Optimizing performance outcomes for bilateral cochlear implant users to offset suboptimal electrode pairings

Staisloff, Hannah Elaine

Permalink

https://hdl.handle.net/2142/114072 Copy

Description

Title

Optimizing performance outcomes for bilateral cochlear implant users to offset suboptimal electrode pairings

Author(s)

Staisloff, Hannah Elaine

Issue Date

2021-12-01

Director of Research (if dissertation) or Advisor (if thesis)

Aronoff, Justin

Doctoral Committee Chair(s)

Aronoff, Justin

Committee Member(s)

• Agrawal , Smita
• Monson, Brian
• Chambers, Ron

Department of Study

Speech & Hearing Science

Discipline

Speech & Hearing Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Cochlear Implant
• Psychoacoustics

Abstract

Individuals who are severely hearing impaired, with lost or damaged hair cells, can benefit from a cochlear implant (CI). Studies have shown that listening with two ears, such as through bilateral cochlear implants (BiCIs), can yield considerable improvements in localizing sounds and understanding speech in noisy environments (Dunn et al. 2008; Eapen et al. 2009; Dunn et al. 2010). Chapter 1 first reviews the background of CIs and how they work, while chapter 2 gives a historical perspective of how the modern-day implants came to be. Although BiCI users often outperform unilateral CI users on localization and understanding speech-in-noise tasks, BiCI users do not receive the full benefit of having two ears (Loizou et al. 2009; Aronoff et al. 2010a). This reduced binaural benefit contributes to their difficulties in complex listening environments. Normal hearing (NH) individuals gain binaural benefits from both the use of interaural time differences (ITDs) and interaural level differences (ILDs), which help them understand speech in complex environments (Aronoff et al. 2011). However, CI users are generally not able to benefit from ITD cues (Aronoff et al. 2010a; Kerber & Seeber 2013) and can have decreased sensitivity to ILD cues (Laback et al. 2004), reducing the benefits they receive from having two ears. Chapter 3 provides an in-depth discussion of these binaural cues and why they are important. The difficulty that CI users display in complex listening environments for speech perception may stem from a variety of factors including but not limited to surgical depth differences of the array in each ear (Fayad et al. 1991; Aschendorff et al. 2005; Svirsky et al. 2015), scalar translocation of the array (Chakravorti et al. 2019; Jwair et al. 2021), neuronal survival in each cochlea (Fayad et al. 1991; Aschendorff et al. 2005; Svirsky et al. 2015), and individual patient factors (Blarney et al. 1996; Deggouj et al. 2007; Budenz et al. 2011). With the increase in adult bilateral implantation, it is important to determine how some individual patient factors such as age at implantation, duration of severe-to-profound deafness, sequential vs. simultaneous BiCIs, and/or pre-lingual vs. post-lingual deafness, affect speech perception performance in this population. Although previous studies have indicated that certain patient factors are negatively associated with the variability seen in speech perception tasks in CI users, of these published studies, many evaluated these factors in a quiet setting and ignored noise (e.g., ). In studies that did include testing in noise, it was typically co-located noise, which is not akin to what is typically seen in the real world (i.e., restaurant settings), where noise generally originates from a different spatial location than the target speech. Publications that have evaluated performance in quiet as well as in noise have done so through either retrospective reviews or experiments conducted across multiple centers. In doing so, various speech tests and testing methods were implemented, a factor which makes it challenging to determine the level of test consistency. Experiments evaluating the effects of noise have varied in which tests and speaker configurations were used. Many of these studies also investigated a wide range of listening conditions including bimodal: one hearing aid (HA) and one CI in the contralateral ear; unilateral: one CI; bilateral: two CIs. As well as variations of testing in these conditions (i.e., with and without HAs used in bimodal and unilateral conditions). Additionally, a wide range of ages (adults and children), small populations of subjects, single-channel and multi-channel CIs, and varying conditions of speech tests were also assessed. Therefore, it may be difficult to delineate if variability in performance on speech tasks can be attributed to the factors assessed, such as duration of deafness or age at implantation, as opposed to the secondary effects listed above (i.e., listening conditions) or a combination of both. Therefore, chapter 4 in this dissertation focuses on evaluating the effect of four common individual factors in adult BiCI users while accounting for spatial set up, speech test, and spatially-separated noise vs. quiet conditions. Since only a small subset of studies reviewed in chapter 4 found that individual patient factors consistently influence speech perception performance, this dissertation then focused on the ways that CIs are programmed, which is another reason why CI users may have difficulty with localization and speech perception in noise. BiCIs are independently programmed in clinics and the frequency bandwidth allocated to each electrode is based on the ordinal position of a given electrode along the electrode array. Electrode pairs that deliver the same frequency bands in the two ears are often not the optimal ones to pair together for binaural abilities based on interaural differences in their location in the cochlea and the neural populations stimulated by them (Poon et al. 2009; Kan et al. 2013; Hu & Dietz 2015). By pairing electrodes based on this method, BiCI recipients may have decreased sensitivity to ITDs and/or ILDs. Since the optimal electrodes to pair together in the left and right array are often not in the same relative location along the two arrays (Kan et al. 2013; Hu & Dietz 2015), there is a need to change how BiCI users are programmed. To do so, customized programs can be created to alter the frequencies allocated to each electrode to match the opposite array based on the optimal pairs. Because the ends of each array are not likely to match well with electrodes in the other ear, some electrodes will be unmatched, and if used, will result in some frequency regions only being delivered by electrodes in one ear or the other (Aronoff et al. 2016b; Aronoff et al. 2016a), which can result in reduced binaural benefits (Aronoff et al. 2014b). Within the HA literature, frequency transposition has been used to preserve the information in frequency regions that cannot effectively be delivered by the HA because of reduced auditory sensitivity (Parent et al. 1997; Kuk et al. 2009; Miller et al. 2016). Chapter 5 examines transposition and the ways it has been implemented in the HA literature. Introducing transposition into the bilateral programing of CI users may allow frequency regions removed in one ear or the other to align stimulation in the two ears to be retained, thus potentially increasing binaural benefits for these users. The goal of chapter 6, therefore, was to explore the feasibility of using transposition to help improve bilateral performance and restore binaural benefits in a CI simulation with NH listeners. Three experiments were conducted using vocoded stimuli comparing the performance on speech-in-noise tasks as well as a localization task across 5 conditions. One condition matched the allocation of frequencies in the output of the simulated left and right array (Matched). Another condition simulated a clinical situation where both arrays were given the same frequency allocation, however, one array was offset in the cochlea, so it covered a larger cochlear region than then other (Mismatched). An additional condition removed 2 channels of frequencies from the left array and was designed to deliver stimuli to only the matched locations in the two arrays (Matched + Deactivated). Another condition took the two previously removed regions from the Matched + Deactivated condition and moved them to the next available cochlear region by means of stacking the envelopes from the deactivated channels on top of those from the next available cochlear region, a form of transposition (Matched + Transposed). Finally, one condition used the same mismatch and frequency allocation that was used in the Matched + Deactivated but widened the frequency bands for the channel adjacent to the deactivated channel to incorporate the removed frequency regions from the deactivated channels. All of these conditions were manipulated at the apical (low frequency) end of the array for experiments 1 and 2 but experiment 3 (localization) additionally evaluated this condition with manipulation at the basal (high frequency) end. Results indicated that, for all experiments, transposition did not improve participants’ performance on either the speech perception or the localization task. Alternative methods such as transposition assume the array is first aligned, therefore it is important to determine the right way to align the two ears and whether different measures of matching the two arrays produce different results. There are several different binaural measures that can potentially be used to determine the optimal way to pair electrodes across the ears. Previous studies suggest that the optimal electrode pairing between the left and right ear may vary depending on the binaural task used (Kan et al. 2013; Hu & Dietz 2015). These studies, however, have only used one reference location or tested a single BiCI user. In both instances, it is difficult to determine if the results that were obtained reflect measurement error or if they instead reflect a systematic difference across binaural tasks. The purpose of chapter 7 was to determine if there were differences in the optimal pairing of electrodes across the measures of ITD, ILD, and pitch. The results in chapter 7 show that mismatches significantly differed in their effects on binaural cues such as ITD and ILD sensitivity, at least for the basal end of the array. This suggests that the improvements seen when using optimally paired electrodes may be tied to the particular percept being measured (ITD, ILD, or pitch) both to determine the electrode pairing and to assess performance, at least for the basal end of the array. Chapter 8, will discuss the overall conclusions from these studies and future directions. A brief synopses of each study in this dissertation is described below. Study 1: Evaluation of the effects of transposing frequencies on co-located speech-in-noise with unilateral input (Chapter 6). Approach: Moving information normally encoded by unmatched electrodes to the next available matched electrode through a process called transposition may allow listeners to increase performance on speech-in-noise tasks. To determine the impact of transposition on unilateral performance for NH listeners using a vocoder, 15 NH listeners listened to vocoded AzBio speech stimuli through headphones, delivered to only one ear, and repeated back what they heard. Measures: Unilateral testing was conducted with speech in noise, with both speech and noise co-located at 0 degrees azimuth. Hypothesis: Transposing frequency regions that were removed will improve speech-in-noise performance compared to the deactivating electrodes, a common practice seen in the clinic, and will yield scores similar to when both the left and right arrays are matched. Results: The results indicate that when information was removed as compared to the matched condition performance decreased. The results also indicated that there was a significant difference between the Mismatched condition and the Matched + Wide condition. Study 2: Evaluation of the effects of frequency transposition on bilateral performance and binaural benefits in the presence of spatially separated speech and noise (Chapter 6). Approach: To determine the impact of transposition on bilateral performance and binaural benefits for NH listeners using a vocoder, 15 NH listeners listened to vocoded AzBio speech stimuli through headphones and repeated back what they heard. Measures: AzBio testing was conducted both unilaterally and bilaterally with speech and noise spatially separated, with speech at 0 degrees and noise at 90 degrees azimuth. Bilateral testing was also conducted with speech and noise co-located at 0 degrees azimuth. These spatial configurations were used to evaluate bilateral performance for both 0 and 90 degrees azimuth as well as 3 binaural benefits: summation, the benefit of adding the full spectrum ear with spatially separated noise, the benefit of moving noise to the good ear. Hypothesis: For all binaural processing assessed, transposing frequency regions that were removed improved speech-in-noise performance more than deactivating electrodes, a common practice seen in the clinic, and yielded scores similar to when both the left and right arrays were matched. Results: The results demonstrated a significant improvement in summation benefit for all conditions compared to the Matched condition, but no significant differences between any of the other conditions. When adding the full spectrum ear with spatially separated noise, there was a significant improvement in performance for all conditions compared to the Matched condition. However, no significant differences were seen between any of the other conditions. When moving noise to the good ear, there was a significant decrease in performance seen for the Mismatched condition compared to that of the Matched condition. Study 3: Evaluation of the effects of frequency transposition on bilateral performance and binaural benefits for localization (Chapter 6). Approach: Sixteen NH listeners listened to vocoded stimuli presented from virtual locations through headphones and determined which virtual speaker they thought the sound came from. Measures: The average performance from the 3 randomized blocks (calculated as the root mean square error in degrees away from the target location) was calculated. Hypothesis: Transposition would improve localization performance more than deactivating electrodes, though only in the basal regions due to access to ILDs. Results: The results indicated that for apical (low frequency) manipulations, performance with Matched + Wide and Matched + Transposed was significantly worse than that of the best condition (Matched). For basal (high frequency) manipulations, performance with Matched + Wide was significantly worse than that of the best condition (Matched), as well as significantly worse than that of Matched + Transposed. Study 4: Determine whether, after experience-dependent adaptation, there are systematic differences in the optimal pairing of electrodes between the two ears at different points along the array for the perception of ITD, ILD, and pitch (Chapter 7) Approach: 7 BiCI users completed an ITD, ILD, and pitch-matching task for 5 locations equally spaced across the array. Measures: The optimal electrode pairs for all 5 locations on each task were collected and used in a comparison analysis. Comparisons were conducted to determine whether the optimal bilateral electrode pairs systematically differed in different regions along the array depending on the task used to measure them (ITD, ILD, or pitch). Comparisons were also conducted to determine how the pairs differed from the pairing in the participants’ clinical programs. Hypothesis: Since CI users have access to ILD cues, the matches between the two arrays would be much smaller than the differences seen for ITDs (a cue to which CI users do not currently have reliable access) and pitch (a cue with which many CI users struggle). Results: The results suggest that optimal electrode pairings differ depending on the cue measured to determine optimal pairing, at least for the basal end of the array. They also suggest that the improvements seen when using optimally paired electrodes may be tied to the particular percept being measured both to determine electrode pairing and to assess performance, at least for the basal end of the array.

Graduation Semester

2021-12

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/114072

Copyright and License Information

Copyright 2021 Hannah Staisloff

Owning Collections