An open-source applied simulation framework for performance evaluation of QRS complex detectors

Abstract: In this study, a mathematical model is described and developed based on a set of algebraic equations which is capable of artificially generating normal events of the electrocardiogram (ECG) such as P-wave, QRS complex, T-wave or synthetic abnormal phenomena like ST-segment episodes (i.e. depression, elevation, and sloped ascending or descending) and repolarization abnormalities such as T-wave alternans (TWA). In order to demonstrate the flexibility of the presented model for the simulation of various ECG signals, fascicular ventricular tachycardia, rate dependent conduction block, and acute Q-wave infarctions of inferior and anterior-lateral walls are simulated. Then, using the proposed artificial ECG signal, two new developed QRS detection algorithms (Modified Hilbert transform-MHT and discrete wavelet curve length-DWCL algorithms) evolved, respectively based on the implementation of some modifications to the conventional Hilbert and discrete wavelet transforms of the ECG signal, are then evaluated. To meet this end, a mathematically-simulated ECG signal with a weak R-wave is embedded into a Gaussian noise with a wide range for variances (powers) and the MHT and DWCL algorithms are applied to the resulted signal to detect QRS complexes. In the next step, an artificial baseline wander is added to the ECG signal and the performance of the MHT and DWCL algorithms in cases of high baseline wander are studied. After conducting numerous simulations, it was concluded that for the values of signal to noise ratio (SNR) larger than −5dB and amplitudes of baseline wanders approximately less than 5 times of R-wave amplitude, the two detection algorithms will have acceptable performance while it is seen that in low (negative) SNR values, performance endurance of the DWCL based detection algorithm is higher than MHT method. The Matlab m-file script of the synthetic ECG generator will be freely available from the Appendix of this paper. [Copyright &y& Elsevier]