Place-based Cochlear Implant Mapping

Academic Medical Organization of Southwestern Ontario, University of North Carolina, Lawson Health Research Institute

The current standard of care approach for programming cochlear implants uses a generalized pitch-map for all patients. This approach fails to account for individualized inner ear anatomy. As a result, many cochlear implant recipients experience place-pitch mismatch. We have recently developed an automated mathematical tool to produce patient-specific, customized cochlear implant pitch-maps (Helpard et al., 2021). In this study, cochlear implant recipients will be randomized to receive either the clinical default pitch-map (the control group) or a place-based pitch-map (the intervention group). Assessments will be conducted at multiple time-intervals to account for patient acclimation and plasticity to both the generalized and individualized pitch-maps. Audiological assessments will be tuned to identify patients' ability to discern pitch scaling and variation in sounds, as well as to understand complexities in speech such as mood and tone. Audiological testing will be conducted in collaboration with the National Centre for Audiology (London, ON) to ensure that the most accurate and relevant metrics are applied.

The cochlea is a spiral-shaped organ of hearing within the inner ear where acoustic vibrations are decomposed into different frequencies to create electrical signals that transmit audio information to the brain. The basilar membrane (BM), which is an internal soft tissue component of the cochlea, mechanically filters different frequencies at different distances along the helical shape. This separation is what allows us to discern different pitches in sound. Due to individual anatomical differences, each person naturally has their own unique pitch-map, or tonotopic map, that maps nerves at specific locations along the basilar membrane to perceived frequencies in the brain. When the cochlea is not functioning properly, cochlear implantation is a successful treatment to restore the sense of sound. A cochlear implant (CI) is a neural-prosthetic device that consists of an external portion that sits behind the ear and a surgically implanted array of electrodes inserted along the cochlea. After surgery, implants are programmed using a process called pitch mapping, whereby each implanted electrode is assigned a specific stimulation frequency. A CI must span the entire length of the cochlea and stimulate with the correct pitch-map (meaning the correct nerves and locations are stimulated with the correct frequencies) to produce full and accurate hearing. When a generalized pitch-mapping approach is used, each electrode within a CI array will stimulate with a pre-specified frequency, independent of a patient's individual tonotopy or postoperative electrode location. Generalized pitch-mapping can result in a place-pitch mismatch of over one octave. This mismatch inhibits the pitch perception required for complex hearing tasks, such as music appreciation or speech recognition. Neural plasticity can allow auditory perception to adapt over time to reduce the effect of cochlear implant pitch-map errors, however this requires long periods of acclimation, is dependent on recipient age and environment, and can only overcome certain sized pitch-map errors. Customization of CI pitch-maps can reduce rehabilitation time and the need for implant acclimation. Patient-specific pitch maps are produced by accurately determining each patient's cochlear duct length (CDL), or more specifically BM length, from diagnostic images. Previous methods to determine CDL have traditionally contained uncertainties at the start- and end-point of the BM, largely due to visualization limitations in the imaging modality used. Measuring an inaccurate BM length may cause an erroneous shift in all tonotopic frequencies. Using an enhanced imaging technique, our team has recently developed an algorithm to automatically and accurately estimate CDL, segment the BM, and determine CI electrode locations from individual patient computed-tomography (CT) scans to produce customized CI pitch-maps, called placed-based mapping (Helpard et al., 2021). The primary objective of this study is to evaluate whether a place-based map improves hearing outcomes for cochlear implant recipients. We will compare the auditory abilities, speech recognition and spatial hearing (speech recognition in spatially separated noise, and sound source localization) for subjects randomized to listen exclusively with a default map versus our novel place-based map. We hypothesize that the majority of CI recipients will experience a faster rate of speech recognition and spatial hearing growth when their cochlear implant is mapped to match the electric stimulation with the tonotopic place frequency (i.e., using the place-based map).

Participants will be randomized into either the intervention group or the control group. In the intervention group, cochlear implants will be programmed according to a custom, place-based pitch map. In the control group, cochlear implants will be programmed according to the standard of care, generalized pitch map. Participants will undergo repeated audiological and music and speech perception testing at baseline, initial activation (~ 6 months post-operative), 1-month, 3-months, 6-months, 7-months, and 1-year post-activation. Each participant will listen to their assigned map during the first 6 months of cochlear implant use. After 6 months, participants will listen to the alternative map for 1 additional month. After the 7th month, participants will have the opportunity to choose their preferred map. Final assessment will take place at 1-year post-implantation. Performance will be compared between the groups and as a function of frequency-to-place mismatch.

CIs will be mapped according to each participant's unique place-based map using MED-EL MAESTRO fitting software. Participants' audiological performance will be assessed in both the experimental and default programs at activation and at 6 months. At 1-month and 3-month visits, performance will be assessed in the participant's assigned program only. Following the 6-month testing interval, participants will listen to the alternative map than the one they were assigned for 1 additional month. At the 7-month visit, participants will be assessed in both programs. After the 7-month visit, participants will be given the choice to move forward with whichever map they prefer. The final assessment interval will take place at 1-year in the participant's chosen map.

CIs will be mapped according to the clinical default program using MED-EL MAESTRO fitting software. Participants' audiological performance will be assessed in both the experimental and default programs at activation and at 6 months. At 1-month and 3-month visits, performance will be assessed in the participant's assigned program only. Following the 6-month testing interval, participants will listen to the alternative map for 1 additional month from the one they were randomly assigned. At the 7-month visit, participants will be assessed in both programs. After the 7-month visit, participants will be given the choice to move forward with whichever map they prefer. The final assessment interval will take place at 1-year in the participant's chosen map.

Pre-operative, 3D CT scans of the temporal bone will be uploaded into a deep learning-based tool which automatically resamples, crops, segments, analyzes, and measures the patient's specific cochlear anatomy. These measurements will be input into an individualized pitch mapping function to determine a patient-specific tonotopic distribution of frequencies (place-based map). At device activation (approximately 1 month post-surgery) CIs will be programmed according to the place-based map.

At device activation (approximately 1 month post-surgery) CIs will be programmed according to the clinical default program.

Change in score on the word and vowel recognition test (Consonant-Nucleus-Consonant (CNC) words/ phonemes)

The CNC word test (Peterson & Lehiste, 1962) consists of 10 lists of 50 monosyllabic (single syllable) words with equal phonemic distribution across lists. Materials will be presented in quiet at 60 decibels (dB) sound pressure level (SPL) in the sound field. Outcomes will be reported as percent correct (%).

Baseline (pre-surgery), device activation (~ 1 month post-surgery), 1 month post-activation, 3 months post-activation, 6 months post-activation, 7 months post-activation, 1 year post-activation.

The AzBio sentence test (Spahr & Dorman, 2012) consists of sentences between 3 and 12 words in length recorded from 2 female and 2 male talkers. Materials will be presented in noise at 60 decibels (dB) SPL with signal-to-noise (SNR) beginning at +10 dB SNR and increasing in difficulty by 5 decibels (dB) increments continuing until a score of 20% or less is achieved. Outcomes will be reported as percent correct (%).

Baseline (pre-surgery), device activation (~ 1 month post-surgery), 1 month post-activation, 3 months post-activation, 6 months post-activation, 7 months post-activation, 1 year post-activation.

The SSQ (Gatehouse & Noble, 2004) is a self-reported questionnaire designed to measure a range of hearing disabilities across several domains. Questions are divided into three sections: speech hearing, spatial hearing, and qualities of hearing. Particular attention is given to hearing speech in a variety of competing contexts, and to the directional, distance and movement components of spatial hearing. In addition, the abilities both to segregate sounds and to attend to simultaneous speech streams are assessed, reflecting the reality of hearing in the everyday world. Qualities of hearing experience include ease of listening, and the naturalness, clarity and identifiability of different speakers, different musical pieces and instruments, and different everyday sounds.

Baseline (pre-surgery), 1 month post-activation, 3 months post-activation, 6 months post-activation, 7 months post-activation, 1 year post-activation.

Sound quality of speech samples will be self-reported according to several dimensions of sound quality (overall impression, loudness, fullness, sharpness and intelligibility) on a scale from 0 (lowest quality) to 10 (highest quality) while seated at a computer. Sound quality dimensions are adapted from Gabrielsson & Kan Sjogren, 1988.

Baseline (pre-surgery), 1 month post-activation, 3 months post-activation, 6 months post-activation, 7 months post-activation.

Change in score on the Multi Stimulus test with Hidden Reference and Anchor (CI-MUSHRA) adapted for cochlear implants

The CI-MUSHRA) adapted for cochlear implants (Roy et al., 2012) is a self-reported measure of sound quality in which participants listen to a set of recorded music samples and rate the sound quality on a sliding scale from "0" (very poor) to "100" (excellent). Samples are presented in an unaltered version (i.e., "reference") and in a highly degraded version (i.e., "anchor"). Participants are asked to rate sound quality differences among each anchor and labeled reference. Music samples will include: (a) 3 genres (classical, jazz, and pop/rock) with 5 stimuli per genre; and (b) high pass filtering at 100 Hz, 200 Hz, 400 Hz, 600 Hz, and 800 Hz frequencies. The anchor will be a bandpass version of the original with 1000 Hz - 1200 Hz passband.

Baseline (pre-surgery), 1 month post-activation, 3 months post-activation, 6 months post-activation, 7 months post-activation.

Sound localization will be measured in an audiometric booth with speakers placed in a horizontal plane at the participant's head level. Noise bursts will be presented to the participant, and the participant will be asked to locate which speaker the sound is coming from by pressing a button on a handheld device. Sound stimuli will be level roved (52, 62, and 72 dB SPL), 200 millisecond broadband presentation, 3 levels, 4 times, and 11 speakers (132 trials), 18 degrees apart.

Baseline (pre-surgery), 1 month post-activation, 3 months post-activation, 6 months post-activation, 7 months post-activation.

Inclusion Criteria: Adults (18 years of age or older) with moderate-to-profound sensorineural hearing loss bilaterally or single-sided deafness (SSD) Approved by the Cochlear Implant Program at London Health Sciences Centre (LHSC) to undergo unilateral cochlear implantation Identify as a native English speaker (recorded speech recognition/perception materials are presented in English) Deny any cognitive issues Willing and able to participate in study procedures (e.g., unaided testing, speech recognition testing, spatial hearing testing) Willing and able to return for follow-up visits (same intervals as recommended clinically for routine follow-up) Exclusion Criteria: Fail the Mini Mental State Examination based on their age and education level

Helpard L, Li H, Rohani SA, Zhu N, Rask-Andersen H, Agrawal S, Ladak HM. An Approach for Individualized Cochlear Frequency Mapping Determined From 3D Synchrotron Radiation Phase-Contrast Imaging. IEEE Trans Biomed Eng. 2021 Dec;68(12):3602-3611. doi: 10.1109/TBME.2021.3080116. Epub 2021 Nov 19.

Nikan S, Van Osch K, Bartling M, Allen DG, Rohani SA, Connors B, Agrawal SK, Ladak HM. PWD-3DNet: A Deep Learning-Based Fully-Automated Segmentation of Multiple Structures on Temporal Bone CT Scans. IEEE Trans Image Process. 2021;30:739-753. doi: 10.1109/TIP.2020.3038363. Epub 2020 Dec 4.