Skip to content

Advertisement

You're viewing the new version of our site. Please leave us feedback.

Learn more

BMC Pregnancy and Childbirth

Open Access
Open Peer Review

This article has Open Peer Review reports available.

How does Open Peer Review work?

Open access intrapartum CTG database

  • Václav Chudáček1Email author,
  • Jiří Spilka1,
  • Miroslav Burša1,
  • Petr Janků2,
  • Lukáš Hruban2,
  • Michal Huptych1 and
  • Lenka Lhotská1
BMC Pregnancy and Childbirth201414:16

https://doi.org/10.1186/1471-2393-14-16

Received: 12 July 2013

Accepted: 6 December 2013

Published: 13 January 2014

Abstract

Background

Cardiotocography (CTG) is a monitoring of fetal heart rate and uterine contractions. Since 1960 it is routinely used by obstetricians to assess fetal well-being. Many attempts to introduce methods of automatic signal processing and evaluation have appeared during the last 20 years, however still no significant progress similar to that in the domain of adult heart rate variability, where open access databases are available (e.g. MIT-BIH), is visible. Based on a thorough review of the relevant publications, presented in this paper, the shortcomings of the current state are obvious. A lack of common ground for clinicians and technicians in the field hinders clinically usable progress. Our open access database of digital intrapartum cardiotocographic recordings aims to change that.

Description

The intrapartum CTG database consists in total of 552 intrapartum recordings, which were acquired between April 2010 and August 2012 at the obstetrics ward of the University Hospital in Brno, Czech Republic. All recordings were stored in electronic form in the OB TraceVue®;system. The recordings were selected from 9164 intrapartum recordings with clinical as well as technical considerations in mind. All recordings are at most 90 minutes long and start a maximum of 90 minutes before delivery. The time relation of CTG to delivery is known as well as the length of the second stage of labor which does not exceed 30 minutes. The majority of recordings (all but 46 cesarean sections) is – on purpose – from vaginal deliveries. All recordings have available biochemical markers as well as some more general clinical features. Full description of the database and reasoning behind selection of the parameters is presented in the paper.

Conclusion

A new open-access CTG database is introduced which should give the research community common ground for comparison of results on reasonably large database. We anticipate that after reading the paper, the reader will understand the context of the field from clinical and technical perspectives which will enable him/her to use the database and also understand its limitations.

Keywords

CardiotocographyIntrapartumCTGDatabaseSignal processingFetal heart rate

Background

Introduction

Fetal heart activity is the prominent source of information about fetal well being during delivery. Cardiotocography (CTG) – recording of fetal heart rate (FHR) and uterine contractions – enables obstetricians to detect fetus with deteriorating status (e.g. ongoing fetal hypoxia), which may occur even in a previously uncomplicated pregnancy. Even though fetus has its own natural defense mechanism to tackle the oxygen insufficiency during the delivery, in some cases only timely intervention can prevent adverse consequences. Hypoxia, with prevalence lying in the region of 0.6% [1] to 3.5% [2], is considered to be the third most common cause of newborn death [3].

Cardiotocography was introduced in late 1960s and is still the most prevalent method of intrapartum hypoxia detection. It did not, however, bring the expected improvements in the delivery outcomes in comparison to previously used intermittent auscultation [4]. Moreover, continuous CTG is the main suspect for growing percentage of cesarean sections (CS) for objective reasons, which increased in the last decade [5].

To tackle the problems of CTG three principal approaches were followed during the years. The first approach focused on improving low inter and intra-observer agreement [6]. In 1986 International Federation of Gynecology and Obstetrics (FIGO) introduced general guidelines [7] based on evaluation of macroscopic morphological FHR features and their relation to the tocographic measurement. But even though the guidelines are available for almost thirty years now, poor interpretation of CTG still persists [810]. Many tweaks to the guidelines were proposed during the years, but with no significant change to the inter-observer variability. For overview of the different guidelines we refer the reader to [11].

The second approach to improve the overall results of CTG looks for technical solutions to add more objective/quantitative evaluation of the state of the fetus using additional measurement techniques. Some used techniques are regionally/country limited in their use – such as fetal blood sampling [12, 13] or fetal pulse oxymetry [14]. The only widespread method is evaluation of averaged repolarization behavior of fetal ECG performed by STAN. Many studies were published comparing STAN-enhanced CTG with use of CTG only. The majority of studies proved that the addition of STAN indeed leads to slightly better fetal outcomes [15, 16] but problems were also reported [17, 18].

Attempts to computer evaluation of the CTG – the third approach – started almost immediately after introduction of the first unified FIGO guidelines. FIGO features became fundamental in most of the first clinically oriented approaches and automatically extracted morphological features played the major role in all automatic systems for CTG analysis [19, 20].

We suggest interested reader to refer to e.g. [21] for more details on CTG preprocessing and to e.g. [22] for more details on different features that had been extracted to characterize the FHR since the focus of this paper is rather on the database presentation.

When reviewing literature on automatic CTG processing, two things are striking. First, there is a large disconnection between approaches and goals in the clinical and technical papers. The clinical papers are mostly looking for applicable solutions to the clinically pressing issues (lack of agreement, critically misclassified recordings). The technical papers often use CTG data as just an another input to the carefully tuned classifiers. Most works use very small ad-hoc acquired datasets, differently sampled with various parameters used as outcome measures, though we have to concede that our previous works [22, 23] were done exactly in the same manner. It is hard to believe that it is more than 30 years when computer processing of CTG has begun [24] and since then, no common database of CTG records is availablea. There is no way how to compare/improve/disregard among different results. And that, in our opinion, hinders any significant progress towards the ultimate goal of a usable and working automated classification of the CTG recordings.

In this paper we present a novel open-access CTG database, which we will call further in the paper CTUUHB databaseb. It consists of CTG records and clinical information. We first provide a comprehensive overview of databases used in literature in the last years. Then we describe development of the CTU-UHB database. We discuss the criteria for selection of records for the database from clinical and technical point of view. At last, we present a detailed description of main clinical and technical parameters, which, in our opinion, are important for understanding and should be taken into account when using the database.

Overview of CTG databases used in literature

We performed a systematic search in the electronic database PUBMED including records up to February 2013. The language of studies was restricted to English. The various combination of the following keywords were used: cardiotocography, fetal heart rate, fetal heart rate monitoring, computer assisted diagnosis, fetal distress, asphyxia, hypoxemia, fetal hypoxia. In the selected articles the references were searched and articles that cited the paper were searched as well.

It is impossible to provide exhaustive review and, therefore, several inclusion criteria were applied to include all relevant works but keep the overview as clear as possible. First, if a CTG database was used in multiple works, we included the paper where the database was described in most detail. If the description was the same, we included the most recent paper, e.g. we preferred paper of Jezewski et al. [25] rather than of Czabanski et al. [26]. Second, only those works that used intrapartum CTG signals were considered, e.g. we did not include the work of H. Ocak [27] since he worked with UCI Cardiotocography Data Setc. Third, we preferred journal papers and works that attempted to show results with regards to objective annotation (pH, base excess, etc.).

Our search of CTG databases used in other studies (with applied selection criteria) resulted in inclusion of 22 works. Due to the space limitation the overview had to be split into two tables, Tables 1 and 2. Table 1 presents used databases regarding the CTG signals and clinical parameters, namely: type of acquisition (ultrasound Doppler (US), direct fetal electrocardiogram measurement (FECG)); antepartum (ante.) or intrapartum (inte.) phase; stage of labor (I. or II.); length of FHR signal; time to actual delivery; use of uterine contractions (UC), description of inclusion criteria; description of clinical data; evaluation type: objective (obj.), subjective (subj.), or combination of both (comb.); number of total cases. The number of cases varies from study to study, the lowest being around 50 cases, and the highest being 7568 cases. Table 2 presents the overview of databases from classification point of view. It is apparent that in each paper different criteria for classes division were used, thus, making any comparison of results between different studies virtually impossible.
Table 1

Overview of databases used in various works I

Reference

Acquisition

Timing

Labor stage

FHR sig. used

Time to delivery

UC used

Incl.

Clinical

Evaluation

# Total

    

[min]

[min]

 

criteria

info.

type

cases

Nielsen et al. 1988 [28]

N/A

intra.

I.

30

N/A

Yes

No

No

obj.

50

Chung et al. 1995 [29]

FECG

intra.

N/A

N/A

N/A

Yes

Yes

Yes

obj.

73

Keith et al. 1995 [30]

N/A

intra.

N/A

> 120

Until del.

Yes

No

Yes

comb.

50

Bernardes et al. 1998 [31]

US, FECG

ante., intra.

I.,II.

Until del.

Yes

No

Yes

obj.

85

Maeda et al. 1998 [32]

N/A

intra.

N/A

50

N/A

No

No

No

subj.

49

Lee et al. 1999 [33]

FECG

intra.

N/A

N/A

Yes

No

No

subj.

53

Chung et al. 2001 [34]

US

ante., intra.

I.,II.

N/A

120

No

No

Yes

comb.

76

Strachan et al. 2001 [35]

FECG

intra.

I.,II.

> 30

Until del.

Yes

No

Yes

obj.

679

Siira et al. 2005 [36]

FECG

intra.

I.,II.

60

95% bellow 9

Yes

Yes

Yes

obj.

334

Cao et al. 2006 [37]

US, FECG

intra.

N/A

30

N/A

Yes

No

No

subj.

148

Salamalekis et al. 2006 [38]

US

intra.

I.,II.

N/A

Until del.

No

Yes

Yes

comb.

74

Georgoulas et al. 2006 [39]

FECG

intra.

I.,II.

20–60

Until del.

No

No

No

obj.

80

Gonçalves et al. 2006 [40]

US, FECG

intra.

I.,II.

32–60

Until del.

No

Yes

Yes

obj.

68

Costa et al. 2009 [41]

FECG

intra.

I.,II.

Until del.

Yes

Yes

Yes

obj.

148

Elliott et al. 2010 [42]

N/A

intra.

I.,II.

> 180

Until del.

Yes

Yes

Yes

subj.

2192

Warrick et al. 2010 [43]

US, FECG

intra.

I.,II.

> 180

Until del.

Yes

Yes

No

obj.

213

Jezewski et al. 2010 [25]

US

ante., intra.

N/A

N/A

Yes

Yes

Yes

obj.

749

Helgason et al. 2011 [44]

FECG

intra.

I.,II.

> 30

Until del.

Yes

No

No

comb.

47

Chudacek et al. 2011 [22]

US, FECG

intra.

I.,II.

20

Until del.

No

Yes

Yes

comb.

552

Spilka et al. 2012 [23]

US, FECG

intra.

I.,II.

20

Until del.

No

Yes

No

obj.

217

Georgieva et al. 2013 [45]

N/A

intra.

I.,II.

Until del.

No

Yes

Yes

obj.

7568

Czabanski et al. 2013 [46]

N/A

ante., intra.

N/A

60

N/A

No

Yes

Yes

subj.

2124

Legend: “N/A” – information not available, “–” – authors used the whole available FHR signal without specifying the length, – 2124 recordings, 333 woman.

Table 2

Overview of databases used in various works II

Reference

Classes (categories)

Division criteria for classes

# classes

# cases in classes

# total cases

Nielsen et al. 1988 [28]

Normal; pathological

Apgar 1 min. < 7 or pH < 7.15 or BE < -10

2

34; 16

50

Chung et al. 1995 [29]

Normal; abnormal

pH < 7.15

2

65; 8

73

Keith et al. 1995 [30]

5-tier scoring system

17 clinicians, pH, BDecf, Apgar

5

38; 12

50

Bernardes et al. 1998 [31]

norm.; susp.; pathol.

pH, Apgar, neonatology

3

56; 22; 7

85

Maeda et al. 1998 [32]

norm.; susp.; pathol.

manual clinical rules

3

12; 18; 19

49

Lee et al. 1999 [33]

Normal CTG; decels.

1 clinician

2

N/A

53

Chung et al. 2001 [34]

Normal; presumed distress; acidemic

norm. FHR; abnorm. & pH > 7.15; abnorm. & pH < 7.15

3

36; 26; 14

76

Strachan et al. 2001 [35]

Normal; abnormal

pH ≤ 7.15 & BDecf > 8

2

608; 71

679

Siira et al. 2005 [36]

Normal; acidemic

pH < 7.05

2

319; 15

334

Cao et al. 2006 [37]

reassuring; NR

2 clinicians

2

102; 44

148

Salamalekis et al. 2006 [38]

Normal; NR [NR & pH > 7.20; NR & pH < 7.20]

FIGO, pH < 7.20

2

32; 42

74

Georgoulas et al. 2006 [39]

Normal; at risk

pH > 7.20; pH < 7.10

2

20; 60

80

Costa et al. 2009 [41]

Omniview-SisPorto 3.5 alerts

pH < 7.05

2

7; 141

148

Elliott et al. 2010 [42]

Normal; abnormal

BDecf ≥ 12 & NE

2

60; 2132

2192

Warrick et al. 2010 [43]

Normal; pathological

BDecf < 8; BDecf ≥ 12

2

187; 26

213

Jezewski et al. 2010 [25]

Normal; abnormal

Apgar N/A min. < 7 or birth weight < 10 th perc. or pH < 7.2

2

28% abnorm.

749

Helgason et al. 2011 [44]

FIGO-TN; FIGO-FP; FIGO-TP

norm. FHR & pH ≥ 7.30; abnorm. & pH ≥ 7.30; abnorm. & pH ≤ 7.05

3

15; 17; 15

47

Chudacek et al. 2011 [22]

norm.; susp.; pathol.

3 clinicians

3

139; 306; 107

552

Spilka et al. 2012 [23]

Normal; pathological

pH < 7.15

2

123; 94

217

Georgieva et al. 2013 [45]

Normal; adverse

pH < 7.1 & neonatology

2

N/A

7568

Czabanski et al. 2013 [46]

Normal; abnormal

Apgar 10 min. < 5

2

306; 27

2124

Abbreviations: NR – non-reassuring, NE – neonatal encephalopathy, – 2124 recordings, 333 woman.

Construction and content

Ethics statement

The CTG recordings and clinical data were matched by anonymized unique identifier generated by the hospital information system. The timings of CTG records were matched to stages of labor (first and second stage) and were made relative to the time of birth, thus also de-identified. This study was approved by the Institutional Review Board of University Hospital Brno; all women signed informed consent.

Data collection

The data were collected between 27 th of April 2010 and 6 th of August 2012 at the obstetrics ward of the University Hospital in Brno (UHB), Czech Republic. The data consisted of two main components, the first were intrapartum CTG recordings and the second were clinical data.

The CTGs were recorded using STAN S21 and S31 (Neoventa Medical, Mölndal, Sweden) and Avalon FM40 and FM50 (Philips Healthcare, Andover, MA) fetal monitors. All CTG signals were stored in an electronic form in the OB TraceVue®;system (Philips) in a proprietary format and converted into text format using proprietary software provided by Philips. Each CTG record contains time information and signal of fetal heart rate and uterine contractions both sampled at 4 Hz. When a signal was recorded using internal scalp electrode it also contained T/QRS ratio and information about biphasic T-wave. From 9164 intrapartum recordings the final database of 552 carefully selected CTGs was created keeping in consideration clinical as well as technical point of view; the details about recordings selection are provided further.

The clinical data were stored in the hospital information system (AMIS) in the relational database. Complete clinical information regarding delivery and fetal/maternal information were obtained. The clinical data included: delivery descriptors (presentation of fetus, type of delivery and length of the first and second stage), neonatal outcome (seizures, intubation, etc.), fetal and neonatal descriptors (sex, gestational week, weight, etc.), and information about mother and possible risk factors. For the final CTU-UHB database, clinical data were exported from relational database and converted into Physionet text format.

Data selection and criteria considered

The selection procedure of the records was based on both clinical and CTG signal parameters and the process is shown in Figure 1.
Figure 1

Selection of recordings for the final database. Flowchart diagram describing the process of data selection for the final database.

Clinical criteria

In the following paragraphs we describe criteria that were used for records exclusion. Additionally we discuss shortly the criteria that were included in the description of the final database but were not used for records exclusion.

Clinical selection criteria

The following parameters were taken into account for inclusion of recordings into the final database. References in this section refer to a description of particular parameter.

  • Women’s Age – although the women’s high age plays significant role in the probability of congenital diseases, for the intrapartum period no significance was found [47]. Low age (maternal age <18 years) could have an adverse effect [48] and such records were therefore excluded.

  • Week of gestation – maturity of the fetus plays significant role in the shape and behavior of the FHR antepartum as well as intrapartum [49]. Therefore the selection was limited to mature fetuses (weeks of gestation ≥37) according to the last menses counting, which was in majority cases confirmed by ultrasound measurement during antepartum check-ups.

  • Known fetal diseases – fetuses with known congenital defects or known intrauterine growth restriction (IUGR) that could influence the FHR and/or outcome of the delivery were excluded from the database. Additionally, postnatally detected defects were consulted and two cases with transposed large veins were left in the set, since these two particular changes should not have influenced the FHR.

  • Type of gravidity – only singleton, uncomplicated pregnancies were included.

  • Type of delivery – the majority of the database consists of vaginal deliveries. Nevertheless to increase the number of cases with pathological outcome in the database, 16 CS recordings with pH≤7.05 were included and consequently a control group consisting of 30 CS with normal outcomes was also included to enable separate evaluation if necessary.

Additional clinical criteria provided

Together with criteria used for selection, the following criteria were considered and are available together with the CTG data:

  • Sex of the fetus – both sexes were included even though the sex of fetus significantly influences the outcome according to Bernardes et al. [50].

  • Parity – although the first delivery can be “more difficult” in general clinical sense [51] it is the same from the point of view of the FHR recording.

  • Risk factors – to be able to describe and identify the a priori high-risk pregnancies. We have included risk factors that could have influenced the state of the baby before entering the labor. For full review of the parameters and further references we recommend paper of Badawi et al. [52]. The final risk factors included in the database were gestational diabetes, preeclampsia, maternal fever (>37.5°C), hypertension and meconium stained fluid.

  • Drugs – especially those administered during delivery were considered only with regard to their influence on FHR. Opiates may influence the FHR directly but are rarely used in the Czech Republic during delivery and were not used in any of the cases included in the database. Therefore, we do not provide information about drugs administration in the database. Note that e.g. oxytocin used for enhancement of the uterine activity influences the FHR in majority indirectly, via increase of uterine activity, and thus can be assessed from the CTG alone.

  • Other criteria – complementary information in order to offer insight why e.g. operative delivery was chosen. These include: induced delivery, type of presentation (occipital/breech), no progress of labor, dystocia cephalocorporal (incoordinate uterine activity), dystocia cephalopelvic.

Labor outcome measures

Since our main intention was to prepare database that could be used for comparison of different automated approaches we have selected only those recordings that included umbilical artery pH. We added all additional outcome measures that were available for the recording in the hospital information system. Some of these measures are often misused and we will discuss their disadvantages below.

Outcome measure selection criteria

To enable objective classification the pH measure was considered as essential for the evaluation of the database.

  • Umbilical artery pH (pH) – is the most commonly used outcome measure, sign of respiratory hypoxia. Records with missing pH were excluded. Following suggestion by Rosen [53] records, which had values of pCO 2 outside 95 th percentile [54] were excluded except those with pH≤7.05, which even according to [54] should be approached with care.

Additional outcome measures provided

Even though the is pH is the most commonly used measure, we included additional measures such as following:

  • Base excess (BE) – is often used in the clinical setting as a sign for metabolic hypoxia, but is often false positive [53].

  • Base deficit in extracellular fluid (BDecf) – is according to Rosén et al. [53] a better measure of metabolic hypoxia than BE. Still pH remains more robust measure and according to last study of Georgieva et al. remains the most informative [45].

  • Neonatology – complete neonatological reports were acquired for all the cases in pre-prepared database. No severe cases of neonatal morbidity were found, no hypoxic ischemic encephalopathy, no seizures (for details on neonatal morbidity see [55]).

  • Subjective evaluation of the outcome of the delivery based on Apgar’s score (Apgar), where five categories are used to assess the newborn child in 1 s t , 5 th and 10 th minute [56].

The complete database was used for inter-intra observer variability study. In this study 9 senior obstetricians evaluated CTG signals. The clinical evaluation will be added to the database as soon as processed.

Signal criteria

When the data were filtered according to the clinical information, we have applied the following criteria on CTG records:

  • Signal length – we have decided to include 90 minutes preceding the delivery, where the delivery time represents also the time when the objective (pH, etc.) evaluation of labor was acquired.

    1. I.

      stage – the length of the I. stage was limited to a maximum of 60 minutes in order to keep recordings easily comparable. The minimal length was dependent on the pH of the records in question – to include as much abnormal records as possible. Thus the minimal length of the I. stage of 30 minutes was required for recording with pH ≤7.15 and 40 minutes for others. The time distance from the end of the I.stage to birth was not allowed to be larger than 30 minutes.

       
    2. II.

      stage &#x2013; based on our previous experience with analysis of the II. stage of labor (active pushing phase), we limited the II. stage to 30 minutes at maximum. This also limits the possibility of adverse events occurring in the II. stage, which could disconnect CTG recording in the I. stage with objective evaluation of the delivery.

       
  • Given the restriction above the signals are 30(40)–90 minutes long depending on a) the length of the II. stage and also b) available signal in the I. stage. No signal ends earlier than 30 minutes before delivery.

  • Missing signal – amount of missing signal was, except for the II. stage, kept to possible minimum. Nevertheless the trade-off between having full signal and having recordings with abnormal outcomes had to be made. No more than 50 % of signal was allowed to be missing in the I. stage.

  • Noise and artifacts – these are a problem especially for the recordings acquired by the ultrasound probe. Certainly in some recordings maternal heart rate is intermittently present. But even though it can pose a challenge for user of the database it also reflects the clinical reality.

  • Type of measurement device – the database is composed as a mixture of recordings acquired by ultrasound doppler probe, direct scalp measurement or combination of both – again reflecting the clinical reality at the obstetrics ward of UHB.

Description of the Database

Records for the CTU-UHB database were selected based on clinical and technical criteria described above. Table 3 provides overview of patient and labor outcome measure statistics and Table 4 presents main parameters regarding the CTG signals. The CTG signals were transformed from proprietary Philips format to open Physionet format [57], all data were anonymized at the hospital and de-identified (relative time) at the CTU side. An example of one CTG record is shown in Figure 2.
Table 3

Patient and labor outcome statistics for the CTG-UHB cardiotocography database

506 – Vaginal (44 – operative); 46 – Caesarean Section

US = 412; DECG = 102; US-DECG = 35; N/A = 3

 

Mean (Median)

Min

Max

Comment

Maternal age (years)

29.8

18

46

Over 36y: 40.

Parity

0.43 (0)

0

7

 

Gravidity

1.43 (1)

1

11

 

Gestational age (weeks)

40

37

43

Over 42 weeks: 2

pH

7.23

6.85

7.47

 

BE

-6.36

-26.8

-0.2

 

BDecf (mmol/l)

4.60

-3.40

  

Apgar 1min

8.26 (8)

1

10

AS1 < 3: 18

Apgar 5min

9.06 (10)

4

10

AS5 < 7: 50

Neonate’s weight (g)

3408

1970

4750

SGA: 17; LGA: 44

Neonate’s sex (F/M)

259 / 293

 

Abbreviations: AS1, AS5 – Apgar score at 1 st and 5 th minute respectively; SGA, LGA – fetus small, large for gestational age.

Table 4

CTG signal statistics for the CTG-UHB cardiotocography database

506 – Vaginal (44 – operative); 46 – Caesarean Section

US = 412; DECG = 102; US-DECG = 35; N/A = 3

 

Mean

Min

Max

Comment

Length of I. stage (min)

225

45

648

 

Length of II. stage (min)

11.87

0

30

 

Dist. SignalEnd to Birth (min)

2.70

0

29

Over 10 min: 9

Noisy data W1 (%)

12.38

0

74

 

Missing data W1 (%)

3.59

0

87

 

Overall W1 (%)

15.98

0

89

Over 50%: 18

Noisy data W2 (%)

13.42

0

49

 

Missing data W2 (%)

0

0

0

 

Overall W2 (%)

13.14

0

49

Over 25%: 98

Noisy data II. stage (%)

22.62

0

91

 

Missing data II. stage (%)

8.47

0

100

 

Overall II. stage (%)

31.26

0

100

Over 50%: 97

W1 – 30 minute-long window beginning 60 minutes before end of the first stage of labor, W2 – 30 minute-long window beginning 30 minutes before end of the first stage of labor.

Figure 2

Record of fetal heart rate (FHR) and uterine contractions (UC). An example record from the database. Fetal Heart Rate at the top, and uterine contractions at the bottom. The end of I. stage of labor is marked with blue line and an arrow.

CTG database – vaginal deliveries

The main part of the CTG database consists of 506 intrapartum recordings delivered vaginally. It means the deliveries got always to the II. stage of labor (fully dilated cervix, periodical contractions), even though not all deliveries had active pushing period. Some were delivered operatively by means of forceps or vacuum extraction (VEX). The main outcome measures are presented in Tables 3 and 4.

Please note the column “Comment”, which gives additional information either with regard to the number of potential outliers or points out interesting features of the database such as number of pathological cases based on certain parameters or quality of the recording in each window.

CTG database – deliveries by Caesarean Section

The database was selected to have the majority of intrapartum recordings with vaginal delivery. Nevertheless due to low number of cases with severely-abnormal outcomes, we have decided to add all recordings delivered by Cesarean Section (CS) with abnormal outcomes that conformed with the requirements mentioned above (16 CS records). Additional 30 CS recordings with normal outcome were randomly selected and added as a control-group. This control should enable the user of the database to evaluate CS recordings separately, if necessary. The details of the CS part of the database can be found in Tables 3 and 4.

Additional perspectives on the database

In addition to the tables presented above four more tables are included in the Additional files section. Additional file 1: Table S5 and Additional file 2: Table S6 show the structure of the database with respect to umbilical cord artery pH values. The main reason for presentation of theses tables is to allow clear visibility of the features’ values and their change with different pH. We are well aware of the article of Yeh [58], which states that there is weak relation between umbilical artery pH after delivery and negative outcome to the baby. Nevertheless the pH still remains the main “objective” parameter – a summary of the delivery – as clearly presented in [45]. Finally, two short Tables S7 and S8 in the Additional file 3 describe different risk factors presented in the database again related to umbilical cord artery pH on vaginal and CS part of the database, respectively.

Utility and discussion

The CTU-UHB database is the first open-access database for research on intrapartum CTG signal processing and analysis. In the following paragraphs we will highlight the subjects, that could, if unobserved, lead to problems with use of the database.

The CTU-UHB users should be aware that there is a possible noise in the clinical data, since some information had to be mined from free text. Even though the whole data was carefully checked it is possible that some noise is present. However, this noise should not significantly disrupt any results obtained. Also we note that, due to the selection process, the database is biased from normal population (disproportional amount of low pH deliveries etc.) but this bias is evident in all other studies and, more importantly, if we would keep the database in the original form, the potential users would be forced to select the data themselves – resulting in different selection criteria and making, again, any comparison across studies infeasible.

From Table 2 it is evident that each study used different outcome measures, or their combinations. Again, this makes any comparison across studies infeasible. There are two main sources of evaluation: objective by e.g. umbilical artery pH, which is prominent example, and subjective evaluation by experts according to their knowledge and/or guidelines used. For the clarity reasons we will focus on non-expert outcome measures, as those are discussed in another recently prepared paper of our research group (Spilka et al.: Analysis of CTG interpretation by expert-obstetricians, submitted).

Exact relation of umbilical pH after delivery to CTG/FHR is so far not fully understood, time between the recording and actual delivery plays a crucial role, and it seems that pH is only weakly correlated to clinical annotation [59, 60]. The best example is the timely CS due to suspect CTG – the CTG is suspect/pathological but the intervention prevented the baby to get into real asphyxia that would be reflected in the pH value. Yeh [58] claims that there is only weak relation between pH and actual outcome for the well-being of the baby. Following work [45] on the same database relates low values of pH to higher risk of clinical complications. From the studies on cerebral palsy in neonates pH and BDecf are recommended as preferred measures [61] even though [62] says the opposite. Additionally intrapartum events and cerebral palsy are very rarely related by the intrapartum hypoxia only [63] and the real outcome of the delivery can be seen only in several years-long follow up [64].

There is no general agreement on the threshold, which should be used to distinguish between the normal and abnormal outcome of the delivery. There are several used/supported possibilities.

  • Cerebral Palsy – pH ≤7 together with BDecf ≤12 [65] was found to be related to significant increase of the cerebral palsy.

  • Pathological – pH ≤7.05 [16] is used as a threshold by most studies. Even though this value is not used unanimously, it is generally accepted as the threshold between pathological and not-pathological delivery outcomes. Combination with BDecf was used e.g. in [66].

  • Abnormal (lower than normal) pH <7.10 [67, 68] – this value is supported by recent works on the large Oxford database as well as used heuristically at the UHB as a sign of severe problems with the delivery.

  • Abnormal (lower than normal) pH <7.15 [69, 70].

  • Abnormal (lower than normal) pH <7.20 [71] this particular value is also mentioned as an threshold to abnormal outcome pH by Irelands’ obstetrician guidelines.

Regarding the Base deficit/base excess values the BDefc established by [72] is the only usable measure for assessment of metabolic hypoxia [53]. Nevertheless in many papers as well as in the clinical practice the base excess (BE) is used erroneously [53].

In general pH is more robust but is affected more by respiratory asphyxia, BDecf is more about metabolic asphyxia. Regarding the reliability of the objective measurements they are in general much more reliable than any expert opinion. Nevertheless biochemical measures are very dependent on the measuring procedure – pH is in general considered to be more robust than the BDecf where it is necessary to use consideration about the pCO 2 measurements [54].

Among undocumented parameters in the database, which could influence the shape and/or different properties of FHR one could count e.g. smoking [73], which can increase the heart rate, or epidural analgesia [74, 75] responsible for intermittent fetal bradycardia due maternal intermittent hypotension. Some risk factors can influence the look of the FHR such as diabetes mellitus, where FHR looks more immature [76]. Also technical parameters can influence the FHR itself – such as size of autocorrelation window for deriving FHR from ultrasound [77], or the derived parameters – such as power spectral density (PSD) of FHR, which can be affected by the type of interpolation [78].

Length of the data used is usually limited by the availability of the data. Really long signals (spanning from the check-in to delivery) enable us to create individualized approach to each fetus with regard to its starting point [53]. We have much more information to analyze, which can be positive [79] or confusing based on the point of view [80]. Short signals such as e.g. 70 min long [81] enables us to try to find direct relation between the features measured and the outcome.

Another question is how to treat the II. stage of labor. Will the length of it confuse the extracted features? General opinion on the second stage is that it is different from the I. stage – in shape of the signal. It is also very often noisy and it differs even in the clinical treatment where obstetricians are much more lenient to apply operative delivery in case of unclear trace [82].

Conclusion

The CTU-UHB database is the first open-access database available for research on intrapartum CTG signal processing and analysis. It is available at the Physionet website. The database is reasonably large and allows researchers to develop algorithms/methods for CTG analysis and classification. Using CTU-UHB database – different approaches can be easily compared one with another in the objective fashion. Intuitively, the use of common database can stimulate the research in CTG signal processing and classification and move the community to the ultimate goal – an automatic analysis of CTG and its possible extension to a decision support system for clinicians.

Availability and requirements

The database is published at http://physionet.org/physiobank/database/ctu-uhb-ctgdb/ as an open-access database on the website dedicated to research in cardiology, heart rate variability and related fields.

The database is free to use for non-commercial purposes given that any publication using the database refers to this paper.

Endnotes

a The only published attempt in this direction was found in [83], but it was discontinued since.

b Czech Technical University – University Hospital Brno.

c UCI Cardiotocography Data Set includes only CTG features not the signals – for more details see [84].

Abbreviations

CTG: 

Cardiotocography

FHR: 

Fetal heart rate

FECG: 

Fetal electrocardiography

UC: 

Uterine contractions

US: 

Ultrasound

BE: 

Base excess

BDecf: 

Base deficit in extracellular fluid

PSD: 

Power spectral density.

Declarations

Acknowledgements

The presented work was partially supported by Ministry of Healthcare of the Czech Republic grant No.NT11124-6/2010, by MH CZ – DRO (FNBr, 65269705) and SGS grant of the CTU SGS13/203/OHK3/3T/13. We would like to thank Philips Healthcare for their software for extraction of the data from the OB TraceVue system to readable format and Ms. Monika Jiráčková from S&T for facilitating the contacts to Philips.

Authors’ Affiliations

(1)
Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague
(2)
Obstetrics and Gynecology clinic, University Hospital

References

  1. Heintz E, Brodtkorb TH, Nelson N, Levin LA: The long-term cost-effectiveness of fetal monitoring during labour: a comparison of cardiotocography complemented with ST analysis versus cardiotocography alone. BJOG. 2008, 115 (13): 1676-1687. 10.1111/j.1471-0528.2008.01935.x.View ArticlePubMedPubMed CentralGoogle Scholar
  2. Strachan BK, van Wijngaarden WJ, Sahota D, Chang A, James DK: Cardiotocography only versus cardiotocography plus PR-interval analysis in intrapartum surveillance: a randomised, multicentre trial. FECG Study Group. Lancet. 2000, 355 (9202): 456-459. 10.1016/S0140-6736(00)82012-7.View ArticlePubMedGoogle Scholar
  3. d’Aloja E, Müller M, Paribello F, Demontis R, Faa A: Neonatal asphyxia and forensic medicine. J Matern Fetal Neonatal Med. 2009, 22 (Suppl 3): 54-56.View ArticlePubMedGoogle Scholar
  4. Alfirevic Z, Devane D, Gyte GML: Continuous cardiotocography (CTG) as a form of electronic fetal monitoring (EFM) for fetal assessment during labour. Cochrane Database Syst Rev. 2006, 3 (3): CD006066-PubMedGoogle Scholar
  5. Melman S, Schoorel E, Dirksen C, Anneke Kwee LS, de Boer F, Jonkers M, Mallory D, Woiski BWM, Doornbos JP, Visser H, Anjoke JM, Huisjes MMP, Delemarre FM, Kuppens SM, Robert Aardenburg IMVD, Vrouenraets FP, Lim FT, Kleiverda G, van der Salm PC, de Boer K, Sikkema MJ, Nijhuis JG, Hermens RP, Scheepers HC: SIMPLE: Implementation of recommendations from international evidence-based guidelines on caesarean sections in the Netherlands. Protocol for a controlled before and after study. Implementation Sci. 2013, 8: 3-10.1186/1748-5908-8-3.View ArticleGoogle Scholar
  6. Lotgering FK, Wallenburg HC, Schouten HJ: Interobserver and intraobserver variation in the assessment of antepartum cardiotocograms. Am J Obstet Gynecol. 1982, 144 (6): 701-705.View ArticlePubMedGoogle Scholar
  7. FIGO: Guidelines for the use of fetal monitoring. Int J Gynecol Obstet. 1986, 25: 159-167.Google Scholar
  8. Bernardes J, Costa-Pereira A, de Campos DA, van Geijn HP, Pereira-Leite L: Evaluation of interobserver agreement of cardiotocograms. Int J Gynaecol Obstet. 1997, 57: 33-37. 10.1016/S0020-7292(97)02846-4.View ArticlePubMedGoogle Scholar
  9. Blix E, Sviggum O, Koss KS, Oian P: Inter-observer variation in assessment of 845 labour admission tests: comparison between midwives and obstetricians in the clinical setting and two experts. BJOG. 2003, 110: 1-5.View ArticlePubMedGoogle Scholar
  10. Blackwell SC, Grobman WA, Antoniewicz L, Hutchinson M, Gyamfi Bannerman C: Interobserver and intraobserver reliability of the NICHD 3-Tier Fetal Heart Rate Interpretation System. Am J Obstet Gynecol. 2011, 205 (4): 378 e1-378.e5.View ArticleGoogle Scholar
  11. de Campos DA, Ugwumadu A, Banfield P, Lynch P, Amin P, Horwell D, Costa A, Santos C, Bernardes J, Rosén K: A randomised clinical trial of intrapartum fetal monitoring with computer analysis and alerts versus previously available monitoring. BMC Pregnancy Childbirth. 2010, 10: 71-10.1186/1471-2393-10-71.View ArticleGoogle Scholar
  12. Henderson Z, Ecker JL: Fetal scalp blood sampling – limited role in contemporary obstetric practice: part I. Lab Med. 2003, 34 (7): 548-553. 10.1309/M86KBYF00HDDJY9L.View ArticleGoogle Scholar
  13. Henderson Z, Ecker JL: Fetal scalp blood sampling – limited role in contemporary obstetric practice: part II. Lab Med. 2003, 34 (8): 594-600. 10.1309/GR5QVQWCJHYU370W.View ArticleGoogle Scholar
  14. Dildy GA, Clark SL, Loucks CA: Intrapartum fetal pulse oximetry: past, present, and future. Am J Obstet Gynecol. 1996, 175: 1-9. 10.1016/S0002-9378(96)70242-9.View ArticlePubMedGoogle Scholar
  15. Norén H, Blad S, Carlsson A, Flisberg A, Gustavsson A, Lilja H, Wennergren M, Hagberg H: STAN in clinical practice–the outcome of 2 years of regular use in the city of Gothenburg. Am J Obstet Gynecol. 2006, 195: 7-15.View ArticlePubMedGoogle Scholar
  16. Amer-Wåhlin I, Maršál K: ST analysis of fetal electrocardiography in labor. Semin Fetal Neonatal Med. 2011, 16: 29-35. 10.1016/j.siny.2010.09.004.View ArticlePubMedGoogle Scholar
  17. Ojala K, Vääräsmäki M, Mäkikallio K, Valkama M, Tekay A: A comparison of intrapartum automated fetal electrocardiography and conventional cardiotocography–a randomised controlled study. BJOG. 2006, 113 (4): 419-423. 10.1111/j.1471-0528.2006.00886.x.View ArticlePubMedGoogle Scholar
  18. Westerhuis M, Kwee A, van Ginkel AA, Drogtrop AP, Gyselaers WJA, Visser GHA: Limitations of ST analysis in clinical practice: three cases of intrapartum metabolic acidosis. BJOG. 2007, 114 (10): 1194-1201. 10.1111/j.1471-0528.2007.01236.x.View ArticlePubMedGoogle Scholar
  19. de Campos DA, 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 (3): 260-264.Google Scholar
  20. Guijarro-Berdiñas B, Alonso-Betanzos A: Empirical evaluation of a hybrid intelligent monitoring system using different measures of effectiveness. Artif Intell Med. 2002, 24: 71-96. 10.1016/S0933-3657(01)00091-4.View ArticlePubMedGoogle Scholar
  21. Neilson DR, Freeman RK, Mangan S: Signal ambiguity resulting in unexpected outcome with external fetal heart rate monitoring. Am J Obstet Gynecol. 2008, 198 (6): 717-724. 10.1016/j.ajog.2008.02.030. http://dx.doi.org/10.1016/j.ajog.2008.02.030. [Monica].View ArticlePubMedGoogle Scholar
  22. Chudáček V, Spilka J, Huptych M, Janků P: Automatic evaluation of intrapartum fetal heart rate recordings: a comprehensive analysis of useful features. Physiol Meas. 2011, 32: 1347-1360. 10.1088/0967-3334/32/8/022.View ArticlePubMedGoogle Scholar
  23. Spilka J, Chudáček V, Huptych M, Georgoulas G, Stylios C, Koucký M: Using nonlinear features for fetal heart rate classification. Biomed Signal Process Control. 2012, 7 (4): 350-357. 10.1016/j.bspc.2011.06.008.View ArticleGoogle Scholar
  24. Dawes GS, Visser GH, Goodman JD, Redman CW: Numerical analysis of the human fetal heart rate: the quality of ultrasound records. Am J Obstet Gynecol. 1981, 141: 43-52.View ArticlePubMedGoogle Scholar
  25. Jezewski M, Czabański R, Wróbel J, Horoba K: Analysis of extracted cardiotocographic signal features to improve automated prediction of fetal outcome. Biocybern Biom Eng. 2010, 30 (4): 29-47.Google Scholar
  26. Czabanski R, Jezewski J, Matonia A, Jezewski M: Computerized analysis of fetal heart rate signals as the predictor of neonatal acidemia. Expert Syst Appl. 1184, 39 (15): 6-11860.Google Scholar
  27. Ocak H: A medical decision support system based on support vector machines and the genetic algorithm for the evaluation of fetal well-being. J Med Syst. 2013, 37 (2): 9913-View ArticlePubMedGoogle Scholar
  28. Nielsen PV, Stigsby B, Nickelsen C, Nim J: Computer assessment of the intrapartum cardiotocogram. II. The value of compared with visual assessment. Acta Obstet Gynecol Scand. 1988, 67 (5): 461-464. 10.3109/00016348809004260.View ArticlePubMedGoogle Scholar
  29. Chung TK, Mohajer MP, Yang ZJ, Chang AM, Sahota DS: The prediction of fetal acidosis at birth by computerised analysis of intrapartum cardiotocography. Br J Obstet Gynaecol. 1995, 102 (6): 454-460. 10.1111/j.1471-0528.1995.tb11317.x.View ArticlePubMedGoogle Scholar
  30. Keith RD, Beckley S, Garibaldi JM, Westgate JA, Ifeachor EC, Greene KR: A multicentre comparative study of 17 experts and an intelligent computer system for managing labour using the cardiotocogram. Br J Obstet Gynaecol. 1995, 102 (9): 688-700. 10.1111/j.1471-0528.1995.tb11425.x.View ArticlePubMedGoogle Scholar
  31. Bernardes J, de Campos DA, Costa-Pereira A, Pereira-Leite L, Garrido A: Objective computerized fetal heart rate analysis. Int J Gynaecol Obstet. 1998, 62 (2): 141-147. 10.1016/S0020-7292(98)00079-4.View ArticlePubMedGoogle Scholar
  32. Maeda K, Utsu M, Makio A, Serizawa M, Noguchi Y, Hamada T, Mariko K, Matsumoto F: Neural network computer analysis of fetal heart rate. J Matern Fetal Investig. 1998, 8 (4): 163-171.PubMedGoogle Scholar
  33. Lee A, Ulbricht C, Dorffner G: Application of artificial neural networks for detection of abnormal fetal heart rate pattern: a comparison with conventional algorithms. J Obstet Gynecol. 1999, 19 (5): 482-485. 10.1080/01443619964256.View ArticleGoogle Scholar
  34. Chung DY, Sim YB, Park KT, Yi SH, Shin JC, Kim SP: Spectral analysis of fetal heart rate variability as a predictor of intrapartum fetal distress. Int J Gynaecol Obstet. 2001, 73 (2): 109-116. 10.1016/S0020-7292(01)00348-4.View ArticlePubMedGoogle Scholar
  35. Strachan BK, Sahota DS, van Wijngaarden WJ, James DK, Chang AM: Computerised analysis of the fetal heart rate and relation to acidaemia at delivery. BJOG. 2001, 108 (8): 848-852.PubMedGoogle Scholar
  36. Siira SM, Ojala TH, Vahlberg TJ, Jalonen JO, Välimäki IA, Rosén KG, Ekholm EM: Marked fetal acidosis and specific changes in power spectrum analysis of fetal heart rate variability recorded during the last hour of labour. BJOG. 2005, 112 (4): 418-423. 10.1111/j.1471-0528.2004.00454.x.View ArticlePubMedGoogle Scholar
  37. Cao H, Lake DE, Ferguson JE 2nd, Chisholm CA, Griffin MP, Moorman JR: Toward quantitative fetal heart rate monitoring. IEEE Trans Biomed Eng. 2006, 53: 111-118. 10.1109/TBME.2005.859807.View ArticlePubMedGoogle Scholar
  38. Salamalekis E, Hintipas E, Salloum I, Vasios G, Loghis C, Vitoratos N, Chrelias C, Creatsas G: Computerized analysis of fetal heart rate variability using the matching pursuit technique as an indicator of fetal hypoxia during labor. J Matern Fetal Neonatal Med. 2006, 19 (3): 165-169. 10.1080/14767050500233290.View ArticlePubMedGoogle Scholar
  39. Georgoulas G, Stylios CD, Groumpos PP: Predicting the risk of metabolic acidosis for newborns based on fetal heart rate signal classification using support vector machines. IEEE Trans Biomed Eng. 2006, 53 (5): 875-884. 10.1109/TBME.2006.872814.View ArticlePubMedGoogle Scholar
  40. Gonçalves H, Rocha AP, de Campos DA, 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): 847-855. 10.1007/s11517-006-0105-6.View ArticlePubMedGoogle Scholar
  41. Costa A, Ayres-de Campos D, Costa F, Santos C, Bernardes J: Prediction of neonatal acidemia by computer analysis of fetal heart rate and ST event signals. Am J Obstet Gynecol. 2009, 201 (5): 464.e1-464.e6.View ArticleGoogle Scholar
  42. Elliott C, Warrick P, Graham E, Hamilton E: Graded classification of fetal heart rate tracings: association with neonatal metabolic acidosis and neurologic morbidity. Am J Obstet Gynecol. 2010, 202 (3): 258.e1-258.e8.View ArticleGoogle Scholar
  43. Warrick P, Hamilton E, Precup D, Kearney R: Classification of normal and hypoxic fetuses from systems modeling of intrapartum cardiotocography. IEEE Trans Biom Eng. 2010, 57 (4): 771-779.View ArticleGoogle Scholar
  44. Helgason H, Abry P, Goncalves P, Gharib C, Gaucherand P, Doret M: Adaptive multiscale complexity analysis of fetal heart rate. IEEE Trans Biomed Eng. 2011, 58: 2186-2193.View ArticleGoogle Scholar
  45. Georgieva A, Payne SJ, Moulden M, Redman CWG: Artificial neural networks applied to fetal monitoring in labour. Neural Comput Appl. 2013, 22: 85-93. 10.1007/s00521-011-0743-y.View ArticleGoogle Scholar
  46. Czabański R, Jezewski J, Wróbel J, Sikora J, Jezewski M: Application of fuzzy inference systems for classification of fetal heart rate tracings in relation to neonatal outcome. Ginekol Pol. 2013, 84: 38-43.PubMedGoogle Scholar
  47. Callaway LK, Lust K, McIntyre HD: Pregnancy outcomes in women of very advanced maternal age. Aust N Z J Obstet Gynaecol. 2005, 45: 12-16. 10.1111/j.1479-828X.2005.00333.x.View ArticlePubMedGoogle Scholar
  48. Berglund S, Grunewald C, Pettersson H, Cnattingius S: Risk factors for asphyxia associated with substandard care during labor. Acta Obstet Gynecol Scand. 2010, 89: 39-48. 10.3109/00016340903418751.View ArticlePubMedGoogle Scholar
  49. Park MI, Hwang JH, Cha KJ, Park YS, Koh SK: Computerized analysis of fetal heart rate parameters by gestational age. Int J Gynaecol Obstet. 2001, 74 (2): 157-164. 10.1016/S0020-7292(01)00423-4.View ArticlePubMedGoogle Scholar
  50. Bernardes J, Gonçalves H, Ayres-De-Campos D, Rocha A: Sex differences in linear and complex fetal heart rate dynamics of normal and acidemic fetuses in the minutes preceding delivery. J Perinatal Med. 2009, 37 (2): 168-176.View ArticleGoogle Scholar
  51. Singh T, Sankaran S, Thilaganathan B, Bhide A: The prediction of intra-partum fetal compromise in prolonged pregnancy. J Obstet Gynaecol. 2008, 28 (8): 779-782. 10.1080/01443610802431857.View ArticlePubMedGoogle Scholar
  52. Badawi N, Kurinczuk JJ, Keogh JM, Alessandri LM, O’Sullivan F, Burton PR, Pemberton PJ, Stanley FJ: Intrapartum risk factors for newborn encephalopathy: the Western Australian case-control study. BMJ. 1998, 317 (7172): 1554-1558. 10.1136/bmj.317.7172.1554.View ArticlePubMedPubMed CentralGoogle Scholar
  53. Rosén KG, Blad S, Larsson D, Norén H, Outram N: Assessment of the fetal bioprofile during labor by fetal ECG analysis. Expert Rev Obstet Gynecol. 2007, 2 (5): 609-620. 10.1586/17474108.2.5.609.View ArticleGoogle Scholar
  54. Kro GAB, Yli BM, Rasmussen S, Norén H, Amer-Wåhlin I, Saugstad OD, Stray-Pedersen B, Rosén KG: A new tool for the validation of umbilical cord acid-base data. BJOG. 2010, 117 (12): 1544-1552. 10.1111/j.1471-0528.2010.02711.x.View ArticlePubMedGoogle Scholar
  55. McIntyre S, Taitz D, Keogh J, Goldsmith S, Badawi N, Blair E: A systematic review of risk factors for cerebral palsy in children born at term in developed countries. Dev Med Child Neurol. 2012,, (in press)Google Scholar
  56. Finster M, Wood M: The Apgar score has survived the test of time. Anesthesiology. 2005, 102 (4): 855-857. 10.1097/00000542-200504000-00022.View ArticlePubMedGoogle Scholar
  57. Goldberger AL, Amaral LA, Glass L, Hausdorff JM, Ivanov PC, Mark RG, Mietus JE, Moody GB, Peng CK, Stanley HE: PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation. 2000, 101 (23): E215-E220. 10.1161/01.CIR.101.23.e215.View ArticlePubMedGoogle Scholar
  58. Yeh P, Emary K, Impey L: The relationship between umbilical cord arterial pH and serious adverse neonatal outcome: analysis of 51,519 consecutive validated samples. BJOG. 2012, 119 (7): 824-831. 10.1111/j.1471-0528.2012.03335.x.View ArticlePubMedGoogle Scholar
  59. Valentin L, Ekman G, Isberg PE, Polberger S, Maršál K: Clinical evaluation of the fetus and neonate. Relation between intra-partum cardiotocography, Apgar score, cord blood acid-base status and neonatal morbidity. Arch Gynecol Obstet. 1993, 253 (2): 103-115. 10.1007/BF02768736.View ArticlePubMedGoogle Scholar
  60. Schiermeier S, von Steinburg SP, Thieme A, Reinhard J, Daumer M, Scholz M, Hatzmann W, Schneider KTM: Sensitivity and specificity of intrapartum computerised FIGO criteria for cardiotocography and fetal scalp pH during labour: multicentre, observational study. BJOG. 2008, 115 (12): 1557-1563. 10.1111/j.1471-0528.2008.01857.x.View ArticlePubMedGoogle Scholar
  61. Pierrat V, Haouari N, Liska A, Thomas D, Subtil D, Truffert P, d’Etudes en Epidémiologie Périnatale G: Prevalence, causes, and outcome at 2 years of age of newborn encephalopathy: population based study. Arch Dis Child Fetal Neonatal Ed. 2005, 90 (3): F257-F261. 10.1136/adc.2003.047985.View ArticlePubMedPubMed CentralGoogle Scholar
  62. Low JA: The current crisis in obstetrics. J Obstet Gynaecol Can. 2005, 27 (11): 1031-1037.View ArticlePubMedGoogle Scholar
  63. Schifrin BS: The CTG and the timing and mechanism of fetal neurological injuries. Best Pract Res Clin Obstet Gynaecol. 2004, 18 (3): 437-456. 10.1016/j.bpobgyn.2004.03.001.View ArticlePubMedGoogle Scholar
  64. Ingemarsson I, Herbst A, Thorngren-Jerneck K: Long term outcome after umbilical artery acidaemia at term birth: influence of gender and duration of fetal heart rate abnormalities. Br J Obstet Gynaecol. 1997, 104 (10): 1123-1127. 10.1111/j.1471-0528.1997.tb10934.x.View ArticlePubMedGoogle Scholar
  65. MacLennan A: A template for defining a causal relation between acute intrapartum events and cerebral palsy: international consensus statement. BMJ. 1999, 319 (7216): 1054-1059. 10.1136/bmj.319.7216.1054.View ArticlePubMedPubMed CentralGoogle Scholar
  66. Westerhuis M, Moons KGM, van Beek E, Bijvoet SM, Drogtrop AP, van Geijn HP, van Lith JMM, Mol BWJ, Nijhuis JG, Oei SG, Porath MM, Rijnders RJP, Schuitemaker NWE, van der Tweel I, Visser GHA, Willekes C, Kwee A: A randomised clinical trial on cardiotocography plus fetal blood sampling versus cardiotocography plus ST-analysis of the fetal electrocardiogram (STAN) for intrapartum monitoring. BMC Pregnancy Childbirth. 2007, 7: 13-10.1186/1471-2393-7-13.View ArticlePubMedPubMed CentralGoogle Scholar
  67. Ross MG: Labor and fetal heart rate decelerations: relation to fetal metabolic acidosis. Clin Obstet Gynecol. 2011, 54: 74-82. 10.1097/GRF.0b013e31820a106d.View ArticlePubMedGoogle Scholar
  68. Fulcher B, Georgieva A, Redman C, Jones N: Highly comparative fetal heart rate analysis. Engineering in Medicine and Biology Society (EMBC), 2012 Annual International Conference of the IEEE. 2012, 3135-3138. http://ieeexplore.ieee.org/xpl/articleDetails.jsp?reload=true%26tp=%26arnumber=6346629%26queryText%3DHighly+comparative+fetal+heart+rate+analysis,View ArticleGoogle Scholar
  69. Georgoulas G, Stylios C, Nokas G, Groumpos P: Classification of fetal heart rate during labour using hidden Markov models. Neural Networks, 2004. 2004, Proceedings. 2004 IEEE International Joint Conference on Volume 3., 2471-2475. http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1381017\%26tag=1,Google Scholar
  70. Salamalekis E, Thomopoulos P, Giannaris D, Salloum I, Vasios G, Prentza A, Koutsouris D: Computerised intrapartum diagnosis of fetal hypoxia based on fetal heart rate monitoring and fetal pulse oximetry recordings utilising wavelet analysis and neural networks. BJOG. 2002, 109 (10): 1137-1142. 10.1111/j.1471-0528.2002.01388.x.View ArticlePubMedGoogle Scholar
  71. Maharaj D: Intrapartum fetal resuscitation: a review. Internet J Gynecol Obstet. 2008, 9 (2): 4800-4808.Google Scholar
  72. Siggaard-Andersen O, Huch R: The oxygen status of fetal blood. Acta Anaesthesiol Scand Suppl. 1995, 107: 129-135.View ArticlePubMedGoogle Scholar
  73. Oncken C, Kranzler H, O’Malley P, Gendreau P, Campbell WA: The effect of cigarette smoking on fetal heart rate characteristics. Obstet Gynecol. 2002, 99 (5 Pt 1): 751-755.PubMedGoogle Scholar
  74. Hill JB, Alexander JM, Sharma SK, McIntire DD, Leveno KJ: A comparison of the effects of epidural and meperidine analgesia during labor on fetal heart rate. Obstet Gynecol. 2003, 102 (2): 333-337. 10.1016/S0029-7844(03)00567-2.PubMedGoogle Scholar
  75. Cleary-Goldman J, Negron M, Scott J, Downing RA, Camann W, Simpson L, Flood P: Prophylactic ephedrine and combined spinal epidural: maternal blood pressure and fetal heart rate patterns. Obstet Gynecol. 2005, 106 (3): 466-472. 10.1097/01.AOG.0000173797.20722.a0.View ArticlePubMedGoogle Scholar
  76. Tincello D, White S, Walkinshaw S: Computerised analysis of fetal heart rate recordings in maternal type I diabetes mellitus. BJOG. 2001, 108 (8): 853-857.PubMedGoogle Scholar
  77. Jezewski J, Roj D, Wrobel J, Horoba K: A novel technique for fetal heart rate estimation from Doppler ultrasound signal. Biomed Eng Online. 2011, 10 (92): http://dx.doi.org/10.1186/1475-925X-10-92,Google Scholar
  78. Cesarelli M, Romano M, Ruffo M, Bifulco P, Pasquariello G, Fratini A: PSD modifications of FHRV due to interpolation and CTG storage rate. Biomed Signal Process Control. 2011, 6 (3): 225-230. 10.1016/j.bspc.2010.10.002.View ArticleGoogle Scholar
  79. Graatsma EM, Jacod BC, van Egmond LAJ, Mulder EJH, Visser GHA: Fetal electrocardiography: feasibility of long-term fetal heart rate recordings. BJOG. 2009, 116 (2): 334-337;. 10.1111/j.1471-0528.2008.01951.x. discussion 337–338View ArticlePubMedGoogle Scholar
  80. Sisco KM, Cahill AG, Stamilio DM, Macones GA: Is continuous monitoring the answer to incidentally observed fetal heart rate decelerations?. J Matern Fetal Neonatal Med. 2009, 22 (5): 405-409. 10.1080/14767050802556059.View ArticlePubMedGoogle Scholar
  81. Schiermeier S, Hatzmann H, Reinhard J: The value of Doppler cardiotocogram computer analysis system 70 minutes before delivery. Z Geburtshilfe Neonatol. 2008, 212 (5): 189-193. 10.1055/s-2008-1077015.View ArticlePubMedGoogle Scholar
  82. Sheiner E, Hadar A, Hallak M, Katz M, Mazor M, Shoham-Vardi I: Clinical significance of fetal heart rate tracings during the second stage of labor. Obstet Gynecol. 2001, 97 (5 Pt 1): 747-752.PubMedGoogle Scholar
  83. Costa Santos C, Ayres de Campos D, da Costa Pereira A, Bernardes J: An interactive web site for research on fetal heart rate monitoring. Obstet Gynecol. 2000, 95 (2): 309-311. 10.1016/S0029-7844(99)00569-4.PubMedGoogle Scholar
  84. Bache K, Lichman M: UCI Machine Learning Repository. UCI Machine Learning Repository University of California, Irvine, School of Information and Computer Sciences. 23 Dec. 2013, http://archive.ics.uci.edu/ml/datasets/Cardiotocography,Google Scholar
  85. Pre-publication history

    1. The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2393/14/16/prepub

Copyright

© Chudáček et al.; licensee BioMed Central Ltd. 2014

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Advertisement