Prediction of neonatal acidemia by computer analysis of fetal heart rate and ST event signals.

Abstract

The objective of the study was to evaluate the accuracy of computer analysis of fetal heart rate (FHR) and ST event signals in prediction of neonatal acidemia.
One hundred forty-eight FHR tracings were evaluated to identify red alerts provided by the system, based on automated analysis of FHR and ST event signals, and compared with the occurrence of umbilical artery acidemia (pH < or =7.05).
The presence of red alerts obtained sensitivity of 1.00 (95% confidence interval [CI], 0.56-1.00), specificity of 0.94 (95% CI, 0.89-0.97), positive predictive value (PPV) of 0.47 (95% CI, 0.22-0.72), negative predictive value (NPV) of 1 (95% CI, 0.96-1.00), positive likelihood ratio (PLR) of 17.6 (95% CI, 9.0-34.5), and negative likelihood ratio (NLR) of 0. When limiting analysis to red alerts that did not include ST data, sensitivity was 0.57 (95% CI, 0.20-0.88), specificity was 0.97 (95% CI, 0.92-0.99), PPV was 0.50 (95% CI, 0.17-0.82), NPV was 0.98 (95% CI, 0.93-0.99), PLR was 20.14 (95% CI, 6.3-64.2), and NLR was 0.44 (95% CI, 0.19-1.04).
Computer analysis of FHR and ST event signals provide higher accuracy in predicting neonatal academia.

Full text

Page 1

1
www.omniview.eu 24/10/2009

Omniview-Sisporto Published Papers’ Highlights

2009 AJOG - American Journal of Obstetrics and Gynecology
Prediction of neonatal acidemia by computer
analysis of fetal heart rate and ST event signals
Antonia Costa, MD; Diogo Ayres-de-Campos, PhD; Fernanda Costa, MD; Cristina Santos, MS; Joao
Bernardes, PhD

OBJECTIVE: The objective of the study was to evaluate the accuracy of computer analysis of fetal heart rate (FHR)
and ST event signals in prediction of neonatal acidemia.
STUDY DESIGN: One hundred forty-eight FHR tracings were evaluated to identify red alerts provided by the system,
based on automated analysis of FHR and ST event signals, and compared with the occurrence of umbilical artery
acidemia (pH <= 7.05).
RESULTS: The presence of red alerts obtained sensitivity of 1.00 (95% confidence interval [CI], 0.56-1), specificity of
0.94 (95% CI, 0.89-0.97), positive predictive value (PPV) of 0.47 (95% CI, 0.22-0.72), negative predictive value (NPV)
of 1 (95% CI, 0.96-1.00), positive likelihood ratio (PLR) of 17.6 (95% CI, 9.0-34.5), and negative likelihood ratio (NLR)
of 0. When limiting analysis to red alerts that did not include ST data, sensitivity was 0.57 (95% CI, 0.20-0.88),
specificity was 0.97 (95% CI, 0.92-0.99), PPV was 0.50 (95% CI, 0.17-0.82), NPV was 0.98 (95% CI, 0.93-0.99), PLR
was 20.14 (95% CI, 6.3-64.2), and NLR was 0.44 (95% CI, 0.19-1.04).
CONCLUSION: Computer analysis of FHR and ST event signals provide higher accuracy in predicting neonatal
acidemia.

Keywords: cardiotocography, computer assisted, electrocardiography, fetal hypoxia, fetal heart rate, fetal monitoring

Web link: http://dx.doi.org/10.1016/j.ajog.2009.04.033

2008 - Journal of Perinatal Medicine
Omniview-SisPorto 3.5 – a central fetal monitoring
station with online alerts based on computerized
cardiotocogram & ST event analysis
Diogo Ayres-de-Campos1,2,*, Paulo Sousa2, Antonia Costa1,2 and Joao Bernardes1,2
1 Hospital de S. Joao - Faculdade de Medicina do Porto, Universidade do Porto, Portugal
2 Instituto de Engenharia Biomedica, Universidade do Porto, Portugal

Abstract
Visual analysis of cardiotocograms is poorly reproducible and is currently recognized as the main weakness of the
STAN_ methodology. The Omniview-SisPorto® 3.5 program is the most recent version of a central monitoring system
that provides visual and sound alerts, based on computer analysis of cardiotocographic and ST event features. This
paper describes the program’s main characteristics and provides an overview of the system’s online alerts. Omniview-
SisPorto® 3.5 is the first central monitoring system to incorporate computerized analysis of cardiotocographic and ST
event features, providing
health professionals with online alerts for minor and major changes in monitored signals. The system is currently
undergoing extensive clinical evaluation.

Keywords: cardiotocography; computer-assisted; electrocardiography; fetal; fetal hypoxia

Web link: http://www.reference-global.com/doi/abs/10.1515/JPM.2008.030

Page 2

esCorrespondence: omniview@speculum.pt
Fetal heart rate (FHR) monitoring remains widely used as a method for detecting changes in fetal oxygenation that
can occur during labor. Yet deaths and long-term disablement from intrapartum hypoxia remain an important cause of
suffering for parents and families, even in industrialized countries. Confidential inquiries in the United Kingdom have
highlighted that as much as 50% of these deaths could have been avoided because they were caused by non
recognition of abnormal FHR patterns, poor communication between staff, or delay in taking appropriate action.

Omniview-SisPorto aims to help decrease the number of Neonatal Encephalopaties related to intrapartum Hypoxia. It
specifically aims to reduce human error in tracing interpretation and possible delays in taking appropriate actions by
providing health professionals with visual and sound alerts to fetal heart rate changes.

The following articles are a description of 3 published papers and connect them to examples and existing commercial
system features and benefits.


Prediction of neonatal acidemia by computer
analysis of fetal heart rate and ST event signals

The aim of this study was to evaluate the predictive capacity of the Omniview-SisPorto 3.5 system, when analyzing the
last hour of intrapartum STAN tracings to identify patients who went on to display umbilical artery acidemia at birth (ie,
umbilical artery pH >= 7.05).

MATERIALS AND METHODS
This prospective observational study was conducted
in a tertiary care university hospital between May
2005 and September 2007. Consecutive cases were
enrolled if they fulfilled the inclusion criteria. In all
cases the umbilical cord was doubly clamped
immediately after birth, and blood was aspirated from
both artery and vein.

Computer analysis of cardiotocogram (CTG) plus ST
tracings by the Omniview-SisPorto 3.5 system was
performed off-line for this study but using a similar
methodology to the one that is used when tracings
are being acquired in real time (ie, only signals
acquired until the present moment are available to
elicit alarms). This was accomplished using a
program feature, by which signals are processed
from the start of the file, 1 minute at a time, only
taking into account signals acquired until then, and a
log of alarms elicited on a minute-by minute basis is
recorded. Thus, no alerts that would be elicited
during real tracing acquisition are lost or changed.

Analysis of alerts was carried out only in the last 60
minutes of the tracing because this was thought to
bear the highest relationship with immediate newborn
outcome. Additional alerts were present in a small
number of patients before that period in both patients
with and without umbilical artery acidemia. However,
these were not included because they could have
elicited an intervention that led to the reversion of
what was only a temporary acidemia.

The last hour of the tracing was reviewed and the red
alerts provided by the system during that period were
compared with the occurrence of umbilical artery
acidemia at birth, defined as a pH value 7.05 or less.


Omniview-Sisporto Published Papers’ Highlights – 24/10/2009esp

2
RESULTS
A total of 193 consecutive patients were enrolled, and
45 were subsequently excluded:
19 because of disconnection of the scalp electrode
conditioning a large tracing-to-delivery interval and/or
signal loss, 21 because of inadequate umbilical blood
gases values, and five because of insufficient tracing
duration. This resulted in a total of 148 cases
available for analysis. 19 had maternal risk factors (8
hypertensive diseases, 8 diabetes, 1 anemia, 2
thrombophilias), 4 had
hydramnios, 1 oligohydramnios, 1 fetal growth
restriction), 9 had heavy meconium staining, 53
patients had an induced labor, and 16 patients had
prolonged rupture of membranes (> 16 hours). The
remaining patients were monitored internally because
of poor FHRsignal quality with external acquisition
methods.

The table below compares the appearance of red
alerts in the last 60 minutes, with the occurrence of
umbilical artery acidemia.

fetal risk factors (2

Page 3

esCorrespondence: omniview@speculum.pt Omniview-Sisporto Published Papers’ Highlights – 24/10/2009esp

3
There were 7 cases of umbilical artery pH of 7.05 or
less, and these are displayed in detail in a table
below.

No cases of neonatal
encephalopathy occurred in this series.

In the 148 studied cases, the occurrence of red alerts
in the last hour of the tracing obtained a sensitivity of
1.00 (95% CI, 0.56-1), a specificity of 0.94 (95% CI,
0.89-0.97), a positive predictive value of 0.47 (95%
CI, 0.22-0.72), a negative predictive value of 1.00
(95% CI, 0.96-1.00) in prediction of newborn
acidemia.

When evaluating red alerts elicited exclusively by
FHR analysis and not incorporating ST data, the
obtained sensitivity was 0.57 (95% CI, 0.20-0.88), the
specificity 0.97 (95% CI, 0.92-0.99), the
positive predictive value 0.50 (95% CI, 0.17-0.82),
the negative predictive value 0.98 (95% CI, 0.93-
0.99).

hypoxic ischemic
COMMENT
This is the first study to evaluate the accuracy of
intrapartum computerized analysis of FHR plus ST
signals. Confidence intervals for sensitivity and
positive predictive value are still wide because of the
occurrence of a small number of patients with
newborn acidemia. Further research is needed to
confirm the high sensitivity found in this study. On the
other hand, the small
surrounding the high specificity and negative
predictive values provide promising results for this
technology.

Recent evidence using
assessment has shown that the addition of ST
segment analysis can enhance the accuracy of fetal
monitoring, and our results also suggest this
hypothesis. Using the system’s red alerts that
incorporate both FHR and ST event features, all 7
acidemic fetuses were identified, and there were 8
false-positive results. Six of these patients had an
umbilical artery pH less than 7.20, suggesting that a
milder degree of hypoxemia was occurring.
confidence intervals
non computerized

Page 4

esCorrespondence: omniview@speculum.pt Omniview-Sisporto Published Papers’ Highlights – 24/10/2009esp

4

When ST event data were excluded, only 4 of the 7
acidemic cases were identified, and 4 false positives.
Thus, 3 acidemic fetuses would not have displayed
red alerts if isolated FHR monitoring had been
performed. It can be seen from the table that all
cases would have been detected if the orange alerts,
the system’s second most ominous alert, had also
been included.

The appearance of this alert should prompt further
investigation, including fetal blood sampling or STAN
monitoring. However, inclusion of orange alerts for
evaluation of the system’s accuracy would also cause
a sharp increase in the number of false positives.
False positives are the main drawback of visual FHR
analysis and also occur with the STAN technology.
They may be caused by fetal adaptations to the
challenges of labor rather than by noxious stimuli.
The system appears to have reduced but not
eliminated these cases, suggesting that further
refinement of some alerts will be required.



In conclusion, computer analysis assures a robust
approach to the issue of accuracy of FHR monitoring
in prediction of neonatal outcome. The present study
demonstrates that a high accuracy may be obtained
with computer analysis of FHR plus ST signals in
prediction of term fetuses that are born with umbilical
artery acidemia. Although this was the largest sample
size used to address this issue, confidence intervals
for specificity and positive predictive values are still
wide, and further studies are needed to confirm this
finding.

The system’s alerts, which appear together with a
description of the criteria that were identified to evoke
them, are expected to
professional’s attention and to provoke a reevaluation
of the case under a new light, but they do not
obviously guarantee a satisfactory outcome. The
effect of introducing such a system into routine
clinical practice requires the conduction of an
adequately sized randomized controlled clinical trial,
which is currently being prepared.
capture the health

Fig. 1: Example of a tracing with red alerts since one hour before delivery. Artery pH was 6,94.


Fig. 2: Example of a non-reassuring tracing but without red alerts. Artery pH was 7,19.



Omniview-SisPorto 3.5 – a central fetal monitoring
station with online alerts based on computerized
cardiotocogram & ST event analysis


This gives a brief description of the Omniview-SisPorto system, a program for automated analysis of CTGs, and
provides an overview of the system’s online alerts.

Description of the system
The Omniview-SisPorto system was designed as a
central monitoring station, allowing the combined
display of up to 16 CTG and/or STAN tracings on
the same computer screen – the main program
window (Figure 3).

Computer
automatically after 11 min of tracing acquisition, and is
updated every minute. This includes estimation of
fetal heart rate (FHR) baseline, identification of
accelerations and decelerations, and quantification of
processing of CTG features starts

Page 5

esCorrespondence: omniview@speculum.pt Omniview-Sisporto Published Papers’ Highlights – 24/10/2009esp

5
short- and long-term variability. A detailed display of
the results can be accessed by maximizing a single
tracing window. A combination of CTG analysis with
ST event features is used to provide online alerts,
which are displayed in the main program window, as
well as in maximized tracing windows.


Fig. 3: Window displaying 5 tracings

Tracing acquisition, display and storage is handled
automatically by the program, as it continuously
detects the connection and disconnection of fetal
monitors transmitting appropriate signal features.

External input is only required for identification of
patients. A multi-client-server architecture ensures
that a practically unlimited number of computers can
be connected via a local intranet to display tracings.
Exams can be printed automatically on tracing end or
at the user’s request, with or without the results of
automated analysis. Previously acquired tracings can
be processed as a whole, or analyzed as if they were
being acquired online, at normal or higher speeds.

Processing of information to produce online alerts
starts automatically after 11 min of acquisition, and is
updated every minute. Definitions used for alerts
were inspired on published consensus guidelines for
interpretation of CTG and combined CTG+ST
signals, but were adapted according to the results of
previous optimization and validation studies.

Alerts Description

The Normality criteria met alert was designed for
tracings for antepartum or as labour admission tests.
When all normality criteria regarding baseline,
accelerations, decelerations and variability are
present it produces a green alert without sound.

Attention alerts occur with changes that are not
commonly associated with fetal hypoxia, such as
signal loss, excessive uterine contractions, and
maternal heart rate monitoring. They produce a blue
visual alert.

Non-reassuring alerts occur with changes in the
FHR that are usually not associated with fetal
hypoxia, but may require continued monitoring and,
occasionally, further investigation: fetal tachycardia;
absent accelerations; repetitive but short-lasting
decelerations. They produce a yellow alert without
sound.
Very non-reassuring alerts occur with more
worrying changes in the FHR, which are usually
associated with a considerable risk of impending fetal
hypoxia: repetitive
decelerations; prolonged
monitoring and further measures are usually required
to avoid progress to relevant degrees of hypoxia.
They produce a blinking orange visual alert with
sound.
and
decelerations.
moderate-lasting
Careful
Pre-terminal alerts occur with signal changes that
are commonly associated with moderate-to-severe
fetal hypoxia: reduced long-term and short-term
variability; very prolonged deceleration or related to
ST events. Rapid measures are required in these
cases to avoid an abnormal fetal outcome, which
may include rapid reversal of the underlying cause or
immediate delivery.
List of system alerts
Reassuring (for
antepartum)
Normality criteria met

Attentioni
Signal loss
Excessive uterine
Maternal heart rate
monitoring?
ST signal loss

Attention extra (sound)
Large signal loss
Maternal heart rate
monitoring?



Non-reassuringi
Basal bradychardia
Baseline shift
Fetal tachycardia Basal
No accelerations
Decelerations
Repetitive decelerations

Very non-reassuring (sound)
Very repetitive decelerations
Prolonged deceleration
Repetitive decelerations and ST
event


Pre terminal (sound)
Reduced long-term variability
Repetitive decelerations and reduced long-term
variability
Tachycardia, decelerations and ST event
Repetitive decelerations and ST event
Decelerations and ST event
Very repetitive decelerations and ST event
Prolonged decelerations and ST event
Reduced short-term variability
Repetitive decelerations and reduced short-term
variability
Very prolonged deceleration
Prolonged deceleration with reduced variability

Page 6

esCorrespondence: omniview@speculum.pt Omniview-Sisporto Published Papers’ Highlights – 24/10/2009esp

6

Discussion

The Omniview-SisPorto system was designed to
incorporate the concepts of central monitoring,
computerized CTG analysis, and online alerts using
combined CTG+ST event data. The main aim of this
technology is to alert health professionals to
situations that may be associated with fetal hypoxia,
thus preventing that
overlooked in busy labor ward settings.



Computer quantification of short-term variability as
an adjunct to fetal electrocardiographic monitoring

Omniview-SisPorto is also intended as an adjunct to clinical decision, especially regarding CTG changes that are
poorly assessed by the human eye or when complex integration of CTG and ST information is needed. One
example that addresses to this issue was published in a response to the paper of Westerhuis et al. Limitations of
ST analysis in clinical practice: three cases of intrapartum metabolic acidosis (BJOG 2007;114:1194–1201).

Results
The paper by Westerhuis et al. illustrates three
cases of intrapartum fetal metabolic acidosis that
were missed or alerted to very late by STAN
analysis. The authors note the marked reduction in
variability occurring in the final segments of each
tracing.

We were interested to quantify these changes, so
we obtained the original fetal heart rate files from the
authors. With their permission, we have quantified
changes in variability over time in each of these
cases and detected episodes of reduced variability
in each of the cases.
such occurrences are

It is also intended as an adjunct to clinical decision,
especially regarding CTG changes that are poorly
assessed by the human eye or when complex
integration of CTG and ST information is needed.

It is hoped that this technology, when associated
with appropriate training of health professionals on
signal interpretation, clinical decision, and adequate
management, will contribute to reduce perinatal
morbidity and mortality consequent to intrapartum
events.

The example depicted below is one of those
cases as it if was acquired and analyzed in real
time in the labour ward with the current version of
Omniview-SisPorto.

The short-term variability, which is poorly
assessed by the human eye, degrades over the
last 1 and a half hour of the tracing. The system
would issue a first red alert for reduced short-term
variability 1h20m before the delivery and
continuously during the last hour.

This is an example that illustrates the difficulties of
CTG interpretation. In this case the newborn was
resuscitated and at 8 months had cerebral palsy.


Fig. 4:Example of a tracing with red alerts since one 1h 20 min before delivery. Artery pH was 6,79.


List of other international peer reviewed papers published and using the Sisporto system:

- Montenegro N, Bernardes J, Ayres-de-Campos D,
Matias A, Areias JC. Monitoring of cardiac-
extracardiac haemodynamics and automated fetal
heart rate preceding intrauterine death. Eur J Obstet
Gynecol Reprod Biol 1996;64:3-6.
- Bernardes J, Costa-Pereira A, Ayres-de-Campos D,
van Geijn HP, Pereira-Leite L. Evaluation of
interobserver agreement of cardiotocograms. Int J
Gynecol Obstet 1997;57:33-7.

Page 7

esCorrespondence: omniview@speculum.pt Omniview-Sisporto Published Papers’ Highlights – 24/10/2009esp

7
- Bernardes J, Costa-Pereira A. The effect of different
sampling intervals on the measurement of intrapartum
fetal heart rate variability (letter). Obstet Gynecol
1997;90(2):318-19
- Bernardes J, Costa-Pereira A. Some concerns about
the new research guidelines for interpretation of
electronic fetal heart rate monitoring (letter). Am J
Obstet Gynecol 1998;179(2);560-1.
- Bernardes J, Ayres-de-Campos D, Costa-Pereira
A, Pereira-Leite L, Garrido A. Objective computerized
fetal heart rate analysis. Int J Gynecol Obstet
1998;62:141-7.
- Bernardes J, Costa-Pereira A. Computerized
estimation of the baseline fetal heart rate in labour:
the low frequency line. Br J Obstet Gynaecol
1998;105(10):1128-9.
- Ayres-de-Campos D, Bernardes J. Early, variable
and late decelerations: can a consensus be reached
in their identification? Int J Gynecol Obstet
1999;65:305-6.
- Ayres-de-Campos D, Bernardes J, Costa-Pereira A,
Pereira-Leite L. Inconsistencies in classification by
experts of cardiotocograms and subsequent clinical
decision. Br J Obstet Gynaecol 1999;106:1307-10.
- Bernardes J, Costa-Pereira A, Calejo L, Ayres-de-
Campos D, Patrício B, Pereira-Leite L. Fetal heart rate
baselines in twins: interobserver agreement in
antepartum estimation.
2000;45(2):105-8.
J Reprod Med
- Ayres-de-Campos D, Bernardes J, Garrido A,
Marques-de-Sá J, Pereira-Leite L. SisPorto 2.0: a
program for automated analysis of cardiotocograms. J
Matern Fetal Med 2000;9(5):311-318.
- Bernardes J, Ayres-de-Campos D, Costa-Pereira A.
Further concerns about the National Institute of Child
Health and Human Development guidelines for
interpretation of electronic fetal monitoring (letter). Am
J Obstet Gynecol 2001;184(7):1587-8.
- Ayres-de-Campos D, Bernardes J, Marsal K,
Nickelsen C, Makarainen L, Banfield P, Xavier P,
Campos I. Can the reproducibility of fetal heart rate
baseline estimation be improved? Eur J Obstet
Gynecol Reprod Biol 2004;112(1):49-54.
- Ayres-de-Campos D, Bernardes J. Comparison of
fetal heart rate baseline estimation by SisPorto ® 2.01
and a consensus of clinicians. Eur J Obstet Gynecol
Reprod Biol 2004; 117:174-8.
- Ayres-de-Campos D, Costa-Santos C, Bernardes J,
for the SisPorto multicentre validation study group.
Prediction of neonatal state by computer analysis of
fetal heart rate tracings: the antepartum arm of the
SisPorto® multicentre validation study. Eur J Obstet
Gynecol Reprod Biol 2005; 118: 52-60.
- Gonçalves H, Rocha AP, Ayres-de-Campos D,
Bernardes J. Internal versus external intrapartum fetal
heart rate monitoring: effect on linear and nonlinear
parameters. Physiol Meas 2006;27:307-19.
- Gonçalves H, Rocha AP, Ayres-de-Campos D,
Bernardes J. Linear and nonlinear fetal heart rate
analysis of normal and acidemic fetuses in the
minutes preceding delivery. Med Biol Eng Comput
2006;44(10):847-55.
- Gonçalves H, Rocha AP, Ayres-de-Campos D,
Bernardes J. Linear and nonlinear analysis of heart
rate patterns associated with fetal behavioural states
in the antepartum period. Early Hum Develop
2007;83:585-91.
- Ayres-de-Campos D, Bernardes J, Kwee A,
Westehuis M, Visser GHA. Computer quantification of
short-term variability as
electrocardiographic monitoring
2007:114(11):1446-7.
an adjunct
(letter).
to fetal
BJOG
-George Georgoulas , Dimitris Gavrilis , Ioannis G.
Tsoulos ,Chrysostomos Stylios , João Bernardes ,
Peter P. Groumpos - Novel approach for fetal heart
rate classification introducing grammatical evolution.
Biomedical Signal Processing and Control (2007)
69–79
-Ayres-de-Campos D, Sousa P, Costa A,Bernardes J.
Omniview-SisPorto® 3.5—a central fetal monitoring
station with online alerts based on computerized
cardiotocogram ST event analysis. J Perinat Med
2008;36:260-4.
- Costa A, Ayres-de-Campos D, Costa F, et al.
Prediction of neonatal acidemia by computer analysis
of fetal heart rate and ST event signals.
Am J Obstet Gynecol 2009;201: