Denoising improves visualization of evoked potentials with limited number of trials

Ensemble averaging is typically used to enhance the signal to noise ratio (SNR) of event related potentials (ERPs). This approach has the disadvantage of extending the recording time. We present an alternative approach to achieve SNRs similar to the ones obtained with ensemble averaging but with fewer trials. Single trial ERPs were denoised by using a wavelet denoising technique. With simulated ERPs we compared the performance obtained with the standard and denoised averages by quantifying the root-mean-square-error (RMSE) between ensemble averages obtained with and without denoising. Denoising decreased the RMSE, giving similar values to the ones obtained with the standard averages but with less than half the number of trials. RMSEs were comparable for larger number of trials, thus showing that denoising did not introduce spurious distortions to the ERPs. Wavelet denoising proved to be an efficient technique to improve evoked potential SNRs. It is useful in clinical settings, where recordings may be constrained by the limited capability to obtain a sufficient number of trials.