Noise Reduction and Delineation of ECG Signal

This work studies the location of characteristic points of an ECG signal (fiduciary points) by using instantaneous phasor transform, which is a technique with low mathematical and computational complexity. Further, it also includes ECG noise attenuation approaches that do not change the ECG morphology. For this purpose, the discrete and stationary wavelet transforms together with thresholding are used. For QRS complex delineation as well as P-and T-wave detection, we apply an adaptive search that allows avoiding false detection, missing points, and locating the onset and offset of the waves. The effectiveness of the procedure is evaluated using QT signals of the MIT-BIH database. For the evaluation of the algorithm, we use 40 QT signals of the MIT-BIH, obtaining as a result a degree of sensitivity (Se) and prediction (P+) of 99:82% and 98:99% respectively. All the algorithms are implemented in Python.