| Abstrakt: |
An improved and versatile filtering approach has become essential for diagnosing cardiovascular diseases. The most crucial noises that can be arises in health monitoring devices are additive white Gaussian noise, baseline wander, power line interference, and motion artifact. The wavelet-based filtering approach is an effective method for electrocardiogram (ECG) signals denoising among the different filtering techniques. However, there is room for improvement of the wavelet-based filtering method in a low signal-to-noise ratio (SNR) environment. In this paper, an orthogonal wavelet-based filtering method is developed by using an optimized filter bank to improve the denoising performance of the ECG signals. The coefficients of the proposed wavelet filter are optimized using the nature-inspired population-based cuckoo search optimization algorithm. The effectiveness of the presented filtering method is experimentally verified using the PhysioBank database. The proposed method is also compared with the empirical wavelet transform, empirical mode decomposition, and Butterworth high pass filter methods. The comparison result indicates satisfactory performance for the removal of various noises present in the ECG signal even at low input SNR. |
Full text from SpringerLink