Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Dec 1.
Published in final edited form as: Hear Res. 2022 Jul 28;426:108584. doi: 10.1016/j.heares.2022.108584

Cochlear implant spectral bandwidth for optimizing electric and acoustic stimulation (EAS)☆

René H Gifford a,*, Linsey W Sunderhaus a, Benoit M Dawant b, Robert F Labadie b,c, Jack H Noble b
PMCID: PMC10036878  NIHMSID: NIHMS1882180  PMID: 35985964

Abstract

Cochlear implantation with acoustic hearing preservation is becoming increasingly prevalent allowing cochlear implant (CI) users to combine electric and acoustic stimulation (EAS) in the implanted ears. Despite a growing EAS population, our field does not have definitive guidance regarding EAS technology optimization and the majority of previous studies investigating hearing aid (HA) and cochlear implant (CI) programming for EAS listeners have been mixed. Thus, the purpose of this exploratory study was to explore the effects of various EAS crossover frequencies—defined as the low-frequency (LF) CI cutoff—relative to the underlying spiral ganglion (SG) characteristic frequency associated with the most distal or apical electrode in the array. Speech recognition in semi-diffuse noise and subjective estimates of listening difficulty were measured for 15 adult CI recipients with acoustic hearing preservation in three listening conditions: 1) CI-alone, 2) bimodal (CI+HA), and best-aided EAS (CIHA+HA). The results showed no effect of LF CI cutoff for any of the three listening conditions such that there was no trend for increased performance or less subjective listening difficulty across LF CI cutoffs, referenced to underlying SG-place frequency. Consistent with past studies, the current results were also consistent with significant speech recognition and subject listening difficulty benefits for both bimodal (CI+HA) and best-aided EAS (CIHA+HA) as compared to CI-alone listening as well as significant additional benefits for best-aided EAS (CIHA+HA) compared to bimodal hearing (CI+HA). Future studies are necessary to investigate the efficacy of SG-place-based fittings for i) large samples of experienced EAS listeners for whom perceptual adaptation has occurred to the frequency mismatch provided by standard CI frequency allocations, and ii) EAS users at or close to CI activation as place-based approaches may ultimately yield greater outcomes, particularly for newly activated CI users for whom SG-place-based approaches may afford a steeper trajectory to performance asymptote.

1. Introduction

Combined Electric and binaural Acoustic Stimulation (EAS), made possible via hearing preservation with cochlear implantation, has become increasingly prevalent. Two of the three FDA-approved cochlear implant (CI) systems have EAS-specific indications allowing individuals with essentially normal low-frequency (LF) hearing through moderate hearing losses in the LF range and precipitously sloping high-frequency (HF) loss to obtain a hybrid or EAS cochlear implant (Cochlear 2014; MED-EL 2016). Further, all three CI manufacturers with FDA-approved systems have integrated hearing aid (HA) circuitry in their ear-level sound processors rendering EAS technology readily accessible to all audiologists managing patients with aidable hearing preservation in the implanted ear(s).

Despite availability of EAS technology and hundreds of peer-reviewed studies reporting hearing preservation in both adult and pediatric populations, programming and personalization of HA and CI parameters for EAS configurations has gained considerably less attention. Though there are studies examining EAS-related frequency allocation for the CI, HA, and the associated EAS boundary, the majority of studies have restricted experimentation to a single manufacturer’s device which has largely included similar insertion depths (Dillon et al., 2021a; Fraysse et al., 2006; Gifford et al., 2017; Gravel et al., 2005; Gstoettner et al., 2008; Imsiecke et al., 2020; Karsten et al., 2013; Kiefer et al., 2005; Plant and Babic, 2016; Simpson et al., 2009; Incerti et al., 2019). In addition, most of these studies have delivered acoustic and electric information in the implanted ear as referenced to a threshold-based criterion for determining the EAS crossover frequency—typically corresponding to the audiometric frequency for which thresholds were in the range of 70–90 dB HL (Gifford et al., 2017; Gstoettner et al., 2008; Karsten et al., 2013; Plant and Babic, 2016; Incerti et al., 2019).

Studies investigating the effect of CI and HA programming for EAS recipients have been mixed, particularly with respect to the EAS boundary and whether or not this includes spectral redundancy for electric and acoustic hearing in the implanted ear. Karsten and colleagues (Karsten et al., 2013) investigated the effect of EAS boundary on speech recognition for 10 short-electrode recipients (10 mm, Hybrid S10 and S12) who had at least 1 year of EAS experience. All listeners were tested on measures of consonant recognition in quiet and spondee recognition in the presence of multi-talker babble in the ipsilateral EAS condition alone (contralateral ear was occluded). They found no group differences in consonant recognition between conditions of spectral overlap for HA and CI programming and a spectral “meet” condition for which the HA and CI bandwidths were immediately adjacent, but did not overlap. However, they did observe a significant decrement in group-level performance with spectral overlap as compared to the spectral meet condition for sentence recognition in multi-talker babble, though there was considerable variability (Karsten et al., 2013). Investigating the impact of EAS boundary programming for HA and CI for conventional, 5 longer electrode recipients1 (Nucleus CI24RE(CA)), Simpson and colleagues (Simpson et al., 2009) reported no differences in monosyllabic word recognition or speech recognition in noise performance between two conditions involving 1) complete spectral overlap between HA and CI, and 2) pitch perception based CI allocation resulting in a spectral “meet” condition.

Vermeire and colleagues (Vermeire et al., 2008) investigated the effect of EAS boundary for a group of 4 experienced CI recipients who had at least 1 month of EAS experience and were implanted with conventional, long electrode arrays (MED-EL standard or medium). They investigated the effects of EAS overlap using two EAS programming approaches 1) full spectral overlap between CI and HA, and 2) minimum spectral overlap with the CI programmed to the frequency at which the unaided audiogram exceeded 65 dB HL which they termed the audiogram “fall-off frequency.” Vermeire et al. (Vermeire et al., 2008) reported significantly higher sentence recognition in noise with the minimal EAS overlap using the audiogram fall-off for the LF CI cutoff as compared to the full spectral overlap for HA and CI.

In a study of 11 experienced CI recipients (13 ears) with Nucleus electrode arrays (slim straight and contour advance), Gifford and colleagues (Gifford et al., 2017) found that minimal spectral overlap between the HA and CI resulted in significantly higher sentence recognition in noise scores as compared to conditions with complete overlap or conditions involving a spectral gap. The spectral overlap condition yielding highest outcomes was achieved by setting the LF CI boundary closest to the audiometric frequency reaching a 70-dB-HL threshold5 —similar to the audiogram fall-off frequency as reported by Vermeire and colleagues (Vermeire et al., 2008). Though this study included experienced CI users with 3.0 years’ experience, on average, just 3 of the study participants’ (5 ears) had chronic EAS experience prior to study enrollment; the remaining 10 participants had no previous EAS experience and were thus, fitted with EAS technology and tested acutely.

Incerti and colleagues (Incerti et al., 2019) also investigated the effect of EAS crossover frequency for a group of 10 experienced adult CI users (5 Hybrid-L24, 3 CI422, 1 CI532, 1 CI24RE(CA)) with acoustic hearing preservation; despite long-term CI experience, the paper did not specify whether the participants had EAS experience, specifically with acoustic amplification in the implanted ear. They investigated the effects of three EAS crossover frequencies defined as the LF CI cutoff associated with the audiogram frequency approximating thresholds of 60, 75, and 90 dB HL termed C60, C75, and C90. Unique to this study was that they also adjusted the bandwidth of acoustic amplification in the CI ear to match the LF CI cutoff—which would result in a programming approach similar to what previous studies termed the “meet” condition (Gifford et al., 2017; Karsten et al., 2013). All study participants were provided with a 4-week period to adjust to the EAS settings prior to assessment. Adaptive speech receptive thresholds (SRTs) for sentences in the presence of multi-talker babble were similar for all three LF CI cutoffs in co-located (S0N0) and spatially separated speech and noise (S0 N90); however, SRTs were lower (i.e. better) for C60 and C75 as compared to C90—an effect not found to be statistically significant. Other measures tested including horizontal plane localization and subjective reports of hearing and communication (Self Evaluation of Listening Function (SELF) and Speech, Spatial, Qualities (SSQ)) were also similar across the three LF CI cutoffs such that there were no significant differences noted. They did, however, also obtain listener preference measures and found that 7 of the 10 participants expressed a preference for an EAS crossover condition which was equivalent to both the condition yielding the listener’s best SRT for speech recognition in noise and the EAS crossover condition that they had been using upon study enrollment. Plant and Babic (Plant and Babic, 2016) also investigated LF CI cutoff preferences for 9 of their 16 enrolled adult EAS users (9 Hybrid-L24). They found that 7 of 9 participants preferred non-overlapping electric and acoustic spectral ranges as opposed to a full CI bandwidth in the implanted ear; however, they did not report speech recognition differences found for complete spectral overlap as compared to a non-overlapping spectrum for electric and acoustic hearing.

Imsiecke and colleagues (Imsiecke et al., 2020) also investigated the effects of three EAS boundaries for 15 adults with at least 10 months of EAS experience who had been implanted with MED-EL Flex electrode arrays (5 Flex 16, 3 Flex 20, 4 Flex 24, and 3 Flex 28). The three programming options included: 1) meet map—the standard MED-EL EAS defaults which includes setting the LF CI cutoff to the audiometric frequency associated with a 65-dB-HL threshold—similar to the audiometric fall-off frequency described above (Vermeire et al., 2008), 2) overlap map—the LF CI cutoff was extended to lower frequencies to allow EAS overlap thereby setting the LF CI cutoff up to 2 octaves below the prescribed EAS cutoff for a given individual, and 3) UNMASKfit—the electric LF CI cutoff was equivalent to that in the meet map, however, charge was reduced to minimize the potential effects of electric on acoustic masking. Their study participants were provided with 4 weeks’ experience with each program prior to assessment via adaptive speech receptive thresholds (SRTs) in the presence of steady-state noise and ICRA fluctuating noise. For the EAS condition in steady-state noise, there was no difference in SRTs for the meet map and UNMASKfit conditions, whereas both conditions yielded better performance than the overlap map. For the EAS condition in ICRA fluctuating noise, there were no significant differences in participant SRTs across the three EAS conditions (Imsiecke et al., 2020).

The majority of the studies in this space have investigated EAS programming adjustments for the LF CI cutoff referenced to the audiogram fall-off frequency; however, there are relatively recent studies that have investigated the effects of providing a tonotopic match for the electric-hearing component for EAS simulations (Fu et al., 2017; Willis et al., 2020) or utilized individualized angular insertion depth measures to offer spiral ganglion (SG) place-based mapping for EAS patients (Dillon et al., 2021a) or applying an SG-place approach for normal hearing (NH) participants listening to EAS simulations (Dillon et al., 2021b). For the studies utilizing SG-place-based programming, researchers obtained intracochlear localization data of the most distal or apical electrode contact and determined a frequency match corresponding to the underlying spiral ganglion (SG) characteristic frequency and/or radial fiber tuning associated with the organ of Corti location, based on individualized cochlear duct length estimates (Dillon et al., 2021a; Dillon et al., 2021b). Place-based methods have been made possible by the emerging availability of postoperative, CT-based localization of the implanted electrode array (Canfarotta et al., 2019; Jiam et al., 2019; Noble et al., 2012; Noble et al., 2013), as well as advanced automated methods affording creation of three-dimensional models of a patient’s cochlea and individualized electrode array placement (Noble et al., 2012; Zhao et al., 2019; Y. Zhao et al., 2018; Y. Zhao et al., 2018). Dillon and colleagues (Dillon et al., 2021b) demonstrated sentence recognition in noise benefit on the order of 20-percentage points for place-based EAS simulations as compared to the default CI/EAS frequency allocation for 17 young adults with normal hearing listening to EAS simulations. In a study of place-based versus default EAS mapping for 2 EAS patients, Dillon et al. (Dillon et al., 2021a) demonstrated individually significant improvement in phoneme recognition with place-based mapping on the order of 24 to 50 rationalized arcsine units.

Despite the promising outcomes described in the SG-place-based EAS fitting papers, clinical application of these data is currently limited given the use of EAS simulations with NH listeners (Dillon et al., 2021b) and the investigation of 2 EAS recipients with place-based and default EAS fitting at initial activation (Dillon et al., 2021a). The 2 EAS recipients at activation were both implanted with the MED-EL Flex 28 electrode with similar angular insertion depths of 649 and 669° (Dillon et al., 2021a); thus, the place-based and default EAS crossover frequencies were quite similar for these participants differing by just 20 to 100 Hz. This is in contrast to shorter electrode arrays with shallower angular insertion depths for which the difference between the default EAS cutoff—as indicated in the manufacturer software—and the SG-place-based cutoff for the most distal or apical electrode is generally an octave or more, thereby limiting transmission of electric low-frequency information. Additionally, there was no comparison completed between the SG or organ of Corti frequency allocation and the standard EAS fitting formula taking into account LF audiometric thresholds.

1.1. Current study

There is great need to investigate the effectiveness of various EAS programming options for HA and CI to provide the maximum auditory potential for CI recipients with acoustic hearing preser-vation. This is particularly true given the poor utilization of EAS technology in adult CI recipients with acoustic hearing preservation despite significant perceptual EAS benefits (Gifford et al., 2017; Plant and Babic, 2016; Dunn et al., 2010; Gifford et al., 2013; Gifford et al., 2014; Rader et al., 2013). Specifically, only 50% (Spitzer et al., 2020) to 69% (Perkins et al., 2021) of adult CI users with hearing preservation are using EAS technology despite high rates of acoustic hearing preservation in the range of 82–94% (Perkins et al., 2021; Gantz et al., 2018; Pillsbury et al., 2018). Investigation of the efficacy of SG place-based mapping for EAS recipients with various electrode arrays and associated angular insertion depths holds high potential for clinical application, particularly for clinical programs with diversity of electrode arrays and angular insertion depths. Thus, the purpose of this exploratory study was to investigate the effect of various EAS crossover frequencies on speech recognition in noise and subjective estimates of listening difficulty as related to the underlying SG characteristic frequency associated with the most distal or apical electrode in the array. The null hypothesis was consistent with no difference in speech recognition performance or subjective estimates of listening difficulty across the different LF CI cutoffs used for assessment of EAS boundary.

2. Material and methods

2.1. Participants

Fifteen adult unilateral CI recipients with acoustic hearing preservation in the implanted ear were recruited for participation. Mean age at testing was 56.7 years (range 29 to 81 years) and all but two participants were experienced CI users with at least 1 year of CI use prior to study enrollment. Participants 14 and 15 had 1 month of CI experience at study enrollment and neither had been previously fitted with EAS. Participants had both short or hybrid electrode arrays (4 Hybrid-L24) as well as standard or conventional length electrode arrays (5 CI422/522/622, 1 mid-scala, 1 CI632, and 4 Flex 28). All but participants 6 and 15 (mid-scala and CI632) had straight or lateral wall electrode arrays. Audiometric thresholds in the implanted and non-implanted ears on the day of study enrollment are shown in Fig. 1.

Fig. 1.

Fig. 1.

Open in a new tab

Individual audiometric thresholds, in dB HL, for the CI and non-CI ears.

All study participants had obtained postoperative computed tomography (CT) imaging allowing us to segment each participant’s cochlear anatomy via fitting with a high-resolution anatomical model created from micro-CT images of cadaver cochleae (Noble et al., 2012; Noble et al., 2013). Following anatomical segmentation, automated processing methods localized the implanted electrode array (Zhao et al., 2019) ultimately resulting in the creation of a 3-dimensional model of a patient’s inner ear with electrode array placement relative to scalar location and electrode-to-modiolus distance for each electrical contact as shown in Fig. 2. Image processing also allowed us to define the angular insertion depth, in degrees, for each electrical contact as well as the underlying spiral ganglion (SG) characteristic frequency associated with each electrode insertion depth (Stakhovskaya et al., 2007). Table 1 displays scalar electrode location and mean electrode-to-modiolus across the entire array as well as age at enrollment, device, electrode array, and angular insertion depth for the most distal or apical electrode. In addition to these data, Table 1 also displays the upper cutoff for the acoustic component (i.e. processor integrated HA) in the implanted ear as well as the experimental EAS crossover frequencies—defined as the LF CI cutoff for the most apical electrode—and the relative distance from the underlying SG characteristic frequency place for the most apical electrode in octave spacing. This study was approved by the Vanderbilt University Medical Center Institutional Review Board and all participants were compensated for their time spent participating.

Fig. 2.

Fig. 2.

Open in a new tab

Three-dimensional model of a patient’s cochlea displayed from a lateral-to-medial view orthogonal to the mid-modiolar axis (left) and from a view along the mid-modiolar axis (right). Scalar array placement, angular insertion depth, and single electrode-to-modiolus distance are displayed. The implanted electrode in this case is a CI632 that has been slightly over inserted such that the mid-section of the array is closer to the lateral wall.

Table 1.

Participant demographic and experimental information including age at testing (years), duration of CI and EAS use (years), electrode array type and length (mm), scalar electrode location, mean electrode-to-modiolus distance (mm), angular insertion depth (deg), SG CF for apical electrode (Hz), upper frequency cutoff for HA in implanted ear (Hz), the presence of “meet” and “gap” conditions between the electric and acoustic bandwidths for the different LF CI cutoffs, and LF CI cutoffs (in Hz) for each participant as referenced to SG CF. The number in parentheses represents the deviation, in Hz, from the exact SG-referenced condition. Participant labels with asterisk indicate those fitted and tested acutely with EAS. The shaded cell for each participant indicates the clinical standard LF CI cutoff equivalent to the audiometric frequency reaching 70 dB HL in the implanted ear. The bolded and outlined cells represent the chronic EAS condition for the 6 participants with long-term EAS experience. LF: low frequency; ST: scala tympani; SV: scala vestibuli; SG: spiral ganglion; CF: characteristic frequency; HA: hearing aid.

Participant
label		 age at
testing
(years)		 duration of
CI & EAS
use
(months)		 electrode array,
length (mm)		 scalar
electrode
location		 mean
electrode-to-
modiolus
distance
(mm)		 angular
insertion
depth (deg)		 SG CF
for most
apical
electrode
(Hz)		 upper
frequency
cutoff for
HA in CI
ear (Hz)		 "meet"
condition
tested?		 "gap"
condition
tested?		 EAS crossover re: SG CF
−4–oct −3–oct −2.5–oct −2–oct −1.5–oct −1–oct −0.75 −0.5–oct −0.25–oct SG
1* 41 26, 0 CI422, 20 ST 1.17 318 1120 685 no no 188 (48) 313 (33) 438 (42) 563 (3)
2 48 15, 14 Hybrid-L24, 16 ST 0.67 221 2218 685 no yes: 813 Hz 188 (49) 313 (36) 438 (46) 563 (9) 813 (29)
3 38 28, 27 Hybrid-L24,16 ST 0.95 232 2056 685 no yes: 813 Hz 188 (60) 313 (56) 438 (75) 563 (49) 813 (86)
4 62 12, 11 Hybrid-L24, 16 ST 0.92 237 2136 866 no no 188 (55) 313 (46) 438 (60) 563 (29) 688 (67)
5 61 13, 12 Hybrid-L24, 16 ST 0.94 251 1637 1000 yes: 938 Hz yes: 1188 Hz 438 (29) 688 (109) 813 (6) 938 (35) 1188 (31)
6 47 37, 36 mid-scala, 18.5 ST 0.51 354 827 685 yes: 690 Hz no 350 (64) 520 (65) 690 (5)
7* 60 12, 0 CI522, 20 ST 1.12 464 588 685 no no 188 (84) 313 (19) 438 (22) 563 (25)
8* 51 12, 0 CI422, 20 ST 1.19 467 514 685 no no 188 (6) 313 (50) 438 (6) 563 (49)
9 81 44, 43 Flex 28, 28 ST 1.24 456 534 430** yes: all but 500 Hz yes: 500 Hz 70 (3) 150 (17) 250 (17) 350 (28) 500 (34)
10* 72 22, 0 Flex 28, 28 ST 1.13 406 658 430** yes: all tested no 70 (12) 150 (15) 250 (17) 350 (21)
11* 56 16, 0 Flex 28, 28 ST-SV 1.24 523 394 430** yes: all tested no 70 (0.4) 150 (11) 250 (29) 350 (44)
12* 71 12, 0 Flex 28, 28 ST 1.20 584 291 430** yes: all tested no 70 (3) 150 (5) 250 (5) 350 (59)
13* 78 15, 0 CI622, 20 ST 1.02 384 738 866 no no 188 (3) 313 (56) 438 (0) 563 (40) 688 (50)
14* 55 1, 0 CI622, 20 ST 1.26 359 781 560 yes: 563 Hz no 188 (7) 313 (38) 438 (26) 563 (11)
15* 29 1, 0 CI632, 18.4 ST-SV 1.15 428 658 1000 no no 188 (24) 313 (16) 438 (27) 563 (8) 781 (0)
Open in a new tab
*

acute EAS fitting and testing.

shaded: EAS crossover = 70 dB HL condition.

bolded: chronic EAS condition.

**

all MED-EL EAS fittings are "meet"; however, this is the maximum audible frequency for the HA, based on the audiogram.

2.2. Stimuli

Participants’ HAs were programmed to meet NAL-NL2 (Keidser et al., 2011) target output for audiometric thresholds up to 90 dB HL to be consistent with our previous study (Gifford et al., 2017) aiming for consistency with the U.S. Nucleus Hybrid clinical trial. For the implanted ear, however, no amplification was provided above 1000 Hz given the goal of providing LF audibility affording access to interaurai time differences (ITDs) (Gifford et al., 2014; Gifford and Stecker, 2020); this was the case even if an individual had audiometric thresholds ≤ 90 dB HL above 1000 Hz as observed for participants 5 and 15. Amplification was provided via the integrated acoustic component for each of the manufacturer processors included in the study including Nucleus 6 and 7, Naida CI Q90, and Sonnet 2 EAS. All HA settings were verified via real-ear measures prior to assessment.

Speech understanding was assessed using a + 5 dB signal-to-noise ratio (SNR) for which AzBio (Spahr et al., 2012) sentences were presented at 67 dB SPL at 0° with a semi-diffuse restaurant noise originating from seven loudspeakers oriented from 45° to 315 ° with 45° spacing (R-SPACE™ system). All Nucleus CI recipients were utilizing Autosensitivity control (ASC) in their everyday program and thus, ASC was left active for experimentation preserving the full +5 dB SNR and speech peaks associated with the 67-dB-SPL signal.

2.3. Procedure

The effect of EAS boundary, via LF CI cutoff, on speech recognition and perceived listening difficulty was assessed in a single test session lasting 1.0 to 2.5 h, including participant breaks. Six of the fifteen participants were full-time, chronic EAS users including bilateral acoustic amplification paired with their CI; the remaining nine participants were fitted with EAS technology and assessed acutely within the same session. Prior to study enrollment, the nine study participants who had not previously used EAS were using CI maps with the manufacturer’s default full CI bandwidth which was 250–8700 Hz for AB, 188–7938 Hz for Cochlear, and 100–8500 Hz for MED-EL.

Table 1 indicates the nine participants who had been fitted with EAS and acutely tested via asterisk next to numerical participant label. Note that in our 2017 study (Gifford et al., 2017), we investigated the effect of acute versus chronic EAS use in three individuals (five ears) and found no effect of listening experience. Thus, in the interest of time and the fact that several participants traveled from a distance to participate, we did not investigate the effect of EAS listening experience, in years, for the nine EAS participants acutely tested.

Speech understanding in the presence of semi-diffuse restaurant noise was evaluated in the bimodal (CI+HA) and best-aided EAS (CIHA+HA) conditions for all fifteen participants. Ten of fifteen participants were also tested in the CI-alone condition utilizing an EAR foam plug in each ear. Participants were also asked to rate perceived listening difficulty for each of the LF CI cutoffs tested. Listening difficulty ratings were obtained using a visual analog scale (VAS) identical to that used in our previous study (Gifford et al., 2017). The VAS displayed 10-equidistant ticks ranging from 1 to 10 with 1 labeled as “no difficulty at all” and 10 labeled as “most difficulty imaginable”. In standard clinical practice, a full CI bandwidth is generally provided for CI recipients utilizing CI-alone or bimodal (CI+HA) hearing configurations. Consequently, it was not a primary aim of the current study to investigate the effect of EAS boundary (i.e. LF CI cutoff) for the bimodal (CI+HA) and CI-alone conditions. Rather, we assessed speech recognition and subjective listening difficulty in all listening conditions for each of the LF CI cutoffs so that we could both compare to existing data in the literature (Gifford et al., 2017) as well as identify the impact of CI spectral bandwidth for scenarios in which a listener may experience a malfunctioning HA in the implanted and/or non-implanted ears.

Study participants 1 through 14 were CT imaged immediately following study consent; however, image processing was not completed for several hours or up to a day later in all cases. Thus we did not have angular insertion depth information prior to beginning speech recognition testing. Participant 15 had been CT imaged intraoperatively and thus we had imaging information and were able to select their exact SG characteristic frequency as a tested condition. For the 14 participants without immediately accessible imaging information, we utilized various standard EAS boundaries for all three CI manufacturers and later referenced these LF CI cutoffs to the underlying SG characteristic frequency for the most apical electrode in the array.

Table 1 displays the different EAS crossover frequencies used for testing for each participant and the relative deviation, in full- and partial-octave steps, to the SG-place frequency for the most apical electrode. The SG relative deviation was calculated as the spectral distance, in octaves, between the SG characteristic frequency for the most apical electrode and the programmed LF CI cutoff for a given test condition. For example, if the SG characteristic frequency was 1200 Hz and the LF CI cutoff was set to 600 Hz, this would represent a 1-octave difference and would be expressed as −1 oct in Table 1. Because we did not have the results of CT imaging prior to commencing data collection for 14 of 15 participants, we did not have an opportunity to create EAS maps for which the LF CI cutoff exactly matched the relative deviation from the SG characteristic frequency in discrete- or partial-octave steps.2 Consequently, we calculated the actual difference, in Hz, between the SG relative deviation and the actual LF CI cutoff for all tested frequencies. This information is included in Table 1 in parentheses next to the EAS crossover frequency representing each of the octave-based deviations from the underlying SG characteristic frequency for the most apical electrode.

For the six participants with chronic EAS experience, the LF CI cutoff to which they were accustomed is listed in bold text in Table 1. The LF CI cutoff corresponding to the manufacturers’ full spectral bandwidth for electric-only programs is 250 Hz, 188 Hz, and 100 Hz for AB, Cochlear, and MED-EL, respectively. Thus, the LF CI cutoffs corresponding to the six chronic EAS listeners’ everyday maps (shown in bold in Table 1) were considerably higher than would be allocated for a full spectral bandwidth typically used for electric-only maps.

For all fifteen participants, the LF CI cutoff most closely matching the audiometric frequency for which thresholds reached 70 dB HL, or the audiometric fall-off frequency, is indicated via gray shading in Table 1. Based on audiometric characteristics and electrode type, different EAS boundary conditions were tested across the study population; however, we were most interested in trends associated with performance and perceived listening difficulty as we deviated further from the SG place frequency. The majority of LF CI cutoffs tested incorporated some degree of spectral-based overlap between the HA and CI bandwidth in the implanted ear—consistent with spectral “overlap” as indicated by previous studies (Gifford et al., 2017; Imsiecke et al., 2020; Karsten et al., 2013). Seven of fifteen participants were tested in a “meet” map defined here as less than 100-Hz overlap between the upper frequency cutoff for the HA and LF CI cutoff for the EAS (CIHA+HA) conditions. Finally, four of the participants were tested in a spectral “gap” condition for which there was a range of frequencies not transmitted by either the HA or CI in the implanted ear. Spectral information for HA and CI as well as the presence of “meet” and “gap” conditions is shown in Table 1. All statistical analyses were completed using Graphpad Prism version 9 (San Diego, CA) and SPSS version 28 (Armonk, NY).

3. Results

3.1. Speech recognition performance in noise

Individual speech recognition scores, in percent correct, for each LF CI cutoff tested are shown in Fig. 3A-3C for the CI-alone, bimodal (CI+HA), best-aided EAS (CIHA+HA) conditions, respectively. The data displayed in Fig. 3 demonstrate that despite some within-subjects variation in performance across EAS boundary, speech recognition performance was fairly consistent across most of the LF CI cutoffs tested for all listening conditions. Because different CI cutoffs were used for the participants, we completed data analysis via nonlinear (quadratic) regression comparing the fit of the nonlinear regression to that of a first order polynomial (i.e. straight line) using an extra sum-of-squares F test with Bonferroni correction for multiple comparisons (α=0.003). The goal of data analysis was to determine whether speech recognition performance varied as the LF CI cutoff deviated further from the SG place frequency for the most apical electrode.

Fig. 3.

Fig. 3.

Open in a new tab

Individual speech recognition in noise scores, in percent correct, plotted against LF CI cutoff relative to the SG CF for the most apical electrode. Scores are shown for the CI-alone, bimodal (CI+HA), and best-aided EAS (CIHA+HA) conditions in panels A, B, and C, respectively. SG: spiral ganglion; CF: characteristic frequency.

For the best-aided EAS listening condition (CIHA+HA) displayed in Fig. 3C, though there were some performance differences across LF CI cutoffs, all 15 participants’ data were best fit via first order polynomial regression (i.e. linear regression) with p values ranging from 0.10 to 0.96 for all analyses. Considering the resultant linear regressions, after adjusting for multiple comparisons, none of the 15 regression slopes were found to be significantly different from zero (p ranging from 0.03 to 0.91 for all analyses) demonstrating no significant effect of LF CI cutoff on speech recognition performance in the best-aided EAS condition (CIHA+HA).

For the bimodal condition (CI+HA) displayed in Fig. 3B, all but 1 participants’ data (participant 5) were best fit via first order polynomial (p ranging from 0.02 to 0.81 for all analyses). For the 14 fitted via linear regression, none of the regression slopes were found to be significantly different from zero with p values ranging from 0.05 to 0.82 for all analyses; thus this finding demonstrated no effect of LF CI cutoff on speech recognition performance in the bimodal condition (CI+HA). For participant 5, whose data were best fit via quadratic regression (F(1,1) = 55,569.0, p = 0.002), the trend was parabolic in nature such that performance decreased the further the LF CI cutoff deviated from SG place frequency as well as for the LF CI cutoffs closest to the SG place. Of note, participant 5 was one of six listeners with chronic EAS experience and was implanted with a short electrode array (Hybrid-L24).

Ten of the fifteen participants also completed CI-alone testing and individual data are displayed in Fig. 3A. All of the participants’ data were best fit via first order polynomial regression (p > 0.05 for all analyses) and none of the linear regression slopes were found to be significantly different from zero with p values ranging from 0.02 to 0.95 for all analyses; again, this demonstrated no effect of LF CI cutoff on speech recognition performance in the CI-alone condition.

3.2. Subjective listening difficulty for speech in noise

Individual subjective reports of listening difficulty for each LF CI cutoff are shown in Fig. 4A-4C for the CI-alone, bimodal (CI+HA), best-aided EAS (CIHA+HA) conditions, respectively. Subjective listening difficulty estimates were obtained for 10 of the 15 enrolled participants in this study. Identical to the analyses completed for Fig. 3, we completed data analysis via nonlinear (quadratic) regression comparing the fit of the nonlinear regression to that of a first order polynomial (i.e. straight line) using an extra sum-of-squares F test with Bonferroni correction for multiple comparisons (α=0.003). As with speech recognition in noise, the goal of this analysis was to determine whether subjective estimates of listening difficulty varied as the LF CI cutoff deviated further from the SG place frequency.

Fig. 4.

Fig. 4.

Open in a new tab

Individual subjective listening difficulty scores on the visual analog scale (VAS) plotted against LF CI cutoff relative to the SG CF for the most apical electrode. VAS scores are shown for the CI-alone, bimodal (CI+HA), and best-aided EAS (CIHA+HA) conditions in panels A, B, and C, respectively. SG: spiral ganglion; CF: characteristic frequency.

For the best-aided EAS condition (CIHA+HA) displayed in Fig. 4C, though there were some subjective differences in listening difficulty across the CI cutoffs tested, all participants’ data were best fit via first order polynomial regression with p values ranging from 0.17 to 0.98. Considering the resultant linear regression, after adjusting for multiple comparisons, none of the linear regression slopes were significantly different from zero (p ranging from 0.01 to 0.78) demonstrating no statistically significant effect of LF CI cutoff on perceived listening difficulty in the best-aided EAS condition (CIHA+HA).

For the bimodal listening condition (CI+HA) displayed in Fig. 4B, all participants’ data were best fitted via linear regression with p values ranging from 0.01 to 0.43. For the resultant linear regression analyses, none of the regression slopes were found be significantly different from zero (p ranging from 0.12 to 0.94) demonstrating no statistically significant effect of LF cutoff on perceived listening difficulty in the bimodal condition (CI+HA).

For the CI-alone condition displayed in Fig. 4A, all 10 participants’ data were best fit via linear regression versus quadratic regression with p values ranging from 0.09 to 0.96 for all analyses. Furthermore, none of the resultant linear regression slopes were significantly different from zero (p ranging from 0.01 to 0.83) demonstrating no significant effect of LF cutoff on perceived listening difficulty in the CI-alone condition.

3.3. EAS benefit (CIHA+HA vs. CI+HA) for speech recognition in noise and perceived listening difficulty

In order to compare the current data set to previous reports in the literature investigating the effectiveness of bimodal hearing (CI+HA) and CI plus binaural acoustic hearing (CIHA+HA) as compared to CI-alone listening, we also plotted speech recognition and subjective listening difficulty for each individual participant in Figs. 5 and 6, respectively. Figs. 5A and 6A display a singular score for each of the 3 listening conditions including CI-alone for the full CI bandwidth tested, bimodal (CI+HA) using the full CI bandwidth, and best-aided EAS (CIHA+HA) using the LF CI cutoff closest to the audiometric frequency reaching 70 dB HL for a given listener. The full CI bandwidth for CI-alone and bimodal (CI+HA) and the 70-dB-HL criterion for determining LF CI cutoff in the CI ear were chosen as these fitting approaches would be most consistent with current clinical practice; however, because there were variations in performance across the EAS (CIHA+HA) boundary conditions—and the fact that well-defined clinical guidelines do not currently exist for EAS programming—we also wanted to investigate individual and mean performance using the LF CI cutoff yielding the best score in each of the 3 listening conditions. Thus, Figs. 5B and 6B display an individual listener’s best scores obtained for each of the 3 listening conditions.

Fig. 5.

Fig. 5.

Open in a new tab

Individual and mean speech recognition in noise scores, in percent correct, for the CI-alone (gray), bimodal (white), and best-aided EAS (black) conditions. Panel A represents scores obtained for the standard clinical fittings including full CI bandwidth for CI-alone and bimodal and the 70-dB-HL frequency for the EAS condition. Panel B represents individual best scores for each listening condition. Error bars represent +1 standard error.

Fig. 6.

Fig. 6.

Open in a new tab

Individual and mean scores of subjective listening difficulties for the CI-alone (gray), bimodal (white), and best-aided EAS (black) conditions. Panel A represents scores obtained for the standard clinical fittings including full CI bandwidth for CI-alone and bimodal and the 70-dB-HL frequency for the EAS condition. Panel B represents individual best scores for each listening condition. Error bars represent +1 standard error.

Focusing first on Fig. 5A displaying speech recognition consistent with current clinical CI and EAS programming practices, the mean scores for CI-alone, bimodal (CI+HA), and best-aided EAS (CIHA+HA) were 38.8%, 59.4%, and 74.4%, respectively. For Fig. 5B, considering the highest individual performance for each listening condition, mean scores for CI-alone, bimodal (CI+HA), and best-aided EAS (CIHA+HA) were 52.0%, 71.2%, and 80.1%, respectively. Speech recognition in noise scores were poorer in all conditions with the CI programmed using standard clinical programming recommendations (i.e. full CI bandwidth for CI-alone and bimodal and LF CI cutoff equal to the 70-dB-HL frequency for EAS) as compared to the “best” individual score for each participant. Indeed, we completed statistical analysis via linear mixed model accounting for repeated measures assessment using both listening condition (CI, bimodal, and best-aided EAS) and LF CI cutoff (standard clinical or individual best score) as independent variables and speech recognition in noise as the dependent variable. We found a statistically significant effect of listening condition (F(2, 48.76) = 27.61, p < 0.001, ηp2 = 0.53), a significant effect of LF CI cutoff (F(1, 51.43) = 9.53, p = 0.003, ηp2 = 0.16), and no interaction (F(2, 48.76) = 0.53, p = 0.59, ηp2 = 0.02). All pairwise multiple comparisons (Sidak) revealed that scores across the 3 listening conditions were all significantly different (p ≤ 0.021 for all comparisons), as expected. For the CI-alone condition, there was a statistically significant difference between clinical and best LF CI cutoff (t = 3.73, p < 0.001). Similarly for the bimodal condition, there was a statistically significant difference between clinical and best LF CI cutoff (t = 3.77, p < 0.001). However, for the best-aided EAS condition, there was not a statistically significant difference between clinical and best LF CI cutoff (t = 1.91, p = 0.07).

We completed similar analyses for subjective reports of listening difficulty across LF CI cutoffs for each listening condition. For Fig. 6A displaying subjective listening difficulty for conditions programmed via standard clinical CI and EAS programming, mean scores for CI-alone, bimodal (CI+HA), and best-aided EAS (CIHA+HA) were 8.9, 7.5, and 5.4, respectively. Considering the highest individual performance obtained for each listening condition as displayed in Fig. 5B, mean subjective listening difficulty scores for CI-alone, bimodal (CI+HA), and best-aided EAS (CIHA+HA) were 7.2, 6.2, and 4.7, respectively (see Fig. 6B). Similar to speech recognition analyses, we completed statistical analysis via linear mixed model accounting for repeated measures assessment using both listening condition (CI, bimodal, and best-aided EAS) and LF CI cutoff (standard clinical or individual best score) as independent variables and subjective listening difficulty as the dependent variable. We found a statistically significant effect of listening condition (F(2, 39.0) = 13.71, p < 0.001, ηp2 = 0.41), a significant effect of LF CI cutoff condition (F(1, 53.3) = 6.72, p = 0.012, pp2 = 0.11), and no interaction (F(2, 39.0) = 0.37, p = 0.70, pp2 = 0.02). All pairwise multiple comparisons (Sidak) were completed revealing that scores across the 3 listening conditions were all significantly different (p ≤ 0.02 for all comparisons). For the CI-alone condition, there was a statistically significant difference between clinical and best LF CI cutoff (t = 3.10, p = 0.005). Similarly for the bimodal condition, there was a statistically significant difference between clinical and best LF CI cutoff (t = 2.69, p = 0.014). However, for the best-aided EAS condition, there was not a statistically significant difference between perceived listening difficulty with the clinical standard and an individual’s best LF CI cutoff (t = 1.25, p = 0.224).

4. Discussion

Despite the increasing prevalence of acoustic hearing preservation with cochlear implantation, definitive clinical guidance regarding HA and CI programming for EAS technology is not currently available and the majority of previous studies investigating the effects of LF CI cutoff for EAS outcomes have been mixed. Some studies have demonstrated significant perceptual benefits with some spectral overlap between the HA and CI frequency range in the implanted ear with LF CI cutoff set to the frequency at which thresholds reach approximately 60–70 dB HL (Gifford et al., 2017; Imsiecke et al., 2020; Incerti et al., 2019; Vermeire et al., 2008). Other studies have demonstrated significant perceptual benefits programming the HA and CI frequency range in the CI ear to meet, but not overlap (Karsten et al., 2013; Vermeire et al., 2008); however, the studies demonstrating superiority of “meet” EAS programming have largely focused on patients with shorter electrode arrays ranging from 10 to 21 mm9,14. Most recently, research has shown significant perceptual benefits for EAS-based speech recognition applying a closer tonotopic match for CI simulations (Fu et al., 2017; Willis et al., 2020), SG-place based EAS programming for NH listeners via EAS simulations (Dillon et al., 2021b) and 2 EAS recipients assessed at initial activation (Dillon et al., 2021a). As described in the introduction, the 2 EAS recipients assessed at activation were both implanted with the MED-EL Flex 28 electrode and had nearly identical and deep angular insertion depths (649 and 669°) (Dillon et al., 2021a) resulting in minimal discrepancy between the place-based and default LF CI cutoff for EAS or even for a standard electric-only map (Dillon et al., 2021a).

In the current study, though we observed some variability for speech recognition and subjective listening difficulty across LF CI cutoff frequency, our results largely demonstrated no effect of LF CI cutoff for the EAS (CIHA+HA), bimodal (CI+HA), or CI-alone conditions. Specifically, the current results revealed no trend for increased performance or less subjective listening difficulty as the LF CI cutoff approached the SG-place frequency. Consequently, we could not reject the null hypothesis for this exploratory study.

Our findings were also consistent with a number of previous studies demonstrating 1) significant benefit for both monaural acoustic amplification (CI+HA) and binaural acoustic amplification (CIHA+HA) as compared to CI-alone listening and, 2) significant speech recognition and subjective perceptual benefits for binaural acoustic amplification (CIHA+HA) as compared to the bimodal hearing condition (CI+HA) with monaural acoustic amplification. Also consistent with our previous study (Gifford et al., 2017), the results demonstrated that considerable speech recognition and subject listening difficulty benefits can be observed at the individual level by deviating from the default frequency allocation for the most apical electrode for all 3 listening conditions—though this did not reach statistical significance for the best-aided EAS condition (CIHA+HA). In other words, limiting CI bandwidth for CI-alone and bimodal conditions yielded significant improvements for speech recognition in noise and subjective estimates of listening difficulty, an effect consistent with previous reports using CI simulations with normal-hearing listeners (Fu et al., 2017; Willis et al., 2020) and experienced CI recipients (Gifford et al., 2017; Fowler et al., 2015). This effect may be related to the spectral characteristics of the noise (Gifford et al., 2017), frequency mismatch between clinical defaults for the most apical electrode and the SG-place (Fu et al., 2017; Willis et al., 2020), and/or reduction of interaural spectral overlap for the bimodal hearing condition—though Green and colleagues (Green et al., 2014) demonstrated no speech recognition benefits for restricting electric bandwidth in bimodal listeners. Finally, our current results were consistent with high performance for LF CI cutoffs resulting in considerable frequency mismatch to the underlying SG place frequency, but were similar to that which the 6 chronic EAS users were accustomed—a finding consistent with the results presented by Incerti and colleagues (Incerti et al., 2019) for a group of experienced EAS users.

At first glance, one may conclude that the current results are inconsistent with previous studies showing a significant effect of LF CI cutoff for EAS performance (Gifford et al., 2017) and significant benefit for SG-place-based fittings (Dillon et al., 2021a). However, there are considerable differences between the current study, our accompanying analyses, and previous SG-place-based studies which likely influenced across-study differences. First, with respect to the effect of LF CI cutoff on EAS performance, the current project did not complete statistical analyses examining the effects of discrete LF CI cutoffs given the variability of electrode arrays, acoustic hearing preservation, and resultant LF CI cutoffs assessed for the 15 participants. Rather, our analyses were focused on 1) the deviation of the LF CI cutoff from the underlying SG-place frequency and 2) performance obtained with the standard clinical EAS programming approach for LF CI cutoff and the LF CI cutoff yielding an individual participant’s best outcomes.

Another difference is that 13 of our 15 study participants were experienced CI users having ≥ 1 year of CI use and 6 participants had chronic EAS experience, as indicated in Table 1. Our remaining two study participants (14 and 15) were enrolled 1 month following CI activation and were fitted with EAS in the implanted ear at the study visit and assessed acutely. In contrast, Dillon and colleagues (Dillon et al., 2021a) studied 2 EAS patients at device activation for whom there was no perceptual adaptation to the frequency mismatch between standard EAS frequency allocation and underlying SG place. Thus, it is quite possible that had we enrolled more study participants at or closer to device activation, we may have observed performance and subjective benefits for EAS boundaries approaching SG-place frequency. Relatedly, our assessments were completed acutely and as such, it is possible that had we allowed for an acclimatization period for each LF CI cutoff tested, we may have observed benefit for conditions offering a closer match to underlying SG tonotopicity. However, previous studies including an acclimatization period showed either no effect of LF CI cutoff (Imsiecke et al., 2020; Incerti et al., 2019) or no difference between speech recognition scores across LF CI cutoff obtained acutely and chronically5—though none of these studies employed a SG-place frequency approach to EAS programming.

Finally, another difference between the current study and previous studies of EAS SG-place fittings is that Dillon and colleagues (Dillon et al., 2021a) provided a full and exact SG tonotopic match across all electrodes—including an expanded HF range for basal electrodes. In the current study, not only did we not provide an exact SG match given the lack of angular insertion depth information at commencement of experimentation for 14 of 15 participants, we did not expand the HF range for basal electrodes. Thus, it is important for future investigations of both newly activated and experienced EAS users to include exact SG matches in the apical region as well as clinically available HF allocation consistent with processor microphone specifications and transmission limitations in the basal electrode array.

4.1. Limitations

There are limitations to the current study which should be noted. As indicated in the Discussion, investigation of LF CI cutoff was assessed acutely and 13 of 15 EAS patients had ≥ 1 year of CI experience. With 1 year of CI experience, perceptual adaptation to the standard CI or EAS frequency allocation had presumably taken place (Aronofff et al., 2019; Reiss et al., 2011) which may have introduced participant bias. Additionally, less than half of our sample had chronic EAS experience with the remaining participants having either long-term experience (n = 7) or 1-month CI experience (n = 2) with a full CI bandwidth without acoustic amplification in the implanted ear. Due to the listening experience and presumable perceptual adaptation to the tonotopic mismatch for those with ≥ 1 year of CI use without EAS fitting (Aronofff et al., 2019; Reiss et al., 2011), this likely impacted our results such that acute EAS listeners demonstrated high levels of performance and low reports of subjective listening difficulty associated with LF CI cutoffs representing the largest deviations from the SG-place match (e.g., −3 and −4 oct). Additionally, the LF CI cutoffs tested were different across participants to be most relevant to one’s individual electrode array characteristics and degree of LF acoustic audibility in the implanted ear; however, this introduced variability and not all participants were ultimately assessed at LF CI cutoffs that were at similar deviations from the SG place frequency for the most apical electrode in the array. Finally, though all previous studies examining EAS programming have included fairly small samples ranging from 2 to 15, the current sample size of just 15 participants with variation in implanted electrode arrays and device experience also limits clinical generalizability.

5. Conclusion

There are multiple approaches for programming the LF CI cutoff for EAS/Hybrid CI recipients that may result in similar scores for speech recognition in noise—good news from a clinical CI/HA programming perspective. Consistent with past studies, we found significant benefits for speech recognition and subjective estimates of listening difficulty for both bimodal (CI+HA) and best-aided EAS (CIHA+HA) conditions as compared to CI-alone listening as well as significant additional benefits of binaural acoustic hearing (CIHA+HA) as compared to bimodal hearing with monaural acoustic hearing (CI+HA). Additionally, we found significant differences for both speech recognition and subjective listening difficulty across the clinical default settings (full BW for CI-alone and bimodal, 70-dB-HL frequency for EAS) and the scores obtained with an individual’s best LF CI cutoff. This effect was significant for CI-alone and bimodal listening such that a slightly restricted CI bandwidth afforded significantly better speech recognition in noise and subjective listening difficulty for the CI-alone and bimodal listening conditions. EAS technology is underutilized and CI recipients with acoustic hearing preservation derive significant benefit from EAS fittings even for experienced CI users for whom perceptual adaptation to the CI-mediated frequency mismatch to underlying SG-place frequencies has already occurred. Thus, the primary take-aways from the current study are that 1) patients with acoustic hearing preservation should be fitted with an acoustic component in the implanted ear to achieve highest outcomes for speech recognition in noise and subjective reports of listening difficulty, and 2) the current clinical approach for EAS fittings setting the LF CI cutoff roughly equivalent to the audiometric threshold reaching approximately 70 dB HL yields significant EAS benefit even beyond that offered by bimodal hearing (CI+HA). Though this was not a primary aim of the current study, the current data also suggest that clinicians may also want to investigate clinical effects of slightly increasing LF CI cutoff for bimodal listeners, particularly for noisy listening conditions as that was shown here and in previous studies to yield significant improvement in speech recognition. Finally, further studies are needed investigating the efficacy of a SG-place fitting for large samples of EAS listeners with various electrode arrays at CI activation as well as for experienced EAS listeners following perceptual adaptation as place-based approaches may ultimately yield greater outcomes.

Acknowledgements

This work was supported by National Institutes of Health (NIH) Grant Nos. R01 DC009404, R01 R01 DC014462, R01 R01DC014037, UL1 TR000445, and an investigator initiated grant from Cochlear.

Footnotes

CRediT authorship contribution statement

René Gifford: Conceptualization, Methodology, Investigation, Formal Analysis, Writing—Original Draft, Writing—Review & Edit-ing, Supervision, Project Administration, Funding Acquisition. Linsey Sunderhaus: Conceptualization, Data curation, Investigation, Writing-Original draft. Benoit Dawant: Software, Writing—Original Draft, Visualization, Funding Acquisition. Robert Labadie: Software, Writing—Original Draft, Visualization, Funding Acquisition. Jack Noble: Software, Writing—Original Draft, Visualization, Resources, Formal Analysis, Funding Acquisition.

*

Financial Disclosures: RHG is an audiology clinical advisory board member for Advanced Bionics, Cochlear, and Frequency Therapeutics as well as a consultant for Akouos.

1

Though the 5 study participants had acoustic hearing preservation in the implanted ear, just 2 of the 5 were using bilateral HAs and were thus assessed in the EAS condition; the remaining 3 participants were assessed in the bimodal condition. For the 2 individuals using EAS technology via CI plus bilateral HAs, an in-the-ear HA was fitted at the time of CI activation and all participants had 12 weeks’ listening experience with each condition prior to assessment.

2

For AB recipients programmed in SoundWave 3.2, there are discrete LF CI cutoffs from which to choose including 250 Hz (full CI spectral bandwidth), 350 Hz, 520 Hz, 690 Hz, 850 Hz, 1010 Hz, 1190 Hz, and 1540 Hz. For Cochlear CI recipients programmed in Custom Sound 6.3, the LF CI cutoffs are assigned in 125-Hz increments relative to the default LF CI allocation for the most apical electrode which is 188 Hz; thus, the chosen EAS crossover frequencies were in multiples of 125 from 188 Hz. For MED-EL CI recipients programmed in Maestro 9.0, the LF CI cutoffs were chosen to represent the lowest possible frequency allocation for electrode 1 (70 Hz) and varied in 50-Hz increments between 150 and 500 Hz for a given participant.

References

  1. Aronofff JM, Staislofff HE, Kirchner A, Lee DH, Stalmach J, 2019. Pitch Matching Adapts Even for Bilateral Cochlear Implant Users with Relatively Small Initial Pitch Differences Across the Ears. J. Assoc. Res. Otolaryngol 20 (6), 595–603. doi: 10.1007/s10162-019-00733-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Canfarotta MW, Dillon MT, Buss E, Pillsbury HC, Brown KD, O’Connell BP, 2019. Validating a new tablet-based tool in the determination of cochlear implant angular insertion depth. Otol. Neurotol 40 (8), 1006–1010. doi: 10.1097/MAO.0000000000002296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Cochlear. Nucleus® HybridTM L24 Cochlear Implant CI24REH. 2014. [Google Scholar]
  4. Dillon MT, Canfarotta MW, Buss E, O’Connell BP, 2021a. Comparison of Speech Recognition With an Organ of Corti Versus Spiral Ganglion Frequency-to-Place Function in Place-Based Mapping of Cochlear Implant and Electric-Acoustic Stimulation Devices. Otol. Neurotol 42, 721–725. doi: 10.1097/MAO.000000000003070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Dillon MT, Canfarotta MW, Buss E, Hopfinger J, O’Connell BP, 2021b. Effectiveness of Place-based Mapping in Electric-Acoustic Stimulation Devices. Otol. Neurotol 42, 197–202. doi: 10.1097/MAO.0000000000002965. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Dunn CC, Perreau A, Gantz B, Tyler RS, 2010. Benefits of localization and speech perception with multiple noise sources in listeners with a short-electrode cochlear implant. J. Am. Acad. Audiol 21 (2010), 44–51. doi: 10.3766/jaaa.21.1.6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Fowler JR, Eggleston JL, Reavis KM, McMillan GP, Reiss LAJ, 2015. Effects of Removing Low-Frequency Electric Information on Speech Perception with Bimodal Hearing. J. Speech Lang. Hear. Res doi: 10.1044/2015_JSLHR-H-15-0247, Published online:epub ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Fraysse B, Macías AR, Sterkers O, Burdo S, Ramsden R, Deguine O, Klenzner T, Lenarz T, Manrique Rodriguez M, von Wallenberg E, James C, 2006. Residual hearing conservation and electroacoustic stimulation with the nucleus 24 contour advance cochlear implant. Otol. Neurotol 27 (5), 624–633. doi: 10.1097/01.mao.0000226289.04048.0f. [DOI] [PubMed] [Google Scholar]
  9. Fu QJ, Galvin JJ, Wang X Integration of acoustic and electric hearing is better in the same ear than across ears. Scientific reports. 2017;7(1):12500. doi: 10.1038/s41598-017-12298-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Gantz BJ, Dunn CC, Oleson J, Hansen MR, 2018. Acoustic plus electric speech processing: Long-term results. Laryngoscope 128, 473–481. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gifford RHH, Davis TJJ, Sunderhaus LWW, et al. , 2017. Combined electric and acoustic stimulation (EAS) with hearing preservation: effect of cochlear implant low-frequency cutoff on speech understanding and perceived listening difficulty. Ear Hear. 38 (5), 539–553. doi: 10.1097/AUD.0000000000000418. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gifford RH, Dorman MF, Skarzynski H, et al. , 2013. Cochlear implantation with hearing preservation yields significant benefit for speech recognition in complex listening environments. Ear Hear. 34 (4). doi: 10.1097/AUD.0b013e31827e8163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gifford RH, Grantham DW, Sheffield SW, Davis TJ, Dwyer R, Dorman MF, 2014. Localization and interaural time difference (ITD) thresholds for cochlear implant recipients with preserved acoustic hearing in the implanted ear. Hear. Res 312. doi: 10.1016/j.heares.2014.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gifford RH, Stecker GC, 2020. Binaural Cue Sensitivity in Cochlear Implant Recipients With Acoustic Hearing Preservation. Hear. Res 390, 107929. doi: 10.1016/j.heares.2020.107929. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gravel JS, White KR, Johnson JL, et al. , 2005. A multisite study to examine the efficacy of the otoacoustic emission/automated auditory brainstem response newborn hearing screening protocol: Recommendations for policy, practice, and research. Am. J. Audiol 14 (2), S217–S228. [DOI] [PubMed] [Google Scholar]
  16. Green T, Faulkner A, Rosen S, 2014. Overlapping frequency coverage and simulated spatial cue effects on bimodal (electrical and acoustical) sentence recognition in noise. J. Acoust. Soc. Am 135 (2), 851–861. [DOI] [PubMed] [Google Scholar]
  17. Gstoettner WK, Van de Heyning P, Fitzgerald O’Connor A, et al. , 2008. Electric acoustic stimulation of the auditory system: results of a multi-centre investigation. Acta Otolaryngol. 128 (9), 968–975. [DOI] [PubMed] [Google Scholar]
  18. Imsiecke M, Kruger B, Buchner A, Lenarz T, Nogueira W, 2020. Interaction Between Electric and Acoustic Stimulation Influences Speech Perception in Ipsilateral EAS Users. Ear Hear. 41 (4), 868–882. doi: 10.1097/AUD.0000000000000807. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Incerti PV, Ching TYC, Cowan R, 2019. The effect of cross-over frequency on binaural hearing performance of adults using electric-acoustic stimulation. Cochlear Implants International 20 (4), 190–206. doi: 10.1080/14670100.2019.1590499. [DOI] [PubMed] [Google Scholar]
  20. Jiam NT, Gilbert M, Cooke D, Jiradejvong D, Barrett K, Caldwell M, Limb CJ, 2019. Association Between Flat-Panel Computed Tomographic Imaging-Guided Place-Pitch Mapping and Speech and Pitch Perception in Cochlear Implant Users. JAMA Otolaryngology - Head and Neck Surgery 145 (2), 109–116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Karsten SA, Turner CW, Brown CJ, Jeon EK, Abbas PJ, Gantz BJ, 2013. Optimizing the combination of acoustic and electric hearing in the implanted ear. Ear Hear. 34 (2), 142–150. doi: 10.1097/AUD.0b013e318269ce87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Keidser G, Dillon H, Flax M, Ching T, Brewer S, 2011. The NAL-NL2 prescription procedure. Audiol. Res 1 (1), e24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kiefer J, Pok M, Adunka O, Sturzebecher E, Baumgartner W, Schmidt M, Tillein J, Ye Q, Gstoettner W, 2005. Combined electric and acoustic stimulation of the auditory system: results of a clinical study. Audiology and Neuro-Otology 10 (3), 134–144. doi: 10.1159/000084023. [DOI] [PubMed] [Google Scholar]
  24. MED-EL. User Manual for SONNET (Me1310) and SONNET EAS (Me1320) Audio Processors. 2016. [Google Scholar]
  25. Noble JH, Gifford RH, Labadie RF, Dawant BM, 2012. Statistical shape model segmentation and frequency mapping of cochlear implant stimulation targets in CT. Medical image computing and computer-assisted intervention : MICCAI. International Conference on Medical Image Computing and Computer-Assisted Intervention, 15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Noble JH, Labadie RF, Gifford RH, Dawant BM, 2013. Image-Guidance enables new methods for customizing cochlear implant stimulation strategies. IEEE Trans. Neural Syst. Rehabil. Eng 21 (5). doi: 10.1109/TNSRE.2013.2253333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Perkins E, Lee J, Manzoor N, et al. , 2021. The Reality of Hearing Preservation in Cochlear Implantation: Who Is Utilizing EAS? Otol. Neurotol 42 (6). doi: 10.1097/MAO.0000000000003074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Pillsbury HC, Dillon MT, Buchman CA, et al. , 2018. Multicenter US Clinical Trial With an Electric-Acoustic Stimulation (EAS) System in Adults: Final Outcomes. Otol. Neurotol 39, 299–305. doi: 10.1097/MAO.0000000000001691. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Plant K, Babic L, 2016. Utility of bilateral acoustic hearing in combination with electrical stimulation provided by the cochlear implant. Int. J. Audiol 2027 (March), 1–8. doi: 10.3109/14992027.2016.1150609. [DOI] [PubMed] [Google Scholar]
  30. Rader T, Fastl H, Baumann U, 2013. Speech perception with combined electric-acoustic stimulation and bilateral cochlear implants in a multisource noise field. Ear & Hearing 34 (3), 324–332. doi: 10.1097/AUD.0b013e318272f189. [DOI] [PubMed] [Google Scholar]
  31. Reiss LAJ, Lowder MW, Karsten SA, Turner CW, Gantz BJ, 2011. Effects of extreme tonotopic mismatches between bilateral cochlear implants on electric pitch perception: a case study. Ear Hear. 32 (4), 536–540. doi: 10.1097/AUD.0b013e31820c81b0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Simpson A, McDermott HJ, Dowell RC, Sucher C, Briggs RJS, 2009. Comparison of two frequency-to-electrode maps for acoustic-electric stimulation. Int. J. Audiol 48 (2), 63–73. doi: 10.1080/14992020802452184. [DOI] [PubMed] [Google Scholar]
  33. Spahr AJ, Dorman MF, Litvak LM, et al. , 2012. Development and validation of the azbio sentence lists. Ear Hear. 33 (1), 112. doi: 10.1097/AUD.0b013e31822c2549. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Spitzer ER, Waltzman SB, Landsberger DM, Friedman DR, 2020. Acceptance and Benefits of Electro-Acoustic Stimulation for Conventional-Length Electrode Arrays. Audiology and Neuro-Otology 26 (1), 17–26. doi: 10.1159/000507975. [DOI] [PubMed] [Google Scholar]
  35. Stakhovskaya O, Sridhar D, Bonham BH, Leake PA, 2007. Frequency map for the human cochlear spiral ganglion: implications for cochlear implants. J. Assoc. Res. Otolaryngol 8 (2), 220–233. doi: 10.1007/s10162-007-0076-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Vermeire K, Anderson I, Flynn M, 2008. Van de Heyning, P. The influence of different speech processor and hearing aid settings on speech perception outcomes in electric acoustic stimulation patients. Ear Hear. 29 (1), 76–88. doi: 10.1097/AUD.0b013e31815d6326. [DOI] [PubMed] [Google Scholar]
  37. Willis S, Moore BCJ, Galvin JJ, Fu QJ, 2020. Effects of noise on integration of acoustic and electric hearing within and across ears. PLoS One 15 (10), e0240752. doi: 10.1371/journal.pone.0240752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Zhao Y, Chakvorti S, Labadie RF, Dawant BM, Noble JH, 2019. Automatic graph-based method for localization of cochlear implant electrode arrays in clinical CT with sub-voxel accuracy. Med. Image Anal 52, 1–12. doi: 10.1016/j.media.2018.11.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Zhao Y, Dawant BM, Labadie RF, Noble JH, 2018. Automatic localization of closely spaced cochlear implant electrode arrays in clinical CTs. Med. Phys 45 (11), 5030–5040. doi: 10.1002/mp.13185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Zhao Y, Labadie RF, Dawant BM, Noble JH, 2018. Validation of automatic cochlear implant electrode localization techniques using μCTs. J. Med. Imag 5 (3), 035001. doi: 10.1117/1.JMI.5.3.035001. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES