Efficient ECG Approximation Using Chebyshev Polynomials

An ECG (Electrocardiogram) signal records electrical activity of heart and is always been chosen for diagnosing the cardiac abnormalities of patients. The morphology of the signal plays very important role in analysis and diagnosis. Noises introduced in the signal changes its morphology and thus prevent accurate analysis, diagnosis and hence need to be reduced. Also massive volume of ECG data is required to gather enough information about the patients. Denoising and compression of ECG signals are the solutions to above problems. In this paper, an efficient model is designed to compress ECG signals of MIT-BIR. ECG signals are first denoised using majorization-minorization optimization technique. The denoised signals are then appropriately divided into segments by well known time series Bottom-Up algorithm. The individual segments thus obtained are interpolated using Chebhshev nodes of suitable degree. The proposed model is evaluated in terms of Maximum Error (MaxError), Root Mean Square Difference (RMSD), Percentage Root Mean Difference (PRD) and Normalized Percentage Root Mean Difference (PRDN). Performance of the proposed model is compared and is found to be better than other existing approximation models.