Multitone Electrocochleography in Cochlear Implant Recipients: An Expert and Algorithm-Based Comparison
Related Experiment Video
Updated: Nov 12, 2026
Performing Intracochlear Electrocochleography During Cochlear Implantation
Published on: March 8, 2022
Hannes S Hefty1, Klaus Schuerch2, Martin Kompis1
1Department of Otorhinolaryngology, Head and Neck Surgery, Bern University Hospital.
Objectives:
This study compared postoperative pure-tone and multitone electrocochleography (ECochG) recordings in cochlear implant recipients regarding signal characteristics and detectability of frequency-specific responses. The reliability of expert visual assessment and a Fourier transform-based binning algorithm were evaluated for detecting ECochG responses and identifying spectral components in multitone recordings.
Study Design:
Prospective pilot cohort study.
Setting:
Tertiary referral hospital.
Patients:
Twenty-one cochlear implant recipients with stable residual hearing at least 6 months postimplantation.
Interventions:
Intracochlear cochlear microphonics potentials were recorded through the implant electrode array using pure tones at 250, 500, 1000, and 2000 Hz and 2 multitone stimuli comprising 3 or all 4 of the aforementioned frequencies. Measurements were conducted at least 6 months postimplantation, with follow-ups at 2 to 48 hours and 2 to 4 months thereafter.
Main Outcome Measures:
Response classification by the expert and the algorithm was compared. Inter-rater agreement, signal-to-noise ratio, sensitivity, and specificity were assessed.
Results:
Pure-tone ECochG responses were detected at all tested frequencies in every participant. Multitone stimulation produced an overall ECochG response in 17 of 21 participants. Response amplitudes and spatial patterns varied with electrode position and residual hearing. While tonotopic trends were observed, maximal responses were frequently shifted toward the basal cochlea. Inter-rater agreement was high for pure tones but clearly lower for frequency-specific components of multitone recordings, particularly at low frequencies. Signal-to-noise ratios were low across all conditions. The binning algorithm showed high sensitivity but limited specificity for pure tones and a frequency-dependent trade-off for multitone stimulation.
Conclusions:
In patients with stable audiometry and electrode position after at least 6 months of cochlear implant insertion, multitone ECochG was feasible and produced detectable responses. However, signal characteristics varied with electrode position and residual hearing, yielding more complex patterns than pure tones. This complexity hindered reliable identification of individual spectral components by experts and by objective methods.
