Analysis and extraction characteristic parameters of ECG signal in real-time for intelligent classification of cardiac arrhythmias

In this paper various adaptive filters have been thoroughly applied to biomedical data processing in the aim to implement a barrier between noise reduction and preservation of the useful information. Electrocardiography (ECG) presents one of the most important indicators, which can be informed of the recognizing approaches to discover heart disease. Due to its inherent importance, it is interesting to develop new technique of prevention and processing medical information. The main goal is to extract necessary information on the state of the heart. The ECG signals are generally contaminated and infected by many parasites, which can be polluted, and in some cases make unrecognizable information. Therefore, an efficient process with good performance (accuracy, speed) is essential. In this papers we describe a comparative study between different applied adaptive filtering algorithms including Normalized Least Mean Square (LMS), Least Mean Square (NLMS), Recursive Least Square (RLS) and Kalman Filter (KF). The Percent Root-Mean-Squared Difference (PRD) and the Signal to Noise Ratio (SNR) are the two basic parameters that used to compare the performances of all algorithms.