Two Simulation Methods of J Wave Signal in ECG

Abstract

The appearance of J waves in electrocardiogram (ECG) may cause malignant arrhythmia, ventricular tachycardia and other diseases, and even bring the risk of sudden death. However, the data on J waves in the current public databases are sparse, and the hospital lacks relevant data due to the urgency of pathological J waves. To address the lack of J wave data, two methods for simulating J wave signal are proposed in this paper. First, by studying the parameter information of the ECG signal, the J wave signal is generated by the superposition of sine wave, triangle wave and sawtooth wave; Second, the inflection point position of the normal ECG signal in the public database is determined by the local transformation method, and the J waves are added to form the J wave signal. The simulation results show that both methods can realize the free adjustment of the occurrence time, duration and amplitude of J waves, which can simulate various common J wave signals. The simulated J wave signals are trained by the UNet++ model and the feasibility of the simulation method is verified, which is helpful to promote the detection of J wave by portable devices.