A Method for Signal Quality Assessment during Ambulatory EMG Recording

Biopotential signal recording is often affected by noise during recording. There is a large body of literature on noise removal techniques for every class of biopotential signals. However, obtaining clean recording is preferable to cleaning a noisy recording. An instantaneous quality measure of the signal to noise ratio during recording would be invaluable in correcting a defective recording situation. Such a quality measure can be used to correct the situation immediately. Ambulatory EMG recording to study complex muscle movement is one such situation where the quality of recording can easily deteriorate when electrodes become loosened or wires break due to entanglement with the subject’s moving body. We propose a model based approach for EMG quality assessment in the presence of most common types of noise. The method has been tested on simulated and real EMG sequences in the presence of numerically synthesized as well as experimentally induced noise. The method is amenable to implementation on line using a low cost microcontroller which could be easily included as a small circuit attached to the preamplifier.