Analysis of cochlear implant artifact removal techniques using the continuous wavelet transform

When patients with cochlear implants (CI) undergo cortical auditory evoked potential (CAEP) tests to evaluate their hearing, a large electrical artifact introduced by the CI obscures the relevant information in the signal. Several methods have been developed for the purpose of removing the CI artifact; however, there is no gold standard (i.e., patient´s auditory response before the CI) to assess the effectiveness of these methods in terms of successful removal of artifact. To address this crucial shortcoming, we employ time-frequency (TF) signal representation (i.e., continuous wavelet transform (CWT)) to evaluate the effectiveness of two recent CI removal techniques, known as the subtraction and polynomial methods. Our results show that polynomial method consistently outperforms the subtraction method in the presence of tone stimulus. These results also indicate a possible CWT-based method for removing the CI artifact from a speech stimuli response, which the subtraction and polynomial methods cannot do.