Fetal ECG Synthetic Database

@article{pollard_physionet_2026, title = {{PhysioNet} as a global platform for biomedical research}, volume = {1}, issn = {3005-0693}, url = {https://doi.org/10.1038/s44360-026-00096-z}, doi = {10.1038/s44360-026-00096-z}, abstract = {Twenty-five years after its first public repository was established, the PhysioNet resource for biomedical and clinical data has enabled tens of thousands of research papers. Further efforts are ongoing to advance secure data sharing, community engagement and scalable research infrastructure.}, number = {8}, journal = {Nature Health}, author = {Pollard, Tom and Moody, Benjamin E. and Lehman, Li-wei H. and Gow, Brian J. and Fernandes, Chrystinne and Xie, Chen and Johnson, Alistair and Mark, Roger G. and Heldt, Thomas}, month = aug, year = {2026}, pages = {792--795},}

Abstract

The FECGSYNDB is a large database of simulated adult and non-invasive fetal ECG (NI-FECG) signals, which provides a robust resource that enables reproducible research in the field. The data is generated using the FECGSYN simulator (visit website).

Experiment/Simulation Details

The simulator represents maternal and foetal hearts as punctual dipoles with different magnitudes and spatial positions. This current version obtains foetal–maternal mixtures by treating each abdominal signal component (e.g. foetal/maternal ECG or noise signals) as an individual source, whose signal is propagated onto the observational points (‘electrodes’). Therefore this database is able to provide separate waveform files for each signal source.

The modeled volume conductor, sampled from 34 channels (32 abdominal and 2 maternal ECG reference channels), is shown in fig1.png. The figure shows side (a) and upper (b) views the of volume conductor. Positions of foetal (small sphere, blue) and maternal (larger sphere, red) hearts shown.

For this database, ten different pregnancies were simulated. For each simulated pregnant subject, seven different physiological events were considered and simulated as described below:

On top of this, there were five different levels of additive noise and 5 repetitions for each combination of settings.

Overall a total of 10 (simulated pregnants) x 7 (cases) x 5 (SNR levels) x 5 (repetitions) = 1750 synthetic signals were produced, totaling 145.8 hours of multichannel data with 1.1 million fetal peaks. Each simulation had a duration of 5 minutes, and was sampled at 250 Hz with a 16-bit resolution. For more technical details, see section 2.1 of Andreotti et al., 2016.

Files

Each subdirectory subXX/snrYYdb/ includes a set of data for simulated pregnancy XX and SNR level YY.

Inside each subdirectory are numerous files in WFDB format. The files names are in the form

subXX_snrYYdb_lZ_WW_VVVV

where:

The file extensions are in the standard WFDB MIT format: .hea for header files, and .dat for signal files. As shown in Fig. 1 above, channels 1–32 are the abdominal FECG channels and channels 33-34 are the maternal reference ECG channels. Header files also contain additional model parameters described in table 2 of Andreotti et al. The units are listed as 'nu' for no units since the signals are simulated, though their amplitudes do correspond to the 1–2mV amplitudes of normal MECG signals. Annotation files .qrs give machine generated QRS locations of their respective signals.

Contact

If you have any questions, suggestions or want to know more about FECGSYNDB, do not hesitate to contact us .

Cited by

177 publications cite this project. The most recent are listed below.

• A systematic review on AI-driven noninvasive fetal ECG processing and analysis methods
  Dragoş Daniel Ţarălungă, Bogdan Cristian Florea, Anusuyah Subbarao et al. Frontiers in Digital Health, 2026.
• Synthesis of multi-channel fetal ECG using generative modeling
  Swedha Sankaranarayanan, Marta Regis, Judith O. E. H. van Laar et al. Physiological Measurement, 2026.
• SVD-UNet: SVD-enhanced UNet for FECG extraction in the time–frequency domain
  Wei Zhong, Chenyang Gong, Ruiwen Li. Biomedical Signal Processing and Control, 2026.
• Transformer-based deep learning for estimating bidirectional maternal-fetal cardiac coupling
  Murad Almadani, Leontios J. Hadjileontiadis, Ahsan Habib Khandoker. Frontiers in Medical Technology, 2026.
• MCA-FM: Robust Non-Invasive Fetal ECG Extraction via Minimal Channel Attention and Flow Matching
  Qingqing Duan, Xinyu Hu, Yuwei Zhang et al. Applied Sciences, 2026.

View all 177 publications

Citation data from OpenAlex. Papers that use the project without citing it may be missing.