The Efficacy of In-Phase and Quadrature Demodulation in Electronic Fetal Heart Rate Monitoring During Labor

Abstract

Objective:

To investigate the efficacy of in-phase and quadrature (IQ) demodulation in electronic fetal heart rate monitoring (EFM) to reduce false reports of fetal heart rate (FHR) doubling or halving.

Methods:

This is a prospective cohort study. A total of 263 full-term pregnant women who delivered at Peking University Shenzhen Hospital between August 2019 and July 2020 were prospectively enrolled in the study. FHR monitoring began when the cervix was dilated to 2–3  cm and continued until delivery. Raw fetal Doppler audio signals and internal and external cardiotocography curves from internal electrode monitoring, EFM with conventional demodulation (external), and EFM with IQ demodulation (external) were acquired to compare FHR doubling and halving time. In cohort 1, FHR was compared between IQ demodulation and conventional demodulation. In cohort 2, FHR was compared between IQ demodulation, conventional demodulation, and internal FHR monitoring. Count data were statistically analyzed using the Chi-squared test, and measurement data were statistically analyzed using t-test for correlation coefficients, and Bland-Altman analysis for concordance ranges.

Results:

To compare IQ demodulation and conventional demodulation, 225 pregnant women were monitored for a total of 835,870  seconds. The beat-to-beat interval of FHRs in raw fetal Doppler audio signals was used as the reference. The results showed a doubling time of 3401  seconds (0.407%, 3401/835,870) and a halving time of 2918  seconds (0.349%, 2918/835,870) with conventional demodulation, compared to 241  seconds (0.029%, 241/835,870) and 589  seconds (0.070%, 589/835,870), respectively, with IQ demodulation. IQ demodulation reduced FHR doubling by approximately 93% (3160/3401) and FHR halving by approximately 80% (2329/2918) compared to conventional demodulation (P < 0.01).

To compare IQ demodulation, conventional demodulation, and internal FHR monitoring, 38 pregnant women were monitored for a total of 98,561  seconds. FHR from internal electrode monitoring was used as the reference. The results showed a doubling time of 238  seconds (0.241%, 238/98,561) and a halving time of 235  seconds 0.238%, 235/98,561) with conventional demodulation, compared with 30  seconds (0.030%, 30/98,561) and 81  seconds (0.082%, 81/98,561), respectively, with IQ demodulation (P < 0.05). No significant difference was observed in doubling or halving time between IQ demodulation and internal FHR monitoring (P > 0.05). IQ demodulation was more closely correlated with internal FHR monitoring than conventional demodulation (0.986 vs. 0.947). The Bland-Altman analysis showed that the concordance range of IQ demodulation vs. internal FHR monitoring was significantly narrower than that of conventional demodulation vs. internal FHR monitoring ((−5.32,6.01) vs. (−10.87,11.46)).

Conclusion:

EFM with IQ demodulation significantly reduces false FHR doubling and halving, with an efficacy similar to that of internal FHR monitoring.

Introduction

An electronic fetal heart rate monitoring (EFM) evaluates fetal safety by monitoring the changes in fetal heart rate (FHR). In the early 1800s, obstetricians first began to assess fetal status with fetal heart auscultation. With the advancements in Doppler ultrasound technology, Dopp ler-based EFM has become the most common method to monitor FHR and has significantly reduced neonatal asphyxia and perinatal morbidity and mortality.^1–3

During labor, EFMs often erroneously report FHR by doubled or halved rates, making it difficult to interpret the data and increasing the likelihood of unnecessary operations such as cesarean section and operative vaginal delivery.^4–6 This false appearance of doubled or halved FHRs on an EFM is related to the mechanism of Doppler ultrasound. In brief, the ultrasonic signals emitted by the ultrasound probe reflects or “echoes” of the beating heart structures which can be used to assess the systolic and diastolic phases of the cardiac cycle. However, EFM with conventional demodulation only uses one in-phase channel to demodulate ultrasonic signals and thus, cannot accurately differentiate between systolic and diastolic echo signals. As a result, false FHR doubling occurs when the duration of systole and diastole are similar, and false FHR halving occurs when the duration of systole and diastole are very short. Regardless, both scenarios result in false-positive errors.

In-phase and quadrature (IQ) demodulation using two in-phase quadrature channels for demodulation can differentiate movements in opposite directions (towards the sound source and away from the sound source) and for this reason, is widely used in radar detection, telecommunication, and ultrasonic detection.^7–10 During FHR monitoring, the heart wall moves closer and then away from the probe during systole and diastole. Based on this, IQ demodulation may distinguish these two movements and theoretically resolve false FHR doubling and halving. In this study, we investigated the efficacy of IQ demodulation in EFM at reducing false FHR doubling and halving.

Materials and methods

Clinical information

This is a prospective cohort study. A total of 263 pregnant women who delivered at Peking University Shenzhen Hospital between August 2019 and July 2020 were prospectively enrolled in the study. The inclusion criteria were: (1) delivery between 37 and 42 gestational weeks; (2) a singleton pregnancy; (3) a head-down fetal position; and (4) no contraindications for vaginal delivery. The exclusion criteria were: (1) severe pregnancy complications; (2) skin disease, infection, or trauma at the measuring site.

Monitoring began when the cervix was 2–3  cm and continued until delivery. Raw fetal Doppler audio signals and internal and external cardiotocography (CTG) curves were recorded to compare the false FHR doubling and halving by internal electrode monitoring, EFM with IQ demodulation (external), and EFM with conventional modulation (external). The study was split into two cohorts. In cohort 1, the incidences of FHR doubling or halving were compared between IQ demodulation and conventional demodulation. In cohort 2, the incidences of FHR doubling or halving were compared between IQ demodulation, conventional demodulation, and internal FHR monitoring.

This study was approved by the Medical Ethics Committee of Peking University Shenzhen Hospital (with the unique ID: PUshenzhenH 140610). Informed consent for the study was obtained from all participants.

Devices

The EFM with IQ demodulation (Edan, Shenzhen, China; F15 EFM with internal monitoring) and the EFM with conventional demodulation (Edan; F3 EFM) were used in this study.

Methods

Simultaneous external monitoring by both EFM with conventional demodulation and EFM with IQ demodulation was established using a split-line ultrasound probe to detect and ensure FHR acquisition was synchronous in the same subject. The split-line ultrasonic probe is shown in Figure 1. The FHR data were uploaded to the central monitoring system for analysis. Clinically indicated pregnant women received internal monitoring with disposable helical electrodes in addition to external monitoring. In these cases, both internal and external CTG curves were recorded on the EFM.

Figure 1.

Split-line ultrasound probe (one end into the electronic fetal heart rate monitor with conventional demodulation and the other end into the electronic fetal heart rate monitor with IQ demodulation). IQ: In-phase and quadrature.

In cohort 1, both FHR acquisitions were obtained from the same monitoring probe. The beat-to-beat interval of FHRs in raw fetal Doppler audio signals was used as the reference to compare the incidences of FHR doubling or halving. Data associated with invalid FHR signals due to probe displacement from the fetal heart position were excluded from the analysis. The incidences of doubling and halving, and the reduction rate of doubling incidences and halving incidences were calculated.

In cohort 2, FHR acquisitions may have varied due to differences in internal vs. external monitoring signals and their associated interference, although the signals were obtained at the same time from the same subject. Data associated with invalid FHR signals from the probe or poor helical electrode contact were excluded from the analysis. The internal monitoring value was used as the reference to compare the incidences of doubling or halving with IQ demodulation or conventional demodulation, as well as the correlation and concordance between IQ demodulation, conventional demodulation, and internal monitoring.

The primary indicator were the incidences of doubling and halving, calculated by n/N × 100%; the reduction rates calculated by (conventional demodulation (n) − IQ demodulation (n))/conventional demodulation (n) × 100%. The secondary indicator: the correlation and concordance between IQ demodulation, conventional demodulation, and internal monitoring. Figure 2 shows FHR doubling and halving.

Figure 2.

FHR doubling and halving. A FHR doubling during contraction; B FHR halving after position change. FHR: Fetal heart rate.

Statistical analysis

IBM SPSS Statistics v25 (IBM Corp., Armonk, New York, USA) was used for data analysis. Count data were analyzed with the Chi-squared test, and measurement data were analyzed with paired t-test and Bland-Altman analysis.¹¹ A P < 0.05 was considered statistically significant. Correlation coefficients and concordance ranges were calculated for IQ demodulation vs. internal monitoring and conventional demodulation vs. internal monitoring.

Results

Comparison between IQ demodulation and conventional demodulation in EFM on FHR doubling and halving

Cohort 1 consisted of 225 pregnant women aged 28.71 ± 3.55  years. Their gestational age was 39.01 ± 1.21 gestational weeks. The beat-to-beat interval of FHRs in the raw fetal Doppler audio signals was used as the reference to compare the incidences of doubling and halving on external monitoring. Figure 3 shows the CTG and corresponding raw fetal Doppler audio signals of a subject with FHR doubling during contraction in the second stage of labor. FHR slowed during contraction, which resulted in a significant variation in CTG curves between IQ demodulation and conventional demodulation (red box). Analysis of the CTG curve (arrow) in Figure 4 showed concordance in the calculated FHR between IQ demodulation and the raw signals, while the calculated value from conventional demodulation was twice as high, which explains the accelerated FHR during EFM with conventional demodulation when in fact FHR slowed during contraction.

Figure 3.

Reduction in false FHR doubling during contraction (FHR slowed during contraction). The grid at the top is the CTG curve (with FHR curves in the top grid and the contraction curve in the bottom grid), and the waveform at the bottom shows raw signals obtained at the position indicated by the arrow. The black FHR curve was obtained with conventional demodulation, and the pink curve was obtained with IQ demodulation (downshift by 20  bpm). The heartbeat cycle was approximately 0.75  seconds, that is, the heart rate was 80  bpm (arrow) based on raw signals, approximately 80  bpm based on IQ demodulation, and approximately 160  bpm based on conventional demodulation. The 160  bpm FHR curve (black) in the red box is the FHR doubling of the 80  bpm FHR. bpm: Beats per minute; CTG: Cardiotocography; FHR: Fetal heart rate; IQ: In-phase and quadrature.

Figure 4.

Scatter diagrams of the durations of doubling and halving. A FHR doubling; B FHR halving. FHR: Fetal heart rate; IQ: In-phase and quadrature.

FHR from all participants was monitored for a total of 835,870  seconds. The results showed a doubling time of 3401  seconds (0.407%, 3401/835,870) and a halving time of 2918  seconds (0.349%, 2918/835,870) with conventional demodulation, and a doubling time of 241  seconds (0.029%, 241/835,870) and a halving time of 589  seconds (0.070%, 589/835,870) with IQ demodulation. IQ demodulation reduced doubling by approximately 93% (3160/ 3401) and halving by approximately 80% (2329/2918) relative to conventional demodulation (P < 0.01).

Table 1 shows the incidences of FHR doubling and halving over time. IQ demodulation significantly reduced doubling and halving at different time points (P < 0.01). Figure 4 shows the scatter diagram of the durations of doubling and halving. With IQ demodulation, the duration of FHR halving was mostly close to 0 second, and occasional doubling was transient. With conventional demodulation, FHR doubling and halving lasted longer.

Table 1.

The incidences of FHR doubling and halving under IQ demodulation vs. conventional demodulation in cohort 1 women (n = 225), n(%).

	FHR doubling(s)				FHR halving(s)
Items	>0	>5	>10	>15	>0	>5	>10	>15
IQ demodulation	26 (11.6)	16 (7.1)	8 (3.6)	4 (1.8)	42 (18.7)	27 (12.0)	15 (6.7)	13 (5.8)
Conventional demodulation	64 (28.4)	62 (27.6)	52 (23.1)	42 (18.7)	68 (30.2)	61 (27.1)	51 (22.7)	43 (19.1)
χ 2	20.056	32.816	37.231	34.966	8.134	16.330	23.011	18.356
P	<0.001	<0.001	<0.001	<0.001	0.006	<0.001	<0.001	<0.001

FHR: Fetal heart rate; IQ: In-phase and quadrature; s: Seconds.

Figure 5 shows the CTG and corresponding raw fetal Doppler audio signals of a subject with FHR halving during the first stage of labor. The subject had a fever (38.2°C) with uterine inertia and fetal tachycardia. An analysis of the CTG curve (arrow) in Figure 5 showed concordance in the calculated FHR between IQ demodulation and raw signals, while the value calculated from conventional demodulation was about half of the raw signals, which explains the halved FHRs observed during EFM with conventional demodulation.

Figure 5.

Reduction in false halving during fetal tachycardia (fetal tachycardia in a pregnant woman with fever). The grid at the top is the CTG curve (with FHR curve in the top grid and the contraction curve in the bottom grid), and the waveform at the bottom shows raw signals obtained at the position indicated by the arrow. The black FHR curve was obtained with conventional demodulation, and the pink curve was obtained with IQ demodulation (downshift by 20  bpm). The heartbeat cycle was approximately 0.3  seconds, that is, the heart rate was 200  bpm (arrow) based on raw signals, approximately 200  bpm based on IQ demodulation, and approximately 100  bpm based on conventional demodulation. The 100  bpm FHR curve (black) in the red box is the FHR halving of the 200  bpm FHR. bpm: Beats per minute; CTG: Cardiotocography; FHR: Fetal heart rate; IQ: In-phase and quadrature.

Comparison between IQ demodulation, conventional demodulation, and internal monitoring on FHR doubling and halving

Cohort 2 consisted of 38 pregnant women aged 29.51 ± 2.61  years. Their gestational age was 39.31 ± 1.11 gesta-tional weeks. The internal monitoring value was used as the reference to compare the incidences of doubling and halving on external monitoring. The FHR from all participants was monitored for a total of 98,561  seconds. The results showed a doubling time of 238  seconds (0.241%, 238/98,561) and a halving time of 235  seconds 0.238%, 235/98,561) with conventional demodulation, and a doubling time of 30  seconds (0.030%, 30/98,561) and a halving time of 81  seconds (0.082%, 81/98,561) with IQ demodulation (Chi-squared test, P < 0.001).

Table 2 shows the incidences of doubling and halving over different times (n = 38). IQ demodulation significantly reduced the incidences of FHR doubling and halving relative to conventional demodulation. There were no significant differences in the incidences of doubling or halving between IQ demodulation and internal monitoring (both P > 0.05), but there were significant differences in the incidences of FHR doubling and halving within 15 seconds between conventional demodulation and internal monitoring (both P < 0.05).

Table 2.

The incidences of FHR doubling and halving with IQ demodulation, conventional demodulation, and internal monitoring in cohort 2 women (n = 38), n (%).

	FHR doubling(s)				FHR halving(s)
Items	>0	>5	>10	>15	>0	>5	>10	>15
Internal monitoring	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
IQ demodulation	3 (7.9)∗	1 (2.6)∗	1 (2.6)∗	1 (2.6)∗	5 (13.2)∗	4 (10.5)∗	2 (5.3)∗	2 (5.3)∗
Conventional demodulation	8 (21.1)†	8 (21.1)†	6 (15.8)†	4 (10.5)‡	8 (21.1)†	8 (21.1)†	8 (21.1)†	5 (13.2)‡

FHR: Fetal heart rate; IQ: In-phase and quadrature; s: Seconds.

^∗Compared with internal monitoring group, P > 0.05.

^†Compared with internal monitoring group, P < 0.05.

^‡Compared with internal monitoring group, P > 0.05.

Figure 6 shows two CTG curves of a subject in the second stage of labor with IQ demodulation, conventional demodulation, and internal monitoring.

Figure 6.

Internal vs. external monitoring. The CTG curve with FHR curves in the top grid and the contraction curve in the bottom grid. The black FHR curve was obtained with internal monitoring, the blue FHR curve with IQ demodulation (upshift by 20  bpm), and the pink FHR curve with conventional demodulation (downshift by 20  bpm). Occasional transient doubling and halving were visible with conventional demodulation (red box), while the curve of IQ demodulation was stable and concordant to internal monitoring. Short transient FHR halving and transient doubling during contraction are visible with conventional demodulation (red box). IQ demodulation and internal monitoring are concordant. bpm: Beats per minute; FHR: Fetal heart rate; IQ: In-phase and quadrature.

Concordance of IQ demodulation, conventional demodulation, and internal monitoring

Concordance analysis of the calculated FHR and a paired t-test comparison of the correlation coefficients between different monitoring methods was assessed in cohort 2. The Bland-Altman analysis was used to calculate the concordance range with the 95% confidence interval of the difference in FHR. The paired t-test showed IQ demodulation was more closely correlated with internal monitoring than conventional demodulation (0.986 vs. 0.947). The Bland-Altman analysis showed the concordance range was significantly narrower for IQ demodulation vs. internal monitoring compared to conventional demodulation vs. internal monitoring ((−5.32, 6.01) vs.(−10.87, 11.46)). Figure 7 shows the scatter plot of the Bland-Altman analysis, where the green dotted lines indicate the concordance range.

Figure 7.

Scatter diagram of Bland-Altman analysis. A IQ demodulation vs. internal monitoring. B Conventional demodulation vs. internal monitoring. The concordance range (green dotted line) is significantly narrower in A than B ((−5.32,6.01) vs. (−10.87,11.46)). bpm: Beats per minute. IQ: In-phase and quadrature.

Discussion

Currently, FHR monitoring from EFMs can be acquired by using electrodes or Doppler ultrasound. Electrodes can be placed internally onto the fetus to directly acquire fetal electrocardio signals or attached externally on the mother's abdomen to indirectly acquire fetal electrocardio signals. Although internal electrode monitoring is a minimally invasive procedure, this technique brings a potential risk of infection because the amniotic membrane must be ruptured to attach a spiral electrode on the fetal scalp to directly acquire fetal electrocardio signals.^12–14 Internal electrode monitoring is used more often in developed countries and is rarely used in developing countries.

Some new technique obtains mixed signals from the mother and fetus through one or more electrodes, which are then processed to isolate the fetal electrocardio signals.^15,16 The external electrode monitoring has no current clinical application because of its susceptibility to maternal myoelectric interference during contraction and is difficult to obtain the signals of fetuses with excess fat.

Doppler ultrasound monitoring is the most commonly used FHR monitoring method and is widely used in pregnant women. However, the current technical limitations on Doppler ultrasound makes it less accurate to calculate FHR when compared to electrode monitor-ing.^17,18 Moreover, Doppler ultrasound is prone to false doubling and halving rates which are associated with maternal movement and fetal entry into the birth canal.

In recent years, many researchers have explored alternative methods to reduce miscalculations of FHR in order to achieve more accurate and reliable FHR monitoring. Varady et al. (Germany)¹⁹ and Chourasia et al. (India)²⁰ proposed a two-channel FHR acquisition device. This device uses one channel to acquire the FHR from a stethoscope, and another channel to acquire ambient noise with a sensor. Together, these signals are then processed to suppress noise and calculate the FHR.

Upon Doppler ultrasound interrogation, two main signals are received at the transducer: (1) the echo signal from rhythmic heart movements during systole and diastole and (2) random interference signals. Therefore, there is a need to accurately extract heart movement signals and suppress random interference. Conventional EFMs only use one channel for demodulation and cannot differentiate between systolic and diastolic movements and as a result, are prone to report false FHR doubling and halving.

In this study, we used an EFM with IQ demodulation to accurately differentiate the heart movements in opposite directions (towards the sound source vs. away from the sound source during systolic vs. diastolic movements) and demodulate echo signals with two quadrature channels, thereby effectively reducing false FHR doubling and halving. We investigated the incidences of FHR doubling and halving in EFM with IQ demodulation compared to other monitoring methods. In cohort 1, external monitoring methods were compared (EFM with IQ demodulation vs. EFM with conventional demodulation). In cohort 2, internal and external monitoring methods were compared. Additionally, the concordance between internal and external monitoring methods was assessed in cohort 2. We found that IQ demodulation reduced FHR doubling and FHR halving. There was no significant difference in the incidences of FHR doubling or halving between IQ demodulation and internal monitoring. The Bland-Altman analysis showed a significantly narrower concordance range for IQ demodulation vs. internal monitoring when compared to conventional demodulation vs. internal monitoring ((−5.32,6.01) vs. (−10.87,11.46)).

This study also has several limitations. In FHR monitoring, the correct judgment of fetal status depends on the reliability of CTG, and a clear and legible CTG curve is closely related to the placement of the probe and the position of the pregnant woman. The EFM based on IQ demodulation can greatly reduce the miscalculation rate, but the method is still an indirect detection of heart rate based on ultrasound Doppler, which will still be limited by the principle of ultrasonic Doppler technology, such as misjudgment of fetal position before monitoring, incorrect placement and poor fixation of the probe, strong fetal movement, descending of the fetal head during delivery, failure to adjust the probe in time when the pregnant woman's position changes, etc, which leads to fetal heart's deviation from ultrasound coverage, so that the monitor cannot obtain a clear and stable fetal signal. Furthermore, the mother's abdominal aortic signal introduction, sitting in the delivery ball, abdominal pressure during childbirth, and other reasons lead to ultrasound probe to obtain the fetal heart signal mixed with other signals, so that the CTG messy, broken and even serious artifacts, reduce the readability of CTG graphics. Therefore, various interference factors should be eliminated before fetal monitoring to create a favorable monitoring environment.

In summary, EFM with IQ demodulation significantly reduces false FHR doubling and halving, and is more concordant to internal FHR monitor than that of conventional demodulation.

Acknowledgments

The authors thank the patients for their participation in this study. The authors thank the following additional investigators for their participation in the study: Jinqun Liu (Edan Instruments, Inc).

Funding

This research was supported by the Shenzhen Science and Technology Innovation Commission (JCYJ20180228162311024).

Author Contributions

Concept and design (Yiheng Liang; Shangrong Fan; Duijin Chen), Literature search (Shaomei Yan), Data acquisition (Ping Liu, Yun Li), Data analysis (Shangrong Fan), Statistical analysis (Shangrong Fan), Manuscript preparation, manuscript editing, and manuscript review (Yun Li, Shangrong Fan, Shaomei Yan). The manuscript has been read and approved by all the authors.

Conflicts of Interest

None.

Editor Note

Dunjin Chen and Shangrong Fan are Editorial Board Members of Maternal-Fetal Medicine. The article was subject to the journal's standard procedures, with peer review handled independently of these editors and their research groups.