Influence of factors related to electrode array placement on the early speech recognition of cochlear implant recipients with and without functional hearing preservation

Cochlear implant (CI) and electric-acoustic stimulation (EAS) device users vary in their early speech recognition. This variability may be due in part to factors related to the electrode array placement, including frequency-to-place mismatch and the array's proximity to the functional acoustic hearing region. The present study reviewed the association of factors related to electrode array placement and early speech recognition in quiet and noise in CI and EAS users. The sample included 34 adults with preoperative thresholds at 250 Hz of ≤ 70 dB hearing level (HL) who were implanted with a 23-mm straight electrode array. For those who lost functional acoustic hearing and listened with a CI device (n = 15), word recognition in quiet was significantly associated with frequency-to-place mismatch at 1.5 kHz, r (15) = 0.61, p = 0.015. For EAS users (n = 18), 14 had their electrode arrays placed outside the functional acoustic hearing region, and 4 had electrodes in the functional acoustic hearing region. For EAS users with the array outside the functional acoustic hearing region, speech recognition in noise was not significantly associated with frequency-to-place mismatch, r (9) = 0.33, p = 0.389. Performance in noise was significantly associated with proximity, r (9) = 0.96, p < 0.001, with poorer scores for EAS users whose electrode arrays were placed farther from the functional acoustic hearing region. These preliminary data suggest that factors related to electrode array placement may contribute to the variability in early outcomes in CI and EAS users and may be important to consider in device selection and mapping to help individuals achieve their best outcomes.

For cochlear implant (CI) candidates with functional low- to mid-frequency hearing in the ear to be implanted, short (e.g., 23-mm) straight electrode arrays are typically recommended to limit cochlear trauma and increase the likelihood of postoperative hearing preservation. When functional acoustic hearing is preserved, CI recipients are fit with an electric-acoustic stimulation (EAS) device that provides amplification of low frequencies and electric stimulation of high frequencies. If functional acoustic hearing is lost, CI recipients are fit with a CI-alone device that electrically stimulates the device's full frequency range. In both scenarios, speech recognition in most CI recipients is significantly improved over preoperative performance with amplification (Adunka et al., 2013; Balkany et al., 2006; Gantz et al., 2016; Helbig et al., 2016; Kiefer et al., 2005; Pillsbury et al., 2018; Roland et al., 2018; Sargsyan et al., 2021; Thompson et al., 2024), although outcomes vary widely across individuals. The variability in speech recognition in CI and EAS users may be due in part to factors related to the electrode array placement, including frequency-to-place mismatch and the proximity of the electrode array to the functional acoustic hearing region.

Frequency-to-place mismatches are discrepancies in the frequency information extracted for an individual channel and the cochlear place frequencies stimulated by the electrode contact. Frequency-to-place mismatches are prevalent in CI and EAS users because the placement (i.e., angular insertion depth [AID]) of an individual's electrode array is not accounted for in the default assignment of the filter frequencies (Canfarotta et al., 2020a). The AID of electrode arrays vary widely across CI recipients due to differences in electrode array design, surgical approach, and the individual's cochlear anatomy (Canfarotta et al., 2020a; Landsberger et al., 2015). For CI users who were conventional CI candidates (e.g., preoperative severe-to-profound hearing loss), some samples have shown significantly poorer early speech recognition as the magnitude of frequency-to-place mismatch increases (Canfarotta et al., 2020a, 2022; DeFreese et al., 2025; Dessard et al., 2024; Lee et al., 2025; Mertens et al., 2022; O'Rourke et al., 2026), although others have observed no significant relationship (e.g., straight array recipients in DeFreese et al., 2025). For CI users who lost functional acoustic low-frequency hearing, there are reports of poorer outcomes for recipients of shorter electrode arrays compared to those with longer arrays (Büchner et al., 2017; Thompson et al., 2024), suggesting that they also experience negative effects of frequency-to-place mismatches. Data on the influence of frequency-to-place mismatches on early outcomes for EAS users are relatively limited, although preliminary data suggest that EAS users with larger mismatches experience less improvement in speech recognition than individuals with smaller mismatches (Dillon et al., 2023). There is a need to expand the understanding of the potential influence of frequency-to-place mismatches on post-activation outcomes for hearing-preservation CI candidates.

For EAS users, another factor specific to electrode array placement that may influence speech recognition is the proximity of the array relative to the functional acoustic hearing region (Nix et al., 2023). Proximity values characterize the AID of the electrode array relative to the functional acoustic hearing region for an individual. One of three scenarios occurs: (1) The electrode array is placed outside of the region of functional acoustic hearing, potentially creating a stimulation gap in the cochlea; (2) the electrode array is placed next to the region of functional acoustic hearing; or (3) one or more electrodes are placed within the functional acoustic hearing region. Nix et al. (2023) reviewed the placement of straight electrode arrays that were 24- (n = 41), 28- (n = 43), and 31.5-mm (n = 21) in length relative to the region of functional hearing and found that 60% of cases had one or more electrode contacts within the functional acoustic hearing region. For EAS users, better sentence recognition in noise was observed when the electrode array was placed close to the functional acoustic hearing region. Speech recognition declined when the proximity was farther away (a larger gap between the AID of the most apical electrode and the functional hearing region) or deeper within the functional acoustic region (overlap). There is a need to understand whether proximity of the electrode array to the functional acoustic hearing region has a significant relationship with speech recognition for other samples of EAS users and/or other devices.

The present study reviewed the association of factors related to electrode array placement (i.e., frequency-to-place mismatches and proximity of the array to the functional acoustic hearing region) with early speech recognition in CI and EAS users who were hearing-preservation CI candidates. The sample had been implanted with a short (23 mm) straight electrode array. It was hypothesized that most recipients would have their electrode array placed close to or outside the functional acoustic hearing region. This exploratory study aimed to review the potential relationship between factors related to electrode array placement and early speech recognition to understand whether these factors should be considered in device selection and mapping for CI and EAS users to support better outcomes.

A retrospective review was conducted of adult recipients of the Advanced Bionics Corporation (Valencia, CA) Slim J version 2 electrode array (IRB: 09-2328). Inclusion criteria were hearing-preservation CI candidates (i.e., ≤ 70 dB HL at 250 Hz) who underwent cochlear implantation between March 2020 and July 2025 and who had an intraoperative X-ray sufficient to determine the AID for the electrode contacts. The ≤ 70 dB HL criterion was used because the reviewed EAS system uses 70 dB HL to determine the crossover between acoustically vs. electrically represented frequency information (Wolfe and Schafer, 2025). Individuals were excluded if they had evidence of retrocochlear pathology (e.g., vestibular schwannoma), had undergone revision surgery, were non-English speakers (test materials presented in English), or did not complete speech recognition assessment by 3 months post-activation.

Hearing-preservation surgical procedures at the study center were a standard posterior tympanotomy approach, the administration of intravenous dexamethasone (10 mg) and topical dexamethasone (10 mg/mL) before opening the round window, and a round window insertion of the electrode array. CI recipients were also prescribed a postoperative steroid course. During the reviewed period, the use of a robotic-assisted insertion device (iotaSOFT, iotaMotion, Inc., St. Paul, MN) was initiated at the study center. For Advanced Bionics Corporation cases, the Slim J electrode array was the preferred device for hearing-preservation cases. The Slim J electrode array is a 16-channel, 23-mm (active length: 20 mm; contact spacing: 1.3 mm) straight electrode array.

Unaided hearing thresholds from 125 to 8,000 Hz were measured behaviorally, with the patient seated in a single-walled sound booth. Pulsed pure tones were presented via insert earphones, with masking presented to the contralateral ear when warranted. A value of 120 dB HL was entered when there was no response to the stimulus at the audiometer's output limit.

Activation of a CI or EAS device occurred approximately 2–4 weeks postoperatively. The criterion at the study center for fitting an Advanced Bionics Corporation EAS device was an unaided threshold in the implanted ear of ≤ 70 dB HL at 250 Hz; individuals who exceeded this criterion were typically fit with a CI device (electric stimulation only). For CI and EAS devices, the most comfortable loudness levels were assigned using either behavioral measures or electrically evoked stapedius reflex thresholds. For CI devices, the extended low-filter frequency assignments (low-filter frequency: 250 Hz) were mapped at activation. For EAS devices, the unaided thresholds were entered into the clinical mapping software, which defines the electric low-filter frequency as the acoustic frequency that exceeds 70 dB HL. This EAS system's acoustic earhook includes a receiver in the canal, which was coupled with a dome or custom earmold. The acoustic output was fit to NAL-NL1 prescriptive targets (Byrne et al., 2001). At post-activation visits, the clinical protocol was to measure unaided thresholds for all CI recipients who were hearing-preservation candidates to evaluate improvements/decrements in thresholds compared to the activation visit (Dillon et al., 2020). EAS users who experienced a post-activation decline in thresholds at 250 Hz yet maintained aidable (e.g., ≤ 70 dB HL) hearing at 125 Hz typically continued to use their EAS device.

Aided speech recognition generally followed the procedures described in the Minimum Speech Test Battery – Version 3 (Dunn et al., 2024). Briefly, recorded materials were presented at 60 dB SPL with the patient seated approximately 1 meter from the loudspeaker (0° azimuth) in a single-walled sound booth. Patients were instructed to repeat each target word or sentence and encouraged to guess if unsure. Speech recognition assessment included one 50-word consonant-nucleus-consonant (CNC) list (Peterson and Lehiste, 1962) in quiet. Most patients were also assessed in noise, using the AzBio sentences (Spahr et al., 2012) in a 10-talker masker at 10 dB signal-to-noise ratio (SNR). Performance for both measures was scored as the percentage of words correctly repeated. The recommended post-activation visits at the study center were 1 and 3 months and then annually. The present study reviewed data from the 3-month post-activation visit because Berg et al. (2023) reported no significant differences in word recognition for CI users between the 3- and 6-month and between the 6- and 12-month post-activation visits.

Electrode array placement

Electrode array placement was determined from each individual's intraoperative X-ray image, using previously described procedures (Giardina et al., 2020). At the study center, the CI recipient's head is typically positioned around 30° to achieve a cochlear view using the X-ray image. The AID was measured for each electrode contact using the helical scala tympani model (see Giardina et al., 2020).

Frequency-to-place mismatch was calculated for the electrode contact at the 1.5-kHz cochlear region, with the rationale that mid-frequency information is critical for speech recognition (ANSI/ASA S3.5-1997, 1997; Bosen and Chatterjee, 2016; Warren et al., 1995). The Target CI clinical programming software (version 1.5.11) lists the low-frequency filter between the 16 adjacent electrode contacts. Frequency-to-place mismatch was calculated as the semitone deviation between the channel's low-frequency filter and the cochlear place frequency for the associated electrode at 1.5 kHz. Negative values indicated a basal shift in frequency information relative to cochlear tonotopy, whereas positive values indicated an apical shift.

For EAS users, electrode array placement was also used to calculate the proximity of the electrode array relative to the functional acoustic hearing region using previously described procedures (Nix et al., 2023; Thompson et al., 2026). For the present study, functional acoustic hearing was defined as ≤ 70 dB HL because this is the fitting criterion for the reviewed EAS system. Proximity of the electrode array relative to the functional acoustic hearing region was calculated as the difference in the AID of the most apical electrode contact and the AID that the individual's unaided threshold was at 70 dB HL. Negative proximity values indicated that the electrode array was placed basal to the acoustic hearing region; positive values indicated one or more electrode contacts were placed within the acoustic hearing region.

Data analyses

The associations of factors related to electrode array placement (i.e., frequency-to-place mismatch and proximity to the functional acoustic hearing region) and speech recognition were analyzed using two-tailed Pearson correlations in SPSS (version 29). Speech recognition scores were converted from percent correct to rationalized arcsine units (RAU; Studebaker, 1985) for the analyses to increase the numerical distance between scores that were at or near floor or ceiling performance levels. Uncorrected p -values were reported due to the present study's exploratory nature.

Thirty-four CI cases (24 female, 19 right ear) met the criteria for study inclusion. Age at surgery ranged from 30 to 84 years (mean: 67 years, SD: 13 years). A robotic-assisted insertion was used for 3 cases. The AID of the most apical electrode contact ranged from 279° to 512° (mean: 434°, SD: 52°). There were four cases of partial electrode array insertions (3 manual insertions; 1 robotic-assisted insertion). For the 30 cases of full insertions, the AID of the most apical electrode contact ranged from 354° to 512° (mean: 447°, SD: 38°).

Figure 1 plots the individual unaided thresholds and the sample mean values at the preoperative (left panel) and 3-month post-activation (right panel) visits. The unaided thresholds for the three cases that had robotic-assisted insertions are indicated in orange; thresholds for the four cases of partial insertion are indicated with dashed lines.

CI users

Fifteen patients had an unaided threshold >70 dB HL at 250 Hz and listened with a CI device. For this subgroup, mean age at surgery was 67 years (range: 34–84 years, SD: 14 years), and the mean AID of the most apical electrode contact was 445° (range: 332°-512°, SD: 52°). In two cases, a partial insertion was observed on the intraoperative X-ray. Six patients had one or more basal electrodes deactivated (three with E16 and three with E15 and E16). Frequency-to-place mismatch at 1.5 kHz ranged from −7 to −18 semitones (mean: −10 semitones, SD: 3 semitones). A mismatch of 12 semitones is a one-octave shift between the presented frequency information and the cochlear place frequency.

Figure 2 plots the speech recognition scores (in RAU) for CNC words (top row) and AzBio sentences in noise (bottom row) as a function of frequency-to-place mismatch at 3 months post-activation for the CI users. The filled symbols indicate the results for the CI users with electrodes deactivated. Word recognition in quiet was significantly poorer with larger magnitudes of frequency-to-place mismatch, r (15) = 0.61, p = 0.015. Three patients did not complete AzBio sentence recognition in noise. For the remaining 12 patients, sentence recognition in noise was not significantly correlated with frequency-to-place mismatch at 3 months post-activation, r (12) = 0.39, p = 0.212.

EAS users

Eighteen patients (1 bilateral, 19 ears) listened with an EAS device(s). For this subgroup, mean age at surgery was 67 years (range: 30–84 years, SD: 13 years), and the mean AID of the most apical electrode contact was 425° (range: 279°–482°, SD: 52°). For two cases, a partial insertion was observed on the intraoperative X-ray. Seven patients had one or more basal electrodes deactivated: three with E16, three with E15 and E16, and one with E13–E16. For the latter case, the four electrodes were deactivated due to electrode array extrusion. This patient was excluded from further analyses. Of the 18 ears, 14 had a threshold ≤ 70 dB HL at 250 Hz. Four EAS users had hearing at 250 Hz, which was >70 dB HL at 3 months post-activation; their thresholds at 125 Hz were 35, 50, 55, and 55 dB HL.

For most ears (n = 14; 78%), the electrode array was placed outside the functional acoustic hearing region. Four ears (22%) had one or more contacts within the functional acoustic hearing region. The proximity of the most apical electrode contact to the functional acoustic hearing region ranged from −218° to 44° (mean: −80°, SD: 74°). Frequency-to-place mismatch at 1.5 kHz ranged from −1 to −15 semitones (mean: −9 semitones, SD: 4 semitones).

Figure 3 plots the speech recognition scores for CNC words (top row) and AzBio sentences in noise (bottom row) as a function of proximity to the functional acoustic hearing region (left panel) and frequency-to-place mismatch (right panel) for the EAS users. Cases with the electrode array placed outside the functional acoustic region are indicated with circles; cases with one or more contacts within the functional acoustic hearing region are indicated with squares. Data from the four cases with electrodes within the functional acoustic hearing region are included in the plot but not in the analyses due to the small sample size and the likelihood of an inverse relationship on performance (see Nix et al., 2023). Word recognition in quiet was not significantly associated with proximity, r (14) = 0.03, p = 0.925, or frequency-to-place mismatch, r (14) = 0.04, p = 0.891. Five patients did not complete AzBio sentence recognition in noise. For the remaining nine cases, AzBio sentence recognition in noise was significantly correlated with proximity, r (9) = 0.96, p < 0.001, but not frequency-to-place mismatch, r (9) = 0.33, p = 0.389. The pattern of statistical significance with proximity remained unchanged during leave-one-out validation (correlation coefficient range: 0.94–0.98, all p < 0.001).

The variability in speech recognition across CI and EAS users may be due in part to factors related to the placement of an individual's electrode array. The present study reviewed the association between early speech recognition and two electrode array placement factors: frequency-to-place mismatch and proximity to the functional acoustic hearing region. For CI users, frequency-to-place mismatch was significantly associated with word recognition in quiet, although not with sentence recognition in noise. For EAS users, proximity to the functional acoustic hearing region was significantly associated with sentence recognition in noise, although not for word recognition in quiet; frequency-to-place mismatch was not significantly associated with early performance on either measure. Taken together, these exploratory data suggest that factors related to electrode array placement may contribute to the variability in early outcomes in CI and EAS users and may be important to consider in device selection and mapping to help individuals achieve their best outcomes.

The finding that speech recognition in CI users may be associated with frequency-to-place mismatches corroborates the results from other studies of CI users (Büchner et al., 2017; Canfarotta et al., 2020b; DeFreese et al., 2025; Dessard et al., 2024; Dillon et al., 2025; Lee et al., 2025; Mertens et al., 2022; O'Rourke et al., 2026; Ramos et al., 2026; Thompson et al., 2024; Walia et al., 2024). For example, Büchner et al. (2017) reviewed early speech recognition in recipients of 20-, 24-, and 28-mm straight electrode arrays who lost functional acoustic hearing and reported better outcomes for the CI users with the longer electrode array. The assumption was that CI users with the longer array had better speech recognition due to less frequency-to-place mismatch than CI users with shorter arrays. With the availability of imaging to calculate AID and frequency-to-place mismatch, others have reported significant correlations between frequency-to-place mismatches and early speech recognition in recipients of straight (Canfarotta et al., 2020b; Lee et al., 2025; Mertens et al., 2022; O'Rourke et al., 2026) and pre-curved (DeFreese et al., 2025; Ramos et al., 2026) electrode arrays. Taken together, this indicates that the issue of frequency-to-place mismatches on CI user outcomes is universal across electrode array designs and CI manufacturers. These findings have clinical implications for hearing-preservation CI candidates, who are likely to receive a shorter electrode array at a shallow AID in an attempt to preserve functional low-frequency hearing. There is a need to understand the long-term (e.g., >12 months post-activation) influence of frequency-to-place mismatches on outcomes for CI users, as some have reported a lack of significant association at later post-activation visits (e.g., Mertens et al., 2022; Sahara et al., 2026), and the effectiveness of CI mapping procedures aimed at minimizing mismatches.

For the EAS users in this sample, the significant association between proximity to the functional acoustic hearing region and early speech recognition corroborates data from another sample of EAS users Nix et al., (2023). Specifically, Nix et al. (2023) reviewed the speech recognition of EAS users who were recipients of a 24-, 28-, or 31.5-mm straight electrode array. For that sample, 60% (n = 63) had one or more electrode contacts within the functional acoustic hearing region. The difference in percentage of electrodes within the functional acoustic hearing region (i.e., 60% vs. 22% for Nix et al. and the present study, respectively) is expected because the samples are similar in acoustic hearing (i.e., mean threshold at 250 Hz was 50 dB HL [range: 0–80 dB HL] and 58 dB HL [range: 20–90 dB HL] for Nix et al. and the present study, respectively), and the present report reviewed recipients of a shorter electrode array (i.e., 23 mm) than reviewed in the Nix et al. study. Both studies found significantly poorer speech recognition in noise, but not quiet, for EAS users with an electrode array placed farther from the functional acoustic hearing region. Nix et al. postulated that this relationship may be due to a stimulation gap created in the cochlea when the electrode array is placed farther from the region of functional acoustic hearing. Clinically, stimulation gaps (area of the cochlea that is not stimulated) are not accounted for in EAS mapping. The current clinical convention for EAS mapping is to limit spectral gaps (frequency output of the acoustic and CI components) by dividing or overlapping the frequency information presented by the acoustic and CI components. This practice is based on evidence of poorer performance with spectral gaps (Gifford et al., 2017; Karsten et al., 2013); however, there are simulation studies suggesting that spectral gaps may not significantly impact performance when the CI filter frequencies match cochlear tonotopicity (Dillon et al., 2021; Fu et al., 2017; Willis et al., 2020). While the present data cannot elucidate the mechanism behind the relationship between proximity to the functional acoustic hearing region and speech recognition outcomes, they extend our understanding of how factors related to electrode array placement influence the early outcomes in challenging listening conditions in EAS users. These preliminary data suggest that proximity to the functional acoustic hearing region may be a worthwhile variable to consider for electrode array selection and EAS programming.

In this sample, EAS users' speech recognition was not significantly associated with frequency-to-place mismatch, which contradicts the findings from studies using EAS simulations and EAS users (Dillon et al., 2023; Fu et al., 2017; Willis et al., 2020). For example, Fu et al. (2017) assessed speech recognition in noise in participants with normal hearing while listening to simulations of EAS with different magnitudes of frequency-to-place mismatch. Performance was best when frequency-to-place mismatches were small. In EAS users, Dillon et al. (2023) reported a significant interaction between post-activation visit (1, 3, and 6 months) and frequency-to-place mismatch at 1.5 kHz for CNC word recognition, with EAS users with larger mismatches experiencing less improvement over time than EAS users with smaller mismatches. The reason for the discrepancy between the present data and previous studies is unclear. The differences could be due to the small sample size and/or retrospective design of the present study. It may also be that electrode array proximity has a larger influence on early speech recognition than frequency-to-place mismatches for EAS users with electrode arrays placed outside the functional acoustic hearing region. Prospective, long-term investigations with larger samples are needed to disentangle the potential effects of factors related to electrode array placement in EAS users both acutely and over time.

The present preliminary data suggest that variability in early speech recognition outcomes in CI and EAS users may be due in part to factors related to electrode array placement; however, this study has limitations that warrant consideration. Because this was a retrospective study design that used clinical data, it is limited to data from those patients who returned for follow-up and completed speech recognition assessment. Patients with poorer performance in quiet are typically not assessed in noise, reducing the available data for that condition. Furthermore, the sample size was small (e.g., 9 EAS users with speech recognition data in noise), limiting our ability to explore the mechanisms behind the present findings. Additionally, only four EAS users had one or more contacts in the functional acoustic hearing region; thus, the potential relationship of electric-on-acoustic masking could not be explored (see Imsiecke et al., 2020; Koka and Litvak, 2017; Krüger et al., 2017; Lin et al., 2011). Also, uncorrected p -values were reported due to the study's exploratory nature. There is a risk of Type 1 error due to the multiple comparisons. The aim was to explore potential relationships to understand whether these factors should be considered in device selection and mapping for CI and EAS users to support better outcomes. Prospective investigations using larger samples are needed to better understand these preliminary findings.

The present study's methodology also has limitations that warrant consideration. Data were obtained within the early months of device use. The recommended clinical follow-up at the study center had been revised to 1, 3, and 12 months post-activation (removed 6 months), with the rationale that performance for CI users does not change significantly between 3 and 6 months post-activation (Berg et al., 2023). The timeline for a plateau in speech recognition may be different for EAS users than for CI users. Long-term data are needed to understand the effects of listening experience in these patient populations. Another limitation is that some CI and EAS users had electrodes deactivated that were not observed with the intraoperative X-ray to be extracochlear. This may be due to measurement error with an intraoperative X-ray; that is, electrodes observed to be within the first 10° of the cochlea may be extracochlear. It may also be that arrays extruded slightly postoperatively. The present report did not include postoperative imaging to review placement. The electrodes' deactivation, however, did not appear to drive the significant relationships found in this study (see the filled symbols in Figures 2 and 3); that is, the reduced number of active channels does not appear to be the primary reason for poorer outcomes. Also, frequency-to-place mismatch was measured for the electrode contact at the 1.5-kHz region. While this allowed for a direct comparison to the effects of frequency-to-place mismatch in other samples, there is a need to understand whether the potential effect of frequency-to-place mismatch varies across frequency for CI and EAS users and how available acoustic cues may influence this.

Factors related to electrode array placement may have significantly influenced the early outcomes in CI and EAS users implanted with a short (23-mm) straight electrode array. Investigations are ongoing regarding the effectiveness of CI and EAS mapping procedures that account for an individual's electrode array placement in the filter frequency assignments and maximum stimulation levels for specific channels/electrodes.