Beat-to-beat P and T wave delineation in ECG signals using a marginalized particle filter

The delineation of P and T waves is important for the interpretation of ECG signals. In this work, we propose a sequential Bayesian detection-estimation algorithm for simultaneous P and T wave detection, delineation, and waveform estimation on a beat-to-beat basis. Our method is based on a dynamic model which exploits the sequential nature of the ECG by introducing a random walk model to the waveforms. The core of the method is a marginalized particle filter that efficiently resolves the unknown parameters of the dynamic model. The proposed algorithm is evaluated on the annotated QT database and compared with state-of-the-art methods. Its on-line characteristic is ideally suited for real-time ECG monitoring and arrhythmia analysis.