ECG Fragment Database for the Exploration of Dangerous Arrhythmia

Creators: Anatoly Nemirko, Liudmila Manilo, Anna Tatarinova, Boris Alekseev, Ekaterina Evdakova

Published: March 17, 2022. Version: 1.0.0

Citation

When using this resource, please cite:
Nemirko, A., Manilo, L., Tatarinova, A., Alekseev, B., & Evdakova, E. (2022). ECG Fragment Database for the Exploration of Dangerous Arrhythmia (version 1.0.0). PhysioNet. https://doi.org/10.13026/kpfg-xs25.

Additionally, please cite the original publication:
L. A. Manilo, A. P. Nemirko, E. G. Evdakova and A. A. Tatarinova, "ECG Database for Evaluating the Efficiency of Recognizing Dangerous Arrhythmias," 2021 IEEE Ural-Siberian Conference on Computational Technologies in Cognitive Science, Genomics and Biomedicine (CSGB), 2021, pp. 120-123, doi: 10.1109/CSGB53040.2021.9496029.

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 database contains a set of 2-second fragments of ECG signals with rhythm disturbances, grouped into separate classes according to the degree of threat to the patient's life. It is intended for practical use to develop and test the efficiency of algorithms for detecting dangerous arrhythmias for continuous monitoring systems.

The “MIT-BIH Malignant Ventricular Ectopy Database” (MVED) was selected as the primary source of ECG records. The signal is a modified limb lead II (MLII).

Background

Reliable detection of life-threatening arrhythmias is critical in monitoring cardiovascular diseases. In continuous ECG monitoring systems, instant alarms are essential for effective resuscitation. There is a need for a standardized database of short ECG fragments to objectively assess algorithm performance. This database offers 2-second ECG fragments classified by arrhythmia risk.

Arrhythmias were ranked based on risk using medical literature and specialist input, covering ventricular fibrillation, high-rate ventricular tachycardias, and others. 22 half-hour recordings from MVED were reviewed, and 6 classes of risk were defined.

Methods

ECG fragments (2s) were manually selected from MVED. Spectral features were computed using FFT. Fragments were classified and annotations corrected when necessary with cardiologist supervision. Spectral data was generated using Daniell's periodogram for different frequency resolutions. 4 files were generated per ECG fragment, corresponding to the raw signal and different spectral representations.

Data Description

Classes included:

• Class 1: Ventricular Flutter (VFL), Ventricular Fibrillation (VF)
• Class 2: Ventricular Tachycardia Torsades de Pointes (VTTdP)
• Class 3: High-Rate Ventricular Tachycardia (VTHR)
• Class 4: Low-Rate VT (VTLR), Bigeminy (B), High Degree Ectopy (HGEA), Ventricular Escape Rhythm (VER)
• Class 5: Supraventricular Arrhythmias (AFIB, SVTA, SBR, BI, NOD)
• Class 6: Sinus Rhythm (BBB, N, Ne)

Dataset Composition

Table 1. Composition:

Table 2. Distribution of Fragments by Recording:

File Naming and Structure

• Example: 418_C_VFL_277s_frag.txt
• frag: 2s signal, full: full spectrum, 15_2: 1Hz spaced spectrum, 10_3: 1.5Hz spaced spectrum.

Project folder organized into 6 main sections (1_Dangerous_VFL_VF, 2_Special_Form_VTTdP, etc.) each with 4 subsections (frag, full, 10_3, 15_2).

Usage Notes

This database allows training and testing algorithms for detecting ventricular fibrillation and ventricular tachycardias using either time or frequency domain features. Initial studies with kNN, LNCH, FLD, SVM classifiers reported accuracy up to 94.8% using cubic SVM.

Limitations

• Only 2-second ECG fragments (limits to online algorithms)
• No demographics (age, gender, weight, medications)
• No ECG signals for asystole or pacemaker cases

Ethics

The authors declare no ethics concerns.

Acknowledgements

Supported in part by the Russian Foundation for Basic Research, project 19-29-01009.

Conflicts of Interest

No conflicts of interest declared.

References

1. Acharya U. R., et al. (2018). Future Generation Computer Systems, 79, 952–959.
2. Bayers de Luna, et al. (1989). Am. Heart J. 117 (1): 151–159.
3. Lown B., Wolf M. (1971). Circulation, 44: 130-142.
4. Bigger J.T. Jr. (1984). Am. J. Cardiol. 54(9): 3D-8D.
5. Priori S.G., et al. (2015). European Heart Journal, 36(41): 2793-2867.
6. Glikson M., et al. (2021). European Heart Journal (2021): 1-94.
7. MIT-BIH Malignant Ventricular Ectopy Database. MIT PhysioNet.
8. Marple S. L. (1987). Digital Spectral Analysis with Applications. Prentice Hall.
9. Manilo L.A., Nemirko A.P. (2016). Pattern Recognition and Image Analysis, 26(4): 782–788.
10. Nemirko, A., et al. (2020). SCITEPRESS.
11. Manilo L.A., Nemirko A.P., Evdakova E.G. (2021). IEEE USBEREIT.
12. Manilo L.A., Nemirko А.P., Evdakova E.G., Tatarinova А.А. (2021). IEEE CSGB.