTIMATION IN A SINGLE ELASTIC TUBE In previous chapters the model used to simulate the arterial tree was described in terms
of its organisation and implementation. Results were also presented for wave
propagation and effects of altering arterial properties. However, no model can be
complete without testing its accuracy in its assumptions and predictions. Similarly, the
implementation of model algorithm and analysis tools have to be tested. The integrity
of the hardware and developed software requires testing so that errors could be spotted
and corrected before proceeding with experimentation.
This chapter describes an experimental investigation using a single elastic tube
terminated with a real positive load aimed at validating the model described previously.
CHAPTER 6
126
This enables estimation of model parameters using an optimisation technique which
allows for global convergence of system parameters to experimental observations.
6.1 SINGLE ELASTIC TUBE EXPERIMENT
The experiment is similar to the classical tube experiments performed by early
investigators of arterial haemodynamics (Taylor 1957b, 1959b; Hardung 1964) , figure
6.1.
A/D and D/A Board
PressureTransducer
PressureTransducer Clamp
Amplifier
Pressure Generator
Data
PC
Elastic Tube
Figure 6.1 The schematic drawing showing arrangement of experimental hardware. A PC coupled with A/D-D/A board controls the experiment. It generates pressure waveforms through a servo controlled pressure generator in the tube. The simultaneous recordings of propagated pressure along the tube are made by identical pressure transducers which are amplified and digitised using the same PC. The digitised data are stored in computer hard disk for subsequent analysis.
This setup was chosen not only because of its ease of construction but also because there
is ample theoretical and experimental evidence for comparison (Milnor 1989; Nichols
and O'Rourke 1990). Briefly, a silicon elastic tube (Silastic, Dow-Corning), two metres
127
in length, was clamped at one end, while the other end was excited by a pressure source.
The tube had an internal diameter of 0.335 cm and a wall thickness of 0.065 cm. The
tube was relatively thicker and shorter than that described in previous reports (Taylor
1957a, 1959; Hardung 1964). Specific waveforms were generated by means of a servo
controlled loudspeaker (Biotek waveform generator Model 601A). Since the volume
displacement that could be attained by the drive mechanism was small (2 ml) within
linear range, it was important to ensure a stiffer tube at this length to restrict volume
displacement. The tube was filled with isotonic saline (0.9% wt/vol) which was boiled
and cooled to eliminate air bubbles in the fluid. The pressure generator with the tube
filled with this fluid was able to generate pressure frequencies up to 20 Hz, in line with
input fed to the waveform generator. The pressure generator was excited by waves
having fundamental frequencies of 0.65, 1.25, 2.0 and 2.5 Hz dumped out from an IBM-
AT computer in analogue form using an Analog to Digital/Digital to Analog board (DT
2801-A, Data Translation, Mass). The same computer and board were used to acquire
signals from pressure transducers that were conditioned by a suitable amplifier (Model
HS9, Medtel Instruments).
C5C4C3C2C1
R5R4R3R2R1
PressureWave
Generator
~
Figure 6.2 The arrangement of sites of pressure recording (R1=0 cm, R2=30 cm, R3=60 cm, R4=90 cm, R5=200 cm) and the sites for clamp application ( C1=100 cm, C2=120 cm, C3=140 cm, C4=160 cm, C5=200 cm).
Pressure waves were recorded using to disposable strain gauge manometers (Admac
100, Spectromedics). The transducers which are directly attached to 22 gauge
hypodermic needles are inserted at recording sites R1-R5 ,Figure 6.2. Since the
128
dynamic equivalence of the manometers was more important than true frequency
characteristics (the waves are all described in relation to pressure waves at site R1), the
test on this factor showed that they are identical up to frequencies of 40 Hz. The
pressures are recorded at sites R1 for input, and R2, R3, R4 and R5. Site R2 is
assumed to be fixed when expressing the influence of increased reflection due to early
wave reflection. To simulate different distances to the wave reflection site, ie early
wave reflection, the tube is clamped at different distances from the input (C1=100 cm,
C2=120 cm, C3=140 cm, C4=160 cm, C5=200 cm).
6.1.1 Estimation of Tube Elastance
The manufacturers specification for the Silastic tube used in this experiment did not
include values of wall elastance therefore it was necessary to determine tube elastance
by experimentation. The calculation of the elastance of the tube was compared with the
model predictions from the optimisation process. Although the calculation could be
achieved by using a single method, three separate methods previously reported (Patel
1972) are employed in the calculation.
Calculation of Elastance from Stress-Strain Relation by Longitudinal
Stretch
Upon completion of the experiment, a segment of the tube 100 cm in length was
subjected to stretch by hanging weights to one end while suspended from the other. The
resultant displacements were recorded. Assuming a Hookean substance for the wall
material, the elastance was derived by fitting a straight line to stress-strain relation
(Attinger 1964):
σ = =
FA
M gA.
(6.1)
129
where M is the mass in kg, g is the gravitational constant 9.81 m/sec2 and σ is the
tensile stress. The strain, ε , is described as
−×= 1100
100 mlε (6.2)
where lm is the length of the tube under mass M. The elastance then becomes
ESS =σε
(6.3)
Calculation of Elastance from Pressure-Volume Relations
A bolus of saline 0.5 ml is injected into the tube. The resultant change in pressure, ∆P ,
is then used to estimate elastance using the formula given by Bergel (Bergel, 1960).
( )2223
oi
oi
i
PV RRRR
RR
PE−∆
∆= (6.4)
where Ri and Ro are the inner and outer radius respectively. Both radii were calculated
after considering a circular cross-sectional area where expansion of tube in z direction is
ignored.
Calculation of Elastance from Wave Velocity:
Wave velocity can be used in conjunction with Moens-Korteweg equation to estimate
elastance
E c Rh
i o= 02 2 ρ (6.5)
130
The value of wave velocity, c0, can either be obtained by foot-to-foot estimates
(Nichols and O'Rourke 1990) or by averaging phase velocities for frequencies above the
first minimum. An alternative method can be devised using the quarter wavelength
formula
c f l0 4= min (6.6)
Since phase velocity minimum, fmin, also denotes the minimum of the impedance
spectrum (see chapter 2.2.6 ), first minimum of the phase velocity spectrum or first
maximum of the transfer function modulus can be used to estimate fmin. Wave
velocities are estimated by foot-to-foot and quarter wavelength method and
subsequently to estimate of elastances of ECff and ECf min respectively.
6.2 ESTIMATION OF MODEL PARAMETERS
The arterial model described in Chapter 4 is used in conjunction with an optimisation
process to estimate various physical parameters of the tube. For this purpose, a uniform
transmission line model of the experimental set up was constructed using the B-tree
model described in chapter 4. The model parameters for the experimental case of
clamping at 200 cm (C5) are given in Table 6.1. The measurable parameters of the
model - diameters, wall thicknesses, and lengths- are entered as known parameters. The
elastance of the tube and reflection coefficient at the termination are assumed to be
unknown and thus estimated by the model.
Using pressure waves generated by the model and the ones actually measured along the
tube, an error function is defined to describe the deviation of model from the elastic
tube. For this purpose, recorded pressure waveform at site R1 is used as input to model
131
at node M. The sum square difference between the pressure waves recorded at sites R2-
R5 and the ones calculated by the model for the corresponding sites is defined as the
forcing quantity. The process is iterated with different values of E and RF until values
less than 2 percent are achieved (Figure 6.3). At this value of E and RF the model is
assumed to represent the elastic tube. The entire iteration procedure is implemented as a
batch process. This enabled to predict model parameters from experiments described in
Section IV.
B-Tree Code
NAME L (cm)
D (cm)
h (cm)
Elastic Modulus
(dyne/cm2) x106
θ (Degrees)
Γ (Units)
M R1 0.01 0.34 0.07 E 0.0 MR R2 30.0 0.34 0.07 E 0.0 MRR R3 30.0 0.34 0.07 E 0.0 MRRR R4 30.0 0.34 0.07 E 0.0 MRRRR R5 110.0 0.34 0.07 E 0.0 RF
TABLE 6.1 Table of elements included in the model representing the entire tube and its measurement sites where the tube is clamped at C5=200 cm. Elastance , E, and reflection coefficient at C5, RF, are estimated by the optimisation process.
E, RF
InputPressure
Wave
RegressionCoefficient
Calculated
Measured
New Model Parameters
Linear Regression
B-Tree Model
Elastic Tube
Pressure
Pressure
Figure 6.3. Block diagram for estimation of elastic tube parameters using an iterative technique and an appropriate model.
132
6.3 RESULTS
6.3.1 Tube Elastance
The stress-strain relation obtained from longitudinal stretch was not perfectly linear,
concave towards strain axis, indicating a higher compliance for larger stresses (figure
6.4). The slope of the linear relation, ESS , was 48.25 x106 dyne/cm2. This value
compares well with the value obtained from pressure-volume relation. Injection of 0.5
ml fluid changed the pressure by 130 mmHg, yielding an EPV value of 44.65 x106
dyne/cm2. The foot-to-foot delay between sites R1 and R5 was 67 milliseconds, giving
a pulse wave velocity of 2985 cm/sec and by equation 6.5 the elastance ECff becomes
45.94 x106 dyne/cm2.
Stress
Figure 6.4. The stress-strain relationship of the elastic tube used in the experiment. The values are obtained by applying tension to the tube and by measuring the elongation.
The value obtained from equation 8.6 for recordings between sites R1 and R5 yields an
ECf min value of 42.04 x106 dyne/cm2 for a minimum of 4.2 Hz (Table 6.2). The close
133
similarity between different methods of calculation of the tube elastances allows one to
use any of the methods for estimation of wall elastance.
METHOD Elastance (x106 dyne/cm2)
Stretch (ESS ) 48.25 Pressure Volume (EPV ) 44.65 Foot-to-foot (ECff ) 45.44 Quarter Wavelength (ECf min ) 42.04 Mean ± SEM 45.09±1.28
Table 6.2. Data showing the comparison of different methods on the estimation of the elastance of the tube.
6.3.2 Convergence of the Model
The error introduced by the model converged to a minimum for different fundamental
frequencies (figure 6.4).
For curves rich in low frequency power, that is at lower fundamental frequencies, the
concavity of the error surface was less than that at higher fundamental frequencies.
However, the errors obtained in the latter were much higher than in the former. These
indicate a high sensitivity to the model parameters E and RF for high frequency
components. Not surprisingly, the predicted value for E by the model varied between
39.61 x106 dyne/cm2 and 47.27 x106 dyne/cm2, with an average of 42.18 x106
dyne/cm2. Convergence always occurred for errors less than 2 %.
134
A B
C D
Figure 6.4. The sum square error surfaces between by the model predicted (Table 6.1) and measured pressure waves at site R5 during which the fundamental frequency of the input wave is varied (A=0.65 Hz, B= 1.25 Hz, C= 2.0 Hz, D= 2.5 Hz). Note the shallow minima at lower fundamental frequencies. The site of tube clamp is at C5 (200 cm) and the pressure recording is made in R2. Convergence always occurs with an error of less than 2 percent.
6.3.3 Predicted Pressure Waves
The pressure waves generated by the model for the optimum values are shown in figure
6.5.
135
A B
C D
Figure 6.5 Pressure waves predicted by the model (solid line) and measured (dotted line) at different fundamental frequencies (A=0.65 Hz, B=1.25 Hz, C=2.0 Hz, D=2.5 Hz) for the optimised model. The individual parameter optimisation process for these waveforms is given in Figure 6.4. The site of clamp (C5) is 200 cm away from the origin.
The upstroke and the contour of the pressure waves are all well matched. The waves
from the model (solid line) are not identical to experimental waveforms at peak systolic
pressures (dotted line). The difference in peak systolic pressure is dependent on the
fundamental rate, showing accentuation at or around 3 Hz (see below). However, this
difference never exceeds 15% of the calculated pulse pressure.
136
6.3.4 Transfer Function
Modulus
Early wave reflection caused by shortened termination increases the amplitude of the
modulus at the maximal point with accompanying rightward shift of the peak (top
figures in figure 6.6 through 6.8).
A B
Figure 6.6 Transfer function moduli (top) and the apparent phase velocities (Bottom) of the model (solid line) and of the experiment (circles) between sites R1 (0 cm) and R2 (30 cm) when termination is at 100 cm (A) and 120 cm (B) from the origin. The dashed line in the phase velocity plot (2955 cm/sec) represents the true wave velocity calculated from the mean elastance value of Table 6.2. Note the difference in scale of moduli in (A).
137
This change in modulus of transfer function between sites R1 and R2 shows the same
trend for both the experiment (circles) and the model (solid line).
A B
Figure 6.7. Transfer function moduli (top) and the phase velocities (Bottom) of the model (solid line) and of the experiment (circles) between sites R1 (0 cm) and R2 (30 cm) when termination is at 140 cm (A) and 160 cm (B) from the origin. The solid line in the phase velocity plot (2955 cm/sec) represents the true wave velocity calculated from the mean elastance value of Table 6.2.
Wave reflection also increases due to early wave reflection caused by shortened
distance. At higher frequencies, it increases the differences between recorded waves
while at lower frequencies the amplitude of augmentation decreases. This rightward
shift of the maxima caused by shortening of the tube beyond R2, can also be interpreted
as the stiffening of the segment between R1 and R2. The higher amplitude at this
138
maxima can also be interpreted as an increase in the reflection coefficient at the
termination R2.
Figure 6.8 Transfer function moduli (Left) and the phase velocities (Right) of the model (solid line) and of the experiment (circles) between sites R1 (0 cm) and R2 (30 cm) when termination is at 200 cm from the origin. The solid line in the phase velocity plot (2955 cm/sec) represents the true wave velocity calculated from the mean elastance value of Table 6.2.
These alternative explanations cannot be accepted readily, unless they are confirmed by
information gained from the phase plots. The model, on the other hand, reproduced the
essential features of the experimental findings. There was some discrepancy, however.
The model tended to underestimate the amount of the pressure wave amplification in
every frequency, sometimes as much as 15%. This was apparent from the synthesised
pressure waves, figure 6.5, where the model could not fit the exact amplitude by up to
15%. This finding is surprising, since there were no losses in the wall, wall viscosity
was zero and the estimated reflection coefficient was close to unity (actually 0.999).
This difference is probably due to the tethering of the tube (Taylor 1959). The
mathematical model uses strong tethering which increases viscous losses due to
increased coupling between wall and fluid, yet the tube was tethered slightly in this
experiment. Since there is substantial tethering in vivo situations, the model should
apply well in those circumstances.
139
Phase Velocity
Phase velocities, between sites R2 and the termination, predicted by the model (Bottom
figures in figures 6.6 through 6.8) agree well with experimental data (circles). They
both start from a high value for low frequencies, then fall to a minimum, followed by
maxima and minima around true wave velocity that is calculated using mean elastance
value in table 6.2. The deviation from this true wave velocity is caused by reflected
waves originating from the closed end termination. This is manifested by the frequency
of the second minimum being three times that of the first minimum (3.8 vs 11.8 Hz for
clamp at C5). When the clamp is moved closer to the measuring site (from C5 to C1),
the first minimum shifted rightward. There was simultaneous increase in the values of
phase velocities at lower frequencies. This agrees with model predictions that the
reflections will increase (higher values) with decreased distance (rightward shift) to the
reflecting site.
In the extreme case, C5, the distance to the reflecting site becomes more difficult to
calculate using the quarter wavelength formula. The very short distance makes the
minimum very dispersed hence making its detection more difficult. Due to introduction
of significant errors in calculations, the phase differences are small, at higher
frequencies the model predictions are different from the experimental findings. The
distances to the reflecting site as calculated from the phase velocity plots, change in the
same direction for model data and experimental data, indicating the implementation of
the B-tree model does not introduce errors.
6.3.5 Reflection Coefficient
The model predicted a terminal reflection coefficient of 0.95 or more for each
optimisation. This value is similar to the experimental setup, given that the reflection
140
coefficient is determined by using phase velocities. This value varied from 0.8 to 0.95
for proximal reflecting sites obtained by clamping the tube proximally. The values
obtained from the model for reflection coefficients for the same situation yielded the
same values, 0.8 to 0.95 as in experiments.
6.4 DISCUSSION
The elastic tube model has been used successfully by previous investigators (Taylor
1959, Hardung 1964) to verify theoretical predictions. They measured pressure and
flow along the elastic tube and calculated various model parameters. In this study, the
experiments are performed with omission of flow measurements. By programming the
experiments in a digital computer, attempts have been made to mimic the real
conditions. However, it is difficult to predict the behaviour of an algorithm
implemented in a computer. Such an implementation may be tested by simulating the
real events with real data. This approach was adapted in these experiments. A
performance criterion was defined and the simulation was optimised until this criterion
was achieved.
The real data was obtained from the dimensional measurements of the elastic tube
material and from the pressure measurements. Since an extra physical quantity,
elastance, is needed as well to define the material, different techniques were utilised to
estimate this quantity. It was found that these techniques estimated similar values with
close agreement. The model estimate closely approximated the actual values of wall
elastance under a wide variety of conditions. The same performance was obtained for
wave reflection coefficient. Although the initial values of these two parameters were
very much different from the actual values, the convergence was definite, free from
local maxima and minima. This suggests that continuous estimation of parameters may
141
be made with minimal computer resources. Since this approach uses only pressure
measurements it avoids the complexities brought by measurement of blood flow
velocity.
Proper implementation of the model in a system should necessitate measurement of the
tube's physical properties. In tube models, distances can be measured quite accurately as
well as contours of pressure waves. The wall thickness, diameters and wall viscosity
parameters could be fixed for certain segments of the arterial tree, freeing computer
resources from estimation of these parameters as well. Normalisation process by setting
a fixed diameter, wall viscosity and wall thickness, does not make much difference in
phase velocity estimates. The variation of diameter will effect the α number, which is
high under normal circumstances. Therefore α related parameters are asymptotic at this
range. The wall viscosity, however, is important in setting the transfer function moduli
values (Taylor 1966b, Avolio 1976a) as demonstrated in this experiment. Even small
viscous forces due to tethering can change the synthesised pulse pressures. In the model
presented here it was less than 15 percent.
It was also shown that the implemented model followed the changes in transfer function
modulus and impedance, as determined from the phase velocity estimates, reliably. This
was achieved under wide variety of conditions which included changes in fundamental
frequency and distance to the reflecting sites. As expected, when distance to the
reflecting site increased, the first minimum of the phase velocity spectrum and the
transfer function modulus shifted leftwards. Since the wave velocity was unchanged,
the distance from the reflecting site could be obtained using quarter wavelength formula.
However, when the true wave velocity is unknown, the use of either of these plots alone
may lead to misinterpretation of the phenomenon; that is, estimations made from the
modulus alone do not differentiate between increase in intensity of reflection coefficient
from distance to the reflecting site or change in wave velocity. The actual reflection
coefficient or the actual wave velocity can be overestimated.
142
Obviously, the parameter estimation could still be achieved if one has to construct
pressure waveforms at site R1 for corresponding inputs of R2-R5. However, in this
calculation, the problem of inverse filtering of a low pass filter occurs (Kenner 1978).
The high frequency components of the pressure wave measured downstream are
attenuated and are close to noise level due to viscous losses. The tube, therefore,
behaves like a low-pass filter at these frequencies and the information lost during this
process would not be regained by inverse filtering. To overcome this problem, a higher
fundamental frequency with enough power would have to be issued. In this study the
highest fundamental frequency used in this study was 2.5 Hz and the components of the
pressure wave at the 6th (15 Hz) and above harmonics were well below system noise.
There is also possibility that parameters estimated using this process may not be unique.
To avoid being attracted to some local minimum in the solution, the estimator first scans
the entire parameter space for detection of local minima. It then adjusts the iteration
steps accordingly. Since this global search approach uses extensive amount of computer
time, once the boundaries of local minima are known, the model assumes this
information as a priori and uses it to limit the parameter space. Iteration maps of the
calculation process are generated for the entire E and RF space, to test the hypothesis
that these calculated values are unique. Once it has been verified that there is a unique
solution by this method, no further maps are generated since it was consuming extensive
amount of computer time.
In summary, the model could reliably predict parameters such as reflection coefficient,
wall elastance and spatial distribution of pressure waves. It is also clear from these
findings that one can use the model to support experimental data to extend the
experimental findings. This can be achieved without significantly increasing the errors
introduced other than by measurement.
143
SECTION III
EXPERIMENTAL STUDIES
In this Section, the features of the arterial pulse, measured in a central or peripheral
artery, are investigated systematically in human subjects. To explain features of the
pulse, methods are described to estimate the intensity of reflected waves using pressure
wave recordings. In general, however, the emphasis is placed on the formation of the
upper limb pressure pulse as upper limb arteries are readily accessible and pressure
pulses can be obtained non-invasively. Experiments were performed using wave
propagation parameters as a substitute for pressure-flow relations in real conditions.
In the first chapter, the upper limb system is described in subjects where the input to the
system could be varied. In a group of 18 subjects with implanted programmable cardiac
pacemakers, the input pulse shape to the upper limb arterial system was altered by
144
changing heart rate. Resultant changes in the output pulse were measured. Using
spectral techniques a unique expression was derived between the central pressure
waveform and the peripheral/radial pressure waveform.
In the second chapter, the effect of reflections from the terminations on upper limb
system were investigated. By applying gradual compression to the vessels of the palm
vasculature and by reducing the mean arterial pressure in the upper limb, the pressure
waveform and the intensity of reflected waves were varied and studied. The transmural
pressure in this vasculature was found to have a systematic effect on the system.
However, under normal circumstances, the system was similar to that observed under
total occlusion of this vasculature.
In third chapter, the effects of vasoactive drugs on this system were investigated.
Although there are documented reduction in secondary pressure oscillations caused by
these drugs , these were found not to be related to changes in upper limb properties.
These drugs affected the proximal vessel properties without altering the reflections
occurring at the termination of the upper limb.
The fourth chapter concentrates on the relationship between secondary oscillations on
the pressure wave and the arterial circulation as a whole. Administration of oral glucose
to a group of subjects caused reductions in secondary oscillations. However, there were
no changes in the upper and lower limb arterial system suggesting a possible role of
vasoactive state of the splanchnic circulation on these oscillations. This suggested a
bifurcation of the aorta into two circulations, one to the splanchnic and the other to the
lower limb. Representation of the arterial tree with a modified T-tube model using this
concept generated realistic pressure and flow waveforms.
145
DETERMINATION OF WAVE PROPAGATION CHARACTERISTICS IN THE HUMAN UPPER L IMB The clinical implications of pressure wave propagation of the upper limb are often
overlooked. Pressure wave in the upper limb is not only different in its amplitude but
also different in its shape to that in the central aorta (Kroeker and Wood 1955;
Remington and Wood 1956, Kelly et al 1989). Due to wave propagation and wave
reflection phenomena the travelling pressure wave is delayed and is amplified. Yet, the
pressure pulse in the upper limb is used to infer central pressure pulse and the left
ventricular afterload ( Pauca et al 1989).
Previously, the comparisons of pressure waves recorded in the upper limb to that in the
central artery are made by comparing the peak and mean pressures obtained using
Chapter 7
146
sphygmomanometric techniques (Borow et al 1982). The invasive data is obtained
during vasodilation (Kelly et al 1990, Simkus and Fitchett 1990), exercise (Rowell et al
1968), normal conditions (Kroeker and Wood 1955; Remington and Wood 1956) and
shock (O'Rourke 1970). However, these studies were limited by the range of heart rate
available to the investigators. Although it was possible to obtain a spectrum of rates in
many patients, this was not useful in investigating the variation brought by different
heart rates. When it was possible to change the heart rate by exercise (Rowell et al
1968), the alteration of global control mechanisms and the vasoactive state brought by
this manoeuvre made it impossible to analyse the data.
In this Chapter, results are presented demonstrating the clinical implications of wave
travel and wave reflection in eighteen subjects with implanted programmable
pacemakers. By pacing the hearts at different rates and taking simultaneous
measurements of pressure waves in the carotid and radial arteries together with cardiac
output, the changes in the central and peripheral pressure wave shapes were
documented. A transfer function was determined for each subject by relating the central
and peripheral radial pulses in the frequency domain. It was found that the
amplification of the pulse from central to peripheral sites depends on heart rate as first
suggested by O'Rourke (1970) and that the upper limb transfer function is essentially
constant in the frequency domain. Results were similar to those predicted by the upper
limb model presented in Chapter 5.
147
7.1 METHODS
7.1.1 Subjects
Subjects for this study were drawn from a group attending a hospital pacemaker clinic as
outpatients. All were equipped with VVIR pacemakers (Telectronics, Optima-MP) and
were undergoing routine pacemaker functionality test. The population consisted of 12
males and 6 females whose age ranged from 19-82 years with a mean age of 57.6 years.
All patients were in good health and signed a consent form to participate in the study.
7.1.2 Measurements
The heart rate and cardiac output of the subjects were measured with a non-invasive
cardiac output meter (Bomed, NCCOM3-R7). This device operates on the principle of
impedance cardiography by injecting currents between two electrodes at 20 to 200 kHz
and detecting the voltage drop across the thorax with a different pair. With proper
calibration, this device has shown to give similar results to existing cardiac output
methods, ie. thermodilution and dye clearance (Herok 1992). Two tonometers, one at
the carotid artery, the other on the radial, continuously recorded the pressure wave at
these sites while a photoplethysmographic blood pressure measurement (FINAPRES)
was performed on the middle digit of the right hand. When the patient was relaxed and
all the recordings were stable, the heart rate was varied by 10 beats/minute from 60
beats/minute to 120 beats/minutes, while the established rate was maintained for 10
seconds to allow for a steady state to be achieved. After completion of the study, the
pacemaker was programmed to run at a rate of 70 beats/minute where another set of data
was collected.
The carotid and radial pressure data at different heart rates digitised on-line via an IBM-
compatible PC equipped with an A/D board (DT 2801-A, Data Translation Mass).
Acquired data (128 Hz, 12 bits per channel) was temporarily stored on the hard disk to
148
be later transferred to a WORM (Write Once Read Many) drive (Maxtor 800S) for
permanent storage. Using in-house software developed for this purpose and using ECG
triggering data segments of 15 second length were averaged.
Calibration of Waves
It is normally impossible to predict the calibration of non-invasive tonometric tracings
due to lack of knowledge of exact applanation pressure and orientation. However,
calibration is possible if one determines the pressure values by an independent technique
and assumes that (i) the mean arterial pressure is equal throughout the arterial tree and
(ii) the diastolic pressure values can be considered to be approximately equal in every
branch. These two assumptions are readily satisfied in this study. Firstly, the
measurements are made in the supine position and secondly at diastolic pressure level
the effects of reflected waves are shown to be minimal (see Chapter 5). In this study,
the calibration of tonometer tracings is performed by equating the mean and diastolic
pressures obtained from the FINAPRES device and the non-invasive recordings.
7.1.3 Data Analysis
Pressure Wave Amplification
To quantify the pressure wave amplification between the carotid pulse pressure, PPc,
and radial pulse pressure, PPr, in the time domain the following relation is defined.
100×=c
ramp PP
PPPP (7.1)
Pressure Wave Shapes
The augmentation index, AI, is used in this study to document changes in the radial
pressure waveform with increase in heart rate. The augmentation index is often used to
express the intensity of reflected waves (Murgo et al 1980b). This index is defined as
149
the ratio of peak systolic pressure increase due to wave reflection to the peak pressure at
around flow and given as:
dPP
PPAI−−
×=1
12100(%) (7.2)
Where P1, P2 and Pd are pressure at late systole, at early systole and at diastole. When
applied to the radial artery waveform, however, this index attains significantly lower
values from the centrally derived indices (Kelly et al 1989a). Because of the
mismatches, reflected waves originating at the palm level increase the first peak in the
radial artery without necessarily affecting the second peak (see Figure 5.17). Thus, the
augmentation index measured in the radial artery actually decreases due to an increase in
locally reflected waves. Furthermore this index attains negative values for the
peripheral waveforms. Therefore, this index is modified throughout the thesis by adding
an offset of 100 to the form defined by Murgo et al (Murgo et al 1980).
Transfer Functions
For an absolute determination of the relationship between the carotid and radial
waveforms transfer function of the upper limb RACAH − is determined. This allows a
unique way of expressing both the pressure pulse amplification and the wave shapes
under different pacing rates. The transfer functions are determined by relating the
harmonic components calculated using Fourier analysis (see Chapter 2).
Parameters Obtained from the Transfer Function
By applying equations derived for the single tube analog of the upper limb transfer
function, the resonant frequency and reflection coefficient can be estimated.
Resonant frequency
150
The resonant frequency of the upper limb is derived from the quarter wavelength
formula
l
cfn 4
0= (7.3)
where c0 is the pulse wave velocity obtained from foot-to-foot measurements and l is
the distance between measurement sites.
Reflection coefficient
An estimate of the complex reflection coefficient is derived from the phase velocities
analogous to equation 2.148 and described as
∞
∞+ +
−=Γ
cccc
i
i0 (7.4)
where ci is the pulse wave velocity at frequency 'i' and c∞ is the foot-to-foot estimate of
true wave velocity obtained from the pressure waves. This technique assumes c∞ to be
proportional to Z0,characteristic impedance, and ci to be proportional to the terminal
impedance, ZT, value at frequency 'i'. Similarly, the pulse wave velocity at very long
wavelengths, c0, or ones close to zero frequency, c0+ , is considered to be related to the
peripheral resistance values. These velocity values were then used to estimate a
compatible reflection coefficient value with those previously reported (O'Rourke and
Taylor 1966). Estimation of the value of c0 for this purpose normally requires
utilisation of random excitation techniques to increase the resolution of the phase
velocity spectrum. Since application of this technique in humans was impossible, c0+ ,
the phase velocity at very low frequencies, is substituted for c0 and estimated by linear
extrapolation of phase velocities of second and first harmonic to zero. This linear
extrapolation technique has already been described in conjunction with estimation of
151
frequency independent values of elastance from frequency dependent terms in smooth
muscle preparations (Gow 1972).
Reduced Model Of the Upper limb Arterial System
Since it was possible to obtain reasonably high fidelity recordings from these patients,
an attempt was made to obtain parameters describing the arterial properties of the upper
limb. For this purpose, a parameter estimation technique similar to that performed in
the elastic tube experiment (Chapter 6) was employed. However, in these sets of
simulations a reduced model of the upper limb involving few branches was constructed
in order to decrease computation time (Table 7.1).
B-Tree Code
NAME L (cm)
D cm
h (cm)
E (x106 ) (Dyne/cm2)
Θ (Degree)
Γ (units)
τ (Sec)
M Asc. Aorta 0.10 0.94 0.08 Em θm Γm τm MR Innominate 9.10 0.94 0.08 Em θm Γm τm MRR Brachial 22.0 0.60 0.18 Em θm Γm τm MRRR Radial 41.0 0.34 0.16 Em θm Γm τm MRRRR Termination 0.10 0.34 0.16 Em θm Γm τm
Table 7.1 Reduced model of the upper limb used in estimation of transmission characteristics. Em , θm , Γm and τm are the model parameters to be estimated from the best fit between pressure waves generated by the model and those measured using linear regression analysis.
Input to this model was considered to be the carotid pressure waveform. Similar to the
elastic single tube experiment, the model was iterated with different elastance, Em , wall
viscosity, θm , reflection coefficient, Γm , and time constant, τm , values to yield a best
fit between model derived and actually measured radial waveform. The model values
obtained this way are considered to be the values representative for the arterial segments
in the upper limb vasculature. Since same elastance values are used for each element of
the model, the wall thickness to diameter ratio is adjusted to account for progressive
increase in pulse wave velocities (Table 4.1). This resulted in seemingly unrealistic
152
values for wall elastance values. However, corrections could be made if the actual wall
thickness to diameter ratios were known by multiplying wall elastic modulus with the
appropriate ratio.
Model limits are then compared with these values obtained by analysing modulus and
phase velocities of transfer functions.
7.1.4 Statistics
Differences from the baseline heart rate are analysed using paired t-test. A p value of
less than 0.05 is accepted statistically significant. Data are presented as mean ± standard
error of mean unless otherwise stated.
7.2 RESULTS
Figure 7.1 shows an example of pressure waveforms recorded in the carotid and radial
arteries for a single patient during the pacing protocol. Under control conditions the
pulse pressure in the radial artery was higher than that found in the carotid artery.
Pacing the heart from 60 to 120 beats/min with increments of 10 beats/min decreased
the pulse pressure in the carotid artery. The pulse pressure in the radial artery either
increased or remained unchanged. This made it difficult to asses the pulse pressure in
the carotid artery from the radial artery. When heart rate was reduced to 70 Beats/min,
the difference in pulse pressures was restored to baseline values. This finding indicates
that heart rate is the only determinant of this amplification. Similar changes are also
observed in all patients (Table 7.2).
153
HEART RATE
Figure 7.1 Pressure recordings obtained from one patient showing carotid (solid line) and radial (dotted line) pressure pulses. Pulses are averaged and calibrated by FINAPRES device. The heart rate increased from 60 to 120 with increments of 10 beats/min, then decreased back to 70 beats/min. Note the decrease in pulse pressure in the carotid while not much increase is observed in the radial.
When data from all subjects are pooled and expressed in terms of AMPPP , it revealed a
non-linear relationship between the heart rate and the pulse pressure difference (Figure
7.2). There was an increase in AMPPP with heart rate (from 114±7% (mean ± SD) at 70
beats/min to 123±15% (mean ± SD) at 110 beats/min).
Pacing of the heart also caused significant changes in the pressure waveforms (Figure
7.3). There was a significant drop in the augmentation index, AI, with increase in
pacing rate (from 79±10% at 70 beats/min to 50±7% at 100 beats/min, p<0.001). At
much higher rates (110 and 120 beats/min) the algorithm used to detect the second peak
(see Chapter 12) failed in all instances due to absence of this peak. This finding alone
154
Carotid Pulse Pressure (mmHg)
Radial Pulse Pressure (mmHg)
P No Age
60 B/min
70 B/min
80 B/min
90 B/min
100 B/min
110 B/min
120 B/min
70 B/min
60 B/min
70 B/min
80 B/min
90 B/min
100 B/min
110 B/min
120 B/min
70 B/min
1 19 34 35 38 43 38 43 * * 43 42 42 44 44 41 * * 2 37 43 39 39 39 36 37 35 39 45 41 46 47 44 45 45 54 3 30 52 44 40 38 38 34 32 45 55 51 51 55 49 51 49 64 4 35 46 45 40 40 39 44 * 35 54 54 54 56 56 59 * 53 5 64 81 75 71 69 90 * * 93 82 80 76 79 100 * * 98 6 59 53 49 48 56 51 49 51 49 56 56 60 65 63 62 60 64 7 65 * 50 45 46 43 36 39 60 * 61 56 56 56 54 53 73 8 59 52 57 57 54 52 47 41 63 56 65 64 66 64 61 59 74 9 45 * 53 49 48 43 43 42 59 * 65 62 64 60 59 54 64 10 61 59 50 41 42 44 47 46 52 59 53 47 46 53 54 53 56 11 71 54 62 59 56 55 63 64 61 62 67 65 64 60 67 72 65 12 82 30 32 31 31 30 30 26 37 34 37 34 34 34 36 29 46 13 74 47 40 37 35 42 40 36 51 53 47 45 42 51 49 46 58 14 71 63 60 49 54 46 48 48 81 70 68 53 58 55 60 62 76 15 77 66 58 56 54 53 54 52 68 68 62 65 62 64 64 62 73 16 82 41 43 46 38 53 52 46 46 41 44 47 39 49 50 49 47 17 23 24 30 28 28 25 25 24 33 32 36 37 38 38 32 32 43 18 78 88 86 75 67 65 62 54 80 98 94 93 86 85 82 72 95 Mean 57.33 51.50 50.44 47.17 46.56 46.83 43.25 42.67 54.50 55.57 56.83 55.39 55.61 56.94 52.75 53.67 63.00 SEM 4.73 4.13 3.32 2.88 2.65 3.30 2.42 2.71 4.05 4.15 3.49 3.29 3.24 3.60 2.86 3.08 3.67
Table 7.2 Patient details and pressure pulses obtained at baseline and during pacing. Missing values (*) are due to lack of baseline data or the impossibility of pacing at a higher rate.
155
PPAMP
Figure 7.2 The pulse pressure amplification between carotid and radial sites during pacing. The pressure wave is normally amplified during its travel in the upper limb. There is also small but significant increase in amplification with higher heart rates. (Bars represent ± 1 SEM, * p<0.05, † p<0.001 compared with 70 beats/min).
RADIAL AI
Figure 7.3 The augmentation index in the radial artery pressure waveform At rates higher than 100 beats/min, the second shoulder was impossible to be detected. (Bars represent ± 1 SEM, † p<0.001 compared with 70 beats/min).
156
HCA RA−
60708090
10011070x
Figure 7.4 (Left) The transfer function between carotid pulses and the radial pulses of Figure 7.1 during pacing the heart form 60 beats/min to 110 beats/min and back to 70 beats/min. (Right) The transfer function between carotid and radial artery sites for the entire study group, compare this to one given in Figure 5.11. Bars represent 95% confidence interval.
reveals that major change in the pressure wave is in its first peak brought by increased
wave reflection from the palm vasculature.
The left panel in Figure 7.4 displays the transfer function modulus and phase for the
recordings given in Figure 7.1. Although there were significant differences in the
amplitude and the shape of individual pulses, radial pulses were related to carotid pulses
in a unique way in the frequency domain. As expected, there is a consistent
amplification of harmonics around 4 Hz. The graph on the right shows all patient data
that are pooled to obtain a generalised transfer function for the study population. The
close correspondence of this transfer function with the model predictions (Figure 5.11)
157
indicates the applicability of arterial model presented in Chapter 5 into upper limb
arterial system.
Patient No
fn (Hz)
Γ0+ (Units)
Em (X106) (dyne/cm2)
θm (Degrees)
Γm (Units)
τm (sec)
1 3.83 0.02 1.69 20.50 0.31 0.07 2 3.23 0.24 1.25 22.88 0.40 0.08 3 4.38 0.37 2.64 13.25 0.59 0.09 4 4.27 0.67 2.84 15.86 0.59 0.06 5 4.50 0.08 2.37 16.00 0.11 0.04 6 3.45 0.24 1.19 11.88 0.35 0.06 7 3.84 0.40 2.35 13.43 0.54 0.02 8 3.69 0.42 1.84 12.75 0.43 0.06 9 3.96 0.46 2.35 10.43 0.57 0.07 10 4.57 0.15 2.63 7.50 0.22 0.05 11 4.51 0.17 2.60 19.63 0.15 0.06 12 3.79 0.23 1.80 13.88 0.24 0.08 13 4.32 0.40 2.44 11.63 0.34 0.08 14 3.85 0.24 1.96 12.00 0.27 0.07 15 4.17 0.13 2.13 15.38 0.21 0.05 16 3.43 -0.06 1.28 25.13 0.04 0.04 17 3.71 0.39 1.90 14.50 0.64 0.07 18 3.85 0.27 1.75 17.25 0.39 0.07 Mean±SEM
3.96±0.09
0.27±0.04
2.06±0.12
15.22±1.03
0.36±0.04
0.06±0.00
Table 7.3 Measured and model generated upper limb model parameters. Resonant frequency fn and Γ0+ reflection coefficient are derived from quarter wavelength formula while Em , θm , Γm and τm are derived from model iterations.
Table 7.3 displays the values estimated from the transfer functions using quarter
wavelength formulas and with the model iterations. The resonant frequency is close to 4
Hz and the reflection coefficient estimate is 0.27 by the analysis of transfer functions.
Similar values are obtained from the model iterations for reflection coefficient.
However, the elastance values are relatively small with respect to documented values.
158
The small value of τm indicates that the termination of upper limb behaves more like a
closed end tube.
7.3 DISCUSSION
These results (Table 7.2) confirm the model predictions which suggests the presence of
pressure wave amplification along the upper limb. In this study, the pressure pulse is
not only amplified but its contour also altered while it travels along the upper limb
vasculature. As heart rate increased by pacing these changes became more dramatic.
These observations indicated the presence of frequency dependent amplification of the
pressure pulse in the upper limb and warranted the utilisation of spectral techniques for
studying this phenomenon.
The transfer function which relates the input to the output, ie central aortic to upper limb
pressure pulse, usually obtained by relating the harmonic components of input and
output pulses. However, the technique is limited by the heart rate involved which would
only resolve the spectrum at fundamental frequencies. In this study the continuous
spectrum is obtained by scanning the transfer function spectrum with alteration in heart
rate. Scanning of arterial impedance spectrum by pacing the heart is employed
previously in determination of ascending aorta (Taylor 1966c) and pulmonary artery
(Bergel and Milnor 1965). Use of this technique in pressure transfer function is new
and enables to compare the model predictions with the experimental findings.
The application of this technique to the pressure pulses revealed the uniqueness of the
transfer function in the upper limb, (figure 7.4, Left). Although, there were differences
in both the amplitude and contour (Figures 7.1, 7.2 and 7.3), the harmonic components
of the pulses are related in a unique manner. Similar results were obtained in each
159
patient by pacing the heart. The transfer functions were found to be similar to the one
predicted by the model (Figure 7.4, Right). As suggested by the model, the peak of the
modulus occurs at around 4 Hz, reaching to a value of around 2.2 at this frequency.
Analysis of transfer functions using single tube analogy indicated a resonant frequency
of 3.97 Hz and a terminal reflection coefficient of 0.27. The iteration of the model to
predict the properties of vessel wall and the terminal element confirmed these results.
Although the predicted wall elastance value is nearly one order of magnitude less than
that reported (2.06 vs 10 to 20 x106 dyne/cm2, Table 7.3 vs Table 4.1), the wall
viscosities are comparable (15.22 vs 10 to 20 degrees, Table 7.3 vs Table 4.1). The
modulus of reflection coefficient (0.36 vs 0.27, Γm vs Γ0+) and time constant (0.06 vs
0.00, θm vs real termination) also agrees. The discrepancy between the model predicted
and reported wall elastances are probably due to inaccurate estimation of wall thickness
to diameter ratio.
Based on the transfer function in the upper limb, a higher dependence of amplification
to the heart rate is expected. This behaviour can be explained by invoking the effects of
pacing on both ventricular filling and ejection. The pressure wave amplification does
not only depend on the transfer function, but also the components of the flow wave
constituting the fundamental harmonic. If the amount of energy contained in the
fundamental harmonic is decreased, the estimate of amplification by this technique will
be blunted. The VVIR method of pacing does not allow the atrium to contract in
synchrony with the left ventricle, thus not enhance left ventricular end-diastolic
dimensions. This decreases the contractile force developed by the left ventricle since it
is proportional to the end diastolic volume (Frank-Starling mechanism). As a result, the
stroke volume and the energy contained in the flow pulse decreases. Normalisation of
pulse pressure measured in the radial artery to the stroke volume, however, takes this
factor into account, Figure 7.5.
160
PP/SV
Figure 7.5 The radial pulse pressure for a given stroke volume indicating the dissociation of pulse pressure to that of stroke volume. (Bars represent ± 1 SEM, † p<0.001 compared with 70 beats/min)
The presence of amplification between the central and upper limb vessels and presence
of similar transfer function across different individuals suggests that radial artery
waveforms are not reliable for patient monitoring. When the heart rate can be as high as
90 beats/min under routine invasive monitoring situations, the peripheral pulse is an
unreliable guide to stroke volume and cardiac work. This assertion becomes more valid
when the heart rate is high and this amplification of the components of the peripheral
pressure pulse becomes as much as 200 per cent of the central pulse (Figure 7.5).
161
QUANTIF ICATION OF THE INTENSITY OF REFLECTED WAVES IN THE UPPER L IMB The upper limb pressure pulse is used in every aspect of patient management and for
diagnosis and treatment of hypertension. Yet in nearly all of these circumstances, the
effect of wave reflection in the upper limb arteries as a determinant of arterial blood
pressure has never been addressed properly. There is a lack of documentation of the
origin of reflections in the upper limb and quantification of relative intensities of
reflected energy. This chapter attempts to document the effect of reflected waves in the
upper limb pressure waves by analysing the pressure waves recorded in the central
(Carotid) and peripheral (Radial) arteries.
It was predicted in Chapter 5 that the systolic part of the upper limb pressure wave is
contaminated with reflected waves originating from upper limb terminations. Being
CHAPTER 8
162
different from the ones in the central aortic waveform they are responsible for peak
systolic pressure in the upper limb.
In the following experiment, these predictions were investigated by altering the intensity
of reflected waves originating from the palm of the hand by altering the impedance
match at this level. Reflection coefficient at this termination were altered by
modifications of terminal impedance by graded compression of the palm vasculature
and/or the characteristic impedance by raising the arm. This in turn resulted in
alteration in the intensity of reflected waves. To estimate the degree of wave reflection,
the same upper limb model employed in Chapter 7 was also used. By predicting
pressure wave propagation parameters and by utilisation of non-invasive techniques to
measure blood pressure waveforms, this model enabled the intensity of reflected waves
to be calculated without the need of employing complex flow measurements.
8.1 METHODS
8.1.1 Subjects
Thirty-three subjects who had undergone routine coronary artery bypass surgery were
recruited from the Cardiothoracic Department, St. Vincent's Hospital (Table 8.1). None
had peripheral vascular disease or arterial stenosis. They were screened for
hypertension (mean arterial pressure <120 mmHg) and hypotension (mean arterial
pressure >60 mmHg). The study was approved by the research ethics committee of the
hospital and all patients gave consent prior to the study.
163
Mean±SEM Range Age (year) 58.2±1.6 42-74 Sex 25 Males and 8 Females Weight (kg) 80.1±2.4 52-102 Height (cm) 171±1.7 148-1.86 MAP (mmHg) 84.1±2.1 63-111 Heart Rate (beats / min) 93.1±2.05 75-122 Type of Operation 26 CAG, 2 MVR, 4 AVR, 1 ASD Time after Operation (Hours) 6:08±0:26 1:45-11:00 Temperature (°C) 37.2±0:26 32.8-38.8
Table 8.1 Summary data for patients enrolled in the study ( MAP = Mean arterial pressure, MVR = Mitral valve replacement, AVR = Aortic valve replacement, ASD = Atrial septal defect, CAG = Coronary artery graft)
8.1.2 Procedure and Data Acquisition
Changes in the terminal load were introduced through graded occlusion of the hand
vasculature by means of a sphygmomanometer cuff. By increasing the pressure in the
cuff, the physical properties of the underlying arterial bed are modified due to
transmission of this external pressure to the arterial wall and hence decreasing the wall
tension. This decrease in wall tension introduces changes in the lumped characteristics
of the distal arterial bed. The changes in the proximal characteristic impedance were
introduced by raising the arm analogous to the cuff occlusion manoeuvre, the reduction
in mean arterial pressure caused by raising the arm decreased the transmural pressure
and thus increased the compliance of the proximal vessels and decreased the
characteristic impedance, since both are inversely related.
The procedure involved wrapping an adult-size sphygmomanometer cuff around the
hand of the arm which had an indwelling radial artery catheter. While the central
pressure waveform was being registered, the cuff was rapidly inflated to 200 mmHg. It
was then deflated slowly (2 mmHg/sec) to zero pressure. This cycle of inflation and
164
deflation was performed three times to allow for data averaging. The same procedure
was repeated while the arm was raised from its baseline level to reduce the registered
mean arterial pressure by about 15 mmHg.
As a routine procedure, ECG (5-lead) and invasive arterial pressure were used to
monitor the heart rate and blood pressure via a bedside monitor (Spacelabs-PC). The
amplifier of this monitor has an output of 1 volt/100 mmHg. The pressure monitoring
catheter system was composed of 220 cm polyethylene tube attached to a disposable
radial arterial cannula (18-20 gauge) and disposable strain gauge transducer. Although
the natural frequency of this system was more than 60 Hz, the attachment of catheters
and connectors greatly reduced this. The investigation of the actual and achievable
frequency response of a typical system was performed by sinusoidal excitation of the
recording assembly with a function generator (Hewlett Packard 3311A) connected to a
pressure generator (Biotek model 501A). The pressure waveforms obtained from this
bench top experiment were subjected to frequency analysis to estimate the amplification
and phase delay (Figure 8.1). The transfer function obtained this way was later used to
correct the manometer system by inverse filtering.
Figure 8.1 The frequency response of the catheter system including the amplifiers and the digital filters.
165
Before any measurements were taken, the catheter was checked and flushed with sterile
normal saline (0.9% wt/vol) to minimise trapped air bubbles and blood clot. An
estimation of frequency response of the catheter system was then performed by the
modified pop test. This test involves tapping the end of the catheter and estimating the
period of the oscillations. Whenever the estimated oscillation was below 20 Hz the
system was considered to be unsuitable. Subsequent flushing and eliminating the
bubbles consistently improved the characteristics of the system. The manometer was
then moved from the rack mounting the entire transducer assembly to the cannulation
site. This process allowed the registration of blood pressure at the radial arterial level
rather than usual heart level. Since the entire measurement took 15 minutes to
complete, routine blood pressure monitoring was performed in the contralateral arm
using an oscillometric technique.
Applanation tonometry was used to record the blood pressure waveform in the carotid
artery (Central). Since it was impossible to predict the calibration of these non-invasive
carotid tracings (due to lack of knowledge of exact applanation pressure and
orientation), they were calibrated by equating the mean and diastolic pressures of the
invasive radial and carotid tracing. This technique assumes that the mean arterial
pressure is equal throughout the arterial tree and at diastolic pressure level the effects of
reflected waves are minimal and thus pressure values can be considered to be
approximately similar in every branch (see Chapter 5). Inevitably this technique led to
errors when the arm was raised, but appropriate change of mean pressure due to this
manoeuvre was taken into account when calibrating these recordings. A separate
tonometer which was calibrated previously was attached to the sphygmomanometer cuff
connectors in order to monitor instantaneous cuff pressure during the procedure
All pressure and ECG signals were digitised on-line with an IBM compatible computer
using an A/D board (DT 2801-A, Data Translation, Mass). Acquired data (128 Hz, 12
bits per channel) was temporarily stored on the hard disk and later transferred to a
166
WORM (Write Once Read Many) drive (Maxtor 800S) for permanent storage. Using
software developed for this purpose data segments of 8 second length during the
manoeuvre were averaged using ECG triggering. The data were then subjected to
transfer function analysis (see Chapter 2) and expressed with respect to transmural
pressure which is defined as
crt PPP −= (8.1)
where Pt, Pr and Pc are the mean transmural, radial and cuff pressures during the pulse
period.
8.1.3 Estimation of Intensity of Reflected Waves
Although the effects of reflected pressure waves on the formation of pressure waves are
well documented (Nichols and O'Rourke 1990; O'Rourke et al 1992) quantification of
wave reflection still presents major challenges. The amount of wave reflection may be
quantified by measuring pressure and flow at a site invasively, and by applying the
equations given in Chapter 2. Historically, these equations have been used either in the
ascending aorta to estimate the amount of global reflections as seen from the heart
(O'Rourke and Taylor 1967a) or in large conduit arteries to quantify the reflections from
major beds (O'Rourke and Taylor 1966; Li et al 1984). However, investigations
directed towards the applicability of these equations in small conduit arteries have been
scarce, not only because of the perceived lack of importance of reflected waves but also,
because of the difficulties associated with flow measurement techniques in small calibre
peripheral arteries.
167
(i) Estimation from Phase Velocities
An estimate of complex reflection coefficient is derived from the phase velocities
similar to one described in chapter 7 is obtained. This equation is analogous to equation
2.148 and described as
∞
∞+ +
−=Γ
cccc
i
i0 (8.2)
Where ci is the pulse wave velocity at frequency 'i' and c∞ is the foot-to-foot estimate
of true wave velocity obtained from the pressure waves. This technique assumes c∞ to
be proportional to Z0, characteristic impedance, and ci to be proportional to the
terminal impedance, ZT, value at frequency 'i'. The pulse wave velocity at very long
wavelengths, c0 , or ones close to zero frequency, c0+ , is considered to be related to the
peripheral resistance values. The c0+ , the phase velocity at very low frequencies is
estimated by linear extrapolation of phase velocities of second and first harmonic to
zero. This technique yields values similar to that obtained with parameter estimation
method (see Chapter 7).
(ii) Estimation from Pulse Pressures
The pulse pressure amplification, AMPPP , between central, PPc is the central pulse
pressure, and peripheral sites, PPr , radial pulse pressure, is also calculated as an index
of reflections
100×=c
rAMP PP
PPPP (8.3)
(iii) Estimation from the Model
Another set of reflection coefficient estimates were obtained by the parameter
estimation technique similar to the one performed in Chapter 7. Briefly, a parameter
168
estimation technique similar to the one performed in elastic tube experiment (Chapter 5)
is employed in a reduced model of the upper limb involving few branches (Table 8.2).
Input to this model is considered to be the carotid pressure waveform. Similar to elastic
single tube experiment, model output is compared against the in vivo measured radial
waveform while model parameters, Em , wall viscosity, θm , reflection coefficient, Γm ,
and time constant, τm , are continuously being altered. The model values representing
the best fit between these two pressures are considered to be the values representative
for the upper limb circulation.
B-Tree Code
NAME L (cm)
D cm
h (cm)
E (x106 ) (Dyne/cm2)
Θ (Degree)
Γ (units)
τ (Sec)
M Asc. Aorta 0.10 0.94 0.08 Em θm Γm τm MR Innominate 9.10 0.94 0.08 Em θm Γm τm MRR Brachial 22.0 0.60 0.18 Em θm Γm τm MRRR Radial 41.0 0.34 0.16 Em θm Γm τm MRRRR Termination 0.10 0.34 0.16 Em θm Γm τm
Table 8.2 Reduced model of the upper limb used in the estimation of transmission characteristics. The model parameters Em , θm , Γm and τm are estimated by the best fit between recorded and model generated pressure waves.
8.1.4 Statistics
Differences of indices from the zero transmural pressure control state are analysed
using Student's paired t-test. A p value of less than 0.05 is accepted statistically
significant. Data are presented as mean ± standard error of mean unless otherwise
stated.
8.2 RESULTS
A typical run obtained during the study on one patient is shown in Figure 8.2.
169
Figure 8.2 A representative tracing obtained in one subject showing the radial (Solid), Carotid (Dotted) pressure waveforms and transmural pressure calculated using equation 8.1. Each pulse in the tracing is the average of 8 second recordings of three runs appended with respect to transmural pressure. (Negative transmural pressures indicates cuff pressures are higher than the intra-arterial mean pressure).
CAROTID RADIAL
Figure 8.3 Representative tracings obtained in one subject showing the carotid and radial pressure waveforms against transmural pressure (TP) in the hand. Arterial pressure data (AP) are stacked and interpolated to obtain continuous values for different transmural pressures (TP).
170
CONTROL
TP (mmHg)
PP (mmHg)
PPAMP (%)
c∞ (m/sec)
c1 55. (m/sec)
c0+ (m/sec)
Γ0+ (Units)
-100.00 52.89±1.63 126.10±2.77 780.23±13.80 926.13±53.58 1138.51±98.72 0.14±0.03 -80.00 51.36±1.64 125.52±2.37 797.92±14.77 904.98±34.05 1117.81±62.65 0.13±0.02 -60.00 51.98±1.77 120.87±2.68 782.42±12.71 835.32±28.42 998.33±49.79 0.10±0.03 -40.00 47.76±1.47 120.00±2.63 798.05±13.37 813.32±28.42 955.73±46.02 0.03±0.03 -20.00 48.12±1.46 115.00±2.16 782.57±12.15 773.31±22.85 891.75±36.40 0.00±0.02 0.00 45.14±1.61 112.20±1.88 808.61±14.18 737.71±20.49 820.50±29.96 0.00±0.02 20.00 42.53±1.53 109.37±1.92 796.90±11.88 704.80±18.30 772.93±30.30 -0.03±0.02 40.00 45.51±1.70 114.09±2.03 801.49±15.43 763.78±23.39 862.60±30.96 0.03±0.02 60.00 44.62±1.26 118.76±2.09 784.84±12.18 776.31±22.40 894.06±33.49 0.05±0.02
ARM RAISED
TP (mmHg)
PP (mmHg)
PPAMP (%)
c∞ (m/sec)
c1 55. (m/sec)
c0+ (m/sec)
Γ0+ (Units)
-100.00 49.89±2.14 119.90±3.21 765.30±15.88 775.99±36.71 916.55±58.10 0.07±0.03 -80.00 49.14±1.99 121.96±2.99 766.27±13.22 798.49±27.22 952.85±53.75 0.07±0.03 -60.00 48.38±1.79 117.37±2.35 769.34±11.76 775.48±28.48 924.91±47.24 0.07±0.02 -40.00 46.32±1.44 115.30±2.56 797.95±13.18 767.03±25.91 897.41±44.30 0.04±0.03 -20.00 43.82±2.02 109.74±2.33 777.21±12.18 699.01±28.12 794.02±42.11 -0.01±0.02 0.00 39.97±1.52 108.61±2.39 774.23±16.99 671.74±20.11 729.36±29.99 -0.06±0.03 20.00 39.99±1.66 107.68±2.36 772.86±13.54 662.09±21.15 702.95±32.42 -0.05±0.02 40.00 42.42±2.28 107.93±2.36 771.45±14.06 654.00±17.37 703.13±22.19 0.01±0.02 60.00 43.79±1.57 113.78±1.65 789.52±11.81 718.12±15.25 823.79±24.25 0.06±0.02
Table 8.3 The effect of transmural pressure on wave reflection indices under control conditions (Top) after raising the arm (Bottom). (Tp = transmural Pressure; PP = Pulse pressure; PPAMP = Pulse pressure amplification between carotid and radial sites; c∞ , c1 55. and c0+ are the foot to foot, at 1.55 Hz and at near zero hertz components of the phase velocities respectively; Γ0+ is the reflection coefficient estimates from
Equation 8.2. (Data are given as mean ± standard error of the estimate).
During the cuff deflation period, ie increase in transmural pressure, the amplitude of
radial pulse pressure initially decreases. It later increases when the cuff pressure falls
below the intra-arterial level. Since no simultaneous pressure swing is observed in the
carotid waveform, this phenomenon is presumed to be local to the upper limb. This is
171
further confirmed after averaging 8 second recordings of carotid (Figure 8.3, Left) and
radial pressure waveforms (Figure 8.3, Right) and expressing them with respect to
transmural pressure in the hand.
The radial systolic pressure decreases initially to reach a saddle point at around zero
transmural pressure then increases subsequently with an increase in transmural pressure
(Figure 8.3, Right) without any significant change in the carotid waveform (Figure 8.3.
Left). This characteristic change in radial pressure waveform was reproducible in all
patients (Table 8.3).
When individual pulse pressure data are pooled against transmural pressure in the hand,
the minimum pulse pressure near zero transmural pressure becomes evident in both
control (Figure 8.4, circles) and arm raised (Figure 8.4, squares) conditions.
PULSE PRESSURE
Figure 8.4 The effect of transmural pressure in the hand on radial pulse pressure under control (circles and solid line) and arm raised conditions (squares and dotted line). The right most points represent the baseline conditions. Data are pooled in bins of 20 mmHg and averaged, bars represent ± 1 standard error,* p<0.05, ** p<0.0001).
172
There is a lesser degree of change in the radial pulse pressure during arm raise than
during control. This is due to different baseline mean arterial pressures (83.11±2.31
mmHg versus 63.73±1.58 mmHg, p<0.0001) and pulse pressures (50.79±1.96 mmHg
versus 44.36±2.61 mmHg, p<0.05). At zero transmural pressure both pulse pressures are
reduced (control 45±1.61 mmHg, raised 39.97± 1.52 mmHg).
PPAMP
Figure 8.5 Changes induced by cuff occlusion on the pulse pressure amplification in the upper limb under control (circles, solid line) and arm raised conditions (circles, solid line). The right most points represent the baseline conditions. Data are pooled in bins of 20 mmHg and averaged, bars in the graph represent ± standard error, * p<0.05,.
Since changes in the actual values of the radial pulse pressure might be contaminated by
the input values (that is alteration in central pulse pressure), they are expressed in terms
of pressure amplification using Formula 8.3 (Figure 8.5). Under control conditions
(Circles) the AMPPP has a value of 125.41±1.12% which decreases to 112±1.88%
(p<0.001) at zero transmural pressure and increases back to 126.10±2.77% at -100
mmHg (p = NS). Raising the arm reduces the AMPPP to 116.32 ± 1.3% with respect to
control (p<0.05) while change in transmural pressure affects the AMPPP the same way
173
as control (108.61±2.39% at 0 mmHg, p<0.0001 and 119.9 ± 3.21% at -100 mmHg
p<0.05) (Squares).
PHASE VELOCITY
CONTROL ARM RAISED
Figure 8.6 Changes induced by cuff occlusion on the phase velocity estimates in the upper limb under control (left) and arm raised conditions (right). The foot-to-foot velocity estimates, c∞ (circles) are constant across different transmural pressures both in the distal bed (abscissas) and the proximal conduit arteries (left and right). The first (c1.55, squares) and the extrapolated zero harmonic (c0+, diamonds) show the maximum variation to both pressure changes. The right most data points in both graphs represent the baseline conditions. Data are pooled in bins of 20 mmHg and averaged, bars represent ± standard error of mean.
The phase velocities at the first harmonic (1.55 Hz) and those extrapolated for the zero
harmonic for both control and arm raised display similar patterns (Figure 8.6). The
foot-to-foot wave velocity estimate, c∞, remains the same both in response to cuff
occlusion and when the mean pressure in the arm is lower (Figure 8.6 ). The first and
extrapolated zero harmonics are greater with respect to the condition where the arm is
raised (Figure 8.6 Right) indicating a higher mismatch between the proximal and distal
174
beds (see below). The distal bed impedance which is proportional to the phase velocity
(Chapter 2) at baseline, estimated from the extrapolated zero harmonic, c0+, is 1.3 times
to the proximal beds characteristic impedance, c∞, under control conditions. Raising
the arm reduces this to 1.1 times that of c∞.
The reflection coefficient estimates calculated at extrapolated zero harmonic phase
velocities using equation 8.2 show changes parallel to that of the pressure pulse
amplification data (Figure 8.7). The reflection coefficient estimate, Γ0+, follows
changes in the PPAMP during control (circles) and arm raised (squares) conditions. It is
lower when the arm is raised than during control and it has the exactly same saddle
point at around zero transmural pressure. This near zero values of the reflection
coefficient estimate at around zero transmural pressure indicates a better match than that
for baseline.
Reflection Coefficient Estimate (Γ0+)
Figure 8.7 Changes induced by cuff occlusion on the reflection coefficient estimate in the upper limb under control (Circles) and arm raised conditions (Squares). The changes due to arm raising and the hand occlusion are in parallel with the ones derived from the pulse pressure amplification (see Figure 8.5). The right most points represent the baseline conditions. Data are pooled in bins of 20 mmHg and averaged, Bars in the graph represent ± standard error.
175
These parallel changes are further confirmed by the linear regression between the pulse
pressure amplification to that of the reflection coefficient estimate (Figure 8.8). There is
a direct relation between the pulse pressure and the reflection coefficient estimate
derived from the phase velocity spectra. Although both are derived in different domains
( AMPPP in time domain and Γ0+ in frequency domain) the regression equations between
the two were y=83.78x+113, R2=0.72 (p<0.0001) for under control (Figure 8.8, Left)
and y=84.37x+112.7, R2=71 (p<0.0001) for arm raised conditions (Figure 8.8, Right).
ALL PATIENT DATA
CONTROL ARM RAISED
Figure 8.8 The regression between the reflection coefficient estimate, Γ0+ , and the pulse pressure amplification, AMPPP , under control (Left, y=83.78x+113, r2=0.72,
p<0.0001) and arm raised conditions (Right, y=84.37x+112.7, r2=71) in the upper limb.
On the average one hundred iterations were required by the model to obtain a regression
coefficient of R2=0.98±0.02 for each pair of pressure pulse (Carotid and Radial). The
summary data for model estimated elastances, wall viscosities, reflection coefficients
and time constants are given in Table 8.4 for each transmural pressure.
176
CONTROL
TP (mmHg)
Em (x106 ) (Dyne/cm2)
θm (Degrees)
Γm (Units)
τm (Seconds)
-100.00 1.21±0.06 23.69±1.88 0.53±0.05 0.13±0.01 -80.00 1.25±0.05 23.88±2.24 0.43±0.08 0.15±0.02 -60.00 1.17±0.07 28.03±2.90 0.51±0.06 0.18±0.03 -40.00 1.19±0.07 26.86±1.86 0.43±0.06 0.15±0.02 -20.00 1.31±0.11 24.65±2.58 0.40±0.05 0.17±0.02 0.00 1.14±0.07 31.94±1.95 0.36±0.04 0.23±0.02 20.00 1.20±0.08 28.61±2.18 0.34±0.05 0.20±0.02 40.00 1.12±0.06 28.90±2.54 0.31±0.04 0.20±0.01 60.00 1.13±0.08 24.31±1.64 0.40±0.05 0.18±0.03
ARM RAISED
TP (mmHg)
Em (x106 ) (Dyne/cm2)
θm (Degrees)
Γm (Units)
τm (Seconds)
-100.00 1.16±0.06 24.93±2.17 0.57±0.05 0.15±0.02 -80.00 1.16±0.06 26.15±2.07 0.51±0.05 0.17±0.02 -60.00 1.16±0.08 27.35±2.34 0.42±0.07 0.17±0.02 -40.00 1.12±0.07 27.60±1.79 0.45±0.05 0.17±0.02 -20.00 1.17±0.09 28.01±2.21 0.47±0.05 0.18±0.02 0.00 1.11±0.07 29.48±1.84 0.37±0.05 0.18±0.02 20.00 1.12±0.08 29.59±2.03 0.36±0.03 0.21±0.02 40.00 0.99±0.06 33.35±2.02 0.40±0.03 0.22±0.02 60.00 1.15±0.08 25.30±1.65 0.37±0.05 0.16±0.02
Table 8.4 Model parameters, Em , θm , Γm and τm estimated from the pressure waves recorded in the radial artery and in the carotid artery (Mean±SEM). Correlation coefficients, R2, for each fit is in excess of 0.98.
Although the estimated reflection coefficients Γm were similar, the elastance Em values
were lower and the wall viscosity θm values were higher than those calculated in
Chapter 7. Similarly, the time constants τm were higher than those in the previous
investigation (Chapter 7). Raising the arm had little effect on each of value
177
investigated. However, in each of the transmural pressures the pulse pressure
amplification is linearly related to model parameters:
for control
mmmampPP θτ ×−×−Γ×+= 01.002.120.038.1 , R2=0.66, p<0.0001
for arm raised
mmmampPP θτ ×−×−Γ×+= 01.083.022.039.1 , R2=0.75, p<0.0001
8.3 DISCUSSION
Often systolic pressure and/or rate of rise in pressure in the upper limb is taken as an
index of myocardial contractility, and pulse pressure as an index of stroke volume (
Pauca et al 1989). However, this study indicates that these indices are definitely
dependent on the distal bed properties specifically the palm vasculature. Previous
studies found that systolic pressure in the radial artery is dependent on heart rate
(O'Rourke 1970). The compression applied to the radial artery distal to the cannulation
also effects the pressure pulse amplification (Pauca et al 1989). This second effect is
not surprising in the light of earlier studies (O'Rourke and Taylor 1967a; Van den Boss
et al 1977; Newman et al 1979). The occlusion of any artery creates a positive
reflection site which is close to unity which in turn augments the reflected waves.
This present study is significantly different from earlier studies. Since the procedure
involved partial occlusion of the arteries under the cuff, it was possible to create a whole
range of terminal impedance mismatches. Intuitively, one would expect this mismatch
178
to be proportional to cuff pressure (higher cuff pressures introducing higher
mismatches). However, the outcome is more complex. The pulse pressure reduces
initially with increase in compression (Figure 8.3). It then increases beyond a minimal
point with further increase in the compression of palm vasculature.
These seemingly paradoxical findings can be explained, however, if one uses transmural
pressure rather than the cuff pressure to express the relation (Figure 8.9). It follows that
at some intermediate cuff pressure the transmural pressure approximates to zero. At this
small positive transmural pressure the wall stress approximates to zero and the vessel
wall becomes more compliant (point B in Figure 8.9) due to non-linear elastance of the
wall (Gow and Taylor 1968). If a modified Windkessel for the palm vasculature is
assumed first order approximation, increased compliance decreases the impedance of
this bed. Given that the proximal properties of the radial artery remains unchanged as
there is no change in pulse wave velocity, the decrease in terminal impedance leads to a
decrease in impedance mismatch. By equation 2.93 this decreases the reflection
coefficient at this point. This will in turn reduces the intensity of reflected waves which
often constitute a major part of the systolic upstroke.
With further increase in external pressure however, the wall stress becomes compressive
and both the intraluminal area of the vessels in the bed and its compliance decreases
(point C in Figure 8.9). As a result, the impedance of the bed will increase both due to
decrease in parent-daughter diameter ratio (Taylor 1966b) and the increased Windkessel
impedance. This in turn increases reflected waves and augments pulse pressure (Figure
8.5).
179
Figure 8.9 The non-linear transmural pressure-volume curve of the arterial bed under the hand cuff. The point A represents the normal distending pressure of the bed where the compliance is lower and thus the reflection coefficient is higher. Increase in cuff pressure decreases the transmural pressure reaching to zero at point B where the bed becomes extremely compliant and the reflection coefficient is at a minimum. Further increases in cuff pressure decrease the volume further, asymptotically approaching a constant value. At those pressures (point C) the bed is again less compliant and thus higher reflections can be expected
These conclusions are confirmed by the estimated Windkessel time constant, τm, which
relates the compliant element to the resistive element (Chapter 5). There was a
significant increase in time constant as determined by the model accompanied by a
decrease in reflection coefficient and increase in wall viscosity at zero transmural
pressure levels (Table 8.4). While deriving this parameter, it was assumed that the
elastic properties of the wall are exponentially related to strain and the effective
resistance of the lumen is constant. Therefore, when all these changes are taken into
account it becomes apparent that the actual change in compliance is underestimated by
this technique.
Interestingly, the pulse pressure attained by maximum compression or vascular
occlusion, is close to the values under control conditions. This indicates that the
impedance of palmar bed is close to the values attained by maximum vasoconstriction.
180
Since the pressure volume curve is S-shaped (Figure 8.9) and under normal
physiological conditions the arteries are positively stressed, the compliance values are
close to those obtained by vascular occlusion (point A versus C in Figure 8.9). Thus, by
altering the transmural pressure in the palm one can scan the pressure volume (stress-
strain) relation of the vessels involved where the effect will be visible from the changes
in pulse pressure or systolic rate of rise.
Under normal circumstances however, pulse pressure, systolic pressure or rate of rise
may not be a quantifiable parameter for estimation of reflection coefficient. This is
because factors other than the vessel wall properties affect these parameters, such as
stroke volume and systemic input impedance. One way to overcome this difficulty is by
expressing the relation in terms of the pulse pressure amplification (Equation 8.3). As
expected, this technique yields similar results (Figure 8.4 vs Figure 8.5).
The phase velocity plot spectrum is similar to the input impedance spectrum and takes
into account the presence of reflected waves (Chapter 2 and Figures 8.3-4). Since this
technique does not require calibrated pressure waveforms it forms an ideal case for non-
invasive systems that are capable of recording pulse waveforms. This study shows the
applicability of this technique to estimate the intensity of wave reflection using
tonometers (Figure 8.6 and Figure 8.7). Estimation of the zero frequency intercept was
pursued in this study to account for changes in heart rate in different individual subjects.
Obviously one could derive reflection coefficient estimates at each frequency using
equation 8.2 but this was necessary for comparisons to be made amongst different
individual subjects with different heart rates. Pulse pressure amplification and the
reflection coefficient estimates obtained this way are closely related (Figure 8.8), an
indication that they are compatible. An inspection of the regression equation reveals
that at the unity value of reflection coefficient estimate, the pulse pressure amplification
reaches 196% which is approximately the theoretical limit of 200%.
181
The values obtained for reflection coefficient estimates by this technique (Table 8.3 and
Figure 8.7) are far less than the ones reported in the literature (O'Rourke 1967). This
difference can be explained in terms of the vasoactive state of the patient group that was
studied and the assumptions made for the equation employed. Nearly all the patients
were receiving vasoactive drugs, nitroglycerine (5 mg/hour) and sodium nitroprusside (5
mg/hour) as a routine after the operation, through which the reflections are expected to
be lower (O'Rourke et al 1992). Although they were allowed to warm up (37.1°C)
before performing the experiments (they were investigated at an average of 6:08 hours
after surgery), they were still hypotensive (MAP = 83 mmHg) and vasodilated, pulse
wave velocity estimates for the arm in this patient population (Table 8.3) were much
lower than the ones reported in the literature (Avolio et al 1983b, 1985) (780 cm/sec vs
1060 cm/sec).
Methodological errors in estimation might also be introduced due to the shape of the
phase velocity spectra and the sampling interval. It was assumed that the phase
velocities are proportional to the frequency thus a linear extrapolation technique was
used for the whole range of frequencies. In fact a linear relationship is more relevant for
these frequencies. However, when exponential extrapolation rather than linear
extrapolation was attempted for this purpose, the reflection coefficient estimates
differed as little as 20%. Another problem might arise due to the presence of minima in
the phase velocity spectrum. The extrapolation technique would produce errors if the
first and second harmonics were separated with a minimum value. This would
introduce an under estimation of the zero intercept. Since the patients recruited for this
study had often a higher heart rate, 93±1 beats/min (Table 8.1), the second harmonic
was at 3.1 Hz. The position of the minimum estimated using quarter length formula
(Equation 2.138) yields a value of 3.3 Hz which is not significantly higher than the
second harmonic. Nevertheless, even under these restrictions equation 8.2 still yields
values of reflection coefficient that change in the appropriate direction with pulse
pressure amplification.
182
The purpose of raising the arm was to decrease the proximal characteristic impedance
and thus increase reflected energy. This manoeuvre caused a minor decrease in pulse
wave velocity in the proximal vessels, which did not reach statistical significance (60
cm/sec, p = NS). Moreover it is a recognised method to decrease the mean pressure in
the distal bed (Warnes et al 1983). This obviously altered the entire relation between
the proximal and distal vessels to an unknown point. Yet, when the cuff inflation was
introduced and results were expressed in terms of transmural pressure, compensation
was expected for the effect of arm raising on properties of distal vessels. However,
there was a slight but consistent decrease in apparent reflection coefficient which
warrants further study (Figure 8.8).
The general arterial model applied to the upper limb circulation was used to simulate
experimental conditions. The model predicted values for reflection coefficients that
were comparable with those estimated from phase velocity plots but were often higher.
Furthermore it was also possible to simulate changes in the compliance of the distal
beds by the time constant parameter of the model. The correspondence between the
model derived and the actual radial waveform was also satisfactory. The model
parameters explained 98 per cent of the features of the radial pressure waveforms. The
remaining 2 per cent could be introduced by experimental noise or the arbitrarily chosen
diameters and lengths or by the inherent assumptions made in the mathematical
abstraction. Nevertheless, the iteration technique helped to estimate parameters of
physiological significance within a reasonable iteration time. This technique can
therefore be extended to real situations.
In general, the pressure waves travelling in the upper limb are reflected back due to a
major reflecting site residing in the palm. This bed often exhibits a higher load for the
radial artery under extreme conditions, vasodilation and maximum compliance, thus
contributing to the formation of the upstroke of the pressure wave. This property of this
specific circulation, that is mostly a positive reflection coefficient led previous
183
investigators to believe that the upper limb behaves like a second order system (Warner
1957; Gardner 1982; Chadwick et al 1986; Schwid et al 1987). For this kind of
approximation to be valid, the terminal load has to be higher than the proximal load. As
shown in the present study, this condition is satisfied even under the extreme
vasodilatory state of these patients. Yet, when there is a change in the transmural
pressure, for example by raising or lowering the arm, the terminal load changes its
characteristic, ie. the lumped parameters of the second order system. The damping
coefficient and resonant frequency of the second order system will reflect these changes.
However, it will not convey the information necessary to interpret the actual
phenomenon.
The transmission line analogue, however, explains these changes correctly. It will point
out the changes in the terminal load as deduced from the reflection coefficient. When
properly estimated from modelling studies, this terminal load characteristic opens new
ways of assessing passive changes from the active ones. Since the transmission line
analogue does not require the absolute amplitude to determine phase velocities, only
high fidelity recording systems are required. Tonometry or other high-fidelity non-
invasive techniques are ideal for this purpose.
This study also implies that one can predict the mean arterial pressure in the radial artery
by a simple manoeuvre as analogous to the oscillometric pressure measurement
technique (Posey et al 1969, Ramsey 1976). After plotting the pulse pressure in relation
to the cuff pressure, mean arterial pressure should correspond to the value of cuff
pressure when the pulse pressure becomes a minimum. When it is not possible to obtain
absolute pulse pressure, this technique can use the relative upstroke of the pressure
waveform. This index should also decrease when cuff pressure is at or close to the
mean arterial pressure.
184
It is also possible to plot stress-strain relation of the arterial bed under the cuff as given
in Figure 8.6, by plotting the reflection coefficient against transmural pressure or cuff
pressure. This will then display the terminal impedance mismatch that can be converted
to the elastance of the bed by simple assumptions. Using this technique, one can then
easily document the effects of drugs and aging on this specific vascular bed.
185
EFFECTS OF VASODILATORS ON UPPER L IMB WAVE PROPAGATION PROPERTIES Vasoactive agents alter the systemic input impedance and thus the left ventricular
afterload in a favourable direction (Nichols and O'Rourke 1990). They affect different
parts of the vasculature both at the level of conduit or cushion arteries, the arterioles or
the veins. These drugs are designed to induce changes on the vessel wall by directly or
indirectly acting on the smooth muscle. By causing changes in vessel wall properties -
diameter, thickness and elastance - these drugs may alter the wave propagation
properties of the vessel. This can become extremely important if they interfere with the
pressure wave transmission in the upper limb where blood pressure is routinely
monitored. In current clinical practice, the implications of wave transmission in the
upper limb are generally overlooked not only under normal conditions (see Chapter 7)
but also under the effect of these drugs. Recent reports on the discrepancies between the
CHAPTER 9
186
central and upper limb pressure measurements before and after the administration of two
very potent vasodilators, nitroglycerine and sodium nitroprusside (Kelly et al 1990b,
Simkus and Fitchett 1990) do not deal with this phenomenon. Although these
investigators attempted to explain these observations by differences in the wave shape in
the arteries, neither a comprehensive explanation of the phenomenon nor a description
of the effects of these drugs on the upper limb wave propagation properties is provided.
In the following experiments, an attempt will be made to quantify the pressure wave
transmission in the upper limb under control conditions and after using nitroglycerine
(NTG) and sodium nitroprusside (SNP). Using techniques developed in previous
chapters parameters of and upper limb model will be calculated. Comparisons will then
be made between the experimental findings and the model predictions made earlier in
Chapter 5.
9.1 METHODS
9.1.1 Nitroglycerine (NTG)
Since it was possible to obtain the original experimental data that appeared in the
original publication (Kelly et al 1990b) thanks to the courtesy of the authors, no attempt
was made to perform the same experiments under identical conditions. The data
comprised the patients' age, sex, and morphometric measurements, digitised recordings
of the invasive micromanometric pressure (the ascending aorta and brachial artery) and
the non-invasive tonometric (Radial ) pressure waves before and after NTG. Calibration
signals were also available for all signals.
187
Analysis of the upper limb transmission properties was performed on averaged
waveforms by relating the harmonic components of the ascending aortic - brachial
BAAAH − and ascending aortic - radial RAAAH − pairs. The transfer functions were
expressed in relation to frequency in modulus and phase format. Because of the
differences in heart rates, the transfer function values were not obtained at similar
frequencies. Thus, to obtain representative spectra for the control and for the NTG
administration, the spectra were clustered into bins of fundamental harmonics derived
from mean heart rates. Averaging the contents of these bins and performing linear
interpolations between these mean values yielded representative spectra for each case.
Comparisons using Student's t-test were also performed on these bins to estimate the
degree of variation due to the effect of NTG.
A model of the upper limb arterial vasculature for each patient was also constructed to
estimate the physical parameters responsible for the measured pulses. This was done by
employing the same approach as was described in Chapter 7. Briefly, the reduced
model of the upper limb arterial system was constructed for each patient (Table 7.1).
Input pressure waves were supplied to this model was that recorded in the central artery
(ascending aorta). The model was then iterated with different elastance, Em , wall
viscosity, θm , reflection coefficient, Γm , and time constant, τm , values to yield a best
fit between the waveform derived from the model and measured distal pressure
waveform (brachial or radial). Values for wall elastance and terminal reflection
coefficients corresponding to the minimum of the sum square difference between
calculated and measured distal waveform are regarded as the model parameters.
An estimate of the complex reflection coefficient is derived from the phase velocities
similar to that described in Chapter 7. The resonant frequency of the upper limb is also
estimated using the technique given in Chapter 7. The effect of NTG on these
parameters calculated from the transfer function and from the model were analysed.
188
9.1.2 Sodium Nitroprusside (SNP)
The data for this agent were obtained from the original publication of the study (Simkus
and Fitchett 1990). The aortic and radial tracings in the publication were digitised using
a digitising tablet (Summasketch, Summagraphic). Since this process involves a noise
level higher than expected due to digitisation by hand, it is taken not to include the noisy
part of the waveform. This was done by excluding the frequency components that fall
below a noise threshold which was estimated by averaging components 9-16 hz.
Because of the absence of the morphometric data for the subjects reported in this
publication, a constant length, 62 cm, was used for ascending aorta to radial artery. This
figure was subsequently used in the modelling part of the study. Apart from these
changes made, the analysis of the waveforms and modelling of the upper limb
proceeded as in the case of NTG.
9.1.3 Statistics
Differences from the control state are analysed using paired t-test. A p value of less than
0.05 is accepted statistically significant. Data is presented as mean ± standard error of
mean unless otherwise stated.
189
9.2 RESULTS
9.2.1 Nitroglycerine
Table 9.1 summarises the parameters that were calculated and derived from the model
for each patient. Since neither the estimated values of time constant nor the wall
viscosity are changed, these values are not reported.
Control
NTG
Patient. No
fn (hz)
Γ0+ (Units)
Em (x106 dyne/cm2)
Γm (Units)
fn (hz)
Γ0+ (Units)
Em (x106 dyne/cm2)
Γm (Units)
1 2.58 0.22 1.12 0.56 1.50 0.24 * * 2 1.79 0.35 1.01 0.78 3.76 0.77 0.85 0.67 3 2.48 0.52 1.35 0.67 1.98 0.43 0.78 0.78 4 1.95 0.38 0.95 0.67 2.85 0.35 1.15 0.84 5 2.12 0.63 1.24 0.84 2.85 0.80 0.82 0.73 6 2.42 0.41 * * 2.58 0.58 0.78 0.84 7 2.53 0.40 1.95 0.62 2.31 0.47 1.11 0.84 8 2.36 0.45 1.12 0.78 3.10 0.91 1.33 0.78 9 2.14 0.71 1.31 0.73 2.33 0.56 0.75 0.73 10 2.50 0.54 1.02 0.73 2.15 0.39 0.86 0.84 11 3.40 0.66 1.10 0.73 2.47 0.45 1.21 0.78 12 2.63 0.49 1.49 0.73 2.59 0.01 0.78 0.62 13 2.18 -0.05 2.49 0.73 2.48 0.36 1.31 0.67 Mean ± SEM
2.39 ± 0.11
0.44 ± 0.06
1.35 ± 0.13 0.71 ± 0.22
2.36 ± 0.15
0.49 ± 0.07
0.98 ± 0.07# 0.76 ± 0.22
Table 9.1 Calculated resonant frequencies, fn and reflection coefficient estimates, Γ0+, and model derived elastances, Em , and reflection coefficients, Γm , under control and under NTG conditions for ascending aorta to radial system. The missing (*) values denote cases where R2 is less then 0.94. (# p<0.05).
The model-derived parameters for ascending aorta to radial artery system demonstrated
a slight reduction in the estimated elastances (Figure 9.1 Left). This change is in line
with decrease in pulse wave velocity apparent from phase of the transfer function. The
190
reflection coefficient estimate either by the model or by the analysis of transfer function
pointed towards a consistent but statistically insignificant increase (Figure 9.1 Left).
NTG
Em Γm
Figure 9.1 The model (ascending aorta to radial) estimated elastance (left) and the reflection coefficients (right) of the group of patients before (open bars) and after NTG (closed bars). Bars indicate 1 SEM).
The grouped, binned and average upper limb transfer function between ascending aorta
to brachial artery BAAAH − (Figure 9.2, Left) indicates a peaking similar to one predicted
by the model (Chapter 5). NTG (open boxes, dotted line) alters the modulus of this
transfer function by an overall reduction of the amplification of the transfer function.
Since he major reflection site in the upper limb (ie palm vasculature, Chapter 8) is distal
to the brachial arterial site the influence of the impedance of the proximal branches will
be more apparent in this transfer function (see below). There is also a slight reduction
in the frequency where the peak modulus occurs. As expected, the upper limb transfer
function RAAAH − (Figure 9.2, Right) was similar to the one predicted in Chapter 5. and
subsequently presented in Chapter 7. It is steeper with frequency for the phase and
191
lower in frequency for the maximum amplitude with respect to BAAAH − . This is
predictable on the basis of greater distance for the radial artery (equation 2.138). NTG
does not considerably alter the modulus of this transfer function in contrast to the
BAAAH − . Similar to RAAAH − there is a slight steepening in the phase indicating a
decrease in pulse wave velocity.
HAA BA− HAA RA−
Figure 9.2 The transfer function HAA BA− (Left) and HAA RA− (Right) of the entire
study group under control conditions (circles and solid line) and after administration of NTG (boxes and dotted line). Bars represent ±1 SEM and dash dot line represents no amplification.
9.2.1 Sodium Nitroprusside
Table 9.2 summarises the model derived and calculated parameters for each patient
under control and after administration of SNP. There was no change in any parameter
derived from the model.
192
When transfer function, RAAAH − , data is pooled and averaged for the control case
(Figure 9.3, Solid line), it is apparent that this population is not very much different
from that in Figure 9.1 (Right, Solid line). It has the same prominent peak at 4 Hz. The
effect of SNP on the propagation properties of the upper limb is similar to that of NTG
(Figure 9.3, dotted line).
Control
SNP
Patient No
fn (hz)
Γ0+ (Units)
Em (x106 dyne/cm2)
Γm (Units)
fn (hz)
Γ0+ (Units)
Em (x106 dyne/cm2)
Γm (Units)
1 3.63 0.73 2.50 0.73 2.49 0.66 1.55 0.78 2 2.49 0.18 1.18 0.56 2.39 0.05 0.92 0.56 3 5.38 0.08 * * 4.98 0.71 2.87 0.78 4 2.74 0.52 1.32 0.62 2.34 0.51 0.88 0.56 5 2.20 0.47 0.85 0.73 1.95 0.51 0.72 0.67 6 2.44 0.57 * * 1.84 0.53 0.79 0.84 7 3.50 0.28 2.51 0.56 3.52 0.45 2.54 0.56 8 3.01 0.38 1.32 0.56 3.09 0.31 1.25 0.62 9 2.85 0.50 1.54 0.67 2.32 0.46 1.02 0.62 10 3.20 0.59 2.10 0.73 2.35 0.49 1.21 0.73 Mean ± SEM
3.14 ± 0.29
0.43 ± 0.06
1.67 ± 0.22 0.65 ± 0.03
2.73 ± 0.30
0.47 ± 0.06
1.38 ± 0.24# 0.67 ± 0.03
Table 9.2 Calculated Resonant frequencies, fn and Reflection Coefficient Estimates, Γ0+, and model derived Elastances, Em , and Reflection coefficients, Γm , under Control and under SNP. The missing values (*) denotes cases where R2 is less then 0.94.
The corresponding elastances and reflection coefficients when plotted as bar graphs are
given in Figure 9.4, indicating that there is no significant change in these parameters
which can be ascribed to SNP.
193
HAA RA−
.
Figure 9.3 The transfer function RAAAH − of the entire study group under control
conditions (open circles, solid line) and after administration of SNP (closed circles, dotted line). Bars represent ± 1 SEM and dash dot line represents no amplification.
Em Γm
Figure 9.4 The ascending aortic-radial model estimated elastance (left) and the reflection coefficients (right) of the group of patients before (open bars) and after (closed bars) SNP (Bars indicate ± 1 SEM)
194
9.3 DISCUSSION
The data presented here are important in its implications on two fronts. Firstly, in
different patient populations the transfer function between the central and upper limb
measurement sites is relatively similar. This similarity is consistent with the model
predictions (Chapter 5) and with the other studies (Chapters 7 and 8). This is most
relevant to clinical situations where the consistency is quite important. Since blood
pressure is mostly measured in the upper limb, these findings imply that the errors in
measurement due to different populations will be minor.
Secondly, the relative stability of the upper limb propagation properties under two very
potent drugs, NTG and SNP, allow one to consider the possibility of using these transfer
functions as constant terms under these conditions.
However, it is rather surprising to find such a consistency in transfer functions during
vasodilation. It may be due to alteration of proximal and distal vessel properties in the
same ratio as evidenced from little change in reflection coefficient. If the changes in
wall thickness to diameter ratio caused by these drugs are accompanied by equal and
opposite changes in the wall elastance by smooth muscle relaxation, then, no change in
the reflection coefficient and the proximal elastances is observed. Actually, Cox (1978)
observed this with smooth muscle activation where changes in wall elastance were
offset by changes in wall thickness diameter ratio. Although it seems paradoxical, it has
been suggested that this might be partly due to arrangement of collagen and elastin in
the vessel wall (Nichols and O'Rourke 1990). Activation of smooth muscle reduces the
diameter but relieves tension in the series collagenous element, thus unloading it. When
compared to the original diameter, this effect can be interpreted as a reduction in
elastance (Gow 1972). In the model employed however, no allowance was made neither
for this non-linear behaviour nor for the alterations in the wall geometry. Thus, any
change caused by the model as changes in elastance might be due to a combination of all
195
these parameters. For example, model used in analysing the effects of SNP on wall
elastance might be obscured by changes in wall thickness to diameter ratio.
The simulation results displayed a reduction in elastance by NTG but not with SNP
although the trend was similar. The reflection coefficient estimates derived from the
model did not decrease. In fact there was a tendency to increase indicating no change in
the terminal vessel properties. These surprising findings might be interpreted as changes
on the passive rather than the active mechanical properties of the vessel wall. Indeed, in
both cases the mean blood pressure was reduced slightly but significantly (from 98±3
mmHg to 90±3 by NTG, from 80±3 to 69±2 mmHg by SNP). Yet, based on the
reflection coefficient estimates, both through model and through transfer functions,
these findings suggest that these vasoactive drugs do not alter the mechanical properties
of the upper limb. Therefore, it is possible that any observed changes in the wave shape
are due to changes in the input to the upper limb system. It is further possible that these
drugs alter the central circulation more than the upper limb one. Based on these
findings, the effect of the drugs should be investigated in the central circulation,
probably trunk area, rather than the upper limb.
196
EFFECT OF THE SPLANCHNIC CIRCULATION ON THE FORMATION OF THE ARTERIAL PULSE The arterial model presented in Chapter 5 suggests that reflections from the lower part
of the aortic trunk play an important role in the formation of the central pressure
waveform. It also suggests that vascular beds in the lower trunk, ie mainly the
splanchnic bed, have considerable influence on systemic input impedance as opposed to
beds of the extremities.
Experimental evidence on vasodilators, NTG and SNP, on upper limb arterial properties
(Chapter 9) suggested that these drugs have little effect on modification of properties of
the brachial-radial arterial system. However, these drugs alter the pressure waveforms
dramatically (Yaginuma et al 1986; Kelly et al 1990b, Simkus and Fitchett 1990)
CHAPTER 10
197
without necessarily altering the systolic pressure measured in the upper limb. These
findings and the model simulations suggest that the circulation in the trunk area,
especially splanchnic, has a special affinity to vasoactive drugs. By the action of these
drugs, substantial changes in the arterial waveforms can be induced which can be
detected in the upper limb.
It is known that the blood flow to the digestive organs increases after ingestion of
glucose (Chou 1983, Jern 1991). Thus, ingestion of glucose can be the convenient
stimulant which might reproduce the action of vasoactive drugs on the arterial
waveforms. In this study, this hypothesis was investigated by administering glucose
orally to human subjects to cause vasodilation in the splanchnic circulation (Chou 1983,
Takagi et al 1988, Kato et al 1989, Jern 1991). These caused reductions in the reflected
waves, as manifested by reductions in the augmentation index, while no changes in the
arterial properties of upper and lower limbs were observed. Subsequent implementation
of a reduced model of the systemic circulation suggested that vasodilation of the
splanchnic circulation, induced either by vasodilators or by glucose, reduces the
reflected waves from this circulation, thereby altering the oscillatory load as seen from
the left ventricle.
10.1 METHODS
10.1.1 Subjects
Ten normal healthy subjects participated in this study. None had a history of myocardial
infarction, stroke or diabetes. The characteristics of the study group are given in Table
10.1. The nature, purpose and risks of the study were explained to the subjects and a
198
signed consent form was obtained from each. The study was approved by the Research
Ethics Committee of St Vincent's Hospital.
Mean±SEM Range Age 43.6 ± 5.15 28 - 77 Sex 5 Male 5 Female Weight (kg) 62 ± 3.52 40 - 78 Height (cm) 165 ± 2.27 155 - 181
Table 10.1 The clinical characteristic of the patients recruited for the study.
10.1.2 Protocol
Experiments were performed in the morning in the post-absorptive state after an
overnight fast (10-12 hours). Each subject was studied on two separate occasions within
a period of two weeks. On each occasion they ingested either glucose (75g glucose + 30
ml of cordial + 300 ml of water) or placebo (30 ml cordial + 300 ml water). The
subjects were studied in a quiet, dimly lit and well-ventilated room whilst in the supine
position. They were asked to relax and avoid unnecessary communication with the
examiners whilst they were allowed to establish a semi-recumbent position during
ingestion of the solution. To eliminate observer and subject bias, the solutions were
randomised in a double-blind fashion. At the end of the analysis of the recordings, the
code assigned to the solutions was broken to allow further analysis.
10.1.3 Measurements and Data Analysis
Electrocardiogram (3-leads) and finger blood pressure (FINAPRES, Ohmeda) were
recorded continuously throughout the study. After instrumentation, the examiners (3 in
all) were allowed to practise non-invasive recordings at carotid, both radial and femoral,
199
and dorsalis pedis sites, to gain familiarity with the arterial site. This practice period
normally lasted for 15 minutes before the test. The subjects were asked to maintain this
relaxed position for at least 30 seconds to allow for recordings to be done without any
interference. After a training period of both patient and examiner, a baseline period of
30 minutes was spent while the tonometric recordings were being taken simultaneously
every five minute. At the end of this baseline period, the solution was administered and
further recordings were taken at each five-minute interval which lasted for sixty
minutes. Each tonometric recording throughout the study was concluded with blood
pressures measured in the arm with a semi-automated oscillometric device (Dynapulse,
Pulsemetric Inc, CA). At the end of the study, the distances from the sub-sternal notch
to the recording sites (carotid, radial, femoral and dorsalis pedis) were recorded to allow
for pulse wave velocity calculations.
The data were digitised on-line at a rate of 128 samples/sec, with a 12-bit A/D converter
(Data Translation, DT 2801-A) using an IBM-AT compatible computer. The data were
then transferred to a Write-Once-Read-Many (WORM, Maxtor 800S) drive for
permanent storage.
Data were analysed off-line after averaging segments of 20 second length of each
recording to obtain representative waveforms for each recording session. Blood
pressure waveforms were calibrated by setting mean and diastolic pressures of each non-
invasive waveform to that of the FINAPRES recordings (see Chapter 5). Pulse wave
velocities of the arm (carotid-radial), trunk (carotid-femoral) and leg (femoral-dorsalis
pedis) were calculated by foot-to-foot estimate method. The augmentation indices in the
carotid and radial artery were calculated to obtain an index of wave reflection. Transfer
functions between the carotid-radial, carotid-femoral and femoral-dorsalis pedis sites
were calculated to estimate the segmental propagation properties. All data were then
grouped into bins of 15 minutes which were later averaged and analysed.
200
10.1.4 Statistics
Statistical analysis was performed between derived indices by 2-way analysis of
variance (ANOVA) for repeated measures with subjects as the random factor. Data
averaged in bins of 15 minutes and expressed as mean ± SEM, unless otherwise noted.
10.2 RESULTS
Although, there was no significant difference between the calculated baseline values one
agent was consistently found to reduce the augmentation index in the carotid artery
(Figure 10.1).
PLACEBO GLUCOSE
Baseline Ingestion
1 Sec
Baseline Ingestion
1 Sec
Figure 10.1 A representative carotid tracing obtained from one patient illustrating the experimental protocol and the changes in pressure wave shapes. After a run in period of 30 minutes either placebo (Left Panel) or Glucose (Right Panel) is ingested at time zero (Arrow). Recordings are made at each 5 minute and subsequently averaged over 15 minute periods. Note the decrease in augmentation of the late systolic peak in carotid waveform after glucose ingestion.
201
This effect of the agent started 15 minutes after the ingestion and lasted longer than 60
minutes. When the code was broken, this was found to be glucose. Under the action of
glucose, the second peaks in both radial and carotid arteries were reduced in amplitude,
which was not reproduced by placebo ingestion. The placebo, however, increased the
mean blood pressure without affecting any other features of the pulses. The cause of
this increase in mean blood pressure was not investigated.
When pressure data from all arteries were binned and pooled, it was possible to observe
the effect of glucose on different arterial pulses (Figure 10.2, Table 10.2). The
secondary fluctuations in the central and upper limb arterial pulses decreased in
amplitude more than the lower limb pressure waveforms. Yet, the foot to foot delays in
all segments remained identical to the baseline indicating no decrease in pulse wave
velocities.
CONTROL GLUCOSE
Figure 10.2 Averaged pressure tracings obtained at baseline (Left) and after glucose ingestion in a subject (Right). There was a significant reduction in the central pressure augmentation, accompanied with reductions in pulse pressure in all arteries except the radial.
202
Accompanying these, there was a slight but significant increase in the heart rate (62.0 ±
2.6 beat/min) to (66.9 ± 3.0 beat/min) (p<0.01) at 30 minutes after ingestion of glucose,
but not with placebo (58.9 ± 2.4 vs 58.4 ± 2.3, p = NS). Blood pressure (systolic,
diastolic, mean) did not change with glucose but did so with placebo (mean BP 90.9 ± 3
mmHg to 98.9 ± 3.1 mmHg) (p<0.01) (Figure 10.3, Left).
Control Glucose p value Systolic BP (mmHg) 135.31±5.76 142.04±6.17 NS Diastolic BP (mmHg) 71.52±3.22 70.53±3.14 NS Mean BP (mmHg) 91.58± 3.06 91.46± 3.32 NS Heart Rate (beats/min) 62.0± 2.6 66.9 ± 3.0 <0.01 Carotid AI (%) 112.3 ± 6.3 102.3 ± 7.4 <0.01 Radial AI (%) 65.2 ± 5.9 59.4 ± 6.9 <0.05 Arm-Pwv (cm/sec) 1039.73± 45.9 1050± 43.41 NS Trunk-Pwv (cm/sec) 811 ± 79.15 741.45± 44.28 NS Leg-Pwv (cm/sec) 1016.78± 42.16 1004.14± 37.51 NS Γ0+ -Arm 0.35± 0.05 0.37± 0.03 NS Γ0+ -Trunk 0.41± 0.04 0.35± 0.06 <0.05 Γ0+ -Leg 0.57± 0.05 0.58± 0.05 NS Γs - model 0.67 ± 0.03 0.42±0.05 <0.001 Γl - model 0.53±0.08 0.68±0.05 NS
Table 10.1 Table of calculated parameters (mean ± SEM) during control and 30 minutes after ingestion of glucose. Γ0+ represents the reflection coefficient estimated from the phase velocity, while Γs and Γl are derived from iteration of a reduced model and represent the reflection coefficients at the splanchnic and lower limb circulations respectively (see text for more explanation).
Although segmental pulse wave velocities are not altered with glucose ingestion, both
carotid and radial augmentation decreased (Figure 10.4, Right). Carotid augmentation
decreased from 112.3 ± 6.3% to 102.3 ± 7.4%, p<0.01; radial decreased from 65.2 ± 5.9
to 59.4 ± 6.9, p<0.05. There was a slight increase in carotid augmentation index after
placebo ingestion but it did not reach statistical significance (114.2 ± 6.2% to 117.6 ±
6.8%, p = NS).
203
Mean BP Heart Rate
* ** *
* **
Figure 10.3 Changes in mean blood pressure (Left ) and heart rate (Right) introduced by placebo (closed symbols, solid line) and by glucose ingestion (open symbols, dotted line). Arrow represents the time of ingestion. Bar represents Mean ± SEM, * p<0.01.
Carotid AI Radial AI
# * *
##
Figure 10.4 Changes in carotid augmentation index (Left) and radial augmentation index (Right) introduced by placebo (closed symbols, solid line) and by glucose ingestion (open symbols, dotted line). Arrow represents the time of ingestion. Bar represents Mean ± SEM, * p<0.01, # p<0.05.
204
The transfer function in the segments investigated did not differ for the arm and the leg
as expected from the modelling studies, yet there was a significant change in the trunk
segment (Figure 10.5). The peak modulus of the transfer function decreases and the
phase velocity showed lesser amplification at lower frequencies indicating decrease in
reflected waves. Since the foot-to-foot pulse wave velocities remained unchanged, the
characteristic impedance of the aorta was assumed to remain the same (Chapter 9,
Nichols and O'Rourke 1990).
Amplification Phase Velocity
Figure 10.5 Changes in carotid to femoral transfer function modulus (Left) and phase velocity (Right) under control (closed symbols, solid line) and by glucose ingestion (open symbols, dotted line). Bars represents Mean ± SEM
Using phase velocity data and equation 8.4, reflection coefficients are estimated for each
compartment. There was no change in the reflection coefficient estimates obtained this
way in the upper limb and the lower limb, yet the reflections for the trunk segment
decreased significantly (0.41±0.04 to 0.35±0.06, p<0.05) (Figure 10.6, Right).
205
Γ0+
Figure 10.6 Changes introduced by ingestion of glucose on reflection coefficient estimate in the trunk segment (left) and the leg segment (right). This estimate is determined from the phase velocity data using equation III.6.4. (Bars represent Mean ± SEM, see text for more explanation).
10.3 DISCUSSION
This study demonstrates that glucose ingestion significantly lowers augmentation index
without altering the propagation properties of the upper and lower limb, as evidenced
from the transfer functions. This is accompanied by a slight but significant increase in
the heart rate, with no change in arterial pressures (systolic, diastolic and mean)
measured in the upper limb. Placebo, on the other hand, induced a slight but not
significant, increase in the augmentation index and a slight increase in the mean arterial
pressure. The changes in wave shapes and segmental properties bear resemblance to
those documented with vasodilators (Kelly et al 1990b; Simkus and Fitchett 1990).
They also support model findings in Chapter 5, where vasodilation of the trunk arteries
caused major changes in second peaks in pressure waves. The experimental findings of
206
Chapter 9 are also reproduced, that is that upper limb propagation properties are not
affected by vasodilator drugs. These seemingly contradictory findings on arterial
haemodynamics can be explained by the action of ingestion of food.
It was shown previously that blood flow to the digestive organs (ie coeliac, superior
mesenteric, Moneta et al 1988; renal beds, Avashti et al 1987) increases after a meal in
humans. Because of decreases in resistances of coeliac, superior mesenteric (Takagi et
al 1988) and gastric (Kato et al 1989) beds, central peripheral resistance decreases
(Chou 1983, Jern 1991). Ingestion also affects heart rate, cardiac output, blood pressure
and left ventricular performance indices (Fagan et al 1986; Cornyn et al 1986; Kaspar et
al 1987). These changes in haemodynamics are dependent on the composition of the
meal (Moneta et al 1988; Jansen et al 1990). The ingestion of glucose causes the
maximum changes (Jern 1991; Jansen et al 1989). The changes documented in this
study due to glucose ingestion, are yet another confirmation of these early findings.
However, this present study appears to be the first to couple these changes with pulsatile
haemodynamic measurements performed non-invasively.
These changes related to ingestion of food with vasodilators are not incidental however.
Murrell, who first suggested the use of nitroglycerine as a remedy for angina pectoris,
quoted in 1879 "I took my pulse, and found it was much fuller than natural, and
considerably over 100. The pulsation was tremendous, and I could feel the beating to
the very tips of my fingers. The pen I was holding was violently jerked with every beat
of the heart" (Murrell 1879). This indicated that nitroglycerine actually increases the
heart rate with accompanying sensation of pulses in the finger tips. This effect was
corroborated in recent studies that showed that NTG affects pre-arteriolar arteries
(O'Rourke et al 1992). Surprisingly, a similar feeling can be observed after eating. In
1908, Oliver noted "Observation has shown that the ingestion of food invariably raises
the arterial pressure in the distal area (last phalanx of finger or thumb). This rise
begins as a rule, within five or ten minutes, after the inception of food and it attains its
207
maximum development of from 15 to 20 mmHg in an hour, then it declines and in the
course of from one and a half hours to two and a half hours it gradually settles down to
its minimum level" (Oliver 1908). If one replaces the word "food" by nitroglycerine, the
above quote actually describes the action of nitroglycerine on the arterial pulse as first
described by Murrell. These historical qualitative accounts are probably the first
descriptions provided for common mechanism for nitroglycerine and food on the arterial
pulse.
The changes brought about by a meal (decrease in splanchnic bed resistance, cardio-
acceleration and decrease in reflected waves) could be attributed to the action of insulin
(Jern 1991). Insulin was shown to induce vasodilation in animals (Liang et al 1982) and
in humans (Scott et al 1988) and isolated arteries in vitro studies (Alexander 1977).
However, direct injection of insulin into the brachial artery was shown not to induce
increases in the forearm blood flow in humans (Natali et al 1990) suggesting that the
action of insulin is elsewhere. It was also reported that abnormal insulin mechanism
may affect the arterial wave shape especially the latter part of the pulse (Lax and
Feinberg 1959). This study extends these early findings by describing the action of
insulin on the arterial waveform by augmentation index. The pressure waveforms in the
central arteries and this index have been shown to be related to the impedance patterns
(Murgo et al 1980b, Takazawa 1987). Higher indices are related to reflected waves that
arise from oscillations of the input impedance spectrum. It was shown that this index
decreases with vasodilators (O'Rourke et al 1992) and with certain manoeuvres (Murgo
et al 1980a), yet increases with aging (Kelly et al 1989a). This study suggests that
ingestion of glucose modifies the vasoactive state of the splanchnic circulation.
The effect of the splanchnic circulation on aortic impedance and thus pressure
waveforms are often overlooked. McDonald in his monograph (1974) pointed out this
fact by stating that "the consideration of these (organ) flows illustrates more
dramatically the functional role of the branches than do the dimensions of the branches.
208
The direct line of the anatomical aorta unconsciously leads us to isolate it conceptually.
From the functional point of view a better perspective would be maintained if the aorta
was thought of as terminating at the diaphragm in a group of large branches supplying
the important organs of the abdomen". This conceptualisation of the lower body has
been ignored in the past by investigators who searched for reflection sites in the human
arterial system. Attempts were often directed towards the terminal aorta and sometimes
to the femoral artery (Nichols and O'Rourke 1990). However, evidence that the aorta
"terminates" at the level of diaphragm is mounting. It was first suggested by Alexander
(1952) who proposed the presence of a standing wave in the aorta with its node at
diaphragm level. Ryan et al (1956) attributed this to influence of vessels supplied by
the abdominal aorta to the peripheral resistance. Later, Latham et al (1985) proposed
that the major reflecting site to be at around the diaphragm level, while Ting et al (1990)
found a reduction in pulse wave velocity around this region. The modelling studies
presented in Chapter 5. also support the view that this region behaves like a functional
origin for reflected waves. It was shown that alteration in reflected waves by decreases
in reflection coefficients of vascular beds around the trunk area generates the most
profound changes in the arterial waveform and in the systemic input impedance. The
same changes could not be observed by altering the terminal loads at the limbs. This is
not accidental. The vessels perfusing the splanchnic circulation and the capacitance
vessels are shown to vasodilate more than other beds in response to SNP infusion (Risoe
et al 1990).
There is also evidence from other sources, pointing out the importance of this region in
formation of the arterial pulse. Chinese folklore, which tries to relate organ flows to the
arterial pulse (Fung 1984, Dai et al 1985, Xue et al 1989), has been shown to have some
basis (Young et al 1988, Wang et al 1991) especially under conditions of ingestion,
stress and temperature. It is possible that all these observations are related to the
alterations of reflected waves originating from this bed.
209
10.3.1 Reduced Model of the Arterial System
The relative importance of the splanchnic bed on the formation of the pulse and
ascending aortic impedance warrants special consideration of this bed in interpreting
and modelling of the pulse and impedance patterns. Usage of traditional reduced
models, T-tube and two-tubes in series, does not take into account the importance of this
bed. Often the aorta has been treated in these models to extend from the heart to the
bifurcation as opposed to its functional role. Both of these models considers the lower
limb as an extension of the trunk. In reality, the aorta can be represented as a single tube
extending from the root to the diaphragm where it splits into two, one being the
splanchnic circulation, the other being the lower limb circulation. The evidence for this
is from the observation of the transfer functions obtained from the upper limb, trunk and
lower limb segments in this study. Although there was no change in the upper limb and
lower limb transfer functions, there was considerable change in the trunk reflections
(Figure 10.6). If the traditional view was true, there is no reason why there should be
changes in the transfer function in the trunk, since the terminal load, ie. lower limbs,
remains unchanged. This can only be possible if the lower limb circulation is parallel to
another circulation, the splanchnic, in which the properties are altered. Using this
interpretation of the arterial system, a reduced model was constructed where the blood
flow to upper limbs and head with respect to that in the lower limbs and splanchnic bed
is ignored (Figure 10.7).
In this model, two parallel branches simulating the splanchnic bed and the lower limbs
and which are terminated with modified Windkessels (see Chapter 4) form the
termination to the trunk element (thoracic aorta). The parameters of this configuration
can readily be estimated from morphological measurements and anatomical relations.
The length of the trunk LT, can be estimated as the distance from sub sternal notch to
the Xiphisternum. The elastance ET, can be estimated from pulse wave velocity of the
210
Reduced Model of Arterial System
Γl
Γs
Trunk
Legs
Splanchnic
Figure 10.7 Reduced model of the arterial system based on the experimental findings of this study. Neglecting the flow to the head and upper limbs the trunk segment bifurcates into two at the diaphragm level one to the splanchnic circulation the other is to the legs. These elements are terminated with modified windkessel represented by complex reflection coefficients.
trunk (sub sternal notch - femoral recording site) and the aortic diameter, DT, from the
body surface area (BSA) as (Keats 1990)
BSADT 717.1= (10.1)
Wall thickness, hT, assumed to be 0.07 of the DT. The length of the splanchnic bed
tube, LS, can be estimated as the distance from Xiphisternum to Umbilicus, and of
lower limb tube, LL, from Xiphisternum to femoral artery. Elastance of lower limb, EL,
is estimated from pulse wave velocity in the leg (femoral-dorsalis pedis) by using
Moens-Korteweg equation, while that of the splanchnic bed, ES, was taken as
(ET+EL)/2. The diameters of the splanchnic element, DS, and of lower limbs, DL, are
obtained as from the distribution of cardiac output as (McDonald 1974)
TS DD 74.0= (10.2)
211
TL DD 61.0= (10.3)
while hS and hL are assumed to be 10 percent of their diameters.
By altering the terminal reflection coefficients, ΓS and ΓL (splanchnic and lower limb
respectively), the model is assumed to simulate leg exercise, baseline and ingestion.
Traditionally, model fits are checked against the synthesis of reasonable pressure
waveforms it generates. However, it has been suggested that the flow waveforms are
better templates for predicting the fit of any model to the arterial system (Beneken
1972). Since measurements made in this study do not allow a flow waveform to be
measured, the error function is defined from the synthesised flow wave instead of
comparing it to the measured wave. For this purpose, a carotid pressure waveform was
used as input to the model and an aortic flow wave was synthesised. Using this flow
wave, the error function is defined as the deviation of the diastolic portion of the flow
wave from zero. The relative flatness and zero value of ascending aortic flow waves
have been used previously for estimation of model parameters (Fry et al 1957).
When model is iterated with different values of ΓS and ΓL ,the error surface obtained
displayed always a single global minimum, (Figure 10.8). The values of ΓS and ΓL
corresponding to this minimum represent the model values where the diastolic portion
of the synthesised aortic wave is flat. Since, the iteration for the entire error surface was
time consuming, a minimum searching algorithm based on partial derivatives was
utilised to locate this point (solid circles on the error map in Figure 10.8) which reached
to the same global minimum with reduced number of iterations.
The corresponding pressures and flows for this map is given in Figure 10.9. Although it
is difficult to ascertain the true shape of the flow wave, the flat diastolic flow is in
agreement with reported recordings. The relatively small positive values can be
attributable to the usage of carotid pressure wave instead of the ascending aortic and to
212
the crude estimates of physical dimensions. These flows suggests that the flow wave
remains similar except a slight peaking and shortened ejection duration.
Iteration Map
Figure 10.8 An example of the entire error surface's contour map during iterating the model parameters ΓS , and ΓL . The numbers on contours indicate the percentage error for a given ΓS - ΓL combination. The concentric contours in the middle indicate the
global minimum for this error surface. The solid dots and arrows indicate the path followed by the recursion steps to reach to this global minimum.
The modulus of corresponding impedances (Figure 10.10) shows reduction in lower
frequencies, in line with phase velocity estimates presented in Figure 10.4. The
modulus of the first harmonic is decreased without significant alterations in the zero
crossing of the phase. When these data are compared to those documented in the
literature dealing with effects of nitroglycerine on pulsatile haemodynamics (Yaginuma
et al 1986, Fitchett et al 1988), the correspondence of alterations is clear. This suggests
that changes are due to similar mechanisms.
213
Carotid Pressure Synthesised Flow
Figure 10.9 The pressure waveforms used to predict the model parameters by iteration of the reflection coefficients (Left). For each pressure pulse (control: solid line and glucose ingestion: dotted line) a flow waveform (Right) is generated by altering the model parameters. The flatness and zero value of the diastolic flow, form the criteria of the best fit.
Figure 10.10 The model impedance patterns derived by iteration obtained under control (solid line, filled symbols) and after glucose ingestion (Dotted line, open symbols). There is a reduction in both the phase angles and of the lower frequency components the modulus (compare this to Figure 5.16).
214
The estimated reflection coefficients by this method for pressure waves at baseline and
30 minutes after ingestion of glucose for each patient are shown in Figure 10.11. Again,
reduction in the reflection coefficient of the splanchnic circulation is the best predictor
of these changes.
REFLECTION COEFFICIENT
Figure 10.11 The model predicted changes introduced by ingestion of glucose on reflection coefficients of splanchnic element (left) and the leg element (right). (Bars represent 1 SEM, see text for more explanation).
In this study, it was demonstrated that ingestion of glucose alters the central wave shape
by reducing reflected waves. Since the reflected waves from both upper and lower
limbs are not altered in the mean time, it is concluded that vascular beds supplying the
splanchnic circulation vasodilate after glucose ingestion. Since these reductions in
reflected waves mimic the ones brought about by vasodilators, the apparent action site
of these vasodilators is in splanchnic circulation.
215
SECTION IV
APPLICATIONS
The amount of information contained in the arterial pulse is often underestimated,
wrongly addressed or at times over-emphasised under clinical conditions (Wright and
Wood 1958; Apter 1963; Bourgeois et al 1974; Wei and Chow 1985; Gerber et al
1987). This has lead to misconceptions mainly due to the lack of information on factors
responsible for the formation of the pulse. The absence of readily accessible tools for
non-invasive measurement of the arterial pressure pulse and interpretation of the
pressure data has exacerbated this problem. However, tools and theories related to the
phenomenon of the arterial pulse have been available in various laboratories for nearly
three decades, but were scarcely applied in the clinical environment (Freis and Kyle
1968). This was because of the complexities involved in the mathematics and
instrumentation in application of these techniques, as well as lack of interest from
216
equipment manufacturers who often respond to specific demands. Because of these
factors, both the clinician and the patient had to resort to sphygmomanometry, and the
information gained from that procedure which has not changed appreciably since its first
introduction in 1905. The only exception to this was the introduction of 24 hour blood
pressure measurement devices due to recent developments in microprocessors. The
opposite trend was encountered in places where accurate invasive measurement of
pressure was possible. Since the sites of measurements and their effects were ignored,
attempts to correlate pressure waves obtained this way to the condition of the patients
was thwarted. Even if this was achieved the conclusions drawn were inconsistent with
laboratory investigations. The most notorious of these inconsistencies is the notion
derived from the Windkessel concept. High systolic pressure was considered good
because it denoted a vigorous heart while the high diastolic pressure was considered bad
since it denoted a high afterload.
Previous chapters demonstrated the consistency of the transfer function in the upper
limb as opposed to lower limb and the trunk under a wide variety of conditions. Using
this feature of the upper limb arterial system, it is possible to synthesise the central
aortic pressure waveform from peripheral upper limb pressure recordings. Previous
chapters also shed light on the features of the pressure waveform and the inherent
information content. One may therefore apply these concepts to design of a system for
automated analysis of the pressure waveform. Coupled with synthesis technique, this
system might find use in clinical environments.
In this section, methods for the synthesis of the central pressure waveform from the limb
pressure measurements are described. These methods are then implemented in a
computer program by which certain features of the arterial pressures are extracted and
reported.
217
THE SYNTHESIS OF THE CENTRAL PRESSURE WAVEFORM FROM THE PERIPHERAL PULSE IN THE UPPER L IMB As demonstrated in the experiments described in Section III, there are substantial
differences between peripheral and central pressure waveforms due to wave propagation
and wave reflection. More importantly, these differences are dependent not only on the
arterial properties as dictated by age, anatomy and vasoactive state, but also on the heart
rate. The effect of dependence on the latter becomes apparent when one considers the
use of ad hoc correction and estimation techniques such as linear regression analysis and
statistical methods for estimation of central from peripheral recordings.
CHAPTER 11
218
It would not be of much concern whether the deviation introduced by these techniques
was large, provided that the central pressure waveform was not carrying more
information than that in the peripheral waveform. The central pressure waveform has
been shown to be important in calculation of left ventricular vascular coupling
parameters (Yin 1987). It is also useful if left ventricular performance indices (Sagawa
1981) could be calculated using this waveform instead of left ventricular pressure. It is
therefore not surprising to find on going attempts to synthesise the central aortic
pressure pulse from peripheral radial/brachial recordings using available techniques.
These varied from synthesis of the pulse from upstroke of the waveform by ECG gated
sphygmomanometers (Rodbard et al 1957; Johnston and Kannar 1974; Marmor et al
1987) to application of standing wave concepts (Warner 1955) and to straightforward
inverse filtering (Lasance et al 1976). When none of these were available, correlations
between the blood pressure measured in the brachial artery to that of central were
investigated using different sphygmomanometric techniques (Borow et al 1982).
In this chapter, pressure waves from previously published central and peripheral sites
are analysed. The analysis is repeated after synthesis of the central pressure waveforms.
The synthesis of central pressure waveform involves the use of the transfer function
derived in chapter 8. It is shown that these transfer functions could be used both in
frequency domain where a single pulse could be synthesised off-line, and in the time
domain where a series of pulses could be obtained on-line. The frequency domain
technique is compared to these reported independently and shown to be in close
agreement both in estimated pressure shape and in peak systolic pressure. Extension of
this technique by convolution windows based on these transfer functions is used to
generate on line ascending aortic pressure waveform form peripheral recordings. It will
be shown that this approach could enhance non-invasive determination of cardiac
properties, when used with other techniques (echo cardiography, radionuclide scans).
219
11.1 METHODS
The resynthesis of central pressure from the peripheral pulse assumes that the system is
in steady state oscillation and linear. This allows the application of linear system
identification techniques. In such a system, a transfer function describes the input-
output relations explicitly similar to that presented in previous chapters for upper limb
wave propagation characteristics. Once the system is characterised by relating the input
and output relations in the frequency domain, it can then subsequently be used to find an
input from an output or vice versa.
11.1.1 Synthesis in the Frequency Domain
Since it was shown in Section III that the upper limb transfer function is relatively
constant under different conditions such as ageing, vasodilation and following physical
manoeuvres, it can be used to calculate an input from an output signal. To perform this
operation, the output (ie the peripheral pressure waves) must be transformed into the
frequency domain and subsequently divided by the transfer function (Equation 11.1),
since the output is obtained by the multiplication operation between the input and
system identifier, transfer function, (Oppenheim et al 1983). The resultant frequency
domain representation of the input can then transformed back to time domain to obtain
synthesised aortic pressure waveform.
×=
− )(1)()(
ωωω
PPAAHPPAA (11.1)
In this expression AA, PP and HAA-PP are the ascending aortic, peripheral pressures
and transfer function between the two. The symbol "× " denotes multiplication.
220
Data used for synthesis are obtained from published recordings where accurate aortic
and/or brachial or radial artery pressure waves were given. These were recorded
simultaneously or at short intervals apart. They were obtained from the studies
representing vasodilation (Simkus and Fitchett 1990), exercise (Rowell et al 1968),
normal conditions (Kroeker and Wood 1955; Remington and Wood 1956) and shock
(O'Rourke 1970) to simulate the wide variety of conditions that an individual might
encounter. The aortic and radial/brachial tracings in these reports were digitised by
hand using a digitising tablet (Summasketch Summagraphic). Since this process
involves a noise level higher than expected due to digitisation by hand, it is taken not to
include the noisy part of the waveform. This was done by excluding the frequency
components in the calculation that are far below a noise threshold that was estimated by
averaging components 9-16 Hz.
To synthesise the central waveforms, the digitised peripheral recordings were first
smoothed through a Hamming window (Hamming 1977) and later transformed into the
Fourier (frequency) domain. After eliminating the frequency components with
amplitude below the noise level, they were divided by the transfer functions as discussed
in Chapter 9 in the complex domain. Transforming the resulting complex values
yielded the central pressure waveform. There was considerable attenuation and scatter
in high frequency components of the transfer functions. When complex division was
applied, the attenuation produces an erroneous amplification of high frequency
components. Because of the low power content of high frequency components of the
arterial pulse, it was justifiable to correct for this artefact by assigning the value of unity
to the frequency components of the inverse transfer function whose values are greater
than unity.
The synthesised central pressure waveforms are then compared in the time domain with
the recorded waves. Since systolic pressure was part of the peripheral pressure
waveform most influenced by wave reflection, linear regression analysis was performed
221
on recorded central and peripheral systolic pressure together with the calculated aortic
systolic pressure. Furthermore, visual comparisons were made to ascertain the fit
between calculated wave shapes.
11.1.2 Synthesis in the Time Domain
The method of synthesis of central pressure waveform from peripheral recordings in
frequency domain requires the entire peripheral pulse to be recorded, transformed into
frequency domain, multiplied by the inverse of the transfer function and then
transformed back to the time domain. These steps are not only time consuming, but it is
also impossible to synthesise the aortic waveform on-line. For practical applications it
is important to obtain a continuous trace of central waveform. In physiological
manoeuvres where beat-to-beat changes can occur, such as tilt table testing, Valsalva
manoeuvre and aortic counterpulsation the former approach needs extensive computer
power. These limitations are overcomed by performing operation in the time domain
since multiplication in the frequency domain becomes convolution in the time domain
and vice versa.
In time domain the frequency domain representation of equation 11.1 is
( )thtPPtAA PPAA−∗= )()( (11.2)
where "∗" denotes convolution operation. The function ( )th PPAA− is the time domain
representation of
− )(1
ωPPAAH which can be obtained by inverse Fourier
transformation
− )(1
ωPPAAH.
222
11.2 RESULTS
11.2.1 Frequency Domain
As already documented in these studies (Kroeker and Wood 1955; Remington and
Wood 1956, Rowell et al 1968, O'Rourke 1970, Simkus and Fitchett 1990), there were
substantial differences between the wave shapes of central and peripheral pulses. These
differences are pronounced at peak systole due to reflected waves effecting specifically
the peripheral systolic pressures
The systolic pressure differences between peripheral and central aortic pulses ranged
from a few mmHg to as high as 60 mmHg during exercise (data from Rowell et al
1968), Figure 11.1, Left.
Before Synthesis After Synthesis
Peripheral Systolic Pressure (mmHg) Peripheral Systolic Pressure (mmHg)
Figure 11.1 The relationship between the peripheral and central systolic pressure difference, ( )AAPA SPSP −∆ , and peripheral systolic pressure before (Left) and after
corrections are made (Right). Although initially a a linear relationship exist, the synthesised pressure becomes independent of peripheral pressure ( 0001.0 67.0 10.2032.0 2 <=−= pRxy ). Dotted lines represent 95% confidence
intervals.
223
Importantly, these differences were linearly dependent on the actual peripheral pressure,
suggesting the influence of reflected waves on peripheral pressure wave contour. At
first, it could be thought that this relation might be useful in determination of the central
systolic pressure from peripheral recordings by simple linear regression analysis.
However, the relatively small correlation between these two pressures (R2=0.67) makes
it almost impossible to employ linear techniques for this purpose. The difference
between the central and calculated systolic pressure became independent of the
peripheral systolic pressure when the resynthesis technique was applied (Figure 11.1,
Right). This indicates that synthesised central pressure waveform correlates better with
the actual pressure and the scatter should be explained by factors other than the wave
reflection.
This finding is also supported when the actual and synthesised pressure waveforms are
compared (Figure 11.2).
CONTROL NITROPRUSSIDE
Figure 11.2 The radial (thin solid line) measured aortic (thick solid line) and the synthesised aortic pressure (dotted line) waves, before (top) and after (bottom) administration of sodium nitroprusside. Although there were significant differences between the radial and central aortic pressure waves, the synthesised aortic wave approximates the actual measured wave. Data are obtained from Simkus and Fitchett 1990.
224
Although significant differences in both the shape and the peak systolic pressures were
present between peripheral radial and central aortic pulses before and after
administration of sodium nitroprusside, the synthesis method eliminated the major
discrepancies. During synthesis of the central pulse, however, the high frequency
components of the pulse are lost due to filtering and windowing effects. This indicates
that more work has to be done to reduce these effects. Yet, the point of incisura as well
as the shoulders and the upstroke were still identifiable as the sudden inflection points in
the pressure wave.
Figure 11.3 Graph showing directly recorded (closed circles) and synthesised (open circles) ascending aortic pressure compared with recorded brachial or radial pressure (open squares) in data published by (A) Simkus and Fitchett 1990 under control and (B) after nitroprusside infusion, (C) by Rowell et al 1968 under control and during graded degrees of exercise, (D) by O'Rourke 1970 during shock, (E) by Remington and Wood 1956 under control and (F) by Kroeker and Wood 1955 under control conditions.
When data were grouped within the sources from which they were obtained (Figure
11.3, Table 11.1) it was seen that the synthesis technique often decreased the systolic
overestimation. However, in four cases it predicted a lower central systolic pressure
225
compared to the actual measured pressure. Close inspection of these recordings revealed
that the mean arterial pressures in the peripheral sites were much lower than that in the
central aorta. This might be caused either by the digitisation technique or by the actual
pressure drop in the upper limb, since the zero hertz components of the modulus of the
transfer functions were always equal to unity.
Systolic Pressure (mmHg) Pulses Radial or
Brachial Asc.Aorta (Calc) Asc.Aorta
(Meas) Simkus and Fitchett ,1990 (Control, n=10)
116±3* 103±3 101±3
Simkus and Fitchett , 1990 (Nitroprusside, n=10)
105±3* 93±3 88±8.3
Rowell et al 1968 (Graded degree of exercise, n=4)
187±18= 143±11 141±12
O'Rourke et al 1970 (Shock, n=1) 85 75 73 Remington end Wood, 1956 (Control, n=1)
139 133 132
Kroecker and Wood, 1955 (Control, n=1)
165 138 142
Total (n=27) 124±7* 106±5 104±5
Table 11.1 Comparison of measured peripheral, measured aortic and synthesised aortic systolic pressures from different investigators. *=p<0.0001, ==p<0.005 compared against cental aortic values).
The maximum reduction in synthesised systolic pressure occurred in the data reported
by Rowell et al (C in Figure 11.3) where the heart rate was more than 120 beats/min. It
was expected, however, that under conditions involving high hearts rates the errors
introduced by the variability of the transfer functions would be apparent. Yet, either due
to harmonic composition of the peripheral pressure wave or due to pre-processing of the
pressure waveforms, this problem does not introduce severe error in the synthesis.
226
11.2.2 Time Domain
The convolution window that has been used to synthesise aortic from brachial is given
in Figure 11.4. It has a zero mean indicating no change in mean value and a delay of
128 milliseconds indicated by the peak of the window. This delay (45º / Hz) has to be
taken into account for each sample acquired.
( )th AABA−
Figure 11.4 Convolution window used for synthesis of ascending aortic from brachial pressure wave.
An example for on-line calculation of aortic from radial pressure waves using
convolution window based on transfer function between radial and aortic is given in
Figure 11.5. The on-line method yields similar results to the off-line method.
Although the difference between the radial (Middle, solid line) and aortic pressures
(Bottom, Solid Line) were high, the window reduced this difference considerably
(Bottom, Dotted line). There was however a linear delay caused by application of
convolution to the data. Although real aortic wave occurs earlier than the peripheral
waves ( of the order of 60-100 milliseconds) the synthesised wave appears on the output
128 milliseconds later. Because of this processing delay it is impossible to obtain the
actual temporal relationship between the two waves. However, for display purposes a
227
value of pulse transmission may be empirically added to the raw signal so that two
waves can be aligned temporally. These findings are confirmed in a study performed in
12 patients. Results are similar to frequency domain synthesis and well compared to
beat-to-beat and patient-to-patient data (Table 11.2, figure 11.6).
ECG
Radial
Aortic
Synthesised
50
mmHg
Figure 11.5 On-line calculation of ascending aortic wave from radial pressure waves. Data is recorded using high fidelity Millar micromanometers. The radial pressure wave(middle panel)is shifted by 50 mmHg up for display purposes. The synthesised aortic pressure wave (bottom panel, dotted line) agrees well with actually recorded aortic pressure wave (bottom panel, solid line), although it introduces a delay of 128 milliseconds to the data.
228
Systolic (mmHg) Diastolic (mmHg) Pulses Radial Asc.Aorta
(Calc) Asc.Aorta
(Meas) Radial Asc.Aorta
(Calc) Asc.Aorta
(Meas) 9.0 133.2±9.0 122.0±8.5 127.2±3.7 67.8±8.0 71.5±8.0 68.2±1.6 6.0 125.7±4.6 120.5±4.3 126.4±5.1 70.1±3.9 74.5±2.7 74.5±5.9 7.0 122.5±4.6* 105.4±4.2 110.0±4.4 62.9±3.9 65.4±3.9 66.3±3.1 7.0 151.1±3.4* 131.8±2.9 133.3±3.5 78.0±1.7* 79.6±1.7* 73.0±1.2 8.0 137.5±1.9* 122.6±2.5= 118.8±1.7 76.2±3.0 79.4±2.8= 76.2±0.7 6.0 123.5±13.6 116.8±13.6 121.3±1.7 64.0±5.1 66.2±4.8 65.3±2.1 6.0 128.8±4.6* 115.0±3.7 109.7±2.1 59.1±2.6* 61.4±2.0= 65.4±1.1 8.0 164.4±5.9* 153.1±5.5 151.3±7.2 70.0±3.2 73.2±3.1 70.1±3.2 8.0 124.9±3.4* 103.1±4.2 102.3±2.4 59.3±4.4= 65.2±4.6 68.7±0.9 8.0 166.4±9.2 156.8±9.6 163.2±3.6 78.0±7.9 81.4±8.4 84.5±2.1 10.0 131.1±3.1= 113.1±2.8* 124.9±3.9 69.2±1.6 71.7±1.8= 68.7±3.6 10.0 123.2±4.4* 99.2±4.1 101.8±2.7 59.0±3.7 62.1±3.4 61.4±1.6
Total 136.0±15.6* 121.6±18.1 124.2±18.6 67.8±7.1 71.0±6.9 70.2±6.2
Table 11.2. Results obtained by time domain synthesis method on series of peripheral radial pressure pulses in 12 patients. The technique reduces both beat to beat (indicated by row values) and patient to patient (indicated by column values) differences in both systolic and diastolic pressures. Radial pressure waves are measured by tonometer and calibrated against brachial pressure. Due to technical difficulties data was not recorded simultaneously. (Data are given as Mean±SD, *=p<0.001, ==p<0.05 compared against cental aortic values).
11.3 DISCUSSION
It is shown in this study that a single transfer function can be used to estimate the central
waveform from the peripheral upper limb pulse. This transfer function which is
obtained from an earlier study (Chapter 9), has a local peak at around 4 Hz which is
predicted by the model (Chapter 5) and subsequently demonstrated by experimentation
(Chapters 7). Being higher than unity at frequencies of physiological significance, it
dictates the amplification of components of the pressure wave.
229
The usage of a single transfer function for this purpose could be questioned. This
transfer function might be expected to vary between individual subjects with ageing,
vasodilation, exercise and under certain physical manoeuvres. However, the simulated
effect of ageing, as produced by increase elastic modulus, or vasodilation, as produced
by decreases in global reflections, on this transfer function is minor (Chapter 5). Some
of these predictions of the model were later confirmed by experiments on human
subjects. It was shown that, although both have profound effect on global reflections
neither the vasodilator agents nor the ingestion of glucose has a major impact on this
transfer function (Chapter 9 and 10). Another major determinant of the transfer
function, the terminal reflection coefficient, was shown in the model to affect the
pressure wave amplification (Chapter 8). However, it was determined that this factor is
stable and varies little under physiological conditions. All these points justify the use of
single transfer functions for aortic to radial and aortic to brachial pressure wave
propagation systems for different subsets of patients.
Transfer functions per se do not dictate the specific contour of the peripheral waveform,
but state the relationship between the input and the output. It is therefore important to
quantify the input to the system used for reconstruction of a pressure waveform.
Clearly, the pressure waveform at the output will only be composed of harmonics of the
input pulse, amplified or attenuated and delayed (ie all as consequences of a linear
system). The relative consistency of the transfer function at lower harmonics makes
more sense, if it was to be used to synthesise the central waveform, when one notices
the power present in these harmonics of the pressure waveform. As already shown
(Nichols and O'Rourke 1990) the pressure pulse often contains much of its energy
(approximately 90 per cent) at frequencies below 4 Hz (Figure 11.7).
This frequency in turn is the one where the transfer function happens to be more reliable
and not showing much scatter (Section III). Thus, if one uses this transfer function, it is
probable that the most of the central pressure wave will be synthesised reliably since
230
both the power content of the pulses and the reliability of transfer function is high at
frequencies below 4 Hz.
Figure 11.7 Graph showing the transfer function between the carotid artery to radial (Solid line , Closed Circles) and the percentage total power in the carotid waveform (Dotted line, Open squares). Although there is amplification of the frequency components above 3 Hz, the power in these harmonics constitute less than 10% of the total power. Thus radial pressure waveform is not sensitive to alteration in the transfer function at higher frequencies.
The resynthesis is inevitably an operation of low pass filtering of the peripheral pulse
(the inverse operation of amplification is attenuation). This operation, unfortunately,
reduces the amount of intermediate frequency components in the peripheral pulse and
the synthesised pressures become rounder (Figure 11.2). Since the transfer function
approximates unity at high harmonics, the features with steep changes are retained. It is
therefore possible to extract these features from the synthesised pulse. In cases where
this is impossible, such as the incisura, values derived from the original pulse can be
used (Gallagher et al 1992). This becomes extremely useful for the purposes described
in the introduction of this chapter.
231
5 10 15 20
1
2
3
-5
0
-10
Frequency (Hz)
Phase (Radians)
|Amp|
Figure 11.8 The transfer function between ascending aorta and brachial artery used by Lasance et al (1976) to synthesise the central from the brachial pressure waveform for calculation of cardiac output. Note the similarity between this and the transfer function of the brachial artery under control conditions displayed in Figure 9.2.
These findings are also in line with these reported in the literature (Falsetti et al 1974;
Krovetz et al 1974) where the effect of harmonic correction by linear techniques was
analysed. The synthesis technique of undistorted pressure waveforms recorded by
catheter systems with poor frequency response is basically similar to that employed in
this study. Although the system that was analysed in this study was practically different
(catheters versus upper limb circulation), the analogy is striking. When similar inverse
filtering technique is used in catheters having resonant frequencies of 7-10 Hz, it was
possible to reduce the systolic overshoot from 27% to 6% of the actual (Krovetz et al
1974). Therefore this study confirms the assumption that the important parameters are
the behaviour of the transfer function and the harmonic content of the pulse. Another
study that was essentially similar to that presented in this chapter was made in 63
subjects to calculate a representative transfer function between the ascending aorta and
the brachial artery (Lasance et al 1976). This generalised transfer function (Figure 11.8)
was then applied to the original data to synthesise the central from the peripheral pulse
232
with the aim of calculating cardiac output. The calculated transfer function and the
correspondence of the synthesised and measured waves were similar to those obtained
independently and presented in this chapter and earlier in this thesis.
In conclusion, it is appears that to a first approximation a single transfer function can be
used in human adults to synthesise central pressure waveform from peripheral upper
limb recordings. The synthesised waveforms bear a close resemblance to measured
waves. However, due to the low pass filtering effect of the process, some features of the
central pulse may not be reproduced. The upper limb transfer function employed in this
technique is considered to be independent of age and vasodilators. In earlier chapters, it
was also shown to be relatively stable even under extreme conditions, such as occlusion
of the hand and reactive hyperaemia. However, performance of the technique might be
further increased if transfer functions are personalised using anatomical and
physiological quantities. Application of time domain representation of transfer function
(convolution windows) enables to calculate central from peripheral recordings on-line.
This procedure could be used where central aortic features are needed on a beat-by-beat
basis.
233
A SYSTEM FOR ON-LINE ANALYSIS OF BLOOD PRESSURE WAVEFORMS Results of investigation described in previous chapters have lead to development of a
system which integrates the current techniques of blood pressure waveform registration
and feature extraction. The system comprises a computer with appropriate input/output
devices, such as screen, keyboard and disk(s); an analogue to digital board to convert
signals from various pressures and triggering sources, and software to gather, analyse
and report the pulses using this hardware. Since it is of vital importance to run in near
real-time in the clinical environment, provisions are made to accelerate the entire
process while minimising costs.
The basis of the system is the acquisition of pulses from a peripheral site (radial, carotid,
femoral, brachial, axillary, subclavian and dorsalis pedis artery) or from ascending aorta
CHAPTER 12
234
by means of a pressure registration device (tonometry, volume clamping or invasive
catheters) and analysing the waves due to existing knowledge about the pulse. Pressure
wave propagation parameters such as reflection coefficients and pulse wave velocities
are calculated to quantify the effects of different drugs or manoeuvres on the arterial
vasculature. It also contains a database management system based on pressure
recordings which can be coupled with other diagnostic and therapeutic tools. To
achieve all these goals, the Diagnostic Applanation Tonometery (DAT) software
includes routines for signal acquisition, signal conditioning, feature extraction,
parameter calculation, database maintenance and report generation (Figure 12.1).
Feature Extraction Unit
ReportWriter
DatabaseEngine
Signal Conditioning Unit
Data Acquisition Unit
The DAT System
HELP
UNIT
Figure 12.1 The components of the Diagnostic Applanation Tonometery (DAT) system showing the hierarchy and interplay of separate units.
This chapter describes the implementation of the system and algorithms used for
analysis. The information on installation, operation and maintenance of the system is
described in detail by the user's manual provided in Appendix I.
235
12.1 IMPLEMENTATION
12.1.1. Data Acquisition
The peripheral pulse is acquired at a rate of 128 Hz using an A/D converter. The
sampling rate is fixed at this frequency since it not only facilitates operation of Fast
Fourier transforms (FFT) using techniques described by Blackman-Tuckey algorithms
(Blackman and Tuckey 1958) but also it is well above the bandwidth of pressure signals.
This choice of frequency is low enough to reduce overhead in data storage and analysis
while high enough from aliasing errors. By doing so, it is guaranteed that the same
harmonic will always lie in the 1 Hz window. The clock pulses necessary to generate
this sample interval is generated by the computers (IBM-PC) timer chip, (Intel 8255)
(although a clock on an A/D board would also suffice) generating an interrupt to be
serviced. The use of the system clock allows utilisation low cost A/D boards which do
not posses internal timer or interrupt generating capabilities. Since the timer interrupts
were used to synchronise the data acquisition steps, they were not available to the
operating system. Before data acquisition module takes control, the vector pointer at
address $0000:0020 Hex for timer interrupt ($08 Hex) handler of BIOS was changed to
the interrupt servicing routine of the DAT system (Figure 12.2). This technique
modified the priority scheme of the timer interrupts. Since same timer interrupts are
also used by DOS in time keeping activities, BIOS and DOS routines are activated at
regular intervals to update the system clock.
At each interrupt, the CPU reads data from the A/D ports representing the pressure (Ps)
and if present, the trigger (Ts), Figure 12.3. The channel numbers for these signals are
assumed to be sequential, the channel for Ps being the first, and given in the
configuration information. The Ps is then copied to a circular buffer of 10 second
length. Thus to fill the buffer once, 1280 samples are necessary. If the trigger signal is
236
PC TimerHardware
Layer
SoftwareLayer
BIOS
DOS
DAT AfterBefore
Figure 12.2 The modification of original interrupt vector table for timer interrupts. Arrows indicate the order of execution of code after each received interrupt. Under normal circumstances the priority is to the BIOS, DOS and then the user program. Alteration of the vector sequence also alters the existing priority.
not present, then Ps is also copied to a buffer of equal type and size for trigger
information, otherwise Ts will be copied into it. On-line calculation of aortic pressure
waveform is also performed during the acquisition step of the pulse. The Ps is directed
to a convolution buffer, where it is convolved with the window representing the inverse
transfer function (see below) to yield an ascending aortic pulse (Cs). The Cs is stored
into the same size circular buffer. If D/A channels are available, an identical copy of Cs
and Ps are dumped out from these ports. The Ps is then displayed on the screen in a
rotating drum sequence. If present the derivative of the trigger signal (Ts) is also
displayed. The same process is repeated continuously until terminated by the user.
For recording sites between the ascending aorta and either carotid, radial, femoral,
brachial, axillary, subclavian or dorsalis pedis a transfer function is determined either by
direct calculation from in vivo data or from the model. From this, a convolution
window is calculated for each transfer function (see chapter 11). To obtain a cut off
237
frequency at 16 Hz, the resultant window is then stored in a look-up table of 33 samples
long for each patient. Normally, this window function yields an estimate of the central
pressure value for each sample after (W+1)/2 convolutions, (W is the window length).
This operation introduces a linear delay of sampling period*(W+1)/2.
Cs
Ts
Ps
Triggersource
PeripheralPressure
Ps
Asc. Aorta
Interrupt
Display
Display
PC Timer
Ts Buffer
Ps Buffer
Cs BufferConvolutionWindow
A/D
D/A
Figure 12.3 Block diagram of the data acquisition section of the DAT system. Peripheral pressure waveform and trigger source is fed into to the respective buffers where they are kept and displayed. Peripheral pressure is convolved with the respective filter and stored as central pressure waveform. Due to presence of D/A converters the pressure signals (raw and convolved) are dumped out. Entire events are synchronised with PC timer interrupts. Dotted lines indicate optional paths.
12.1.2 Signal Conditioning
Upon the user's request data from buffers (Cs, Psand Tc) are transferred to the database
records, excluding the last 2 seconds to allow for data interruption which may occur
during the initiation of the request. The PC timer is then stopped and updated by the
system parameters while real-time clock registers are also updated. The contents of all
buffers are smoothed using 7-point moving average filters to limit the bandwidth of the
signals to 20 Hz (Oppenheim et al 1983). The data string in buffer Tc is then
238
differentiated and further smoothed using first forward differentials and 3-point moving
averages to find the triggering points and to eliminate the noise amplified during
differentiation. Maximal and minimal thresholds are defined as the 60% of the
maximum and minimum derivatives in the entire buffer content. This value for
threshold is determined empirically after trying for several pulses. The positions of the
onset of pulses are then determined by comparing them against the maximum threshold.
The segments within these maximal and minimal thresholds are marked. The
corresponding data from Ps and Cs are averaged to yield averaged recordings of
peripheral and synthesised waveforms. The averaged Ps is then calibrated using given
pressure values. The numerical values for Ps and Cs are then used to estimate overall
gain and offset of the system. The calibration values obtained this way are subsequently
used to calibrate the synthesised aortic pulse.
12.1.3 Feature extraction
Wiggers (1928) described 14 different features present in the arterial pulse contour.
Investigation described in previous chapters indicates that meaningful information can
be obtained a much smaller subset, This system extracts five time relative points on the
waveform from which parameters relating to the heart and arterial system are
determined, (Figure 12.4). They form the positions describing the foot, first shoulder,
second shoulder, incisura and the duration of the pulse. Since these features are
geometric and could be described as inflection points along a curve, they can be
identified using differentials of different orders (Figure 12.5). Of all the points on the
pressure waveform, particular importance is given to the systolic onset (foot of the
pulse), maximum of the first derivative (upstroke of the pulse) and systolic point
(maximum of the pressure pulse). All other points can then be expressed in relation to
these points, thus providing consistency in comparisons.
239
Features of the Arterial Pulse
Second Shoulder
First Shoulder
Foot of the Pulse
Pulse Duration
T 1
T 2
T i
T f
T T
Aortic valve closure(Incisura)
Figure 12.4 The basic features of the arterial pulse. After the foot of the pulse indicating the onset of ejection determined from the trigger source, the pressure wave rises to an initial peak where it forms a shoulder. It then proceeds to a second shoulder which often constitutes the peak pressure in the elderly. The former point is related to timing of peak flow while the second shoulder to reflected waves. The end of ejection is associated with closure of the aortic valve which is often seen as a distinct incisura on the aortic pressure pulse.
The flow chart describing the feature extraction process is shown in Figure 12.6. The
point Tp is the systolic pressure point and is found by scanning the averaged pressure
pulse for its maximum point. Max dp/dt is the maximum point on the first derivative
curve and systolic onset can be determined by taking the first derivative of the pulse and
locating a zero crossing from negative-to-positive that precedes the Max dp/dt point.
The point Zc3rd+- is the third derivative's first zero crossing point from positive to
negative after the Max dp/dt point and it correspond to the peak flow in an "elderly"
(Type A) pulse Figure 12.7 (Murgo et al 1980b; Kelly et al 1989a).
240
Third Derivative
Pressure Waveform
TfTi
T Max dp/dt
Tp
T zc3rd-+
T zc3rd+-
First Derivative
T Max3rd
Figure 12.5 A typical waveform (solid line) together with the first and third derivatives (dotted and dash dot line) used in extracting features (see text for more explanation).
Tp>Tmax3rd+60msec
(Tzc3rd-+)<(Tp +60
msec) or (Tzcrd-+ )>
T2= Tzc3rd-+T2=0T2 =Tp
T1= Tmax3rd T1 =Tp
Type= B or CType=A
Find Tmax3rd, Tzc3rd+
Find 1st Derivative Max Point(Tmaxdpdt) and Wave
Foot (Tf)
Find Systolic Max (Tp)
START
(Tf+300 msec)
Figure 12.6 Flow chart for extraction of time domain features in the pulse.
241
The Zc3rd-+ point is the point where the second shoulder is expected to occur in a
"young" pulse (Type B or C) and could be determined as the first negative-to-positive
zero crossing point after the Tp on the third derivative curve. The point of incisura
representing the end of ejection can be determined once the second shoulder is found. It
is often at the Zc3rd+- point after the second shoulder. Once these times are found, all
the other important features can be determined from them using the flow chart.
Type A Type B or C
Figure 12.7 The types of pulses defined by Murgo et al (1980b) in describing the typical waveform in central aorta. Type A pulses have a distinct shoulder in late systole which augments the systolic pressure and seen mostly in elderly patients. Type B or C pulses are seen in young patients where no augmentation in the pulse can be observed. Although these descriptions are categorical there is a continuos spectrum of pulses can be observed due to different degrees of reflected waves present in the composite pulse. Augmentation index describes the amount of reflected wave in the pressure pulse.
12.1.4 Calculated Parameters
From averaged peripheral and synthesised central pulses the following values are
calculated:
Ejection Duration
242
ED Secs T Ti f( ) = − (12.1)
Heart Rate
HR Beats TT( min) = 60 (12.2)
Augmentation Index
AI P PP Pd
d(%) = ×−−
100 2
1 (12.3)
Pressure at First Shoulder
P mmHg P T1 1( ) = (12.4)
Pressure at Second Shoulder
P mmHg P T2 2( ) = (12.5)
Pressure at Diastole
P mmHg P Td f( ) = (12.6)
Pressure at Systole
P mmHg P Ts p( ) = (12.7)
Mean Diastolic Pressure
243
MDP mmHg
P
T T
ii T
T
T i
i
T
( )( )
=−
=∑
(12.8)
Mean Arterial Pressure
MAP mmHgP
T
ii
T
T
T
( ) = =∑
1 (12.9)
Mean Systolic Pressure
MSP mmHg
P
T T
ii T
T
i f
f
i
( )( )
=−
=∑
(12.10)
Tension Time Index
( )fi TTMSPHRBeatsmmHgTTI −××=)min.( (12.11)
Diastolic Time Index
( )iT TTMDPHRBeatsmmHgDTI −××=)min.( (12.12)
Subendocardial Viability Ratio
( )TTIDTISVI ×= 100% (12.13)
Augmented Pressure
AP mmHg P T P T( ) = −2 1 (12.14)
244
Reflection Transit Time
RT Sec T T( ) = −2 1 (12.15)
Maximum Rate of Rise
=
dtdPMaxSecmmHgdtdPMax )( (12.16)
Reference Age ( ) 81.33100642.0RA(Years) +−×= AI (12.17)
Reference Age is the age corresponding to the augmentation index as derived from
ageing studies, Kelly et al 1989a and where "P[]" indicates the element of the pressure
waveform array.
12.1.5 Database Engine
Since it is assumed that there will be further enquires regarding the recordings in the
future based on epidemiological and short term research, a database engine is installed
as part of the software. The data preserved are the raw and averaged data for each
peripheral and calculated central aortic waveforms as well as indices and values derived
from them. A separate database is linked to the recording's database by the name of the
patient and the time of visit (Table 12.1). This second database keeps the information
regarding the patient particulars such as age, sex and anatomical and diagnostic data
(Table 12.2). Type definitions for each record are given in tables 12.3 and 12.4.
To facilitate retrieval of archived information, the transaction time is kept to a minimum
by indexing both the "PATIENTS" and "RECORDS" databases on their key fields
(Table 12.1) using AVL-trees (Adelson-Velski and Landis 1962). AVL-trees are
245
derivations from the B-trees (Chapter 4) where it is ensured that a maximum of log2N
searches would be sufficient to access a particular record (N is the maximum number of
elements in the tree). B-trees lack this insurance due to the unbalancing of the tree when
sequential insertions are made. AVL trees however, take this into account by balancing
the B-tree at each insertion and deletion so that there will be log2N levels present in a B-
tree at all times. Using this algorithm, the software is able to locate a record from a list
of 65536 recordings in at most 16 attempts (Shammas 1988). Provisions are also made
to ensure integrity and distribution of the database across several platforms by obeying
strict rules imposed by the system in accessing the database using indexes.
FIELD TYPE Patient Name Small String Date of visit Small String Medication Array[1..2] OF Small String Notes Small String Operator Id Word Trigger Present Boolean Diagnosed Boolean Gain of System Real Offset of System Real Trigger Series Array[1..128] of Integer Recorded Pulses Array[Raw..Convolved] of Pulse
Buffer Heart Rate Real Augmented Pressure Real Reference Age Real Tension Time Index Real Diastolic Time Index Real Subendocardial Viability Ratio Real Mean Systolic Pressure Real Mean Diastolic Pressure Real End Systolic Pressure Real
Table 12.1 The template for record of database file "RECORDS". This file is indexed with Patient Name field and Date of visit and linked to PATIENTS database. For explanation of defined types see tables 12.3 and 12.4.
246
FIELD TYPE Patient Name Text String Patient ID String[8] Patient Age Integer Sex String[2] Address Small String Distances Array [Radial.. Aortic] of real Convolutions Convolution Types
Table 12.2 The template for record of database file "PATIENTS". This file is indexed with Patient Name field and related to RECORDS database. For explanation of defined types see tables 12.3 and 12.4.
FIELD TYPE Pulse Name Pulse Types Tonometer Input Series Pulse Stream Averaged Pulse Pulse Length of Averaged Pulse Integer Systolic Pressure Real Diastolic Pressure Real Mean Pressure Real Time to Foot Of Wave Real Ejection Duration Real Max dp/dt: Real Systolic Onset Real Time to Incisura Real Reserved Real Time to First Shoulder Real Time to Second Shoulder Real Reverse Shoulder Index Real
Table 12.3 The type defined for any pulse. It contains information about the original series of pulse, averaged pulse and the derived indices. This buffer is duplicated for both raw and convolved signals and corresponds to PS and CS buffers in Figure 12.3.
247
New Type Definition Small String String[40] Pulse Array[0..255] of Real Pulse stream Array[0..1023] of Integer Convolution Window Array [-16..16] of Real Pulse Types Radial, Carotid, Femoral, Brachial,
Axilla, Subclavian, Dorsalis Pedis, Aortic
Convolution Types Array[Radial..Aortic] of Convolution Window
Signal Types Trigger, Raw, Convolved
Table 12.4 The type definitions based on ordinal definitions supplied by Pascal language used in record definitions.
12.2 THE TRIAL OF THE SYSTEM IN A CLINICAL ENVIRONMENT
This system is currently being evaluated in several laboratories around the world. Early
implementation of this system in an outpatient clinic in Sydney has accumulated more
than 6000 recordings of nearly 800 patients over two years. Clinical utilisation of this
system and the database has been undertaken by Dr. David Gallagher as a component of
his thesis for Doctor of Medicine in collaboration with Prof. MF O'Rourke in this
laboratory. Experience has shown that the system is reliable. An experienced user
spends 20 seconds to record and analyse 10 seconds run of pulses using an IBM-XT. A
hardcopy of the report is generated in 90 seconds with conventional printer hardware.
The average access time to any recording is undetectable (less than 12 lookups).
Currently it is undergoing extensive usage to detect effects of vasodilators. It is hoped
that following these investigations, the system with the necessary hardware and software
will be implemented as a stand alone unit in the near future, thus enabling clinicians to
complement conventional sphygmomanometer recording using sophisticated pulse
waveform analysis.
248
SECTION V
SUMMARY AND CONCLUSIONS
The models employed and experimental studies performed and reported in this thesis
were directed towards a better understanding of the human arterial pulse as a carrier of
information relating to physical and wave propagation properties of arterial system. To
understand the causes of alterations in pulse contour, the phenomenon of pressure wave
propagation was investigated by characterising the arterial system in terms of pressure-
pressure relations. This approach was different from many previous studies in pulsatile
haemodynamics where the system was characterised in terms of pressure-flow relations.
The investigation concluded with approaches that not only describe the information
content of the pulse but also ways to quantify it.
249
Throughout the thesis, the arterial system is considered to be operating within the
physiological range of pressures. This approach enabled the use of linear techniques in
employing tubular models of the arterial system. A linear approach has been shown to
be adequate to quantify the specific features of the propagating arterial pulse under
normal physiological conditions. The extra information that might be gained by
adopting a non-linear approach did not justify the increased complexities of
computation. Non-uniformity of the arterial system was simulated by connecting
uniform tubes with different physical properties. Although the computational procedure
used in this model is similar to that used in earlier models, there are basic differences in
both the model construction and its implementation. For a better computational
efficiency, the present model uses a binary tree structure for the arterial tree. The
anatomical dimensions and physical parameters of the arteries were also updated using
recent data. There were also major differences in the terminations of the branches
which were loaded with modified Windkessels in contrast to previous models where
real elements were used explicitly. Using complex reflection coefficients derived from
this approach a wide variety of interventions were simulated. As part of the
experimental investigation the model was used to estimate significant model parameters
using optimisation techniques.
The advantage of using models in experimental settings was demonstrated by exploring
the propagation of reflected waves in the arterial system. The intensity of reflected
waves was modulated by altering the reflection coefficients at specific terminations.
Although limited to major arteries, this technique has been successfully applied in many
previous experimental investigations. In these previous studies however, individual
reflected waves were not isolated and their travel was not specifically tracked in the
arterial system. Using maps similar to those employed in weather forecasting, the
overall propagation of reflected waves from different origins was expressed as a
function of spatial dimension. The investigation concentrated on both the entire arterial
system and the arterial vasculature of the upper limb.
250
In the first part, the effects of wave travel and wave reflection were simulated. The
pressure pulse in different segments and the reflected waves from the upper limbs, trunk
and lower limbs were isolated. The study into the propagation of these waves
demonstrated the importance of vascular beds around the trunk on global arterial
circulation. The waves originating from the trunk segment formed the secondary
pressure oscillations in every arterial branch. In contrast, the effect of both limbs on
global function was found to be minor. They dictated the initial peak of the pressure
waveforms. These two findings indicate that pulse shape in different arteries may react
differently to different interventions.
In the second part, the formation of the upper limb pressure waveform is examined in
detail. Since these waveforms are widely utilised in clinical situations this detail
treatment was deemed necessary. Under normal conditions the pulses in this limb were
found to be considerably different from those in the central aorta. These differences
were also augmented under certain circumstances such as ageing, vasodilatation and
exercise. Although a second order system representation has often been suggested
previously, the upper limb system was found to be substantially different from a second
order system. It was found to be better modelled as an elastic tube terminated with a
complex load. Positive wave reflections from this termination induce a peaky transfer
function, similar to that of second order systems. However, the resultant phase and
modulus of this transfer function differ considerably from the resonant behaviour of the
second order system. The study then concentrated on the sensitivity of upper limb
system to altered arterial parameters. The upper limb system was found to be generally
insensitive to alterations in wall properties and the terminations within physiological
limits. The changes in peripheral waveforms and the transfer functions were also found
to be minor.
Since no model can be complete without testing its accuracy in its assumptions and
predictions, an experiment was arranged. The same experiment was also used in testing
251
the hardware and developed software. The software had to be tested to verify the coding
of the algorithms. Using iterative techniques the model accurately estimated pressure
waveforms and the physical properties.
The transfer function in the upper limb was characterised in subjects where the input
could be varied. Although in response to changes in heart rates both the input pulse and
the output pulse were altered, a unique expression of transfer function between the
central waveform and the peripheral radial waveforms was derived by resorting to
spectral techniques. Results of this study confirmed the assumptions made about the
linearity of the upper limb arterial system. Model values describing the upper limb
arterial system were derived by analysing the transfer function. Similar values were also
calculated by using parameter estimation techniques.
Reflections from the terminations of the upper limb were altered by compressing the
vessels of the palm vasculature. The systolic pressure and the intensity of reflected
waves increased in parallel with occlusion pressure. This finding confirmed that the
initial part of the upper limb waveform is contaminated by reflected waves originated
from the palm vasculature. Reduction in mean arterial pressure in the upper limb
caused by raising the arm decreased the intensity of reflections. However, when the
externally applied pressure to these vessels became equal to the intra-arterial pressure,
the intensity of reflected waves was decreased. This is consistent with known non-
linear pressure volume relations of the arterial wall. These findings indicate that under
normal conditions the intensity of reflections from this bed is comparable to that
induced by total occlusion of these arteries thus not subject to great variation.
Further investigations were made on patients receiving vasoactive drugs, in order to
assess their effects on wave shapes and propagation properties on the upper limb.
Nitroglycerine (NTG) and sodium nitroprusside (SNP) reduce secondary pressure
oscillations in all arteries. Results obtained in the upper limb indicated that these affect
252
mainly the proximal vessel properties. There was no evidence to suggest reductions in
the reflections occurring at the termination of the upper limb. There was little change in
the primary wave amplitude while the secondary wave amplitude was reduced indicating
the reduction in the intensity of reflected waves in the trunk. This indicates that these
drugs have a differential action by altering the properties of the vascular beds of the
trunk.
These experimental observations on the nature of the primary and secondary peak led
toinvestigations into the trunk segment. In a group of subjects, the oral administration
of glucose significantly altered both the intensity of second peak and the transfer
function of the trunk segment. However, the transfer functions of the upper and lower
limb segments changed little. This differential action of glucose is basically similar to
the action of vasodilators indicating involvement of common mechanisms. To explore
this possibility, a modified T-tube model was constructed to represent the arterial tree.
In this reduced model the lower T-tube was further bifurcated into two circulations, one
to the splanchnic and the other to the lower limb. This reduced model explained these
observations while generating realistic flow waveforms for each state.
The relative stability of the transfer function in the upper limb was exploited to generate
central aortic pressures from pressure recording in the radial artery. When used as a first
order approximation, inverse filtering using a generalised transfer function yielded
reasonable central pressure waveforms. This approach was found to be applicable under
conditions such as vasodilator therapy, exercise or certain physical manoeuvres.
The on-line resynthesis technique together with feature extraction, analysis and storage
of pulses recorded from different sites of the arterial system is an important practical
outcome of this investigation. A cheap, reliable and methodologically sound system
was designed and implemented using available hardware and software tools. This
system has been tested in an out-patient clinic for assisting in current pharmacological
253
management of cardiovascular disease. It was also extended to the critical care area,
where the extra information provided by features of the arterial pulse aid in assessing
patient's condition.
In summary the following conclusions are made:
1. Because of the experimental difficulties in isolating the individual components of
the reflected pressure waves, mathematical models of the human arterial system are
sine qua non in investigating the consequences of pressure wave propagation. It is
also possible to characterise a real system using mathematical models in conjunction
with parameter estimation methods.
2. A functional origin of reflected waves resides in the aortic trunk segment of the
human arterial system. The vasoactive state of the arterial beds of this segment is
the major determinant of the impedance patterns found in the central aorta in
contrast to the vasculature of upper and lower limbs.
3. Reflected waves originating from the arterial beds of the trunk travel into every
arterial branch and constitute the secondary oscillations seen in the central aortic and
upper limb pulses. In contrast, reflected waves from upper and lower limbs are
mainly confined to individual segments and form the initial peak of the pressure
waveforms in these arterial beds. These two distinct components of the arterial
pulse could explain the well documented differences in central and peripheral upper
limb pulses.
4. By relating the spectral components of input and output pressure waves, the wave
propagation properties of any vascular system can be explicitly described by transfer
functions. The information about the intensity and timing of reflected waves can be
obtained not only by analysis of the modulus but also from the phase of the transfer
254
functions. The latter method is superior to the former when the pressure waves are
uncalibrated but recorded with high fidelity instruments. Being such an instrument,
the tonometer is best suited for this method in measurement of altered transmission
properties.
5. In the upper limb arterial system the intensity and timing of reflected waves, ie
reflection coefficient at the palm vasculature, is the most important determinant of
the peak amplitude of the modulus of transfer function. Other factors modify the
characteristics of the transfer function by altering the components at high
frequencies. However, these have little effect on the peripheral wave contour since
the central pressure waveform is band limited.
6. The peaky transfer function in the upper limb has severe implications in clinical
monitoring. Although a linear relationship between the systolic pressure and the
stroke volume is often assumed, simple alterations in heart rate causes significant
deviations from this linear assumption.
7. Under normal circumstances the terminal reflection coefficient in the palm
vasculature approximates to that obtained by total occlusion of this bed. It reduces
to zero or even becomes negative when the transmural pressure experienced by this
bed approximates to zero. This indicates relative stability of the upper limb transfer
function when intra arterial pressure is within normal or above normal range.
8. Vasoacting agents such as nitroglycerine and sodium nitroprusside do not produce a
significant decrease in the reflection coefficient in the palm vasculature. In fact,
they produce a functional increase in reflection coefficient due to increased proximal
arterial compliance. They, however, decrease the reflection coefficient in the trunk
vasculature. This explains the paradoxical disparity of wave shapes after the
administration of these drugs.
255
9. Ingestion of glucose causes similar changes in the arterial pressure wave contours
observed after administration of vasodilators. It causes reductions in the secondary
peak without necessarily effecting the primary peak. These changes in the wave
shapes can be substantiated by analysis of transfer functions of the upper limb, trunk
and lower limb segments which indicates significant reductions in the properties of
the vasculature of the trunk segment.
10. The human arterial tree can be modelled as a modified T-Tube to explain the wide
variety of wave shapes and impedance patterns. As opposed to T-Tube models
constructed for other species, in this model of the human arterial tree the lower
element of the T further bifurcates into two circulations, one to the splanchnic and
the other to the lower limb. This model can be used to obtain realistic impedance
patterns and possible stroke volumes from measurements of arterial pressure waves
along the tree.
11. It is possible to generate central aortic pressures from the peripheral upper limb
pressure recordings. This method uses the relatively stable transfer function of the
upper limb. The technique can be applied to synthesise on-line central pressure
waveforms under conditions such as vasodilator therapy, exercise or certain
physiological manoeuvres. This enables the determination of central aortic
waveform indices from recordings of peripheral upper limb pressure wave contour.
256
REFERENCES
Adelson-Velskii GM and Landis EM (1962). An algorithm for the organization of the information. Dokl. Akad. Nauk SSSR. Mathemat. 146:263-266.
Alexander RS (1953). The genesis of the aortic standing wave. Circ Res 1: 145-151.
Alexander WD and Oake RJL (1977). The effect of insulin on vascular reactivity to norepinephrine. Diabetes. 26:611-614.
Anliker M, Stettler JC, Niederer P and Holenstein R (1978). Prediction of shape changes of propagating flow and pressure pulses in human arteries. In "The arterial system: dynamics, control theory and regulation". Bauer RD and Busse R (eds). Springer Verlag Berlin.
Aperia A (1940). Haemodynamical studies. Scand Arch Physiol 83 Suppl: 1-230.
257
Apter HT (1963). An analysis of the aortic pressure curve during diastole. Bull. Math. Biophys. 25:325
Arts MGJ (1978). A mathematical model of the dynamics of the left ventricle and the coronary circulation. PhD Thesis. Rijksuniverstaeit Limburg. The Netherlands.
Attinger EO (1964). Elements of theoretical hydrodynamics. In "Pulsatile blood flow". Attinger EO (ed).. McGraw Hill, NY.
Atabek HB (1968). Wave propagation through a viscous fluid contained in a tethered, initially stressed, orthotropic elastic tube. Biophys. J. 8:626-649.
Avasthi PS, Greene ER and Voyles WF (1987). Non-Invasive doppler assessment of human postprandial renal blood flow and cardiac output. Am. J. Physiol 252:F1167-F1174
Avolio AP (1976a). Haemodynamic studies and modelling of the mammalian arterial system. PhD Thesis, University of New South Wales.
Avolio AP, O'Rourke MF, Mang K, Bason PT and Gow BS (1976b). A comparative study of arterial haemodynamics in rabbits and guinea pigs. Am J Physiol 230:868-875.
Avolio AP (1980). Multi-branched model of the human arterial system. Med. & Biol. Eng. & Comp. 18: 709-718.
Avolio AP, O'Rourke MF, Bulliman BT, Webster ME and Mang K (1982). Systemic arterial haemodynamics in the Diamond Python (Morelia Spilotes). Am J Physiol 243:R205-R212.
Avolio AP, O'Rourke MF and Webster ME (1983a). Pulse propagation in the arterial system of the Diamond Python (Morelia Spilotes). Am J Physiol 245:R831-R836.
Avolio AP, Chen S-G, Wang R-P, Zhang C-L, Li M-F and O'Rourke MF (1983b). Effects of ageing on changing arterial compliance and left ventricular load in a Northern Chinese urban community. Circulation 68: 50-58.
Avolio AP, Deng F-Q, Li W-Q, Lu Y-F, Huang Z-D, Xing L-F and O'Rourke MF (1985). Effects of ageing on arterial distensibility in populations with high and low
258
prevalence of hypertension: comparison between urban and rural communities in China. Circulation 71: 202-210.
Balar SD, Rogge TR and Young DF (1989). Computer simulation of blood flow in the human arm. J. Biomechanics 22:691-697.
Beneken JEW (1972). Some computer models in cardiovascular research. In "Cardiovascular Fluid Dynamics" Bergel DH (ed). Acad. Press NY.
Bergel, DH, McDonald DA and Taylor MG (1958). A method for measuring arterial impedance using a differential manometer. J. Physiol. 141:17-18p.
Bergel DH (1960a). The static elastic properties of the arterial wall. J. Physiol. 156: 445-457.
Bergel DH (1960b). The dynamic elastic properties of the arterial wall. J. Physiol. 156:458-469.
Bergel DH and Milnor WR (1965). Pulmonary vascular impedance in the dog. Circ. Res. 16: 401-415.
Bergel DH (1978). Mechanics of the arterial wall in health and disease. In "The arterial system: dynamics, control theory and regulation". Bauer RD and Busse R (eds). Springer Verlag Berlin.
Blackman RB and Tuckey JW (1958). The measurement of power spectra. Dover.
Borow KM and Newburger JW (1982). Noninvasive estimation of central aortic pressure using the oscillometric method for analysing systemic artery pulsatile blood flow: Comparative study of indirect systolic, diastolic and mean brachial artery pressure with simultaneous direct ascending aortic pressures. Am Heart J 103:879-886
Bramwell JC and Hill AV (1922). Velocity of transmission of the pulse wave and elasticity of arteries. Lancet 891-892.
Brin KP and Yin FCP (1984). Effect of nitroprusside on wave reflections in patients with heart failure. Annals Biomed Eng 12: 135-150.
259
Broadbent WH (1890). The Pulse. London: Cassell & Co. Ltd.
Bourgeois M, Gilbert BK, Donald DE and Wood EH (1974). Characteristics of aortic diastolic pressure decay with application to the continuos monitoring of changes in peripheral vascular resistance. Cardiovasc. Res. 35:56-66.
Burattini R (1989a). Reduced models of the systemic arterial circulation. In "Vascular dynamics". Westerhof N and Gross DR (eds). Plenum Press
Burattini R, Knowlen GG and Campbell KB (1989b). Two arterial effective reflecting sites may appear as one to the heart. Circ. Res. 68:85-99
Campbell KB, Lewis CL, Frasch HF and Noordergraaf A (1989). Pulse reflection sites and effective length of the arterial system. Am J Physiol. 256:H1684-1689.
Carew TE, Vaishnav RN and Patel DJ (1968). Compressibility of the arterial wall. Circ Res. 23:61-68.
Caro CG, Pedley TJ, Schroter RC and Seed WA (1978). The mechanics of the circulation. Oxford University Press.
Cevenini G, Barbini P, Cappello A and Avanzolini G (1987). Three element model for total systemic circulation: Emphasis on the accuracy of parameter estimates. J. Biomech. Eng. 9:374-378.
Chang KC, Hsieh JS, Kuo TS and Chen HI (1990). Effects of nifedepine on systemic hydrolic vascular load in patients with hypertension. Cardiovasc. Res. 24:719-726.
Chadwick RS, Goldstein DS and Keiser HR (1986). Pulse-wave model of brachial arterial pressure modulation in ageing and hypertension. Am. J. Physiol. 251:H1-H11.
Chiu YC, Arand PW, Shroff SG, Feldman T and Carroll JD (1991). Determination of pulse wave velocities with computerized algorithms. Am. Heart J. 121:1460-1470.
Chou CC (1983). Splanchnic and overall cardiovascular haemodynamics during eating and digestion. Fed. Proc. 42:1658-1661
260
Conrad WA (1969). Pressure flow relationships in collapsible tubes. IEEE Trans. Biomed. Eng. BME-16:284-295.
Cope FW (1965). Elastic reservoir theories of the human circulation with applications to clinical medicine and to computer analysis of the circulation. In "Advances in Biol. Med. Phys." Vol 10. Acad. Press, N.Y.
Cornyn JW, Massie BM, Unverferth DV and Leier CV (1986). Haemodynamic changes after meals and placebo treatment in chronic congestive heart failure. Am. J. Cardiol. 57:238-241.
Corey PD, Wemple RR and Werff TJV (1975). A combined left ventricular/ Systemic arterial model. J Biomech. 8:9-15.
Cox RH (1969). Comparison of linearized wave propagation models for arterial blood flow analysis. J. Biomechanics 2:251-265
Cox RH (1971). Determination of the true phase velocity of arterial pressure waves in vivo. Circ Res 29: 407-418.
Cox RH (1978). Arterial smooth muscle mechanics. In "The arterial system: dynamics, control theory and regulation". Bauer RD and Busse R (eds). Springer Verlag Berlin.
Crandall I (1927). Theory of vibrating systems and sound. Van Nostrand, NY.
de Pater L and Van den Berg (1963). An electrical analogue of the entire human circulatory system. Medical Elecronics. 797-808.
Dai K, Xue H, Dou R and Fung YC (1985). On the detection of messages carried in arterial pulse waves. J. Biomech. Eng. 107:268-273
Dick DE, Kendrick JE, Watson GL and Rideout VC (1968). Measurement of nonlinearity in the arterial system of the dog by a new method. Circ. Res. 22:101-111
Dinnar U (1981). Cardiovascular fluid dynamics. CRC Press, Florida.
261
Dontas AS (1960). Comparison of simultaneously recorded intra-arterial and extra-arterial pressure pulses in man. Am. Heart. J. 59:576-589
Drzewiecki GM and Noordergraaf A (1980). Positioning effects in arterial tonometry. 33rd Annual conference on Engineering in Medicine and Biology, Washington, DC. p60.
Drzewiecki GM, Melbin J and Noordergraaf A (1983). Arterial tonometry: review and analysis. Journal of Biomechanics 16: 141-153.
Einav S, Aharoni S and Manaoach M (1988). Exponentially tapered transmission line model of the arterial system. IEEE. Trans. Biomed. Eng. 35:333-339.
Elad D, Foux and A Kivity Y (1988). A model for the nonlinear elastic response of large arteries. J. Biomech. Eng. 110:185-189.
Fagan TC, Phyllis RS, Gourley LA, Lee JT and Gaffney TE (1986). Postprandial alterations in haemodynamics and blood pressure in normal subjects. Am. J. Cardiol. 58:536-641
Falsetti HL, Mates RE, Carroll RJ, Gupta MS and Bell AC (1974). Analysis and correction of wave distortion in fluid filled catheter systems. Circulation 49:165-172
Fich S, Welkowitz W and Hiltor R (1966). Pulsatile blood flow in the aorta. In " Biomedical fluid mechanics symp". ASME
Fishman AP and Richards DW (1964). Circulation of the blood: Men and Ideas. Oxford University Press, NY.
Fitzhugh R (1969). Mathematical models of excitation and propagation in nerve. In "Biological Engineering", Schwan HP (Ed), McGraw-Hill, NY
Fitchett D, Simkus G, Beaudry J and Marpole D (1988). Reflected pressure waves in the ascending aorta: effect of glyceryl trinitrate. Cardiovasc Res 22: 494-500.
Frank O (1899). Die grundfurm des arteriellen puls. Zeitschreiff fur Biologie 37: 483-526.
262
Frank O (1903). Kritik der elastischen manometer. Zeitschreiff fur Biologie Munich 44: 445-613.
Frank O (1905). Der puls in den arterien. Zeitschreiff fur Biologie 45: 441-553.
Frank O (1930). Schatzung des schlargvolumens des menschlichen herzens auf grund der Wellenund Windkessel theorie. Zeitschreiff fur Biologie 90: 405-409.
Franklin KJ (1979). The circulation of the blood, William Harvey's. JM Dent and Sons, London.
Freis ED and Kyle MC (1968). Computer analysis of carotid and brachial pulse waves. Effects of age in normal subjects. Am J Cardiol 22: 691-695.
Fry DL Nobel FW and Mallos AJ (1957). Electric device for instantenous and continouos computation of aortic blood velocity. Circ. Res. 5:75
Fry DL and Greenfield JC Jr (1964). The mathematical approach to haemodynamics with particular reference to Womersley's theory. In "Pulsatile blood flow".Attinger EO(ed). McGraw Hill, New York.
Fung YC (1984). Biodynamics: Circulation. Springer Verlag, NY.
Fujii M, Yaginuma T, Takazawa K, et al (1987). Non-invasive detection for reflection wave in the arterial system. Automedica 9: 49.
Gabe IT (1965). The measurement of oscillatory blood flow and impedance in the human external iliac artery. Clin. Sci. 29:45-58
Gabe IT (1972). Pressure measurement in experimental physiology. In "Cardiovascular Fluid Dynamics" Bergel DH (ed). Acad. Press NY.
Gardner RM (1982). Blood pressure monitoring: Sharing common elements, problems. EMB Magazine March:28-31
263
Geddes LA, Hoff HE and Badger AS (1966). Introduction of the ascultatory method of measuring blood pressure: Including a translation of Korotkoff's original paper. Cardiovasc. Res. Center. Bull. 5:57-74
Geddes LA (1970). Direct and indirect measurement of blood pressure. Chicago. Year Book Medical Publishers.
Gerber MJ, Hines RL and Barash P (1987). Arterial waveforms and systemic vascular resistance: Is there a correlation. Anesthesiology. 66:823-825.
Gow BS and Taylor MG (1968). Measurement of viscoelastic properties in the living dog. Circ. Res. 23:111-122.
Gow BS (1972). The influence of vascular smooth muscle on the viscoelastic properties of blood vessels. In "Cardiovascular Fluid Dynamics". Bergel DH (Ed). Acad. Press, NY.
Greenfield JC (1966). Pressure gradient technic. In "Methods in medical research". Rushmer RF and Fry DL (ed) Vol 11. Year Book Med. Publ, Chicago.
Guier WH (1981). A haemodynamic model for relating phasic pressure and flow in the large arteries. IEEE Trans. Biomed. Eng. 8:479-482.
Gundel W, Cherry G, Rajagopalan B, Tan LB, Lee G and Shultz D (1981). Aortic input impedance in man: acute response to vasodilator drugs. Circulation 63: 1305-1314.
Haimovici H (1981). Arterial circulation of the extremities. In "Structure and function of the circulation". Schwartz (Eds), Plenum Press. NY.
Hales S (1733). Haemastatiks. Hafner publishing Co, London.
Hamilton WF (1944). The patterns of the arterial pulse. Am J Physiol 141: 235-241.
Hamming RW (1977). Digital filters. Prentice-Hall International, London.
264
Hardung (1964). Input impedance and reflection of pulse waves. In "Pulsatile blood flow". Attinger EO (ed).. McGraw Hill, NY.
Harris CRS (1981). The arteries in Greco-Roman medicine. In "Structure and function of the circulation". Schwartz CJ et al (eds), Plenum Press. NY.
Harvey W (1628). "Esercitato anatomica de motu cordis sanguinis in animalibus". Franklin KJ (1979). Blackwell, Oxford
Hayward C (1986). Non-invasive pressure pulse detection in the cardiovascular system. BSc(Med) Thesis, University of New South Wales, Australia.
Helal MA, Watts KC, Marble AE and Sarwal SN (1990). Theoretical model for assessing haemodynamics in arterial networks which include bypass grafts. Med. Biol. Eng. Comp, 28:465-473
Herok G (1992). Changes in cardiac output and in arterial pulse waveform in subjects with implantable programmable pacemakers using impedance cardiography and arterial tonometry. MBiomedE. Thesis. Center for Biomedical Engineering. University of New South Wales. Sydney.
Ho KK (1982). Effects of ageing on arterial distensibility and left ventricular load in an Australian population. BSc(Med) Thesis, University of New South Wales, Australia.
Hodgin AL and Huxley AF (1952). A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117:500-544.
Iberal AS (1950). Attenuation of oscillatory pressures in instrument lines. J. Res. Natn. Bur. Stan. 45:85-108.
Imura T, Yamamoto K, Satoh T, Kanamori K, Mikama T and Yasuda H (1990). In vivo viscoelastic behavior in the human aorta. Circulation 66:1413-1419.
Jager GN, Westerhof N and Noordergraaf A (1965). Oscillatory flow impedance in an electrical analogue of the arterial system: representation of sleeve effect and non-newtonian properties of blood. Circ. Res. 26:121-133.
265
Jansen RWMM Peeters TL Van Lier HJJ and Hoefnagels WHL (1990). The effect of oral glucose, protein fat and water loading on blood pressure and the gastrointestinal peptides VIP and somatostatin in hypertensive elderly subjects. Eur.J. Clin. Invest. 20:192-198
Jern S (1991). Effects of acute carbohydrate administration on central and peripheral haemodynamic responses to mental stress. Hypertension 18:790-797
Johnston KW and Kakkar VV (1974). Noninvasive measurement of systolic pressure slope. Arch. Surg. 108:52-56.
Kannel WB, Wolfe PA, McGee DI, Dawber TR, McNamare P, Castelli WP (1981). Systolic blood pressure, arterial rigidity and risk of stroke (The Framingham Study). J Am Med Assoc 1225-1229.
Karamanoglu M, Avolio AP and O'Rourke MF (1990). Simulation of the effects of lower body wave reflection on aortic and brachial pulse in man: Implications to therapy. Proc.of IEEE-EMBS 12:1154-1155
Karamanoglu M, O'Rourke MF, Avolio AP and Kelly RP (1992). An analysis of the relationship between central aortic and peripheral upper limb pressure waves in man. European Heart J (in press)
Kaspar L, Karnik R and Slany J (1987). Postprandial increase in cardiac output in patients with acute myocardial infarction. Int. J. Cardiol 15:177-183
Kato M, Naruse S, Takagi T and Shionoya S (1989). Postprandial gastric blood flow in conscious dogs. Am. J. Physiol. 257:G111-G117
Kawasaki T, Sasayama S, Yagi S, Asakawa T and Hirai T (1987). Non-invasive assessment of the age related changes in stiffness of major branches of the human arteries. Cardiovasc. Res. 21:678-687.
Keats TE (1990). Atlas of Roentgenographic measurement (6th Ed). Mosby Yearbook,Baltimore.
Kelly RP, Hayward CS, Avolio AP and O'Rourke MF (1989a). Non-invasive determination of age-related changes in the human arterial pulse. Circulation 80:1652-1659.
266
Kelly RP, Hayward CS, Ganis J, Daley JE, Avolio AP and O'Rourke MF (1989b). Non-invasive registration of the arterial pressure pulse waveform using high-fidelity applanation tonometry. J. Vasc. Med. Biol. 1:142-149.
Kelly RP (1990a). Systemic arterial function in health and disease: Clinical determinants and measurement. MD Thesis. University of New South Wales. Sydney
Kelly RP, Gibbs HH, Morgan JJ, Daley JE, Mang K, Avolio AP and O'Rourke MF (1990b). Nitroglycerine has more favourable effects on left ventricular afterload than apparent from measurement of pressure in a peripheral artery. European Heart J 11: 138
Kenner T (1978). Models of the arterial system. In "The arterial system: dynamics, control theory and regulation". Bauer RD and Busse R (eds). Springer Verlag Berlin.
Korotkoff NC (1905). On the subject of methods of determining blood pressure. Wien med Wschir 11: 365.
Kroeker EJ and Wood EH (1955). Comparison of simultaneously recorded central and peripheral arterial pressure pulses during rest, exercise and tilted position in man. Circ. Res. 3: 623-632.
Krovetz J, Jennings RB and Goldbloom S (1974). Limitation of correction of frequency dependent artefact in presure recordings using harmonic analysis. Circulation. 50:992-997.
Langewouters GJ, Wesseling and Goedhard (1984). The static elastic properties of 45 human thoracic and 20 abdominal aortas in vitro and the parameters of a new model. J. Biomech. 17:425-435.
Lasance HAJ, Wesseling KH and Ascoop CA (1976). Peripheral pulse contour analysis in determining stroke volume. Progress Report 5. pp59-62 Inst Med Phys DA Costake 45, Utrecht,.
Laskey WK, Kussmaul WG, Martin JL, Kleaveland JP, Hirshfield JW and Shroff S (1985). Characteristics of vascular hydraulic load in patients with heart failure. Circulation 72: 61-71.
267
Laskey WK and Kussmaul WG (1987). Arterial wave reflection in cardiac failure. Circulation 75: 711-722.
Latham RD, Westerhof N, Sipkema P, Rubal BJ, Reuderink P and Murgo JP.(1985). Regional wave travel and reflections along the human aorta: a study with six simultaneous micromanometric pressures. Circulation 72: 1257-1269.
Latson TW, Hunter WC, Katoh N and Sagawa K (1988). Effect of nitroglycerin on aortic impedance, diameter and pulse wave velocity. Circ Res 62: 884-890.
Lax H and Feinberg AW (1959). Abnormalities of the arterial pulse wave in young diabetic subjects. Circulation 20:1106-1110
Learoyd BM and Taylor MG (1966). Alterations with age in the viscoelastic properties of human arterial walls. Circ Res 18: 278-292.
Liang CS, Doherty, JU, Faillace R, Maekawa K, Arnold S, Gavras H and Hood WB (1982). Insulin infusions in consious dogs. Effects on systemic and coronary haemodynamics, reginal bloood flows and plasma catecholamines. J. Clin. Invest. 69:1321-1336.
Li JK, Melbin J and Noordergraaf A (1984). Directional disparity of pulse reflection in the Dog. Am. J. Physiol. 247:H95-H99.
Liu Z, Shen F and Yin FCP (1989). Impedance of arterial system simulated by viscoelastic T tubes terminated in Windkessels. Am. J. Physiol. 256:H1087-1096.
Ling BS and Atabek HB (1972). A nonlinear analysis of blood flow in arteries. J. Fluid Mech. 55:493-511.
Mackenzie J (1902). The study of the pulse: arterial, venous and hepatic, and of the movements of the heart. Edinburgh: Young J. Pentland.
Mahomed FA (1874). The aetiology of Bright's disease and the prealbuminuric stage. Med Chir Trans 57: 197-228.
Mahomed FA (1892). The physiology and clinical use of the sphygmograph. Medical Times and Gazette 1: 62-64, 128-130, 220-222.
268
Marey E.J (1860). Recherches sur le pouls au moyen d'un nouvel apareil enregistreur - le sphygmographe. Paris: E Thunot et Cie.
Marmor AT, Blondheim DS, Gozlan E, Navo E and Front D (1987). Method for noninvasive measurement of central aortic systolic pressure. Clin. Cardiol. 10:215-221.
McDonald DA (1955). The relationship of pulsatile pressure to flow in arteries. J Physiol 127: 532-552.
McDonald DA (1960). Blood flow in arteries. 1st ed. London, Edward Arnold.
McDonald DA (1968). Regional pulse wave velocity in the arterial tree. J Appl. Physiol. 24: 73-78.
McDonald DA (1974). Blood flow in arteries. 2nd ed. London, Edward Arnold.
McIlroy MB, Seitz WS and Targett RC (1986). A transmission line model of the normal aorta and its branches. Cardiovasc Res 20: 581-587.
Merillon JP, Fontenier G, Leralluit JF, et al (1984). Aortic input impedance in heart failure: comparison with normal subjects and its changes during vasodilator therapy. Eur Heart J 5: 447-55.
Mills CJ, Gabe IT, Gault JH, et al (1970). Pressure-flow relationships and vascular impedance in man. Cardiovasc Res 4: 405-417.
Milnor WR, Bergel DH and Bargainer JD (1966). Hydrolic power associated with pulmunary blood flow and its relation to heart rate. Circ. Res. 19:467-480.
Milnor WR. (1978). Influence of arterial impedance on ventricular function. In "The arterial system: dynamics, control theory and regulation". Bauer RD and Busse R (eds). Springer Verlag Berlin.
Milnor WR. (1989). Haemodynamics. 2nd Edition. Williams and Wilkins. Baltimore.
Morgan GW and Kiely JP (1954). Wave propagation in a viscous liquid contained in a flexible tube. J Acoust. Soc. Am. 26:323-328.
269
Moneta GL, Taylor DC, Helton WS, Mulholland MW and Strandness DE (1988). Duplex ultrasound measurement of postprandial intestinal blood flow: Effect of meal composition. Gastroenterology 95:1294-1301
Murgo JP, Westerhof N, Giolma JP and Altobelli SA (1980a). Effects of exercise on aortic input impedance and pressure wave shapes in normal man. Circ. Res. 48: 334-343.
Murgo JP, Westerhof N, Giolma JP and Altobelli SA. (1980b). Aortic input impedance in normal man: relationships to pressure wave forms. Circulation 62: 105-116.
Murgo JP, Westerhof N, Giolma JP and Altobelli SA. (1981). Manupulation of ascending aortic pressure and flow wave reflections with the Valsalva manouvre: Relation to input impedance. Circulation 63:122-132.
Murrel W (1870). Nitroglycerine as a remedy for angina pectoris. Lancet Jan-Feb:80-227
Newman DL, Greenwald SE and Bowden NL (1979). An in vivo study of the total occlusion method for the analysis of forward and backward pressure waves. Cardiovasc Res 13: 595-600.
Niederer P and Schilt W (1988). Experimental and theretical modelling of intra-aortic baloon pump operation. Med.& Biol.& Eng. Comput. 26:167-174
Nichols WW, O'Rourke MF, Avolio AP, et al (1987). Age-related changes in left ventricular /arterial coupling. In "Ventricular/vascular coupling". Yin FCP (ed). Springer-Verlag, NY.
Nichols WW and O'Rourke MF (1990). McDonald's Blood Flow in Arteries. 3rd edition. Arnold, London.
Noble MIM (1979). Left ventricular load, arterial impedance and their inter-relation. Cardiovasc Res 13: 183-198.
Noordergraaf A (1969). Haemodynamics. In "Biological Engineering", Schwan HP (Ed), McGraw-Hill, NY
270
Natali A, Buzzigoli G, Minaker KL, Stevens Al, Pallota J and Landsberg L (1990). Effects of insulin on haemodynamics and metobolism in human forearm. Diabetes. 39:490-500.
Oliver G (1908). Studies in blood pressure: physiological and clinical. London: Lewis, HK.
Oppenheim AV, Willsky AS and Young IT (1983). Signals and systems. Prentice-Hall
O'Rourke MF (1965). Pressure and flow in arteries. MD Thesis, The University of Sydney.
O'Rourke MF, Taylor MG (1966). Vascular impedance of the femoral bed. Circ. Res. 18:126-139.
O'Rourke MF, Taylor MG (1967a). Input impedance of the systemic circulation. Circ. Res. 20:365-380.
O'Rourke MF (1967b). Pressure and flow waves in systemic arteries and the anatomical design of the arterial system. J Appl Physiol 23: 139-149.
O'Rourke MF, Blazek JV, Morreels CL and Krovetz LJ (1968). Pressure wave transmission along the human aorta. Changes with age and in arterial degenerative disease. Circ Res 23: 567-579.
O'Rourke MF (1970). Influence of ventricular ejection on the relationship between central aortic and brachial pressure pulse in man. Cardiovasc Res. 4: 291-300.
O'Rourke MF and Avolio AP (1980). Pulsatile flow and pressure in human systemic arteries. Studies in man and in a multibranched model of the human systemic arterial tree. Circ. Res. 46: 363-372.
O'Rourke MF (1982). Arterial function in health and disease. Churchill Livingstone, Edinburgh.
271
O'Rourke MF, Avolio AP and Nichols WW (1987). Left venticular-systolic arterial coupling in humans and strategies to improve coupling in disease states. In "Ventricular/vascular coupling" Yin FCP (ed). Springer Verlag, New York.
O'Rourke MF (1988). What is the arterial pressure? Aust NZ J Med 18: 649-650.
O'Rourke MF, Kelly RP and Avolio AP (1992). The arterial pulse. Williams & Wilkins, Baltimore.
Osler W (1921). The evolution of modern medicine, Yale University Press. New Haven.
Pauca Al, Hudspedth AS, Wallenhaupt SL, Tucker WY, Kon ND, Mills SA and Cordell AR (1989). Radial artery-to-aorta pressure difference after discontinuation of cardiopulmunary bypass. Anesthesiology. 70:935-941
Papageorgiou GL and Jones NB (1988). Wave reflection and hydrolic impedance in the healhy arterial system: a controversial subject. Med. & Biol. Eng. & Comput. 26:237-242.
Parker KH and Jones CJH (1990). Forward and backward running waves in the arteries: Analysis using the method of characteristics. J. Biomech. Eng. 112:322-326.
Patel DJ, Freitas FM and Fry DL (1963). Hydrolic input impedance to aortic and pulmonary artery in dogs. J. Appl. Physiol. 18: 134-140.
Patel DJ, Greenfield JC, Austen WG, Morrow AG and Fry DL (1965). Pressure-flow relationships in the ascending aorta and femoral artery in man. J. Appl. Physiol. 20: 459-463.
Patel DJ and Vaisnav RN (1972). The rheology of large blood vessels. In "Cardiovascular Fluid Dynamics" Bergel DH (ed). Acad. Press NY.
Penaz J (1973). Photoelectric measurement of blood pressure, volume, and flow in the finger. Digest of the 10th International Conference on Medical Engineering, Dresden, Germany.
272
Pepine CJ, Nichols WW, Curry RC Jr and Conti CR (1979). Aortic input impedance during nitroprusside infusion: a reconsideration of afterload reduction and beneficial action. J Clin Invest 64: 643-654.
Pickering GW (1982). Systemic arterial hypertension. In: "Circulation of the blood, men and ideas". Fishman AP and Richards DW (Eds). Waverly Press. Baltimore.
Porenta G, Young DF and TR Rogge (1986). A finite-element model of blood flow in arteries including taper, branches and obstructions. J. Biomech. Eng. 108:161-167.
Porjé IG (1946). Studies of the arterial pulse wave, particularly in the aorta. Acta Physiologica Scandinavica 13: Supplement 42: 168.
Posey JA, Geddes LA, Williams H and Moore AG (1969). The meaning of the point of maximal oscillation in cuff pressure in the indirect measurement of blood pressure. Cardiovasc. Re. Cnt. Bull. 8: 15-25.
Ramsey M III (1976). Device for indirect noninvasive automatic mean arterial pressure. Proc.of IEEE-EMBS 29:79
Remington JW and Hamilton WF (1947). Quantitative calculation of the time course of cardiac ejection from the pressure pulse. Am J Physiol 148: 25-34.
Remington JW, Noback CR, Hamilton WF and Gold IJ (1948). Volume elasticity characteristics of the human aorta and prediction of the stroke volume from the pressure pulse. Am. J. Physiol 153-25.
Remington JW and Wood EH (1956). Formation of peripheral pulse contour in man. J Appl Physiol 9:433-442.
Remington JW and Obrien LJ (1970). Construction of aortic flow pulse from pressure pulse. Am. J. Physiol. 218:437-447
Remington JW (1974). Pressure-flow relations in the arterial system. In "Medical Engineering", Ray CD (ed), Year Book Publ Inc, Chicago.
Riva-Rocci S (1896). Un nuovo sfigmomanometro. Gass med ital., Torino 47: 981.
273
Risoe C, Simonsen S, Rootwelt K and Smiseth OA (1990). Nitroprusside redistributes blood to intestinal capacitance vessels. Circulation 82:suppl III p386.
Rockwell RL (1969). Nonlinear analysis of pressure and shock waves in blood vessels. PhD Thesis. Stanford University. California.
Rowell LB, Brengelmann GL, Blackmon Jr and Murray JA (1968). Disparities between aortic and peripheral pulse pressures induced by upright exercise and vasomotor changes in man. Circulation 37: 954-964.
Rodbard S, Rubeinstein HM and Rosenblum S (1957). Arrival time and calibrated contour of the pulse wave, determined indirectly from recordings of arterial compression sounds. Am. Heart. J., Feb:205-212
Ryan JM, Stacy RW and Watman RN (1956). Role of abdominal aortic branches in pulse wave contour changes. Circ. Res. 4:676-679.
Sagawa K (1981). The end-systolic pressure-volume relation of the ventricle: definition, modifications and clinical use. Circulation 63: 1223-1227.
Salotto AG, Muscarella LF, Melbin J, Li JKJ and Noordergraaf A (1986). Pressure pulse transmission into vascular beds. Microvasc. Res. 32:152-163.
Salans AH, Katz LN, Graham GR, Gordon A, Elisberg EI and Gerber A (1951). A study of central and peripheral arterial pressure pulse in man: correlation with simultaneously recorded electrokymograms. Circulation 4: 510.
Schechter DC, Lillehei CW and Soffer A (1969). Sphygmology and heart block. Dis. Chest 55:Suppl 1:535-579.
Schonfeld JC (1953). analogy of hydraulical, mechanical acoustic and electrical systems. Appl.Sci.Res. 3:417-450
Schwartz CJ, Werthessen NT and Wolf S (1981). Structure and function of the circulation, Plenum Press, NY.
Schwid HA, Taylor LA and Smith NT (1987). Computer model analysis of the radial artery pressure wave. J. Clin. Monitor. 3: 220-228.
274
Scott AR, Bennet T and Macdonald IA (1988). Effects of hyperinsulinaemia on the cardiovascular responses to graded hypovolaemia in normal and diabetic subjects. Clin. Sci.. 75:85-92.
Shammas N (1988). Applying Turbo Pascal library units. Wiley.
Simkus GJ and Fitchett DH (1990). Radial arterial pressure measurements may be a poor guide to the beneficial effects of nitroprusside on left ventricular systolic pressure in congestive heart failure. Am J Cardiol 66:323-326.
Sipkema P and Westerhof N (1975). Effective length of the arterial system. Annals of Biomedical Engineering 3: 296-307.
Skalak R (1972). Synthesis of a complete circulation. In "Cardiovascular Fluid Dynamics", Bergel DE (ed), Academic Press, NY.
Snellen HA (1980). Marey EJ and Cardiology: Physiologist and Pioneer of Technology (1830-1904). Kooyker Scientific Publications, Rotterdam.
Sphygmomanometers (1989). Australian Standard AS 3655. Standards Australia, Sydney
Sunagawa K, Maughan WL, Burkhoff D and Sagawa K (1983). Left ventricular interaction with arterial load studied in isolated canine ventricle. Am J Physiol 245: H773-H780.
Takagi T, Nanuse S and Shinoya S (1988). Postprandial celiac and superior mesentric blood flows in conscious dogs. Am. J. Physiol 255:G522-G528
Takazawa K (1987). A clinical study of the second component of left ventricular systolic pressure. J Tokyo Med College. 45: 256-270.
Taylor MG (1957a). An approach to an analysis of the arterial pulse wave: I. Oscillations in an alternating line. Phys. Med. Biol. 1:258-269.
Taylor MG (1957b). An approach to the analysis of the arterial pulse wave II. Fluid oscillations in an elastic tube. Phys in Med Biol 1: 321-329.
275
Taylor MG (1959). An experimental determination of the propagration of fluid oscillation in a tube with a viscoelastic wall; together with an analysis of the characteristics required in an electrical analoque. Phys Med Biol 4: 63-82.
Taylor MG (1964). Wave travel in arteries and the design of the cardiovascular system. In "Pulsatile blood flow". Attinger EO (ed).. McGraw Hill, NY.
Taylor MG (1965). Wave travel in a non-uniform transmission line, in relation to pulses in arteries. Phys in Med and Biol 19: 539-550.
Taylor MG (1966a). The input impedance of an assembly of randomly branching tubes. Biophys. Journal 6: 29-51.
Taylor MG (1966b). Wave transmission through an assembly of randomly branching tubes. Biophys. Journal 6: 697-716.
Taylor MG (1966c). Use of random excitation and spectral analysis in the study of frequency dependent parameters of the cardiovascular system. Circ. Res. 18: 585-595.
Taylor MG (1991). Opening Address. In "Proceedings of the 8th Biennial Conferance". Suttor JR and Belnave R. (Eds), Cumberland Collage Health Sciences, Sydney.
Ting CT, Chang MS, Wang SP, Chiang BN and Yin FCP (1990). Regional pulse wave veolcities in hypertensive and normotensive humans. Cardiovasc. Resc. 24:865-872.
Ur A and Gordon M (1970). Origin of Korotkoff sounds. Am. J. Physiol. 218:524-529.
Van Bergen FH, Weatherhead DS, Treloar AE, Dobkin AB and Buckley JJ (1954). Comparison of indirect and direct methods of measuring arterial blood pressure. Circulation 10: 481-490.
Van Den Bos GC, Westerhof N, Elzinga G and Sipkema P (1976). Reflection in the systemic arterial system: effects of aortic and carotid occlusion. Cardiovasc Res 10: 565-573.
276
Wang YYL, Chang SL, Wu YE, Hsu TL and Wang WK (1991). Resonance. The missing phenomenon-in haemodynamics. Circ. Res. 69:246-249.
Warner HR (1955). Synthesis of central arterial pulse contour from recording of radial artery pressure in man. Am. J. Physiol. 183:670.
Warner HR (1957). A study of the mechanisms of pressure wave distortion by arterial walls using an electrical analogue. Circ. Res. 5:79-84
Warnes CA, Harris PC and Fritts HW (1983). Effect of elevating the wrist on the radial pulse in aortic regurgitation: Corrigan revisited. Am. J. Cardiol. 51:1551-1553.
Watt TB and Burrus C (1976). Arterial pulse contour analysis for estimating human vascular properties. J. Appl. Physiol. 40:171-176.
Weber KT and Janicki JS (1977). Myocardial oxygen consumption: the role of wall force and shortening. Am J Physiol 233: H421-H430.
Webster C (1981). Medieval and renaissance interpretations of the cardiovascular system. In "Structure and function of the circulation". Schwartz CJ et al (Eds), Plenum Press. NY.
Wei LY and Chow P (1985). Frequency distribution of human pulse spectra. IEEE. TRans. Biomed. Eng. BME-32:245-246.
Westling H, Jansson L, Jonson B and Nilsen R (1984). Vasoactive drugs and elastic properties of human arteries in vivo, with special reference to the action of nitroglycerine. Eur Heart J 5: 609-16.
Wesseling KH, Settels JJ, Van Der Hoeven MA, Nijboer JA, Butijn MWT and Dorlas JC (1985). Effects of peripheral vasoconstriction on the measurement of the blood pressure in the finger. Cardiovasc Res 19: 139-145.
Wesseling KH, de Wit B, Settels JJ and Klawer WH (1982). On the indirect registration of blood pressure after Peñáz. Funkt Biol Med 1:245-250.
Westerhof N, Bosman F, De Vries CJ and Noodergraaf A (1969). analogue Studies of the human systemic arterial tree. Journal of Biomechanics 2: 121-143.
277
Westerhof N and Noodergraaf A (1970). Arterial viscoelasticity: a generalised model. Effect on input impedance and wave travel in the systemic tree. Journal of Biomechanics 3: 357-379.
Westerhof N, Sipkema P, Van Den Bos GC and Elzinga AG (1972). Forward and backward waves in the arterial system. Cardiovasc Res. 6: 648-656.
Westerhof N, Elzinga AG, Sipkema P and Van den Bos GC (1977). Quantitative analysis of the arterial system and heart by means of pressure-flow relations. In "Cardiovascular flow dynamics and measurements". Hwang NHC and Normann NA (eds). University Park Press. Baltimore.
Wetterer E (1954). Flow and pressure in the arterial system, their haemodynamic relationship and the principles of measurement. Minnesota Medicine 37: 77-86.
Whirlow DK and Rouleau WT (1965). Periodic flow of a viscous liquid in a thick-walled elastic tube. Bull. Math. Biophys. 27:355-370.
Wiggers CJ (1928). Pressure pulses in the cardiovascular system. Longmans, Green and Company, London.
Witzig K (1914). Uber erzwunge wellenbewegung en zaherm inkombressibler flussigkeiten in elastichen rohren. Inaug. Diss. Bern. Bern Wyss.
Womersley JR (1955a). Method for calculation of velocity, rate of flow and viscous drag in arteries when the pressure gradient is known. J Physiol. 127:553-563.
Womersley JR (1955b). Oscillatory motion of a viscous liquid in a thin walled elastic tube. I. Linear approximation for long waves. Phil Mag. 46:199-221.
Womersley JR (1957a). Oscillatory flow in arteries: the constrained elastic tube as a model of arterial flow and pulse transmission. Phys Med Biol 2: 178-187.
Womersley JR (1957b). The mathematical analysis of the arterial circulation in a state of oscillatory motion. Wright air development center, technical report, WADC-TR-56-614.
278
Womersley JR (1958). Oscillatory flow in arteries: The reflection of the pulse wave at junctions and rigid inserts in the arterial system. Phys. Med. Biol. 2:313-323
Wood EH, Fuller J, Clagett OT (1951). Intraluminal pressures recorded simultaneously from different arteries in man. Am J Physiol 167: 838-839.
Wright JL and Wood EH (1958). The value of cantral and peripheral intra-arterial pressures and pulse contours in cardiovascular diagnosis. Minnesota Medicine 41:215-222.
Xue H and Fung YC (1989). What Nei Jing and Mai Jing say about arterial pulse waves and our attempts to illustrate some of their statements. J. Biomech. Eng. 111:88-91.
Yaginuma T, Avolio A, O'Rourke M et al (1986). Effect of glyceryl trinitrate on peripheral arteries alters left ventricular hydraulic load in man. Cardiovasc Res 20: 153-160.
Yamakoshi KI, Shimazu H, Togawa T (1980). Indirect measurement of instantaneous arterial blood pressure in the human finger by the vascular unloading technique. IEEE Trans. on Biomed. Eng. BME-27: 150-155.
Yen RT, Zhuang FY, Fung YC, Ho HH, Tremer H and Sobin SS (1984). Morphometry of cat's pulmonary arterial tree. J. Biomech. Eng. 106:131-136.
Yin FCP, Guzman PA, Brin KP, et al (1983). Effect of nitroprusside on hydraulic vascular loads on the right and left ventricles of patients with heart failure. Circulation 67: 1330-1339.
Yin FCP (1987a). Ventricular/vascular coupling: clinical, physiological and engineering aspects. Springer, New York.
Young ST Wang WK, Chang LS and Kuo TES (1988). Specific frequency properties of renal and superior mesentric arterial beds in rats. Cardiovasc. Res. 23:465-567
Young T (1808). Hydrolic investigations subservient to an intended Croonian lecture on the motion of the blood. Phil. Trans. Roy.Soc. 98:164-186
Zamir M (1978). Nonsymmetrical bifurcations in arterial branching. J. Gen. Physiol. 72:837-845.
279
PUBLICATIONS
(M.Karamanoglu)
Full Journal Papers
1. Karamanoglu M, O'Rourke MF, Avolio AP and Kelly RP (1992). An analysis of the relationship between central aortic and peripheral upper limb pressure waves in man. European Heart J (in press)
2. Kelly RP, Karamanoglu M, Gibbs HH, Avolio AP and O'Rourke MF (1990). Non-invasive blood pressure registration as an indicator of ascending aortic pressure. J Vasc Med Biol 1:241-247
280
Letters
1. O'Rourke MF, Avolio AP, Karamanoglu M, Gallagher D and Schyvens C (1990). Brachioradial delay. Lancet, 336:1377-1379
Papers In Conference Proceedings
1. Karamanoglu M, Avolio AP, O'Rourke MF (1990). Simulation of the effects of lower body wave reflection on aortic and brachial pulse in man: Implications to therapy. Proc.of IEEE-EMBS 12:1154
Book Chapters
1. O'Rourke MF, Avolio AP, Kelly RP and Karamanoglu M (1992). Systolic pressure in central and peripheral arteries; Monitoring of central from the brachial or radial pulse. In "Vasodilatation in conduit arteries: Logical strategy in cardiovascular disease". O'Rourke MF, Safar M and Dzau V(eds). Edward Arnold, London.
Abstracts
1. Avolio AP, Kelly RP, Karamanoglu M, O'Rourke MF (1989). Measurement of brachial artery systolic pressure underestimates systemic effects of nitroglycerin: role of pulse wave transmission in the upper limb. J Vasc Med Biol 1:157
2. Karamanoglu M, Gravlee G, Kelly RP, Avolio AP, O'Rourke MF (1989). Pressure wave propagation in the upper limb- clinical implications. Aust NZ J Med 19:548
3. Karamanoglu M, Kelly RP, Gravlee G, Avolio AP, O'Rourke MF (1989). Clinical implications of pressure wave propagation in the upper limb. Circulation 80:Suppl 2:p542
4. Avolio AP, Karamanoglu M, O'Rourke MF (1990). Simulation of effects of vasodilators as detected from measurement of arterial pressure in the arm. Aust. NZ. J. Med Suppl I:328
5. O'Rourke MF, Avolio AP, Karamanoglu M, Kelly RP (1990). Derivation of ascending aortic pressure waveform from the brachial pressure pulse in man. Aus. NZ. J. Med 20:Suppl I:329
281
6. Karamanoglu M, Avolio AP, O'Rourke MF (1991). Real-Time estimation of aortic pressure wave contour from non-invasive measurements of the peripheral pulse in the upper limb. Aust. NZ. J. Med. 20:Suppl I:329
7. Karamanoglu M, Gallagher DE, Schyvens C, Avolio AP, O'Rourke MF (1991). A system of on-line analysis of pressure waveforms. Aust. NZ. J. Med 21:526
8. Gallagher DE, Karamanoglu M, Huang G-H, Avolio AP, O'Rourke MF (1991). Assessment of interobserver variation on peripheral pressure waveform analysis using a semi-automated system. Aust. NZ. J. Med 21:527
9. O'Rourke MF, Avolio AP, Karamanoglu M, Kelly RP (1991). Derivation of ascending aortic pressure waveform from the brachial pressure pulse in man. Medical & Biol Eng & Comp 29:Suppl I,157.
10. O'Rourke MF, Avolio AP, Karamanoglu M (1992). Use of transfer functions to determine central from peripheral pulse. J Am. Coll. Cardiol 19:227A.
11. O'Rourke MF, Karamanoglu M, Avolio AP (1992). A modelling analysis of aging and vasoactive drugs on the human arterial pulse. J. Am. Coll. Cardiol. 19:74A
12. Gallagher DE, O'Rourke MF, Avolio AP, Karamanoglu M, Kelly RP (1992). Determination of left ventricular ejection time from the radial pulse. J Am. Coll. Cardiol 19:74A.
13. Gallagher DE, Karamanoglu M, Avolio AP, O'Rourke MF (1992). Functional origin of wave reflection in the human arterial system. Aust. NZ. J. Med. (in press)
14. Kuchar DL, Platt HL, Karamanoglu M, Gallagher DE, Thorburn CW (1992). Effect of vagal blockade on baroreceptor responsiveness assessed using the Valsalva manoeuvre. Aust NZ. J. Med (in press).
15. Kuchar DL, Platt HL, Karamanoglu M, Thorburn CW (1992). Power spectrum analysis of the electrocardiogram does not predict the occurrence of syncope during tilt testing. Aust NZ. J. Med (in press)
282
16. O'Rourke MF, Karamanoglu M, Lee L, Cutfield G, Avolio AP, Gallagher DE, (1992). Derivation of central aortic from radial artery systolic pressure using a non-occlusive cuff to attenuate wave reflection in the hand. Aust NZ. J. Med (in press)
17. Gallagher DE, Karamanoglu M, Herok G, Avolio AP, Baird DD, O'Rourke MF (1992). Peripheral pressure measurements may be unreliable for calculation of cardiac properties with change in heart rate. Aust NZ. J. Med (in press).
18. Gallagher DE, O'Rourke MF, Avolio AP, Karamanoglu M, Kelly RP (1992). Left ventricular ejection time can be determined from radial pressure pulse. Aust NZ. J. Med (in press).
19. Gallagher DE, Karamanoglu M, Avolio AP, O'Rourke MF (1992). Accuracy of a finger clamp device (Finapres) for assessment of arterial pulse wave contour and of wave velocity. Aust NZ. J. Med (in press).
20. Gallagher DE, Karamanoglu M, Young JA, Kelly RP, Avolio AP and O'Rourke MF (1992). Apparent alteration in the upper limb pressure wave transmission with changing heart rate. Circulation 86:1830.
283
APPENDIX I
284
DIAGNOSTIC APPLANATIONTONOMETRY
For IBM Compatible Computers
Version 1.0
$GGUHVV
)RU&RUUHVSRQGDQFH
0XVWDID
.DUDP
DQRJOX
3RVW0'RFWRUDO5
HVHDUFK)HOORZ
&DUGLRORJ\
'HSDUWP
HQW
6W9LQFHQ
W*V+RVSLWDO
16:5343
6\GQ
H\$XVWUDOLD
HPDLO=0
1.DUDP
DQRJOX
#XQVZ1HGX
1DX
&RS\ULJK
W4<<3/4<<5
E\0XVWDID
.DUDP
DQRJOX
/6\GQ
H\$XVWUDOLD1
$OOULJK
WVUHVHUYHG1
,%0
DQG
,%0
3&2$
7/+HUFX
OHV/060'
26/'DWD
7UDQVODWLRQ/
0LOODU
7RQRP
HWHU/3RVWVFULSW/
+3/DVHU-HW
DQG
3&/DUH
WKHWUDGHP
DUNVRU
UHJLVWHUHGWUDGHP
DUNVRIUHVSHFWLYH
KROGHUV1
i
TABLE OF CONTENTS
INTRODUCTION 1
GETTING STARTED 3Hardware Requirements 3Software Requirements 3Installing the Data Acquisition Boards 3Installing the Software 4Executing the Program 5
OVERVIEW 7What is DAT ? 7Theory of Operation 7Features of DAT 7
PATIENT INFORMATION INPUT FORM 9Introduction 9Information Fields and Their Usage 10Keys and Functions 11
INPUT OF THE PULSE 13Acquisition of the Pulse 13Calibration of the Pressure Pulses 13Sound Generation 13Triggering 14Selecting the Pulse 14Feature Extraction 14Keys and Functions 15
REPORT GENERATION 17Introduction 17Viewing of Patient Recordings 17Generating Reports of Recorded Pulses 17Printing a Hard Copy 19Explanation of Report 19Keys and Functions 21
ii
ADVANCED TOPICS 23Configuration File 23Command Line Interface 25
TECHNICAL CONCEPTS 29Overview 29Data Acquisition 30Signal Conditioning 31Feature Extraction 32Calculated Values 33Database Engine 34
HOW TO ? 37Import Existing Tape Data to DAT 37Conduct Different Studies Simultaneously 37Combine Different Databases 37Perform Statistical Analysis of the Data 38Print Unattended 39Use Reports in Word Processors 39Record Waveforms Without Actually Seeing Them 40Measure Pulse wave Velocity 40
WHAT TO DO IF ? 41Cannot Run DAT Program 41DATHELP.MSG is Missing 41Cannot Register the Pulse. The System Crashes 41Cannot See Any Data on the Screen 42The System Crashes 42Cannot Find a Recording Registered Previously 42
iii DIAGNOSTIC APPLANATION TONOMETRY INTRODUCTION •••• 1
INTRODUCTION
$SSODQDWLRQ WRQRPHWU\ TXDQWLILHV WKH DJH ROG SUDFWLFH RI SDOSDWLQJWKH SHULSKHUDO SXOVH/ DQG UHSUHVHQWV D VXEVWDQWLDO WHFKQLFDOLPSURYHPHQW RYHU WKH PHFKDQLFDO VSK\JPRJUDSKV +0DUH\/0DKRPHG/ 'XGJHRQ/ DQG 0DFNHQ]LH0/HZLV, XVHG VRPH KXQGUHG\HDUV DJR1 3UHYLRXVO\/ LQIRUPDWLRQ LQ WKH FRQWRXU RI WKH DUWHULDOSUHVVXUH ZDYH UHPDLQHG KLGGHQ +D, EHFDXVH RI WKH DUWHIDFWV LQUHFRUGHG WUDFLQJV +E, EHFDXVH RI LQDELOLW\ WR GLJLWL]H DQG SURFHVVWKH ZDYHV DQG +F, WKURXJK ODFN RI WKHRUHWLFDO EDVLV WR DQDO\]H WKHZDYHV1
7\SH $ 7\SH % RU &
7KH W\SLFDO ZDYHIRUPV LQ FHQWUDO DRUWD1 7\SH $ SXOVHV KDYH DGLVWLQFW VKRXOGHU LQ ODWH V\VWROH ZKLFK DXJPHQWV WKH V\VWROLFSUHVVXUH DQG VHHQ PRVWO\ LQ HOGHUO\ SDWLHQWV1 7\SH % RU & SXOVHVDUH VHHQ LQ \RXQJ SDWLHQWV ZKHUH QR DXJPHQWDWLRQ LQ WKH SXOVH FDQEH REVHUYHG1 $OWKRXJK WKHVH GHVFULSWLRQV DUH FDWHJRULFDO WKHUH LV DFRQWLQXRV VSHFWUXP RI SXOVHV FDQ EH REVHUYHG GXH WR GLIIHUHQWGHJUHHV RI UHIOHFWHG ZDYHV SUHVHQW LQ WKH FRPSRVLWH SXOVH1$XJPHQWDWLRQ LQGH[ GHVFULEHV WKH DPRXQW RI UHIOHFWHG ZDYH LQ WKHSUHVVXUH SXOVH +DGRSWHG IURP 0XUJR HW DO +4<;3,1 &LUFXODWLRQ95=4380449,1
7KH XVH RI DSSODQDWLRQ WRQRPHWU\ LQ DVVHVVLQJ FKDQJHV LQ WKHSHULSKHUDO SXOVH FRQWRXU ZLWK DJHLQJ +VHH ILJXUH, DQG LQ UHVSRQVHWR GLIIHUHQW WKHUDSHXWLF DJHQWV KDV EHHQ YDOLGDWHG SUHYLRXVO\1 ,QDGGLWLRQ/ JHQHUDOL]HG DOJRULWKPV KDYH EHHQ GHYHORSHG DQG
DIAGNOSTIC APPLANATION TONOMETRY INTRODUCTION •••• 2
YDOLGDWHG WR UHODWH FKDQJHV LQ SHULSKHUDO SUHVVXUH FRQWRXU WRFKDQJHV LQ FHQWUDO SUHVVXUH FRQWRXU1
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
$IWHU V\QWKHVLVLQJ WKH DVFHQGLQJ DRUWLF SXOVH/ UHOHYDQWSK\VLRORJLFDO SDUDPHWHUV DUH GHULYHG DXWRPDWLFDOO\1 7KLVLQIRUPDWLRQ FDQ DVVLVW WKH FOLQLFLDQ WR HYDOXDWH WKH FRXSOLQJEHWZHHQ WKH V\VWHPLF FLUFXODWLRQ DQG OHIW YHQWULFOH/ DQG FDQHQKDQFH VXEVWDQWLDOO\ WKH UHODWLYHO\ OLPLWHG LQIRUPDWLRQ DYDLODEOHIURP WKH XVH RI VSK\JPRPDQRPHWHU1
2WKHU IHDWXUHV RI WKH V\VWHP DUH=
• 2Q0OLQH DQDO\VLV
• 'DWDEDVH IRU SDWLHQW ILOHV
• 5HSRUW IXQFWLRQ
7KLV PDQXDO DVVXPHV \RX KDYH EDVLF NQRZOHGJH RQ RSHUDWLRQV RI\RXU FRPSXWHU V\VWHP1 7KH PDQXDO LV RUJDQLVHG DV IROORZV=
• &KDSWHU 4 ,QVWDOODWLRQ RI WKH '$7 VRIWZDUH
• &KDSWHU 5 $Q RYHUYLHZ RI WKH '$7 6\VWHP
• &KDSWHU 6 3DWLHQW LQIRUPDWLRQ LQSXW IRUP
• &KDSWHU 7 $FTXLVLWLRQ RI WKH DUWHULDO SXOVHV
• &KDSWHU 8 *HQHUDWLRQ RI D UHSRUW
• &KDSWHU 9 $GYDQFHG XVHUV
• &KDSWHU : 7HFKQLFDO JXLGH WR RSHUDWLRQV
• &KDSWHU ; $QVZHUV WR FRPPRQ TXHVWLRQV DVNHG DERXW KRZ WRJHW PRVW IURP WKH '$7
• &KDSWHU < *XLGH WR FRPPRQ FDXVHV RI WURXEOH DQG KRZ WR IL[WKHP
,I \RX KDYH DQ\ TXHULHV DERXW WKH LQQHU ZRUNLQJV RI WKH VRIWZDUHRU LI \RX QHHG IXUWKHU LQIRUPDWLRQ DERXW WKH V\VWHP/ SOHDVH IHHOIUHH WR FRUUHVSRQG WR WKH DGGUHVV JLYHQ LQ WKH IURQW SDJH
DIAGNOSTIC APPLANATION TONOMETRY GETTING STARTED •••• 3
GETTING STARTED
%HIRUH XVLQJ WKH '$7 VRIWZDUH/ PDNH VXUH \RX KDYH WKHIROORZLQJ KDUGZDUH DQG VRIWZDUH1
Hardware Requirements• $Q ,%00$7 SHUVRQDO FRPSXWHU RU FRPSDWLEOH ZLWK PDWK0
FR0SURFHVVRU
• $ PRQRFKURPH RU FRORXU PRQLWRU ZLWK +HUFXOHV RU(*$29*$ JUDSKLF FDUG
• $W OHDVW 845 .LORE\WHV RI IUHH PHPRU\
• $ KDUG GLVN RI D FDSDFLW\ RI 53 0% RU PRUH
• $ 0LOODU 7RQRPHWHU 6\VWHP RU RWKHU DFFXUDWH GHYLFH IRUSHULSKHUDO SXOVH UHJLVWUDWLRQ ZKLFK RXWSXWV 49ROWV2433PP+J
• $ 45 ELW $2' V\VWHP ZLWK RSWLRQDO '2$ FKDQQHOV ZLWK DWOHDVW 833 +] DFTXLVLWLRQ UDWH
• $ GRW PDWUL[ RU ODVHU SULQWHU UHFRJQLVLQJ 3RVW6FULSW RU 3&/ODQJXDJHV
Software Requirements7KH '$7 VRIWZDUH UXQV RQ WKH 9HUVLRQ 613 RU ODWHU YHUVLRQV RIWKH 3&0'26 RU 060'26 RSHUDWLQJ V\VWHP1
Installing the Data Acquisition Boards6XSSRUWHG GDWD DFTXLVLWLRQ KDUGZDUH LV 'DWD 7UDQVODWLRQ+'DWD 7UDQVODWLRQ/ ,QF1 0DVV1, ERDUGV=
'75;34/ '75;340$/ '75;3428:49/ '75;38/ '75;3828:49/'75;3;/ '75;4;
CHAPTER 1
DIAGNOSTIC APPLANATION TONOMETRY GETTING STARTED •••• 4
,QVWDOO WKH DFTXLVLWLRQ ERDUG DV VXJJHVWHG E\ WKHPDQXIDFWXUHU DQG VHW WKH EDVH DGGUHVV RI GDWD DFTXLVLWLRQV\VWHP WR 5(& ++H[DGHFLPDO,1
$WWDFK SUHVVXUH ZDYHIRUP LQSXW FDEOH WR FKDQQHO 3 +]HUR, RI$2' ERDUG1 2SWLQDOO\ \RX PD\ DWWDFK FKDQQHO 4 RI WKH $2'ERDUG WR /HDG ,, RI DQ (&* RU D UHIHUHQFH WULJJHU VRXUFH WRWLPH WKH RQVHW RI WKH SUHVVXUH ZDYH1
Installing the Software7KLV VHFWLRQ GHVFULEHV KRZ WR LQVWDOO WKH '$7 V\VWHP RQWR\RXU KDUG GLVN1 <RX RQO\ QHHG WR LQVWDOO WKH V\VWHP RQFH1
%HIRUH XVLQJ '$7 IRU WKH ILUVW WLPH/ SOHDVH IROORZ WKHVH VWHSV=
• 6WDUW WKH FRPSXWHU V\VWHP
:KHQ '26 GLVSOD\V LWV SURPSW/ SODFH WKH '$7 GLVNHWWH LQWRGULYH $1 ,I WKH SURPSW LV QRW $= W\SH $= ?(QWHU!1 7KH V\VWHPZLOO WKHQ GLVSOD\ WKH $= SURPSW1
• 7\SH '$7 , DQG SUHVV ?(QWHU!1
7KH '$7 GLVSOD\V WKH IROORZLQJ SURPSW=
Drive: [C]
• 7\SH WKH OHWWHU LGHQWLI\LQJ WKH KDUG GLVN ZKHUH \RX ZDQW WRLQVWDOO WKH '$7 V\VWHP DQG SUHVV ?(QWHU!1 7R DFFHSW WKHFXUUHQW YDOXH 'ULYH &/ VLPSO\ SUHVV ?(QWHU!1 7KH '$7 ZLOOGLVSOD\ WKH IROORZLQJ SURPSW=
Directory: [\DAT]
• 7\SH WKH QDPH RI WKH GLUHFWRU\ ZKHUH \RX ZDQW WKH '$7 WRSODFH WKH '$7 VRIWZDUH1 ,I WKH GLUHFWRU\ GRHV QRW H[LVW/ '$7ZLOO FUHDWH LW IRU \RX1 7R LQVWDOO WKH '$7 V\VWHP LQ WKH URRWGLUHFWRU\/ VLPSO\ SUHVV ?(QWHU!1
'$7 WKHQ DVNV WKH KDUGZDUH LQIRUPDWLRQ E\ LVVXLQJ WKHIROORZLQJ SURPSW=
Printer types:
1=9 pin dot matrix
2=24 pin dot matrix
3=HP LaserJet
4=Postscript
Enter Number:
• 7\SH WKH UHOHYDQW QXPEHU DQG SUHVV ?(QWHU! WR LQVWDOO WKHDSSURSULDWH SULQWHU GULYHU1
Printer Port:
1=LPT1
2=LPT2
DIAGNOSTIC APPLANATION TONOMETRY GETTING STARTED •••• 5
3=COM1
4=COM2
5=FILE
Enter Number:
• 7\SH WKH UHOHYDQW QXPEHU DQG SUHVV ?(QWHU! WR LQVWDOO WKHDSSURSULDWH SULQWHU SRUW1
A/D Board type:
1=DT280X series
2=Other
3=Nil
Enter Number:
• 7\SH WKH UHOHYDQW QXPEHU DQG SUHVV ?(QWHU! WR LQVWDOO WKHDSSURSULDWH VLJQDO DFTXLVLWLRQ GULYHU1
)LQDOO\ '$7 DVNV IRU FRQILUPDWLRQ
CONFIRM. Do you wish to install DAT? (Y/N) [N]
• 7\SH < DQG SUHVV ?(QWHU! WR LQVWDOO WKH V\VWHP1 '$7 ZLOOLQIRUP \RX ZKHQ WKH LQVWDOODWLRQ LV FRPSOHWH DQG ZLOOUHWXUQ \RX WR WKH '26 SURPSW1
3OHDVH QRWH=
41 $V ZLWK DOO DSSOLFDWLRQV UXQQLQJ XQGHU '26/ '$7*VSHUIRUPDQFH FDQ EH HQKDQFHG LI \RX VSHFLI\%8))(56 53 DQG ),/(6 53 LQ WKH &21),*16<6FRQILJXUDWLRQ ILOH1 5HIHU WR \RXU 'LVN 2SHUDWLQJ 6\VWHPPDQXDO IRU LQIRUPDWLRQ RQ WKH %8))(56 DQG WKH ),/(6FRPPDQGV1
51 ,I \RX PDNH D PLVWDNH ZKLOH DQVZHULQJ WKH '$7 SURPSWV/SUHVV ?&RQWURO0&!1 <RX ZLOO WKHQ UHFHLYH WKH '26 SURPSWZKLFK ZLOO DOORZ \RX WR UH0VWDUW WKH LQVWDOODWLRQ SURFHVVDJDLQ1
61 $V GLIIHUHQW GULYHUV IRU GLIIHUHQW KDUGZDUH EHFRPHDYDLODEOH WKHVH ZRXOG EH LQFOXGHG LQ IXWXUH XSJUDGHV RIWKH VRIWZDUH1
Executing the Program7R XVH WKH '$7 SURJUDP/ PDNH VXUH \RX DUH LQ WKH GLUHFWRU\ZKHUH WKH '$7 ILOHV ZHUH LQVWDOOHG/ WKHQ W\SH
•••• '$7 DQG SUHVV ?(QWHU!1
DIAGNOSTIC APPLANATION TONOMETRY GETTING STARTED •••• 6 DIAGNOSTIC APPLANATION TONOMETRY OVERVIEW •••• 7
OVERVIEW
What is DAT ?'$7 LV D QHZO\ GHYHORSHG FRPSXWHULVHG GLDJQRVWLF WRRO IRU WKHFOLQLFDO DVVHVVPHQW RI SXOVH LQGLFHV1 7KLV V\VWHP LV GHVLJQHGWR GHWHUPLQH FHQWUDO KDHPRG\QDPLF LQGLFHV IURP SHULSKHUDOSUHVVXUH ZDYH PHDVXUHPHQWV XVLQJ DSSODQDWLRQ WRQRPHWU\17KH V\VWHP FDQ DOVR EH XVHG ZLWK LQYDVLYHO\ GHWHUPLQHGSUHVVXUH ZDYHV IURP WKH UDGLDO RU EUDFKLDO DUWHU\1
Theory of Operation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
Features of DAT• $ SUHVVXUH ZDYH DFTXLVLWLRQ V\VWHP +FXUUHQWO\ 0LOODU
WRQRPHWHU V\VWHP,1
• $Q RQ0OLQH DQDO\VLV IHDWXUH1
• $ FRPSUHKHQVLYH UHSRUW V\VWHP1
• $ GDWDEDVH V\VWHP IRU DFFHVVLQJ SDWLHQW LQIRUPDWLRQ DQGWKHLU UHFRUGV1
• $ FRQWH[WXDO RQ0OLQH KHOS V\VWHP1
• $ FRPPDQG OLQH LQWHUIDFH WR H[HFXWH FRPPDQGV DW WKH'26 SURPSW1
CHAPTER 2
DIAGNOSTIC APPLANATION TONOMETRY OVERVIEW •••• 8 DIAGNOSTIC APPLANATION TONOMETRY PATIENT INFORMATION INPUT FORM •••• 9
PATIENT INFORMATION INPUT FORM
Introduction3DWLHQW ,QIRUPDWLRQ ,QSXW )RUP LV WKH EDVLV RI WKH '$7V\VWHP1 7KH 3& RSHUDWHV ERWK DV D ILOH VHUYHU DQG D GDWDDQDO\VLV V\VWHP/ KHQFH WKH ILUVW VWDJH RI RSHUDWLRQ LV WKHFUHDWLRQ RI D GDWD ILOH IRU D SDWLHQW1 7KLV UHTXLUHV WKH HQWU\ RILQIRUPDWLRQ RQWR WKH VFUHHQ ZKHQ SURPSWHG +)LJXUH 614,1 7KHRYHUDOO UHODWLRQ EHWZHHQ WKH PHQX V\VWHPV DUH VKRZQ LQ)LJXUH 6151
PATIENT INFORMATION INPUT FORM
Name : JOHN CITIZEN
Sex : M
Age : 52
ID : 00000000
Adress : 111 Johnston Ave St Pauls NSW 2031
Measurement: Rad Car Fem ECG Bra Axl S.C. D.P A.A
Distances : 0 10 0 0 0 0 0 0
Sp (mmHg) : 0
Dp (mmHg) : 0
Medication : Nil
Notes : Normal
Operator ID: 0
F1-Help F2-SaveToFile PgDn-InputPulse Alt-D-Delete End-Report Esc-Quit
:
)LJXUH 6141 'DWD HQWU\ IRUP IRU LQSXW RI SDWLHQW GDWD DQG YLVLWLQIRUPDWLRQ1
CHAPTER 3
DIAGNOSTIC APPLANATION TONOMETRY PATIENT INFORMATION INPUT FORM •••• 10
Patient Information Patient Recording Details
<PgDn> <Esc, F2>
Input Pulses Menu
<End> <Esc>
Report Menu
PATIENT INFORMATION INPUT FORM
)LJXUH 615 'LDJUDP VKRZLQJ WKH LQWHUSOD\ RI PHQX V\VWHPVZLWK UHVSHFW WR 3DWLHQW ,QIRUPDWLRQ ,QSXW )RUP1
Information Fields and Their Usage
NAME7KH QDPH RI WKH SDWLHQW1 1RWH WKDW DOO WKH SDWLHQWV DUH LQGH[HGDFFRUGLQJ WR WKLV HQWU\1 7KHUHIRUH LW LV D JRRG SUDFWLFH WR XVHHLWKHU D ILUVW QDPH0VXUQDPH FRPELQDWLRQ RU D VXUQDPH0ILUVWQDPH FRPELQDWLRQ FRQVLVWHQWO\ LQ DOO WKH HQWULHV1 2PLW DOOSXQFWXDWLRQ GXULQJ WKH NH\ HQWULHV1 ,I WKH SDWLHQW LQIRUPDWLRQKDV DOUHDG\ EHHQ UHFRUGHG LQ WKH V\VWHP/ WKHQ E\ SUHVVLQJ WKH?5HWXUQ! NH\ DIWHU HQWHULQJ WKH SDWLHQW*V QDPH/ WKHUHPDLQGHU RI WKH SDWLHQW*V LGHQWLI\LQJ GDWD ZLOO EH GLVSOD\HG1
SEX(LWKHU 0 IRU PDOH RU ) IRU IHPDOH1
AGE7R WKH FORVHVW ELUWKGD\/ HQWHU WKH DJH LQ QXPEHU RI \HDUV1
ID7KH KRVSLWDO LGHQWLILFDWLRQ QXPEHU IRU WKDW SDWLHQW RU DQDUELWUDU\ QXPEHU JLYHQ E\ WKH V\VWHP XVHU1
ADDRESS7KH KRPH DGGUHVV RI WKH SDWLHQW1 (QWHU DV 6WUHHW1XPEHU26WUHHW 1DPH26XEXUE23RVWFRGH1
7KH IROORZLQJ SDUWLFXODUV IRU WKH FXUUHQW H[DPLQDWLRQ PXVW EHHQWHUHG=
DIAGNOSTIC APPLANATION TONOMETRY PATIENT INFORMATION INPUT FORM •••• 11
SITE, DISTANCES and ORDER,I UHTXLUHG IRU FDOFXODWLRQ RI SXOVH ZDYH YHORFLW\ +VHH SDJH 67,/WKH FKRVHQ VLWH LV QRPLQDWHG E\ SUHVVLQJ WKH ?(QWHU! NH\ZKHQ WKH FXUVRU LV VXSHULPSRVHG RYHU WKH QRPLQDWHG VLWHZKLFK ZLOO WKHQ EH KLJKOLJKWHG1 'LVWDQFHV VKRXOG EH IURPVXSUDVWHUQDO QRWFK WR WKH PHDVXUHPHQW VLWH LQ FHQWLPHWUHV1
0RUH WKDQ RQH SHULSKHUDO VLWH PD\ EH QRPLQDWHG IRU D JLYHQH[DPLQDWLRQ ZLWK WKH VHTXHQFH RI H[DPLQDWLRQ FKRVHQSUHYLRXVO\1 $OWHUQDWLYHO\/ QH[W H[DPLQDWLRQ VLWH PD\ EHHQWHUHG LQGLYLGXDOO\ DIWHU HDFK H[DPLQDWLRQ1 6HOHFWLRQ RI D VLWHPD\ EH FDQFHOOHG E\ SUHVVLQJ WKH ?(QWHU! NH\ ZLWK WKH FXUVRURYHUO\LQJ WKH VLWH QDPH1 7KH RSHUDWLRQ LV FRQILUPHG E\ WKHGLVDSSHDUDQFH RI KLJKOLJKWLQJ RYHU WKH VLWH QDPH1
SYSTOLIC and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
MEDICATION7ZR OLQHV DUH DYDLODEOH IRU WKLV HQWU\/ DQG HQWHUHG GDWD ZLOODSSHDU RQ WKH UHSRUW IRUP1
CLINICAL NOTES7KLV LQIRUPDWLRQ ZLOO QRW DSSHDU RQ WKH UHSRUW/ WKHUHIRUH LW LVRQO\ IRU UHIHUHQFH1
OPERATOR ID7KLV YDOXH LV IRU WKH LGHQWLILFDWLRQ RI RSHUDWRUV1 $ YDOXHEHWZHHQ 40<<1 7KH ]HUR YDOXH LV DVVLJQHG IRU XQLGHQWLILHGRSHUDWRUV1
Keys and Functions
Editing Keys7KH IROORZLQJ NH\V SHUPLW WKH HGLWLQJ RI GDWD=
• ?⇐! 0RYHV FXUVRU OHIW
• ?⇒! 0RYHV FXUVRU ULJKW
• ?⇑! 0RYHV FXUVRU XS
DIAGNOSTIC APPLANATION TONOMETRY PATIENT INFORMATION INPUT FORM •••• 12
• ?⇓! 0RYHV FXUVRU GRZQ• ?¤!'HOHWHV WKH FKDUDFWHU RQ WKH OHIW VLGH RI WKH FXUVRU•••• ?'HO! 'HOHWHV WKH FKDUDFWHU XQGHU WKH FXUVRU•••• ?,QV! 7RJJOHV LQVHUW PRGH1 ,I LW LV RQ/ LW DOORZV
FKDUDFWHUV WR EH LQVHUWHG DW WKH FXUVRU SRLQW1•••• ?7DE! 0RYHV ; FKDUDFWHUV DW D WLPH•••• ?5HWXUQ! $FFHSWV DQ HQWU\•••• ?(QWHU! $FFHSWV DQ HQWU\
Command Keys
?)4! +(/3 IXQFWLRQ NH\1
7KLV NH\ FDQ EH XVHG DW DQ\ GDWD HQWU\ SRLQW ZKHQ WKH XVHU LVXQFOHDU DV WR ZKDW QHHGV WR EH HQWHUHG1
?)5! 6$9( 72 ),/( IXQFWLRQ NH\1
)ROORZLQJ WKH HQWU\ DQG2RU HGLWLQJ RI SDWLHQW GDWD/ WKLV NH\PXVW EH SUHVVHG LQ RUGHU WR VDYH WR D SDWLHQW ILOH LQ WKHFRPSXWHU GLVN1 2WKHUZLVH DQ\ DOWHUDWLRQ RU PRGLILFDWLRQV ZLOOEH SHUPDQHQWO\ ORVW1
?3J'Q! ,1387 38/6( IXQFWLRQ NH\1
7KLV NH\ LQLWLDWHV UHFRUGLQJ DQG GLDJQRVLV SURFHGXUH RI WKHVHOHFWHG SXOVHV1 )RU PRUH LQIRUPDWLRQ RI WKLV RSWLRQ VHH&KDSWHU 71
?(QG! 5(3257 IXQFWLRQ NH\1
7KLV NH\ SHUPLWV WKH FRPPHQFHPHQW RI WKH UHSRUWLQJSURFHGXUH RI WKH SDWLHQW1 7KH UHSRUW RI WKH SDWLHQW ZLOO EHJLQIURP WKH PRVW UHFHQW UHFRUGLQJ1 8SRQ UHWXUQ IURP WKLV RSWLRQWKH LQIRUPDWLRQ DERXW WKH ODVW YLHZHG UHFRUGLQJ LV DYDLODEOHIRU YLHZLQJ DQG2RU HGLWLQJ1 )RU PRUH LQIRUPDWLRQ RI WKLVRSWLRQ VHH &KDSWHU 81
?$OW0'! '(/(7( IXQFWLRQ NH\1
7KLV NH\ LV IRU GHOHWLQJ D SDWLHQW LQIRUPDWLRQ LQFOXGLQJ WKHUHFRUGLQJV IURP WKH ILOH V\VWHP1 ,W LV XVHIXO ZKHQ HUURQHRXVSDWLHQW LQIRUPDWLRQ KDV EHHQ W\SHG DQG VDYHG1
?$OW0(! (',7 IXQFWLRQ NH\1
7KLV NH\ LV IRU HGLWLQJ D SDWLHQW QDPH LQFOXGLQJ WKH UHFRUGLQJVIURP WKH ILOH V\VWHP1 ,W LV XVHIXO ZKHQ HUURQHRXV SDWLHQWLQIRUPDWLRQ KDV EHHQ W\SHG DQG VDYHG1
?(VF! 48,7 IXQFWLRQ NH\1
7KLV NH\ LV XVHG WR SHUPLW WKH RSHUDWRU WR TXLW WKH SURJUDPDQG UHWXUQ WR WKH RSHUDWLQJ V\VWHP1
DIAGNOSTIC APPLANATION TONOMETRY INPUT OF THE PULSE •••• 13
INPUT OF THE PULSE
Acquisition of the Pulse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
Calibration of the Pressure Pulses%HIRUH DSSO\LQJ WKH WRQRPHWHU/ HVWDEOLVK ]HUR DQG 433 PP+JFDOLEUDWLRQ WKURXJK LQWHUQDO HOHFWULFDO VHWWLQJV RI WKHWRQRPHWHU ER[ E\ SUHVVLQJ ?$OW06! NH\1 )LUVW/ LW ZLOO TXHU\ WKH3 PP+J SUHVVXUH VLJQDO DQG WKHQ WKH 433 PP+J SUHVVXUHVLJQDO1 7KH V\VWHP ZLOO DYHUDJH WKH FRUUHVSRQGLQJ YDOXHV RISUHVXUH IRU 425 VHFRQGV WR HOLPLQDWH QRLVH1 %\ XVLQJ WKHVHWZR YDOXHV WKH JDLQ DQG RIIVHW RI WKH WKH V\VWHP ZLOO EHFDOFXODWHG DQG ODWHU ZLOO EH XVHG WR VFDOH WKH SUHVVXUHZDYHIRUPV1
Sound Generation,Q FDVH RI GLIILFXOW\ LQ REVHUYLQJ WKH VFUHHQ GXULQJ VLJQDOUHJLVWUDWLRQ/ D VRXQG JHQHUDWRU LV 3&*V VSHDNHU WR IUHTXHQF\PRGXODWH WKH DFTXLUHG VLJQDO1 7KXV LW SURYLGHV D IHHGEDFN WRWKH RSHUDWRU RQ WKH DPSOLWXGH DQG IUHTXHQF\ RI WKH SUHVVXUHZDYH1 7KLV JHQHUDWRU LV WRJJOHG E\ ?$OW0'! NH\1 7KH VRXQGVLPXODWHV WKRVH JHQHUDWHG IURP D GRSSOHU IORZ PHWHU/ ZLWKORZ0WR0KLJK SLWFK VRXQG FRUUHVSRQGLQJ WR SUHVVXUH VLJQDOJRLQJ IURP D ORZ WR D KLJKHU VLJQDO OHYHO1
CHAPTER 4
DIAGNOSTIC APPLANATION TONOMETRY INPUT OF THE PULSE •••• 14
Triggering7KH SXOVH WUDLQ LV WULJJHUHG E\ LWVHOI GXULQJ VLJQDO DQDO\VLV1 ,I/KRZHYHU/ DQ (&* ZDYHIRUP KDV DOVR EHHQ VHOHFWHG LQ WKH3DWLHQW ,QIRUPDWLRQ ,QSXW )RUP DV D WULJJHULQJ VRXUFH IRUVLJQDO DQDO\VLV/ WKH ILUVW GHULYDWLYH RI WKH (&* ZDYHIRUP +RURWKHU VLJQDO VXFK DV DQRWKHU SUHVVXUH ZDYH, ZLOO DOVR EHGLVSOD\HG RQ WKH XSSHU KDOI RI WKH VFUHHQ1 7KH GDWD UHFRUGLQJV\VWHP H[SHFWV (&* VLJQDO LQ WKH VHFRQG FKDQQHO IROORZLQJWKH SUHVVXUH VLJQDO RI WKH $2' FRQYHUWHU +E\ GHIDXOW FKDQQHO4,1 %\ SUHVVLQJ ?.20! NH\V D XVHU FDQ DGMXVW WKH DPSOLWXGH RIWKH (&* VLJQDO/ HOLPLQDWLQJ RYHUULGLQJ VLJQDOV RQ WKH VFUHHQ1
Selecting the Pulse7KH SRVLWLRQ DQG SUHVVXUH RQ WKH WRQRPHWHU SUREH VKRXOG EHDGMXVWHG XQWLO WKH ZDYHIRUPV DUH UHSURGXFLEOH IURP EHDW WREHDW DQG RI JRRG FRQWRXU DQG DPSOLWXGH +XVXDOO\ LQ WKH UDQJHRI WKH 3DWLHQW ,QIRUPDWLRQ ,QSXW )RUP V\VWROLF DQG GLDVWROLFSUHVVXUH YDOXHV,/ DQG ZLWK D VWDEOH EDVHOLQH1 7KH SULQFLSOH RIDSSODQDWLRQ WRQRPHWU\ UHTXLUHV WKDW WKH IURQW ZDOO RI WKHDUWHU\ EH IODWWDQHG E\ WKH LQVWUXPHQW1
Feature Extraction:KHQ DSSUR[LPDWHO\ WZR DQG D KDOI VFUHHQV +43 VHFRQGV, RIFRQWLQXRXV ZDYHV IXOILOOLQJ WKH DERYH FULWHULD KDYH EHHQREWDLQHG/ WKH RSHUDWRU VKRXOG SUHVV WKH ?)5! NH\1 7KH VFUHHQLV WKHQ IUR]HQ DQG WKH '$7 V\VWHP DQDO\VHV ; VHFRQGV RISUHVVXUH GDWD E\ H[FOXGLQJ WKH YHU\ ODVW 425 VFUHHQ +5VHFRQGV, RI GDWD1 7KLV H[FOXVLRQ LV QHFFHVVDU\ WR HOLPLQDWHSRVVLEOH DUWHIDFWV FDXVHG E\ GHOD\ RI D VLQJOH RSHUDWRU LQXVLQJ WKH NH\ERDUG1 $ FKDUW LOOXVWUDWLQJ WKH ZKROH SURFHVV LVJLYHQ LQ )LJXUH 7141
<Esc> <F2>
Storing of Results
Analysis of the Pulses
<Alt-S>
Calibration Sound on/off
<Alt-D>
Acquisition of Pulses
PATIENT INFORMATION INPUT FORM
)LJXUH 714 &RQQHFWLRQV EHWZHHQ WKH FRPPDQG VWUXFWXUHV RI,QSXW 3XOVHV 0HQX1
DIAGNOSTIC APPLANATION TONOMETRY INPUT OF THE PULSE •••• 15
Keys and Functions
Command Keys
?)4! +(/3 IXQFWLRQ NH\1
7KLV NH\ FDQ EH XVHG DW DQ\ GDWD HQWU\ SRLQW ZKHQ WKH XVHU LVXQFOHDU DV WR ZKDW QHHGV WR EH HQWHUHG1
?)5! ',$*126( IXQFWLRQ NH\1
7KLV NH\ LV XVHG ZKHQ VDWLVIDFWRU\ ZDYHIRUP KDV EHHQDFTXLUHG IRU DW OHDVW 43 VHFRQGV1 7KH VFUHHQ IUHH]HV DQG WKH'$7 V\VWHP DQDO\VHV DSSUR[LPDWHO\ WKH ODVW ; VHFRQGV RIGDWD1
?$OW06! &$/,%5$7( IXQFWLRQ NH\1
7KLV NH\ LV XVHG ZKHQ WKH FDOLEUDWLRQ RI SXOVHV LV UHTXLUHG1$IWHU VHWWLQJ H[WHUQDO SUHVVXUH VRXUFH FRQQHFWHG WR WKHWUDQVGXFHU WR 3 PP+J/ SUHVV ?(VF! NH\ ZKHQ LQVWUXFWHG17KHQ/ VHW WKH H[WHUQDO SUHVVXUH WR 433 PP+J DQG SUHVV?(VF! NH\1
?$OW0'! 6281' IXQFWLRQ NH\1
7KLV NH\ LV XVHG WR WRJJOH VRXQG JHQHUDWLRQ 21 RU 2))1
?.2 0! 75,**(5 IXQFWLRQ NH\1
7KHVH NH\V DUH XVHG LQ FRQMXQFWLRQ ZLWK WKH (&* WULJJHUVRXUFH ZKHQ WKLV LV EHLQJ XVHG1 ,I \RXU (&* VLJQDO LV WRR KLJKDQG WKHUHIRUH REVWUXFWLQJ WKH YLHZLQJ RI WKH SUHVVXUH VLJQDO\RX FDQ DOWHU (&* ZDYHIRUP DPSOLWXGH E\ SUHVVLQJ WKHVHNH\V1
?(VF! 48,7 IXQFWLRQ NH\1
7KLV NH\ LV XVHG WR SHUPLW WKH RSHUDWRU WR TXLW WKH ,QSXW3XOVHV 0HQX DQG UHWXUQ SDWLHQW LQIRUPDWLRQ LQSXW IRUPZLWKRXW DQDO\VLQJ WKH GDWD1
DIAGNOSTIC APPLANATION TONOMETRY INPUT OF THE PULSE •••• 16 DIAGNOSTIC APPLANATION TONOMETRY REPORT GENERATION •••• 17
REPORT GENERATION
Introduction7KH UHSRUW+V, RI D SDWLHQW PD\ EH YLHZHG DW DQ\ WLPH E\SUHVVLQJ WKH ?(QG! NH\ IURP WKH SDWLHQW LQIRUPDWLRQ IRUP1
Viewing of Patient Recordings8SRQ HQWU\ WR WKH UHSRUW V\VWHP LW ZLOO GLVSOD\ WKH FXUUHQW RUPRVW UHFHQW UHFRUGLQJ1 7KH RSHUDWRU FDQ VHOHFW WKH SUHYLRXV/QH[W/ ILUVW DQG ODVW UHFRUGLQJV RI WKH VDPH SDWLHQW E\ SUHVVLQJ?3J8S!/ ?3J'Q!/ ?+RPH! RU ?(QG! NH\V UHVSHFWLYHO\1
Generating Reports of Recorded Pulses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
CHAPTER 5
DIAGNOSTIC APPLANATION TONOMETRY REPORT GENERATION •••• 18
111 Johnston Ave St Pauls Sydney
6
)LJXUH 814 $ VDPSOH RI JUDSKLF UHSRUW GLVSOD\HG RQ WKH VFUHHQ17RS SDQHO VKRZV WKH UHFRUGHG +WRS, DQG WKH DYHUDJHG +ERWWRP,SHULSKHUDO DQG FDOFXODWHG FHQWUDO SXOVHV1 $OWHUQDWH VFUHHQREWDLQHG E\ SUHVVLQJ WKH VSDFH EDU +%RWWRP SDQHO, VKRZV WKHSDWLHQW LQIRUPDWLRQ WRJHWKHU ZLWK FDOFXODWHG FHQWUDO LQGLFHV1 7KHLQVHW GLVSOD\V WKH DYHUDJHG FHQWUDO SUHVVXUH ZDYHIRUP ZLWKFRUUHVSRQGLQJ IHDWXUHV RI WKH ILUVW SHDN/ VHFRQG SHDN DQGLQFLVXUD1 1
DIAGNOSTIC APPLANATION TONOMETRY REPORT GENERATION •••• 19
Printing a Hard Copy,I D SULQWHG UHSRUW LV QHHGHG/ D KDUG FRS\ RI WKH VHOHFWHG SXOVHIURP WKH PRQLWRU +)LJXUH 815, FDQ EH REWDLQHG E\ VLPSO\SUHVVLQJ ?)5! NH\1 (QVXUH WKDW WKH SULQWHU/ LWV FRQQHFWLRQVDQG WKH JUDSK SDSHU DUH DOO LQ SODFH FRUUHFWO\ EHIRUH SUHVVLQJ?)5! NH\1
)LJXUH 815 $ VDPSOH RI KDUG FRS\ RI WKH UHSRUW1
Explanation of Report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
DIAGNOSTIC APPLANATION TONOMETRY REPORT GENERATION •••• 20
EORRG SUHVVXUH LV WKDW GHWHUPLQHG E\ VSK\JPRPDQRPHWHU DEGPHDQ LV GHWHUPLQHG IURP LQWHJUDWLRQ RI WKH ZDYH1 7KH DRUWLFSUHVVXUHV DUH GHWHUPLQHG IURP WKH SHDN/ QDGLU DQG LQWHJUDWHGPHDQ RI WKH V\QWKHVLVHG ZDYH/
7KH IODJ DW WKH ERRWRP RI WKH HQVHPEOH DYHUDJHG ZDYHUHSUHVHQWV WKH WLPLQJ RI WKH LQFLVXUD/ 7KH VDPH WLPLQJ LVDSSOLHG WR WKH V\QWKHVLVHG ZDYH1
&HQWUDO SUHVVXUH ZDYH LQGLFHV DW WKH ERWWRP RI WKH UHSRUW DUH=
Augmented Pressure3UHVVXUH GLIIHUHQFH EHWZHHQ WKH ILUVW SHDN +RU VKRXOGHU, RI WKHV\QWKHVLVHG ZDYH DQG WKH VHFRQG SHDN1 7KLV LV XVXDOO\SRVLWLYH EXW PD\ EH QHJDWLYH LQ \RXQJ DGXOWV/ ZLWK YDVRGLODWRUWKHUDS\/ ZLWK WDFKLFDUGLD RU HLWK K\SRWHQVLRQ1 ,I HLWKHU WKHILUVW RU VHFRQG SHDN FDQQRW EH LGHQWLILHG D +"", LQGLFDWRU ZLOODSSHDU1
Systolic (Tension) Pressure Time Index7KLV LV WKH LQHJUDO RI SUHVVXUH [ WLPH WKURXJKRXW V\VWROH DQGH[SUHVVHG RYHU D PLQXWH +VHH 6DUQRII 6M HW DO +4<8;, $P -3K\VLRO 4<5=47;0489,
Diastolic Pressure Time index7KLV LV WKH LQHJUDO RI SUHVVXUH [ WLPH WKURXJKRXW GLDVWROH DQGH[SUHVVHG RYHU D PLQXWH +VHH %XFNEHUJ HW DO +4<:5,1 &LUF 5HV53= 9:0;4,
Subendocardial Viability Ratio5DWLR RI GLDVWROLF SUHVVXUH WLPH LQGH[ WR V\VWROLF SUHVVXUH WLPHLQGH[1 $ YDOXH RI 433 ( LV RIWHQ DVVRFLDWHG ZLWK LPSDLUHGVXEHQGRFDUGLDO EORRG IORZ1 +VHH %XFNEHUJ HW DO +4<:5,1 - 7KRU&DUG 6XUJ 497=99<09;8,
Mean Systolic Pressure$YHUDJH SUHVVVXUH EHWZHHQ WKH ZDYH IRRW DQG LGHQWLILHGLQFLVXUD +VHH 2*5RXUNH +4<9:,1 &DUGLRYDVF 5HV 4=6460649,
Mean Diastolic Pressure$YHUDJH SUHVVVXUH EHWZHHQ WKH LQFLVXUD DQG EHJLQQLQJ RI WKHQH[W ZDYH +VHH 2*5RXUNH +4<9:,1 &DUGLRYDVF 5HV 4=6460649,
3DWLHQW LGHQWLILFDWLRQ GDWD DSSHDUV DW WKH WRS LI WKH UHSRUW DVHQWHUHG E\ WKH RSHUDWRU1 ,PPHGLDWHO\ EHORZ WKLV LV SULQWRXW
+HDUW UDWH LQ EHDWV2PLQ
(MHFWLRQ GXUDWLRQ +WLPH IURP ZDYH IRRW WR LQFLVXUD, LQPLOOLVHFRQGV
DIAGNOSTIC APPLANATION TONOMETRY REPORT GENERATION •••• 21
5HIHUHQFH DJH1 7KLV LV GHULYHG IURP FRPSDULQJ DXJPHQWDWLRQRI WKH UHFRUGHG ZDYH ZLWK DXJPHQWDWLRQ GHWHUPLQHG D JURXSRI 4338 QRUPDO VXEMHFWV +VHH .HOO\ HW DO +4<;<,1 &LUFXODWLRQ;3=49850498<,
Keys and Functions
Command Keys
?)4! +(/3 IXQFWLRQ NH\1
7KLV NH\ FDQ EH XVHG DW DQ\ WLPH ZKHQ WKH XVHU LV XQFOHDU DVWR ZKDW QHHGV WR EH GRQH1
?)5! +$5' &23< IXQFWLRQ NH\1
7KH )5 NH\ FDQ EH XVHG LI D SULQWHG FRS\ RI WKH FXUUHQWO\YLHZHG UHSRUW LV UHTXLUHG1
?3J8S! 35(9,286 IXQFWLRQ NH\1
7KLV NH\ LV XVHG WR GLVSOD\ SUHYLRXV UHSRUWV1 7KH UHSRUWV DUHILOHG LQ RUGHU RI HQWU\ ZLWK WKH PRVW UHFHQW UHSRUW EHLQJ WKHILUVW WR EH GLVSOD\HG1
?3J'Q! 1(;7 IXQFWLRQ NH\1
7KH ?3J'Q! NH\ LV GHSUHVVHG LI WKH XVHU ZLVKHV WR YLHZ PRUHUHFHQW UHSRUWV1
?+RPH!),567 IXQFWLRQ NH\1
7KLV NH\ LV UHVHUYHG IRU VHOHFWLQJ WKH ILUVW UHFRUGLQJ RI WKHSDWLHQW1
?(QG! /$67 IXQFWLRQ NH\1
7KLV NH\ LV UHVHUYHG IRU VHOHFWLQJ WKH ODVW UHFRUGLQJ RI WKHSDWLHQW1
?$OW0'! '(/(7( IXQFWLRQ NH\1
'HOHWH WKH GLVSOD\HG UHFRUG1
?6SDFHEDU!72**/( IXQFWLRQ NH\1
,W LV XVHG WR WRJJOH EHWZHHQ WKH JUDSKLF DQG WH[W FRPSRQHQWVRI WKH UHFRUGLQJ1
?(VF! 48,7 IXQFWLRQ NH\1
7KLV NH\ LV XVHG WR SHUPLW WKH RSHUDWRU WR TXLW WKH 5HSRUW0HQX DQG UHWXUQ WR WKH 3DWLHQW ,QIRUPDWLRQ ,QSXW )RUP1 7KH3DWLHQW ,QIRUPDWLRQ ,QSXW )RUP ZLOO QRZ GLVSOD\ WKHLQIRUPDWLRQ UHODWHG WR WKLV UHFRUGLQJ1
DIAGNOSTIC APPLANATION TONOMETRY REPORT GENERATION •••• 22 DIAGNOSTIC APPLANATION TONOMETRY ADVANCED TOPICS •••• 23
ADVANCED TOPICS
Configuration File'$7 OHDUQV DERXW WKH HQYLURQPHQW XQGHU ZKLFK LW LV UXQQLQJE\ H[DPLQLQJ WKH '$71&)* ILOH1 7KLV ILOH LV DQ $6&,, ILOH ZKLFKFRQWDLQV LQIRUPDWLRQ DERXW WKH $2' %RDUG/ 3ULQWHU VHWXS DVZHOO DV WKH SDWKV WR WKH GDWD ILOHV1 ,Q RUGHU WR LQIRUP WKH '$7IRU WKH SDWK WR WKH GLUHFWRU\ RI '$71&)*/ \RX FDQ HLWKHU W\SH6(7 '$73$7+ 3$7+ LQ FRPPDQG OLQH RU XVH ('/,1 RU DOLQH HGLWRU WR HGLW $872(;(&1%$7 ILOH1
first line7KLV OLQH LQ '$71&)* ILOH VSHFLILHV WKH SURJUDP WR DFTXLUH WKHGDWD1 ,W FDQ EH D EDWFK SURJUDP WR DFTXLUH DQG GLVSOD\ WKHGDWD1 %\ GHIDXOW WKLV OLQH LV VHW WR %'75;3; 3 5%/ LQGLFDWLQJ DSURJUDP ZLWK 'DWD 7UDQVODWLRQ ERDUGV XVLQJ FKDQQHO ]HUR DVWKH SUHVVXUH VRXUFH ZLWK LQWHUQDO JDLQ RI 51 )RU LQWHUQDOSURJUDPPDEOH JDLQ VHH RSHUDWLRQV PDQXDO RI WKH 'DWD7UDQVODWLRQ ERDUG1 ,W LV KRSHG LQ WKH IXWXUH GLIIHUHQW GULYHUVZKLFK ZRXOG VXSSOHPHQW WKH DFTXLVLWLRQ PRGXOH1 &XUUHQWO\'$7 FRPHV ZLWK D GULYHU LQWHUIDFH WR $2' ERDUGVPDQXIDFWXUHG E\ 'DWD 7UDQVODWLRQ &RUS1
second line7KLV OLQH LV WKH JDLQ RI WKH HQWLUH DQDORJXH FLUFXLW DW WKH IURQWHQG RI WKH $2' ERDUG1 7KLV YDOXH LV XVHIXO LI RQH XVHV GLIIHUHQWDQDORJXH HTXLSPHQWV WR FDSWXUH WKH SUHVVXUH DQG (&*ZDYHIRUPV1
third line7KLV OLQH LV WKH RIIVHW RI WKH DQDORJXH FLUFXLW DW WKH IURQW HQG RIWKH $2' ERDUG1 7KLV YDOXH LV XVHIXO LI RQH XVHV GLIIHUHQWDQDORJXH HTXLSPHQWV WR FDSWXUH WKH SUHVVXUH DQG (&*ZDYHIRUPV1
CHAPTER 6
DIAGNOSTIC APPLANATION TONOMETRY ADVANCED TOPICS •••• 24
fourth line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
&23< +($'(51(36 /374=
3676&537 !/374=
ZKHUH +($'(51(36 LV DQ HQFDSVXODWHG 3RVW6FULSW ILOHFRQWDLQLQJ WKH KHDGHU1 6LPLODUO\ WKH RXWSXW FDQ EH GLUHFWHG WRD ILOH ZKLFK FRXOG WKHQ EH LQVHUWHG WR D ZRUG SURFHVVLQJSURJUDP1
fifth line7KLV OLQH LV WKH GDWDEDVH SDWK ZKHUH WKH 3$7,(17 GDWD ILOHXVHG E\ WKH '$7 V\VWHP UHVLGHV1 7KLV IHDWXUH LV XVHIXO LI RQHZLVKHV WR VHSDUDWH GDWD IRU GLIIHUHQW WDVNV1 )RU H[DPSOH GDWDIURP WKH LQWHQVLYH FDUH XQLW PLJKW UHVLGH LQ GLUHFWRU\ ,&8ZKLOH WKH FOLQLFLDQ*V RIILFH GDWD PLJKW UHVLGH LQ 2)),&( VXEGLUHFWRU\1 7KH PDVWHU GDWDEDVH ZKLFK FRPELQHV WKH ERWKPLJKW EH LQ VXE GLUHFWRU\ 0$67(5'%1 7KLV NLQG RI VHSDUDWLRQRI GDWDEDVHV LQFUHDVHV WKH VHFXULW\ DQG JRRG PDLQWHQDQFH RIGDWD1
sixth line7KLV OLQH LV WKH GDWDEDVH SDWK ZKHUH WKH 5(&25'6 GDWD ILOHXVHG E\ WKH '$7 V\VWHP UHVLGHV1 7KLV IHDWXUH LV XVHIXO LI RQHZLVKHV WR VHSDUDWH GDWD IRU GLIIHUHQW WDVNV1 )RU H[DPSOH GDWDIURP WKH LQWHQVLYH FDUH XQLW PLJKW UHVLGH LQ GLUHFWRU\ ,&8ZKLOH WKH FOLQLFLDQ*V RIILFH GDWD PLJKW UHVLGH LQ 2)),&( VXEGLUHFWRU\1 7KH PDVWHU GDWDEDVH ZKLFK FRPELQHV WKH ERWKPLJKW EH LQ VXE GLUHFWRU\ 0$67(5'%1 7KLV NLQG RI VHSDUDWLRQRI GDWDEDVHV LQFUHDVHV WKH VHFXULW\ DQG JRRG PDLQWHQDQFH RIGDWD1
seventh line7KLV OLQH LV WKH GDWDEDVH SDWK ZKHUH WKH LQGH[ ILOHV RI3$7,(17 GDWD ILOH XVHG E\ WKH '$7 V\VWHP UHVLGHV DV LQ OLQHILYH
eighth line7KLV OLQH LV WKH GDWDEDVH SDWK ZKHUH WKH LQGH[ ILOHV 5(&25'6GDWD ILOH XVHG E\ WKH '$7 V\VWHP UHVLGHV DV LQ OLQH VL[1
DIAGNOSTIC APPLANATION TONOMETRY ADVANCED TOPICS •••• 25
Command Line Interface7KH FRPPDQG /LQH ,QWHUIDFH LV XVHG WR KDQGOH LQIRUPDWLRQIURP VSHFLILHG GDWDEDVH VRXUFHV1 7KH FRPPDQG OLQH RSWLRQVDUH LQYRNHG E\ WKH V\QWD[=
'$7 ?RSWLRQ! ?6WDQGDUG ,QSXW !6WDQGDUG 2XWSXW
+HUH ?RSWLRQ! VWDQGV IRU HLWKHU $/ &/ ' RU 1=
?$!LV IRU DSSHQGLQJ GDWD IURP RWKHU GDWDEDVHV1 7KLV RSWLRQ LVXVHIXO LI RQH NHHSV D PDVWHU GDWDEDVH IRU SDWLHQWV DQGUHFRUGLQJV DQG D GLIIHUHQW GDWDEDVH LQ GLIIHUHQW PHGLD DQGZLVKHV WR DSSHQG WKH ODWWHU WR WKH IRUPHU1 '$7 FKHFNV WKHH[LVWHQFH RI WKH VDPH UHFRUGV LQ WKH PDVWHU GDWDEDVH DQG LI LWIDLOV WR ILQG RQH/ LW DSSHQGV WKH UHFRUGLQJ RQWR WKH PDVWHUGDWDEDVH1
1RWH= 7KH PDVWHU GDWDEDVH GLUHFWRU\ VKRXOG EH JLYHQ LQ'$71&)* ILOH1
?&!LV IRU UH0GLDJQRVLQJ RI WKH H[LVWLQJ GDWDEDVH ZLWK FKDQJHGSDUDPHWHUV1 7KLV LV HVSHFLDOO\ XVHIXO IRU IXWXUH XSGDWLQJ RIWKH '$7 SURJUDP RQ WKH H[LVWLQJ GDWDEDVHV1 ,W DVVXPHV WKHLQGH[ NH\ WR WKH UHFRUGLQJV EH JLYHQ WKURXJK VWDQGDUG LQSXWGHYLFH1 1RUPDOO\ WKLV NH\ LV WKH RXWSXW RI RSWLRQ ?1! +VHHEHORZ,1
?'!LV IRU GHOHWLQJ FHUWDLQ UHFRUGV IURP WKH GDWDEDVH1 ,WDVVXPHV WKH LQGH[ NH\ WR WKH UHFRUGLQJV EH JLYHQ WKURXJKVWDQGDUG LQSXW GHYLFH1 1RUPDOO\ WKLV NH\ LV WKH RXWSXW RIRSWLRQ ?1! +VHH EHORZ,1
?,!LV IRU LQVWDOOLQJ '$7 WR D GLUHFWRU\1 7KLV IHDWXUH ILQGV XVHZKHQ D ILUVW WLPH RU UHSHDWHG LQVWDOODWLRQV DUH QHHGHG1
?1!LV IRU GXPSLQJ QDPHV LQ WKH GDWDEDVH1 7KLV IHDWXUH LV QRWRQO\ XVHIXO LQ GXPSLQJ WKH LQGH[ NH\ WR WKH FRQWHQWV RIGDWDEDVHV EXW DOVR XVHIXO IRU RWKHU FRPPDQG OLQH RSWLRQV WREH XVHG DV DQ LQSXW1 '$7 UHVSRQGV WR WKLV FRPPDQG E\DVNLQJ WKH QDPH RI SDWLHQWV WR EH GXPSHG1 )RU H[DPSOH LQUHVSRQVH WR WKH FRPPDQG
'$7 1 ?(QWHU!
ZKHUH '$7 UHVSRQGV ZLWK
1DPH '$9
DOORZV RQH WR GXPS WKH QDPHV DQG UHFRUGLQJV RI WKH SDWLHQWVZKR KDV QDPHV VWDUWLQJ ZLWK '$91 7KH RXWSXW RI WKH '$7EHFRPHV
3 a 5HFRUG 1XPEHU a 3DWLHQW 1DPH
ZKHUH 3 LV D OHWWHU LQGLFDWLQJ DQ HQWU\ WR WKH 3DWLHQWGDWDEDVH/ a LV D VHSDUDWRU/ 5HFRUG 1XPEHU LV WKH QXPEHUDVVRFLDWHG ZLWK WKH GDWDEDVH DQG
5 a 5HFRUG 1XPEHU a 3DWLHQW 1DPH a 'DWH RI,QVSHFWLRQ
DIA
GN
OS
TIC
AP
PLA
NA
TIO
N T
ON
OM
ET
RY
A
DV
AN
CE
D T
OP
ICS
••• • 26
ZKHUH
5LVDOHWWHU
LQGLFDWLQ
JDQ
HQWU\
WRWKH
5HFRUGLQ
JVGDWDEDVH/a
LVDVHSDUDWRU/5
HFRUGQXPEHU
LVWKH
QXPEHU
DVVRFLDWHGZLWK
WKHGDWDEDVH1
?2!7K
LVRSWLRQ
WHOOV'$7WR
RXWSXWWKHDYHUDJHG
UHFRUGHGSHULSK
HUDOSXOVH1
7KHGDWD
DUHZULWWHQ
WRILOHV
ZKLFK
DUHFUHDWHG
E\WKH'$71
7KHILOHV
DUHQDP
HGDV
)LOHQDPH1(
[WHQVLRQZKHUH
)LOHQDPHLVWKHQDP
HRIUHFRUGLQ
JVLWH
DQG([WHQVLRQ
LVWKHQXPEHU
VWDUWLQJIURP
%333%DQ
GLQFUHP
HQWLQ
JE\
RQHXSWR%<<<%1
7KHVH
QXPEHUV
FRUUHVSRQG
WRWKHRX
WSXWREWDLQ
HGE\?1
!RSWLRQ
1
?3!,VIRU
XQDWWHQ
GHGSULQWLQJ1
'$7UHSHDWHGO\
FDOOVOLQ
HIRX
URI'
$71&
)*SURJUDP
WRRX
WSXWWR
WKHSULQ
WGHYLFH1
7KH
LQSX
WZLOOEH
WKHLQGH[
NH\WR
WKHUHFRUGLQ
JVZKLFK
LVWKH
RXWSX
WREWDLQ
HGE\?1
!RSWLRQ
1
?5!7K
LVRSWLRQ
WHOOV'$7WR
UHSRUWWKHH[WUDFWHG
LQIRUP
DWLRQDERX
WWKHSDWLHQ
WVDQ
GUHFRUGLQ
JVLQ
$6&,,IRUP
DW17K
HLQSX
WZLOOEH
WKHLQGH[
NH\WR
WKHUHFRUGLQ
JVZKLFK
LVWKHRX
WSXW
REWDLQHG
E\?1
!RSWLRQ
17K
HRX
WSXWLVVHSDUDWHG
ZLWK
FRPPDV
DQGLQ
WKHRUGHU
RI=
411DP
H51
6H[
61$JH
712SHUDWRU
,'81
0HGLFDWLRQ
91'DWH
RI,QVSHFWLRQ
:1+HDUW
5DWH
;1)RRW
RIWKHZDYH>5
DZ@
<1)RRW
RIWKHZDYH>&
HQWUDO@
431)LUVWSHDN
WLPH>5DZ
@4413UHVVX
UH$W)LUVW
3HDN>5DZ
@451)LUVW
SHDNWLP
H>&HQWUDO@
4613UHVVXUH
$W)LUVW
3HDN>&HQWUDO@
4716HFRQ
G3HDN
7LPH>5DZ
@4816
HFRQG3HDN
7LPH>&HQWUDO@
4916\VWROLF
3UHVVXUH
>5DZ
@4:16
\VWROLF3UHVVX
UH>&HQWUDO@
4;1'LDVWROLF
3UHVVXUH
>5DZ
@4<1'
LDVWROLF3UHVVX
UH>&HQWUDO@
5310HDQ
3UHVVXUH
>5DZ
@5410
HDQ3UHVVX
UH>&HQWUDO@
551(MHFWLRQ
'XUDWLRQ
>5DZ
@561(
MHFWLRQ'XUDWLRQ
>&HQWUDO@
571$XJP
HQWDWLRQ
,QGH[
>5DZ
@581$
XJP
HQWDWLRQ
,QGH[
>&HQWUDO@
5910D[LP
XP
GS2GW>5DZ
@5:10
D[LPXP
GS2GW>&HQWUDO@
5;17HQVLRQ
7LPH,QGH[
5<1'LDVWROLF
7LPH,QGH[
6316XEHQ
GRFDUGLDO9LDELOLW\
5DWLR
6410HDQ
6\VWROLF
3UHVVXUH
6510HDQ
'LDVWROLF
3UHVVXUH
DIA
GN
OS
TIC
AP
PLA
NA
TIO
N T
ON
OM
ET
RY
A
DV
AN
CE
D T
OP
ICS
••• • 27
661(QG6\VWROLF
3UHVVXUH
671$XJP
HQWHG
3UHVVXUH
6815HFRUGLQ
J6LWH
691(VWLP
DWHG5HIHUHQ
FH$JH
?8!LV
IRUXSJUDGLQJ
WKHGDWDEDVH
WRDUHFHQ
WYHUVLRQ
17K
LVRSWLRQ
LVXVHG
LIWKHYHUVLRQ
RI'$7LVODWHU
WKDQ
WKHH[LVWLQ
JRQ
H17K
H'$7WKHQ
DVNVWKHQDP
HRIWK
HGDWDEDVH
WREH
XSJUDGHG1
,WWKHFUHDWHV
WZRGDWDEDVHV
ZLWK
H[WHQVLRQ
%1(:%
ZKLFK
FRQWDLQ
VWKHXSJUDGHG
YHUVLRQVRIWK
HROG
YHUVLRQV1
2QHFDQ
WKHQ
HLWKHU
HGLWWKH'$71&
)*ILOH
GHQRWHG
E\'$73$
7+VWDWHP
HQWWR
UHQDP
HWKHVH
ILOHVDV
QHZ
GDWDEDVHILOHV
RIUHQDP
HWKHH[WHQ
VLRQ%1(:%DV
WKHROG
GDWDEDVHV1,W
LVDGYLVHG
WKDW
WKLVFRP
PDQ
GVKRX
OGEH
LVVXHG
LQDQHZ
GLUHFWRU\WR
HOLPLQDWH
FKDQ
FHVRIFRQ
IXVLRQ
16LQFH
WKLVRSWLRQ
GRHVQRW
FKHFN
WKHSUHVHQ
FHRIWK
HROG
ILOH/LWLVWKHXVHUV
UHVSRQVLELOLW\
WRFRQ
ILUPWKHSUHVHQ
FHRIROG
GDWDEDVH1
?;!LV
IRUH[WUDFWLQJ
GDWDIURP
FXUUHQ
WGDWDEDVHV17K
LVRSWLRQ
LVXVHIX
OLIRQHNHHSV
DPDVWHU
GDWDEDVHIRU
SDWLHQWV
DQG
UHFRUGLQJV
DQGZLVK
HVWR
H[WUDFWSDWLHQ
WLQIRUP
DWLRQDQ
GUHFRUGLQ
JLQIRUP
DWLRQIURP
LWWR
DQRWK
HUGDWDEDVH1
,WDVVX
PHV
WKHLQGH[
NH\WR
WKHUHFRUGLQ
JVEH
JLYHQWKURX
JKVWDQ
GDUGLQSX
WGHYLFH11
RUPDOO\
WKLVNH\
LVWKHRX
WSXWRIRSWLRQ
?1!
+VHHDERYH,1
DIAGNOSTIC APPLANATION TONOMETRY ADVANCED TOPICS •••• 28 DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 29
TECHNICAL CONCEPTS
Overview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
Feature Extraction Unit
ReportWriter
DatabaseEngine
Signal Conditioning Unit
Data Acquisition Unit
The DAT System
HELP
UNIT
)LJXUH :14 7KH FRPSRQHQWV RI WKH '$7 V\VWHP VKRZLQJ WKHKLHUDUFK\ DQG LQWHUSOD\ RI VHSDUDWH XQLWV1
CHAPTER 7
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 30
Data Acquisition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
PC TimerHardware
Layer
SoftwareLayer
BIOS
DOS
DATAfterBefore
)LJXUH :15 7KH PRGLILFDWLRQ RI RULJLQDO LQWHUUXSW YHFWRU 7DEOHIRU WLPHU LQWHUUXSWV1 $UURZV LQGLFDWH WKH RUGHU RI H[HFXWLRQ RIFRGH DIWHU HDFK UHFHLYHG LQWHUUXSW1 8QGHU QRUPDO FLUFXPVWDQFHVWKH SULRULW\ LV WR WKH %,26/ '26 DQG WKHQ WKH XVHU SURJUDP1$OWHUDWLRQ RI WKH YHFWRU VHTXHQFH DOVR DOWHUV WKH H[LVWLQJ SULRULW\1
$W HDFK LQWHUUXSW/ WKH &38 UHDGV GDWD IURP WKH $2' SRUWVUHSUHVHQWLQJ WKH SUHVVXUH +Ps, DQG LI SUHVHQW/ WKH WULJJHU +Ts,/)LJXUH :161 7KH FKDQQHO QXPEHUV IRU WKHVH VLJQDOV DUHDVVXPHG WR EH VHTXHQWLDO/ WKH FKDQQHO IRU Ps EHLQJ WKH ILUVW/DQG JLYHQ LQ WKH FRQILJXUDWLRQ LQIRUPDWLRQ1 7KH Ps LV WKHQFRSLHG WR D FLUFXODU EXIIHU RI 43 VHFRQG OHQJWK1 7KXV WR ILOO WKHEXIIHU RQFH/ 45;3 VDPSOHV DUH QHFHVVDU\1 ,I WKH WULJJHU VLJQDOLV QRW SUHVHQW/ WKHQ Ps LV DOVR FRSLHG WR D EXIIHU RI HTXDO W\SHDQG VL]H IRU WULJJHU LQIRUPDWLRQ/ RWKHUZLVH Ts ZLOO EH FRSLHGLQWR LW1 2Q0OLQH FDOFXODWLRQ RI DRUWLF SUHVVXUH ZDYHIRUP LV DOVRSHUIRUPHG GXULQJ WKH DFTXLVLWLRQ VWHS RI WKH SXOVH1 7KH Ps LVGLUHFWHG WR D FRQYROXWLRQ EXIIHU/ ZKHUH LW LV FRQYROYHG ZLWK WKH
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 31
ZLQGRZ UHSUHVHQWLQJ WKH LQYHUVH WUDQVIHU IXQFWLRQ +VHH EHORZ,WR \LHOG DQ DVFHQGLQJ DRUWLF SXOVH +Cs,1 7KH Cs LV VWRUHG LQWRWKH VDPH VL]H FLUFXODU EXIIHU1 ,I '2$ FKDQQHOV DUH DYDLODEOH/DQ LGHQWLFDO FRS\ RI Cs DQG Ps DUH GXPSHG RXW IURP WKHVHSRUWV1 7KH Ps LV WKHQ GLVSOD\HG RQ WKH VFUHHQ LQ D URWDWLQJGUXP VHTXHQFH1 ,I SUHVHQW WKH GHULYDWLYH RI WKH WULJJHU VLJQDO+Ts, LV DOVR GLVSOD\HG1 7KH VDPH SURFHVV LV UHSHDWHGFRQWLQXRXVO\ XQWLO WHUPLQDWHG E\ WKH XVHU1
)LJXUH :16 %ORFN GLDJUDP RI WKH GDWD DFTXLVLWLRQ VHFWLRQ RI WKH'$7 V\VWHP1 3HULSKHUDO SUHVVXUH ZDYHIRUP DQG WULJJHU VRXUFHLV IHG LQWR WR WKH UHVSHFWLYH EXIIHUV ZKHUH WKH\ DUH NHSW DQGGLVSOD\HG1 3HULSKHUDO SUHVVXUH LV FRQYROYHG ZLWK WKH UHVSHFWLYHILOWHU DQG VWRUHG DV FHQWUDO SUHVVXUH ZDYHIRUP1 'XH WR SUHVHQFHRI '2$ FRQYHUWHUV WKH SUHVVXUH VLJQDOV +UDZ DQG FRQYROYHG, DUHGXPSHG RXW1 (QWLUH HYHQWV DUH V\QFKURQLVHG ZLWK 3& WLPHULQWHUUXSWV1 'RWWHG OLQHV LQGLFDWH RSWLRQDO SDWKV1
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
Signal Conditioning8SRQ WKH XVHU*V UHTXHVW GDWD IURP EXIIHUV +Cs, PsDQG Tc, DUHWUDQVIHUUHG WR WKH GDWDEDVH UHFRUGV/ H[FOXGLQJ WKH ODVW 5VHFRQGV WR DOORZ IRU GDWD LQWHUUXSWLRQ ZKLFK PD\ RFFXUGXULQJ WKH LQLWLDWLRQ RI WKH UHTXHVW1 7KH 3& WLPHU LV WKHQVWRSSHG DQG XSGDWHG E\ WKH V\VWHP SDUDPHWHUV ZKLOH UHDO0WLPH FORFN UHJLVWHUV DUH DOVR XSGDWHG1 7KH FRQWHQWV RI DOOEXIIHUV DUH VPRRWKHG XVLQJ :0SRLQW PRYLQJ DYHUDJH ILOWHUV WROLPLW WKH EDQGZLGWK RI WKH VLJQDOV WR 53 +]1 7KH GDWD VWULQJ LQEXIIHU Tc LV WKHQ GLIIHUHQWLDWHG DQG IXUWKHU VPRRWKHG XVLQJILUVW IRUZDUG GLIIHUHQWLDOV DQG 60SRLQW PRYLQJ DYHUDJHV WR ILQGWKH WULJJHULQJ SRLQWV DQG WR HOLPLQDWH WKH QRLVH DPSOLILHGGXULQJ GLIIHUHQWLDWLRQ1 0D[LPDO DQG PLQLPDO WKUHVKROGV DUHGHILQHG DV WKH 93( RI WKH PD[LPXP DQG PLQLPXP GHULYDWLYHV
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 32
LQ WKH HQWLUH EXIIHU FRQWHQW1 7KLV YDOXH IRU WKUHVKROG LVGHWHUPLQHG HPSLULFDOO\ DIWHU WU\LQJ IRU VHYHUDO SXOVHV1 7KHSRVLWLRQV RI WKH RQVHW RI SXOVHV DUH WKHQ GHWHUPLQHG E\FRPSDULQJ WKHP DJDLQVW WKH PD[LPXP WKUHVKROG1 7KHVHJPHQWV ZLWKLQ WKHVH PD[LPDO DQG PLQLPDO WKUHVKROGV DUHPDUNHG1 7KH FRUUHVSRQGLQJ GDWD IURP Ps DQG Cs DUHDYHUDJHG WR \LHOG DYHUDJHG UHFRUGLQJV RI SHULSKHUDO DQGV\QWKHVLVHG ZDYHIRUPV1 7KH DYHUDJHG Ps LV WKHQ FDOLEUDWHGXVLQJ JLYHQ SUHVVXUH YDOXHV1 7KH QXPHULFDO YDOXHV IRU Ps DQGCs DUH WKHQ XVHG WR HVWLPDWH RYHUDOO JDLQ DQG RIIVHW RI WKHV\VWHP1 7KH FDOLEUDWLRQ YDOXHV REWDLQHG WKLV ZD\ DUHVXEVHTXHQWO\ XVHG WR FDOLEUDWH WKH V\QWKHVLVHG DRUWLF SXOVH1
Feature Extraction'$7 H[WUDFWV ILYH WLPH UHODWLYH SRLQWV RQ WKH ZDYHIRUP IURPZKLFK SDUDPHWHUV UHODWLQJ WR WKH KHDUW DQG DUWHULDO V\VWHP DUHGHWHUPLQHG/ +)LJXUH :17,1
)HDWXUHV RI WKH $UWHULDO 3XOVHSecond Shoulder
First Shoulder
Foot of the Pulse
Pulse Duration
T 1
T 2
T i
T f
T T
Aortic valve closure(Incisura)
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
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 33
Calculated Values)URP DYHUDJHG SHULSKHUDO DQG V\QWKHVLVHG FHQWUDO SXOVHV WKHIROORZLQJ YDOXHV DUH FDOFXODWHG=
Ejection DurationED Secs T Ti f( ) = −
Heart RateHR Beats TT( min) = 60
Pressure at Diastole
P mmHg P Td f( ) =Pressure at Systole
P mmHg P Ts p( ) =Mean Arterial Pressure
MAP mmHg
P
T
ii
T
T
T
( ) = =∑
1
Pressure at First Shoulder
P mmHg P T1 1( ) =Pressure at Second Shoulder
P mmHg P T2 2( ) =Augmented Pressure
AP mmHg P T P T( ) = −2 1Mean Diastolic Pressure
MDP mmHg
P
T T
ii T
T
T i
i
T
( )( )
=−
=∑
Mean Systolic Pressure
MSP mmHg
P
T T
ii T
T
i f
f
i
( )( )
=−
=∑
Augmentation Index
AIP PP Pd
d(%) = × −−
100 2
1Tension Time Index
TTI mmHg Beats HR MSP T Ti f( . min) = × × −1 6Diastolic Time Index
DTI mmHg Beats HR MDP T TT i( . min) = × × −1 6Subendocardial Viability Ratio
SVIDTITTI
%0 5 = ×100
Reflection Transit TimeRT Sec T T( ) = −2 1
Maximum Rate of Rise
Max dP dt mmHg Sec MaxdPdt
( ) = çåäâ
5HIHUHQFH $JH
RA(Years) = × − +0 642 100 33 81. .AI0 5
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 34
5HIHUHQFH $JH LV WKH DJH FRUUHVSRQGLQJ WR WKH DXJPHQWDWLRQLQGH[ DV GHULYHG IURP DJHLQJ VWXGLHV1 %3>@% LQGLFDWHV WKHHOHPHQW RI WKH SUHVVXUH ZDYHIRUP DUUD\1
Database Engine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
),(/' 7<3(3DWLHQW 1DPH 6PDOO 6WULQJ'DWH RI YLVLW 6PDOO 6WULQJ0HGLFDWLRQ $UUD\>4115@ 2) 6PDOO
6WULQJ1RWHV 6PDOO 6WULQJ2SHUDWRU ,G :RUG7ULJJHU 3UHVHQW %RROHDQ'LDJQRVHG %RROHDQ*DLQ RI 6\VWHP 5HDO2IIVHW RI 6\VWHP 5HDO7ULJJHU 6HULHV $UUD\>41145;@ RI ,QWHJHU5HFRUGHG 3XOVHV $UUD\>5DZ11&RQYROYHG@ RI
3XOVH %XIIHU+HDUW 5DWH 5HDO$XJPHQWHG 3UHVVXUH 5HDO5HIHUHQFH $JH 5HDO7HQVLRQ 7LPH ,QGH[ 5HDO'LDVWROLF 7LPH ,QGH[ 5HDO6XEHQGRFDUGLDO 9LDELOLW\5DWLR
5HDO
0HDQ 6\VWROLF 3UHVVXUH 5HDO0HDQ 'LDVWROLF 3UHVVXUH 5HDO(QG 6\VWROLF 3UHVVXUH 5HDO
7DEOH :14 7KH WHPSODWH IRU UHFRUG RI GDWDEDVH ILOH%5(&25'6%1 7KLV ILOH LV LQGH[HG ZLWK 3DWLHQW 1DPH ILHOG DQG'DWH RI YLVLW DQG OLQNHG WR 3$7,(176 GDWDEDVH1 )RU H[SODQDWLRQRI GHILQHG W\SHV VHH 7DEOHV :16 DQG :171
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 35
,(/' 7<3(3DWLHQW 1DPH 7H[W 6WULQJ3DWLHQW ,' 6WULQJ>;@3DWLHQW $JH ,QWHJHU6H[ 6WULQJ>5@$GGUHVV 6PDOO 6WULQJ'LVWDQFHV $UUD\ >5DGLDO11 $RUWLF@ RI UHDO&RQYROXWLRQV &RQYROXWLRQ 7\SHV
7DEOH :15 7KH WHPSODWH IRU UHFRUG RI GDWDEDVH ILOH%3$7,(176%1 7KLV ILOH LV LQGH[HG ZLWK 3DWLHQW 1DPH ILHOG DQGUHODWHG WR 5(&25'6 GDWDEDVH1 )RU H[SODQDWLRQ RI GHILQHGW\SHV VHH 7DEOHV :16 DQG :171
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
),(/' 7<3(3XOVH 1DPH 3XOVH 7\SHV7RQRPHWHU ,QSXW 6HULHV 3XOVH 6WUHDP$YHUDJHG 3XOVH 3XOVH/HQJWK RI $YHUDJHG 3XOVH ,QWHJHU6\VWROLF 3UHVVXUH 5HDO'LDVWROLF 3UHVVXUH 5HDO0HDQ 3UHVVXUH 5HDO7LPH WR )RRW 2I :DYH 5HDO(MHFWLRQ 'XUDWLRQ 5HDO0D[ GS2GW= 5HDO6\VWROLF 2QVHW 5HDO7LPH WR ,QFLVXUD 5HDO5HVHUYHG 5HDO7LPH WR )LUVW 6KRXOGHU 5HDO7LPH WR 6HFRQG 6KRXOGHU 5HDO5HYHUVH 6KRXOGHU ,QGH[ 5HDO
7DEOH :16 7KH W\SH GHILQHG IRU DQ\ SXOVH1 ,W FRQWDLQVLQIRUPDWLRQ DERXW WKH RULJLQDO VHULHV RI SXOVH/ DYHUDJHG SXOVHDQG WKH GHULYHG LQGLFHV1 7KLV EXIIHU LV GXSOLFDWHG IRU ERWK UDZDQG FRQYROYHG VLJQDOV DQG FRUUHVSRQGV WR 36 DQG &6 EXIIHUV LQ)LJXUH :161
DIAGNOSTIC APPLANATION TONOMETRY TECHNICAL CONCEPTS •••• 36
1HZ 7\SH 'HILQLWLRQ6PDOO 6WULQJ 6WULQJ>73@3XOVH $UUD\>311588@ RI 5HDO3XOVH VWUHDP $UUD\>3114356@ RI ,QWHJHU&RQYROXWLRQ :LQGRZ $UUD\ >0491149@ RI 5HDO3XOVH 7\SHV 5DGLDO/ &DURWLG/ )HPRUDO/
%UDFKLDO/ $[LOOD/ 6XEFODYLDQ/'RUVDOLV 3HGLV/ $RUWLF
&RQYROXWLRQ 7\SHV $UUD\>5DGLDO11$RUWLF@ RI&RQYROXWLRQ :LQGRZ
6LJQDO 7\SHV 7ULJJHU/ 5DZ/ &RQYROYHG
7DEOH :17 7KH W\SH GHILQLWLRQV EDVHG RQ RUGLQDO GHILQLWLRQVVXSSOLHG E\ 3DVFDO ODQJXDJH XVHG LQ UHFRUG GHILQLWLRQV1
DIAGNOSTIC APPLANATION TONOMETRY HOW TO ? •••• 37
HOW TO ?
Import Existing Tape Data to DAT'$7 GRHV QRW GLIIHUHQWLDWH EHWZHHQ D UHDO WLPH VLJQDO DQG DQRII0OLQH VLJQDO1 7KHUHIRUH LW LV SRVVLEOH WR UHFRUG VLJQDOV IURPPDJQHWLF WDSHV DV UHDO WLPH VLJQDOV1 6LPSO\ SOXJ LQ WKHDSSURSULDWH SUHVVXUH DQG WULJJHU VLJQDO WR WKH GDWD DFTXLVLWLRQERDUG/ W\SH LQ WKH SDUWLFXODU GHWDLOV RI WKH SDWLHQW DQG SURFHHGDV XVXDO1 7KLV IHDWXUH LV TXLWH XVHIXO WR DUFKLYH WKH H[LVWLQJDQDORJ WDSHV LQ GLJLWDO IRUP XVLQJ GDWDEDVH IHDWXUHV RI WKH'$7 V\VWHP1
Conduct Different Studies Simultaneously,I \RX ZDQW WR FRQGXFW VWXGLHV LQYROYLQJ GLIIHUHQW SURFHGXUHVDQG2RU GLIIHUHQW VHWV RI VXEMHFWV \RX FDQ LQVWDOO WKH '$7V\VWHP LQ GLIIHUHQW GLUHFWRULHV1 )RU H[DPSOH LI \RX KDYH DVWXG\ WR H[SORUH WKH HIIHFWV RI $&( LQKLELWRUV DQG 1,75$7(6/LQVWDOO WKH '$7 LQWR VXE GLUHFWRULHV FDOOHG
• &=?$&( DQG
• &=?1,75$7(6
'XULQJ SHUIRUPLQJ WKH $&( H[SHULPHQW \RX FDQ LVVXHFRPPDQG
• 6(7 '$73$7+ &=?$&(
LQ '26 SURPSW DQG ZRUN LQ $&( HQYLURQPHQW ZKLOH LVVXLQJ
• 6(7 '$73$7+ &=?1,75$7(6
WDNHV \RX WR 1,75$7( HQYLURQPHQW1
Combine Different Databases:KHQ \RX KDYH WZR RU PRUH '$7 GDWDEDVHV VXFK DV WKH RQHJLYHQ LQ WKH SUHYLRXV H[DPSOH DQG \RX ZDQW WR DSSHQG WKHP
CHAPTER 8
DIAGNOSTIC APPLANATION TONOMETRY HOW TO ? •••• 38
WR D PDVWHU GDWDEDVH \RX FDQ XVH WKH '$7 FRPPDQG OLQHLQWHUIDFH WR SHUIRUP WKLV WDVN1
)RU H[DPSOH/ DVVXPH WKDW \RX KDYH D PDVWHU GDWDEDVH LQGULYH '= DQG VXE GLUHFWRU\ 0$67(5'% WKH FRPPDQGV WKDWQHHG WR EH LVVXHG DUH=
41 6(7 '$73$7+ '=?0$67(5'%
51 '$7 $
'$7 UHVSRQGV ZLWK=
$SSHQG 3DWLHQWV IURP
7\SH=
• &=?$&(?3$7,(1761'$7
DQG
$SSHQG 5HFRUGV IURP
7\SH=
• &=?$&(?5(&25'61'$7
'$7 ZLOO DSSHQG GDWD IURP $&( VWXG\ WR WKH 0$67(5'%11LWUDWHV GDWDEDVH FDQ EH DSSHQGHG VLPLODUO\
Perform Statistical Analysis of the Data$OWKRXJK WKH UHVXOWV RI WKH UHSRUWV FDQ EH W\SHG LQWR DVWDWLVWLFDO DQDO\VLV SURJUDP/ LW LV PRUH FRQYHQLHQW WR LPSRUWWKH GDWD WR WKH VDPH SURJUDP DV $6&,, LQSXW1
)RU H[DPSOH DVVXPH WKDW \RX ZDQW WR DQDO\VH WKH GDWD RISDWLHQW+V, ZLWK LQLWLDOV 61 7KH VWHSV WR SHUIRUP WKLV LV
41 6HOHFW WKH DSSURSULDWH GLUHFWRU\ XVLQJ 6(7 '$73$7+ FRPPDQG
51 7\SH '$7 1 !1$0(6
'$7 UHVSRQGV ZLWK
1DPH
61 7\SH $-
7KH QDPHV DQG UHFRUGLQJ WLPHV RI SDWLHQW QDPHV DUH QRZGXPSHG LQWR ILOH QDPHG 1$0(61 1RWH= ,I \RX ZDQW WR GXPSWKH HQWLUH QDPHV W\SH MXVW ?(QWHU! RU ?5HWXUQ!
71 7\SH '$7 5 ?1$0(6 !5(68/76
7KH UHVXOWV RI WKHVH UHFRUGLQJV ZLOO EH LQ ILOH 5(68/76 ZKLFKFRXOG EH LPSRUWHG LQWR DQ\ VWDWLVWLFDO SDFNDJH XVLQJ $6&,,
DIAGNOSTIC APPLANATION TONOMETRY HOW TO ? •••• 39
LPSRUW ILOWHUV1 7KH FRQWHQWV RI GXPSHG ILHOGV DUH H[SODLQHG LQ&KDSWHU 91
Print Unattended$OWKRXJK \RX FDQ SULQW WKH UHSRUWV ZKLOH YLHZLQJ WKHP LW LVDOVR SRVVLEOH DQG SUREDEO\ PXFK HDVLHU WR SULQW WKHP ODWHU17KLV FDQ EH GRQH LQ WZR GLIIHUHQW ZD\V=
• 'XULQJ LQWHUDFWLYH UHFRUGLQJ
41 8VH D WH[W HGLWRU WR HGLW '$71&)* ILOH WR FKDQJH SULQWHUSRUW LQ OLQH 7 IURP !/374 WR !!5(32576
51 5XQ '$7 DV XVXDO DQG SULQW UHSRUWV DV XVXDO1 +RZHYHU/UHVXOWV RI WKH UHSRUWV ZLOO QRW EH SULQWHG LPPHGLDWHO\ EXWDSSHQGHG WR D ILOH 5(32576
61 ([LW '$7
71 7\SH &23< 5(32576 /3741 /374 LV WKH GHIDXOW SULQWHUSRUW
81 5HYHUW EDFN WR WKH GHIDXOW SRUW E\ QRWLQJ WKH GHIDXOW SRUWLQ OLQH 7 DV LQ VWHS 4
• 'XULQJ FRPPDQG OLQH VHVVLRQ
41 &RQVLGHU \RX REWDLQHG D ILOH FDOOHG 1$0(6 IROORZLQJ WKHSUHYLRXV H[DPSOH1 ,W LV SRVVLEOH WR SULQW WKHVH UHFRUGLQJVXQDWWHQGHG1
51 7\SH '$7 3 ?1$0(6
'$7 ZLOO SULQW WKH UHFRUGLQJV LQ 1$0(6 ILOH +LH WKH OLQHV WKDWVWDUWV ZLWK 5a, WR WKH GHYLFH DQG SRUW JLYHQ LQ OLQH IRXU LQ'$71&)* ILOH1 3OHDVH PDNH VXUH WKDW WKH SULQWHU KDV QRW UXQRXW RI SDSHU VLQFH QR FKHFNLQJ RI DYDLODELOLW\ RI SDSHU LV PDGHGXULQJ SULQWLQJ1
Use Reports in Word Processors:RUG SURFHVVRUV XVLQJ 3RVW6FULSW SULQWHUV DOORZ RQH WR HPEHG(36 +(QFDSVXODWHG 3RVW6FULSW, ILOHV LQWR GRFXPHQWV1 7KHJUDSK JLYHQ LQ )LJXUH 815 LV DQ H[DPSOH RI VXFK SURFHVV1 7RSHUIRUP WKLV=
41 8VH D WH[W HGLWRU WR HGLW '$71&)* ILOH1 &KDQJH OLQH 7 WR3676&5,37 !5(3257
51 7\SH '$7
61 9LHZ WKH UHFRUG WR EH HPEHGGHG
71 3UHVV ?)5! WR VDYH LW
81 4XLW IURP '$7
91 5XQ \RXU ZRUG SURFHVVRU DQG HPEHG 5(3257 ILOH LQWR\RXU GRFXPHQW1
DIAGNOSTIC APPLANATION TONOMETRY HOW TO ? •••• 40
Record Waveforms Without Actually Seeing Them6RPHWLPHV LW PD\ EH KDUG WR YLHZ WKH SUHVVXUH SXOVHV HLWKHUGXH WR LQDSSURSULDWH OHYHOV RI SUHVVXUH SXOVHV RU GXH WRUHPRWHQHVV RI WKH FRPSXWHU VFUHHQ1 7KH XVH RI VRXQG XQGHUWKHVH FLUFXPVWDQFHV FDQ KHOS D WUDLQHG RSHUDWRU VLQFH WKHJHQHUDWHG VRXQG LV IUHTXHQF\ PRGXODWHG LQ FRQMXQFWLRQ ZLWKSUHVVXUH SXOVH DPSOLWXGH1 6LPSO\ SUHVV ?$OW0'! WR WXUQ WKHVRXQG RQ DQG RII1
Measure Pulse wave Velocity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
)RU H[DPSOH IRU WKH SXOVH ZDYH YHORFLW\ LQ WKH DUP=
• PHDVXUH WKH FDURWLG SXOVH ZLWK (&* WULJJHULQJ1 1RWHWKH GLVWDQFH WR VXEVWHUQDO QRWFK/ lc/ DQG UHFRUG WKH V\VWROLFRQVHW WLPH / tc
• PHDVXUH WKH UDGLDO SXOVH ZLWK (&* WULJJHULQJ1 1RWHWKH GLVWDQFH WR VXEVWHUQDO QRWFK/ lr/ DQG UHFRUG WKH V\VWROLFRQVHW WLPH / tr1
• SXOVH ZDYH YHORFLW\ LQ WKH DUP EHFRPHV=
• pwv arml lt tr c
r c( ) = −
−
DIAGNOSTIC APPLANATION TONOMETRY WHAT TO DO IF ? •••• 41
WHAT TO DO IF ?
Cannot Run DAT Program
Database is corrupt'HOHWH DOO ILOHV HQGLQJ ZLWK 1'; DQG UH0UXQ '$71
DATHELP.MSG is Missing
Path to DATHELP.MSG is invalid(GLW '$71&)* WR LQFOXGH WKH SDWK
Cannot Register the Pulse. The System Crashes
A/D converter is not installed properly,QVWDOO WKH $2' FRQYHUWHU DW $GGUHVV 5(& +KH[DGHFLPDO,1
Not enough memory,QFUHDVH PHPRU\ RU UHPRYH PHPRU\ FRQVXPLQJ GULYHUVDQG2RU WHUPLQDWH DQG VWD\ UHVLGHQW SURJUDPV
The clock speed of the PC is higher than
12 MHz5HGXFH WKH FORFN VSHHG RI WKH 3&
CHAPTER 9
DIAGNOSTIC APPLANATION TONOMETRY WHAT TO DO IF ? •••• 42
Cannot See Any Data on the Screen
Gain and offset have not been set properly8VH ?$OW06! WR UHFHOHEUDWH WKH SXOVH LQ DFTXLVLWLRQ PHQX
Channel number does not match the
analogue input(GLW OLQH 4 RI '$71&)*
The System Crashes, no Feature is Extracted or Results areWrong
Unexpected termination of the registration
of pulses5HSHDW WKH SXOVH DFTXLVLWLRQ SURFHGXUH ZLWK OHVV LQWHUUXSWLRQLQ WKH ODVW 43 VHFRQGV EHIRUH SUHVVLQJ ?)5! IRU GLDJQRVLV1
The PC system is too slowLQFUHDVH WKH FORFN VSHHG RU DGG D PDWK FR0SURFHVVRU
Cannot Find a Recording Registered Previously
System date and time settings are wrong8VH 'DWH DQG 7,0( &RPPDQGV LQ '26 FRPPDQG OLQH WR VHWWKH GDWH DQG WLPH SURSHUO\1
The path to DAT.CFG file is not correct8VH 6(7 FRPPDQG WR VHW D SDWK WR '$71&)* ILOH
DIAGNOSTIC APPLANATION TONOMETRY WHAT TO DO IF ? •••• 43