| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 388985 | Expert Systems with Applications | 2008 | 8 Pages |
Time-frequency distributions have been widely utilized in the analysis of transients’ nature of biomedical signals. In these applications, the time-frequency components with very small amplitude values cannot be displayed clearly. This drawback results from a masking effect on these components by the presence of high-energy slow waves and sharp patterns in the signal that produces large values in the time-frequency distributions. In this paper, we proposed an effective signal preprocessing method-using kurtosis-based denoising of wavelet coefficients.This method enhances the time-frequency distributions so that the masking effect is greatly reduced, while the original time-frequency signatures of the input signal are preserved. Experimental studies on ECG signal coming from MIT-BIH database, with and without preprocessing methods have shown a clear improvement in observability and sensivity in the time-frequency distributions.
