Introducing a combined approach of empirical mode decomposition and PCA methods for maternal and fetal ECG signal processing

Objectives: The purpose of this study was to evaluate the effectiveness of the non-linear adaptive combined approach on the analysis of maternal and fetal signals corrupted by noise. Methods: Empirical mode decomposition (EMD) is a non-linear adaptive technic for data analysis and has been widely used in biomedical data. When we apply EMD on ECG signals, the number of modes that contain cardiac information may vary regarding the subject, type of the signal and recording conditions. This fact can cause some difficulties in signal reconstruction and noise removal using the derived modes. For overcoming this issue, we designed a method to combine principal component analysis (PCA) method with EMD to remove the correlation between the calculated modes and provide a smaller set of uncorrelated orthogonal. Results: We have developed a combined method that proves the power of using PCA on the output of EMD method. The combined method reduces the power of oscillatory artifacts of the baseline. Thus, the PCA-EMD combination provides a noise-free signal. Conclusions: The combination of EMD and PCA methods worked well in being adaptive (from of EMD) and reconstruction (from PCA). It has been proved that this combined method is helpful in separating the signal components, especially in extracting the pure data from the baseline fluctuations.