A new combinatorial algorithm for QRS detection

QRS detection is the most important step for analyzing electrocardiogram (ECG) signals. In this paper, a new combinatory algorithm based on Pan-Tompkins´ method and wavelet transform is proposed to keep their strengths and avoid the weakpoints. It consists of two stages: preprocessing and decision. In the preprocessing stage, the wavelet transform is used to remove undesired details of the ECG signal and amplify QRS complexes. Then, the local energy of the signal is computed by using a Gaussian kernel. Finally, in the decision stage, similar to Pan-Tompkins´ method, an adaptive threshold selection algorithm is employed to detect QRS complexes. Experimental results on the MIT-BIH database demonstrated outstanding solution quality of the proposed algorithm compared to a number of QRS detection algorithms. Furthermore, interesting properties such as low computational burden and straightforward implementation nominate the proposed algorithm as an appropriate candidate for real-time QRS detection applications.