Toward Automated Cochlear Implant Fitting Procedures Based on Event-Related Potentials.

Area of Science:

• Neuroscience
• Audiology
• Biomedical Engineering

Background:

• Cochlear implants (CIs) restore hearing via electrical nerve stimulation, requiring precise fitting for optimal function.
• Current CI fitting relies heavily on subjective user feedback, posing challenges for non-verbal users like young children.
• Objective measures are needed to assess auditory perception in CI users, particularly for fitting and closed-loop systems.

Purpose of the Study:

• To investigate the feasibility of using electroencephalogram (EEG) to objectively determine auditory perception in cochlear implant (CI) users.
• To evaluate the efficacy of EEG-based classification for differentiating auditory stimuli in CI users.
• To explore the application of EEG in objective CI fitting and closed-loop auditory systems.

Main Methods:

• An auditory oddball paradigm was employed with 12 CI users, presenting standard and deviant auditory stimuli.
• EEG data were recorded and analyzed using shrinkage linear discriminant analysis for classification.
• The study assessed CI artifact removal techniques and the reusability of trained classifiers across sessions.

Main Results:

• EEG classification performance was significantly above chance level for all participants, though individual variability was noted.
• Effective removal of CI artifacts from EEG data did not compromise classification accuracy.
• Reusing trained classifiers across sessions resulted in only a minor decrease in classification performance.

Conclusions:

• This study provides initial evidence for the successful single-trial classification of EEG data in CI users.
• EEG artifact correction and classifier stability were demonstrated across multiple sessions, simplifying application.
• The findings suggest that EEG can objectively assess auditory perception in CI users, with potential to automate and refine CI fitting procedures.