Evaluating the Noise in Electrically Evoked Compound Action Potential Measurements

Electrically Evoked Compound Action Potentials (ECAPs) are widely used to study the excitability of the auditory nerve (AN) and stimulation properties in cochlear implant (CI) users. However, ECAP detection can be difficult and very subjective at near- threshold stimulation levels or when stimulation parameters are modified (e.g. electrode position, rate and pulse width (PW)). In this study, we evaluated the statistical properties of the background noise (BN) and the post-average residual noise (RN) in the ECAP measurement in order to determine an objective detection criterion. The BN and the RN were estimated by a method currently used in auditory brainstem response (ABR) measurements. The potential benefit of using weighted (Bayesian) average was also examined. All estimations were performed from a set of approximately 400 ECAP measurements recorded from 5 human CI users of the CII or HiRes90K device (Advanced Bionics). Results demonstrate that the BN was normally distributed and the RN decreased accordingly. No additional benefit was observed by using weighted averaging. The noise power was not significantly different either at different stimulation intensities or across recording electrodes along the cochlea. The analysis of the statistical properties of the noise indicated that an signal-to- noise ratio (SNR) of 1.7 dB as a detection criteria corresponds to a false positive detection rate of 1% with the used measurement setup.status: publishe