Brain stem auditory evoked potentials: the use of noise estimate.

One of the main difficulties with the interpretation of evoked potential data is the question of how much noise is contained within the average. This question is crucial in deciding whether any particular peak component is a genuine signal, or whether it is a random occurrence due to noise. This is particularly crucial in abnormal cases, and with stimuli near sensory threshold. A simple method is described which allows on-line objective statistical determination of the presence of significant non-random peak components. It is shown that high signal-to-noise (S/N) estimates are seen with 'clean' and reproducible BAEPs and low S/N estimates are seen with poorly reproducible BAEPs. This provides an objective measure to judge the acceptability of any single BAEP, or to indicate that further averaging is required. The effect of noise on wave shape and peak latency is discussed. This can be quite marked, with gross distrotion of morphology and latency as a result. The technique lends itself to evoked response audiometry (ERA) and sensory threshold determination. Further, it can be applied to evoked potentials from other stimulus modalities.