A bootstrap approach to the detection of auditory evoked potentials

Auditory Evoked potentials are often evaluated subjectively (by visual inspection) and considerabledifferences between interpretations can occur. Objective, automated methods are normally based oncalculating one or more parameters from the data, but only some of these techniques can provide statisticalsignificance (p-values) for the presence of a response. The bootstrap method can be used to provide suchp-values; it is based on randomly resampling (with replacement) the original data and gives an estimate ofthe probability that the response obtained is due to random variation in the data rather than a physiologicalresponse [1]. It can compare estimators for detecting responses and it can also estimate a criterion valuefor response detection corresponding to any chosen significance level, so avoiding the need for elaboratetheoretical calculations of criterion values. We have applied the bootstrap approach to detect hearingthresholds with both the Auditory Brainstem Response (ABR) [2] and the Slow Vertex Response (SVR). Forthe ABR data, time-domain detection algorithms including (amongst others) the Fsp [3] and ± difference [4]were compared. For the SVR data both time-domain and frequency-domain techniques (such asmagnitude-squared coherence [5]) were compared. This paper summarises the findings of these analyses.The flexibility of the approach is illustrated, showing how it can be used with different parameters, numbersof stimuli and with user-defined false-positive rates. The bootstrap method provides a new, simple and yetpowerful means of detecting evoked potentials, which is very flexible and readily adapted to a wide varietyof signal parameters.