Enhanced detection of envelope-following responses for objective fitting of cochlear-implant users.

Abstract

Objective: Electrically evoked auditory steady-state responses (EASSRs) are potential neural responses for objectively determining stimulation parameters of cochlear implants (CIs). Unfortunately, they are difficult to detect in electroencephalography (EEG) recordings due to the electrical stimulation artifacts of the CI. This study investigates a novel stimulation paradigm hypothesized to improve artifact removal efficacy via system identification (SI), and therefore to improve response detection and clinical applicability.

Approach: An amplitude-modulated (AM) CI stimulation pulse train with a step-wise increase in modulation frequency is created (referred to as SWEEP stimulation). Another stimulation is created by randomly shuffling modulation frequencies of the SWEEP stimulation (referred to as Shuffled- SWEEP stimulation). AM pulse trains with fixed modulation frequency (referred to as conventional AM stimulation), which elicit EASSRs, are also created for comparison. EEG data is collected from four CI users. A supra-threshold stimulation condition is used to investigate whether the SWEEP and Shuffled- SWEEP stimulation can elicit envelope-following responses (EFRs). A sub- threshold stimulation condition allows the collection of artifact-only EEG data, which is used to compare the SI accuracy on recordings from the SWEEP and the conventional AM stimulation.

Main results: In all CI users, neural responses, following the SWEEP, Shuffled-SWEEP, and conventional AM stimulation are detected after artifact removal with SI. The validation with artifact-only EEG data shows higher F1 scores when comparing recordings with SWEEP stimulation (F1 = 0.9) to recordings with conventional AM stimulation (F1 = 0.82).

Significance: Being able to accurately identify the response within one EEG recording enables the development of effective, online, objective fitting protocols. The increased neural response detection sensitivity with SWEEP stimulation reduces clinical recording time on average by a factor of 2.07. Detecting EFRs following complex stimulation paradigms offers a potential advancement in the systematic assessment of the temporal envelope processing in CI users.