Effectiveness of Advanced ECG Noise Reduction Filters

The study examined various filtering methods applied to ECG signals; the MIT-BIH Arrhythmia dataset included two channels to reduce noise and artifacts while maintaining critical waveform characteristics. Initial filtering included a notch filter and a bandpass filter. Advanced filters such as Savitzky-Golay, Wavelet, Kalman, Wiener, and combined Wiener-Golay were used for noise reduction. Results showed that the Kalman filter performed best with an average SNR of 21.40 dB, while the Wavelet method worked well on channel 2. The Wiener filter was less effective due to its limited adaptability to changing noise. The combined Wiener-Savitzky-Golay filter showed a balanced performance but did not outperform the Kalman filter. Visual inspections of ECG signals from five patients revealed that the Kalman and Wavelet filters successfully maintained crucial morphological characteristics. The Savitzky-Golay filter offered effective smoothing but sometimes dampened the QRS complex, while the Wiener filter introduced high-frequency artifacts.