Motion artifact reduction of electrocardiograms using multiple motion sensors

Abstract

An electrocardiogram (ECG) is a measurement of the electrical signal produced by the heart as it beats. This is a signal very commonly used by medical professionals, as it gives an indication of an individual’s heart rate and can further be used to detect specific abnormalities within the heart. There are a number of sources of noise that can corrupt the ECG signal, the most problematic being that of motion artifacts. As an individual wearing a surface ECG moves, their movements will add noise to the signal. This noise is particularly difficult to remove, as it will change depending on the movements of the user and will often fall in the same spectrum as the ECG signal itself.

The effectiveness of the adaptive filtering method in reducing motion artifacts is investigated using multiple motion sensors on key locations of the body and by combining the motion data through the use of various blind source separation methods. An adaptive filter is a filter that can use a reference signal in order to readjust itself to a constantly changing noise signal and is commonly used to clean ECG signals. The adaptive filter uses noise estimations based on the reference signal as well as previous noise estimations in order to continually clean the noisy signal. Since motion artifacts are based directly off the movements of the user, collected motion data will be directly correlated with the noise being introduced to the ECG, and can therefore be used in the adaptive filter to produce a desirable ECG signal.