Purpose: The electrocardiogram (ECG) segmentation step establishes the basis of the cardiac pathologies' classification. We propose here to detect the essential basic forms of this signal to optimize the cost calculation and ensure real-time application.

Methods: An automatic approach for R wave's location of the ECG signal based on the Entropy Criterion (EC) of the Wavelet Transform (WT) is introduced in this paper and implemented on MATLAB platform. The method uses automatic placement of analysis window and adjusting its width by measuring entropy for signal in three localized sub-windows. The WT is applied to the ECG signal at analysis window with a first derivative Gaussian wavelet. The R-wave corresponds to the zero crossing between the two maxima of the WT.

Results: The above-mentioned method (EC-WT) was tested with most noisy signals within QT databases and compared with two methods: the R wave detection method proposed by Martinez et al. and Zhang and Yong. EC-WT achieved good results attaining a sensitivity about 99.94% and a predictivity over 99.8%.

Conclusion: Through this method, the problem of adjustable thresholds for R wave detection is resolved as we apply a dynamic window that fits R peak's parameters. This ensures optimization and efficiency detection in terms of computational cost and complexity.