Denoising of ECG Signals Using Fuzzy Based Singular Spectrum Analysis

Automatic detection of various arrhythmia from an Electrocardiography signal(ECG) is a challenging task due to the corruption of ECG signal by various noises such as power line interference noise, high frequency noises, motion artifacts etc. An efficient de-noising technique should be used to remove the noise from ECG signal. Singular spectrum analysis (SSA) is a data adaptive and non parametric method which decomposes the signals into physically meaningful components like sinusoidal signals, noise etc. Effective embedding and grouping algorithms in SSA leads to better de-noising performance. In this paper a fuzzy C means based automatic SSA with inverse of correlation as measure of dissimilarity is proposed for de-noising. The proposed method results in better de-noising of the ECG signal compared to SSA with K-means and SSA with hierarchical clustering.