Improving the ensemble average of visual evoked potentials. II. Simulations and experiments.

Ensemble averaging is generally used for the estimation of Evoked Potentials. This paper deals with the assessment of correction procedures for the time variability of the ensemble components, this time variability reduces the improvement of the signal-to-noise ratio (SNR) by averaging. Evoked potentials were estimated by ensemble averaging, synchronized to a periodic stimulus. It is assumed that VEP-instability is partly caused by time-variability of the evoked potentials. Two time-variate models were used, from which procedures were derived to correct the single VEP-responses prior to ensemble averaging. The models are: (1) variation in response delay (jitter), (2) variable compression/expansion of the time scale of the response (wow). The Spectral Phase Difference method was applied to estimate both the delay time jitter and the wow factor of single responses with respect to a template (conventional ensemble average). The effects of the devised correction on the average VEP waveform and on the SNR of the ensemble were investigated by using data from realistic simulations and from experiments (n = 23) with a number of healthy human volunteers (n = 17). Jitter- and wow-corrections were effective on simulations with time variability due to delay time jitter and time scale distortion (wow), respectively. Both wow- and jitter correction of the single responses improved the SNR of the VEP measurements significantly and to the same amount. A combined wow-jitter approach resulted in significantly better results than the exclusive application of jitter- or wow correction.