Impact of Wavelet Denoising on the Signal Processing of Muscle Sympathetic Nerve Activity (MSNA)

Objective This paper discusses the characteristics of a wavelet denoising method in terms of its efficiency and usefulness in processing of MSNA signals. Methods To address the issue of whether wavelet denoising could make a poor SNR signal into one of better quality, two multiunit recordings of varying signal‐to‐noise (SNR) were obtained from each of two individuals, during the same recording session, by slight adjustments in the microelectrode position within the peroneal nerve recording site. The subjects were supine and resting when the recordings were made. The wavelet denoising was conducted on the wavelet toolbox in Matlab 6.5. Results When the original raw MSNA signals from two subjects have very low SNR, the wavelet based denoising approach was not able to separate the nerve activity and noise despite our attempts to use a variety of mother wavelets and coefficient thresholds. However, the denoised MSNA from the high SNR signals provided good results with reductions in noise residuals as identified using Gaussian distribution analysis. Conclusions The ability to denoise MSNA signals depends on the quality of the original signal. Also, the choice of the different mother wavelets can cause different denoised results and an inappropriate choice of noise threshold can also impair the denoised result. Despite these limitations, careful application of this technique to strong SNR recordings may enhance signal detection in this noisy but important biopotential signal. Supported by NSERC‐CHRP.