Waveform variance and latency jitter of the visual evoked potential in childhood.

PURPOSE: Recording the visual evoked potential (VEP) in young children is challenging due to movement artifacts with variable fixation or attention. This study examined the effects of latency jitter, noise, and waveform consistency on the averaging of the VEP across childhood age.
METHODS: Stimuli were contrast-reversing (1.4 Hz) checkerboards of 163 arc minutes and pattern-onset-offset of 0.5 cycle/degree horizontal sine-wave gratings. Subjects were 79 normal children (0.3-16 years age; mean 6.9). Results were compared to recordings of EEG noise only (noise controls). Epochs underwent four averaging methods: (1) latency jitter correction using cross-correlation, (2) correction of phase shifts across a limited bandwidth in the Fourier domain, (3) selection of epochs based on consistency in the time domain, and (4) selection of epochs based on phase consistency in the Fourier domain. Signal-to-noise ratios (SNR) were estimated in both the time and Fourier domains.
RESULTS: Compared to standard averaging, all methods improved the amplitude of the primary peak (P100) while generating mild changes in latency. All methods also increased amplitudes of residual peaks in noise controls. In VEPs with an adequate SNR, selective averaging in the Fourier domain provided the greatest improvement in amplitude (61 % increase; p < 0.0001) without prolongation in latency. Correction of latency jitter did not consistently improve amplitude but caused latency prolongation in 24 % of subjects. There was no age-related effect of any averaging method for either stimulus.
CONCLUSIONS: Since latency jitter correction does not improve VEP amplitude more than selective averaging, recording artifacts in children are dominated by random phase components rather than inducing latency jitter.