Article
Reference
Speech perception with novel stimulation strategies for combined cochleo-vestibular systems
LANTHALER, David, et al.
Abstract
Cochlear implants are very well established in the rehabilitation of hearing loss and are regarded as the most successful neuroprostheses to date. While a lot of progress has also been made in the neighboring field of specific vestibular implants, some diseases affect the entire inner ear, leading to both hearing and vestibular hypo-or dysfunction. The proximity of the cochlear and vestibular organs suggests a single combined implant as a means to alleviate the associated impairments. While both organs can be stimulated in a similar way with electric pulses applied through implanted electrodes, the typical phase durations needed in the vestibular system seem to be substantially larger than those typically needed in the cochlear system. Therefore, when using sequential stimulation in a combined implant, the pulse stream to the cochlea is interrupted by comparatively large gaps in which vestibular stimulation can occur. We investigate the impact of these gaps in the auditory stream on speech perception. Specifically, we compare a number of stimulation strategies with different gap lengths and distributions and evaluate [...]
LANTHALER, David, et al . Speech perception with novel stimulation strategies for combined cochleo-vestibular systems. IEEE Transactions on Neural Systems and Rehabilitation Engineering , 2021, vol. 29, p. 1644-1650
DOI : 10.1109/TNSRE.2021.3105271 PMID : 34398757
Available at:
http://archive-ouverte.unige.ch/unige:155364
Disclaimer: layout of this document may differ from the published version.
1 / 1
Speech Perception With Novel Stimulation Strategies for CombinedCochleo-Vestibular
Systems
David Lanthaler , Andreas Griessner, Viktor Steixner , Patrick P. Hübner , Maurizio Ranieri, Samuel Cavuscens, Anissa Boutabla , Clemens M. Zierhofer , and Angélica Pérez Fornos
Abstract —Cochlear implants are very well established in the rehabilitation of hearing loss and are regarded as the most successful neuroprostheses to date. While a lot of progress has also been made in the neighboring field of specific vestibular implants, some diseases affect the entire inner ear, leading to both hearing and vestibular hypo- or dysfunction. The proximity of the cochlear and vestibular organs suggests a single combined implant as a means to alleviate the associated impairments. While both organs can be stimulated in a similar way with electric pulses applied through implanted electrodes, the typical phase durations needed in the vestibular system seem to be substantially larger than those typically needed in the cochlear system.
Therefore, when using sequential stimulation in a combined implant, the pulse stream to the cochlea is interrupted by comparatively large gaps in which vestibular stimulation can occur. We investigate the impact of these gaps in the auditory stream on speech perception. Specifically, we compare a number of stimulation strategies with different gap lengths and distributions and evaluate whether it is feasible to use them without having a noticeable decline in perception and quality of speech. This is a prerequisite for any practicable stimulation strategy of a combined system and can be investigated even in recipients of a normal cochlear implant. Our results show that there is no signif- icant deterioration in speech perception for the different strategies examined in this paper, leaving the strategies as viable candidates for prospective combined cochleo- vestibular implants.
Index Terms—Cochlear implant, cochleo-vestibular implant, speech perception, stimulation strategies, vestibular implant.
Manuscript received December 30, 2020; revised June 1, 2021 and July 30, 2021; accepted July 31, 2021. Date of publication August 16, 2021; date of current version August 26, 2021.(Corresponding author:
David Lanthaler.)
This work involved human subjects or animals in its research. Approval of all ethical and experimental procedures and protocols was granted by the Cantonal Research Ethics Commission in Geneva under Approval No. CCER 2018-01938.
David Lanthaler, Andreas Griessner, Viktor Steixner, and Clemens M. Zierhofer are with the Department of Mechatronics, University of Innsbruck, 6020 Innsbruck, Austria (e-mail: david.lanthaler@uibk.ac.at).
Patrick P. Hübner is with MED-EL GmbH, 6020 Innsbruck, Austria.
Maurizio Ranieri, Samuel Cavuscens, Anissa Boutabla, and Angélica Pérez Fornos are with the Division of Otorhinolaryngology Head and Neck Surgery, Geneva University Hospitals, University of Geneva, 1211 Geneva, Switzerland.
Digital Object Identifier 10.1109/TNSRE.2021.3105271
I. INTRODUCTION
H
EARING loss is a wide spread condition that dramati- cally affects the quality of life [1]. In 2013, the World Health Organization estimated the number of people with disabling hearing loss globally at 360 million [2]. With the help of a cochlear implant (CI) the sense of hearing can be rehabilitated to a large degree. In a CI, hearing impressions are evoked by applying electrical pulses to stimulate the auditory nerve via implanted electrodes inside the cochlea.Apart from hearing loss, a significant reduction in quality of life can also be caused by impairment or total loss of the vestibular function, the second sensory system located in the inner ear. The vestibular system provides fundamental infor- mation to the sense of balance [3]. Analogous to the cochlear system, the physiological operating principle of the vestibular system is based on hair cells triggering action potentials in nerve fibers. Hence, vestibular loss can be treated with a similar approach as hearing loss by introducing electrodes close to the afferent nerves, for example, into the semicircular canals. First devices for vestibular stimulation are already being developed and implanted [4]–[6]. The sensory organs for hearing and balance are located right next to each other and thus functionality of both can be affected, e.g., by infections or ototoxic medication. Remedying the resulting impairment motivates research of a combined cochleo-vestibular implant.
First steps towards the development of such a system have already been performed [5]–[7]. In these kinds of combined implants both the auditory and the vestibular nerve fibers are stimulated via implanted electrodes in a continuous interleaved paradigm, where each electrode is activated sequentially within each stimulation frame. Typically, vestibular pulses with a length of approximately a few hundred microseconds [7] are distinctly longer than cochlear stimulation pulses, typically shorter than 50µs. Therefore, in a sequential stimulation paradigm where both vestibular and cochlear electrodes are addressed within a stimulation frame, relatively long gaps are inevitably introduced in the auditory stream to accommodate the vestibular pulses. Since the feasibility of a combined implant would depend on speech perception performance not to deteriorate substantially, the possible impact of these
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/
TABLE I SUBJECTDEMOGRAPHICS
auditory gaps needs to be studied. This is possible even in CI users without a vestibular implant.
The aim of the current study is to investigate the perception and quality of speech with auditory stimulation strategies as they could be implemented in prospective combined systems.
To this end, we use six different strategies which are based on a standard continuous interleaved sampling (CIS) [8] paradigm with different lengths and distributions of gaps. For the evalu- ation of the different strategies, first of all, speech perception was evaluated using consonant identification tests. In addition, the response times for this test were evaluated as an indicator of listening effort. Finally, the subjective quality and clarity of speech were assessed using self-reported ratings.
In the methods section, we present the subject data, the used hard- and software as well as the specific testing strategies and procedures. In the next sections, our findings and the statistical evaluation of the obtained data are shown and discussed.
II. METHODS
A. Subjects
A total of eight adult participants were enrolled in the study, but only seven completed the entire testing protocol. One sub- ject dropped out of the study voluntarily. All subjects suffered from bilateral sensorineural deafness of varying etiology and had been implanted with MED-EL devices. The average time of implant use at the time of testing was 8 years (range:
0.9 – 17 years) and the average age at the time of surgery was 52.7 years (range: 2.5 – 68.5 years) (see Table I). Two subjects were recipients of a COMBI 40+ implant, one of a PULSAR100CI implant, two of a CONCERTO implant and two of a SYNCHRONY implant. All subjects had at least nine active electrodes in their standard clinical fitting. The demographics of all subjects who completed the study are shown inTable I.
B. Stimulation Hardware and Software
All tests were performed with an OPUS 2 audio test processor. Programming and configuring the audio processor was done with a Personal Computer and the MED-EL devices Diagnostic Interface Box II (DIB II) and MAX programming interface. The clinical fitting software MAESTRO 8 and MATLAB (R2017b The MathsWorksTM, Inc.) were used to
TABLE II
SUBJECTELECTRODEINFORMATION. AVAILABLEELECTRODESARE ACTIVEELECTRODES IN THESUBJECTS’ CLINICALFITTING. THE THIRDCOLUMNSHOWS THEELECTRODESTHATWERESPECIFICALLY
DEACTIVATED FOROURTESTS
generate study specific configurations based on subject related clinical settings.
C. Electrode Addressing
In a combined vestibular-cochlear system, up to three elec- trodes are needed for vestibular stimulation, one for each semicircular canal. This implies that some of the stimulation channels available in the implant are not available for cochlear stimulation. For example, in the 12-channel MED-EL implants used in this study, nine channels can be used for audio stimulation. To simulate the effect of this limitation and to obtain comparable results across patients, the number of activated electrodes was reduced for subjects with more than nine clinically activated electrodes.
For these subjects, the deactivated electrodes were chosen evenly distributed over the available array. For example, if the electrodes one to twelve were available, the third, sixth and tenth were deactivated. Data about the clinically activated electrodes and the specific electrodes that were deactivated for the tests in this study are given for each subject in Table II.
D. Stimulation Strategies
To evaluate the effect of combined cochleo-vestibular stimu- lation on speech perception, six different strategies were used in this study. All of them were based on a sequential CIS paradigm for the nine activated cochlear electrodes. During
Fig. 1. CIS stimulation strategy patterns used in this study. The reference strategy P1 shows the usual CIS pattern of stimulation, where the different electrodes are activated sequentially, with each blue arrow corresponding to a single electrode stimulation pulse. We define a frame as a part of the overall stimulation stream in which all electrodes have been stimulated exactly once. In strategies P2-P6, gaps are inserted at different places of a frame, denoted by orange arrows. The distribution of the electrode stimuli and the gaps within a frame is defined as a stimulation block pattern. In strategies P2 and P3 as well as P5 and P6, the block pattern only differs in the length of the individual gaps, i.e.
200µs for P2 and P5 and 400µs for P3 and P6. The frame duration of P4 is the same as P3, but the 400µs gap at the end of the block pattern of P3 is instead equally distributed between the stimulation pulses as depicted for P4.
periods when the vestibular electrodes would be stimulated, a stimulation of the auditory electrodes is not possible. The resulting gaps in the auditory stimulation sequence can be distributed in different block patterns as shown in Figure 1.
Block patterns that contain the stimulation of all nine auditory electrodes exactly once, together with specific gap patterns, form a stimulation frame.
As a reference strategy, the standard CIS strategy P1 with maximum frame rate and no gaps was chosen. In the stim- ulation paradigms P2 and P3, a single gap of 200 µs and 400 µs per frame, respectively, was introduced at the end of the frame. The duration of the gaps corresponds to a phase duration of 100 µs and 200 µs, respectively, which is similar to those typically used in current studies on vestibular stimulation [6], [7], [9]–[13]. While P3 contains one long gap per frame, P4 has the same frame length, but the 400µs gap is divided into small equidistant gaps over the whole frame. This allows evaluation of the influence of the distribution of periods without auditory stimulation within a frame. In P5 three gaps of 200 µs each were equally distributed over one frame, and in P6 three gaps of 400 µs were equally distributed over one frame. In one possible scenario, each of these three gaps could be used for the stimulation of one of the three semicircular canals within one frame.
The pulse phase duration of the cochlear stimulation was consistently set to 40.83µs with an interphase gap of 2.1µs for all subjects except S4 and S7, who were implanted with a C40+. For these subjects, the phase duration was set to 40 µs and the interphase gap is zero, leading to slightly different frame rates. The frame rate for the reference strategy P1 was set to the highest available value, while the effective frame
TABLE III
PARAMETERS OF THETESTEDSTRATEGIES. SINCEGAPSINCREASE THEFRAMEDURATION,THEFRAMERATEDECREASES FORHIGHER
CUMULATIVEGAPLENGTHS. FRAMERATESAREGIVEN FOR SUBJECTSWITH AC40+IMPLANT ANDSUBJECTSWITHOTHER
IMPLANTS, RESPECTIVELY
rates for P2-P6 were reduced due to the introduction of gaps (seeTable III). Stimulation pulses were biphasic with cathodic phase first. The gaps in the auditory stimulation were modeled by pulses with an amplitude of zero on non-used electrodes.
None of the subjects used any of the tested strategies in their clinical settings, which reduces the risk of introducing unwanted bias.
E. Procedure
1) Preparatory Configuration of the Test Processor: For all subjects, the most recent clinical fitting data was obtained before testing. This allowed us to prepare the programming of the processor ahead of time, leading to a reduced fitting time for the subjects. The clinical fitting data was exported from MAESTRO 8 to MATLAB, where the study relevant parame- ters like electrode sequences, number of gaps and durations were configured. Additionally, electrodes were deactivated as described above and strategies P1 to P6 were personalized, depending on the clinically deactivated electrodes. The final configurations were then written to the OPUS 2 test processor via the DIB II.
2) Loudness Balancing: To avoid a possible bias caused by different volume levels, the used strategies were loudness balanced prior to the speech tests in a two-step procedure. For a rough estimate, in the first step the investigator repeated a sentence while maintaining approximately the same speaking volume. A comfortable listening level was first determined for strategy P1. For each subsequent strategy, the volume was adjusted based on the loudness perceived by the subject to roughly match the loudness to the previous strategy.
In a second step, the balancing was refined in a sound-proof room. A studio speaker (AD-S52, QSC Audio Products, Inc.) playing white noise at 65dB SPL was positioned in front of the subject. The strategies were switched through in the order shown in Table III, playing the reference P1 every second time as comparison. This was repeated until no differences in loudness were noticeable for the subject.
In particular, for each strategy, the linear volume factor of the processor was used to adjust the amplitude of all active channels with the same factor. In this way, the amplitude of each channel can be varied in principle between threshold and
Fig. 2. Individual subject results of the consonant identification test. The different symbols represent independent blocks of all six strategies.
maximum comfortable loudness. In the balancing procedure in this study, the volume for strategies P2-P6 was increased by up to 25% compared to the reference strategy P1.
3) Testing Procedure:After the loudness balancing had been completed, a French consonant identification test at 65dB SPL in quiet [14] was performed as a free field test in a sound-proof room with a frontal studio speaker using strategies P1-P6. The test was performed in three blocks, each containing all six strategies. To prevent bias caused by training effects, the order of presentation of the six strategies was randomized within a block and across subjects.
In each test run, 56 logatomes, i.e. consonants preceded and followed by the vowel /a/, e.g. /ala/ were presented to the subject. The subject was then asked to choose one out of the 14 possible answers/b/,/d/,/f/,/g/,/k/,/l/,/m/,/n/,/p/, / /,/s/,/t/,/v/and/z/(IPA symbols) [15]. Feedback of the recognized sample was given by the subject via a graphical user interface (GUI) with clickable buttons for each possible consonant presented on a touchscreen. Audio signals were created by a sound card (Fireface UC, RME Intelligent Audio Solutions) controlled by a PC. The touchscreen panel used for subject feedback was connected to the same PC.
The duration between starting the playback sound and feedback by the subject on the touchscreen was recorded and accumulated for each run. These durations are further denoted as response time and accumulated response time, respectively.
Indeed, the response time has been suggested as a measure for the listening effort as an additional objective value for the quality of speech [16], [17].
After the consonant test was completed, subjects were asked for a quality assessment of their subjective hearing impres- sions. This was done in a single measure blinded test where a sentence was repeatedly spoken with an approximately constant volume level, while all strategies were compared one by one to the reference strategy. Subjects were then asked how they would rate the quality and the clarity of each strategy on a visual analog scale from 1 (worst) to 10 (best).
4) Ethical Considerations:All experiments were performed according to the clinical study protocol approved by the Cantonal research ethics commission in Geneva (CCER 2018- 01938). All patients provided written informed consent to participate in the study.
III. RESULTS
The individual subject results of the consonant identification test are presented inFigure 2. This figure shows results of the test as a percentage of correct answers over the 56 presented logatomes for the six strategies on the x-axis and the blocks, i.e. the three repetitions of the tests, as different symbols. The figure also shows relatively small intra-subject variability, with an expected higher inter-subject variability, and no evidence of flooring or ceiling effects in the data.
The group results for the different strategies of the con- sonant identification test are shown in Figure 3. To better reflect the relation between the different strategies for each subject, results have been normalized with respect to the mean value of the reference strategy P1 for that particular subject.
At first glance, no obvious differences between the strategies P1-P6 are noticeable.
Fig. 3. Mean values across the subjects of the correct answers results of the consonant identification test are presented for the six tested strategies. Results have been normalized to subjects’ P1-mean-value individually. Red crosses represent outliers.
Fig. 4. Mean response times per logatome for the six tested strategies.
To investigate these differences for the tested strategies in a more quantitative way, a 2-way repeated-measures analysis of variance (RM-ANOVA) with the factors strategy andrun was performed. Statistical analyses have been performed with SigmaPlot 14 (Systat Software Inc.). No statistically signif- icant difference for the main effect of strategy was found for the mean results of the consonant identification test (p= 0.286, F=1.312). For the main effect ofrun, no statistically significant (p=0.079, F=3.162) difference was found either.
Finally, no interaction betweenstrategyandrunwas observed (p=0.683, F=0.741). Thus, the effect of different strategies on speech perception does not depend on the order of the presented run, i.e., no learning effect could be seen.
In Figure 4, the mean response times per consonant and tested strategy are shown. Again, no apparent differ- ences between the tested strategies are noticeable. However, a repeated measures ANOVA shows statistically significant differences between the strategies (p = 0.003, F = 4.611).
Post-hoc pairwise multiple comparison tests (t-tests) show no differences between strategies P1-P5 though, and only a small but significant difference (p < 0.05) for P6 vs. P5, with a Bonferroni-corrected p-value of p =0.045.
The results of the quality assessments which were per- formed after the speech tests are shown in Table IV and Table V. Subjects were asked to rate the quality (Table IV) and clarity (Table V) of each of the tested strategies on a scale
TABLE IV
SUBJECTIVEQUALITYASSESSMENT BY THESUBJECTSFROM1 (POOREST)TO10 (BEST)
TABLE V
SUBJECTIVEASSESSMENT FOR THECLARITY OF THEPERCEIVED SOUNDRATED BY THESUBJECTSFROM1 (POOREST)TO10 (BEST)
from 1 to 10, with 1 indicating the poorest and 10 the best perceived quality. Subjects S1-S6 seemed to perceive a very similar quality and clarity for all tested strategies, S7 rated the quality of the reference strategy P1 as the best and P6 as the worst. A repeated measures ANOVA showed no significant differences for both the quality (p=0.48, F=0.92) and the clarity (p=0.52, F =0.85).
IV. DISCUSSION
In this feasibility study, several CIS cochlear stimulation strategies with different distributions of time-fixed gaps were compared to a standard CIS strategy without gaps in terms of speech perception. The reason for inserting gaps is to mimic a possible mode of operation in future cochleo-vestibular implants, where vestibular electrodes would be stimulated during the gaps in cochlear stimulation. We have tested two strategies with a single gap per frame in the auditory stream, as well as two strategies with three gaps. We have also studied the influence of a single gap per frame as compared to distributing this gap equally through the whole frame. All strategies used nine activated cochlear channels.
Earlier psychophysical studies have already demonstrated that 4-8 active channels are sufficient for recognizing speech sounds in quiet [18]. The influence of stimulation rate on speech perception has also been studied previously [19]–[22].
While, e.g., [19] found little variation of speech perception as a function of pulse rate, [20] compared speech perception using 720 and 1800 pps and found that some subjects performed better on certain tests with the lower pulse rate and others performed better with the higher pulse rate [21] and [22]
studied, among other things, the performance of subjects in consonant tests with a low (400pps), medium (800pps)
and high (≥1400pps) pulse rates with different phonological features. While [22] did not find any influence of pulse rate on consonant recognition, [21] found that subjects were more sensitive to pulse rate variations when the consonants were presented in a /iCi/ or a /uCu/ background as compared to the commonly used /aCa/ background. Despite the extensive work on the influence of pulse rate on speech perception, with quite a bit of variability in the results, the specific impact of irregular gaps in the auditory stimulation sequence, as studied in the current work, has never been investigated before.
Consonants were selected as test materials to allow for a rapid comparison of strategies. Previous studies [23] have shown significant correlations in CI users between consonant identification scores and scores from open-set speech tests, including bisyllabic word and sentence in noise tests. Thus, consonant test results are a good predictor for the outcomes of open-set tests in CI users. Vowel test performance, on the other hand, did not correlate with open-set speech test scores in the studies by [23]. While we have used the consonant tests in the commonly used /aCa/ context, it might be of interest to also look at the consonant performance in different vowel contexts due to a higher sensitivity on pulse rates as reported in [22].
The results of the performed consonant identification tests in our study show that the subjects performed better than chance in terms of speech perception for all tested strategies. Repeated measures ANOVA scores did not indicate a significant dete- rioration due to the new strategies. These results suggest that the tested strategies could, in general, be used for combined stimulation of the cochlea and the vestibular system in the future while preserving speech perception performance.
In addition to speech perception, we also looked at response times measured during the consonant identification test, which are, as an indicator, related to the listening effort. RM-ANOVA scores indicated a significant effect of the stimulation strategy.
However, post-hoc pairwise comparisons showed, that only for strategy P6 with the longest total gap duration, i.e. three gaps with 400 µs, there is a significant, albeit small differ- ence compared to strategy P5 with the same pulse grouping, but only 200 µs gaps. This might hint at a tendency, that listening effort could rise for very long gap durations. All other strategies do not show significant differences in response times.
In the third assessment of this study, the subjects were asked to rate the clarity and quality of a spoken sen- tence while using the different strategies and compare it to the reference strategy. The statistical analysis also does not indicate any significant differences between the tested strategies.
The response times, together with the subjective assessment, indicate that the listening effort for the new tested strategies does not noticeably increase and subjects do not experience a distinctly different quality of speech.
The subject population included in the current study consists of CI only users, the vestibular system was not stimulated. Pre- vious research indicates possible interactions between cochlear and vestibular stimulation in a combined stimulation para-
digm [24]. In our study design we have used a sequential stim- ulation paradigm, where the auditory system is only stimulated when there is no vestibular stimulation. We believe, that in this way, a possible influence of the vestibular stimulation on speech perception could be reduced in prospective combined cochleo-vestibular systems. In future studies the impact of real vestibular stimulation on speech perception is of great interest and has to be investigated in a subject population implanted with combined cochleo-vestibular systems.
Since the performed tests demand a high level of focus and concentration by the subjects for a period of up to 4 hours, albeit with breaks, a relatively small number of participants was chosen for this study. The power of the performed tests would naturally have been improved by testing a larger group of subjects. However, the primary goal of the study was to assess whether an immediate deterioration of speech perception would be introduced by gaps in the stimulation strategies that are long enough for vestibular stimulation pulses. In addition, the performance of subjects will potentially improve when they get used to the strategy, e.g., in home use. The study results suggest that one of the tested strategies could be used for combined stimulation of the cochlea and the vestibular system in the future.
V. CONCLUSION
In this study, the influence of introducing specific gaps in the auditory stream for a cochlear implant on speech perception were evaluated. The results do not show significant deterioration of speech perception even in strategies with gaps up to 400µs. Additionally, a subjective assessment of clarity and quality for the tested strategies did not yield significant differences. While these findings were obtained in regular CI recipients, they give rise to an avenue of future strategies that could be used in combined cochleo-vestibular implants.
In such systems, the vestibular stimulation pulses could be applied during the gaps in the auditory stream, which allows for sequential combined stimulation.
ACKNOWLEDGMENT
The authors would like to thank the patients for their time and motivation.
REFERENCES
[1] D. S. Dalton, K. J. Cruickshanks, B. E. K. Klein, R. Klein, T. L. Wiley, and D. M. Nondahl, “The impact of hearing loss on quality of life in older adults,”Gerontologist, vol. 43, no. 5, pp. 661–668, Oct. 2003, doi:
10.1093/geront/43.5.661.
[2] Multi-Country Assessment of National Capacity to Provide Hearing Care, WHO, Geneva, Switzerland, 2013, pp. 1–49.
[3] D. Q. Sun, B. K. Ward, Y. R. Semenov, J. P. Carey, and C. C. D. Santina,
“Bilateral vestibular deficiency: Quality of life and economic impli- cations,” JAMA Otolaryngol. Head Neck Surg., vol. 140, no. 6, pp. 527–534, 2014, doi:10.1001/jamaoto.2014.490.
[4] K. N. Hageman et al., “A CMOS neural interface for a multichannel vestibular prosthesis,”IEEE Trans. Biomed. Circuits Syst., vol. 10, no. 2, pp. 269–279, Apr. 2016, doi:10.1109/TBCAS.2015.2409797.
[5] J.-P. Guyot and A. P. Fornos, “Milestones in the development of a vestibular implant,” Current Opinion Neurol., vol. 32, no. 1, pp. 145–153, 2019.
[6] J. T. Rubinstein, L. Ling, A. Nowack, K. Nie, and J. O. Phillips,
“Results from a second-generation vestibular implant in human subjects:
Diagnosis may impact electrical sensitivity of vestibular afferents,”
Otol. Neurotol., vol. 41, no. 1, pp. 68–77, Jan. 2020, doi: 10.1097/
MAO.0000000000002463.
[7] C. Crétallazet al., “Influence of systematic variations of the stimulation profile on responses evoked with a vestibular implant prototype in humans,”J. Neural Eng., vol. 17, no. 3, Jun. 2020, Art. no. 036027.
[8] B. S. Wilson, C. C. Finley, D. T. Lawson, R. D. Wolford, D. K. Eddington, and W. M. Rabinowitz, “Better speech recognition with cochlear implants,” Nature, vol. 352, no. 6332, pp. 236–238, Jul. 1991, doi:10.1038/352236a0.
[9] R. van de Berget al., “The vestibular implant: Frequency-dependency of the electrically evoked vestibulo-ocular reflex in humans,”Frontiers Syst. Neurosci., vol. 8, p. 255, Jan. 2015.
[10] T. A. K. Nguyen et al., “Characterization of cochlear, vestibular and cochlear-vestibular electrically evoked compound action potentials in patients with a vestibulo-cochlear implant,”Frontiers Neurosci., vol. 11, p. 645, Nov. 2017.
[11] P. J. Boutroset al., “Continuous vestibular implant stimulation partially restores eye-stabilizing reflexes,”JCI Insight, vol. 4, no. 22, Nov. 2019, Art. no. e128397, doi:10.1172/jci.insight.128397.
[12] A. P. Fornoset al., “Artificial balance: Restoration of the vestibulo-ocular reflex in humans with a prototype vestibular neuroprosthesis,”Frontiers Neurol., vol. 5, p. 66, Apr. 2014, doi:10.3389/fneur.2014.00066.
[13] T. A. K. Nguyenet al., “Characterization of pulse amplitude and pulse rate modulation for a human vestibular implant during acute electrical stimulation,”J. Neural Eng., vol. 13, no. 4, 2016, Art. no. 046023.
[14] M. Pelizzone, C. Boëx, and P. Montandon, “Vowel and consonant iden- tification tests can be used to compare performances in a multilingual group of cochlear implant patients,”ORL, vol. 55, no. 6, pp. 341–346, 1993, doi:10.1159/000276452.
[15] B. Pompino-Marschall, Einführung in die Phonetik. Berlin, Germany:
Walter de Gruyter, 2009.
[16] C. Pals, A. Sarampalis, and D. Ba¸skent, “Listening effort with cochlear implant simulations,” J. Speech, Lang. Hearing Res., vol. 56, no. 4, pp. 1075–1084, 2013.
[17] T. Baer, B. C. J. Moore, and S. Gatehouse, “Spectral contrast enhance- ment of speech in noise for listeners with sensorineural hearing impair- ment: Effects on intelligibility, quality, and response times,”J. Rehabil.
Res. Dev., vol. 30, p. 49, Jan. 1993.
[18] R. V. Shannon, F.-G. Zeng, V. Kamath, J. Wygonski, and M. Ekelid, “Speech recognition with primarily temporal cues,”Science, vol. 270, no. 5234, pp. 303–304, Oct. 1995, doi: 10.1126/science.
270.5234.303.
[19] R. M. Bonnet, P.-P.-B. M. Boermans, O. F. Avenarius, J. J. Briaire, and J. H. M. Frijns, “Effects of pulse width, pulse rate and paired electrode stimulation on psychophysical measures of dynamic range and speech recognition in cochlear implants,” Ear Hear- ing, vol. 33, no. 4, pp. 489–496, Jul. 2012, doi: 10.1097/AUD.
0b013e31824c761a.
[20] L. K. Holden, M. W. Skinner, T. A. Holden, and M. E. Demorest,
“Effects of stimulation rate with the nucleus 24 ACE speech coding strategy,” Ear Hearing, vol. 23, no. 5, pp. 463–476, Oct. 2002, doi:
10.1097/00003446-200210000-00008.
[21] P. C. Loizou, O. Poroy, and M. Dorman, “The effect of parametric variations of cochlear implant processors on speech understanding,”
J. Acoust. Soc. Amer., vol. 108, no. 2, pp. 790–802, Aug. 2000, doi:
10.1121/1.429612.
[22] C. A. Verschuur, “Effect of stimulation rate on speech perception in adult users of the Med-El CIS speech processing strategy efectos de la tasa de estimulación en la percepción del lenguaje en usuarios adultos de la estrategia de procesamiento Med-El CIS,”Int. J. Audiol., vol. 44, no. 1, pp. 58–63, Jan. 2005, doi:10.1080/14992020400022488.
[23] C. Wilson, B. Lawson, and D. Finley, “Speech processors for auditory prostheses: Comparison of compressed analog and continuous inter- leaved sampler processors in tests with Symbion subjects,” Neural Pros- thesis Program, Nat. Inst. Health, Bethesda, MD, USA, Fourth Quart.
Prog. Rep. NIH Project N01-DC-9-2401, 1990.
[24] J. O. Phillips, L. Ling, A. Nowack, B. Rebollar, and J. T. Rubinstein,
“Interactions between auditory and vestibular modalities during stim- ulation with a combined vestibular and cochlear prosthesis,” Audiol.
Neurotol., vol. 25, nos. 1–2, pp. 96–108, 2020, doi: 10.1159/
000503846.
