Deformation Imaging
AN ABSTRACT OF THE THESIS OF
Marco L. Margiocco for the degree of Master of Science in Veterinary Science
presented on December 17, 2007.
Title: Deformation Imaging.
Abstract approved:
D. David Sisson
Recent developments in the field of clinical echocardiography allow the
ultrasonographer to objectively quantify both regional and global myocardial
function. Regional deformation (Strain) and deformation rate (Strain Rate) are
novel indices of ventricular function that can be estimated non-invasively by
ultrasonographic interrogation of the heart. Deformation Imaging (DI)
represents a family of echocardiographic techniques that can be employed to
detect, quantify and display the characteristics of physical deformation of the
myocardium. The objective of the studies described in this thesis was to
investigate Doppler-derived deformation imaging in dogs.
This Master’s Degree thesis is structured in two separate studies. The initial
study compared Doppler-derived DI measures of ventricular function with
more traditional invasive indices of cardiovascular function in healthy adult
anesthetized dogs over a range of hemodynamic conditions created by serial
pharmacologic
manipulations.
Five
adult
healthy
dogs
underwent
simultaneous cardiac catheterization and transthoracic echocardiography
under general anesthesia. The following invasive indices were monitored
during the study: cardiac output (CO), femoral arterial (FA) and left ventricular
(LV) pressures, +dP/dtmax, -dP/dtmax, right atrial (RA) and pulmonary arterial
(PA) pressures, and pulmonary capillary wedge pressure (PCWP). Sequential
manipulations of systolic function, afterload and preload were performed,
respectively, by means of dobutamine, nitroprusside, and hetastarch infusions.
Significant changes were induced in the following invasive hemodynamic
parameters: cardiac output (p<0.0001, range 2.600 - 9.340 L/min), +dP/dtmax
(p=0.0152, range 953.7 - 3822 mmHg/s), left ventricular end-diastolic pressure
(LVEDP) (p<0.0001, range 0.210 -16.43 mmHg), mean right atrial pressure
(p<0.0004, range -2.490 - 9.920 mmHg), systemic vascular resistance (SVR)
(p<0.0001, range 510.0 - 2652 dyne*sec/cm5), and pulmonary vascular
resistance (PVR) (p<0.05, range 256.4 -1635 dyne*sec/cm5). All dogs
underwent a complete echocardiographic exam before anesthesia, after
induction of general anesthesia and after each hemodynamic manipulation. A
total of 750 regions of interest (ROIs) were included in the final analysis.
Measurable plots were obtained in 741 (98.67%) out of 750 ROIs. Peak
systolic strain rate (SSR) values obtained at the 6 ROIs did not differ
significantly from each other and the avSSR (average of the 6 ROIs)
negatively correlated with +dP/dtmax (r=0.9792, p=0.0208). Individual values of
SSR from the different ROIs also negatively correlated with +dP/dtmax, with the
exception of ROI 4 (basal portion of the interventricular septum). The lack of
correlation was attributed to high variability in the measurements obtained
from this specific ROI. The avSSR was significantly reduced, i.e. systolic
function was increased, by the dobutamine infusion, but was not significantly
changed during nitroprusside and hetastarch infusions. The LVEDP was
significantly reduced during infusion of nitroprusside, apparently not
influencing avSSR, and suggesting the possible preload independence of this
parameter in the hemodynamic range generated in the study. Nitroprusside
infusion effected a significant reduction of systemic vascular resistance without
alteration of avSSR, suggesting a possible afterload independence of this
parameter, at least in the hemodynamic range obtained in this study.
Study 2 was conducted on healthy adult non-sedated dogs divided into four
groups (group 1: Body weight (BW) < 15 kg, group 2: BW 15 - 30 kg, group 3:
BW > 30 kg, and group 4: Doberman pinscher dogs). The aim of the study
was to collect information on the feasibility, repeatability and reproducibility of
Doppler-derived DI performed in a clinical setting. Of the initial 56 dogs from
which echocardiographic data were collected, 7 dogs (12.5%) were excluded
because of poor image quality, and data from 49 (87.5%) dogs were included
in the final analysis. Of a total of 1470 ROIs analyzed, 4.65% yielded good
quality curves, 88.3% were characterized by a significant amount of noise but
retained a discernible pattern, and 7.14% were considered non-interpretable.
Some pair-wise comparisons of the SSR values obtained from the 6 ROIs
reached statistically significant difference. DI values in the Doberman pinscher
group did not differ significantly compared to values obtained from a BWmatched group. DI values in group 1 dogs, but not the traditional indices FS
and EF, were significantly higher compared to groups 2 and 3. Results from a
Bland-Altman analysis revealed overall poor clinical repeatability and
reproducibility, as a result of the wide variability of the data.
Based on the results of study 1 we conclude that Doppler-derived SSR
represents a useful load-independent index of global systolic function, as
demonstrated in anesthetized dogs. However study 2 suggests that, in a
population of non-sedated dogs, with a relatively wide range of breeds and
BW, Doppler-derived DI measurements are characterized by less than optimal
intra-operator and inter-operator variability which may limit its value as a tool
to evaluate myocardial function in a clinical setting. Comparison to other newly
developed
non-invasive
techniques
to
estimate
regional
myocardial
deformation is needed to determine the eventual clinical utility of the method.
©Copyright by Marco L. Margiocco
December 17, 2007
All Rights Reserved
Deformation Imaging
by
Marco L. Margiocco
A THESIS
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Master of Science
Presented December 17, 2007
Commencement June 2008
Master of Science thesis of Marco L. Margiocco
presented on December 17, 2007.
APPROVED:
Major Professor, representing Veterinary Science
Head of the Department of Clinical Sciences
Dean of the Graduate School
I understand that my thesis will become part of the permanent collection of
Oregon State University libraries. My signature below authorizes release of my
thesis to any reader upon request.
Marco L. Margiocco, Author
ACKNOWLEDGEMENTS
I wish to express sincere appreciation to the many individuals who have
offered support, inspiration, and encouragement throughout this research
effort. I like to thank Dr. Michele Borgarelli, who first introduced me to the
world of veterinary cardiology, and still teaches me a lot about cardiology and
friendship. A special thank goes to the Students, Staff, Faculty, breeders and
owners who volunteered their dogs to the study. Immense gratitude goes to
Ms. Robyn Panico, who devoted innumerable hours of technical support and
who helped me throughout my whole residency: thank you Robyn for being a
good and sincere friend to me. Dr. Craig Mosley and Dr. Rolando Quesada
provided anesthesia support during the long hours of the invasive study, and
without their assistance the study would have been much more challenging.
The author would also like to thank the senior students, too numerous to be
listed, who helped with data collection. Dr. Paul Rist and Dr. Chris Cebra also
participated in my thesis committee and provided constructive criticism during
the course of the study. Dr. Donna Champeau kindly accepted to serve as
Graduate Council Representative. Dr. Randolph P. Martin and Dr. Andreas
Kalogeropoulos at Emory University Hospital shared with me their experience
and perplexities with the technique object of this investigation. Statistical
assistance was provided by Dr. Roby Joehanes, Department of Statistics, and
Dr. Ronette Gehring, Department of Clinical Sciences, both at Kansas State
University. Dr. Barret Bulmer was instrumental in planning and conducting the
study and was always willing to provide help and assistance. I feel particularly
fortunate for the opportunity of working with him. My deepest expression of
appreciation goes to my mentor Dr. David Sisson, who accepted me (twice)
into his program, helped me find my way in the “New World”, continuously
offered encouragement and advice, and shared with me his knowledge and
love for science, decorated with pearls of surfer wisdom. And, above all, the
final thank-you goes to my wife Patrizia, who always supported my dreams
and contributed making them come true.
TABLE OF CONTENTS
Page
Introduction ...................................................................................................... 1
Literature review............................................................................................... 3
Motion versus deformation imaging .............................................................. 3
Review of M-mode and 2D-based echocardiographic techniques used to
evaluate ventricular function ......................................................................... 3
Tissue Velocity Imaging (Motion Imaging) and its natural evolution:
Deformation Imaging .................................................................................... 6
The physical concepts of Strain and Strain Rate ........................................ 10
Strain and Strain Rate by ultrasonography: Strain Rate Imaging (SRI) ...... 13
Technical limitations of SRI ........................................................................ 15
Technical strengths of SRI.......................................................................... 17
Current and future clinical applications of SRI ............................................ 18
Normal SRI tracings.................................................................................... 24
Doppler-derived DI in dogs ......................................................................... 24
Aims of the Project ......................................................................................... 33
Materials and methods ................................................................................... 34
Study 1 - Invasive indirect validation of Doppler-derived Strain Rate imaging
in anesthetized healthy adult dogs.............................................................. 34
Hypothesis .............................................................................................. 34
Inclusion criteria and study protocol ........................................................ 34
Statistical methods .................................................................................. 39
Study 2 – Deformation Imaging in normal adult dogs ................................. 39
Hypotheses ............................................................................................. 40
Inclusion criteria and study protocol ........................................................ 41
Estimation of the intra-operator variability ............................................... 42
Estimation of the inter-operator variability ............................................... 42
Statistical methods .................................................................................. 42
TABLE OF CONTENTS (Continued)
Page
Results ........................................................................................................... 46
Study 1 ....................................................................................................... 46
Study 2 ....................................................................................................... 62
Discussion...................................................................................................... 98
Study 1 ....................................................................................................... 98
Study 2 ..................................................................................................... 103
Conclusion ................................................................................................... 112
Bibliography ................................................................................................. 114
LIST OF FIGURES
Figure
Page
1: Comparison between single- and bi-plane methods for the estimation of left
ventricular volumes ................................................................................. 25
2: The Doppler principle: stationary target...................................................... 26
3: The Doppler principle: target in relative motion .......................................... 26
4: Velocity/amplitude relationship of backscatters from blood and tissues ..... 27
5: Example of one-dimensional S................................................................... 27
6: The four components of bidimensional S ................................................... 28
7: Relationships between Cartesian axes and cardiac longitudinal (L), radial
(R) and circumferential (C) axes ............................................................. 28
8: Schematic of Doppler-derived Strain Rate Imaging.................................... 29
9: Short axis view showing radial (R) and circumferential (C) axes................ 29
10: Long axis view showing the longitudinal (L) axis ...................................... 30
11: The cosine function .................................................................................. 30
12: Comparison between TVI and SRI with respect to the effects of translation
of the whole heart.................................................................................... 31
13: Schematic representation of the reciprocating pump principle, with
emphasis on the systolic phase .............................................................. 31
14: Normal longitudinal S (solid line) and reduced S (dashed line) as a typical
example of pathologic S.......................................................................... 32
15: Normal SR (solid line) and pathologic SR (dashed line) as commonly
found with ischemia................................................................................. 32
16: Schematic representation of the apical 4-chamber view .......................... 45
LIST OF FIGURES (Continued)
Figure
Page
17: Results of a repeated measure ANOVA on the variable RR between the 5
stages of the study .................................................................................. 49
18: Results of a repeated measure ANOVA on the variable FS between the 5
stages of the study .................................................................................. 50
19: Results of a repeated measure ANOVA on the variable CO during the 3
pharmacologic manipulations.................................................................. 50
20: Results of a repeated measure ANOVA on the variable SV during the 3
pharmacologic manipulations.................................................................. 51
21: Results of a repeated measure ANOVA on the variable +dP/dtmax during
the 3 pharmacologic manipulations......................................................... 51
22: Results of a repeated measure ANOVA on the variable LVEDP during the
3 pharmacologic manipulations............................................................... 52
23: Results of a repeated measure ANOVA on the variable mean RAP during
the 3 pharmacologic manipulations......................................................... 52
24: Results of a repeated measure ANOVA on the variable SVR during the 3
pharmacologic manipulations.................................................................. 53
25: Multiple comparison test on the variable PVR between the 3
pharmacologic manipulations.................................................................. 53
26: Multiple comparison test on the variable -dP/dtmax between the 3
pharmacologic manipulations.................................................................. 54
27: Multiple comparison test on the variable mean FA pressure between the 3
pharmacologic manipulation ................................................................... 54
28: Multiple comparison test on the variable PCWP between the 3
pharmacologic manipulations.................................................................. 55
29: Result of a linear regression analysis with Pearson correlation test,
between LiDCO® and Thermodilution techniques ................................... 55
LIST OF FIGURES (Continued)
Figure
Page
30: Result of a linear regression analysis with Pearson correlation test,
between 2D-FS and +dP/dtmax ................................................................ 56
31: Results of a repeated measure ANOVA comparing the SSR values
between the 6 ROIs ................................................................................ 56
32: Results of a repeated measure ANOVA comparing the avSSR during the
five phases of the study .......................................................................... 57
33: Result of a linear regression analysis with Pearson correlation test,
between 2D-FS and avSSR .................................................................... 57
34: Result of a linear regression analysis with Pearson correlation test,
between SSR and +dP/dtmax ................................................................... 58
35: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 1 and +dP/dtmax .................................... 58
36: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 2 and +dP/dtmax .................................... 59
37: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 3 and +dP/dtmax .................................... 59
38: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 5 and +dP/dtmax .................................... 60
39: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 6 and +dP/dtmax .................................... 60
40: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 4 and +dP/dtmax .................................... 61
41: Sex distribution in the four study groups .................................................. 71
42: Age distribution in the four study groups .................................................. 71
43: BW distribution in the four study groups................................................... 72
LIST OF FIGURES (Continued)
Figure
Page
44: Distribution of plot quality scores in the four study groups ....................... 72
45: Distribution of the DI variable SSR in the six ROIs................................... 73
46: Distribution of the DI variable S in the six ROIs........................................ 74
47: Distribution of the DI variable SRE in the 6 ROIs ..................................... 75
48: Distribution of the DI variable SRA in the 6 ROIs ..................................... 75
49: Distribution of the DI variable SSR in the 6 ROIs in group 1 .................... 76
50: Distribution of the DI variable SSR in the 6 ROIs in group 2 .................... 76
51: Distribution of the DI variable SSR in the 6 ROIs in group 3 .................... 77
52: Distribution of the DI variable SSR in the 6 ROIs in group 4 .................... 77
53: Distribution of the HR values in the four study groups.............................. 78
54: Distribution of the HR values during the two echocardiographic
examinations performed by the same operator ....................................... 78
55: Distribution of the HR values during the two echocardiographic
examinations performed by the two different operators .......................... 79
56: Distribution of the BP values in the 4 study groups .................................. 79
57: Comparison between a BW-matched subgroup 3 and group 4................ 80
58: Comparison between the EF values in the BW-matched subgroup 3 and
group 4.................................................................................................... 80
59: Comparison between the avSSR values in the BW-matched subgroup 3
and group 4 ............................................................................................. 81
LIST OF FIGURES (Continued)
Figure
Page
60: Comparison between the avS values in the BW-matched subgroup 3 and
group 4.................................................................................................... 81
61: Comparison between the avSSR values in the three BW groups ............ 82
62: Same analysis as in Figure 36, excluding dogs from group 4 .................. 82
63: Comparison between the FS values in the three BW groups ................... 83
64: Comparison between the EF values in the three BW groups ................... 83
65: Simple linear regression analysis between avSSR and BW..................... 84
66: Simple linear regression analysis between avS and BW.......................... 84
67: Simple linear regression analysis between avSSR and FS...................... 85
68: Simple linear regression analysis between avS and FS........................... 85
69: Simple linear regression analysis between avSSR and EF...................... 86
70: Simple linear regression analysis between avS and EF........................... 86
71: Simple linear regression analysis between avSSR and HR ..................... 87
72: Simple linear regression analysis between avS and HR .......................... 87
73: Bland-Altman plots of the repeatability sub-analysis for avSSR............... 88
74: Bland-Altman plots of the repeatability sub-analysis for avSRE............... 89
75: Bland-Altman plots of the repeatability sub-analysis for avSRA............... 90
76: Bland-Altman plots of the repeatability sub-analysis for avS.................... 91
77: Bland-Altman plots of the reproducibility sub-analysis for avSSR ............ 93
78: Bland-Altman plots of the reproducibility sub-analysis for avSRE ............ 94
LIST OF FIGURES (Continued)
Figure
Page
79: Bland-Altman plots of the reproducibility sub-analysis for avSRA ............ 95
80: Bland-Altman plots of the reproducibility subanalysis for avS .................. 96
LIST OF TABLES
Table
Page
1: Imaging views and modalities utilized in the echocardiographic acquisition
protocol ................................................................................................... 44
2: Imaging views and modalities utilized in the echocardiographic off-line
analysis ................................................................................................... 45
3: Summary table indicating the most important hemodynamic changes
obtained during the pharmacologic manipulations .................................. 49
4: Descriptive statistics of the 49 dogs included in the final analyses ............ 69
5: DI parameters from the six ROIs (Mean values) ........................................ 70
6: DI parameters as average of the six ROIs ................................................. 70
7: Results of a Dunn’s Multiple Comparison Test for the variable SSR,
between the 6 ROIs ................................................................................ 73
8: Results of a Dunn’s Multiple Comparison Test for the variable S, between
the 6 ROIs............................................................................................... 74
9: Results of a Bland-Altman analysis of the data from the repeatability
subanalysis ............................................................................................. 92
10: Results of a Bland-Altman analysis of the data from the reproducibility
subanalysis ............................................................................................. 97
LIST OF ABBREVIATIONS
+dP/dtmax: Maximum rate of pressure rise
-dP/dtmax: Maximum rate of pressure decay
2D imaging: two-dimensional imaging
2D-SRI: Two-Dimensional Strain Rate Imaging
4CH: Four-chamber view
5CH: Five-Chamber view
A wave: Late diastolic filling wave
AV plane: Atrio-ventricular plane
AVC: Aortic Valve Closure
AVO: Aortic Valve Opening
avS: Average Strain
avSRA: Average SRA
avSRE: Average SRE
avSSR: Average SSR
BSA: Body Surface Area
BW: Body Weight
C.I.: Confidence Intervals
CO: Cardiac Output
CP: Constrictive Pericarditis
CRI: Constant Rate Infusion
LIST OF ABBREVIATIONS (Continued)
CRT: Cardiac Resynchronization Therapy
CV: Coefficient of Variability
CW: Continuous Wave Doppler
DCM: Dilated Cardiomyopathy
DI: Deformation Imaging
DICOM: Digital Imaging and Communications in Medicine
DP: Doberman pinscher
E wave: Early diastolic filling wave
E’: TVI Early diastolic filling wave
EDV: End Diastolic Volume
EF: Ejection Fraction
ESV: End Systolic Volume
FA: Femoral Artery
fe= Emitted frequency
fr= Reflected frequency
fps: Frames Per Second
FS: Fractional Shortening
HCM: Hypertrophic Cardiomyopathy
HR: Heart Rate
IV: Intravenous injection
LIST OF ABBREVIATIONS (Continued)
IVC: Isovolumetric Contraction
IVR: Isovolumetric Relaxation
KS test: Kolmogorov-Smirnov test
LA: Left Atrium
LiDCO©: Lithium Dilution Cardiac Output
LV: Left Ventricle
LVDd: Left Ventricular end-diastolic Diameter
LVEDP: Left Ventricular end-diastolic Pressure
LVDs: Left Ventricular end-systolic Diameter
LVOT: Left Ventricular Outflow Tract
LVSP: Left Ventricular end-systolic Pressure
MRI: Magnetic Resonance Imaging
MVC: Mitral Valve Closure
MVG: Myocardial Velocity Gradient
MVO: Mitral Valve opening
NIBP: Non Invasive Blood Pressure
PA: Pulmonary Artery
PCWP: Pulmonary Capillary Wedge Pressure
PSS: Post Systolic Shortening
PVR: Pulmonary Vascular Resistance
LIST OF ABBREVIATIONS (Continued)
PW: Pulsed Wave Doppler
RA: Right Atrium
RCM: Restrictive Cardiomyopathy
R&R analysis: Repeatability and Reproducibility analysis
ROI: Region of Interest
RR Interval: R wave to R wave Interval
RV: Right Ventricle
S: Strain
SC: Subcutaneous injection
SD: Standard Deviation
SSR: peak Systolic Strain Rate
SR: Strain Rate
SRA: Strain Rate late diastolic wave
SRE: Strain Rate early diastolic wave
SRI: Strain Rate Imaging
SV: Stroke Volume
SVR: Systemic Vascular Resistance
TD: Thermodilution
TMF: Trans Mitral Flow
TSI: Tissue Synchronization Imaging
LIST OF ABBREVIATIONS (Continued)
TT: Tissue Tracking
TVI: Tissue Velocity Imaging
Introduction
Assessment of regional and global myocardial function, during both systole
and diastole, is essential for the evaluation of patients suspect for heart
disease. Such evaluations are important in the diagnosis and management of
patients with most forms of heart disease including coronary atherosclerosis,
myocardial and valvular heart diseases, and most of the common congenital
cardiac disorders.
With the introduction of cardiac ultrasonography into
routine clinical practice, such evaluations can, for the first time, be
accomplished non-invasively and at low cost without the need for sedation or
anesthesia. Since the initial introduction of echocardiography to clinical
practice, a variety of different imaging techniques, including M-Mode
echocardiography, real-time two-dimensional (2D) imaging, and blood velocity
(Doppler) imaging has been developed to improve the assessment of cardiac
function. While many of these techniques have been extensively studied and
embraced by both medical and veterinary cardiologists, other methodologies
have not enjoyed such widespread acceptance despite some apparent
advantages over traditional imaging. Tissue velocity imaging (TVI), although
better established in the human cardiology, is not yet widely available in
veterinary practice; and deformation imaging (DI), also referred to as strain
rate imaging (SRI), has received even less attention from the veterinary
medical community. Deformation imaging was developed to display and
2
quantify the phenomenon of physical deformation of an object, in particular the
changes in shape of myocardium. This term is used to differentiate this
modality from other motion imaging techniques, such as spectral Doppler,
color-Doppler and tissue velocity imaging (TVI). The overall objective of the
research undertaken in this thesis was to investigate the accuracy and utility of
Doppler-derived deformation imaging to evaluate myocardial function in dogs.
3
Literature review
Motion versus deformation imaging
Of the various imaging modalities that are currently available for the study of
myocardial dynamics, an important distinction exists between motion imaging
and deformation imaging. Both techniques can be based on TVI technology,
which is able to estimate and display the velocity of a target, relative to the
imaging probe. In motion imaging, the relative velocities of the interrogated
object are displayed and post-processed without further calculations. With
Deformation Imaging, the relative velocities of neighboring areas within the
target are compared to each other, allowing detection and quantification of a
regional velocity gradient. The presence of a velocity gradient between
different areas of an object indicates that a change in shape (i.e. deformation)
is taking place. In turn, the term deformation implies that different interrogated
areas are characterized by a different velocity relative to a common point.
Review of M-mode and 2D-based echocardiographic techniques used to
evaluate ventricular function
The non-invasive study of left and right ventricular function is based on a
series of assumptions that allow the adoption of simplified models, but, of
course, decrease the reliability of our estimates. Among the different
techniques, the most commonly adopted parameter of systolic performance is
4
Fractional Shortening (FS%). It represents a linear measurement of ventricular
systolic function, defined as:
(1)
FS =
LVDd − LVDs
%
LVDd
Where LVDd is the end-diastolic diameter
and LVDs is the end-systolic diameter of the left ventricle. The limitations of
this functional index are that it only estimates the contraction along the short
axis (i.e. radial and circumferential shortening) and only at a specific region of
the left ventricle (standard sampling is performed at the level of the chordae
tendineae). For example, it does not accurately reflect global chamber function
in the presence of dyskinesia or focal hypokinesia.
Another commonly used index of left ventricular function is the ejection fraction
(EF%). This parameter represents an attempt to investigate simultaneously
the behavior of the radial, circumferential and longitudinal components of the
left ventricle. It is calculated from estimates of left ventricular volume at endsystole and end-diastole. A variety of different algorithms can be used to
estimate the volume of the ventricle from echocardiographic images.(1) Among
them, the modified Simpson’s rule, also known as the method of the discs,
treats the LV as a stack of discs. In the formula, the LV volume is obtained by
summation of the volumes of a number (usually 20) of equal cylinders whose
areas are determined from estimated diameters of the LV from two orthogonal
long-axis views and whose lengths are 1/20 of the long axis of the LV. Another
commonly adopted method is the single plane area-length method. Usually
5
applied from the apical two-chamber view, it can also be applied to a right
parasternal long axis four chamber view. When this method is utilized, the
area and the length of the long axis of the left ventricle are measured, and the
length of the short axis of the left ventricle is calculated (estimated) from these
data, on the assumption that the left ventricle is a spheroid. Figure 1 shows a
comparison between the two methods.
Once the chamber volumes are estimated, the EF% can be calculated as
follows:
(2)
EF =
EDV − ESV
%
EDV
Where EDV represents the end-diastolic
volume and ESV represents the end-systolic volume of the left ventricle.
Among the major limitations of these methods are the already mentioned
geometric assumptions that might not be fulfilled in the presence of moderateto-severe
cardiac
diseases
associated
with
substantial
remodeling.
Furthermore these 2D-based techniques rely on the accurate detection of the
endocardial borders to obtain accurate estimates of the chamber diameters.
Alternatively, tissue velocity (TVI) and strain rate imaging (SRI) can provide
measurements of ventricular function that are independent of geometric
assumptions and which do not require endocardial tracing.
6
Tissue Velocity Imaging (Motion Imaging) and its natural evolution:
Deformation Imaging
Myocardial velocities can be estimated by ultrasound-based techniques
derived from the phenomenon of the Doppler shift as first described in the
nineteenth century by Christian Andreas Doppler (1803-1853). The Doppler
phenomenon can be simply described as the change in frequency (or
wavelength) of a wave that is perceived by an observer in relative motion
compared to the interrogated object. Spectral Doppler and tissue Velocity
Imaging are techniques based on the concept of Doppler shift. When
ultrasound waves are reflected by an object that is not moving compared to
the detector (probe), the reflected frequency (fr) equals the emitted frequency
(fe) as shown in Figure 2. In contrast, if the object is in relative motion
compared to the probe (Figure 3), the reflected frequency differs compared to
the emitted frequency and the difference between these two quantities is the
Doppler shift (Df), as predicted by the Doppler equation:
(3)
Df =
2 × f e × v × cos α
c
Where fe is the transmitted frequency, v is the relative velocity of the object
compared to the probe, α is the angle between the ultrasound beam and the
trajectory of the object, and c is the speed of sound in the medium (the
average speed of sound in body tissues is 1540 m/s).
7
Equation (4) shows the Doppler equation solved for v:
(4)
V=
Df × c
2 × f e × cos α
Applying this equation, the velocity of an object, relative to the probe, can be
estimated by means of Doppler ultrasonography. The Doppler principle has
been extensively applied to detect, quantify and visualize the motion of blood
through normal and pathologic cardiovascular structures.
Approximately 20 years ago Isaaz et al.(2) published the first report on the use
of pulsed and color Doppler for the measurement of velocities of solid
structures. By inverting the filter settings for blood flow Doppler, so that the
low-velocity, high-amplitude signals from solid structures were now selectively
displayed instead of the high-velocity, low-amplitude signals originating from
the red blood cells (Figure 4), it was possible to detect and visualize direction
and velocity of motion of the myocardium. Over the last decade, tissue velocity
imaging modalities have been improved and optimized, so that velocities
calculated from a desired region of interest (ROI) of the myocardium can be
displayed either as spectral Doppler graphs (velocity versus time) or as colorcoded velocity maps overlaying the real time two dimensional images. As
importantly, the validity of TVI has been established in vitro and in vivo, by
comparison with sonomicrometry.(3) Moreover, TVI has been shown to provide
clinically useful measures of systolic and diastolic function, providing a more
8
complete non-invasive evaluation than can be achieved by the isolated
application of more traditional echocardiographic imaging techniques. TVI has
certain desirable attributes that make it particularly useful in clinical practice. In
contrast to traditional grey-scale imaging, which is reliant on the amplitude of
reflected waves to construct an interpretable image, TVI relies on the detection
of Doppler-shifts, permitting the acquisition of useful data from areas of the
myocardium that may not provide satisfactory grey-scale information.(4) For the
study of systolic function, TVI has certain advantages over other techniques,
such as two-dimensional imaging, because it is not reliant on boundary
detection. For the study of diastolic function, TVI provides functional measures
that are less flow-dependent than blood flow-derived Doppler indices.(5,6,7) In
addition, the time resolution of myocardial velocity events is superior to blood
velocity events providing an important advantage for evaluating diastolic
function. In some instances, TVI data may complement blood velocity data, as
demonstrated by the calculation of the E/E’ ratio that shows a high correlation
with the mean left atrial pressure.(8) This ratio allows a quick and non-invasive
estimate of the filling pressures of the left heart that is particularly useful in the
diagnosis of left sided congestive heart failure.
There are important limitations inherent to TVI techniques, which must be
appreciated. Inasmuch as TVI detects relative velocity, that is motion of the
ROI with respect to the probe, it is subject to certain unavoidable errors.
Tissue velocities measured by TVI can be influenced by translational motion,
9
either the result of the movement of the heart during the cardiac cycle or the
result of voluntary or respiratory movements. These translational velocities will
add or subtract to the inherent velocity of the myocardium, resulting in falsely
elevated or reduced myocardial velocities, or in an exaggerated beat-to-beat
variability. Another limitation of TVI is its inability to differentiate between
“tethering”, which represents passive motion of a portion of myocardium that is
pulled or pushed by adjacent muscle, and active motion. This limitation is
particularly significant in the evaluation of focal dyskinesia or hypocontractile
states. In order to overcome these limitations, researchers at the Norwegian
University of Science and Technology have worked to develop an imaging
technique based on a different concept. Instead of detecting and visualizing
relative velocities, these investigators endeavored to develop a technology to
detect changes in shape, or deformation; hence the term deformation imaging
(DI). Soon after the introduction of TVI, Doppler M-mode and color Doppler
studies revealed the existence of differences in velocities across the thickness
of the left ventricular wall.(9) The systematic evaluation of these myocardial
velocity gradients (MVG), eventually led to the idea of studying myocardial
deformation by manipulating data obtained from TVI studies. Subsequent
technical improvements allowed the quantification of myocardial deformation,
at first as an off-line elaboration of raw TVI data with the use of specific
software, and more recently in real time as part of the software capabilities of
advanced ultrasound units. In this regard, DI represents an evolution of TVI.
10
The DI application does not require specialized hardware but is based on the
analysis, by dedicated software, of raw TVI data. The application of this
specific software to TVI velocity data permits the nearly instantaneous
calculation of estimates of strain (S) and strain rate (SR). Useful DI studies
require high frame rate TVI imaging and high computational power.
The physical concepts of Strain and Strain Rate
In solid-state physics, linear S (one-dimensional S) is defined as the change in
length (Δl) of an object relative to its original length (l0), produced by the
application of a stress (force per unit cross-sectional area). This corresponds
to the so-called Lagrangian S, as defined by the Lagrangian equation (5).
(5)
ε=
l − lo Δl
=
lo
lo
Where ε is strain (S), l0 = baseline length, l is the
instantaneous length at the time of the measurement (Figure 5).
In practice, it is not possible to measure l0 as “unstressed” length, because
some degree of “stress” is constantly applied to myocardial fibers during the
whole cardiac cycle. Hence the concept of “unstressed-based” S is not
applicable to a clinical setting. In the beating heart l0 is approximated by the
end-diastolic length, so that deformation is detected from end diastole to end
systole. S is a dimensionless ratio, and is most commonly expressed in
percent (S%). Importantly, S can be measured in all three spatial dimensions
11
(normal S) and, by definition, positive S is lengthening or stretching. A value of
S of 10% implies a lengthening of 10% in relation to baseline length. Negative
S is shortening or compression. The Lagrangian formula only describes S in
one dimension. In two dimensions, S has four components: two normal strains
and two shear strains (Figure 6). Shear strain is defined as relative
displacement of one object border with respect to the object’s parallel border.
In three dimensions S is composed of nine components: three normal strains
and six shear strains. If the object is incompressible, as is the case with the
myocardium, all the components of S occur simultaneously. An object is
defined incompressible if the mass conserves its volume; hence it is said to
follow the law of conservation of volume. Based on this law we would expect
to see a balance between the different components of S: strain in one direction
has to be balanced by inverse S in other normal directions. In this case the
sum of S in the three dimensions is zero under the assumption that no shear S
occurs:
(6)
εx + ε y + ε z = 0
Where ε is strain, and x, y and z are the three
orthogonal Cartesian coordinates.
The concept of myocardial strain, to study myocardial deformation, was
introduced in the early seventies.(11) Instead of adopting the three orthogonal
Cartesian coordinate system (x, y, and z) it is more practical to divide the heart
into longitudinal (l), radial (r) and circumferential (c) axes as shown in Figure 6.
12
These three coordinates are perpendicular to each other; hence the same
geometric relationships apply. It must be recognized that S alone is not able to
provide a complete description of the pattern of deformation of an object.
Inasmuch as deformation is a dynamic event, two objects may undergo the
same extent of deformation at two different rates. By measuring the rate at
which S occurs, this time dependent dimension of deformation can be
quantified. Thus, strain rate (SR) is a measure of the change of S per time
unit:
(7)
SR =
Δε
Δt
Where Δε indicates change in S, and Δt represents
change in time.
The unit of SR is cm/s/cm or s-1. SR has the same direction as S, i.e. negative
SR is associated with shortening, positive SR with lengthening.
As already mentioned, Fleming et al. introduced the concept of Myocardial
Velocity Gradient (MVG) in 1994.(9) It is defined as the slope of the linear
regression line of myocardial velocities along the M-mode line perpendicular to
the ventricular wall:
(8)
MVG =
Vendoc. − Vepic.
W
=
ΔV
W
Where W = wall thickness, Vendoc. =
endocardial velocity, Vepic. = epicardial velocity.
Inasmuch as SR represents the rate of change in wall thickness, it follows that:
13
(9)
SR =
ΔW Wd ΔW Δt ΔV
≈
=
Δt
W
W
Hence MVG is an estimate of SR, where strain per time unit is an
approximation of velocity per length unit. The reason why it represents an
approximation will be described in the next paragraph.
Strain and Strain Rate by ultrasonography: Strain Rate Imaging (SRI)
Strain and Strain Rate can be estimated non-invasively by the manipulation of
TVI data (Figure 8). It is important to note that in TVI myocardial baseline
“object” length (l0) is not calculated (it is an undefined entity). Instead S is
calculated “by comparing instantaneous length to the length shortly (ideally
infinitesimally) earlier”.(12) In fact, for small (<0.1%) instantaneous changes in
length, SR can be calculated from the spatial velocity gradient as shown in
equations (9) and (10). Using a TVI unit (hardware) able to achieve high frame
rates, it is possible to record the velocity of two points (a and b) at a known
distance (d) in the direction of the (one) ultrasound beam (Figure 8).
(10)
SR ~
=
Va − Vb
d
Where Va-Vb represents the difference of instantaneous
myocardial velocities of points a and b, and d represents the distance between
these two points, the “offset”, often referred to as Strain Length (SL).
To obtain an accurate impression of myocardial deformation, SR is routinely
calculated along the radial (r), circumferential (c), and longitudinal (l) axes as
14
shown in Figures 9 and 10. By summing the SR of all frames times the
sampling interval from each frame, S in each pixel can be calculated
(estimated), as shown in equation (11).
(11)
⎛ Δε
⎞
∑ (SR × Δt ) = ∑ ⎜⎝ Δt × Δt ⎟⎠ = ∑ Δε = ε
Where ε is S
In other words, if we assume that the interval (Δt) between consecutive frames
is infinitesimally short then we can obtain S by summing (i.e. integrating)
instantaneous SR values, from a starting time point (t0) to an end time point (t).
This physical entity is called Eulerian Strain or Natural Strain, and it provides
an estimate of instantaneous deformation, as shown, in integral notation, in
equation (12).
t
(12)
ε n = ∫ SRdt
Where εn represents Natural S or Eulerian S
to
Current technology allows the extraction of both S and SR from stored color
TVI cine-loops. Most units also allow on-line acquisition of real-time S and SR
color-coded images. Regional expansion corresponds to positive S and SR,
and is coded in blue. It corresponds with the systolic thickening that occurs
along the radial axis and the diastolic lengthening visible along the longitudinal
axis. Negative S and SR are coded in yellow-red, and denote regional
compression. It corresponds with diastolic thinning in the radial axis and
systolic shortening in the longitudinal axis. Scanning along the circumferential
15
axis shows systolic shortening and diastolic lengthening. Off-line postprocessing is available that allows the generation of S and SR time-velocity
plots, enabling the operator to position several ROIs in the desired myocardial
locations. Another imaging modality that is currently available in TVI and SRI
is curved M-mode color display. It is a reconstructed color M mode recording
along a manually traced line. It provides an easily interpretable visual display
of segmental asynchrony between different segments.
Technical limitations of SRI
Analogous to other Doppler-derived measurements, SRI data are only
obtainable along the path of the ultrasound beam. As a result, only
deformation in the direction of the ultrasound beam is recorded, while in reality
deformation occurs in all directions. It must also be appreciated that errors
resulting from angulation of the interrogating beam relative to the motion of the
object being interrogated are inherent to the method. The relationship between
the Doppler shift and the angle (α) between the ultrasound beam and the
direction of motion of the object is shown in equation (3), wherein the
estimated velocity is inversely proportional to cos α. Figure 11 shows the
behavior of the function y = cos x . When x =
π
2
(a 90 degrees angle), cos x = 0 ,
the detected Doppler shift becomes zero and the movement of the object
cannot be studied. Not only is SRI affected by the angle dependency typical of
16
the Doppler technique, but also the problem is accentuated by the spatial
components comprising S. Inasmuch as the myocardium is an incompressible
material, for every component of S there is a perpendicular component that
has opposite direction, due to the law of conservation of volume. Hence, with
suboptimal alignment, it is possible to sample both components. In this way,
the perpendicular (or transverse) component will subtract to the original
component with the reduction of the value of S proportional to the cosine of
the transverse angle β. As predicted by the Doppler shift equation (3) TVI
velocity becomes zero at an angle of 90 degrees. However SRI is more
sensitive to misalignment, giving values of zero at angles close to 45 degrees
and negative values for angles wider then 45 degrees. Another important
limitation of the Doppler method in general is a consequence of aliasing of the
signal. When the velocity of an interrogated object exceeds the Nyquist limit, it
appears to be transposed to the other end of the velocity scale (opposite
direction). In SRI imaging two velocities are simultaneously sampled. If both
velocities exceed the Nyquist limit, the velocity difference remains unchanged
and no aliasing of SR occurs. However if only one velocity exceeds the
Nyquist limit, the resultant SR will be false. In a practical setting noise is an
important cause of aliasing, but in most instances both velocities are affected.
SRI is seriously affected by noise because the SR is calculated as the
difference between two velocities, and the total error is the sum of the errors in
estimating each of these two velocities. Thus, the signal-to-noise ratio is less
17
favorable than in TVI, where velocities are calculated separately. One method
of decreasing noise is to increase offset distance between the sampling points.
Because noise is a random phenomenon, increasing the offset will increase
the velocity difference (Va-Vb) and the distance (d) as expressed in equation
(8). This will increase the signal-to-noise ratio, giving “temporal smoothing” of
the SR tracings, albeit at the expense of reducing spatial resolution. Spatial
resolution describes how closely two reflectors, or scattering regions, can be
to one another and remain identifiable as different reflectors. As a general rule,
SRI has a lower spatial resolution compared to TVI, because of the need for
sampling two velocities instead of one. A 5- to 10-mm Strain Length appears
to represent a reasonable trade-off between requirements of spatial resolution
and low levels of noise. S appears less prone to noise because integration of
SR tends to minimize the influence of noise.(13) The reliability of SRI is also
highly dependent on the frame rate adopted to generate the images. High
frame rate minimizes noise by increasing the number of points used to
generate the time-velocity plot. Thus, the maximal obtainable frame rate
should be adopted for image acquisition.
Technical strengths of SRI
In contrast to TVI, wherein relative velocity is calculated, SRI uses calculations
of the difference between two velocities. Hence S and SR measure local
myocardial lengthening and shortening rather than myocardial translation with
18
respect to the transducer (Figure 12). As shown by inspection of equations
(10) and (13), translational velocities (Vt) annul each other. As a result, SRI is
not affected by translational motion created by respiration, motion of the heart,
motions of the patient, or involuntary motion of the transducer.
(13)
SR =
Va ± Vt − (Vb ± Vt ) Va − Vb
=
d
d
Where: ± Vt − (± Vt ) = 0
Translational movement of the ROI is also important when a portion of
myocardium is non-actively contracting. In this scenario the akinetic area is
passively dragged by neighboring myocardium. This phenomenon is referred
to as “tethering”. For the same reason that SRI is not sensitive to translational
motion, it is also resistant to tethering. In fact, the velocity due to tethering can
be considered as a translational velocity (Vt). Hence SRI is not affected by
tethering and is able to reliably identify areas of myocardium that are not
actively participating to the systolic or diastolic function of the heart, but that
are being passively dragged by normal neighboring myocardium.
Current and future clinical applications of SRI
Myocardial S and SR imaging has been validated in vitro using compressed
gelatin phantoms(14,15) and in vivo using sonomicrometry(16), pressure-volume
loops(17), and MRI(18) as gold standards. It has been shown that S correlates
with fiber shortening and SR correlates with the velocity of fiber shortening,
which is a measure of intrinsic myocardial contractility. The technical
19
superiority of SRI over TVI for tracking local systolic and diastolic function has
been demonstrated by several independent investigators, mainly because SRI
is resistant to translation and tethering.(19,20,21,22) Cardiac contraction is the
product of the mechanical properties of a complex myocardial fiber
architecture in which three major fiber orientations have been identified:
longitudinal, radial and spiral. Primitive and inappropriate investigative models
suggested that systole is mainly the result of an inward squeezing motion and
this false impression has, unfortunately, persisted for decades. In fact, the
longitudinal motion of the heart appears to be the dominating mechanical
event that drives the cardiac pump function.(23,24) The concept that the heart
functions as a double pump, with the atrioventricular (AV) plane acting as a
piston, dates back to Leonardo da Vinci (1452-1519). There is a wealth of
theoretical and experimental evidence supporting the assertion that an
important component of the systolic and diastolic excursion of the heart is in
the longitudinal plane. In this model the cardiac apex and the outer ventricular
contour remain relatively stationary and the AV plane shows a wide range of
motion. If the heart pumped only by inward squeezing, significantly reducing
its outer contour, this would be extremely unfavorable from the energetic point
of view. In this scenario, the myocardium would have to win the inertia of the
surrounding tissues both during systole and during diastole. Experimental
studies, performed as early as 1932 suggested that the heart worked mainly
by movement of the heart base (AV plane) toward the apex in systole, and
20
away from the apex in diastole, while the apex itself remained almost
stationary and the outer contour of the heart relatively constant.(23) This
hypothesis is referred to as the principle of the reciprocating pump, and was
subsequently confirmed by Hoffman and Ritmann in CT studies in dogs in
1985, in which they demonstrated a stationary apex, relatively constant outer
contour and motion of the AV plane.(24) Figure 13 shows the motion of the AV
plane towards the apex during systole: minimal reduction of the outer contour
associated with substantial translation of the AV plane, accounting for a
significant percentage of the generated stroke volume (SV). The opposite
occurs during diastole. The important excursion of the AV plane is easily
appreciable with real time 2D imaging from the apical view. In 1967, Harvey
Feigenbaum first suggested that the estimation of the excursion of the AV
plane, performed by M-mode imaging, provided an index of left ventricular
systolic function.(25) With modern echocardiographs, the excursion of the mitral
valve annulus (located on the AV plane) can be studied by TVI and several
independent groups have validated the use of this technique for the
assessment of global ventricular function.(26,27) In one notable clinical study,
the peak mitral annular descent velocity from the apical 4-chamber view
correlated most closely, as an individual index, with the radionuclide
ventriculographic LV ejection fraction (r=0.85).(28) Such studies confirm that the
longitudinal motion of the myocardium and cardiac structures most closely
reflects the pump function of the heart as an organ.
21
Based
on
its
ability
to
discriminate
between
passive
and
active
deformation/motion, SRI, is a particularly promising tool for the study of
myocardial dynamics. One of the most important differences between
myocardial velocities and myocardial velocity gradients (i.e. SR), refers to the
pattern of distribution of these physical entities along the ventricular wall.
Compared with the apex-to-base gradient of tissue velocities, longitudinal SR
is more homogeneously distributed.(29) In fact, the longitudinal velocities
decrease from base to apex.(30) However S and SR do not significantly change
along the ventricular wall.(31) This proves that the longitudinal tissue velocity
gradient is not the result of local differences in tissue velocities, as
demonstrated by the absence of a longitudinal SR gradient. Rather, it appears
to be a function of the greater motion of the whole heart at the base, compared
with the mid and apical regions. This is an example of how SRI might
represent a powerful tool to study the mechanical properties of the
myocardium, both in normal and abnormal hearts.
In 2001 Jamal et al. demonstrated that peak systolic SR, but not S, was
linearly related to global measures of LV contractility.(32) One of the most
important areas of application of SRI is the diagnosis and management of
ischemic heart disease. Such disorders are characterized by variable degrees
of systolic and diastolic impairment, which are most often focal in nature and
SRI appears particularly well suited to the study of such changes. Progressive
delay and reduced amplitude in longitudinal systolic shortening (S and SR)
22
occur early during myocardial ischemia. At the same time the physiologic
uniformity of longitudinal distribution of S and SR is lost.(33) Peak systolic S
and SR decrease and tend to shift towards end-systole or early diastole, i.e.
beyond aortic valve closure (AVC). This finding, called post-systolic shortening
(PSS) or post-systolic thickening, appears as a high and abnormal SR during
the isovolumetric relaxation period (IVR). It has been suggested that this
observation
represents
a
relatively
sensitive
marker
of
ischemia.(33)
Unfortunately, specificity is reportedly not high because similar changes can
be detected in a substantial population of normal patients. To increase the
specificity of this finding, Jamal et al. used the ratio of PSS to maximal
segmental deformation, and found that it was the best quantitative parameter
to identify stress-induced ischemia as an objective marker of ischemia during
dobutamine stress echocardiography in a porcine model.(32) Complete
transmural infarctions do not exhibit appreciable S or SR, either at rest or
during dobutamine stimulation. This aptly demonstrates the superiority of SRI
to TVI, as the former is being able to differentiate between passive and active
deformation.(34) SRI was also investigated in an experimental model of altered
myocardial
bioenergetics
in
which
myofibrillar
creatine
kinase
and
glyceraldehyde dehydrogenase were inhibited by iodoacetamide.(35) The SRI
findings were similar to acute ischemia and infarction. TVI and SRI have also
been investigated in patients affected by congenital and acquired nonischemic cardiomyopathies. Several studies showed that SRI has the potential
23
of detecting early changes occurring in the subclinical phase of the disease
(the so-called “occult” phase). In a transgenic murine model of human
hypertrophic cardiomyopathy (HCM), TVI was able to detect early diastolic and
systolic abnormality, before an appreciable hypertrophy was visible.(36) The
same author obtained equivalent results in a group of clinical patients.(37) Early
evidence of impaired systolic and diastolic function has been demonstrated by
means of SRI in patients affected by Friedreich’s ataxia with normal LV
volumes and preserved EF.(38) Reduced peak systolic radial SR was observed
in patients with normal EF, undergoing anthracycline therapy.(39) Therefore
SRI might represent a sensitive tool in the early recognition of myocardial
toxicity in veterinary patients treated with doxorubicin. SRI might also prove a
valuable aid in differentiating restrictive cardiomyopathy (RCM) versus
constrictive pericarditis (CP). Systolic SR during ventricular ejection and rapid
(early) diastolic filling were lower in RCM compared with CP.(40) In particular,
SR during IVR was positive in RCM patients, and negative in patients affected
by CP, while early diastolic SR was low in RCM compared with CP and normal
controls. Furthermore, SRI appears able to differentiate between primary and
secondary concentric hypertrophy. In fact, systolic myocardial SR values were
significantly lower in patients affected by HCM, compared with athletes,
hypertensive patients and normal controls.(41)
24
Normal SRI tracings
Figure 14 shows a normal longitudinal S tracing (solid line) and an example of
reduced S (dashed line). Figure 15 shows a normal longitudinal SR tracing
(solid line) and an example of SR associated with ischemia (dashed line)
showing reduced amplitude systolic SR and PSS during IVR. E and A waves
represent early and late diastolic filling. In normal human patients longitudinal
S and SR values are lower in the left ventricle (LV) compared to the right
ventricle (RV) or compared to radial deformations. In fact radial SR is
approximately twofold higher than longitudinal SR.(42) A transmural gradient
exists in normal myocardium, with higher SR occurring in the subendocardium.
Doppler-derived DI in dogs
Doppler-derived DI has been evaluated by a French study group in 2006.(43)
The study consisted in an assessment of the intra-operator variability
conducted on 6 healthy Beagle dogs, followed by the assessment of normal
values in a population of 30 adult healthy dogs from different breeds. The
conclusions of the authors were that Doppler-derived DI provides reasonably
acceptable intra-operator variability. G. Wess presented an abstract at the 16th
ECVIM-CA Congress (2006) where he evaluated Doppler-derived DI in a
population of 42 healthy adult Doberman pinscher dogs.(44) Normal ranges for
longitudinal Strain and Strain Rate were reported in this abstract.
25
Figure 1: Comparison between single- and bi-plane methods for the estimation
of left ventricular volumes
26
fe
V=0
fr=fe
Figure 2: The Doppler principle: stationary target
fe
V°0
fr ° fe
Figure 3: The Doppler principle: target in relative motion
α
27
Tissue
Amplitude
Blood
Velocity
Figure 4: Velocity/amplitude relationship of backscatters from blood and
tissues
l0
l
Figure 5: Example of one-dimensional S
28
y
y
x
x
x
x
y
y
y
y
x
x
Figure 6: The four components of bidimensional S
y
r
c
l
x
z
Figure 7: Relationships between Cartesian axes and cardiac longitudinal (L),
radial (R) and circumferential (C) axes
29
Va
d
Vb
Figure 8: Schematic of Doppler-derived Strain Rate Imaging
c
r
Figure 9: Short axis view showing radial (R) and circumferential (C) axes
30
l
l
Figure 10: Long axis view showing the longitudinal (L) axis
Figure 11: The cosine function
31
TVI
SRI
Va
Vt
d
Vb
Figure 12: Comparison between TVI and SRI with respect to the effects of
translation of the whole heart
Figure 13: Schematic representation of the reciprocating pump principle, with
emphasis on the systolic phase
32
Strain(%)
0
%
Figure 14: Normal longitudinal S (solid line) and reduced S (dashed line) as a
typical example of pathologic S
Strain
Rate
E
A
0 s-1
AVO
S
IVR
AVC
Figure 15: Normal SR (solid line) and pathologic SR (dashed line) as
commonly found with ischemia
33
Aims of the Project
Deformation Imaging might represent an interesting addition to the currently
available techniques for the diagnosis and management of both congenital
and acquired cardiovascular diseases in veterinary patients. At the same time
it will surely enrich our knowledge on the dynamics of diastolic and systolic
function in normal patients. The techniques described in the preceding
paragraphs are currently available at our Institution and our group has already
gained a decent level of basic practical experience.
This Master’s Degree thesis is structured into two studies:
Study 1 - Invasive indirect validation of Doppler-derived Strain Rate Imaging in
anesthetized healthy adult dogs: will correlate DI with invasive techniques in
the assessment of global systolic function in normal anesthetized dogs
undergoing pharmacologic manipulations.
Study 2 - Deformation Imaging in normal adult dogs: will provide normal DI
values in normal non-sedated dogs and will also allow the analysis of the intraoperator (repeatability) and inter-operator (reproducibility) variability (R&R
analysis).
34
Materials and methods
Study 1 - Invasive indirect validation of Doppler-derived Strain Rate imaging in
anesthetized healthy adult dogs
There is clinical and experimental evidence in the medical literature that S and
SR imaging accurately reflect the changes in ventricular function induced by
pharmacologic interventions.(45) The aim of this study was to evaluate, in a
canine model, the changes in Doppler-derived DI data during manipulations of
the inotropic state and loading conditions.
Hypothesis
We hypothesized that S and SR imaging accurately reflect, in a loadindependent manner, the changes in the LV systolic function induced by the
pharmacologic manipulation of intrinsic contractility, as documented by
changes in the maximum rate of pressure rise (+dP/dtmax) obtained by
analyzing hi-fidelity LV pressure tracings by means of cardiac catheterization.
Inclusion criteria and study protocol
Five adult clinically healthy dogs were enrolled in the study. The study protocol
was approved by the Institutional Animal Care and Use Committee in
compliance with the Public Health Service (PHS) Policy on the Humane Care
and Use of Laboratory Animals, the US Department of Agriculture’s (USDA)
35
Animal Welfare Act & Regulations (9CFR Chapter 1, 2.31), and the United
States Government Principles for the Utilization and Care of Vertebrate
Animals Used in Research, Teaching and Testing. All dogs underwent a
complete physical exam, a non-invasive blood pressure measurement, and a
ten-lead electrocardiogram. A device based on the oscillometric method was
utilized to obtain non-invasive blood pressure measurements (Cardell®, Sharn
Veterinary Inc. Tampa, FL). Systemic hypertension was defined as a systolic
pressure higher than 170 mmHg.(46,47) A total of five consecutive NIBP
measurements were recorded, the readings with the highest and lowest mean
blood pressure values were discarded, and the average of the three remaining
measurements was used. An echocardiogram was acquired before sedation,
based on the following protocol (Tables 1 and 2, Figure 16). A Vivid 7
Dimension Cardiovascular Ultrasound System (GE Healthcare, Milwaukee,
WI, USA) with Doppler-derived SRI software was used for the acquisition of
the echocardiographic data. A 3Mhz probe with octave strain imaging
application was used for image acquisition. For traditional imaging modalities,
a set of 7 consecutive beats was digitally stored and measurements were
made as the average of 3 good quality beats. For the acquisition of DI data,
two sets of 7 consecutive beats were digitally stored and measurements were
made as the average of 5 good quality beats. Dedicated left apical long axis
views of the left ventricular free wall and interventricular septum were obtained
in order to optimize the alignment with the longitudinal myocardial fibers and to
36
reduce the sector width, so to maximize the frame rate. With this technique, a
minimum frame rate of 220 fps was obtained in all the cineloops. DI data were
collected with the preceding RR interval. Table 1 shows the imaging views and
modalities
included
in
the
echocardiographic
acquisition
protocol.
Measurements with the traditional echocardiographic modalities were
performed
as
previously
described.(48,49)
The
off-line
analysis
(post-
processing) of the DI data was performed using the software embedded in the
ultrasound unit (Vivid 7 Dimension Cardiovascular Ultrasound System, GE
Healthcare, Milwaukee, WI, USA) and a dedicated workstation (EchoPAC
Dimension ’06, GE Healthcare, Milwaukee, WI, USA). Oval sample volumes of
5 mm x 10 mm with Strain Length of 10 mm were used as ROIs for obtaining
SR-time plots and S-time plots (Figure 16). Temporal gaussian smoothing was
set at 40 ms. The Cine Compound function was disabled. Off-line analysis of
DI data was performed following the protocol indicated in Table 2. Figure 16
shows positions and labeling of the 6 ROIs and displays the two narrow-sector
dedicated views of the interventricular septum and left ventricular free wall,
used to generate the DI plots.
All the dogs were found to be free of evidence of cardiac disease and were
defined normal. The dogs were pre-medicated with Diazepam (0.1 mg/kg IV)
and Morphine (1 mg/kg SC). General anesthesia was induced with Thiopental
(10 mg/kg IV) and maintained with 2% isoflurane in 100% oxygen. Meloxicam
(0.2 mg/kg SC) was used to control surgical pain. The dogs were placed in left
37
lateral recumbency on an echocardiographic table. The right jugular vein was
accessed percutaneously, using a Seldinger technique, and a 7Fr introducerdilator set (Check-Flo® Performer® Introducer set, Cook Inc. Bloomington, IN)
was inserted. A 6 Fr Swan-Ganz catheter (Arrow International, Inc., Reading,
PA, USA) was positioned in the main pulmonary artery to record right atrial
(RA) pressure, pulmonary artery (PA) pressure, pulmonary capillary wedge
pressure (PCWP) and to allow the estimation of cardiac output (CO) by
thermodilution technique. The right carotid artery was surgically exposed,
using the Sones technique, and a 6 Fr introducer-dilator set inserted. The left
ventricle was catheterized with a 5 Fr high-fidelity microtransducer pressure
catheter (Mikro-Tip®, Millar Instruments, Inc., Houston, TX, USA) connected
with a physiologic monitor (Biopac® MP100WSW, Biopac System Inc., Goleta,
CA, USA) to record left ventricular pressure tracings and derive +dP/dtmax and
-dP/dtmax. Simultaneous quantification of cardiac output by lithium-dilution
technique was also performed using a LiDCO® Plus Hemodynamic Monitor
(LiDCO Ltd., Cambridge, UK). Correct positioning of the catheters was
confirmed by fluoroscopy. An arterial catheter was inserted in the dorsal pedal
artery, to record a peripheral arterial pressure tracing. A baseline set of
echocardiographic measurements (standard echocardiography and DI), and
invasive measurements were recorded at steady state, defined as stabilization
of +dP/dtmax, -dP/dtmax, HR and systemic blood pressure. The invasive
parameters acquired or calculated during each phase of the study were the
38
following: cardiac output by thermodilution and Lithium-dilution, stroke volume
(SV), left ventricular systolic pressure (LVSP), left ventricular end-diastolic
pressure (LVEDP), +dP/dtmax, -dP/dtmax, mean right atrial pressure, pulmonary
capillary wedge pressure (PCWP), systemic vascular resistance (SVR),
pulmonary vascular resistance (PVR), femoral artery mean pressure,
pulmonary artery mean pressure. After the initial set of measurements was
obtained, a sequence of three pharmacologic manipulations was performed.
At first a constant rate infusion (CRI) of Dobutamine (10 µg/kg/min) was
initiated, and a set of invasive and echocardiographic data was collected at
steady state.(50) After discontinuation of the infusion, a wash-out period of at
least 20 minutes allowed the return of the +dP/dtmax, HR, and mean blood
pressure to baseline. The infusion of dobutamine was performed in order to
achieve a positive inotropic effect. The second manipulation was performed
starting a CRI of Nitroprusside (3.0 µg/kg/min), with the goal of reducing the
loading conditions of the left ventricle, by means of inducing balanced
vasodilation.(51) At steady state the same set of measurements was repeated.
A wash-out period of at least 20 minutes allowed the return to baseline and a
final pharmacologic manipulation was performed by means of administering a
bolus of the plasma volume expander hetastarch (10 ml/kg/15 min), with the
goal of increasing preload.(52) A final set of measurements was obtained at
steady state. At the end of the study the right carotid artery was ligated and
39
the soft tissues sutured in the usual fashion. The dogs recovered uneventfully
from anesthesia.
Statistical methods
Most data sets were normally distributed, as confirmed by the result of a
Kolmogorov-Smirnov test (KS test) with Dallal-Wilkonson-Lilliefor p value, a
D'Agostino & Pearson omnibus normality test, and a Shapiro-Wilk normality
test. For the normally distributed data sets, repeated measures ANOVA with
Bonferroni corrections was used to compare the hemodynamic parameters
recorded at baseline (under anesthesia) with the parameters recorded during
the three pharmacologic manipulations. When the assumption of normality
was not fulfilled, and data transformations proved ineffective, a non-parametric
repeated-measures analysis (Friedman test with Dunn’s comparisons between
groups) was performed. Simple linear regression analysis was used to study
the correlation between peak systolic SR and several invasive parameters,
when the assumptions of normality, equal variance and linearity were met.
Statistical significance was set at p<0.05.
Study 2 – Deformation Imaging in normal adult dogs
The aim of this study was to investigate the feasibility of DI in non-sedated
dogs and to provide DI values in normal dogs divided into three groups based
on body weight. A fourth group of Doberman Pinchers (DP) was also included.
40
The reason for the segregation of dogs belonging to this specific breed to a
separate study group is the high prevalence of dilated cardiomyopathy (DCM),
and the high number of DP dogs with borderline echocardiographic indices of
systolic function. DCM in DP is characterized by a long period during which
echocardiography is unable to definitely diagnose the disease, the so-called
“occult phase”. In this phase arrhythmias may be present and a relatively high
risk of sudden cardiac death has been documented. The study was also
designed to allow the analysis of the intra-operator (repeatability) and interoperator (reproducibility) variability.
Hypotheses
With Study 2 we aimed at investigating the following hypotheses:
1. DI is an echocardiographic technique that is feasible, reproducible and
repeatable, in a veterinary clinical setting.
2. S and SR represent parameters of ventricular function that are
independent of body weight.
3. S and SR values are different in Doberman Pinchers compared to weightmatched non-Doberman controls.
4. S and SR values are uniform throughout the left ventricle in healthy adult
dogs (absence of a significant apical-basal gradient).
41
Inclusion criteria and study protocol
Clinically healthy adult dogs were enrolled on a voluntary basis and were
divided into 4 groups. Inclusion in the first three groups was dependent on
body weight (BW) (Group 1: BW < 15 kg, Group 2: BW 15 - 30 kg, Group 3:
BW > 30 kg). A fourth group enrolled Doberman pinschers. An informed
consent form signed by the owner was obtained in all dogs prior to the study.
The study protocol was approved by the Institutional Animal Care and Use
Committee in compliance with the Public Health Service (PHS) Policy on the
Humane Care and Use of Laboratory Animals, the US Department of
Agriculture’s (USDA) Animal Welfare Act & Regulations (9CFR Chapter 1,
2.31), and the United States Government Principles for the Utilization and
Care of Vertebrate Animals Used in Research, Teaching and Testing. Dogs
were considered eligible for the study if their age was between 12 months and
8 years, there was no history of cardiovascular disease, and were not on any
treatment that could have affected their cardiovascular function (i.e. thyroid
hormone supplementation, phenylpropanolamine, etc). All dogs underwent a
complete physical exam, a non-invasive blood pressure (NIBP) measurement,
and a ten-lead electrocardiogram. The normal blood pressure values for dogs
are breed-specific.(46) Where appropriate, breed-specific data were adopted in
the definition of normal systemic blood pressure.(46) In case data were not
available for the specific breed, systemic hypertension was defined as a
systolic pressure higher than 170 mmHg.(47) In dogs in which the results of
42
these
preliminary
studies
were
within
normal
limits,
a
complete
echocardiographic examination was performed, following the same protocol
used in study 1.
Estimation of the intra-operator variability
Four (about 40%) randomly selected dogs in each group underwent an
echocardiographic study twice by the same operator, with an interval of at
least 20 minutes between the two studies, in order to collect data on
repeatability in the phase of both image acquisition and post-processing.
Estimation of the inter-operator variability
Four (about 40%) randomly selected dogs in each study group were examined
by two operators in order to collect data to estimate an index of inter-operator
variability (reproducibility) in the phase of image acquisition and postprocessing of the DI data.
Statistical methods
Most data were normally distributed, as confirmed by the results of a
Kolmogorov-Smirnov test (KS test) with Dallal-Wilkonson-Lilliefor p value, a
D'Agostino & Pearson omnibus normality test, and a Shapiro-Wilk normality
test. For the normally distributed data sets, paired or unpaired t-Tests or Oneway ANOVA with Bonferroni correction were used to make comparisons
43
between various subsets of data. When the assumption of normality was not
fulfilled, and transformation of data proved to be ineffective, non-parametric
tests such as Mann-Whitney-Wilcoxon test or Friedman test with Dunn’s
comparisons between groups were performed. Simple linear regression
analysis was used to study potential correlations between DI data and other
parameters, when the assumptions of normality, equal variance and linearity
were met. Statistical significance was set at p<0.05. For the repeatability and
reproducibility analysis, Bland and Altman plots were utilized to compare the
intra- and inter-observer variability, which was quantified by using the bias
(average of the differences between the two studies) and its standard
deviation, from which the 95% limits of agreement were calculated.
44
Table 1: Imaging views and modalities utilized in the echocardiographic
acquisition protocol
R. parasternal
window
Short
axis
LV apex
Papillary muscle
Chordae
tendineae
MV
Ao root
Long
axis
4CH (4 chamber
view)
LVOT
L. parasternal
window
Apical
view
4 CH view
5 CH view
2D cineloop
TVI cineloop
2D cineloop
TVI cineloop
2D cineloop
TVI cineloop
M-mode
2D cineloop
TVI cineloop
M-mode
2D cineloop (Ao and LA)
M-mode (Ao and L auricle)
Color-Doppler PV
PW PV
Color-Doppler TV
2D cineloop
Color-Doppler MV
Color-Doppler TV
2D cineloop
Color LVOT
2D cineloop
Color-Doppler MV
PW of Trans-Mitral Flow
(TMF)
Flow propagation velocity
(FPV)
TVI cineloop
PW-TVI lateral MV annulus
PW-TVI medial MV annulus
Color-Doppler TV
PW Trans-Tricuspid Flow
(TTF)
PW-TVI lateral TV annulus
PW-TVI medial TV annulus
2D cineloop
Color-Doppler LVOT
PW Ao
45
Table 2: Imaging views and modalities utilized in the echocardiographic off-line
analysis
L. parasternal
window
Apical
view
4 CH view
Optimized S view free
wall
S tracings at locations 1-2-3
Optimized SR view
free wall
SR tracings at locations 1-2-3
Optimized S view IV
septum
S tracings at locations 4-5-6
Optimized SR view IV
septum
SR tracings at locations 4-5-6
6
5
4
3
2
1
Figure 16: Schematic representation of the apical 4-chamber view. Numbers 1
through 6 represent the positions of the six ROIs used during the off-line
analysis of DI data. The two “optimized” narrow-sector views are shown
46
Results
Study 1
The results of study 1 have been previously presented as an oral abstract at
the 25th Annual Forum of the American College of Veterinary Internal
Medicine.(53) Five dogs were enrolled in the invasive study. There were three
females and two males, with ages ranging between 2 and 7 years (mean 3.80,
median 2.90). The breeds included mixed breed hound (3 dogs) and mixed
breed Labrador retriever (2 dogs). Their BW ranged from 22.60 to 40.20
kilograms, with a mean BW of 28.88 kilograms and a median BW of 27.89
kilograms. Repeated measures ANOVA confirmed that the following
parameters were significantly changed during at least some of the
pharmacologic manipulations, compared to baseline under anesthesia (Table
3): RR interval (p<0.01, range 333.2 - 898.9 ms) (Figure 17), Fractional
shortening (2D-FS) (p<0.001, range 15 - 53%) (Figure 18), cardiac output
(P<0.0001, range 2.600 - 9.340 L/min) (Figure 19), stroke volume (p<0.001,
range 18.92 - 45.68 ml) (Figure 20), +dP/dtmax (p=0.015, range 953.7 - 3822
mmHg/s) (Figure 21), left ventricular end diastolic pressure (p<0.0001, range
0.210 - 16.43 mmHg) (Figure 22), Mean right atrial pressure (p<0.0004, range
-2.490 - 9.920 mmHg) (Figure 23), and Systemic Vascular Resistance
(p<0.0001, range 510.0 - 2652 dyne*sec/cm5) (Figure 24). The following
parameters
did
not
change
significantly
during
the
pharmacologic
47
manipulations, compared to baseline under anesthesia: Pulmonary Vascular
Resistance (p>0.05, range 256.4 - 1635 dyne*sec/cm5) (Figure 25), -dP/dtmax
(p>00.5, range -1872.0 - -984.20 mmHg/s) (Figure 26), mean femoral artery
pressure (p>0.05, range 51.77 - 101.00 mmHg) (Figure 27), and pulmonary
capillary wedge pressure (p>0.05, range 0.06 - 13.59 mmHg) (Figure 28). A
good correlation between thermodilution and LiDCO® techniques in estimating
cardiac output was observed (p=0.0002, Pearson r=0.993) (Figure 29). The
results of a repeated measure ANOVA revealed that the fractional shortening
(FS) significantly increased (p<0.001, mean difference 24.8%) (Figure 18)
during the dobutamine infusion and returned to baseline during nitroprusside
and hetastarch administration. FS was positively correlated with +dP/dtmax
(p=0.004, r squared=0.991 from a Pearson correlation test) (Figure 30). The
analyses of the DI data yielded the following results. A total of 750 ROIs were
included in the final analyses, 150 in each of the 5 dogs. Measurable DI
curves were obtained in 741 out of 750 ROIs (98.8%). The peak systolic strain
rate (SSR) values obtained from the 6 ROIs did not differ significantly from
each other (p=0.072 from a Kruskal-Wallis test) (Figure 31). The average SSR
(avSSR), calculated as the average value from the 6 ROIs, did not change
significantly between the baseline echocardiogram before anesthesia and the
steady state after induction of anesthesia (p>0.05 from a one-way analysis of
variance) (Figure 32). A good correlation was found between 2D-derived FS%
(2D-FS) and avSSR (p=0.040, r squared=0.922) (Figure 33). The avSSR
48
changed significantly during dobutamine infusion versus both baseline preanesthesia and baseline after induction of anesthesia (p<0.001 from a
repeated measure ANOVA) (Figure 32). The variable avSSR negatively
correlated with +dP/dtmax (p=0.021, r=0.979,) (Figure 34). Individual values of
SSR at the level of the basal portion of the free wall (ROI 1; Figure 35)
(p=0.011, r=0.989), mid portion of the free wall (ROI 2; Figure 36) (p=0.002,
r=0.997), apical portion of the free wall (ROI 3; Figure 37) (p=0.011, r=0.989),
mid portion of the septum (ROI 5; Figure 38) (p=0.018, r=0.981), and apical
portion of the septum (ROI 6; Figure 39) (p=0.049, r=0.951) were also
negatively
correlated
with
+dP/dtmax.
However,
statistically
significant
correlation was not reached between SSR obtained at the level of the basal
portion of the interventricular septum (ROI 4; Figure 40) and +dP/dtmax
(p=0.170, r=0.830).
49
Table 3: Summary table indicating the most important hemodynamic changes
obtained during the pharmacologic manipulations
Mean±SD
CO
Anesthesia
Dobutamine
3.11±0.40†
7.62±1.09†
29.82±4.66†
8.65±2.67†‡
Nitroprusside
Hetastarch
P-value
3.67±0.75
P<0.001†
60.24±10.81† 24.85±4.31
36.79±6.61
P<0.001†
1.24±3.63†
10.08±4.89
1754±291.1
P < 0.001†
P < 0.01‡
P < 0.001†
1463±167.2
P < 0.001†
3.59±0.37
(L/min)
SV
(ml)
LVEDP
3.44±3.08‡
(mmHg)
SVR(dyne* 2191±426.2†
5
927.4±378.2† 1568±285.6†
sec/cm )
+dP/dtmax 1546 ±375.5† 3157±464.7†
1675±432.9
(mmHg/s)
RR
900
800
P < 0.01
700
P < 0.01
600
500
400
300
Baseline
Anesthesia
Dobutamine
Nitroprusside
Hetastarch
Manipulations
Figure 17: Results of a repeated measure ANOVA on the variable RR
between the 5 stages of the study
50
2D-FS
55
50
P < 0.001
45
40
35
30
25
20
15
10
5
0
Baseline
Anesthesia
Dobutamine Nitroprusside
Hetastarch
Manipulations
Figure 18: Results of a repeated measure ANOVA on the variable FS between
the 5 stages of the study
CO (TD)
9.5
9.0
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
P<0.0001
Anesthesia
Dobutamine Nitroprusside
Hetastarch
Figure 19: Results of a repeated measure ANOVA on the variable CO during
the 3 pharmacologic manipulations
51
SV
80
70
60
P < 0.001
50
40
30
20
10
0
Anesthesia
Dobutamine Nitroprusside
Hetastarch
Manipulations
Figure 20: Results of a repeated measure ANOVA on the variable SV during
the 3 pharmacologic manipulations
+dP/dtmax
4000
3000
P<0.0001
2000
1000
0
Anesthesia
Dobutamine
Nitroprusside
Hetastarch
Manipulations
Figure 21: Results of a repeated measure ANOVA on the variable +dP/dtmax
during the 3 pharmacologic manipulations
52
LVED Pressure
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
-1
-2
-3
P < 0.05
P < 0.01
Anesthesia
Dobutamine
Nitroprusside
Hetastarch
Manipulations
Figure 22: Results of a repeated measure ANOVA on the variable LVEDP
during the 3 pharmacologic manipulations
Mean RA Pressure
10
9
8
7
P < 0.001
6
5
4
3
2
1
0
Anesthesia
Dobutamine Nitroprusside
Hetastarch
Manipulations
Figure 23: Results of a repeated measure ANOVA on the variable mean RAP
during the 3 pharmacologic manipulations
53
SVR
2750
P > 0.05
P < 0.01
2500
2250
P < 0.001
2000
1750
1500
1250
1000
750
500
250
0
Anesthesia
Dobutamine
Nitroprusside
Hetastarch
Manipulations
Figure 24: Results of a repeated measure ANOVA on the variable SVR during
the 3 pharmacologic manipulations
Pulmonary Vascular Resistance
No post tests. P > 0.05
1700
1600
1500
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
Anesthesia
Dobutamine
Nitroprusside
Hetastarch
Manipulations
Figure 25: Multiple comparison test on the variable PVR between the 3
pharmacologic manipulations
54
-dP/dt max
-750
No post tests. P > 0.05
-1000
-1250
-1500
-1750
-2000
-2250
-2500
Anesthesia
Dobutamine Nitroprusside
Hetastarch
Manipulations
Figure 26: Multiple comparison test on the variable -dP/dtmax between the 3
pharmacologic manipulations
Mean FA pressure
No post tests. P > 0.05
105
100
95
90
85
80
75
70
65
60
55
50
Anesthesia
Dobutamine
Nitroprusside
Hetastarch
Manipulations
Figure 27: Multiple comparison test on the variable mean FA pressure
between the 3 pharmacologic manipulation
55
PCWP
No post tests. P > 0.05
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
-1
Baseline
Dobutamine Nitroprusside Hetastarch
Manipulations
Figure 28: Multiple comparison test on the variable PCWP between the 3
pharmacologic manipulations
P value
0.0002
Pearson r
0.9929
P value (two-tailed)
0.0071
LiDCO vs Thermodilution
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
1
2
3
4
5
6
7
8
9
10
Thermodilution (L/min)
Figure 29: Result of a linear regression analysis with Pearson correlation test,
between LiDCO® and Thermodilution techniques
56
FS vs +dP/dT max
3750
R squared
0.9915
Pearson r
0.9957
P value
0.0043
3500
3250
3000
2750
2500
2250
2000
1750
1500
1250
1000
750
500
250
0
0
10
20
30
40
50
60
FS %
Figure 30: Result of a linear regression analysis with Pearson correlation test,
between 2D-FS and +dP/dtmax
SSR
No post tests. P > 0.05
-0.50
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
-2.75
1
2
3
4
5
6
ROIs
Figure 31: Results of a repeated measure ANOVA comparing the SSR values
between the 6 ROIs
57
avSSR
-0.50
-0.75
P > 0.05
-1.00
-1.25
-1.50
-1.75
P < 0.001
-2.00
-2.25
P < 0.001
-2.50
-2.75
-3.00
-3.25
Baseline
Anesthesia
Dobutamine Nitroprusside
Hetastarch
Manipulations
Figure 32: Results of a repeated measure ANOVA comparing the avSSR
during the five phases of the study
FS vs SSR
-0.25
R squared
0.9221
Spearman r
-0.8000
-0.50
P value
0.0397
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
-2.75
-3.00
-3.25
-3.50
0
10
20
30
40
50
60
FS %
Figure 33: Result of a linear regression analysis with Pearson correlation test,
between 2D-FS and avSSR
58
avSSR vs +dP/dTmax
-0.50
-0.75
r squared
0.9588
Pearson r
-0.9792
P value
0.0208
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
-2.75
-3.00
-3.25
0
250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3000 3250 3500 3750
+dP/dTmax (mmHg/s)
Figure 34: Result of a linear regression analysis with Pearson correlation test,
between SSR and +dP/dtmax
ROI 1
-0.50
r squared
0.9778
-0.75
P value
0.0111
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
-2.75
-3.00
-3.25
-3.50
0
500
1000
1500
2000
2500
3000
3500
4000
max +dP/dT
Figure 35: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 1 and +dP/dtmax
59
ROI 2
-0.25
-0.50
r squared
0.9950
P value
0.0025
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
0
500
1000
1500
2000
2500
3000
3500
4000
max +dP/dT
Figure 36: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 2 and +dP/dtmax
ROI 3
-0.25
r squared
0.9788
-0.50
P value
0.0106
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
-2.75
-3.00
0
500
1000
1500
2000
2500
3000
3500
4000
max +dP/dT
Figure 37: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 3 and +dP/dtmax
60
ROI 5
-0.5
r squared
0.9634
P value
0.0185
-1.0
-1.5
-2.0
-2.5
-3.0
-3.5
-4.0
0
500
1000 1500 2000 2500 3000 3500 4000
max +dP/dT
Figure 38: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 5 and +dP/dtmax
ROI 6
0.0
-0.5
r squared
0.9051
P value
0.0486
3500
4000
-1.0
-1.5
-2.0
-2.5
-3.0
-3.5
-4.0
-4.5
0
500
1000
1500
2000
2500
3000
max +dP/dT
Figure 39: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 6 and +dP/dtmax
61
ROI 4
0.0
r squared
0.6889
P value
0.1700
3500
4000
-0.5
-1.0
-1.5
-2.0
-2.5
-3.0
-3.5
-4.0
0
500
1000
1500
2000
2500
3000
max +dP/dT
Figure 40: Result of a linear regression analysis with Pearson correlation test,
between SSR measured at ROI 4 and +dP/dtmax
62
Study 2
A total of 65 dogs was initially evaluated. Nine dogs did not completely meet
the inclusion criteria because of the presence of a murmur (3 dogs), systemic
hypertension (1 dog), ongoing thyroid hormone supplementation (1 dog),
ongoing treatment with phenylpropanolamine (1 dog), echocardiographic
evidence of mild tricuspid valve dysplasia (1 dog), and echocardiographic
evidence of reduced systolic function (2 dogs). The remaining 56 dogs were
enrolled in the study. There were a total of 4 intact females, 6 intact males, 18
spayed females, and 21 neutered males, with ages ranging between 1 and 8
years. The breeds included American Staffordshire Terrier (1 dog), Australian
shepherd (4), Beagle (1), Border collie (1), Chihuahua (1), Cocker spaniel (1),
Doberman pinscher (10), French Bulldog (1), Golden retriever (2), Great Dane
(2), Greyhound (1), Heeler mix (1), Italian Greyhound (1), Labrador Retriever
(2), Mastiff (1), mixed breed (19), Pitbull (1), Poodle (1), Pug (2), Rhodesian
Ridgeback (1), Saint Bernard (1), and Yorkshire Terrier (1). Their body weights
(BW) ranged from 2.4 to 63.2 kilograms. Echocardiographic data from seven
dogs were not included in the final analysis because of overall poor image
quality. Table 4 shows the descriptive statistics of the 49 dogs included in the
final analysis, as they were divided into the four study groups. There was a
tendency toward a higher number of intact dogs in group 4 (4 intact females
and 5 intact males) compared to the other groups, but it did not reach
63
statistical significance (p=0.956 from a Kruskal-Wallis test) (Figure 41).
Although a tendency for dogs in group 4 to have a younger age was identified,
it did not reach statistical significance either (p=0.162 from a Kruskal-Wallis
test) (Figure 42). Figure 43 shows the body weight distribution of the study
population. As expected, being part of the inclusion criteria, there was a
statistically significant difference between groups 1, 2, and 3 (p<0.001 from an
ANOVA test). There was also a significant difference (p<0.05 from an ANOVA
test, mean difference -8.11 kg) between group 4 and group 3. As already
mentioned, of the initial 56 dogs from which echocardiographic data were
available, 7 dogs (12.5%) were excluded because of poor image quality.
Reasons included poor patient cooperation (2 dogs), poor acoustic window (2
dogs), sinus tachycardia (1 dog), tachypnea (1 dog), and an operatordependent error in storing the images (1 dog). In fact, DI data can only be
post-processed from Hi-Definition DICOM files (HD-DCM files); depending on
the media utilized for storage and/or transfer of images, different compression
algorithms can be involved, in some instances negating further analysis of DI
data. The final analysis included DI data from 49 total dogs. Fourteen
randomly selected dogs (28.6% of the 49 total dogs) were included in the
repeatability sub-analysis, and underwent two echocardiographic studies
performed by the same operator. Seventeen randomly selected dogs (34.7%
of the 49 total dogs) underwent two echocardiographic exams performed by
two operators, to provide data on reproducibility (interoperator variability in the
64
phase of image acquisition and off-line analysis). The quality of the DI plots
was subjectively quantified by a single operator, attributing a score to each
curve
(i.e.
heart
beat)
analyzed,
defined
as
follows:
0
=
non-
measurable/interpretable plot; 1 = significant noise is present but the plot is
still interpretable; 2 = good noise-to-signal ratio. The seven dogs that were
excluded from the final analysis had images from which it was not possible to
generate recognizable DI curves. The score “0” was attributed to cases in
which it was not possible to generate a minimum of 5 good quality plots from
five separate cardiac cycles. A total of 1,470 ROIs were included in the final
analysis, excluding the echocardiograms that were repeated as part of the
R&R analysis. Only 5 dogs (10.2% of the 49 total dogs) received a score 2 in
some of the plots, for a total of 71 plots (4.84% of 1,470 plots). None of them
received a score 2 in all of their plots included in the final analysis. The
majority of plots (1294, equal to 88.02% of 1,470 curves analyzed) were given
a score of 1. The remainder of the curves received a score of 0 (105, equal to
7.14% of 1,470 curves). It means that, of the total number of DI values
obtained from the 49 dogs, 7.14% is not the average of five good quality
beats, but the result of the average of 4 good quality beats. Figure 44 shows
the distribution of scores in the four study groups. There was a tendency
towards a higher number of dogs receiving a score 1 in groups 2 and 3,
although it did not reach statistical significance (p=0.988 from a Kruskal-Wallis
test). Tables 5 and 6 show a summary of the DI parameters in the 49 dogs
65
included in the final analysis. The values are the average of 4-5 heart beats.
Table 5 shows values of individual ROIs, while in Table 6 the average of the 6
ROIs, standard deviation (SD), lower and upper 95% confidence interval, and
% coefficient of variability are indicated. Figure 45 shows the distribution of the
DI variable SSR in the six ROIs. There was a significant difference in the
medians from a Kruskal-Wallis test, between some groups, as indicated in
Table 7. Similar analyses were performed on the S, SRE, and SRA data, to
evaluate the presence of differences between the six ROIs. Figure 46 shows
the distributions of DI parameter S. There was a significant difference in the
means (p<0.0001 from an ANOVA test) between some ROIs, as indicated in
Table 8. Figure 47 and Figure 48 show the distributions of the variables SRE
and SRA, between the six ROIs. There were no statistically significant
differences in the medians (p<0.05 from Kruskal-Wallis tests) for both SRE
and SRA. Figures 49, 50, 51, and 52 show the distributions of the DI
parameter SSR in the six ROIs, respectively in groups 1, 2, 3, and 4. There
were statistically significant differences between some ROIs in group 1 and 2,
but not in group 3 and 4. The instantaneous heart rate was measured from the
RR interval, and recorded for each cardiac cycle evaluated in the final
analysis. There were no significant differences in HR between the four groups
(p<0.05 from an ANOVA test) (Figure 53). The HR was also evaluated in the
subset of data utilized in the R&R analysis. Figure 54 shows the distribution of
HR during the two echocardiographic examinations performed by the same
66
operator. There were no significant differences (Two-tailed p=0.429 from a
paired t Test) in HR between the two echocardiograms. Same results were
obtained from the analysis of the HR distribution between the two
echocardiographic examinations used in the reproducibility analysis. Figure 55
shows the distribution of HR during the two echocardiographic examinations
performed by the two operators. There were no significant differences (Twotailed p=0.431 from a paired t Test) in HR between the two sets of data. To
rule out differences in NIBP as a potential confounding variable in the
assessment of ventricular performance, the mean NIBP values were
compared in the four study groups. There were no differences between the
groups (p=0.176 from a Kruskal-Wallis test), as shown in Figure 56. To identify
potential differences in the group of Doberman pinschers versus other breeds,
controlling for differences in BW, initially a weight-matched subgroup of dogs
was identified and compared with group 4. As previously reported dogs in
group 3 were significantly different regarding their BW, compared to dogs in
group 4. Figure 57 shows a comparison between a BW-matched subgroup 3
and group 4. There was no significant difference between the BW in these two
groups (p=0.302 from an unpaired t Test). Initially the EF was compared
between these two groups. There was no difference between the two groups
(Two-tailed p=0.325 from an unpaired t Test) (Figure 58). The DI variable
avSSR was compared between these two groups (Figure 59). There was no
significant difference between the two groups (p=0.065 from a Mann Whitney
67
test). Similar analysis was conducted on the variable avS. There was no
difference between the two groups (p=0.250 from an unpaired t Test) (Figure
60). The potential effect of BW on DI data was also investigated. For this
analysis dogs were initially re-grouped into three BW groups, independently on
their breed (i.e. dogs in group 4 were included in other groups based on their
BW). Figure 61 shows the effect of BW on avSSR. There was a statistically
significant difference between group 1 and 2 (p<0.01 from an ANOVA test)
and between group 1 and 3 (p<0.05 from an ANOVA test). There was no
significant difference between group 2 and 3 (p>0.05 from an ANOVA test).
When the data from the Doberman pinscher group were excluded from the
analysis, the results of the comparison did not change (Figure 62). The three
BW groups were also compared based on the traditional echocardiographic
indices of systolic function, FS and EF (Figures 63 and 64, respectively). Both
FS (p=0.069 from an ANOVA test) and EF (p=0.245 from an ANOVA test) did
not differ significantly between the three BW groups. To further characterize
the potential effects of BW on avSSR, simple linear regression analysis
between avSSR and BW was conducted (Figure 65). No significant correlation
was documented with this model (r squared=0.011). Same analysis was
conducted on avS. The absence of a significant correlation that can be
explained with a simple linear regression model (r squared=0.013) was
documented as well (Figure 66). A potential correlation between traditional
indices of systolic function (FS and EF) and DI indices of systolic function
68
(avSSR and avS) was investigated constructing a simple linear regression
model. Figures 67, 68, 69, and 70 show the results of these models. It was not
possible to document a significant correlation between avSSR and FS (r
squared=0.196), avSSR and EF (r squared=0.047), avS and FS (r
squared=0.079), and avS and EF (r squared=0.080), in the set of data from
this study. The potential effects of HR on avSSR and avS were investigated
using the same regression model. It was not possible to document a
significant correlation between avSSR and HR (r squared=0.314) (Figure 71)
and between avS and HR (r squared=0.026) (Figure 72). The R&R analysis
was conducted using Bland-Altman analysis. This analysis generates plots
that display the differences between the two measurements as a function of
the average of each sample, which can be considered as the best estimate of
the true value. The bias (which corresponds to the average of the differences
between the two sets of data), the SD of the differences between the two sets
of data, and the 95% limits of agreement are provided in the tables. Figures
73, 74, 75, and 76 show the Bland-Altman plots of the repeatability subanalysis, and Table 9 shows bias, SD and 95% limits of agreement for the
variables avSSR, avSRE, avSRA and avS in the repeatability subanalysis.
Figures 77, 78, 79, and 80 show the Bland-Altman plots of the reproducibility
sub-analysis, and Table 10 reports bias, SD and 95% limits of agreement for
the variables avSSR, avSRE, avSRA and avS in the reproducibility
subanalysis.
69
Table 4: Descriptive statistics of the 49 dogs included in the final analysis
Variable
Group 1 Group 2 Group 3 Group 4
Sex
Females
Males
Spayed females
Neutered males
0
0
2
8
0
0
9
6
0
1
6
7
4
5
1
0
Mean
SD
Median
Minimum
Maximum
9.67
3.52
10.15
2.40
14.50
24.25
3.84
26.00
16.20
29.52
41.69
9.21
41.15
31.00
63.21
33.58
4.97
32.30
28.00
43.60
Mean
SD
Median
Minimum
Maximum
4.2
2.73
4.0
1.0
8.0
4.26
2.22
4.0
2.0
8.0
4.21
2.15
3.5
2.0
8.0
2.42
0.84
2.0
1.0
4.0
10
15
14
10
BW
Age
Total
70
Table 5: DI parameters from the six ROIs (Mean values)
SSR
SRE
SRA
S
ROI 1 ROI 2 ROI 3 ROI 4
-1.40 -1.21 -1.38 -1.55
1.69
1.45
1.71
1.52
1.27
1.25
1.45
1.19
-12.38 -12.00 -12.92 -14.25
ROI 5
-1.88
1.51
1.30
-16.70
ROI 6
-1.52
1.69
1.45
-14.22
Table 6: DI parameters as average of the six ROIs
Mean
avSSR -1.49
avSRE 1.60
avSRA 1.32
avS
-13.74
SD Lower 95%C.I. Upper 95% C.I.
0.31
-1.58
-1.40
0.43
1.47
1.72
0.35
1.22
1.42
2.09
-14.34
-13.15
CV%
20.79
26.90
26.25
15.18
71
Sex distribution
Group 1
Group 2
Group 3
Group 4
P value
9
0.9561
8
7
6
5
4
3
2
1
0
F
M
SF
NM
Sex status
Figure 41: Sex distribution in the four study groups
Age distribution
P value
8.5
0.1617
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Group 1
Group 2
Group 3
Figure 42: Age distribution in the four study groups
Group 4
72
BW
P value
70
P<0.0001
60
50
40
30
20
10
0
Group 1
Group 2
Group 3
Group 4
Figure 43: BW distribution in the four study groups
Plot scores
450
Group 1
Group 2
Group 3
Group 4
400
P value
0.9883
350
300
250
200
150
100
50
0
Score 0
Score 1
Score 2
Figure 44: Distribution of plot quality scores in the four study groups
73
SSR (ROIs 1-6)
-0.10
-0.20
-0.30
-0.40
-0.50
-0.60
-0.70
-0.80
-0.90
-1.00
-1.10
-1.20
-1.30
-1.40
-1.50
-1.60
-1.70
-1.80
-1.90
-2.00
-2.10
-2.20
-2.30
-2.40
-2.50
-2.60
-2.70
-2.80
-2.90
-3.00
-3.10
-3.20
-3.30
-3.40
-3.50
-3.60
-3.70
1
2
3
4
5
6
ROIs
Figure 45: Distribution of the DI variable SSR in the six ROIs
Table 7: Results of a Dunn’s Multiple Comparison Test for the variable SSR,
between the 6 ROIs
Parameter
Dunn's Multiple Comparison Test
1 vs 2
1 vs 3
1 vs 4
1 vs 5
1 vs 6
2 vs 3
2 vs 4
2 vs 5
2 vs 6
3 vs 4
3 vs 5
3 vs 6
4 vs 5
4 vs 6
5 vs 6
Value
Difference in rank sum
-29.34
2.796
28.59
76.78
15.95
32.13
57.93
106.1
45.29
25.80
73.98
13.15
48.18
-12.64
-60.83
Data Set-B
P value
P > 0.05
P > 0.05
P > 0.05
P < 0.001
P > 0.05
P > 0.05
P < 0.05
P < 0.001
P > 0.05
P > 0.05
P < 0.001
P > 0.05
P > 0.05
P > 0.05
P < 0.01
Data Set-C
Summary
ns
ns
ns
***
ns
ns
*
***
ns
ns
***
ns
ns
ns
**
74
S (ROIs 1-6)
-2.00
-3.00
-4.00
-5.00
-6.00
-7.00
-8.00
-9.00
-10.00
-11.00
-12.00
-13.00
-14.00
-15.00
-16.00
-17.00
-18.00
-19.00
-20.00
-21.00
-22.00
-23.00
-24.00
-25.00
-26.00
-27.00
-28.00
-29.00
-30.00
1
2
3
4
5
6
ROIs
Figure 46: Distribution of the DI variable S in the six ROIs
Table 8: Results of a Dunn’s Multiple Comparison Test for the variable S,
between the 6 ROIs
Parameter
Bonferroni's Multiple Comparison Test
1 vs 2
1 vs 3
1 vs 4
1 vs 5
1 vs 6
2 vs 3
2 vs 4
2 vs 5
2 vs 6
3 vs 4
3 vs 5
3 vs 6
4 vs 5
4 vs 6
5 vs 6
Value
Mean Diff.
-0.3745
0.5471
1.872
4.325
1.845
0.9216
2.247
4.699
2.219
1.325
3.778
1.298
2.453
-0.02714
-2.480
Data Set-B
t
0.4942
0.7220
2.470
5.707
2.434
1.216
2.964
6.201
2.929
1.748
4.985
1.712
3.237
0.03582
3.272
Data Set-C
P value
P > 0.05
P > 0.05
P > 0.05
P < 0.001
P > 0.05
P > 0.05
P < 0.05
P < 0.001
P > 0.05
P > 0.05
P < 0.001
P > 0.05
P < 0.05
P > 0.05
P < 0.05
Data Set-D
95% CI of diff
-2.618 to 1.869
-1.696 to 2.790
-0.3711 to 4.115
2.082 to 6.568
-0.3982 to 4.088
-1.321 to 3.165
0.003414 to 4.490
2.456 to 6.943
-0.02373 to 4.463
-0.9182 to 3.568
1.535 to 6.021
-0.9454 to 3.541
0.2097 to 4.696
-2.270 to 2.216
-4.723 to -0.2369
75
SR E (ROIs 1-6)
4.50
No post tests. P > 0.05
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
1
2
3
4
5
6
ROIs
Figure 47: Distribution of the DI variable SRE in the 6 ROIs
SR A (ROIs 1-6)
No post tests. P > 0.05
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0.00
1
2
3
4
5
6
ROIs
Figure 48: Distribution of the DI variable SRA in the 6 ROIs
76
SSR (ROIs 1-6) Group 1
-0.50
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
P < 0.01
-2.75
P < 0.05
-3.00
-3.25
1
2
3
4
5
6
ROIs
Figure 49: Distribution of the DI variable SSR in the 6 ROIs in group 1
SSR (ROIs 1-6) Group 2
-0.25
-0.50
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
P < 0.01
-2.25
-2.50
P < 0.001
P < 0.01
-2.75
1
2
3
4
5
6
ROIs
Figure 50: Distribution of the DI variable SSR in the 6 ROIs in group 2
77
SSR (ROIs 1-6) Group 3
No post tests. P > 0.05
-0.10
-0.20
-0.30
-0.40
-0.50
-0.60
-0.70
-0.80
-0.90
-1.00
-1.10
-1.20
-1.30
-1.40
-1.50
-1.60
-1.70
-1.80
-1.90
-2.00
-2.10
-2.20
-2.30
-2.40
-2.50
-2.60
-2.70
-2.80
-2.90
-3.00
-3.10
-3.20
-3.30
-3.40
-3.50
-3.60
-3.70
1
2
3
4
5
6
ROIs
Figure 51: Distribution of the DI variable SSR in the 6 ROIs in group 3
SSR (ROIs 1-6) Group 4
-0.50
No post tests. P > 0.05
-0.60
-0.70
-0.80
-0.90
-1.00
-1.10
-1.20
-1.30
-1.40
-1.50
-1.60
-1.70
-1.80
-1.90
-2.00
-2.10
1
2
3
4
5
6
ROIs
Figure 52: Distribution of the DI variable SSR in the 6 ROIs in group 4
78
HR
No post tests. P > 0.05
175
170
165
160
155
150
145
140
135
130
125
120
115
110
105
100
95
90
85
80
75
70
65
60
Group 1
Group 2
Group 3
Group 4
Figure 53: Distribution of the HR values in the four study groups
HR (Repeatabilty)
P value
0.4290
175
170
165
160
155
150
145
140
135
130
125
120
115
110
105
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
1
2
Echocardiogram
Figure 54: Distribution of the HR values during the two echocardiographic
examinations performed by the same operator
79
HR (Reproducibility)
P value
0.4317
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
HR Op1
HR Op2
Figure 55: Distribution of the HR values during the two echocardiographic
examinations performed by the two different operators
Mean BP
No post tests. P > 0.05
155
150
145
140
135
130
125
120
115
110
105
100
95
90
85
80
75
Group 1
Group 2
Group 3
Group 4
Figure 56: Distribution of the BP values in the 4 study groups
80
BW-matched Group 3 vs Group 4
47
46
45
44
43
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
P value
BW-matched Group 3
0.3019
Group 4
Figure 57: Comparison between a BW-matched subgroup 3 and group 4
EF BW-matched Group 3 vs Group 4
P value
0.3252
77.5
75.0
72.5
70.0
67.5
65.0
62.5
60.0
57.5
55.0
52.5
50.0
47.5
45.0
42.5
40.0
"BW-matched" Group 3
Group 4
Figure 58: Comparison between the EF values in the BW-matched subgroup 3
and group 4
81
avSSR BW-matched Group 3 vs Group 4
P value
0.0653
-0.6
-0.7
-0.8
-0.9
-1.0
-1.1
-1.2
-1.3
-1.4
-1.5
-1.6
-1.7
-1.8
-1.9
"BW-matched" Group 3
Group 4
Figure 59: Comparison between the avSSR values in the BW-matched
subgroup 3 and group 4
avS BW-matched Group 3 vs Group 4
P value
-9
0.2515
-10
-11
-12
-13
-14
-15
-16
-17
-18
BW matched Group 3
Group 4
Figure 60: Comparison between the avS values in the BW-matched subgroup
3 and group 4
82
avSSR vs BW groups
-0.6
-0.7
-0.8
-0.9
-1.0
-1.1
-1.2
-1.3
-1.4
-1.5
-1.6
-1.7
-1.8
-1.9
-2.0
P < 0.01
-2.1
-2.2
-2.3
P < 0.05
-2.4
Group 1
Group 2
Group 3
Figure 61: Comparison between the avSSR values in the three BW groups
(For this analysis groups 4 dogs were redistributed to other groups based on
their BW)
avSSR vs BW groups (excluding Dobermans)
-0.9
-1.0
-1.1
-1.2
-1.3
-1.4
-1.5
-1.6
-1.7
-1.8
-1.9
-2.0
-2.1
P < 0.01
-2.2
P < 0.01
-2.3
-2.4
Group 1
Group 2
Group 3
Figure 62: Same analysis as in Figure 36, excluding dogs from group 4
83
FS vs BW groups
No post tests. P > 0.05
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Group 1
Group 2
Group 3
Figure 63: Comparison between the FS values in the three BW groups
EF vs BW groups
No post tests. P > 0.05
82.5
80.0
77.5
75.0
72.5
70.0
67.5
65.0
62.5
60.0
57.5
55.0
52.5
50.0
47.5
45.0
42.5
40.0
37.5
35.0
32.5
30.0
27.5
Group 1
Group 2
Group 3
Figure 64: Comparison between the EF values in the three BW groups
84
avSSR vs BW
avSSR
-0.50
r squared
0.01106
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
0
10
20
30
40
50
60
70
BW (kg)
Figure 65: Simple linear regression analysis between avSSR and BW
avS vs BW
-8
r squared
0.01300
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
0
10
20
30
40
50
60
70
BW (kg)
Figure 66: Simple linear regression analysis between avS and BW
85
avSSR vs FS
-0.7
-0.8
r squared
0.1960
-0.9
-1.0
-1.1
-1.2
-1.3
-1.4
-1.5
-1.6
-1.7
-1.8
-1.9
-2.0
-2.1
-2.2
-2.3
-2.4
-2.5
0.0
2.5
5.0
7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
FS%
Figure 67: Simple linear regression analysis between avSSR and FS
avS vs FS
-8
-9
r squared
0.07943
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
0.0
2.5
5.0
7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
FS%
Figure 68: Simple linear regression analysis between avS and FS
86
avSSR vs EF
r squared
-8
0.08088
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
30
35
40
45
50
55
60
65
70
75
80
85
EF
Figure 69: Simple linear regression analysis between avSSR and EF
avS vs EF
r squared
-8
0.08088
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
30
35
40
45
50
55
60
65
70
75
80
85
EF
Figure 70: Simple linear regression analysis between avS and EF
87
avSSR vs HR
r squared
-0.50
0.3143
avSSR
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
60
70
80
90
100
110
120
130
140
150
160
170
180
HR (bpm)
Figure 71: Simple linear regression analysis between avSSR and HR
avS vs HR
r squared
-8
0.02634
avS
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
60
70
80
90
100
110
120
130
140
150
160
170
180
HR (bpm)
Figure 72: Simple linear regression analysis between avS and HR
88
avSSR (Repeatability)
-0.9
-1.0
-1.1
-1.2
-1.3
-1.4
-1.5
-1.6
-1.7
-1.8
-1.9
-2.0
-2.1
-2.2
-2.3
-2.4
Echo 1
Echo 2
Bland-Altman (Repeatability): Difference vs average
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
-0.00
-0.05
-0.10
-0.15
-0.20
-0.25
-0.30
-0.35
-0.40
-0.45
-0.50
-0.55
-0.60
-0.65
-0.70
-2.2
-2.1
-2.0
-1.9
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
-1.1
-1.0
Average
Figure 73: Bland-Altman plots of the repeatability sub-analysis for avSSR
89
avSRE (Repeatability)
2.4
2.3
2.2
2.1
2.0
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
0.9
Echo 1
Echo 2
egareva sv ecnereffiD :)ytilibataepeR( namtlA-dnalB
54.0
04.0
53.0
03.0
52.0
02.0
51.0
01.0
50.0
00.050.001.051.002.052.003.053.004.054.005.055.006.056.007.05.2
4.2
3.2
2.2
1.2
0.2
9.1
8.1
7.1
6.1
5.1
4.1
3.1
2.1
1.1
0.1
9.0
egarevA
Figure 74: Bland-Altman plots of the repeatability sub-analysis for avSRE
90
avSRA (Repeatability)
2.2
2.1
2.0
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
0.9
Echo 1
Echo 2
Bland-Altman (Repeatability): Difference vs average
0.85
0.80
0.75
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
-0.00
-0.05
-0.10
-0.15
-0.20
-0.25
-0.30
-0.35
-0.40
-0.45
-0.50
-0.55
-0.60
-0.65
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
2.2
2.3
Average
Figure 75: Bland-Altman plots of the repeatability sub-analysis for avSRA
91
avS (Repeatability)
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
Echo 1
Echo 2
Bland-Altman (Reproducibility) :Difference vs average
4
3
2
1
0
-1
-2
-3
-4
-5
-6
-18
-17
-16
-15
-14
-13
-12
-11
-10
Average
Figure 76: Bland-Altman plots of the repeatability sub-analysis for avS
92
Table 9: Results of a Bland-Altman analysis of the data from the repeatability
subanalysis
Parameter (repeatability)
Bias
SD
95% Limits of agreement
From:
To:
avSSR
0.076
0.301
avSRE
-0.056
0.264
avSRA
-0.023
0.395
avS
-0.124
2.598
-0.514
0.666
-0.574
0.461
-0.797
0.751
-5.215
4.968
93
avSSR (Reproducibility)
-0.75
-1.00
-1.25
-1.50
-1.75
-2.00
-2.25
-2.50
Op. 1
Op. 2
Bland-Altman (Reproducibility): Difference vs average
0.6
0.5
0.4
0.3
0.2
0.1
-0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-2.3
-2.2
-2.1
-2.0
-1.9
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
-1.1
Average
Figure 77: Bland-Altman plots of the reproducibility sub-analysis for avSSR
94
avSRE (Reproducibility)
3.50
3.25
3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
Op. 1
Op. 2
Bland-Altman (Reproducibility): Difference vs average
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
-0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
2.2
2.3
2.4
2.5
2.6
2.7
Average
Figure 78: Bland-Altman plots of the reproducibility sub-analysis for avSRE
95
avSRA (Reproducibility)
3.50
3.25
3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
Op. 1
Op. 2
Bland-Altman (Reproducibility): Difference vs average
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
-0.00
-0.05
-0.10
-0.15
-0.20
-0.25
-0.30
-0.35
-0.40
-0.45
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
Average
Figure 79: Bland-Altman plots of the reproducibility sub-analysis for avSRA
96
avS (Reproducibility)
-9
-10
-11
-12
-13
-14
-15
-16
-17
Op. 1
Op. 2
Bland-Altman (Reproducibility): Difference vs average
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
-3.0
-3.5
-17
-16
-15
-14
-13
-12
-11
-10
-9
Average
Figure 80: Bland-Altman plots of the reproducibility subanalysis for avS
97
Table 10: Results of a Bland-Altman analysis of the data from the
reproducibility subanalysis
Parameter (reproducibility)
Bias
SD
95% Limits of agreement
From:
To:
avSSR
0.043
0.272
avSRE
0.624
0.586
avSRA
0.000
0.254
avS
-0.650
1.680
-0.491
0.577
-0.524
1.772
-0.497
0.497
-3.942
2.642
98
Discussion
Study 1
The comparisons between the hemodynamic variables obtained at steady
state after induction of anesthesia and during the different pharmacologic
manipulations suggest that significant changes in the inotropic state, preload
and afterload were reached during the study. Hence, the study was able to
generate a model in which a relatively wide range of inotropic and loading
conditions was created. The significance of these variations was of both
statistical and clinical importance, as documented not only by a p-value of less
then 0.05, but also by a wide range of variations in many hemodynamic
parameters, as shown in Table 3. The dobutamine infusion was able to
increase the SV from 29.82 ± 4.66 ml at steady state after induction of
anesthesia, to 60.24 ± 10.81 ml. The RR interval was significantly shortened
during the dobutamine and nitroprusside infusions, compared to baseline
under anesthesia, indicating that the HR was significantly increased. In fact,
the RR interval is a parameter linearly and inversely correlated to the HR, as
shown by the relationship RR = 60,000/HR, where 60,000 is the number of ms
in one minute, and the HR is expressed in beats per minute (BPM). Therefore
one could comment that the increase in CO (where CO = HR x SV) could have
been merely the result of an increase in HR. Hence, it is important to focus on
99
the SV, because it represents an index of LV systolic function that is relatively
independent of the HR. Dobutamine also increased the +dP/dtmax from 1546 ±
375.5 mmHg/s to 3157 ± 464.7 mmHg/s. Both changes indicate that a
significant positive inotropic effect was obtained, making this study an
adequate model to correlate invasive and non-invasive indices of systolic
function. Both CO and SV returned to baseline during nitroprusside and
hetastarch administration. Pharmacokinetic data for dobutamine in domestic
animals are apparently unavailable.(50) Data from the human literature
describe dobutamine as a drug that is metabolized rapidly in the liver and
other tissues and has a plasma half-life of approximately 2 minutes. These
facts should support the appropriateness of a wash-out period of a minimum of
20 minutes, as it was adopted in the study. Dobutamine is a sympathomimetic
drug capable of exerting a positive lusitropic effect (improved relaxation).(50)
Nitroprusside, a balanced vasodilator, has the ability of reducing the
preload.(51) The LVEDP was significantly reduced during dobutamine and
nitroprusside infusions, and returned to baseline after the hetastarch bolus.
These findings indicate that significant changes in the preload have been
obtained during the study. Systemic vascular resistances were significantly
reduced during nitroprusside infusion, documenting that a reduction in the
afterload was obtained in the study. Mean RA pressure was significantly
increased during the hetastarch infusion, indicating that volume expansion,
with secondary increase in the right ventricular preload, was obtained.
100
However, the pulmonary capillary wedge pressure did not change significantly.
This may indicate, together with the fact that the left ventricular end-diastolic
pressure was not significantly elevated during the hetastarch infusion, that this
model was unable of creating a significant increase in the LV preload. An
alternative explanation could be that, at the time of the hetastarch infusion, the
LV was still somehow benefiting from a residual positive lusitropic effect of
dobutamine, enabling it to accommodate the increased venous return, without
significant increases in the LVEDP. Against this hypothesis is the fact that the
invasive parameter -dP/dtmax did not change significantly throughout the study,
indicating that, although the LVEDP (an end-diastolic event) was reduced, the
maximum rate of relaxation during the isovolumic relaxation phase (an early
diastolic event) was not significantly increased. Although there was a trend
toward more negative values of -dP/dtmax during dobutamine infusion, this
change did not reach statistical significance. Because of this we cannot
document that this study created an adequate model to correlate invasive and
non-invasive indices of diastolic function, because what would have been our
“gold-standard” for the quantification of the intrinsic diastolic function did not
reach statistical significance. This is the most important reason why we
focused on indices of systolic function (mainly avSSR and avS). This study
also focused on the longitudinal systolic function of the LV, for reasons
explained in an earlier chapter; findings from this research cannot be
extrapolated to the study of the radial and circumferential LV function. The
101
parameter avSSR was significantly reduced, i.e. systolic function was
increased, by the dobutamine infusion, but was not significantly different
during nitroprusside and hetastarch infusion. It is worth noting that the LVEDP
was significantly reduced during the nitroprusside infusion, apparently not
influencing avSSR, and suggesting the possible preload independence of this
parameter, in the hemodynamic range generated in the study. Furthermore,
SVR remained significantly reduced during nitroprusside administration, but
appeared
not
to
influence
avSSR, suggesting
a
possible
afterload
independence of this parameter, in the hemodynamic range obtained in this
study.
A significant percentage (98.67%) of ROIs included in the final analysis
yielded interpretable Doppler-derived DI curves. In order to correctly interpret
the value of this finding it is important to consider that a total of 14 cardiac
cycles were available for off-line analysis, providing the operator with an
amount of data to choose from, which is significantly greater compared to what
is stored in a routine echocardiographic study (usually between 2 and 5
cardiac cycles).
The SSR values obtained from the 6 ROIs and analyzed separately over the
hemodynamic range generated in the study did not differ significantly from
each other. For this reason the results of many analyses focusing on single
ROIs are not included. This finding supports the hypothesis that longitudinal
SSR is homogeneously distributed along the long axis of the LV, negating the
102
presence of a significant longitudinal gradient for this parameter. Theoretically,
this finding by itself suggests that the functional information obtainable with
SRI could be independent of the location of the specific ROI along the long
axis of the LV, with the only exception being a patient with LV dyskinesia.
Both FS% (Figure 18) and avSSR (Figure 32) did not change significantly
between the baseline echocardiogram before anesthesia and the steady state
after induction of anesthesia. The level of anesthesia was monitored closely by
a board-certified anesthesiologist and always maintained as light as possible.
The procedures performed on the animals were not particularly invasive,
hence surgical pain was limited. The light level of anesthesia definitely played
a role in the absence of detectable negative inotropic effects, as documented
by the absence of changes in FS and peak systolic SR. These data may be
useful in the interpretation of data from echocardiographic examinations
conducted under a similar anesthetic protocol.
103
Study 2
The analysis of the population of dogs enrolled in study 2 could not identify
significant differences between the four groups, beyond the obvious
differences in BW and breed, which were part of our inclusion criteria.
Because it was difficult to enroll the minimum number of Doberman pinschers
defined in the study protocol, breeders in the area were contacted. This led to
the fact that the majority of Doberman pinschers were relatively young, intact
dogs; animals for which the breeder had an interest in obtaining a complete
cardiovascular work-up. This bias in the phase of recruitment is reflected by
the trend for group 4 toward a younger age and a higher number of intact
individuals. This trend did not reach statistical significance with the statistical
model adopted in this study, but should be considered as a potential limitation
of the study. Seven dogs out of the initial 56 (12.5%) were excluded from the
off-line analysis because the quality of the DI curves was considered
unacceptable. These dogs had acceptable traditional echocardiographic
images. Furthermore, of the 1,470 ROIs analyzed, only 10.2% received a
score of 2 (good noise-to-signal ratio), and this level of quality was never
found in all of the four/five beats included in the final analysis. It was not
possible to analyze 7.14% of the ROIs and the majority of the curves were
assigned a score 1, indicating that their noise-to-signal ratio was high, but that
systolic, early diastolic (E) and late diastolic (A) waves were still discernible. A
104
high level of noise can cause two major problems in terms of off-line analysis.
The noise can add to or subtract from the real parameter, causing falsely
elevated or falsely reduced values and contributing to the overall variability of
these parameters. Noise also causes the appearance of spurious waves,
which makes the identification of the real wave pattern challenging, potentially
leading to an incorrect interpretation of the tracing. There appeared to be a
trend for larger breed dogs toward higher plot quality scores, although it did
not reach statistical significance. A higher number of dogs might have yielded
a different result. This trend can be explained by the fact that medium size
dogs often have a better acoustic window, and frequently are also more willing
to cooperate, compared to small breed dogs, hypothetically because of
different temperament and/or different training. The hypothesis of the absence
of a left ventricular longitudinal gradient for DI parameters could not be
confirmed by the results of study 2. The analyses of the systolic phase DI
parameters SSR and S identified the presence of a difference between some
of the ROIs. The same difference could not be documented for the diastolic
phase DI variables SRE and SRA. The finding of a significant difference in
SSR and S values between different ROIs in study 2 is in disagreement with
the results of study 1, and with the medical literature on the topic. One
explanation for this contradictory finding is that the high level of variability in
the DI data collected in this group of non-sedated dogs might have
confounded the results of the statistical analysis. Other explanations include
105
some systematic sampling error or a true deformation difference that might be
body weight or heart size dependent. Why the SRE and SRA waves were
apparently not affected by this phenomenon, remains to be explained. It was
not possible to document a difference in avSSR between Doberman pinschers
and BW-matched non-Doberman dogs. The p value from a Mann Whitney test
was 0.060; it can be hypothesized that a higher number of dogs might have
yielded a different result. However this moderate lack of significance is in
agreement with the convincing lack of significance (p=0.325) obtained
comparing the EF values between the same two groups. It can be probably
concluded that there was no detectable difference in systolic function between
the group of Doberman pinschers enrolled in the study and the BW-matched
non-Doberman group. Interestingly the study documented a significant
difference (p<0.01) in avSSR in dogs weighting less than 15 kilograms (group
1) compared to group 2. Relatively convincing difference (p<0.05) was also
found between group 1 and 3. These findings did not change when Doberman
pinscher dogs were excluded from the analysis. The mean difference ranged
from -0.45 s-1 (between group 1 and 2) and -0.33 s-1 (between group 1 and 3),
showing a tendency for small breed dogs toward higher values of systolic
function, when evaluated based on avSSR. This mean difference is not only
statistically significant, but it also appears to be clinically significant, when
compared to the 95% C.I. for the parameter. No differences were found when
FS, EF and S values were compared between the three BW groups. It could
106
be hypothesized that avSSR has a higher sensitivity in detecting subtle
changes in systolic performance between normal dogs. Simple linear
regression models could not identify any correlation between the DI
parameters avSSR and avS and the population parameters BW and HR. At
the same time, it was not possible to document a correlation between the DI
indices of systolic function avSSR and avS and the traditional indices FS and
EF. This finding is in contradiction with the results of study 1, in which
relatively convincing evidence supporting the existence of a linear correlation
(r squared=0.922, p=0.039) between avSSR and FS was found. The most
plausible explanation for this discrepancy is the high level of variability (notice
the high coefficient of variability of the DI parameters in Table 6) encountered
in study 2 relative to the relatively modest range of contractile states
encountered in a population of healthy dogs. This may have confounded most
of the linear regression models constructed during the analysis. The results of
the R&R analysis based on the Bland-Altman plots allow for an interpretation
of the data with a strong clinical emphasis. Many papers in the veterinary
literature used various statistical approaches to estimate intra- and interoperator variability and several references in the literature comment on the
inappropriateness of some of these statistical approaches.(54,55,56,57,58) In
particular, Bland and Altman in the 1980’s criticized the use of these statistical
tools for clinical purposes.(58) In this paper, the use of correlation, regression
and the difference between means (such as a paired t-Test) are defined
107
inappropriate models to answer the question of interest: what is the level of
agreement between two sets of results? To summarize the points made by the
authors we can quote them as follows: “The correlation coefficient is not a
measure of agreement; it is a measure of association”. Furthermore, criticizing
the use of tests that compare the means of two data sets, the authors state
that by this criterion: “the greater the measurement error, and hence the less
chance of a significant difference, the better”. Finally the authors comment on
the use of linear regression, stating that: “this is equivalent to testing the
correlation coefficient against zero, and the above remarks apply”. Bland and
Altman conclude stating that: “a simple approach to the analysis may be the
most revealing way of looking at the data” and emphasize that this is a
question of estimating, both error and bias, and that no single statistic can
estimate both. The analysis they developed provides estimates of both error
and bias. Bland-Altman plots are generally interpreted informally, without
further analysis. There are three major questions the clinician should try to
answer. The first question pertains to how big the average discrepancy is
between the two sets of measurements (i.e. the bias). Is the discrepancy large
enough to be important? This question must be interpreted from the clinical,
not the statistical standpoint. Second question: Is there a trend? Does the
difference between measurements tend to get larger (or smaller) as the
average increases? Final question: is the variability consistent across the
graph? In other words: does the scatter around the bias line gets larger as the
108
average gets higher? The two last questions can be better assessed if the
number of measurements is moderately high, and is not feasible with small
samples.
Another common statistical parameter used to estimate intra-operator and
inter-operator variability is the coefficient of variation (CV), customarily
presented as a percentage (CV%). It represents a statistical measure of
dispersion of data points, calculated as:
(14)
cv =
σ
× 100 Where σ is the standard deviation (SD) and μ is the
μ
mean of the distribution.
The CV% is a useful statistic for comparing the degree of variation from one
data series to another, even if the means are drastically different from each
other. In fact, specifying the SD alone is more or less useless without the
additional specification of the mean value (and of course the type of
distribution). An important limitation of this statistic is that, when the mean
value is near zero, the coefficient of variation becomes almost entirely
dependent to changes in the standard deviation, limiting its usefulness.(59) The
mean differences between the two sets of measurements in both the
repeatability and reproducibility sub-analyses were, in most instances,
numbers close to zero (range from -0.650 to 0.624). This is the reason why the
CV% was deemed to be not the ideal statistic in the R&R analysis.
In the repeatability analysis (Table 9 from a Bland-Altman analysis) the intra-
109
operator bias ranged from -0.023 for avSRA to 0.076 for avSSR, which by
themselves represent relatively small numbers. However, when their SD is
taken into account we can appreciate how wide the 95% limits of agreement
are compared to the 95% confidence intervals of the corresponding DI
parameters, as shown in Table 6. If we consider, for instance, the variable
avS, the mean value from the four study groups was -13.74%, with a standard
deviation of 2.09, from which we can calculate the lower and upper 95%
confidence intervals (respectively -9.77 and -17.71). It means that, with 95%
confidence we can predict that, in the population of normal dogs enrolled in
the study, the variable will fall within this interval. The results of this study
suggest that, when this parameter is measured twice by the same operator,
there is 95% confidence that the difference between the two measurements
will range from -5.21 to 4.97. Assuming that the 95% C.I. obtained from the
four study groups are a good approximation of the normal range (at least in
the study population), it can be inferred that there is a high chance that
repeating the measurement in a hypothetical, normal animal could “shift” the
result out of the “normal” range; that is, that the intra-operator variability is
greater then the apparent natural variability of the parameter. This
interpretation holds true for the other parameters (avSSR, avSRE, and
avSRA), although, from inspection of Table 9, it appears that some
parameters performed slightly better than others. The results of study 2
suggest a relatively poor repeatability for the technique, when investigated
110
with a Bland-Altman analysis, which can be explained by the high variability of
the measurements.
The reproducibility analysis gave similar results (Table 10). Again, the 95%
limits of agreement for the four DI parameters are wider then the estimated
normal variability of each corresponding variable. Interestingly there appears
not to be a significant difference between the intra-operator and inter-operator
variability. Given the relatively small number of data points, the visual
inspection of the plots does not provide convincing evidence of a pattern, with
the exception of the plot on reproducibility of avSRE (Figure 78), which shows
a tendency toward a systematic bias. Measurements taken by operator 2
tended to be consistently lower then measurements taken by operator 1. A
potential explanation for the poor reproducibility in the assessment of SRE is
that these deflections occur in the early diastolic phase and there is not a
specific marker that enables their unequivocal identification. This is in contrast
to SRA waves, which occur immediately after the P wave on the surface ECG.
This reason could somewhat contribute to poor reproducibility, although it
does not provide a convincing justification for the systematic bias noticed on
the Bland-Altman plot, for which we do not have a good explanation.
Theoretically, the adoption of a marker indicating the timing of mitral valve
opening (MVO) could help improving the identification of SRE.
111
112
Conclusion
Based on the results of study 1, we conclude that Doppler-derived avSSR
represents a useful load-independent index of global systolic function as
demonstrated in anesthetized dogs, although time-consuming in the phase of
post-processing, sensitive to noise and, as described in the medical literature,
highly angle-dependent. However study 2 suggests that, in a population of
non-sedated dogs, with a relatively wide range of breeds and body weight,
Doppler-derived DI is characterized by less than optimal levels of intraoperator and inter-operator agreement which may limit its value as a tool to
evaluate myocardial function in a clinical setting. This study reports a wider
intra-operator variability compared to the results published by Chetboul et
al.(43) A possible explanation is that in our study repeatability was investigated
in a population of 49 dogs, characterized by a wide range of breeds and body
weight, while the study by Chetboul et al. assessed repeatability in a more
homogeneous group of six adult Beagles. Comparisons on reproducibility
cannot be made, because Chetboul et al. did not investigate inter-operator
variability. The DI values obtained from group 4 (Doberman pinschers) are
relatively similar to the results obtained by Wess et al.(44) However, the mean
age of the dogs enrolled in the two studies was different (2.4 ± 0.84 years in
study 2 and 4.3 ± 2.5 in the study by Wess et al.), potentially representing a
113
confounding variable in comparing the results of the two studies, given that the
prevalence of systolic dysfunction secondary to dilated cardiomyopathy
increases with age in Doberman pinschers.(60) The dogs enrolled in these
studies were not randomly selected from a broader population; hence these
results may not necessarily apply to a population of clinical patients. The
results of study 2 are in agreement with the limitations inherent with the
technique (low signal-to-noise ratio and high angle-dependence) described in
the medical literature.(13) The question as to whether experience with the
technique would improve its overall reliability cannot be answered by this
study. However, by the time the final analysis was conducted, both operators
had gained a total exposure to the technique of about two years, which,
compared with the learning curve typical of other echocardiographic
techniques, appears to be a reasonably appropriate time. The results of these
two studies indicate that further research efforts should be finalized toward the
evaluation of more sophisticated forms of DI, in particular the recently
introduced two-dimensional Strain Rate Imaging (2D-SRI), based on speckletracking technology. This new imaging modality represents a non-Doppler
technique, apparently non-angle dependent, more robust to noise, and
characterized by better repeatability and reproducibility.(61)
114
Bibliography
1.
Folland ED, Parisi AF, Moyniham PF, et al. Assessment of left ventricular
ejection fraction and volumes by real-time, two-dimensional echocardiography.
A comparison of cineangiographic and radionuclide techniques. Circulation
1979; 4(60): 760-766
2.
Isaaz K, Thompson A, Ethevenot G, et al. Doppler echocardiographic
measurements of flow velocity motion of left ventricular posterior wall. Am J
Cardiol 1989; 64: 66-75
3.
Sutherland GR, Stewart MJ, Groundstroem KW, et al. Color Doppler
myocardial imaging: a new technique for the assessment of myocardial
function. J Am Soc Echcardiogr 1994; 7: 441-45
4.
Lange A, Palka P, Caso P, et al. Doppler myocardial imaging vs. B-mode
grey-scale imaging: a comparative in vitro and in vivo study into their relative
efficacy in endocardial boundary detection. Ultrasound Med Biol 1997; 23: 6975
5.
Sohn DW, Chai IH, Lee DJ, et al. Assessment of mitral annulus velocity by
Doppler tissue imaging in the evaluation of left ventricular diastolic function. J
Am Coll Cardiol 1997; 30: 474-80.
6.
Nagueh SF, Mikati I, Kopelen HA, et al. Doppler estimation of left
ventricular filling pressure in sinus tachycardia, a new application of tissue
Doppler imaging. Circulation 1998; 98: 1644-50.
7.
Nagueh S, Middleton K, Kopelen HA, et al. Doppler tissue imaging: a
noninvasive technique for evaluation of left ventricular relaxation and
estimation of filling pressures. J Am Coll Cardiol 1997; 30: 1527-33
8.
Oyama MA, Sisson DD, Bulmer BJ, et al. Echocardiographic estimation of
mean left atrial pressure in a canine model of acute mitral valve insufficiency. J
Vet Intern Med 2004; 18(5): 667-72
9.
Fleming AD, McDicken WN, Sutherland GR, et al. Myocardial velocity
gradients detected by Doppler imaging. Br J Radiol 1994; 67: 679-88
10.
Pan C et al. Tissue tracking allows rapid and accurate visual evaluation of
left ventricular function. Europ J Echocardiography 2001; 2(3): 197-202
115
11.
Mirsky I, Parmley WW. Assessment of passive elastic stiffness for isolated
heart muscle and the intact heart. Circ. res. 1973; 33(2): 233-43
12.
Voigt JU, Flachskampf FA. Strain and strain rate. New and clinically
relevant echo parameters of regional myocardial function. Z Kardiol 2004; 93:
249-58
13.
Urheim S, Edvardsen T, Torp H, et al. Myocardial strain by Doppler
echocardiography: validation of a new method to quantify regional myocardial
function. Circulation 2000; 102: 1158-165
14.
Belohlavek M, Bartleson VB, Zobitz ME. Real-time strain rate
imaging:validation of peak compression and expansion rates by a tissue
mimicking phantom. Echocardiography 2001; 18: 565-71
15.
Heimdal A, D’Hooge J, Bijnens B, et al. In vitro validation of in-plane strain
rate imaging. A new ultrasound technique for evaluating regional myocardial
deformation based on tissue Doppler imaging. Echocardiography 1998; 15:
S40
16.
Teske AJ, De Boeck BW, Melman PG et al. Echocardiographic
quantification of myocardial function using tissue deformation imaging, a guide
to image acquisition and analysis using tissue Doppler and speckle tracking.
Cardiovasc Ultrasound 2007; 5: 27-47
17.
Greenberg NL, Firstenberg MS, Castro PL, et al. Doppler-derived
myocardial systolic strain rate is a strong index of left ventricular contractility.
Circulation 2002; 105: 99-105
Edvardsen T, Gerber BL, Garot J, et al. Quantitative assessment of intrinsic
regional myocardial deformation by Doppler strain rate echocardiography in
humans: validation against three-dimensional tagged magnetic resonance
imaging. Circulation 2002; 106: 50-6
18.
Tsutsui H, Uematsu M, Shimizu H, et al. Comparative usefulness of
myocardial velocity gradient in detecting ischemic myocardium by a
dobutamine challenge. J Am Coll Cardiol 1998; 31: 89-93
19.
Armstrong G, Pasquet A, Fukamachi K, et al. Use of peak systolic strain as
an index of regional left ventricular function: comparison with tissue Doppler
velocity during dobutamine stress and myocardial ischemia. J Am Soc
Echocardiogr 2000; 13: 731-37
116
20.
Greenberg NL, Firstenberg MS, Castro PL, et al. Doppler-derived
myocardial systolic strain rate is a strong index of left ventricular contractility.
Circulation 2002; 105: 99-105
21.
Kukulski T, Jamal F, D’Hooge J, et al.: Acute changes in systolic and
diastolic events during clinical coronary angioplasty: a comparison of regional
velocity, strain rate, and strain measurement. J Am Soc Echocardiogr 2002;
15: 1-12
22.
Hamilton WF, Rompf JH. Movements of the base of the ventricle and
relative constancy of the cardiac volume. Am J Physiol 1932; 102: 559-65
23.
Hoffman EA, Ritman EL. Invariant total heart volume in the intact thorax.
Am J Physiol 1985; 249: 883-90
24.
Feigenbaum H, Zaky A, Nasser WK. Use of ultrasound to measure left
ventricular stroke volume. Circulation 1967; 38: 1092-99
25.
Jones CJH, Raposo L, Gibson DG. Functional importance of the long axis
dynamics of the human left ventricle. Br Heart J 1990; 63: 215-20
26.
Pai RG, Bodenheimer MM, Pai SM, Koss JH, et al. Usefulness of systolic
excursion of the mitral anulus as an index of left ventricular systolic function.
Am J Cardiol 1991; 67: 222-4
27.
Gulati VK, Katz WE, Follansbee WP, et al. Mitral annular descent velocity
by tissue Doppler echocardiography as an index of global left ventricular
function. Am J Cardiol 1996; 77: 979-84
28.
C Pislaru, T.P. Abraham, M Belohlavek. Strain and strain rate
echocardiography. Curr Opin Cardiol 2002; 17: 443-54
29.
Galiuto L, Ignone G, DeMaria AN. Contraction and relaxation velocities of
the normal left ventricle using pulsed-wave tissue Doppler echocardiography.
Am J Cardiol 1998; 81: 609-14
30.
Edvardsen T, Gerber BL, Garot J, et al. Quantitative assessment of
intrinsic regional myocardial deformation by Doppler strain rate
echocardiography in humans: validation against three-dimensional tagged
magnetic resonance imaging. Circulation 2002; 106: 50-6
117
31.
Jamal F, Kukulski T, Strotmann J, et al. Quantification of the spectrum of
changes in regional myocardial function during acute ischemia in closed chest
pigs: an ultrasonic strain rate and strain study. J Am Soc Echocardiogr 2001;
4: 874-84
32.
Derumeaux G, Ovize M, Loufoua J, t al. Assessment of nonuniformity of
transmural myocardial velocities by color-coded tissue Doppler imaging:
characterization of normal, ischemic, and stunned myocardium. Circulation
2000; 101: 1390-5
33.
Weidemann F, Dommke C, Bijnens B, et al. Defining the transmurality of a
chronic myocardial infarction by ultrasonic strain-rate imaging: implications for
identifying intramural viability: an experimental study. Circulation 2003; 107:
883-8
34.
Anagnostopoulos PC, Belohlavek M, Dzeja P, et al. Strain-rate
echocardiography: a new quantitative tool for measurement of regional
myocardial function during experimentally altered myocardial energetics. J Am
Coll Cardiol 2001; 37(Suppl A): 443A
35.
Nagueh SF, Kopelen HA, Lim DS, et al. Tissue Doppler imaging
consistently detects myocardial contraction and relaxation abnormalities,
irrespective of cardiac hypertrophy, in a transgenic rabbit model of human
hypertrophic cardiomyopathy. Circulation 2000; 102: 1346-50
36.
Nagueh SF, Bachinski LL, Meyer D, et al. Tissue Doppler imaging
consistently detects myocardial abnormalities in patients with hypertrophic
cardiomyopathy and provides a novel means for an early diagnosis before and
independently of hypertrophy. Circulation 2001; 104: 128-30
37.
Dutka DP, Donnelly E, Palka P, et al. Echocardiographic characterization
of cardiomyopathy in Friedreich’s ataxia with tissue Doppler
echocardiographically derived myocardial velocity gradients. Circulation 2002;
102: 1276-82
38.
Takenaka K, Kuwada Y, Sonoda M, et al. Anthracycline-induced
cardiomyopathies evaluated by tissue Doppler tracking system and strain rate
imaging. J Cardiol 2001; 37(Suppl 1): 129-32
39.
Palka P, Lange A, Donnelly JE, et al. Differentiation between restrictive
cardiomyopathy and constrictive pericarditis by early diastolic Doppler
myocardial velocity gradient at the posterior wall. Circulation 2000; 102: 65562
118
40.
Lange A, Fleming AD, et al. Differences in myocardial velocity gradient
measured throughout the cardiac cycle in patients with hypertrophic
cardiomyopathy, athletes and patients with left ventricular hypertrophy due to
hypertension. J Am Coll Cardiol 1997; 30: 760-8
41.
Weidemann F, Eyskens B, Jamal F, et al. quantification of regional left and
right ventricular radial and longitudinal function in healthy children using
ultrasound-based strain rate and strain imaging. J Am Soc Echocardiogr 2002;
15: 20-8
42.
Chetboul V, Carlos Semperdano C, Gouni Vassiliki, et al. Ultrasonographic
assessment of regional radial and longitudinal systolic function in healthy
awake dogs. J Vet Intern Med 2006; 20: 885-93
43.
Wess G, Killich M, Butz, Hartmann K. Evaluaiton of the new tissue Doppler
imaging methods strain and strain rate in Doberman pinschers. Abstract
presented at the 16th ECVIM-CA Congress, Amsterdam, Holland, September
14 – September 16, 2006
44.
Weidemann F, Jamal F, Sutherland GR, et al. Myocardial function defined
by strain rate and strain during alterations in inotropic states and heart rate.
Am J Physiol Heart Circ Physiol 2002; 283: H792-99
45.
Egner B, Carr A, Brown S. Canine normal values. In Essential facts of
blood pressure in dogs and cats. A reference guide. BE Vet Verlag,
Babenhausen, 2003, pg. 12
46.
Stepien RL, Rapoport GS, Henik RA, et al. Comparative diagnostic test
characteristics of oscillometric and Doppler ultrasonographic methods in the
detection of systolic hypertension in dogs. J Vet Intern Med. 2003; 17(1): 6572.
47.
Cheitlin MD et al. ACC/AHA/ASE 2003 guideline update for the clinical
application of echocardiography: summary article: a report of the American
College of Cardiology/American Heart Association Task Force on Practice
Guidelines (ACC/AHA/ASE Committee to Update the 1997 Guidelines for the
Clinical Application of Echocardiography). Circulation. 2003; 108(9): 1146-62
48.
Quinones MA et al. ACC/AHA clinical competence statement on
echocardiography: a report of the American College of Cardiology/American
Heart Association/American College of Physicians-American Society of
119
Internal Medicine Task Force on Clinical Competence. Circulation 2003;
108(9): 1146-62
49.
Plumb DC. Dobutamine. In Plumb’s Veterinary Drug Handbook 5th Ed.
2005; pg. 272
50.
Plumb DC. Nitroprusside. In Plumb’s Veterinary Drug Handbook 5th Ed.
2005; pg. 566
51.
Plumb DC. Hetastarch. In Plumb’s Veterinary Drug Handbook 5th Ed. 2005;
pg. 388
52.
Margiocco ML, Bulmer BJ, Mosley CA, Sisson DD. Invasive validation of
Doppler-derived strain rate imaging in anesthetized healthy adult dogs.
Abstract presented at the 25th Annual ACVIM Forum, Seattle, WA, June 6 June 9, 2007
53.
Storaa C. Reproducibility and Interpretation in Tissue Doppler
Echocardiography, PhD dissertation. Karolinska Institutet, Division of Medical
Engineering. Stockholm, 26th July 2004
54.
Altman DG, Bland JM. Measurements in medicine: the analysis of method
comparisons studies. The Statistician 1983; 32: 307-31
55.
Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1986; 1: 307-10
56.
Bland JM, Altman DG. Applying the right statistics: analyses of
measurement studies. Ultrasound Obstet Gynecol 2003; 22: 85-93
57.
Bland JM, Altman DG. Comparing methods of measurement: why plotting
difference against standard method is misleading. Lancet 1995; 346: 1085-7
58.
Livers JJ. Some Limitations to Use of Coefficient of Variation. J of Farm
Economics 1942; 24: 892-5
59.
Calvert CA, Pickus CW, Jacobs GJ, et al. Signalment, survival, and
prognostic factors in Doberman pinschers with end-stage cardiomyopathy. J
Vet Intern Med 1997; 11: 323-6
60.
Perk G, Tunick PA, Kronzon I. Non-Doppler two-dimensional strain imaging
by echocardiography - from technical considerations to clinical applications. J
Am Soc Echocardiogr 2007; 20(3): 234-43