This is a useful aid in interpreting the quality of evoked responses that are seen after gathering and then averaging together a series of epochs in order to cancel the noise. Examples include ABRs, long latency responses, ECochGs and OAEs.
The recorded EEG signal (or acoustical signal, in the case of OAEs) consists of the response of interest (this is time locked to the stimulus) and the randomised noise. Averaging enough epochs will cause the noise to cancel, leaving the signal of interest.
A key question is how reliable is the averaged data. Does it accurately reflect the underlying evoked response of interest, or is it too contaminated with noise to be a reliable reflection? Of course, the more reliable it is, the greater confidence one has with its interpretation and therefore clinical decision making such as when estimating threshold or making other diagnoses.
The wave reproducibly measure helps establish the reliability of the measurement. Individual epochs are allocated alternately into two separate memory banks; the A and B buffer. A mathematical calculation known as cross-correlation between the responses in each buffer is then performed, and this provides a measure of the similarity of the two sub-averages; A and B. If a response is reliably present, and the residual noise is sufficiently low, the similarity between A + B data i.e. the wave reproducibility will be high.
If there is no signal present, and/or the residual noise is not sufficiently low, then the wave reproducibility will be low.
Please note that in this sense, when the waveform reproducibility is high it helps increase tester confidence in interpreting a response as being present. However, when the waveform reproducibility is low then this information alone is not enough to differentiate between an absent response and an inconclusive result whereby the residual noise is too high to make a firm conclusion on presence or absence.
Another feature must be used in conjunction with wave reproducibility to differentiate between an absent evoked response or inconclusive findings. Checking the differential (A – B) curve is one method. This provides an indication of the residual noise whereby a flat curve indicates low residual noise. Simply “eyeballing” the difference between A + B traces provides a similar, albeit subjective, indication.
More precise indications of residual noise are available when performing ABR measures where a precise value of residual noise is obtained.
For more information please refer to the Interacoustics Eclipse user manual and Additional Materials handbook.