ECG noise reduction using empirical mode decomposition based on combination of instantaneous half period and soft-thresholding

The electrocardiogram (ECG) signal is widely used for diagnosis of various types of cardiac diseases. However, in practical cases, the signal is corrupted by artifacts through the recording process. Thus, denoising of this type of biological signals seems necessary. Several methods have been suggested in recent years for the purpose of ECG denoising; some of which have been based on Empirical Mode Decomposition (EMD). In this paper, an EMD-based approach is proposed which uses the time interval between two adjacent zero crossings within an Intrinsic Mode Function (IMF), defined as Instantaneous Half Period (IHP), to distinguish noise components from the main ECG signal. Noisy signal is decomposed into several IMFs through an iterative algorithm, called the Sifting process. IMFs are later tested by a predefined threshold; and waveforms with a comparatively small IHP are filtered using soft-thresholding technique. The method is fully data-driven, with the optimum threshold derived from a criterion called Consecutive Mean Square Error (CMSE). The accuracy of the developed EMD-based approach is tested on real ECG data from the MIT-BIH database and compared with one of the previously suggested EMD-based methods. Both quantitative and qualitative results of the work are presented.