Induced Cesarean EHG DataSet (ICEHG DS): An open dataset with electrohysterogram records of pregnancies ending in induced and cesarean section delivery

Creator: Franc Jager

Published: Oct. 8, 2023. Version: 1.0.1

Citation

When using this resource, please cite:
Jager, F. (2023). Induced Cesarean EHG DataSet (ICEHG DS): An open dataset with electrohysterogram records of pregnancies ending in induced and cesarean section delivery (version 1.0.1). PhysioNet. https://doi.org/10.13026/zw34-n382.

Additionally, please cite the original publication:
Jager, F. (2023). An open dataset with electrohysterogram records of pregnancies ending in induced and cesarean section delivery. Scientific Data. 10, Article number: 669. https://doi.org/10.1038/s41597-023-02581-6

Please include the standard citation for PhysioNet:
Goldberger, A., Amaral, L., Glass, L., Hausdorff, J., Ivanov, P. C., Mark, R., ... & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation [Online]. 101 (23), pp. e215–e220.

Abstract

The surface electrohysterogram (EHG) has emerged as a promising diagnostic tool for non-invasive automated preterm birth prediction. This dataset contains 126 30-minute EHG records, recorded either early (23rd week) or later (31st week) during pregnancy, from pregnancies that ended in induced or cesarean section delivery. Data were collected at the University Medical Center Ljubljana, Slovenia, and are intended to support studies into improving the robustness of automated preterm birth prediction methods.

Background

Existing preterm birth prediction methods often ignore induced and cesarean deliveries. With the increasing frequency of such deliveries, it is critical to account for these cases. This dataset, ICEHG DS, was created to enhance understanding of EHG signal characteristics and support the development of better predictive models.

Methods

Data Collection

EHG records were collected between 1997 and 2006 at the Clinical Department of Perinatology, University Medical Center Ljubljana. They were recorded during routine checkups or hospital admissions for impending preterm labor. Women provided written informed consent, and ethical approval was obtained from the National Medical Ethics Committee of the Republic of Slovenia (No. 32/01/97).

Recording Protocol

EHG signals were collected using four Ag/AgCl electrodes placed symmetrically around the navel. Data were sampled at 20 Hz with a 16-bit resolution. Signals were filtered before recording to remove noise and aliasing effects.

Data Processing

The original signals were digitally filtered (0.08 Hz to 5.0 Hz band-pass Butterworth filter) to remove slow drifts. Filtered versions were included alongside the original signals in the dataset.

Data Description

The ICEHG DS contains:

126 EHG records from 91 pregnancies.
Records are categorized into early (23rd week) and later (31st week) datasets based on delivery mode (induced, cesarean, induced-cesarean).
Stored in PhysioNet's WFDB format and also available in MATLAB format.

Each record includes:

A figure showing raw and filtered signals.
A binary signal file (.dat).
A header file (.hea) with clinical and recording information.

Usage Notes

The ICEHG DS supports:

Studying physiological mechanisms leading to induction or cesarean delivery.
Prediction of delivery mode early in pregnancy.
Improved classification models for term and preterm birth prediction.

Limitations:

Unipolar electrode signals are not stored.
Filtered signals may have transient artifacts at the start and end (first and last 150 seconds).

Ethics

The collection was approved by the National Medical Ethics Committee of Slovenia. Participants provided written consent.

Acknowledgements

This research was funded by the Slovenian Research Agency (ARRS) under the project "Metabolic and inborn factors of reproductive health, birth III".

Conflicts of Interest

No conflict of interest was declared. The funder had no role in study design, data collection, analysis, or publication.

References

[References 1-11 available in original documentation.]