A new method for enhancement of ECG signals using cumulant based AR modeling

Electrocardiogram (ECG) signal gets corrupted with artifacts, such as 50/60 Hz power line interference (PLI), electromyogram (EMG) and baseline wander, making it difficult to diagnose the cardiac abnormalities. This paper presents autoregressive (AR) modelling of cumulants for enhancement of ECG signals. Higher order spectral (HOS) cumulants possess many properties that make it an effective tool for the analysis of nonlinear and non stationary signals like ECG. Novelty of the proposed algorithm is that it combines the higher order spectral cumulant properties in conjunction with AR modelling. Results on the records of MIT-BIH data base revealed that the proposed algorithm is effective in reducing the noise and enhancing the ECG data with low root mean square error (RMSE), root mean square variance (RMSV), root mean square deviation (RMSD) and improved signal to Noise Ratio (SNRI).