Motion artifact suppression in the ECG signal by successive modifications in frequency and time.

Ambulatory electrocardiogram signals can be contaminated with various types of noise. Among these, electrode motion 'em' artifacts are considered particularly undesired as they can be mistaken for ectopic beats. Unfortunately, 'em' noise has proved difficult to tackle using ordinary filtering techniques. In this paper, we explore a novel filtering alternative, and show that it could be considered as a potential candidate for dealing with electrode motion artifacts. The proposed system is composed of two simple parts: a frequency filter and a time window, interconnected in series. The two components are designed such that the overall system operates optimally in the mean square error sense. Experimentation on signals obtained from the MIT-BIH database demonstrates the superiority of the above approach over optimal Fourier filtering.