This paper presents a noise reduction method for electrocardiographic (ECG) signals with real-time application. The method was evaluated using digitized signals with resolutions of 14, 16, and 24 bits, durations ranging from 30 to 60 seconds, and a sampling frequency of 1000 Hz. Prior to its application, power line interference (220V/60Hz) is attenuated if present. The approach is based on signal decomposition via the Stationary Wavelet Transform (SWT), analyzing 10 detail levels and 10 approximation levels. It was identified that noise is primarily concentrated in the first four levels, occasionally extending to the fifth. To differentiate between useful signal and noise, histograms of the detail coefficients were analyzed, and key metrics such as kurtosis, relative wavelet energy (RWE), and standard deviation were computed to establish adaptive thresholds for effective noise removal. T...