Peripheral Neural Synchrony in Postlingually Deafened Adult Cochlear Implant Users.

OBJECTIVES: This paper reports a noninvasive method for quantifying neural synchrony in the cochlear nerve (i.e., peripheral neural synchrony) in cochlear implant (CI) users, which allows for evaluating this physiological phenomenon in human CI users for the first time in the literature. In addition, this study assessed how peripheral neural synchrony was correlated with temporal resolution acuity and speech perception outcomes measured in quiet and in noise in postlingually deafened adult CI users. It tested the hypothesis that peripheral neural synchrony was an important factor for temporal resolution acuity and speech perception outcomes in noise in postlingually deafened adult CI users. DESIGN: Study participants included 24 postlingually deafened adult CI users with a Cochlear™ Nucleus® device. Three study participants were implanted bilaterally, and each ear was tested separately. For each of the 27 implanted ears tested in this study, 400 sweeps of the electrically evoked compound action potential (eCAP) were measured at four electrode locations across the electrode array. Peripheral neural synchrony was quantified at each electrode location using the phase-locking value (PLV), which is a measure of trial-by-trial phase coherence among eCAP sweeps/trials. Temporal resolution acuity was evaluated by measuring the within-channel gap detection threshold (GDT) using a three-alternative, forced-choice procedure in a subgroup of 20 participants (23 implanted ears). For each ear tested in these participants, GDTs were measured at two electrode locations with a large difference in PLVs. For 26 implanted ears tested in 23 participants, speech perception performance was evaluated using consonant-nucleus-consonant (CNC) word lists presented in quiet and in noise at signal to noise ratios (SNRs) of +10 and +5 dB. Linear Mixed effect Models were used to evaluate the effect of electrode location on the PLV and the effect of the PLV on GDT after controlling for the stimulation level effects. Pearson product-moment correlation tests were used to assess the correlations between PLVs, CNC word scores measured in different conditions, and the degree of noise effect on CNC word scores. RESULTS: There was a significant effect of electrode location on the PLV after controlling for the effect of stimulation level. There was a significant effect of the PLV on GDT after controlling for the effects of stimulation level, where higher PLVs (greater synchrony) led to lower GDTs (better temporal resolution acuity). PLVs were not significantly correlated with CNC word scores measured in any listening condition or the effect of competing background noise presented at an SNR of +10 dB on CNC word scores. In contrast, there was a significant negative correlation between the PLV and the degree of noise effect on CNC word scores for a competing background noise presented at an SNR of +5 dB, where higher PLVs (greater synchrony) correlated with smaller noise effects on CNC word scores. CONCLUSIONS: This newly developed method can be used to assess peripheral neural synchrony in CI users, a physiological phenomenon that has not been systematically evaluated in electrical hearing. Poorer peripheral neural synchrony leads to lower temporal resolution acuity and is correlated with a larger detrimental effect of competing background noise presented at an SNR of 5 dB on speech perception performance in postlingually deafened adult CI users.

Peripheral neural synchrony in post-lingually deafened adult cochlear implant users.

Objective: This paper reports a noninvasive method for quantifying neural synchrony in the cochlear nerve (i.e., peripheral neural synchrony) in cochlear implant (CI) users, which allows for evaluating this physiological phenomenon in human CI users for the first time in the literature. In addition, this study assessed how peripheral neural synchrony was correlated with temporal resolution acuity and speech perception outcomes measured in quiet and in noise in post-lingually deafened adult CI users. It tested the hypothesis that peripheral neural synchrony was an important factor for temporal resolution acuity and speech perception outcomes in noise in post-lingually deafened adult CI users. Design: Study participants included 24 post-lingually deafened adult CI users with a Cochlear™ Nucleus® device. Three study participants were implanted bilaterally, and each ear was tested separately. For each of the 27 implanted ears tested in this study, 400 sweeps of the electrically evoked compound action potential (eCAP) were measured at four electrode locations across the electrode array. Peripheral neural synchrony was quantified at each electrode location using the phase locking value (PLV), which is a measure of trial-by-trial phase coherence among eCAP sweeps/trials. Temporal resolution acuity was evaluated by measuring the within-channel gap detection threshold (GDT) using a three-alternative, forced-choice procedure in a subgroup of 20 participants (23 implanted ears). For each ear tested in these participants, GDTs were measured at two electrode locations with a large difference in PLVs. For 26 implanted ears tested in 23 participants, speech perception performance was evaluated using Consonant-Nucleus-Consonant (CNC) word lists presented in quiet and in noise at signal-to-noise ratios (SNRs) of +10 and +5 dB. Linear Mixed effect Models were used to evaluate the effect of electrode location on the PLV and the effect of the PLV on GDT after controlling for the stimulation level effects. Pearson product-moment correlation tests were used to assess the correlations between PLVs, CNC word scores measured in different conditions, and the degree of noise effect on CNC word scores. Results: There was a significant effect of electrode location on the PLV after controlling for the effect of stimulation level. There was a significant effect of the PLV on GDT after controlling for the effects of stimulation level, where higher PLVs (greater synchrony) led to lower GDTs (better temporal resolution acuity). PLVs were not significantly correlated with CNC word scores measured in any listening condition or the effect of competing background noise presented at a SNR of +10 dB on CNC word scores. In contrast, there was a significant negative correlation between the PLV and the degree of noise effect on CNC word scores for a competing background noise presented at a SNR of +5 dB, where higher PLVs (greater synchrony) correlated with smaller noise effects on CNC word scores. Conclusions: This newly developed method can be used to assess peripheral neural synchrony in CI users, a physiological phenomenon that has not been systematically evaluated in electrical hearing. Poorer peripheral neural synchrony leads to lower temporal resolution acuity and is correlated with a larger detrimental effect of competing background noise presented at a SNR of 5 dB on speech perception performance in post-lingually deafened adult CI users.

WaveNet-based approximation of a cochlear filtering and hair cell transduction model.

Computational auditory models are important tools for gaining new insights into hearing mechanisms, and they can provide a foundation for bio-inspired speech and audio processing algorithms. However, accurate models often entail an immense computational effort, rendering their application unfeasible if quick execution is required. This paper presents a WaveNet-based approximation of the normal-hearing cochlear filtering and inner hair cell (IHC) transduction stages of a widely used auditory model [Zilany and Bruce (2006). J. Acoust. Soc. Am. 120(3), 1446-1466]. The WaveNet model was trained and optimized using a large dataset of clean speech, noisy speech, and music for a wide range of sound pressure levels (SPLs) and characteristic frequencies between 125 Hz and 8 kHz. The model was evaluated with unseen (noisy) speech, music signals, sine tones, and click signals at SPLs between 30 and 100 dB. It provides accurate predictions of the IHC receptor potentials for a given input stimulus and allows an efficient execution with processing times up to 250 times lower compared to an already optimized reference implementation of the original auditory model. The WaveNet model is fully differentiable, thus, allowing its application in the context of deep-learning-based speech and audio enhancement algorithms.

A comparative study of eight human auditory models of monaural processing.

A number of auditory models have been developed using diverging approaches, either physiological or perceptual, but they share comparable stages of signal processing, as they are inspired by the same constitutive parts of the auditory system. We compare eight monaural models that are openly accessible in the Auditory Modelling Toolbox. We discuss the considerations required to make the model outputs comparable to each other, as well as the results for the following model processing stages or their equivalents: Outer and middle ear, cochlear filter bank, inner hair cell, auditory nerve synapse, cochlear nucleus, and inferior colliculus. The discussion includes a list of recommendations for future applications of auditory models.

Comparison of response properties of the electrically stimulated auditory nerve reported in human listeners and in animal models.

Cochlear implants (CIs) provide acoustic information to implanted patients by electrically stimulating nearby auditory nerve fibers (ANFs) which then transmit the information to higher-level neural structures for further processing and interpretation. Computational models that simulate ANF responses to CI stimuli enable the exploration of the mechanisms underlying CI performance beyond the capacity of in vivo experimentation alone. However, all ANF models developed to date utilize to some extent anatomical/morphometric data, biophysical properties and/or physiological data measured in non-human animal models. This review compares response properties of the electrically stimulated auditory nerve (AN) in human listeners and different mammalian models. Properties of AN responses to single pulse stimulation, paired-pulse stimulation, and pulse-train stimulation are presented. While some AN response properties are similar between human listeners and animal models (e.g., increased AN sensitivity to single pulse stimuli with long interphase gaps), there are some significant differences. For example, the AN of most animal models is typically more sensitive to cathodic stimulation while the AN of human listeners is generally more sensitive to anodic stimulation. Additionally, there are substantial differences in the speed of recovery from neural adaptation between animal models and human listeners. Therefore, results from animal models cannot be simply translated to human listeners. Recognizing the differences in responses of the AN to electrical stimulation between humans and other mammals is an important step for creating ANF models that are more applicable to various human CI patient populations.

Phenomenological model of auditory nerve population responses to cochlear implant stimulation.

BACKGROUND: Models of auditory nerve fiber (ANF) responses to electrical stimulation are helpful to develop advanced coding for cochlear implants (CIs). A phenomenological model of ANF population responses to CI electrical stimulation with a lower computational complexity compared to a biophysical model would be beneficial to evaluate new CI coding strategies. NEW METHOD: This study presents a phenomenological model which combines four temporal characteristics of ANFs (refractoriness, facilitation, accommodation and spike rate adaptation) in addition to a spatial spread of the electric field. RESULTS: The model predicts the performances of CI subjects in the melodic contour identification (MCI) experiment. The simulations for the MCI experiment were consistent with CI recipients' experimental outcomes that were not predictable from the electrical stimulation patterns themselves. COMPARISON WITH EXISTING METHODS: Previously, no phenomenological population model of ANFs has combined all four aforementioned temporal phenomena. CONCLUSIONS: The proposed model would help the further investigations of ANFs responses to different electrical stimulation patterns and comparison of different sound coding strategies in CIs.

Measuring temporal response properties of auditory nerve fibers in cochlear implant recipients.

Auditory nerve fibers' (ANFs) refractoriness and facilitation can be quantified in electrically evoked compound action potentials (ECAPs) recorded via neural response telemetry (NRT). Although facilitation has been observed in animals and human cochlear implant (CI) recipients, no study has modeled this in human CI users until now. In this study, recovery and facilitation effects at different masker and probe levels for three test electrodes (E6, E12 and E18) in 11 CI subjects were recorded. The ECAP recovery and facilitation were modeled by exponential functions and the same function used for +10 CL masker offset condition can be applied to all other masker offsets measurements. Goodness of fit was evaluated for the exponential functions. A significant effect of probe level was observed on a recovery time constant which highlights the importance of recording the recovery function at the maximum acceptable stimulus level. Facilitation time constant and amplitude showed no dependency on the probe level. However, facilitation was stronger for masker level at or around the threshold of the ECAP (T-ECAP). There was a positive correlation between facilitation magnitude and amplitude growth function (AGF) slope, which indicates that CI subjects with better peripheral neural survival have stronger facilitation.

Envelope following responses, noise exposure, and evidence of cochlear synaptopathy in humans: Correction and comment.

A correction and comment are provided for a recent article by Paul, Waheed, Bruce, and Roberts [(2017). J. Acoust. Soc. Am. 142(5), EL434-EL440].

Perceptual and Model-Based Evaluation of Ideal Time-Frequency Noise Reduction in Hearing-Impaired Listeners.

State-of-the-art hearing aids (HAs) try to overcome the deficit of poor speech intelligibility (SI) in noisy listening environments using digital noise reduction (NR) techniques. The application of time-frequency masks to the noisy sound input is a common NR technique to increase SI. The binary mask with its binary weights and the Wiener filter with continuous weights are representatives of a hard- and a soft-decision approach for time-frequency masking. In normal-hearing listeners, the ideal Wiener filter (IWF) outperforms the ideal binary mask (IBM) in terms of SI and speech quality with perfect SI even at very low signal-to-noise ratios. In this paper, both approaches were investigated for hearing-impaired (HI) listeners. Perceptual and auditory model-based measures were used for the evaluation. The IWF outperformed the IBM in terms of SI. Quality-wise, there was no overall difference between the NR algorithms perceived. Additionally, the processed signals were evaluated based on an auditory nerve model using the neurogram similarity metric (NSIM). The mean NSIM values were significantly different for intelligible and unintelligible sentences. The results suggest that a soft-mask seems to be promising for application in HAs.

A phenomenological model of the synapse between the inner hair cell and auditory nerve: Implications of limited neurotransmitter release sites.

Peterson and Heil [Hear. Res., In Press] have argued that the statistics of spontaneous spiking in auditory nerve fibers (ANFs) can be best explained by a model with a limited number of synaptic vesicle docking (release) sites (∼4) and a relatively-long average redocking time (∼16-17 ms) for each of the sites. In this paper we demonstrate how their model can be: i) generalized to also describe sound-driven ANF responses and ii) incorporated into a well-established and widely-used model of the entire auditory periphery [Zilany et al., J. Acoust. Soc. Am. 135, 283-286, 2014]. The responses of the new model exhibit substantial improvement in several measures of ANF spiking statistics, and predicted physiological forward-masking and rate-level functions from the new model structure are shown to also better match published physiological data.

Subcortical amplitude modulation encoding deficits suggest evidence of cochlear synaptopathy in normal-hearing 18-19 year olds with higher lifetime noise exposure.

Noise exposure and aging can damage cochlear synapses required for suprathreshold listening, even when cochlear structures needed for hearing at threshold remain unaffected. To control for effects of aging, behavioral amplitude modulation (AM) detection and subcortical envelope following responses (EFRs) to AM tones in 25 age-restricted (18-19 years) participants with normal thresholds, but different self-reported noise exposure histories were studied. Participants with more noise exposure had smaller EFRs and tended to have poorer AM detection than less-exposed individuals. Simulations of the EFR using a well-established cochlear model were consistent with more synaptopathy in participants reporting greater noise exposure.

Predicting the quality of enhanced wideband speech with a cochlear model.

Objective measures are commonly used in the development of speech coding algorithms as an adjunct to human subjective evaluation. Predictors of speech quality based on models of physiological or perceptual processing tend to perform better than measures based on simple acoustical properties. Here, a modeling method based on a detailed physiological model and a neurogram similarity measure is developed and optimized to predict the quality of an enhanced wideband speech dataset. A model capturing temporal modulations in neural activity up to 267 Hz was found to perform as well as or better than several existing objective quality measures.

Predictions of Speech Chimaera Intelligibility Using Auditory Nerve Mean-Rate and Spike-Timing Neural Cues.

Perceptual studies of speech intelligibility have shown that slow variations of acoustic envelope (ENV) in a small set of frequency bands provides adequate information for good perceptual performance in quiet, whereas acoustic temporal fine-structure (TFS) cues play a supporting role in background noise. However, the implications for neural coding are prone to misinterpretation because the mean-rate neural representation can contain recovered ENV cues from cochlear filtering of TFS. We investigated ENV recovery and spike-time TFS coding using objective measures of simulated mean-rate and spike-timing neural representations of chimaeric speech, in which either the ENV or the TFS is replaced by another signal. We (a) evaluated the levels of mean-rate and spike-timing neural information for two categories of chimaeric speech, one retaining ENV cues and the other TFS; (b) examined the level of recovered ENV from cochlear filtering of TFS speech; (c) examined and quantified the contribution to recovered ENV from spike-timing cues using a lateral inhibition network (LIN); and (d) constructed linear regression models with objective measures of mean-rate and spike-timing neural cues and subjective phoneme perception scores from normal-hearing listeners. The mean-rate neural cues from the original ENV and recovered ENV partially accounted for perceptual score variability, with additional variability explained by the recovered ENV from the LIN-processed TFS speech. The best model predictions of chimaeric speech intelligibility were found when both the mean-rate and spike-timing neural cues were included, providing further evidence that spike-time coding of TFS cues is important for intelligibility when the speech envelope is degraded.

Predicting phoneme and word recognition in noise using a computational model of the auditory periphery.

Several filterbank-based metrics have been proposed to predict speech intelligibility (SI). However, these metrics incorporate little knowledge of the auditory periphery. Neurogram-based metrics provide an alternative, incorporating knowledge of the physiology of hearing by using a mathematical model of the auditory nerve response. In this work, SI was assessed utilizing different filterbank-based metrics (the speech intelligibility index and the speech-based envelope power spectrum model) and neurogram-based metrics, using the biologically inspired model of the auditory nerve proposed by Zilany, Bruce, Nelson, and Carney [(2009), J. Acoust. Soc. Am. 126(5), 2390-2412] as a front-end and the neurogram similarity metric and spectro temporal modulation index as a back-end. Then, the correlations with behavioural scores were computed. Results showed that neurogram-based metrics representing the speech envelope showed higher correlations with the behavioural scores at a word level. At a per-phoneme level, it was found that phoneme transitions contribute to higher correlations between objective measures that use speech envelope information at the auditory periphery level and behavioural data. The presented framework could function as a useful tool for the validation and tuning of speech materials, as well as a benchmark for the development of speech processing algorithms.

Predictions of the Contribution of HCN Half-Maximal Activation Potential Heterogeneity to Variability in Intrinsic Adaptation of Spiral Ganglion Neurons.

Spiral ganglion neurons (SGNs) exhibit a wide range in their strength of intrinsic adaptation on a timescale of 10s to 100s of milliseconds in response to electrical stimulation from a cochlear implant (CI). The purpose of this study was to determine how much of that variability could be caused by the heterogeneity in half-maximal activation potentials of hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels, which are known to produce intrinsic adaptation. In this study, a computational membrane model of cat type I SGN was developed based on the Hodgkin-Huxley model plus HCN and low-threshold potassium (KLT) conductances in which the half-maximal activation potential of the HCN channel was varied and the response of the SGN to pulse train and paired-pulse stimulation was simulated. Physiologically plausible variation of HCN half-maximal activation potentials could indeed determine the range of adaptation on the timescale of 10s to 100s of milliseconds and recovery from adaptation seen in the physiological data while maintaining refractoriness within physiological bounds. This computational model demonstrates that HCN channels may play an important role in regulating the degree of adaptation in response to pulse train stimulation and therefore contribute to variable constraints on acoustic information coding by CIs. This finding has broad implications for CI stimulation paradigms in that cell-to-cell variation of HCN channel properties are likely to significantly alter SGN excitability and therefore auditory perception.

Evidence that hidden hearing loss underlies amplitude modulation encoding deficits in individuals with and without tinnitus.

Damage to auditory nerve fibers that expresses with suprathreshold sounds but is hidden from the audiogram has been proposed to underlie deficits in temporal coding ability observed among individuals with otherwise normal hearing, and to be present in individuals experiencing chronic tinnitus with clinically normal audiograms. We tested whether these individuals may have hidden synaptic losses on auditory nerve fibers with low spontaneous rates of firing (low-SR fibers) that are important for coding suprathreshold sounds in noise while high-SR fibers determining threshold responses in quiet remain relatively unaffected. Tinnitus and control subjects were required to detect the presence of amplitude modulation (AM) in a 5 kHz, suprathreshold tone (a frequency in the tinnitus frequency region of the tinnitus subjects, whose audiometric thresholds were normal to 12 kHz). The AM tone was embedded within background noise intended to degrade the contribution of high-SR fibers, such that AM coding was preferentially reliant on low-SR fibers. We also recorded by electroencephalography the "envelope following response" (EFR, generated in the auditory midbrain) to a 5 kHz, 85 Hz AM tone presented in the same background noise, and also in quiet (both low-SR and high-SR fibers contributing to AM coding in the latter condition). Control subjects with EFRs that were comparatively resistant to the addition of background noise had better AM detection thresholds than controls whose EFRs were more affected by noise. Simulated auditory nerve responses to our stimulus conditions using a well-established peripheral model suggested that low-SR fibers were better preserved in the former cases. Tinnitus subjects had worse AM detection thresholds and reduced EFRs overall compared to controls. Simulated auditory nerve responses found that in addition to severe low-SR fiber loss, a degree of high-SR fiber loss that would not be expected to affect audiometric thresholds was needed to explain the results in tinnitus subjects. The results indicate that hidden hearing loss could be sufficient to account for impaired temporal coding in individuals with normal audiograms as well as for cases of tinnitus without audiometric hearing loss.

The history and future of neural modeling for cochlear implants.

This special issue of Network: Computation in Neural Systems on the topic of "Computational models of the electrically stimulated auditory system" incorporates review articles spanning a wide range of approaches to modeling cochlear implant stimulation of the auditory system. The purpose of this overview paper is to provide a historical context for the different modeling endeavors and to point toward how computational modeling could play a key role in the understanding, evaluation, and improvement of cochlear implants in the future.

Phenomenological modelling of electrically stimulated auditory nerve fibers: A review.

Auditory nerve fibers (ANFs) play a crucial role in hearing by encoding and transporting the synaptic input from inner hair cells into afferent spiking information for higher stages of the auditory system. If the inner hair cells are degenerated, cochlear implants may restore hearing by directly stimulating the ANFs. The response of an ANF is affected by several characteristics of the electrical stimulus and of the ANF, and neurophysiological measurements are needed to know how the ANF responds to a particular stimulus. However, recording from individual nerve fibers in humans is not feasible and obtaining compound neural or psychophysical responses is often time-consuming. This motivates the design and use of models to estimate the ANF response to the electrical stimulation. Phenomenological models reproduce the ANF response based on a simplified description of ANF functionality and on a limited parameter space by not directly describing detailed biophysical mechanisms. Here, we give an overview of phenomenological models published to date and demonstrate how different modeling approaches can account for the diverse phenomena affecting the ANF response. To highlight the success achieved in designing such models, we also describe a number of applications of phenomenological models to predict percepts of cochlear implant listeners.

Temporal Considerations for Stimulating Spiral Ganglion Neurons with Cochlear Implants.

A wealth of knowledge about different types of neural responses to electrical stimulation has been developed over the past 100 years. However, the exact forms of neural response properties can vary across different types of neurons. In this review, we survey four stimulus-response phenomena that in recent years are thought to be relevant for cochlear implant stimulation of spiral ganglion neurons (SGNs): refractoriness, facilitation, accommodation, and spike rate adaptation. Of these four, refractoriness is the most widely known, and many perceptual and physiological studies interpret their data in terms of refractoriness without incorporating facilitation, accommodation, or spike rate adaptation. In reality, several or all of these behaviors are likely involved in shaping neural responses, particularly at higher stimulation rates. A better understanding of the individual and combined effects of these phenomena could assist in developing improved cochlear implant stimulation strategies. We review the published physiological data for electrical stimulation of SGNs that explores these four different phenomena, as well as some of the recent studies that might reveal the biophysical bases of these stimulus-response phenomena.

Evidence for differential modulation of primary and nonprimary auditory cortex by forward masking in tinnitus.

It has been proposed that tinnitus is generated by aberrant neural activity that develops among neurons in tonotopic of regions of primary auditory cortex (A1) affected by hearing loss, which is also the frequency region where tinnitus percepts localize (Eggermont and Roberts 2004; Roberts et al., 2010, 2013). These models suggest (1) that differences between tinnitus and control groups of similar age and audiometric function should depend on whether A1 is probed in tinnitus frequency region (TFR) or below it, and (2) that brain responses evoked from A1 should track changes in the tinnitus percept when residual inhibition (RI) is induced by forward masking. We tested these predictions by measuring (128-channel EEG) the sound-evoked 40-Hz auditory steady-state response (ASSR) known to localize tonotopically to neural sources in A1. For comparison the N1 transient response localizing to distributed neural sources in nonprimary cortex (A2) was also studied. When tested under baseline conditions where tinnitus subjects would have heard their tinnitus, ASSR responses were larger in a tinnitus group than in controls when evoked by 500 Hz probes while the reverse was true for tinnitus and control groups tested with 5 kHz probes, confirming frequency-dependent group differences in this measure. On subsequent trials where RI was induced by masking (narrow band noise centered at 5 kHz), ASSR amplitude increased in the tinnitus group probed at 5 kHz but not in the tinnitus group probed at 500 Hz. When collapsed into a single sample tinnitus subjects reporting comparatively greater RI depth and duration showed comparatively larger ASSR increases after masking regardless of probe frequency. Effects of masking on ASSR amplitude in the control groups were completely reversed from those in the tinnitus groups, with no change seen to 5 kHz probes but ASSR increases to 500 Hz probes even though the masking sound contained no energy at 500 Hz (an "off-frequency" masking effect). In contrast to these findings for the ASSR, N1 amplitude was larger in tinnitus than control groups at both probe frequencies under baseline conditions, decreased after masking in all conditions, and did not relate to RI. These results suggest that aberrant neural activity occurring in the TFR of A1 underlies tinnitus and its modulation during RI. They indicate further that while neural changes occur in A2 in tinnitus, these changes do not reflect the tinnitus percept. Models for tinnitus and forward masking are described that integrate these findings within a common framework.