Mathematical modeling of clean and noisy ECG signals in a level-crossing sampling context

In this paper, a state-of-the-art review of electro cardiographic (ECG) signal modelling techniques is presented. The mathematical Gaussian modelling is selected as the best technique to generate realastic ECG signals regarding its ability to preserve relevant information of ECG databases signals. The generated Gaussian ECG signal model is applied to a level-crossing converter to measure the accuracy and the efficiency of sampling and to evaluate the effects of the signal noise on the sampling paradigm. The performances of the conversion were evaluated in terms of quality of the reconstructed signal. A percentage root mean square difference (PRD) around 2 % is obtained. It is shown that the LC-ADC is robust against respiratory and movement noise. However, for low signal distortions the electromyogram (EMG) noise should not exceed 20 μν which is less than IEC 60601-2-47 specifications.