PRENATAL TISSUE VELOCITY IMAGING OF THE HEART A new approach to assess fetal myocardial function
Nina Elmstedt
Doctoral Thesis
Division of Medical Engineering
School of Technology and Health
KTH Royal Institute of Technology
ii
TRITA‐STH Report 2013:3 ISSN 1653‐3836
ISRN/KTH/STH/2013:3 ‐SE ISBN 978‐91‐7501‐705‐1
© Nina Elmstedt, Stockholm, 2013
ABSTRACT
The general aim of this thesis has been to evaluate color‐coded tissue velocity imaging (TVI) as an approach to developing a new, non‐invasive assessment method for fetal myocardial function. Such a method could hypothetically give early indications of fetal pathology, as myocardial dysfunction is often the consequence when the circulation tries to adapt to deteriorating situations. This would be beneficial in clinical decision making when evaluating fetal well‐being in a wide range of pregnancy associated conditions, to facilitate risk assessment and to monitor the benefit of therapeutic interventions.
TVI is an ultrasound technique that enables quantification of longitudinal myocardial motion with high temporal resolution, which is essential in the identification of fetal myocardial movements of short duration. Furthermore, the longitudinal motion is mainly determined by subendocardial fibers that usually become abnormal in the very early stages of cardiac dysfunction as they are sensitive to milder degrees of hypoxia. Thus, TVI has the potential to give early indications of impaired fetal myocardial function and hypothetically facilitate the detection of intrauterine hypoxia. Hypoxia is a common phenomenon of many pathological conditions in pregnancy, from which a substantial number of children either die or acquire permanent brain injury during delivery every year.
After having established optimal sampling requirements and ensured an acceptable reproducibility for TVI measurements of the fetal myocardium, normal reference values were determined feasible and sensitive enough to provide insight into maturational changes in myocardial function. This provided a foundation that should enable further investigations and was partly accomplished using the cardiac state diagram (CSD) to accurately time the myocardial events during a cardiac cycle according to the motion shifts of the atrioventricular plane.
The demonstrated results are promising and the general conclusion of this thesis is that TVI contributes to increasing the knowledge and understanding of fetal myocardial function and dysfunction. Used together with CSD this technique has great potential as an assessment method. However, further testing of the clinical potential is needed in larger study populations concerning the pathological or physiological questions at issue, and additional development of the method is required to render the method simple enough to be of potential aid in clinical practice.
DISSERTATION
The thesis is based on these appended papers
I. Temporal frequency requirements for tissue velocity imaging of the fetal heart. Nina Elmstedt, Britta Lind, Kjerstin Ferm‐Widlund, Magnus Westgren & Lars‐ Åke Brodin.
Ultrasound Obstet Gynecol 38, 413‐417 (2011).
II. Reproducibility and variability in the assessment of color‐coded tissue velocity imaging of the fetal myocardium. Nina Elmstedt, Britta Lind, Kjerstin Ferm‐Widlund, Magnus Westgren & Lars‐Åke Brodin.
J Biomed Graph Comput 3, 2 (2013).
III. Fetal cardiac muscle contractility decreases with gestational age: a color‐coded tissue velocity imaging study. Nina Elmstedt, Kjerstin Ferm‐Widlund, Britta Lind, Lars‐Åke Brodin & Magnus Westgren. Cardiovasc Ultrasound 10, 19 (2012).
IV. Reference values for fetal tissue velocity imaging and a new approach to evaluate fetal myocardial function. Nina Elmstedt, Jonas Johnson, Britta Lind, Kjerstin Ferm‐Widlund &
Magnus Westgren, Lars‐Åke Brodin. Submitted (2013).
DIVISION OF WORK BETWEEN AUTHORS
I. NE performed the offline analysis, participated in the interpretation of data and drafted the manuscript. KFW performed the ultrasound image acquisition and revised the manuscript.
BL, MW and LÅB participated in the design of the study, the interpretation of data and revised the manuscript critically. All authors read and approved the final manuscript.
II. NE and BL performed the offline measurements. NE drafted the manuscript. KFW performed the ultrasound image acquisition and revised the manuscript. BL, MW and LÅB participated in the design of the study, the interpretation of data and revised the manuscript critically. All authors read and approved the final manuscript.
III. NE performed the offline measurements, statistical analysis and drafted the manuscript.
KFW performed the ultrasound image acquisition and revised the manuscript. BL and LÅB participated in the design of the study, the interpretation of data and revised the manuscript critically. MW drafted the Discussion section, participated in the interpretation of data and revised the manuscript critically. All authors read and approved the final manuscript.
IV. NE, JJ, BL, KFW, LÅB and MW participated in the design of the study and the interpretation of data. NE performed the offline measurements, statistical analysis and drafted the manuscript. JJ performed the software development, defined the timing of the mechanical events during a cardiac cycle and revised the manuscript. KFW performed the ultrasound image acquisition and revised the manuscript. BL, LÅB and MW revised the manuscript critically. All authors read and approved the final manuscript.
OTHER SCIENTIFIC CONTRIBUTIONS
Publications
The cardiac state diagram as a novel approach for evaluation of pre‐ and post ejection phases of the cardiac cycle in asphyxiated fetal lambs. Wågström E, Johnson J, Ferm‐Widlund K, Elmstedt N, Liuba K, Lind B, Brodin L.Å, Lundbäck S, Westgren M. Ultrasound in Medicine and Biology (2013) Å
Conference contributions
Temporal frequency requirements for tissue velocity imaging of the fetal heart. Elmstedt N, Lind B, Ferm‐Widlund K, Westgren M and Brodin L.Å. EuroEcho, Copenhagen (2010)
Samplingsfrekvenskrav för vävnadsdopplermätningar av fosterhjärtat. Elmstedt N, Lind B, Ferm‐
Widlund K, Westgren M and Brodin L.Å. Medicinteknikdagarna, Umeå (2010)
Fetal heart contractility. Elmstedt N, Lind B, Ferm‐Widlund K, Brodin L.Å and Westgren M. Frontiers of CardioVascular Biology, London (2012)
Fetal tissue velocity imaging. Elmstedt N, Johnson J, Lind B, Ferm‐Widlund K, Westgren M and Brodin L.Å. The 2nd International Congress on Cardiac Problems in Pregnancy, Berlin (2012)
Prenatal vävnadsdoppler. Elmstedt N, Lind B, Johnson J, Ferm‐Widlund K, Westgren M and Brodin, L.Å.
Mediceinteknikdagrana, Lund (2012)
Color‐coded tissue velocity imaging of the fetal heart. Elmstedt, N. Philipsdagen Lund (2012)
PREFACE AND ACKNOWLEDGEMENTS
PREFACE
The work presented in this thesis is the result of a five‐year collaboration between the Department of Medical Engineering at KTH Royal Institute of Technology and the Centre of Fetal Medicine at Karolinska University Hospital Huddinge. The past two years this research project has gratefully been receiving support from the Swedish Heart‐Lung foundation.
The academic dissertation is presented at the School of Technology and Health, with permission from KTH Royal Institute of Technology, to be publicly reviewed in lecture hall 4‐221, Alfred Nobels Allé 12, Huddinge, Sweden, on Tuseday May 7 2013 at 13:00 for the degree of Doctor of Technology.
ACKNOWLEDGEMENTS
I would like to thank all the women who participated as volunteers in this research project. I would like to thank my supervisor Lars‐Åke Brodin for enthusiastically sharing his research ideas, as well as a slapstick joke too many. You never seem to worry and always look ahead. Sometimes too far ahead to remember our meetings, but your opinion has always been worth waiting for. Thank you for introducing me to a very interesting research area. I have learnt a lot from your expertise in both clinical medicine and medical engineering. I would like to thank my co‐supervisor Britta så är livet Lind for your excellent teachings in ultrasound imaging as well as gender equality, and your meticulous corrections (probably the only time I will use that word appreciatively). You also keep a much needed critical eye on our research, as well as on the professor. Thank you for helping me and answering all of my questions, even throughout that first year when you did not know you were my supervisor (Lars‐Åke does not always share all of his brilliant research ideas). Your thoughtfulness and support have meant a lot to me, especially during the last couple of months. I would like to thank Magnus Westgren for an inspiring collaboration and additional supervision. Even though you haven’t been my supervisor on paper there was never any doubt thereof, always giving me valuable insights and feedback. It has been a pleasure discussing and writing papers with you. I hope I will have the good fortune to continue doing so in the future. I would like to thank Kjerstin Ferm‐Widlund for performing the ultrasound investigations.
Without your excellence in image acquisition our collaboration would not have been as fruitful (maybe
this should have been stated as a requirement or lack thereof a limitation), nor would my computer
screen light up with such beautiful and exotic animal photos. I would like to thank Jonas Johnson for
trying to make sense of the dynamic adaptive displacement pump also referred to as the heart. Thank
you for always making time in your busy schedule to present physiology from a mechanical perspective
and custom‐making fetal applications, as well as being a friend over lunches and conference dinners. I
would like to thank those who have read and commented on this work in addition to Lars‐Åke, Britta,
Magnus, Kjerstin and Jonas. I would like to thank Peta Sjölander and Björn‐Erik Erlandsson for
language editing and proofreading. I would like to thank Erik Elmstedt, Tanya Jukkala and Viktor
Söderlind for additional readings ‐ your comments have been most valuable. I would like to thank
Staffan Larsson for teaching me some valuable settings in Photoshop and for trying to make the best
of my most blurred charts. I would like to thank Ebba Åkerman for the beautiful illustration of the fetal
circulation (no need for Photoshop there).
WORDS OF GRATITUDE
I would like to thank all of my colleagues for making the third floor at ANA 10 a nice, supportive and most days a fun place to work, three of them in particular. Mats Nilsson ‐ the co‐worker who took the lead in this year’s dissertation race. You have been several much needed steps ahead, giving good advice and pushing the limits further than ever before. Thank you for making it possible for me to “att göra en Mats”, i.e. a digital spikning. Frida Lindberg ‐ the co‐worker who became a friend who turned out to be a relative! You turn all water under the bridge of Flemingsberg into wine. Thank you for making even the worst of things laughable in a trice. I look forward to a lifelong friendship, lots of future laughter and all the bottles of wine that we’ll be sharing as Doctors this summer! Mattias Mårtensson ‐ the co‐worker who to my surprise always comes up with yet another conversation topic beyond my imagination. The ”hollywood” and ”lyxfällan” lunches became dinners, and you become a dear friend.
Thank you for giving me your perspective on things. I look forward to future dinner conversations.
Furthermore, I wish to express my sincerest gratitude to my friends! In particular the ones who have been there listening in on my struggles the last couple of months (in order of appearance, more or less) Josefina witty and wise, Zara always make sense, Marianne the sabering sniper, Li I would choose you anytime over that twenty‐forth lady, Stina observant to the details of disorder, Maria oooooooh stämning!, Emma fashioned with joy, Douglas absolutely fabulous, Ebba the universe most optimistic little helper, Marie‐Louise enigmatically thoughtful, Siri yes is more, Charlotta meget håpefuld, Signhild planning the unscheduled working model, Anna New York’s Eve, Josef the KTH mentor, Noemi even makes unwashed hair stylish, Camilla that 70th, Fredrik time is pastry, Katriina no logo, Anna trendsetting ***dancer, Asha lovely resolute, Therese Sweetheart, Kicki miss knock‐out, Helena skäggbiffen Bettan, En genuinely curious, Alex brilliantly warm, Ellen nakedly energetic, Saga bubbling aware, Tanya peace and love, Vilma the athlete, Teresa getting the hen of things, Cissi the perfect match, Jill a thousand words. Thank you for being my friends! A special thanks to Zara and Henke for hosting the dinner and celebrations at Cloud Nine, the best restaurant in town.
Finally, the greatest of thanks to my family! My parents Anne and Erik, and my brother Petter for always being there for me, for their company and encouragement throughout half a lifetime of enjoyable dinners, numerous bottles of champagne, festivities, skiing, shopping sprees, theaters, amusing conversations and love, and my boyfriend Viktor, for literally keeping me on track wherever we go, making every day a great one, filled with laughter and love despite arguments and the basic boring stuff, and for the anger management I’ve needed the last couple of months. Big smiley face. A special thanks to Viktor’s family, for their warm welcoming and additional support.
Nina Elmstedt
Stockholm, April 2013
TABLE OF CONTENTS
ABSTRACT ... iii
DISSERTATION ... v
DIVISION OF WORK BETWEEN AUTHORS ... vii
PREFACE AND ACKNOWLEDGEMENTS ... ix
INTRODUCTION ... 1
AIMS ... 1
THE FETAL CARDIOVASCULAR SYSTEM ... 2
CARDIOCASCULAR DEVELOPMENT ... 2
FETAL CIRCULATION ... 2
ECHOCARDIOGRAPHY ... 4
COLOR‐CODED TISSUE VELOCITY IMAGING ... 5
CARDIAC STATE DIAGRAM ... 5
METHODOLOGY ... 7
STUDY SUBJECTS ... 7
IMAGE AQUISITION ... 7
OFFLINE ANALYSIS AND DATA PRESENTATION ... 8
RESULTS ... 11
TEMPORAL FREQUENCY REQUIREMENTS ... 11
REPRODUCIBILITY AND VARIABILITY ... 11
GESTATIONAL AGE SPECIFIC REFERENCE VALUES ... 12
DISCUSSION ... 17
CONCLUSIONS ... 21
ABBREVIATIONS ... 23
OTHER SCIENTIFIC CONTRIBUTIONS ... ix
REFERENCES ... 25
1
INTRODUCTION
A substantial number of children either die or acquire permanent brain injury during delivery every year as a cause of intrauterine fetal hypoxia ‐ a pathological condition in which the fetus is deprived of adequate oxygen supply. Additionally, oxygen deprivation is a common phenomenon in many pathological conditions in pregnancy
1and affects fetal well‐being in a wide range of prenatal complications. The fetal circulation is a very flexible circulatory system that responds to a variety of conditions and prenatal challenges
2‐4. Oxygen deprivation is met by augmentation of coronary blood flow
5and induces a hemodynamic adaptation
6by redistributing the arterial circulation to maintain an adequate oxygen supply to both the brain and the heart
4. Myocardial dysfunction is often the consequence of such cardiovascular adaptation and can thus be an early sign of fetal pathology
4,7,8. Consequently, accurate assessment of fetal myocardial function could be of crucial importance when evaluating fetuses at risk or suffering from intrauterine fetal hypoxia
1,3,9‐13.
Quantitative assessment of fetal cardiac function has traditionally been based on echocardiography measurements of ventricular dimensions or blood flow derived indices of fetal hemodynamics
9,14‐17. At their best, these measurements are indirect markers of myocardial function
9,18, and functional abnormalities are often not detectable until late in the disease process when global ventricular dysfunction is obvious
1,17. Recent developments in adult echocardiography have enabled the quantification of regional myocardial function using color‐coded tissue velocity imaging (TVI). This technique facilitates quantitative assessment of longitudinal myocardial motion with high temporal resolution, which is essential to identify the myocardial movements of short duration, an issue that becomes even more important when taking into consideration the high fetal heart rate. Furthermore, the longitudinal motion is mainly determined by subendocardial fibers
17,19that usually become abnormal in the very early stages of cardiac dysfunction
17,20,21as they are more distal from the coronary arteries, and the blood has to overcome tissue pressure passing through the myocardium. Thus, they are more sensitive to milder degrees of hypoxia ‐ reflecting an early reduction in coronary flow.
TVI should have the potential to detect early and subtle changes also in fetal myocardial performance, hypothetically giving early indications of impaired fetal myocardial function and intrauterine hypoxia.
This would be beneficial in clinical decision making when evaluating fetal well‐being in a wide range of pregnancy associated conditions, to facilitate risk assessment and to monitor the benefit of therapeutic interventions.
AIMS
The general aim of this thesis was to evaluate TVI as an approach to developing a new, non‐invasive assessment method for fetal myocardial function, with the specific aims to:
‐ establish sampling requirements for TVI measurements of longitudinal myocardial peak velocities and time intervals during the fetal cardiac cycle.
‐ assess the reproducibility and intra‐ and inter‐observer variability for offline analysis as well as echocardiography investigations using TVI.
‐ determine gestational age specific reference values for TVI measurements of longitudinal myocardial peak velocities, time intervals and displacement during the cardiac cycle.
‐ evaluate fetal myocardial function and gestational age related changes of systolic and diastolic
performance, using the cardiac state diagram (CSD) to further study the mechanical timing of events.
THE FETAL CARDIOVASCULAR SYSTEM
CARDIOCASCULAR DEVELOPMENT
In the human embryo, cardiovascular development occurs between 3 and 6 weeks after ovulation
22, and the heart tube completes its process of septation, valve development, and conotruncal division at 10 weeks
23. Due to the heart’s ability to generate pressure and move blood before the circulation is fully formed, the earliest heart beats start within 4 weeks of gestation
24. In this early circulation interchamber wall thickenings function as valves
25.
The embryo enters the fetal stage at the beginning of the eleventh week of gestation. By this week all organs are present and the fetus is typically 3 cm in crown to rump length and weighs about 8 grams.
The position of the fetal heart within the chest is similar to that in later gestation
26, still the orientation of the heart in the fetus differs from that in postnatal life. The long axis of the left ventricle is more horizontal in the fetus than in the neonate
27. The myocardium continues to grow by cell division until birth and a continued growth thereafter comes with cell enlargement
2. This means that all adult working myocardial cells are the same as those that were present during the prenatal period
25.
FETAL CIRCULATION
The fetal heart differs both structurally and functionally from the postnatal heart. In fetal circulation, unlike postnatal circulation, the ventricles pump in parallel, i.e. the left and right ventricles both pump blood into the systematic circulation. The dominant right ventricle ejects approximately 60% of the combined ventricular output
22, with less than 10% passing through the lungs
28. The lungs, as well as the kidneys and gastrointestinal organs, do not begin to function fully until after birth. Instead the fetus obtains oxygen and nutrients and eliminates carbon dioxide and other wastes through the umbilical cord and placental circulation
29. The heart and the placenta are the main organs determining fetal hemodynamics. Essential distributional arrangements, making fetal circulation adaptive and flexible, are enabled through three shunts connecting the two parallel circulations that unite these organs, i.e.
foramen ovale, ductus venosus and ductus arteriosus. The fetal circulation is illustrated in Figure 1.
The part of the oxygenated blood returning from the placenta through the umbilical vein that does not perfuse the liver, is shunted through ductus venosus that connects to the inferior vena cava (IVC) at its inlet to the heart, opens the foramen ovale and enters the left atrium. Since this blood is well oxygenated and loaded with higher kinetic energy than the blood it mingles with in IVC, it will be this blood that predominantly enter the left atrium
2. Conversely, the deoxygenated blood from IVC, returning from the lower body regions, enters the right atrium. The atrium fills continuously during the cardiac cycle, causing a gradual increase in arterial pressure that once it exceeds ventricular pressure causes the atrioventricular (AV) valves to open and blood to enter the ventricles. Ventricular filling can be divided into the sub phases of rapid filling, slow filling (diastase) and atrial contraction. As the ventricles become filled with blood the intraventricular pressure rises and causes the AV‐valves to close. The transition phase between ventricular filling and ventricular ejection, during which the AV‐
valves will close and the semilunar (SL) valves will be about to open, is referred to as the pre ejection
phase. During the end of this phase the interventricular pressure increases most rapidly and exceeds
the pressure in the aorta and the pulmonary artery, causing the SL‐valves to open and blood to be
ejected into the systematic circulation. Subsequently the pressure in the aorta and the pulmonary
artery exceeds the intraventricular pressure and the SL‐valves close. The transition phase between
ventricular ejection and ventricular filling, when the valves are closed and the myocardium starts to
relax, is referred to as the post ejection phase. The ventricular myocardium relaxes, the intraventricular
pressure drops rapidly and a new cardiac cycle begins. Both the left and right ventricle eject blood into
the systematic circulation as most of the blood pumped into the pulmonary trunk from the right
ventricle bypasses the nonfunctioning fetal lungs through the ductus arteriosus, a vessel that connects the pulmonary trunk with the aorta. However, the distribution between the two ventricular outputs in the fetal body differs. The upper body and coronary circulation is perfused by the left ventricle, whereas the lower body and placenta are largely perfused by the right ventricle.
From 18 weeks of gestation to term, fetal heart rate normally ranges between 130 and 160 beats per minute (bpm). This is considerably higher than the average heart rate at rest in adults (about 70 bpm) and corresponds to a cardiac cycle length of about 375‐460 ms. The interrelationship between the different phases of the cardiac cycle also differ between prenatal and postnatal life. During ventricular filling the diastase, also referred to as the slow filling phase, is almost completely absent, and the pre and post ejection phases constitute 15‐20 % of the total cardiac cycle in the fetus, compared to 5% of the total cardiac cycle in the adult
30.
Figure 1 The fetal circulation. Illustration by Ebba Åkerman.
ECHOCARDIOGRAPHY
Ultrasound is an oscillating sound pressure wave with a frequency above 20 kHz, i.e. sound waves to high in frequency to be detected by the human ear. Diagnostic ultrasound is in the range of 1‐12 MHz.
Sound waves are longitudinal mechanical waves of compression and decompression of the medium through which they propagate, and ultrasound imaging is based on the fact that these waves are partly reflected at the boundary between two media with different acoustic impedance. Acoustic impedance is the product of the density and the speed of sound in the medium. Ultrasound cardiography – later renamed echocardiography – was first applied in Lund in 1953 when Inge Edler and Helmuth Hertz made the first non‐invasive description of moving structures within the heart
31. This one‐dimensional imaging of motion over time is referred to as M‐mode and was also the onset for fetal echo‐
cardiography when Allan et al. made the first assessment of moving fetal cardiac structures in 1982
32. In echocardiography, ultrasound is transmitted and received through conversion from electricity to mechanical waves and vice versa using different transducers also referred to as probes. Different anatomical tissue structures will reflect signals with different amplitude. The amplitude of the reflection depends on the difference in acoustic impedance between the anatomical tissue structures, the reflecting tissue surface (i.e. wall and fiber direction in the myocardium) and the angle of incidence.
As the propagation speed of sound in tissue is fairly constant, ranging from 1450 to 1585 m/s
33, a fixed mean value in the range of 1540 – 1560 m/s is used by most manufacturers, and the distance to the origin of reflection can be determined by measuring the time elapsed from transmission to reception.
Structures at different depths can be resolved accordingly, where the intensity of the reflected signal is displayed as brightness (B‐mode). This laid the foundation for M‐mode imaging, where multiple B‐
mode images are recorded, frame by frame, along a time axis to simulate continuous linear waves as the depth to the reflected objects vary over time. Later on, in the 1980s, real time two‐dimensional echocardiography, built up by several successive B‐mode lines, was introduced into clinical practice.
This technique enabled cross‐sectional images of cardiac structures, and is still used alone or superimposed with Doppler derived velocity data. Doppler imaging utilizes the fact that the frequency of sound is shifted when reflected against a moving object (Doppler shift) and that the size of the shift is directly related to the velocity of the moving object. The velocity is recorded in one dimension, in the direction towards and away from the transducer, and color‐coded in red and blue, respectively.
Quantitative assessment of fetal cardiac function has hitherto been based on cross‐sectional M‐mode measurements of ventricular dimensions or Doppler derived blood flow measurements. However, at their best, these are indirect markers of myocardial function
9,18, and functional abnormalities are often not detectable until late in the disease process when global ventricular dysfunction is obvious
1,17.
For assessment of left ventricular function in adults, the ejection fraction by the modified Simpson’s
rule has been considered the gold standard. However, to delineate the endocardial borders could be
rather cumbersome. Instead, based on the hypothesis that the heart pumps with a reciprocating
motion, the importance of the longitudinal motion of the AV‐plane in terms of heart function was
initially described by Lundbäck in 1986
34. As an alternative approach the AV‐plane displacement was
correlated to global left ventricular function
35and its measurement has become well‐established in
adult cardiology
35‐37. The first measurements were made using M‐mode imaging. Since then,
developments in echocardiographic imaging have enabled the quantification of regional longitudinal
myocardial motion using tissue Doppler, where the displacement of the AV‐plane can be evaluated by
measuring basal myocardial velocities as the time integral of the velocity describes the distance of the
movement relative to the transducer, and basal velocities reflect the myocardial long‐axis shortening
from base to apex
38‐40.
COLOR‐CODED TISSUE VELOCITY IMAGING
Tissue Doppler imaging is a technique for direct analysis of myocardial movement, initially described by Satomura et al. in 1956
41, re‐introduced by Isaaz et al. in 1989
42, and reported feasible in the fetus by Harada et al. in 1999
43. This technique is based on the fact that the velocity of moving blood is higher than the velocity of moving tissue. Consequently, there will be a lower Doppler shift in the myocardial wall than in the circulating blood. Furthermore, the tissue signal will have stronger amplitude because of more scatter returning from blood than from tissue
44. Technically, low pass filtering or gain damping could be applied to calculate the tissue velocities, filtered out when calculating the velocity of the blood. Color‐coded estimates of tissue velocity were introduced by McDicken et al. in 1992
45, and multiple line acquisition has since provided this technique with high temporal resolution (>200 frames/s).
Using color‐coded tissue velocity imaging (TVI), the myocardial tissue velocity is determined by estimating the phase shift using auto‐
correlation
44,46, rather than determining the Doppler shift directly by frequency analysis. TVI enables complete mapping of all velocities within the imaged myocardium using high temporal resolution. This makes it possible to detect the rapid cardiac movements of short duration and to calculate mean tissue velocity in any pixel of the acquired image sequence ‐ allowing comparison of several segments within the same cardiac cycle. TVI is already a well‐established method for analysis of longitudinal myocardial motion and deformation in adult cardiology
44, providing quantitative assessment of both global and regional function.
An example displaying the longitudinal myocardial tissue velocity curve extracted from the basal segment of fetal inferoseptum can be seen in Figure 2. Amongst the marked phases, the most commonly used measurements include the peak velocities of ventricular ejection, rapid filling and atrial contraction. Cardiac time intervals have also been evaluated as a measure of cardiac performance, where the Tei index (defined as the sum of pre and post ejection divided by the time of ventricular ejection) has been described as a marker of fetal cardiac dysfunction
13,47‐51.
CARDIAC STATE DIAGRAM
The longitudinal myocardial motion of the AV‐plane during a cardiac cycle can be expressed as coordinated events in a cardiac state diagram (CSD). The CSD is a novel visualization tool that provides quantitative analysis of the timing of mechanical events in the cardiac cycle, displaying the different time intervals coordinated in a circular diagram. This tool present comparative data of systolic and diastolic performance to provide a more complete overview of the cardiac function
52based on the hypothesis that the heart’s pumping and auto regulating function works with a reciprocating motion, according to a pump‐principle today recognized as the Dynamic Adaptive Piston Pump (DAPP)
34,53.The feasibility and potential use of the CSD has previously been demonstrated in a study of non ST‐
elevation myocardial infarction in adults
54and a study of experimental hypoxia in fetal sheep
55.
Figure 2 An example displaying the longitudinal myocardial tissue velocity profile extracted from the basal segment of inferoseptum using TVI, marked with the most common phase
METHODOLOGY
STUDY SUBJECTS
Characteristics of the studied fetal populations are presented in Table 1. Ethical approval was obtained from the regional ethics committee in Stockholm, Sweden. All women gave their informed consent to participate and had been referred to the centre of fetal medicine at the Karolinska University Hospital Huddinge during 2008‐2012.
Paper I and Paper II evaluated TVI with regard to temporal frequency and observer variability; hence there was no need to account for the indication of the ultrasound investigation in these studies. Most of the women included had experienced a normal pregnancy at the time of investigation while others were investigated because of different complications such as post‐term pregnancy, diabetes mellitus or intrauterine growth restriction. Paper III and Paper IV evaluated fetal myocardial function and gestational age specific characteristics; thus all women enrolled in these studies were healthy and experienced a normal singleton pregnancy. Gestational age was determined according to ultrasound in the 16
thto 18
thweek of gestation.
IMAGE AQUISITION
Tissue Doppler echocardiography recordings were obtained with GE Vivid‐i equipment (GE Vingmed, Norway), shown in Figure 3, using a 3S‐RS transducer. All recordings were performed with the transducer positioned perpendicular to the apex, keeping the angle of insonation as small as possible, to provide a view of the fetal myocardium equivalent to an apical four‐chamber view. The four‐chamber view is considered the basic view and standard of care
23and can be seen in Figure 4. In Paper I the 2D and TVI sector widths were minimized to obtain the highest possible frame rate, in most cases displaying only inferoseptum. In Papers II – IV TVI recordings were acquired with frame rate >200 frames/s, which in Paper I was established as necessary for adequate reconstruction of TVI data for the fetal myocardium
56. The TVI recordings were stored as cineloops of five to 10 consecutive cardiac cycles for offline analysis using EchoPAC (PC’08, GE Vingmed), GHLab (Gripping Heart AB, Sweden) and MATLAB (R2007b, MathWorks, USA).
Table 1 Characteristics of the studied fetal populations.
Paper Number of subjects
Sex
female/male GA GA at delivery MA BW
I 30 14/16 34 (23‐41) 39 (30‐42) 29 (19‐39) 3355 (2060‐4730)
II 21 12/9 32 (27‐41) 39 (35‐42) 30 (22‐37) 3430 (2060‐4730)
III 55 25/30 32 (18‐42) 40 (34‐42) 30 (18‐41) 3650 (2406‐4900)
IV 125 62/63 30 (18‐42) 40 (34‐42) 30 (19‐46) 3525 (2538‐4815)
GA, gestational age (weeks); MA, maternal age (years) and BW, birth weight (g) presented as median (range).
Figure 3 Ultrasound scanner, Vivid‐i (Courtesy of GE Healthcare).
The TVI recordings were imported to EchoPAC, which is software that allows offline analysis of ultrasound images obtained with GE ultrasound systems. Longitudinal myocardial tissue velocity was extracted from a fixed region of interest (ROI), placed above the AV plane
57in the basal segment
58of the myocardial wall during the end of systole. The ROI was set between 1‐3 mm, depending on image characteristics such as thickness of the myocardial wall and interference from valve motion or fetal/maternal movement, and was maintained in the ventricular myocardium, not entering the atrium during any phase of the cardiac cycle.
The mean velocity within the ROI is extracted and displayed over time, as can be seen in a recording from Paper IV shown in Figure 4, where the longitudinal myocardial velocity was extracted from inferoseptum as well as the anterolateral and right ventricular free wall. In Papers II and III the offline analysis were made directly in EchoPAC. In Papers I and IV the velocity data, stored as text files, was imported to additional software for analysis. MATLAB was used in Paper I to enable a gradual reduction in frame rate. GHLab was used in Paper IV to enable identification of the different time intervals in the absence of a concurrent ECG, as well as to generate CSD’s according to DAPP‐technology.
Measured variables Myocardial time intervals
The cardiac cycle was divided into six main phases: atrial contraction, pre ejection, ventricular ejection, post ejection, rapid filling and slow filling, as displayed in Figure 5 (a). The cardiac cycle is traditionally divided into a phase of ventricular ejection (systole) and a phase of ventricular filling (diastole), where systole comprises pre ejection and ventricular ejection, and diastole comprises post ejection, rapid filling, atrial contraction and slow filling. In Papers I ‐ III the time intervals were identified based on the zero axis crossing points of the velocity curve. This is usually carried out in combination with reference points in the ECG signal, and the occasional anatomical M‐mode image, in order to acquire a precise electrophysiological timing. This is not possible when screening the fetus, which precludes accurate timing of the mechanical events. In Paper IV time intervals were measured according to DAPP‐
technology, where the different phases, detected using GHLab, are defined based on the motion shifts of the AV‐plane. This makes identification of the time intervals possible without a concurrent ECG signal. Using GHLab, the velocity, acceleration and displacement curves can either be displayed separately, as in Figure 5, or overlap to enable identification of the different phases according to the changes from static to dynamic work.
Figure 4 A TVI recording performed with the transducer positioned to provide a view of the fetal myocardium equivalent to an apical four‐chamber view. Lower left: gray scale. Upper left: superimposed with color‐coded tissue velocity data.
Right: the longitudinal velocity profile extracted from ROI’s in the basal segment of inferoseptum (yellow), the anterolateral (red) and right ventricular free wall (turquoise).
Myocardial velocities
Longitudinal myocardial peak velocities during atrial contraction (a’), pre ejection (pre’
pos,pre’
neg), ventricular ejection (s’), post ejection (post’
pos, post’
neg) and rapid filling (e’) were measured. The velocity during pre and post ejection show a biphasic motion pattern, but single phase movements occur as well (more common during pre ejection). In Paper IV, the velocity during the slow filling phase was also measured.
Atrioventricular displacement
To estimate the AV‐piston displacement, i.e. the distance covered during ventricular ejection, the time integral of the myocardial velocity curve was measured. In Paper II this was calculated from an average of two or three of the heart cycles stored in the cineloops in EchoPAC and in Paper IV this was calculated automatically using GHLab post processing software. According to DAPP‐technology the term AV‐piston was used in Paper IV, and henceforth, instead of the conventional term AV‐plane, better describing the cardiac mechanics.
Statistical analysis
The results in Paper I were expressed as percentage with a 5% deviation. In Paper II the agreement of measurements was tested and illustrated by using Bland‐Altman plots
59and reproducibility and intra‐
and inter‐observer variability were expressed as a standard error of the mean. The data acquired in Papers III and IV were correlated using linear regression analysis and the results were expressed as mean values with a 95% confidence interval or as median values within a range. The Pearson correlation coefficient (r‐value) was used to examine their relationship. For the additional observations non‐parametric hypothesis testing was applied. The main part of the statistical analysis was performed using SPSS (PASW Statistics 18) and p‐values lower or equal to 0.05 were considered significant.
Figure 5 An example displaying the different time intervals during a cardiac cycle based on DAPP‐
technology. The velocity (a), acceleration (b), displacement (c) and the cardiac state diagrarm (CSD) (d) extracted from basal inferoseptum during the same cardiac cycle. The horizontal yellow line is the zero axis.
P1 Atrial contraction
P2 Pre ejection P3 Ventricular ejection P4 Post ejection P5 Rapid filling P6 Slow filling
RESULTS TEMPORAL FREQUENCY REQUIREMENTS
Paper I established optimal sampling requirements for TVI measurements of the fetal myocardium, and determined that a decrease in frame rate for the acquisition of data was accompanied by a loss of both temporal and velocity information. At low frame rates, all systolic and diastolic velocities were underestimated, and for the measured time intervals, there was a clear tendency to underestimate pre and post ejection, as well as a clear tendency to overestimate ventricular ejection, rapid filling and atrial contraction. An acceptable ≤5% deviation corresponded to measurements obtained at above 150–200 frames/s. The frame rate requirements for the measured peak velocities and time intervals during a cardiac cycle are presented in Table 2.
REPRODUCIBILITY AND VARIABILITY
Paper II assessed reproducibility and intra‐ and inter‐observer variability for offline analysis as well as echocardiographic investigations using TVI. The coefficients of variation for the measured systolic and diastolic peak velocities and time intervals during a cardiac cycle are stated in Table 3. The variation of reproducibility and inter‐observer variability for the echocardiography investigation ranged from 1.5%
to 9.8%. The duration of ventricular ejection and the peak velocity of rapid filling (e’) being the most robust variables measured. The variation of intra‐ and inter‐observer variability for the offline analysis ranged from 1.2% to 10.4%. Least robust were the events of shortest duration, including pre and post ejection. However, the coefficient of variation did not exceed more than 10.
Table 2 Frame rate requirements for TVI measurements of the fetal myocardium.
Variable Frames/s
Time intervals
Atrial contraction > 200
Pre ejection > 200
Ventricular ejection > 175
Post ejection > 200
Rapid filling > 200
Peak velocities
Atrial contraction > 150
Pre ejection > 200
Ventricular ejection > 175
Post ejection > 200
Rapid filling > 175
Table 3 Coefficients of variation for the systolic and diastolic peak velocities and time intervals.
Variable Reproducibility E
Inter‐observer echocardiography E
Intra‐observer offline analysis E
Inter‐observer offline analysis E
Time intervals
Atrial contraction 3,0 2,9 2,4 9,6
Pre ejection 7,8 5,5 2,3 10,4
Ventricular ejection 2,0 2,3 1,3 3,0
Post ejection 9,3 6,5 2,9 7,7
Rapid filling 5,6 4,2 2,0 6,9
Peak velocities
Atrial contraction 2,0 3,1 1,8 3,6
Pre ejection 6,6 3,0 3,4 4,4
Ventricular ejection 4,2 1,6 1,2 5,0
Post ejection 9,8 8,4 7,4 8,8
Rapid filling 2,2 1,5 3,7 8,3
E, standard error of the mean (%), i.e. the coefficient of variation.
Measured variables and observed gestational age related changes
Paper III and Paper IV evaluated fetal myocardial function and gestational age related changes of systolic and diastolic performance using TVI. Heart rate (HR), e’/a’ ratio and AV‐piston displacement of interventricular septum were measured in both studies and the results, shown in Table 4, corresponded well to each other. HR decreased, while the e’/a’ ratio and AV‐piston displacement increased with gestational age. Additional results from Paper IV showed that the displacement was more prominent for the right ventricular free wall, as was the maturational rate for the e’/a’ ratio.
Paper IV described normal reference values for gestational age specific longitudinal myocardial TVI measurements, including peak velocities, time intervals and displacement, obtained from inferoseptum as well as the anterolateral and right ventricular free wall. The results demonstrated that the measurements were feasible and sensitive enough to yield insight into maturational changes in myocardial function. The longitudinal peak velocities of inferoseptum showed a linear increase with gestational age, as did the peak velocities of the anterolateral and right ventricular free wall, except for the peak velocity of post ejection. The duration of systole did not vary significantly with gestational age, while the duration of diastole showed a linear increase. During diastole, the duration of rapid filling and atrial contraction increased while the duration of the post ejection phase decreased. Multiple regression analysis showed that a combination of HR and gestational age enhanced the correlations for atial contraction and rapid filling. However, no significant relationship was observed between HR and the post ejection phase or the systolic durations. The myocardial time intervals and velocities measured during a cardiac cycle, as well as the Tei index and the e’/a’ ratio, extracted from inferoseptum, the anterolateal and right ventricular free wall, are presented in Table 5. The variables that significantly changed throughout gestation are presented as mean values, where the nominator represents the mean value at 18 weeks of gestation and the denominator represents the mean value at term. The variables that did not show any significant change with gestational age are presented as median values within a range.
Table 4 Gestational age related changes of heart rate, e’/a’ ratio and AV‐piston displacement.
Variable 18 weeks of gestation 38‐40 weeks of gestation
r‐value Mean (95 % CI)
Paper III
Heart rate (bpm) 146 (141 ‐ 152) 135 (128 ‐ 148) 0,502
e’/a’ ratio (IS) 0,56 (0,50 ‐ 0,57) 0,82 (0,73 ‐ 0,92) 0,688
AV disp (IS) 1,8 (1,3 ‐ 2,3) 2,8 (2,5 ‐ 3,1) 0,825
Paper IV
Heart rate (bpm) 147 (137 ‐ 154) 137 (133 ‐ 140) 0,401
e’/a’ ratio (IS) 0,54 (0,47 ‐ 0,62) 0,84 (0,73 ‐ 0,95) 0,316 e’/a’ ratio (RV) 0,45 (0,39 ‐ 0,46) 0,79 (0,53 – 1,00) 0,438
AV disp (IS) 1,5 (1,2 ‐ 1,7) 3,0 (2,6 ‐ 3,3) 0,615
AV disp (RV) 2,1 (1,6 ‐ 2,5) 4,7 (3,9 ‐ 5,4) 0,534
IS, interventricular septum; RV, right ventricular free wall; AV, AV‐piston displacement (mm);
bpm, beats per minute; r‐value, correlation coefficient; p < 0,001.
Paper III suggested a decrease in fetal cardiac muscle contractility with gestational age. This assumption was based on the observed longitudinal shortening seen in Figure 6, and the uniform development of left ventricle dimensions throughout gestation. Left ventricular length and width increased with 20 mm and 10 mm, respectively, from 18 weeks of gestation to term. AV‐piston displacement increased in compliance with left ventricular length. Nonetheless, the ratio of AV‐piston displacement and left ventricular length decreased as pregnancy advanced.
Table 5 The mean values from 18 weeks of gestation to term for the myocardial time intervals and velocities during a cardiac cycle, as well as the Tei index and e’/a’ ratio.
Variable Abbreviation Infeoseptum Anterolateral wall Right ventricular fee wall
Time intervals (ms)
Atrial contraction P1 67 / 83 (r = 0,344) 61 / 84 (r = 0,317) 81 / 94 (r = 0,214)
Pre ejection P2 23 (11 ‐ 56) 24 (9 ‐ 54) 26 (0 ‐ 76)
Ventricular ejection P3 180 (95 ‐ 277) 184 (126 ‐ 245) 182 (95 ‐ 303)
Post ejection P4 63 / 41 (r = 0,441) 61 / 36 (r = 0,353) 49 (17 ‐ 87)
Rapid filling P5 71 / 109 (r = 0,403) 74 / 100 (r = 0,393) 72 / 98 (r = 0,416)
Slow filling P6 8 / 110 (r = 0,287 ) 4 (0 ‐ 44) 3 (0 ‐ 36)
Tei index 0,42 (0,13 ‐ 0,64) 0,41 (0,17 ‐ 0,85) 0,41 (0,14 ‐ 0,89)
Peak velocities
(cm/s)
Atrial contraction a’ ‐2,4 / ‐3,9 (r = 0,387) ‐1,9 / ‐4,0 (r = 0,375) ‐4,0 / ‐6,2 (r = 0,421) Pre ejection (pos) pre’pos 0,4 / 1,1 (r = 0,427) 0,3 / 2,0 (r = 0,489) 0,7 / 3,0 (r = 0,553) Pre ejection (neg) pre’neg ‐0,1 (‐2,8 ‐ 1,5) ‐0,1 (‐1,3 ‐ 0,3) ‐0,1 (‐1,3 ‐ 3,7) Ventricular ejection s’ 1,5 / 2,9 (r = 0,566) 1,1/ 2,8 (r = 0,512) 1,7 / 4,7 (r = 0,599) Post ejection (neg) post’neg ‐0,3 (‐1,8 ‐ 1,0) ‐0,3 (‐4,2 ‐ o,9) ‐0,4 (‐6,9 ‐ 0,8) Post ejection (pos) post’pos 0,3 / 1,0 (r = 0,395) 0,3 (‐0,8 ‐ 2,8) 0,4 (‐5,0 ‐ 2,7) Rapid filling e’ ‐1,3 / ‐3,3 (r = 0,509) ‐1,2 / ‐3,1 (r = 0,528) ‐1,4 / ‐4,6 (r = 0,527)
Slow filling p6’ ‐0,5 / ‐1,3 (r = 0,373) ‐0,7 (‐4,9 ‐ 2,8) ‐1,9 (‐6,7 ‐ 3,5)
e’/a’ 0,54 / 0,84 (r = 0,316) 0,56 / 0,84 (r = 0,358) 0,45 / 0,79 (r = 0,438)
The variables that significantly changed throughout gestation are presented as mean values, where the nominator represents the mean value at 18 weeks of gestation and the denominator represents the mean value at term. The variables that did not show any significant change with gestational age are presented as median values within a range.
Figure 6 The longitudinal shortening observed in Study III. AV disp, AV‐piston displacement; LV, left ventricular. A linear regression analysis with 95 % CI (r=0,778, p< 0,001.)
In Paper IV the offline analysis was made using GHLab and the feasibility of the CSD, previously demonstrated by Larsson et al.
54and Wågström et al.
55, was confirmed ‐ introducing this technique as a potential aid in the evaluation of fetal myocardial function. Figure 7 presents two CSD plots generated from the velocity curve extracted from inferoseptum of (a) a fetus at 18 weeks of gestation and (b) a fetus at term. The different time intervals during a cardiac cycle are expressed as coordinated events in a circular diagram. Full circle represents one cardiac cycle.
Additional observations and an example of application
Using the CSD for diagnostic purposes should facilitate visual interpretation and the evaluation of the relationship between the different mechanical events during a cardiac cycle. The following preliminary results provide an example of which, where Figure 8 (a) displays the CSD of an intrauterine growth restricted (IUGR) fetus compared to (b) the CSD of a healthy fetus at 36 weeks of gestation. The duration of post ejection was significantly longer for the IUGR fetus (p=0.02). No differences were observed for the systolic durations, and accordingly the Tei index showed less significance (p=0.03).
IUGR was defined as an estimated fetal weight <‐23%.
Figure 7 Example displaying two cardiac state diagrams (CSD) generated from the velocity curve extracted from inferoseptum of (a) a fetus at 18 weeks of gestation and (b) a fetus at term.
P1 atrial contraction P2 pre ejection P3 ventricular ejection P4 post ejection P5 rapid filling P6 slow filling
Figure 8 Example displaying the cardiac state diagram (CSD) generated from the velocity curve extracted from inferoseptum of an intrauterine growth restricted fetus (IUGR) (a) compared to a healthy fetus (b) at 36 weeks of gestation.
P1 atrial contraction P2 pre ejection P3 ventricular ejection P4 post ejection P5 rapid filling P6 slow filling
Figure 9 displays the longitudinal shortening observed in the IUGR fetus in comparison with controls from Paper III. There was no significant difference in left ventricular length. Even so, the decrease in longitudinal shortening was more significant (p=0.01) than the decrease in AV‐piston displacement (p=0.03). The only significant difference observed amongst the myocardial velocities was the lower peak velocity during atrial contraction (a’) for the IUGR fetus. However, this did not significantly alter the e’/a’ ratio.
Figure 9 The longitudinal shortening observed in IUGR fetuses relative to controls from Study III, a linear regression analysis with 95 % CI. AV disp, AV‐piston displacement; LV, left ventricular.