Comparison of Wavelet Family Performances in ECG Signal Denoising

The heart is considered the most important organ of our body that controls the circulation of blood throughout the body. Measured heartbeat signals can be further analyzed in order to know the health condition of a person. The challenge of ECG signal measurement and analysis is how to remove the noises imposed on the signal that is interfered from many different sources, such as internal noise in sensor devices, power line interference, muscle activity, and body movements. This paper implemented wavelet transform to reduce the noise imposed on the ECG signal to get a closely actual heart signal. ECG data used in this research are three digitized recorded ECG data obtained from MIT-BIH Arrhythmia Database. The first step is generating the noisy ECG signal as the input system by adding 1W WGN signal into the original ECG signal. Then DWT is applied to extract the noisy ECG signal. Some DWT's parameters, threshold selection (rule, type, rescaling), decomposition level, and desired wavelet family are varied to get the best denoised output signal. All results are recorded to be compared. Based on the results, the best DWT parameter for ECG signal denoising is obtained by Symlet wavelet when the decomposition level is set to 3, with soft thresholding, in rigrsure thresholding rule.