| Popis: |
Lung sounds (LS) are important for diagnosis of the pulmonary system but interference signals such as heart sounds (HS) complicate their computer analysis. Sub-band decompositions have provided promising schemes for filtering HS; however, the mother wavelet and the number of level of decomposition have to be selected. On the contrary, the Empirical Mode Descomposition (EMD) offers an adaptive, local, complete, and orthogonal decomposition for dealing with nonstationary and nonlinear signals. In this work a combination of EMD, linear adaptive filtering (AF) and time warping (TW) procedure is proposed for filtering HS. The proposed scheme was applied to simulated and real signals. The scheme reduced significantly the power of HS in both simulated and real signals. Therefore, we concluded that the combination of EMD and AF is a promising scheme to deal with heart sound interference in lung sounds analysis. |
