Electrically evoked compound action potential artifact rejection by independent component analysis: Technique validation

Idrick Akhoun, Colette M. McKay, Wael El-Deredy

    Research output: Contribution to journalArticlepeer-review

    227 Downloads (Pure)

    Abstract

    The electrically-evoked compound action potential (ECAP) is the synchronous whole auditory nerve activity in response to an electrical stimulus, and can be recorded in situ on cochlear implant (CI) electrodes. A novel procedure (ECAP-ICA) to isolate the ECAP from the stimulation artifact, based on independent component analysis (ICA), is described here. ECAPs with artifact (raw-ECAPs) were sequentially recorded for the same stimulus on 9 different intracochlear recording electrodes. The raw-ECAPs were fed to ICA, which separated them into independent sources. Restricting the ICA projection to 4 independent components did not induce under-fitting and was found to explain most of the raw-data variance. The sources were identified and only the source corresponding to the neural response was retained for artifact-free ECAP reconstruction. The validity of the ECAP-ICA procedure was supported as follows: N1 and P1 peaks occurred at usual latencies; and ECAP-ICA and artifact amplitude-growth functions (AGFs) had different slopes. Concatenation of raw-ECAPs from multiple stimulus currents, including some below the ECAP-ICA threshold, improved the source separation process. The main advantage of ECAP-ICA is that use of maskers or alternating polarity stimulation are not needed. © 2013 The Authors.
    Original languageEnglish
    Pages (from-to)60-73
    Number of pages13
    JournalHearing Research
    Volume302
    DOIs
    Publication statusPublished - Aug 2013

    Keywords

    • AN
    • CI
    • CI24RE
    • ECAP
    • ECAP-FM
    • ECAP-ICA
    • IC
    • ICA
    • JADE-R
    • MP1
    • MP2
    • N1P1
    • RMS
    • SNR

    Fingerprint

    Dive into the research topics of 'Electrically evoked compound action potential artifact rejection by independent component analysis: Technique validation'. Together they form a unique fingerprint.

    Cite this