Assessment and quantification of foetal electrocardiography and heart rate variability of normal foetuses from early to late gestational periods 3

Clinical application of non-invasive foetal ECG 1291 Introduction Cardiac arrhythmias arise as a result of abnormal cardiac electrical conduction.. However, the baseline foetal heart ra

Trang 1

Clinical application of non-invasive foetal ECG 128

CHAPTER 8 CLINICAL APPLICATION OF

NON-INVASIVE FOETAL ECG

Trang 2

1 Introduction

Cardiac arrhythmias arise as a result of abnormal cardiac electrical

conduction The gold standard for the diagnosis of arrhythmias has always been via

the ECG Foetal arrhythmias occur in 1 to 2 percent of pregnancies These range from

benign premature atrial or ventricular contractions to life-threatening supraventricular

and ventriculartachycardia Foetal arrhythmias are currently being diagnosed by

M-mode echocardiography, which is labour-intensive and indirectly detects the cardiac

mechanical rather than electrical activity Non-invasive abdominal foetal ECG

(fECG) is a convenient, inexpensive and effective technique for the detection of the

foetal heart’s intrinsic electrical activity Although it has not yet been integrated into

clinical practice, it seems promising for future clinical application in the diagnosis

and management of foetal arrhythmias

2 Case report of a foetus with premature ventricular contractions

The following is a case report that illustrates the clinical application of

non-invasive fECG in a foetus with premature ventricular contractions (PVCs), which was

missed by cardiotocography (CTG) monitoring Using non-invasive fECG, the true

foetal heart rate was determined This led to the probability of avoiding unnecessary

operative intervention to deliver the foetus under emergency conditions

2.1 Antenatal foetal ECG

A young and healthy 17-year-old primigravida at 37 weeks’ gestation

underwent a routine CTG monitoring during an antenatal visit The pregnancy had

Trang 3

thus far been uneventful However, the baseline foetal heart rate (fHR) on the CTG

measured 60-70 beats per minute (bpm), with occasional, sudden “jumps” to 120 bpm

(Figure 8-1) The foetus is otherwise well with active movements noted The mother

had no history of Systemic Lupus Erythematosus (SLE), which is known to be

associated with heart blocks in the foetus With the mother’s consent, the fHR was

determined using the FEMO fECG monitoring system (Medco Electronics Limited,

Israel) The method of measurement of fECG has been described in Chapter 6 Using

FEMO, baseline fHR was reflected as 130 bpm (Figure 8-2) Since the FEMO

monitor documented a stable fHR tracing, the decision was not to intervene unless the

foetus started to show signs of compromise Doppler and M-mode echocardiography

revealed the presence of frequent PVCs, mainly occurring in bigeminy (Figure 8-3)

During follow-up assessment in the following weeks, fHR was still recorded

as 60-70 bpm and 120-140 bpm by the CTG and FEMO, respectively Upon closer

examination of the raw abdominal ECG trace of FEMO, it was discovered that PVCs

could be observed Figure 8-4 shows an abdominal strip with PVCs occurring in

bigeminy In addition, measurement of the averaged fECG complex revealed a

prolonged QRS duration (mean QRS duration = 76 ms) when compared to the cohort

of healthy foetuses of the same gestational age in this study (mean QRS duration = 53

ms) (Figure 8-5) A widened QRS is one of the characteristic ECG findings of PVCs

in adults and children PVC indicates depolarization that arises in either ventricle

before the next expected sinus beat In such circumstances, the normal sequence of

ventricular depolarization is altered, and the two ventricles depolarize sequentially

Trang 4

Clinical application of non-invasive foetal ECG 131

FM= foetal movement; Toco= uterine activity

Figure 8-1: The CTG at 38 th week of gestation showed the baseline foetal heart rate (fHR) as 60-70 bpm

Trang 5

Figure 8-2: The foetal heart rate (fHR) and maternal heart rate (mHR) as displayed by FEMO at 38 th week of gestation The baseline fHR is about 130 bpm with good beat to beat variability

Trang 6

NB – normal beat; PVC – premature ventricular contraction

Figure 8-3: Ventricular bigeminy as demonstrated by Doppler

echocardiography

Trang 7

PVC= premature ventricular contractions

Figure 8-4: Raw abdominal ECG strip showing PVCs occurring in bigeminy

Trang 8

(a)

(b)

Figure 8-5: Average foetal ECG waveform of a healthy term foetus (a) with average QRS duration of 53 ms and that of the foetus with PVC with an average QRS duration of 76 ms (b)

Trang 9

rather than simultaneously, thus resulting in a wide QRS complex This is in contrast

to premature atrial contractions (PACs), where the QRS durations are not prolonged

and the P waves have a configuration different from the normal P wave In this case

study, the abnormally wide QRS duration and normal P wave suggest that the

ectopics were ventricular (and not supraventricular) in origin

2.2 Intrapartum foetal ECG

During labour, it was noted that the CTG tracing during labour remained at 70

bpm and was not recordable most of the time Foetal scalp ECG monitoring was

performed but it failed to record either the fECG or the fHR The fHR of a healthy

foetus is generally in the range of 110-150 bpm A fHR<100 bpm is defined as foetal

bradycardia and fHR<80 bpm indicates possibly severe, chronic hypoxia that

demands immediate delivery However, FEMO recording showed that the fHR was

120-140 bpm with good variability Hence, there was no indication for Caesarean and

the mother progressed with the spontaneous vaginal delivery of a male neonate

Postnatal ECG confirmed the diagnosis of PVCs, showing mainly

quadrigeminy, i.e., one PVC occurring after every 3 normal beats (Figure 8-6) It

appeared that the arrhythmic condition of the patient had improved and PVCs were

occurring less frequently The occurrence of PVCs had possibly progressed from

bigeminy as observed by echocardiography and fECG during the antenatal period to

quadrigeminy seen in the neonatal period

Trang 10

Figure 8-6: Postnatal ECG confirming the diagnosis of premature ventricular contractions (PVCs) Block arrows indicate the occurrence of PVCs occurring every fourth or fifth beat

Trang 11

3 Discussion

CTG utilizes the Doppler ultrasound technology, which is based on the

detection of cardiac mechanical activity The Doppler audio signal contained in a

single heart beat is very complex Fourier analysis has linked its contents to six

cardiac events, namely, four valve and two wall movements (Shakespeare SA et al.,

2001) One known problem of the Doppler ultrasound signals is the effect of HR

doubling when two strong audio signals are present within a single cardiac cycle

(Peters M et al., 2001) On the contrary, when the cardiac stroke volumes are of

different magnitude, as in this case, the ultrasound signals generated by the PVCs

were much weaker than those triggered by the normal heartbeats It was possible that

the CTG could not recognize signals generated by the PVCs, and was only detecting

signals from normal heartbeats (occurring half of the time) This led to the detection

of the fHR as 60 bpm when it should have been 120 bpm The maternal heart rate

ranged between 80-90 bpm, so the recorded HR is unlikely to be maternal

To alleviate the problems of erroneous and missed beats, most current CTG

monitors contain the autocorrelation feature This is a signal processing method that

compares the incoming heartbeat signal with a time-delayed version of itself

Filtering is then applied to capture frequencies and amplitudes likely to be associated

with foetal cardiac motions This method is especially useful in removing “noise”

from arterial blood flow, foetal movements, maternal movements or other muscular

contractions It is likely that in this case, the PVCs, which were of much smaller

amplitudes, and occurring at not exactly the same periodic intervals, had been

Trang 12

eliminated as “noise” This may explain the failure of the CTG to detect the true heart

rate of the foetus

On the other hand, the FEMO system records the foetal ECG, which measures

the foetal cardiac electrical activity Foetal heart rate is derived from the R wave of

the foetal ECG, which is clearly recognizable as the peak amplitude of the ECG This

provided an ideal detection point It was thus able to reflect true beat-to-beat fHR

data Hence, it may be due to this reason that the foetal ECG was successful in

detecting both the normal and ectopic beats and thus gave an accurate measure of the

true fHR The case history highlights the usefulness of non-invasive fECG in the

determination of foetal heart rate and ECG and its advantage over CTG and scalp

ECG In circumstances such as this, reliance on CTG and scalp ECG was not able to

reflect the true fHR and electro-cardiological situation

Trang 13

Development of a novel HRV software 140

CHAPTER 9 DEVELOPMENT OF A NOVEL

HRV SOFTWARE

Trang 14

1 Introduction

In collaboration with the Computer Science Department of National University of Singapore, a software (F-EXTRACT) was developed to compute the foetal heart rate variability (HRV) from the RR-interval data obtained from FEMO abdominal foetal ECG monitor The development of F-EXTRACT arose from the growing interest in HRV and the lack of reliable HRV analysis software tools suitable for the analysis of foetal HRV With the provision of domain-knowledge, a combination of signal-processing tools and techniques were employed to extract the RR-interval data recorded by the FEMO foetal ECG monitor The objectives of this new system were to provide visualization of the foetal HRV power spectrum as well

as to automate the extraction of foetal HRV values

2 Overview of F-EXTACT

The F-EXTRACT system was developed using MatLab 6.1 (Release 12.1) MatLab, which stands for "Matrix Laboratory" is a commercial software package for numerical matrix analysis It operates as an interactive programming environment that enables the development, implementation and analysis of computer algorithms All programs that are written with MatLab will require it to run Thus, F-EXTRACT also requires MatLab installation to run MatLab software was chosen for developing the novel HRV program because it contains built-in mathematical functions This provided great convenience in implementing complex algorithms, such as statistical measures, matrix operations and spectral analysis algorithms

Trang 15

The objectives of F-EXTRACT were primarily to automate the extraction of RR-interval data from foetal ECG recorded from FEMO, and to display the power spectra of foetal HRV so as to provide insightful visualization of the data An automated computation of time- and frequency-domain parameters of foetal HRV was necessary because manual measurement of RR-intervals from raw foetal ECG signal is time-consuming and subjected to errors In addition, artifact-removal tools appropriate for eliminating the high degree of errors typically-found in foetal RR-

mainly for the purpose of foetal HRV analysis in this study It is not packaged as a commercial system, and is thus not comparable to commercial systems with excellent Graphical User Interface such as easy-to-use graphs generation, threshold adjustments through mouse-click, and easy management of patients’ records

Although commercial HRV software packages are available, they are specifically developed for calculating HRV in adults These tools are not likely to be suitable for analyzing foetal HRV unless they allow the modification of frequency bands so as to match those that are used for HRV analysis in the foetus Another point

to consider is that the detection rate of R peaks from foetal ECG is considerably lower than that of adult ECG, and artifacts may occur frequently This is because the foetal heart is considerably smaller than the adult heart and is also further away from the recording electrodes placed on the maternal abdomen Hence, foetal RR-intervals require thorough artifact-correction before HRV computation The development of a new HRV analysis system allows the incorporation of a more rigorous correction

Trang 16

algorithm appropriate for foetal RR-intervals Various commercial softwares have their own different set of algorithms for artifact- correction, which are mainly targeted at adult RR-intervals and may not be suitable for foetal RR-intervals

2.1 Software operating procedures

A summary of the basic steps for HRV analysis on F-EXTRACT is as follows: Before performing the HRV analysis, all foetal RR-interval data were first exported as single-column ASCII files and stored in the computer hard disk drive for later use When HRV analysis was to be conducted, the MatLab program was initiated and relevant programming instructions were typed in the Command Window

to load the foetal RR-interval data file

Next, the corresponding RR interval tachogram was plotted in a new window This would enable the viewing of the foetal RR-interval data file with RR-interval length as the Y-axis and beat number as the X-axis From here, any missing, spurious beats could be visualized, so that an appropriate period free of artifacts could be selected for performing the HRV evaluation After typing suitable commands to load the HRV analysis, the results that included the HRV power spectrum and statistical calculations were then displayed in a single pop-up window The results were saved

in the computer, and the next foetal RR-interval file to be analyzed could be loaded

Trang 17

2.2 Data input format

The first step in the HRV analysis was to acquire a series of RR-intervals This was performed by FEMO foetal ECG system, which recorded and saved the RR-interval data as FEMO’s own proprietary file formats In order for the RR-interval data to be recognized and evaluated by F-EXTRACT, it had to be in a universal format Thus, an additional step was taken to export the RR-interval data from FEMO file formats to some commonly used format, such as the universal ASCII text file It was decided that the RR-interval data be exported as a single column vector in ASCII format, which was recognized by MatLab Hence, each of the foetal RR-interval series was first exported as ASCII files before they were evaluated on F-EXTRACT

2.3 Algorithm to remove artifacts

After the foetal RR-interval files had been exported as a format recognizable

by MatLab, the exported files were first subjected to an algorithm to remove artifacts

in the foetal RR-intervals or heart beats The correction algorithm was written by programming language using MatLab software This was done in the same manner as writing the formulae for the time- and frequency-domain calculations of HRV As a rule, the algorithm would instruct the F-EXTRACT system to correct all artifacts in the RR-interval data before performing the HRV calculations

Artifact-correction is necessary because besides the normal heart beats generated by the sinoatrial node pacemaker, ectopic (non-sinus) beats may sometimes arise from latent pacemakers in the heart It was shown that approximately one third

Trang 18

of healthy men have one or more ventricular ectopic beats during a one hour ECG recording (Bikkina M et al., 1992) As these ectopic beats are often premature and are followed by a compensatory delay, they generate a shorter than normal R-R interval followed by a longer than normal R-R interval, thereby causing interruptions and errors in the HRV analysis Hence, it is necessary to incorporate tools in the F-EXTRACT system that corrects these ectopic beats

In fact, the FEMO foetal ECG acquisition software enables the scanning and tagging of ectopic, spurious or missing beats for their identification and subsequent correction or deletion However, the visual inspection of each RR interval and the manual editing of aberrant beats can be tedious, time-consuming, subjective and prone to errors Thus the automated correction by F-EXTACT with properly adjusted thresholds and linear interpolation may result in greater convenience, less operator bias and higher accuracy in the correction of aberrant beats Whenever possible, segments of RR-intervals free of artifacts were selected for HRV analysis But as a further measure, the automated correction algorithm for these beats was incorporated into the F-EXTRACT system

As mentioned, the purpose of the correction algorithm was to eliminate ectopic, spurious and missing beats in the RR-interval data Two approaches were used in the correction algorithm The first approach was to delete each erroneous beat and replace it by an average value interpolated from the beats before and after it An

RR interval was interpreted as erroneous if it deviated from the previous qualified

Trang 19

intervalby more than 30% (Kamath MV and Fallen EL, 1995) The second approach

by the correction algorithm served to eliminate extreme values of RR-interval durations This removed the foetal heart beats that exceed the high and low threshold values of 170 and 110 beats per minute (equivalent to mNN intervals of 353 and 545 milliseconds, respectively) This range of values was chosen because foetal heart rates above 170 or below 110 beats per minute would be considered as foetal tachycardia and bradycardia, respectively, and thus best to be excluded from the HRV analysis

2.4 User interface

The user-interface of F-EXTRACT is the standard user-interface of the MabLab software, which comprises of only one window This user-interface is divided into two sections: the Command Window on the right half of the screen, and the Command History on the left half (Figure 9-1) The Command Window is where the programming instructions were typed for example, “load 1234567-005.frr” enabled the selection of the 5th serial recording of the foetal RR-interval data of this particular foetus whose clinic ID is 1234567 The Command History records all previously typed instructions for easy referral and confirmation

2.5 Mathematical computation of HRV parameters

The F-EXTRACT system is able to perform foetal HRV analyses in both time-domain and frequency-domain The calculated parameters are discussed in the following sections

Trang 20

Development of a novel HRV software 147

Figure 9-1: The MatLab user-interface that is used for F-EXTRACT, showing the Command Window on the right and Command History

on the left

Trang 21

2.5.1 Time-domain analysis

The time-domain parameters are calculated directly from the raw RR-interval time series These parameters include the mean heart rate (HR), mean RR-interval (mNN), standard deviation of RR intervals (SDNN), root mean square of the differences between consecutive RR intervals (RMSSD), and percentage of consecutive RR-intervals that differ by more than 27 ms (pNN27)

2.5.2 Frequency-domain analysis

In the frequency-domain analysis, the power spectral density of the foetal RR time series was calculated using the Fast Fourier Transformation (FFT), which decomposes the foetal RR-interval data into its underlying oscillations with different frequencies and power The frequency bands selected for analysis of foetal HRV were

as follows: very low frequency (VLF: 0.003-0.04 Hz), low frequency (LF: 0.04-0.15 Hz) and high frequency (HF: 0.15-1.0 Hz) The FFT-based HRV spectral powers were quantified by integrating the spectrum over these frequency bands, which measured the area in these bands

The frequency-domain parameters generated by F-EXTRACT included the total power (TP), VLF power, LF and HF power in both absolute and normalized units, and the LF/HF ratio Total power refers to the total power in the HRV spectrogram Chapter 6 contains a more detailed description of the derivation of the time-and frequency-domain HRV parameters A summary of all the calculated parameters is shown in Table 9-1

Trang 22

Development of a novel HRV software 149

Table 9-1: A summary of the HRV parameters calculated by F-EXTRACT

squares of differences between adjacent NN intervals

by more than 27 ms Total power (TP) msec2 Variance of all NN intervals

Very low frequency

(VLF) power

msec2 Power in the 0.003-0.04 Hz range

Low frequency (LF)

power

High frequency

(HF) power

Trang 23

2.6 Display of HRV results

After the HRV analyses have been performed by F-EXTRACT, the results were displayed in a separate pop-up window (Figure 9-2) This window shows the foetal HRV power spectrum as well as the calculated time- and frequency-domain statistics on one single page The patient details such as name, clinic ID, date of recording and week of gestation were also shown on the top of the window The user menu of the result window allowed functions such as zooming in/out, copying, saving and printing of the statistical and graphical results

2.7 Software limitations

F-EXTRACT was developed to be used in a research environment and is not

"user-friendly" in many ways Firstly, it requires Matlab, which consumes much computing resources like disk storage and memory space Hence it is not suitable for older versions of computers Secondly, in order to perform the HRV analysis, several formatted commands must be typed into the Command Window This is because F-EXTRACT utilizes the MatLab software to run and MatLab can only recognize instructions in programming language

3 Summary

A MatLab-based HRV software, F-EXTRACT, was developed for use in this study Since commercial HRV softwares are specifically-designed for analyzing adult HRV, their algorithms and thresholds may not be appropriate for processing foetal RR-interval data and computing foetal HRV The objective of F-EXTRACT was to

Định dạng
Số trang	46
Dung lượng	3,18 MB