Development of a segmentation algorithm for ECG signals, simultaneously applying continuous and discrete wavelet transform

This paper presents the development of a segmentation algorithm for ECG signals, using simultaneously discrete and continuous wavelet transform. Here, it has been proposed the use of the discrete transform in order to make the pre-processing, signal filtering and detection of QRS complex; also the continuous transform to correct R and S wave detection as well as for the T wave, enabling the characteristics components identification in ECG signal. The algorithm was implemented in MATLAB® software and the QT Database was used for validation, considering the detection and delineation in signals with small or negative polarity QRS complex; and inverted T waves or with depression or elevation in the ST segment. For QRS complex it was found a sensibility Se=99,8% and a positive predictivity of P+=99,8%; and for the T wave a sensibility of Se=97,6 and a positive predictivity of P+=97,4%. The results show that the algorithm can be applied in signals with different morphologies even with low Signal to noise ratio and baseline.