Assessment and quantification of foetal electrocardiography and heart rate variability of normal foetuses from early to late gestational periodsb 1

Studies on fECG using electrodes placed on the foetal scalp have proven to be successful in identifying intrapartum foetal hypoxia and in reducing unnecessary operative interventions for

ASSESSMENT AND QUANTIFICATION OF FOETAL ELECTROCARDIOGRAPHY AND HEART RATE VARIABILITY OF NORMAL FOETUSES FROM EARLY

TO LATE GESTATIONAL PERIOD

ELAINE CHIA EE LING

(B.Sc (Hons.), NUS)

A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF

PHILOSOPHY DEPARTMENT OF PHYSIOLOGY NATIONAL UNIVERSITY OF SINGAPORE

2006

ACKNOWLEDGEMENTS

I would like to express my sincere gratitude to the following people, without whose contributions this thesis would not have been possible:

• A/Prof Ho Ting Fei for the invaluable guidance, advice and

encouragement that she has provided me throughout my time as her PhD student

• A/Prof William Yip for his helpful discussions and comments

• A/Prof Mary Rauff for her guidance and for kindly allowing me to

conduct the study at her clinic

• Dr Chang Ee-Chien and Gao Ping from the Department of Computer

Science for their help in the development of F-EXTRACT

• Dr Zhang Niu for her assistance in acquiring the intrapartum ECG

• A/Prof YC Wong, A/Prof PC Wong and A/Prof Roy Joseph for their

help in this study

• Mdm Heng Ye Yong for her assistance in various technical aspects of the

• The admin staff of Department of Physiology for their secretarial support

PUBLICATIONS

The following publications have resulted from the present study:

International Referred Journals:

1) Chia EL, Ho TF, Rauff M, Yip WCL Cardiac time intervals of normal fetuses

using noninvasive fetal electrocardiography Prenat Diagn 2005; 25(7): 546-52

2) Chia EL, Ho TF, Wong YC, Yip WCL Ventricular bigeminy misdiagnosed as fetal bradycardia by cardiotocography- the value of non-invasive fetal

electrocardiography J Perinat Med 2004; 32(6): 532-4

Conference Paper:

1) Chia EL, Ho TF, Rauff M, Yip WCL Cardiac time intervals of normal fetuses

using non-invasive fetal electrocardiography NHG Annual Scientific Congress;

October 9, 2004; Singapore

TABLE OF CONTENTS

ACKNOWLEDGMENTS……… i

PUBLICATIONS……… ii

TABLE OF CONTENTS……… iii

SUMMARY……… x

LIST OF TABLES……… xiii

LIST OF FIGURES……… xiv

LIST OF ABBREVIATIONS……… xvi

Chapter 1 The foetal electrocardiogram……… 1

1 Historical development of the foetal ECG……… 2

2 Measurement of the foetal ECG……… 4

2.1 Invasive techniques - foetal scalp electrodes……… 4

2.1.1 Clinical application of scalp foetal ECG………… 5

2.2 Non-invasive techniques - maternal abdominal electrodes… 6

2.2.1 Foetal ECG studies utilizing abdominal electrodes… 6 2.2.2 Technical difficulties with abdominal electrodes… 7

2.2.3 Vernix caseosa and abdominal foetal ECG signal… 8

2.2.4 Acquisition and processing of abdominal foetal ECG……… 11

2.2.5 Clinical application of abdominal foetal ECG…… 13

Chapter 2 The foetal ECG waveform……… 14

1 Morphology and time intervals of the foetal ECG……… 15

1.1 P wave……… 15

1.2 PR interval……… 18

1.3 QRS complex……… 21

1.4 QT interval……… 23

1.5 ST segment……… 25

1.6 T wave……… 25

2 Foetal ST segment and T wave- Animal studies……… 26

3 Foetal ST segment and T wave- Human studies……… 31

Chapter 3 Heart rate variability……… 35

1 Definition of heart rate variability……… 36

2 History of heart rate variability……… 36

3 Measurement of heart rate variability……… 37

3.1 Time-domain methods……… 38

3.2 Frequency-domain methods……… 40

3.3 Geometrical methods……… 44

3.4 Non-linear methods……… 47

4 Physiological significance of heart rate variability……… 48

4.1 Components of heart rate variability……… 49

4.2 Estimate of vagal activity……… 50

4.3 Estimate of sympathetic activity……… 51

4.4 Estimate of sympatho-vagal balance……… 52

5 Factors affecting HRV……… 52

6 Clinical application of heart rate variability……… 53

Chapter 4 Heart rate variability in the foetus……… 57

1 Development of the foetal heart rate……… 58

2 Regulation of the foetal heart rate……… 60

2.1 Regulation by the sympathetic nervous system……… 61

2.2 Regulation by the parasympathetic nervous system………… 61

2.3 Central control of foetal heart rate……… 62

2.4 Different maturation rates of the autonomic branches……… 63

3 Causes of foetal heart rate variability… ……… 64

4 Measurement of foetal HRV……… 65

5 Characteristics of foetal HRV……… 67

6 Factors affecting foetal heart rate variability……… 68

6.1 Gestational age……… 68

6.2 Hypoxia……… 68

6.3 Foetal activity……… 69

6.4 Foetal breathing movements……… 70

6.5 Drugs……… 71

7 Clinical significance of foetal heart rate variability……… 72

Chapter 5 Scope of study……… 75

1 Background……… 76

2 Hypotheses……… 77

3 Aims and objectives……… 77

Chapter 6 Materials and methods……… 80

1 Patient selection……… 81

1.1 Study subjects……… 81

1.2 Exclusion criteria……… 81

1.3 Patient withdrawal……… 81

2 Methodology……… 82

2.1 Foetal ECG acquisition procedures……… 82

2.2 Foetal ECG equipment description and operation………… 86

2.3 Measurement of foetal ECG parameters……… 91

2.4 Neonatal ECG acquisition and measurement……… 93

2.5 Foetal HRV measurement……… 94

2.6 Comparison of F-EXTRACT and Nevrokard HRV softwares……… 96

2.7 Correction of aberrant beats……… 97

2.8 Statistics……… 97

Chapter 7 Cardiac time intervals of healthy foetuses……… 98

1 2 3 4 5 6 Introduction……… 99

Study population……… 101

Method……… 103

Results……… 103

4.1 Success rates of foetal ECG recording……… 103

4.2 Cardiac time intervals and gestational age……… 105

4.3 Cardiac time intervals and gender……… 112

4.4 Intrapartum cardiac time intervals……… 112

Discussion……… 115

Summary……… 127

Chapter 8 Clinical application of foetal electrocardiography……… 128

1 Introduction……… 129

2 Case report of a foetus with premature ventricular contractions…… 129

2.1 Antenatal foetal ECG……… 129

2.2 Intrapartum foetal ECG……… 136

3 Discussion……… 138

Chapter 9 Development of a novel HRV software……… 140

1 Introduction……… 141

2 Overview of F-EXTACT……… 141

2.1 Software operating procedures……… 143

2.2 Data input format……… 144

2.3 Algorithm to remove artifacts……… 144

2.4 User interface……… 146

2.5 Mathematical computation of HRV parameters……… 146

2.5.1 Time-domain analysis……… 148

2.5.2 Frequency-domain analysis……… 148

2.6 Display of HRV results……… 150

2.7 Software limitations… ……… 150

3 Summary……… 150

Chapter 10 Heart rate variability of healthy foetuses……… 153

1 Introduction……… 154

2 Study population……… 155

3 Methods……… 155

4 Results……… 157

4.1 Foetal HRV (time-domain analysis) at different gestational ages……… 157

4.2 Foetal HRV (frequency-domain analysis) at different gestational ages ……… ……… 160

4.3 Foetal HRV of male and female foetuses……… 164

5 Discussion……… 164

6 Summary……… 172

Chapter 11 Comparison of novel versus commercial HRV softwares… 174 1 Introduction……… 175

2 Method……… 175

2.1 Nevrokard system description/operation……… 175

3 Statistics……… 182

4 Results……… 184

4.1 Mean measurements obtained by Nevrokard and F-EXTRACT……… 184

4.2 Comparison of time-domain parameters between Nevrokard and F-EXTRACT using Bland-Altman method……… 187

4.3 Comparison of frequency-domain parameters between Nevrokard and F-EXTRACT using Bland-Altman method… 191 5 Discussion……… 195

6 Summary……… 201

Chapter 12 Limitations of the study and future directions……… 203

1 Clinical applications……… 204

2 Limitations of study/ equipment……… 206

3 Recommendations for future studies……… 208

REFERENCES……… 210

APPENDIX A……… 250

SUMMARY

The foetal electrocardiogram (fECG) provides important information about the foetal cardiac electrical activity As in adults and neonates, fECG provides valuable diagnostic information in a wide variety of cardiac conditions Studies on fECG using electrodes placed on the foetal scalp have proven to be successful in identifying intrapartum foetal hypoxia and in reducing unnecessary operative

interventions for foetal distress (Rosen KG, 2005; Noren H et al., 2003; Amer-Wahlin

I et al., 2001) However, scalp fECG is invasive and can only be performed during labour after the rupture of membranes Abdominal fECG is a non-invasive technique utilizing electrodes placed on the maternal abdomen

This study utilized a non-invasive foetal ECG monitor known as FEMO (Medco Electronic Systems Ltd., Israel), which enabled adequate reduction of noise and maternal ECG signals so that foetal cardiac time intervals may be measured Serial fECGs were recorded in 100 foetuses from 18 weeks of gestation till full-term, and also in a separate cohort of 197 foetuses during the first stage of labour The cardiac time intervals of P wave, PR interval, QRS complex, QT interval, QTc interval and T wave in healthy foetuses were established and found to increase with increasing foetal age No gender difference was observed in these cardiac intervals Intrapartum fECG was successfully recorded during the first stage of labour and may suggest that the foetuses were not exposed to hypoxia

Heart rate variability (HRV) is a non-invasive technique of assessing cardiac autonomic modulation It is based on measuring the variability of the time intervals

between R-to-R waves of the ECG HRV has not been well studied in foetuses, and

serially-obtained foetal HRV data is hitherto not available This study is a pioneering effort in using abdominal fECG to derive foetal HRV derived from serially-recorded fECG using a newly-developed HRV system specifically-designed for processing foetal RR-interval data and computing foetal HRV in both time- and frequency-domains Since HRV enables quantitative assessment of the different branches of the autonomic nervous system, longitudinal analysis of foetal HRV would allow an indirect assessment of the normal development and maturation of the foetal cardiac autonomic nervous system with gestational age

Results show that short-term time-domain HRV indices increased with foetal age For frequency-domain analysis, an increase in high frequency (HF) power, as well as reductions in total power, low frequency (LF) power, and LF/HF ratio were observed with advancing gestation All these indicate a change from a sympathetically-predominant to a parasympathetically-predominant cardiac autonomic control as the foetus develops The presence of LF peaks in HRV spectrograms of all foetuses and the appearance of HF peaks only in those of mature foetuses demonstrates that foetal cardiac sympathetic control is fully-developed by the early second trimester whereas the parasympathetic control becomes completely mature only by the late third trimester Finally, this study also compared algorithms between the foetal HRV system and those of a commercial system, and demonstrated

the importance of the processing of artifact beats in foetal RR-intervals for accurate foetal HRV determination

LIST OF TABLES

Table 3-1 Definitions of time domain indices ………… ……… … 39 Table 3-2 Definitions of frequency domain indices……… 45 Table 7-1 Percentage of successful fECG waveform measured ………… 104 Table 7-2 Cardiac time intervals in male and female foetuses ……… 106 Table 7-3 Foetal cardiac time intervals in relation to gestational age …… 107 Table 7-4 Linear Regression model ……… 111 Table 7-5 Intrapartum cardiac time intervals ……… 114 Table 7-6 Cardiac time intervals measured by fECG and fMCG …….… 118 Table 9-1 HRV parameters calculated by F-EXTRACT ……… 149 Table 10-1 Time-domain variables in relation to gestational age ……… 158 Table 10-2 Frequency-domain variables in relation to gestational age … 162 Table 10-3 Mean HRV parameters in male and female foetuses ………… 165 Table 11-1 Time-domain statistics displayed by Nevrokard software … 180 Table 11-2 Bland-Altman analysis (mean difference) ….……… 188 Table 11-3 Bland-Altman analysis (mean percent difference) ….………… 189

LIST OF FIGURES

Figure 2-1 Reference points for measurement of foetal ECG ……… … … 16

Figure 3-1 A typical HRV power spectrum in a resting adult ……… 43

Figure 4-1 The changes in foetal heart rates during gestation ……… 59

Figure 6-1 Electrode placement for abdominal fECG recording ……… 83

Figure 6-2 Patient unit of FEMO system and 3 recording electrodes …….… 84

Figure 6-3 Setup of FEMO system ……….……… 85

Figure 6-4 Block diagram of R wave detection algorithm in FEMO ……… 88

Figure 6-5a Foetal heart rates from scalp and abdominal electrodes … …… 89

Figure 6-5b Foetal ECG complexes from scalp and abdominal electrodes … 90

Figure 6-6 Raw ECG strip, heart rates and averaged foetal ECG complex … 92

Figure 7-1 Foetal cardiac time intervals at various gestational ages ……… 109

Figure 7-2 Scatter plots of cardiac time intervals versus gestational age … 110

Figure 7-3 Relationship between foetal cardiac time intervals and gender … 113

Figure 8-1 Foetal heart rate as shown by CTG …… ……… 131

Figure 8-2 Maternal and foetal heart rates as shown by FEMO……… …… 132

Figure 8-3 Ventricular bigeminy as shown by Doppler echocardiography … 133

Figure 8-4 Raw abdominal ECG showing PVCs occurring in bigeminy … 134

Figure 8-5 Average foetal ECG waveforms comparing QRS durations …… 135

Figure 8-6 Postnatal ECG confirming the diagnosis of PVCs ……… 137

Figure 9-1 The user-interface of F-EXTRACT ……… … 147

Figure 9-2 The results view of F-EXTRACT ……….……… 151

Figure 10-1 Time-domain HRV at various gestational ages ……….……… 159

Figure 10-2 HRV power spectra of foetuses at 20 and 37 weeks ………… 161

Figure 10-3 Frequency-domain HRV at various gestational ages ………… 163

Figure 10-4a Relationship between time-domain HRV and gender …… … 166

Figure 10-4b Relationship between frequency-domain HRV and gender ……167

Figure 11-1 Graph of RR-interval versus time on Nevrokard software ……… 177

Figure 11-2 Artifact-correction on Nevrokard software ……….……… 178

Figure 11-3 HRV power spectrum and results on Nevrokard software …… 181

Figure 11-4 Time-domain HRV from Nevokard and F-EXTRACT ……… 185

Figure 11-5 Frequency-domain HRV from Nevokard and F-EXTRACT …… 186

Figure 11-6 Bland-Altman plots (absolute bias) for time-domain HRV …… 190

Figure 11-7 Bland-Altman plots (percent bias) for time-domain HRV.……… 192

Figure 11-8 Bland-Altman plots (absolute bias) for frequency-domain HRV 193

Figure 11-9 Bland-Altman plots (percent bias) for frequency-domain HRV 194

Appendix Figure 1 Average foetal ECG complexes of different foetuses… 211

Appendix Figure 2 Average foetal ECG complexes of a single foetus…… 215

LIST OF ABBREVIATIONS

ANOVA Analysis of variance

ANS Autonomic nervous system

ApEn Approximate entropy

AV Atrioventricular

bpm beats per minute

CHD Congenital heart defect

CHF Congestive heart failure

CNS Central nervous system

CRC Cardioregulatory center

CTG Cardiotocography

ECG Electrocardiogram

fECG Foetal ECG

FEMO Foetal ECG monitor (Medco Electronic Systems Ltd., Israel)

FFT Fast Fourier transformation

fHR Foetal heart rate

fMCG Foetal magnetocardiography

HR Heart rate

HRV Heart rate variability

IUGR Intrauterine growth restriction

LVEF Left ventricular ejection fraction

mECG Maternal ECG

mHR Maternal heart rate

MI Myocardial infarction

mNN Mean RR interval

ms Milliseconds

NICHD National Institute of Child Health and Human Development

NN50 Number of pairs of adjacent RR intervals differing by more than 50ms

n.u Normalized units

pNN27 Percentage of adjacent RR intervals differing by more than 27 ms

pNN50 Percentage of adjacent RR intervals differing by more than 50 ms

PVC Premature ventricular contraction

QTc corrected QT interval

r2 Coefficient of determination

RCT Randomized controlled trial

REM Rapid eye movement

rMSSD Root mean square of successive differences

RSA Respiratory sinus arrhythmia

s Seconds

SA Sinoatrial

SD index Mean of the standard deviation of RR intervals for each 5-minute

segment of the entire recording

SDANN Standard deviation of the mean of RR intervals for each 5-minute

segment of the entire recording

SDNN Standard deviation of all RR intervals

SIDS Sudden infant death syndrome

SLE Systemic Lupus Erythematosus

STAN ST Analyser (Cinventa AB, Sweden)

SVD Singular value decomposition

TINN Triangular interpolation of NN interval histogram

ULF Ultra low frequency

VLF Very low frequency

CHAPTER 1 THE FOETAL ELECTROCARDIOGRAM

1 Historical development of the foetal ECG

The first foetal electrocardiogram (fECG) was discovered in 1906 by Cremer, who used a combination of abdominal and vaginal electrodes attached to a simple string galvanometer (Symonds EM et al., 2001) The original tracing actually showed

a foetal QRS complex that is quite remarkable considering the crude system used However, over the next 24 years, only four papers (using various combinations of abdominal and vaginal electrodes) appeared in the literature on the subject of fECG, and none of them could repeat Cremer’s earlier success

The continuing work on abdominal fECG involved the use of a high-pass filter to record the fECG, but which obliterated the P and T waves All work in this field was hampered by problems of poor signal acquisition, as well as technical problems in amplification and filtering It was not until 1957 that the P wave was observed on the fECG for the first time Southern (Southern EM, 1957) used abdominal electrodes and described the relationship between ECG changes and oxygen saturation at the time of delivery He demonstrated that foetal distress was associated with an increase in P wave amplitude, prolongation of the PR interval and depression of the ST segment This study again faced the problems of using filters that provided low frequency cut-offs, which distorted the P and T waves

A year later, Kaplan and Toyama (Kaplan S and Toyama S, 1958) used a cardiac catheterization electrode placed inside the amniotic sac in proximity to the foetus after the rupture of membranes Using another electrode attached to the

maternal abdomen, they were able to define certain cardiac time intervals although the PR interval differed slightly from that of Southern’s

By the 1960s, the first use of computer techniques in the analysis of fECG emerged Introducing a scalp clip that enabled direct application of an electrode to the foetus during labour, Hon and Lee (Hon EH and Lee ST, 1963) achieved computer averaging by inserting a triggering signal after the R wave, recording the data on tape and then playing the tape in reverse to initiate the averaging process But this signal was not useful in real time

Due to the better signal obtained from scalp electrodes as compared to abdominal electrodes, studies from the 1970s till recent years were focused on intrapartum scalp fECG, specifically on the fECG changes with cord blood acid-base and electrolyte measurements Their main findings in relation to foetal hypoxia and cord blood acidosis were QT prolongation and T wave inversion (Symonds EM, 1971), shortened PQ interval and biphasic or absent P waves (Pardi et al., 1974), elevation of both the ST segment and the T wave with increasing hypoxia (Rosen KG and Lindecrantz K, 1989; Jenkins et al., 1986; Rosen KG et al., 1976; 1975), and increase in T/QRS ratio (Lilja et al., 1985) These T wave changes preceded the fall in arterial pressure and the onset of myocardial failure

In response to these results in fECG waveform changes during foetal hypoxia, two groups in Göteborg and Nottingham worked on signal isolation and measurement

techniques that enabled the development of computerized systems that operate in real time and accurately quantify any changes in the intrapartum fECG The Göteborg group concentrated on developing a system (now known as STAN) that recognizes waveform configuration and in particular the ST segment and T wave height (Rosen

KG and Lindecrantz K, 1989) The Nottingham group worked on a complete data acquisition system and concentrated on the P wave, PR interval and T wave (Murray

HG, 1986)

2 Measurement of the foetal ECG

As mentioned, there are two methods of recording foetal ECG (fECG) The first one relies on placing an electrode in direct contact with the scalp of the foetus This is an invasive technique that can only be used during labour The second method

of fECG recording is non-invasive and involves attaching electrodes on the maternal abdomen

2.1 Invasive techniques - foetal scalp electrodes

Invasive techniques can only be performed during labour and after the rupture

of membranes since the electrodes are placed directly on the foetus The first such electrode was introduced by Hon (Hon EH, 1963), who recorded the fECG by directly attaching a scalp clip to the foetal scalp This served as a precursor to the current modern scalp electrodes All scalp electrodes work on the same common principle by attachment to the presenting part of the foetus by penetrating the skin surface The spiral electrode is screwed into the scalp whilst the Copeland clip works

by skewering a piece of skin by a spring-loaded clip A second electrode is attached

to the vaginal vault and cervix, and the fECG is obtained by determining the voltage difference between the two electrodes with the reference electrode usually attached to the maternal thigh

Due to the concerns of injury to foetal scalp, attempts have been made to produce “non-invasive” scalp electrodes whereby penetration of the foetal skin is not required Hofmeyr (Hofmeyr GJ et al., 1993) and Spencer (Spencer et al., 1998) reported on single-use probes, which rely on suction to maintain surface contract Antonucci et al (Antonucci MC et al., 1997) developed a plastic probe that uses uterine pressure to maintain contact It contains multiple sensors that allowed foetal and maternal ECG as well as amniotic fluid and intrauterine pressure to be recorded Another development is the balloon probe, which inflates with saline, and wedges the foetal sensor between the uterine wall and the foetal head

2.1.1 Clinical application of scalp foetal ECG

Scalp fECG is currently available for use in intrapartum foetal monmitoring The STAN foetal monitor (ST Analyser, Cinventa AB, Sweden) is a computerized scalp fECG system that measures and displays the ST segment and calculates the T/QRS ratio This system was developed after studies demonstrated that elevation of the ST waveform reflects myocardial stress and a switch to anaerobic metabolism, and a progressive rise of the ST waveform (quantified by T/QRS ratio) indicates prolonged anaerobic metabolism (Rosen KG and Luzietti R, 1994; Rosen KG and

Lindecrantz K, 1989; Jenkins et al., 1986; Rosen KG, 1986a; Lilja et al., 1985; Rosen

KG et al, 1976; Rosen KG et al., 1975) More detailed information on studies utilizing scalp fECG can be found in Chapter 2

2.2 Non-invasive techniques - maternal abdominal electrodes

Non-invasive techniques can be performed during pregnancy as they utilize electrodes placed on the maternal abdomen Abdominal electrodes were discovered nearly a century ago, when it was shown that the fECG could be detected by placing electrodes on the maternal abdominal surface And before the 1960s, almost all fECG studies were performed using electrodes attached on the maternal abdomen However, the only usable information from the fECG had been the peak of the QRS complex, used in the measurement of foetal heart rate (fHR), and this was superceded by the emergence of the Doppler ultrasound technology

2.2.1 Foetal ECG studies utilizing abdominal electrodes

Over the last 30 years, studies were conducted to quantify changes in the fECG configurations and cardiac time intervals of the human foetus using electrodes placed on the maternal abdomen Despite the data obtained from animal experiments suggesting configuration changes in relation to foetal hypoxia, the problems of electrical noise and signal distortion have restricted the application of fECG for routine clinical monitoring of hypoxia in the human foetus (Symonds EM, 1986)

Beyond fHR computation, Pardi et al (Pardi G et al., 1986) have shown the value of the analysis of foetal QRS complex obtained from abdominal fECG They found a decrease in the QRS duration of foetuses with foetal growth retardation and QRS lengthening in foetuses with severe Erythroblastosis Fetalis (Rh disease) In addition, they also associated ventricular hypertrophy and hypoplasia with increased and decreased QRS duration, respectively, suggesting that abdominal fECG does provide important auxiliary information for the diagnosis of congenital heart defects (CHDs)

2.2.2 Technical difficulties with abdominal electrodes

Pardi’s findings widened the potential field of application to antenatal as well

as intrapartum assessment However, the main problem faced in abdominal fECG studies is the technical difficulties in the acquisition and processing of a fECG that provides more information than just the fHR For many years, technical and practical difficulties in signal acquisition and processing restricted research on the fECG For example, the fECG obtained is superimposed and hidden by the maternal electrocardiogram (mECG), which is many times higher in intensity than the fECG Other major sources of interference include maternal electromuscular activity and external electrical interferences (50 Hz noise), as well as the intrinsic noise of the equipment The high impedance of the maternal abdominal skin and the capacitive properties of the feto-maternal interface contribute further to the problem of low signal-to-noise ratio These difficulties are compounded by the sensitivity of the fECG signal amplitude to foetal position Moreover, because the electrodes are not

placed directly on the foetus, the topographical relationship between the foetal heart and the selected leads cannot be established

2.2.3 Vernix caseosa and abdominal foetal ECG signal

Another major drawback of the clinical use of abdominal fECG lies in the reported absence of signals in approximately 50% of patients between 20 and 34 weeks of gestation, and an even lower success rate between about 28 and 32 weeks (Wheeler T et al., 1978) Explanations have been given for the low signals obtained during this period using volume conduction experiments It has been suggested that this decrease in signal amplitudes is caused by the insulating effect of vernix caseosa, which surrounds the foetus at this stage Based on fECG recordings at multiple sites, Roche and Hon (Roche JB and Hon EH, 1965) concluded that after the 30th week of gestation, the fECG current generated by the foetal heart is conducted towards the maternal abdomen via preferred pathways through holes in the vernix, and suggested the oro-nasal cavity of the foetus as a distinct pathway

Oldenburg and Macklin (Oldenburg JT and Macklin M, 1977) suggested that there is uniform volume conduction during the beginning of the second half of gestation, but preferred pathways play a dominant role in the signal transmission during the last weeks of gestation They found that between 20-28 weeks, the fECG signal transmission was consistent with simple volume conduction Between 28 and

34 weeks however, the signal transmission was inconsistent with volume conduction and the fECG amplitude was much lower They hypothesized that these effects were

due to the increasing amounts of vernix and amniotic fluid as a barrier to radial conduction After 34 weeks, the fECG signal increased again as the foetal head is engaged in the maternal pelvis relatively close to the surface, consistent with transmission via the facial orifice not covered by vernix

In 1986, Oostendorp (Oostendorp TF et al., 1986) studied the potential distribution generated by the foetal heart on the maternal abdomen by recording abdominal fECG using 32 leads for a full three-dimensional description of the geometry of the volume conductor Their observation disagreed with the low-impedance preferred pathways, which would produce fixed places of potential extremes but these were not observed in the study

In a further, more detailed study by Oostendorp et al., (Oostendorp TF et al., 1989a; 1989b) it was observed that in early pregnancy less than 28 weeks, the current conducted by the fECG can be described by an electrically homogeneous model whereby the fECG is independent of the position of the foetus Around 28 weeks, a layer of vernix caseosa starts to develop, and almost completely isolates the foetus electrically, so the magnitude of the fECG falls to a very low value As pregnancy continues, gaps appear in the vernix (from about 32 weeks), and the fECG magnitude increases again In late pregnancy, the conduction is largely determined by the size and unpredictable positions of the gaps

Kahn (Kahn AR, 1963) provided further evidence that the vernix interrupts with the fECG current He discovered that the impedance of the foetal vernix is approximately 100 times that of the amniotic fluid and uterine muscle The high impedance was reduced by washing away the vernix and replacing it with a conductive cream or gel

In 1989, Oostendorp et al (Oostendorp TF, 1989c) measured the conductivity

of human foetal membranes, Wharton’s jelly and vernix caseosa, which are involved

in the transmission of current from the foetal heart to the maternal abdominal surface The conductivity of foetal membranes (0.40 Ω-1 m-1) was found to be of the same order of magnitude as the conductivity of the overall tissue (0.222 Ω-1 m-1) This, in addition to the small thickness of the membranes (<0.5 mm), makes it feasible not to include the membranes in a volume conduction model describing the conduction of the fECG As for the Wharton’s jelly, its conductivity is of the same magnitude as the amniotic fluid (1.66 Ω-1 m-1) As a consequence, any current reaching the blood in the umbilical cord will freely flow through the Wharton’s jelly into the amniotic fluid Hence, it is again not necessary to include the umbilical cord in a volume conduction model The conductivity of the vernix was found to be (1.8 ± 0.3) x10-6 Ω-1 m-1, being

of the same order as the value of 1.4 x 10-6 Ω-1 m-1 (at 37˚C) as found by Bolte in a German literature in 1961 Hence, the conductivity of vernix differs by a factor of roughly one million from the conductivities of the other foetal tissues involved Thus,

it is clear that the layer of vernix, which starts to develop around the 28th week and disappears in the very last stage of gestation, is the sole factor interrupting with the

conduction of the fECG current, thereby causing the decline in the fECG signals during 28th –34th gestational weeks

In 2000, Wakai et al (Wakai RT et al., 2000) described the relationship between vernix and fECG signals They performed fECG measurements on a foetus with ectopia cordis, a rare condition in which the foetal heart lies outside the chest cavity In this way, fECG signals bypassed transmission through the foetal skin and vernix The most obvious difference between the ECG of a foetus with ectopia cordis and a normal foetus was that the former was of very high amplitude, ranging from 35

to 65µV During 28 to 34 weeks, a period when the fECG signal would normally be very low and often failed, the fECG amplitude in the foetus with ectopia cordis remained as high as those in near-term foetuses This strongly implies that the decrease in fECG signals during this period in normal foetuses is due to the insulating effects of the vernix The fECG transmission was also consistent with volume conduction properties such that the fECG amplitude and configuration varied with electrode placement and foetal position

2.2.4 Acquisition and processing of abdominal foetal ECG

Due to the low signal-to-noise ratio of the abdominally-derived fECG, it has

to be digitally-processed in order to obtain clear fECG signals from which diagnostic parameters can be derived To overcome the problem of low signal-to-noise ratio, the signal processing algorithm should ideally remove the mECG compexes, reduce the

effects of motion artifact and muscle noise, and then enhance and amplify the fECG complexes in order to obtain fECG signals displaying clear P, QRS and T waves

The largest source of interference in measuring abdominal fECG is the comparatively stronger mECG signal, which must be eliminated before further analysis of the fECG In the past, many methods have been developed to solve this problem The earliest attempts involved a direct scaled subtraction of a thoracic or near-thoracic mECG from an abdominally-measured composite of maternal and fECG signals (Wheeler T et al., 1978; Longini RL et al., 1977; Reichert TA et al., 1977)

Newer and more complex methods used for cancellation of the mECG and thus improve the signal-to-noise ratio of the fECG have based on wavelet transform

techniques (Mochimaru F et al., 2004; Khamene A and Negahdaripour S, 2000),

singular value decomposition (SVD) techniques (Kanjilal PP et al., 1997; Callaerts D

et al., 1990, 1989, 1986; Vanderschoot J et al., 1987), matched and spatial filtering (Gibson NM et al., 1997; van Oosterom A, 1986; Bergveld P et al., 1986, 1981) auto- and cross-correlation (Budin N and Abboud S, 1994; Abboud S et al., 1992; van Bemmel JH, 1968), blind source separation (BSS) techiniques (Zarzoso V and Nandi

AK, 2001), state space projection technique (Richter M et al., 1998), adaptive filtering and noise cancellation (Camps-Valls G et al., 2004; Almenar V et al., 1999; Ferrara ER et al., 1982)

2.2.5 Clinical application of abdominal foetal ECG

Abdominal fECG is not currently used in clinical practice today, despite the above methods used to enhance its signal detection and processing Being safe, comfortable, inexpensive, and easy to operate, abdominal fECG provides a noninvasive technique for fHR monitoring and to assess foetal cardiac well being, especially to identify foetuses with cardiac arrhythmias Clinical evaluation of foetal arrhythmias is generally performed by Doppler and M-mode echocardiography Assessment of any disturbances in rhythm is done based on contractile and flow behaviour But there are certain limitations of this technique For example, it is heavily-dependent on the investigator’s experience It also does not allow analysis of the electrical signal morphology, which may make differential diagnosis of the arrhythmia difficult Non-invasive fECG offers an alternative as it enables the examination of the foetal cardiac conduction system on the basis of electrophysiological signal Recently, a case of slow-rate ventricular tachycardia has been successfully diagnosed in a 32 week-old foetus by the use of non-invasive fECG (Yumoto Y et al., 2004) With further improvements in signal processing power, it is likely that in the near future, non-invasive fECG would have greater potential in the routine antenatal screening for cardiac arrhythmias, monitoring and possible treatment of these high-risk foetuses

CHAPTER 2 THE FOETAL ECG

WAVEFORM

1 Morphology and time intervals of the foetal ECG

Like the adult ECG, the foetal ECG consists of P, QRS and T waves, separated by PR and ST segments These waves represent the summation of electrical events within the heart as seen from the body’s surface Figure 2-1 shows the clinically accepted reference points for time intervals and amplitude measurements These reference points on the foetal ECG are adopted from definitions used in quantitative measurement of the adult ECG (Symonds EM et al., 2001) The following sections describe these ECG time intervals in adults and foetuses

1.1 P wave

The P wave of the ECG represents the sequential activation or depolarization

of the right and left atria But because the atria have relatively little muscle, both atria generate a single small P wave Atrial depolarization is indicated by an upward deflection of the baseline Since the SA node lies in the right atrium, the right atrium begins to depolarize first and finishes earlier as well Hence, the first part of the P wave predominantly represents right atrial depolarization, and the second part left atrial depolarization When the entire mass of the atria is depolarized, the ECG returns to the isoelectric baseline The spread of atrial depolarization thus creates the

P wave

In the adult, the P wave duration normally lasts less than 0.11 seconds An abnormally long P wave occurs whenever it takes extra time for the electrical wave to

reach the entire left atrium This occurs in left atrial enlargement The height of the P wave is normally less than 0.25 millivolts An abnormally tall P wave is seen when larger amounts of electricity are moving over the atrium This usually indicates hypertrophy of the right atrium The P wave may also be decreased in height by hyperkalemia

P wave duration in the term foetus had been studied since the introduction of scalp electrodes in the 1960s The average value for the duration of the P wave in a healthy full-term foetus was found to be 50-52 milliseconds (ms) Data from various studies has been largely similar In 1974, Pardi (Pardi G et al., 1974) described biphasic and absent P waves suggestive of foetal distress, but Marvell (Marvell CJ et al., 1980) found that changes in the shape of P wave were not uncommon in their study of 37 normal foetuses monitored during labour

While P wave duration and the P wave area were shown to be related to respiratory acidaemia in the foetus and correlated with umbilical venous noradrenaline levels, there was only weak correlation with foetal acidaemia at birth and it did not appear to be of any practical value for clinical use (Jenkins HM et al., 1986)

In 1995, Mohajer (Mohajer MP et al., 1995) noticed that the area and amplitude of the P wave seemed to diminish and disappear during episodes of profound bradycardia Examination of the raw fECG complexes showed that the P

waves had not disappeared but they no longer bore any relationship to the R wave This indicated a disruption in conduction between the SA and AV node, leading to a second or third degree heart block that may have a hypoxic basis

PR segment occurs in the AV node

The period of time from the onset of the P wave to the beginning of the QRS

is termed the PR interval, which represents the time between the onset of atrial depolarization and the onset of ventricular depolarization It is measured from Pon to

Qon (Fig 2-1)

PR interval in normal adults lasts between 0.12 to 0.2 seconds The PR interval may be prolonged when conduction of the electrical wave through the AV node is slow This may be seen in degenerative disease of the AV node, or with digoxin, hyperkalemia, hypercalcemia, or hypothermia The PR interval may be unusually short when conduction is rapid A mildly short PR interval may be seen

with hypokalemia or hypocalcemia An artificially-short PR interval occurs when the QRS complex begins early, as happens with an extra conduction pathway between the atria and ventricles in Wolff-Parkinson-White (WPW) Syndrome

In the term foetus, the PR interval lasts for about 103-109 ms Foetal ECG studies suggested that relative changes in the PR and RR interval of the foetal ECG may help to predict foetal compromise Normally, there is a positive correlation between the PR interval and RR interval (i.e., a negative correlation between PR interval and foetal heart rate (fHR)), such that when the heart rate increases, both the

PR and RR intervals shorten But in the hypoxaemic foetus, a paradoxical shortening

of the PR interval occurs despite lengthening of the RR interval (foetal bradycardia)

In 1986, Murray (Murray HG, 1986) observed that in healthy foetuses during labour, the PR interval was normally positively correlated (r=0.66) to RR interval This positive correlation changed to a negative one when the foetus became acidotic,

so that a shorter PR interval occurs (presumably from the effect of catecolamines) with a longer RR interval (slower heart rate) Murray introduced the term “conduction index” (CI), which is the Pearson’s correlation coefficient of the PR interval with fHR A positive CI for longer than 20 minutes was related to foetal deterioration in the form of abnormal umbilical artery blood gas measurements

Another index called the “ratio index” (RI) is a modification of the CI so as to provide a cumulative expression of the CI over the whole course of labour The RI

reflects chronic foetal deterioration in contrast to the CI, which reflects more acute changes There were significant correlations between RI and biochemical indices of foetal hypoxia such as the umbilical blood artery pH, lactate, noradrenaline and hypoxanthine, which led the authors to suggest that intrapartum RI index measurement may be useful for determining foetal hypoxia (Mohajer et al., 1994)

Studies on human foetuses indicate that the use of CI and RI in conjunction with CTG interpretation (as compared to CTG interpretation alone) significantly reduced unnecessary foetal blood sampling, unsuspected foetal acidosis and instrumental deliveries (Reed NN et al., 1996; van Wijngaarden WJ et al., 1996b)

Similar results were obtained by animal studies Experimental sheep models

by which foetal hypoxia was induced via uterine artery occlusion showed that both brief episodes of hypoxia and prolonged continuous hypoxia were associated with a change from a positive to a negative PR/RR ratio, thus further suggesting that a changing relationship between the PR and RR interval had potential use in the

detection of foetal distress (Keunen H et al., 2000; van Wijngaarden WJ et al., 1996a;

Widmark C et al., 1992)

However, Luzietti (Luzietti R et al., 1997) found that a negative PR-RR relationship occurred with all decelerations of more than 40 bpm and seemed to be nothing more than an indicator of decelerations during labour This was further supported by Westgate (Westgate JA et al., 1998), who identified that the PR/RR

correlation changed from positive to negative at a time when the foetus was neither hypotensive nor acidic This suggests that the paradoxical PR shortening at the onset

of a fHR deceleration is likely to be a reflex-mediated response unrelated to foetal acidosis or hypotension In agreement, Strachan (Strachan BK et al., 2000) compared the use of standard CTG with CTG plus PR interval analysis and found that with either method of foetal monitoring, there was no significant difference in the neonatal outcome or in the number of operative deliveries for suspected foetal distress

1.3 QRS complex

In normal sinus rhythm, each P wave is followed by a QRS complex The QRS complex represents the time it takes for depolarization of the ventricles, measured from Qon to Stm The ventricles depolarize in a specific manner The first initial area to depolarize is the ventricular septum This usually depolarizes from the left side across to the right side, corresponding to the Q wave which is not always present Next, the depolarization wave spreads through the right and left ventricles simultaneously and gives rise to the R and S waves These waves will be recorded as positive deflections on the ECG if the depolarization current is moving towards the positive electrode of the recording lead Hence, the morphology of the QRS complex depends on the location of the positive electrode with respect to the direction of current flow

In the adult, the QRS complex is normally less than 0.10 seconds in length Lengthening of the QRS indicates some blockage of the electrical signal in the