Detection of ECG beat using ensemble empirical mode decomposition

Abstract

The electrocardiogram (ECG) is the most common non-invasive test possessing a significant clinical value to cardiac disease diagnostic. In this paper, we developed a robust ECG delineation algorithm based on Ensemble Empirical Mode Decomposition (EEMD) which has very interesting properties on pseudo periodic signals. The proposed algorithm consists to identify an optimum and appropriate set of IMFs to reconstruct a complex or wave from ECG signal. First, QRS complexes are delineated by detecting the peaks of individual waves. Then the determination of P and T wave peaks was performed. We evaluated the reliability of our method on two manually annotated databases: QT database and MIT-BIH Arrythmia database. It proves particularly effective in cases of very noisy recordings or with abnormal morphology. A comparison is also made between the proposed method and other delineation system. The results showed a better or comparable performance and accurate fiducial points detection.