A modified Bayesian filtering framework for ECG beat segmentation

In this paper, we have presented a modified EKF structure based on the previously introduced signal decomposition based ECG Dynamic Model (EDM) for ECG beat segmentation. The new EKF can simultaneously estimate each of the ECG components including P, Q, R, S and T waveforms as well as the ECG signal. In this framework, instantaneous Gaussian functions of the P, Q, R, S and T components are considered as hidden state variables that are distinctly estimated from sample to sample. The result have shown that each of the CWs have been accurately estimated from multiple ECG beats. The proposed EDM can also be useful for synthetic ECG generation and ECG denoising applications.