Application of wavelet on quasi-periodic physiologic signals

Electrocardiogram (ECG) signals signify the electrical activity of the heart. The scrutiny of these signals is extremely crucial as it ascertains the robustness of the heart and accordingly the health of the person. An ideal ECG signal is generated synthetically in MATLAB to interpret its morphology. This research describes an innovative method to classify the normality or abnormality of the ECG signals by considering interval estimation, specifically the P-R interval. Along with the P-R interval, an R-R interval, and a P-P interval is also considered. The intervals play a more significant role compared to the amplitudes of the waves in the ECG signals because the intervals remain invariant with respect to the ECG Recording Machine (ERM) while amplitudes can be manually altered. The characteristics of the ECG signals are analyzed with the help of a wavelet analysis. A novel algorithm is developed using a db4 wavelet to check the 50 Hz interference as well as eliminate the Wandering Baseline (WB) artifact and White Gaussian Noise (WGN) from the signal for an accurate estimation of the intervals. A tolerance of ±2 units is observed in the calculation of the intervals. The ECG signals 103, 105, 107, and 119 are chosen from the MIT-BIH Arrhythmia database, and the ECG signals 16420, 16272, 17052, and 17453 are chosen from the MIT-BIH Normal Sinus Rhythm Database. The developed algorithm demonstrates precise results with the simulations performed in MATLAB to conclude the normality or abnormality of the ECG signals