from PhysioNet, the research resource for complex physiologic signals


ECGSYN: A realistic ECG waveform generator

[Typical trajectory generated by the model]

ECGSYN generates a synthesized ECG signal with user-settable mean heart rate, number of beats, sampling frequency, waveform morphology (P, Q, R, S, and T timing, amplitude,and duration), standard deviation of the RR interval, and LF/HF ratio (a measure of the relative contributions of the low and high frequency components of the RR time series to total heart rate variability). Using a model based on three coupled ordinary differential equations, ECGSYN reproduces many of the features of the human ECG, including beat-to-beat variation in morphology and timing, respiratory sinus arrhythmia, QT dependence on heart rate, and R-peak amplitude modulation. The output of ECGSYN may be employed to assess biomedical signal processing techniques which are used to compute clinical statistics from the ECG.

ECGSYN was contributed to PhysioNet by Patrick McSharry from the Department of Engineering Science, University of Oxford, and by Gari Clifford of the Laboratory for Computational Physiology at MIT. Three implementations are available:

All three forms of the code are freely available under the GNU GPL (General Public License), and can be downloaded as a single gzip-compressed tar archive, ecgsyn.tar.gz.

The algorithms used by ECGSYN are described in

McSharry PE, Clifford GD, Tarassenko L, Smith L. A dynamical model for generating synthetic electrocardiogram signals. IEEE Transactions on Biomedical Engineering 50(3): 289-294; March 2003.

Current implementations of ECGSYN allow the user to modify the morphology of the P-QRS-T cycle, which was not a feature of the original ECGSYN described in the paper. The angle of each attractor (P, Q, R, S and T) around the limit cycle is set by ti (initially, [-70 -15 0 15 100]*pi/180). Their positions above or below the z=0 plane are set by bi and the widths of the waveform components are given by ai. Since ti=0 defines the placement of the R-peak, the ordering of each element of ti, ai and bi is [P Q R S T]. The bi and the ti are stretched by the square root of the reciprocal mean RR interval, as suggested by Bazett's (empirical) formula relating the QT interval to the heart rate. This transformation does not cancel out the reduction of the inter-attractor angular distance that arises 'naturally' from augmented heart rates in this model.

Of related interest is ECGwaveGen, a Matlab/Octave application that generates (non-realistic) ECG-like test waveforms with well-defined characteristics as specified in ANSI/AAMI EC13:1992 (American National Standard: Cardiac Monitors, Heart Rate Meters, and Alarms).