Performance analysis of adaptive algorithms for removal of low frequency noise from ECG signal

Electrocardiogram (ECG) signal is the electrical activity of the human heart. The ECG contains important information about the overall performance of the human cardiac system. Therefore, accurate examination of the ECG signal is very important but challenging task. ECG signal is often very low amplitude and contaminated with different types of noises due to its measurement process e.g. power line interference, amplifier noise and baseline wander. Baseline wander is a type of biological noise caused by the random movement of patient during ECG measurement and distorts the ST segment of the ECG waveform. In this paper, we present a comprehensive comparative study of five widely used adaptive filtering algorithms for the removal of low frequency noise. We perform extensive experiments on the Physionet MIT BIH ECG database and compare the signal to noise ratio (SNR), convergence rate, and time complexity of these algorithms. It is found that modified LMS has better performance than others in terms of SNR and convergence rate.