Auditory cortical images of cochlear-implant stimuli: coding of stimulus channel and current level

Tonotopic Selectivity in Cats and Humans: Electrophysiology and Psychophysics

We describe a scalp-recorded measure of tonotopic selectivity, the "cortical onset response" (COR) and compare the results between humans and cats. The COR results, in turn, were compared with psychophysical masked-detection thresholds obtained using similar stimuli and obtained from both species. The COR consisted of averaged responses elicited by 50-ms tone-burst probes presented at 1-s intervals against a continuous noise masker. The noise masker had a bandwidth of 1 or 1/8th octave, geometrically centred on 4000 Hz for humans and on 8000 Hz for cats. The probe frequency was either - 0.5, - 0.25, 0, 0.25 or 0.5 octaves re the masker centre frequency. The COR was larger for probe frequencies more distant from the centre frequency of the masker, and this effect was greater for the 1/8th-octave than for the 1-octave masker. This pattern broadly reflected the masked excitation patterns obtained psychophysically with similar stimuli in both species. However, the positive signal-to-noise ratio used to obtain reliable COR measures meant that some aspects of the data differed from those obtained psychophysically, in a way that could be partly explained by the upward spread of the probe's excitation pattern. Our psychophysical measurements also showed that the auditory filter width obtained at 8000 Hz using notched-noise maskers was slightly wider in cat than previous measures from humans. We argue that although conclusions from COR measures differ in some ways from conclusions based on psychophysics, the COR measures provide an objective, noninvasive, valid measure of tonotopic selectivity that does not require training and that may be applied to acoustic and cochlear-implant experiments in humans and laboratory animals.

Recent Advances in the Application of Artificial Intelligence in Otorhinolaryngology-Head and Neck Surgery

This study presents an up-to-date survey of the use of artificial intelligence (AI) in the field of otorhinolaryngology, considering opportunities, research challenges, and research directions. We searched PubMed, the Cochrane Central Register of Controlled Trials, Embase, and the Web of Science. We initially retrieved 458 articles. The exclusion of non-English publications and duplicates yielded a total of 90 remaining studies. These 90 studies were divided into those analyzing medical images, voice, medical devices, and clinical diagnoses and treatments. Most studies (42.2%, 38/90) used AI for image-based analysis, followed by clinical diagnoses and treatments (24 studies). Each of the remaining two subcategories included 14 studies. Machine learning and deep learning have been extensively applied in the field of otorhinolaryngology. However, the performance of AI models varies and research challenges remain.

Machine Learning and Cochlear Implantation-A Structured Review of Opportunities and Challenges

OBJECTIVE

The use of machine learning technology to automate intellectual processes and boost clinical process efficiency in medicine has exploded in the past 5 years. Machine learning excels in automating pattern recognition and in adapting learned representations to new settings. Moreover, machine learning techniques have the advantage of incorporating complexity and are free from many of the limitations of traditional deterministic approaches. Cochlear implants (CI) are a unique fit for machine learning techniques given the need for optimization of signal processing to fit complex environmental scenarios and individual patients' CI MAPping. However, there are many other opportunities where machine learning may assist in CI beyond signal processing. The objective of this review was to synthesize past applications of machine learning technologies for pediatric and adult CI and describe novel opportunities for research and development.

DATA SOURCES

The PubMed/MEDLINE, EMBASE, Scopus, and ISI Web of Knowledge databases were mined using a directed search strategy to identify the nexus between CI and artificial intelligence/machine learning literature.

STUDY SELECTION

Non-English language articles, articles without an available abstract or full-text, and nonrelevant articles were manually appraised and excluded. Included articles were evaluated for specific machine learning methodologies, content, and application success.

DATA SYNTHESIS

The database search identified 298 articles. Two hundred fifty-nine articles (86.9%) were excluded based on the available abstract/full-text, language, and relevance. The remaining 39 articles were included in the review analysis. There was a marked increase in year-over-year publications from 2013 to 2018. Applications of machine learning technologies involved speech/signal processing optimization (17; 43.6% of articles), automated evoked potential measurement (6; 15.4%), postoperative performance/efficacy prediction (5; 12.8%), and surgical anatomy location prediction (3; 7.7%), and 2 (5.1%) in each of robotics, electrode placement performance, and biomaterials performance.

CONCLUSION

The relationship between CI and artificial intelligence is strengthening with a recent increase in publications reporting successful applications. Considerable effort has been directed toward augmenting signal processing and automating postoperative MAPping using machine learning algorithms. Other promising applications include augmenting CI surgery mechanics and personalized medicine approaches for boosting CI patient performance. Future opportunities include addressing scalability and the research and clinical communities' acceptance of machine learning algorithms as effective techniques.

Evaluation of Possible Effects of a Potassium Channel Modulator on Temporal Processing by Cochlear Implant Listeners

Temporal processing by cochlear implant listeners is degraded and is affected by auditory deprivation. The fast-acting Kv3.1 potassium channel is important for sustained temporally accurate firing and is also susceptible to deprivation, the effects of which can be partially restored in animals by the molecule AUT00063. We report the results of a randomised placebo-controlled double-blind study on psychophysical tests of the effects of AUT00063 on temporal processing by CI listeners. The study measured the upper limit of temporal pitch, gap detection, and discrimination of low rates (centred on 120 pps) for monopolar pulse trains presented to an apical electrode. The upper limit was measured using the optimally efficient midpoint comparison (MPC) pitch-ranking procedure; thresholds were obtained for the other two measures using an adaptive procedure. Twelve CI users (MedEl and Cochlear) were tested before and after two periods of AUT00063 or placebo in a within-subject crossover study. No significant differences occurred between post-drug and post-placebo conditions. This absence of effect occurred despite high test-retest reliability for all three measures, obtained by comparing performance on the two baseline visits, and despite the demonstrated sensitivity of the measures to modest changes in temporal processing obtained in other studies from our laboratory. Hence, we have no evidence that AUT00063 improves temporal processing for the doses and patient population employed.

Intracochlear near infrared stimulation: Feasibility of optoacoustic stimulation in vivo

Intracochlear optical stimulation has been suggested as an alternative approach to hearing prosthetics in recent years. This study investigated the properties of a near infrared laser (NIR) induced optoacoustic effect. Pressure recordings were performed at the external meatus of anaesthetized guinea pigs during intracochlear NIR stimulation. The sound pressure and power spectra were determined. The results were compared to multi unit responses in the inferior colliculus (IC). Additionally, the responses to NIR stimulation were compared to IC responses induced by intracochlear electric stimulation at the same cochlear position to investigate a potentially confounding contribution of direct neural NIR stimulation. The power spectra of the sound recorded at the external meatus (n = 7) had most power at frequencies below 10 kHz and showed little variation for different stimulation sites. The mean spike rates of IC units responding to intracochlear NIR stimulation (n = 222) of 17 animals were significantly correlated with the power of the externally recorded signal at frequencies corresponding to the best frequencies of the IC units. The response strength as well as the sound pressure at the external meatus depended on the pulse peak power of the optical stimulus. The sound pressure recorded at the external meatus reached levels above 70 dB SPL peak equivalent. In hearing animals a cochlear activation apical to the location of the fiber was found. The absence of any NIR responses after pharmacologically deafening and the comparison to electric stimulation at the NIR stimulation site revealed no indication of a confounding direct neural NIR stimulation. Intracochlear optoacoustic stimulation might become useful in combined electro-acoustic stimulation devices in the future.

Technology and the Future of Cochlear Implants

Cochlear implants are one of the dramatic success stories of the bioengineering enterprise. Although these prostheses are used extensively, they still can be improved substantially. We suggest that high-density electrode designs will permit field shaping and field steering to an extent not presently possible with the arrays that are used today. Those opportunities will make it possible to make use of the phase information that is richly available to normal listeners. Although this information makes possible more precise location of sound sources in the auditory environment, and will likely improve the recognition of intelligent sound in noise, it will require the consumption of additional power in future cochlear prostheses. Those opportunities and trade-offs provide the designers of cochlear implants with exciting goals for the future.

Evaluation of focused multipolar stimulation for cochlear implants in acutely deafened cats

OBJECTIVE

The conductive nature of the fluids and tissues of the cochlea can lead to broad activation of spiral ganglion neurons using contemporary cochlear implant stimulation configurations such as monopolar (MP) stimulation. The relatively poor spatial selectivity is thought to limit implant performance, particularly in noisy environments. Several current focusing techniques have been proposed to reduce the spread of activation with the aim towards achieving improved clinical performance.

APPROACH

The present research evaluated the efficacy of focused multipolar (FMP) stimulation, a relatively new focusing technique in the cochlea, and compared its efficacy to both MP stimulation and tripolar (TP) stimulation. The spread of neural activity across the inferior colliculus (IC), measured by recording the spatial tuning curve, was used as a measure of spatial selectivity. Adult cats (n = 6) were acutely deafened and implanted with an intracochlear electrode array before multi-unit responses were recorded across the cochleotopic gradient of the contralateral IC. Recordings were made in response to acoustic and electrical stimulation using the MP, TP and FMP configurations.

MAIN RESULTS

FMP and TP stimulation resulted in greater spatial selectivity than MP stimulation. However, thresholds were significantly higher (p < 0.001) for FMP and TP stimulation compared to MP stimulation. There were no differences found in spatial selectivity and threshold between FMP and TP stimulation.

SIGNIFICANCE

The greater spatial selectivity of FMP and TP stimulation would be expected to result in improved clinical performance. However, further research will be required to demonstrate the efficacy of these modes of stimulation after longer durations of deafness.

Cochlear-implant spatial selectivity with monopolar, bipolar and tripolar stimulation

Sharp spatial selectivity is critical to auditory performance, particularly in pitch-related tasks. Most contemporary cochlear implants have employed monopolar stimulation that produces broad electric fields, which presumably contribute to poor pitch and pitch-related performance by implant users. Bipolar or tripolar stimulation can generate focused electric fields but requires higher current to reach threshold and, more interestingly, has not produced any apparent improvement in cochlear-implant performance. The present study addressed this dilemma by measuring psychophysical and physiological spatial selectivity with both broad and focused stimulations in the same cohort of subjects. Different current levels were adjusted by systematically measuring loudness growth for each stimulus, each stimulation mode, and in each subject. Both psychophysical and physiological measures showed that, although focused stimulation produced significantly sharper spatial tuning than monopolar stimulation, it could shift the tuning position or even split the tuning tips. The altered tuning with focused stimulation is interpreted as a result of poor electrode-to-neuron interface in the cochlea, and is suggested to be mainly responsible for the lack of consistent improvement in implant performance. A linear model could satisfactorily quantify the psychophysical and physiological data and derive the tuning width. Significant correlation was found between the individual physiological and psychophysical tuning widths, and the correlation was improved by log-linearly transforming the physiological data to predict the psychophysical data. Because the physiological measure took only one-tenth of the time of the psychophysical measure, the present model is of high clinical significance in terms of predicting and improving cochlear-implant performance.

Cochlear infrastructure for electrical hearing

Although the cochlear implant is already the world's most successful neural prosthesis, opportunities for further improvement abound. Promising areas of current research include work on improving the biological infrastructure in the implanted cochlea to optimize reception of cochlear implant stimulation and on designing the pattern of electrical stimulation to take maximal advantage of conditions in the implanted cochlea. In this review we summarize what is currently known about conditions in the cochlea of deaf, implanted humans and then review recent work from our animal laboratory investigating the effects of preserving or reinnervating tissues on psychophysical and electrophysiological measures of implant function. Additionally we review work from our human laboratory on optimizing the pattern of electrical stimulation to better utilize strengths in the cochlear infrastructure. Histological studies of human temporal bones from implant users and from people who would have been candidates for implants show a range of pathologic conditions including spiral ganglion cell counts ranging from approximately 2% to 92% of normal and partial hair cell survival in some cases. To duplicate these conditions in a guinea pig model, we use a variety of deafening and implantation procedures as well as post-deafening therapies designed to protect neurons and/or regenerate neurites. Across populations of human patients, relationships between nerve survival and functional measures such as speech have been difficult to demonstrate, possibly due to the numerous subject variables that can affect implant function and the elapsed time between functional measures and postmortem histology. However, psychophysical studies across stimulation sites within individual human subjects suggest that biological conditions near the implanted electrodes contribute significantly to implant function, and this is supported by studies in animal models comparing histological findings to psychophysical and electrophysiological data. Results of these studies support the efforts to improve the biological infrastructure in the implanted ear and guide strategies which optimize stimulation patterns to match patient-specific conditions in the cochlea.

Combining acoustic and electric stimulation in the service of speech recognition

The majority of recently implanted, cochlear implant patients can potentially benefit from a hearing aid in the ear contralateral to the implant. When patients combine electric and acoustic stimulation, word recognition in quiet and sentence recognition in noise increase significantly. Several studies suggest that the acoustic information that leads to the increased level of performance resides mostly in the frequency region of the voice fundamental, e.g. 125 Hz for a male voice. Recent studies suggest that this information aids speech recognition in noise by improving the recognition of lexical boundaries or word onsets. In some noise environments, patients with bilateral implants can achieve similar levels of performance as patients who combine electric and acoustic stimulation. Patients who have undergone hearing preservation surgery, and who have electric stimulation from a cochlear implant and who have low-frequency hearing in both the implanted and not-implanted ears, achieve the best performance in a high noise environment.

Partial tripolar cochlear implant stimulation: Spread of excitation and forward masking in the inferior colliculus

This study examines patterns of neural activity in response to single biphasic electrical pulses, presented alone or following a forward masking pulse train, delivered by a cochlear implant. Recordings were made along the tonotopic axis of the central nucleus of the inferior colliculus (ICC) in ketamine/xylazine anesthetized guinea pigs. The partial tripolar electrode configuration was used, which provided a systematic way to vary the tonotopic extent of ICC activation between monopolar (broad) and tripolar (narrow) extremes while maintaining the same peak of activation. The forward masking paradigm consisted of a 200 ms masker pulse train (1017 pulses per second) followed 10 ms later by a single-pulse probe stimulus; the current fraction of the probe was set to 0 (monopolar), 1 (tripolar), or 0.5 (hybrid), and the fraction of the masker was fixed at 0.5. Forward masking tuning profiles were derived from the amount of masking current required to just suppress the activity produced by a fixed-level probe. These profiles were sharper for more focused probe configurations, approximating the pattern of neural activity elicited by single (non-masked) pulses. The result helps to bridge the gap between previous findings in animals and recent psychophysical data.

Current steering and current focusing in cochlear implants: comparison of monopolar, tripolar, and virtual channel electrode configurations

OBJECTIVES

To compare the effects of Monopole (Mono), Tripole (Tri), and "Virtual channel" (Vchan) electrode configurations on spectral resolution and speech perception in a crossover design.

DESIGN

Nine experienced adults who received an Advanced Bionics CII/90K cochlear implant participated in a crossover design using three experimental strategies for 2 wk each. Three strategies were compared: (1) Mono; (2) Tri with current partly returning to adjacent electrodes and partly (25 or 75%) to the extracochlear reference; and (3) a monopolar "Vchan" strategy creating seven intermediate channels between two contacts. Each strategy was a variant of the standard "HiRes" processing strategy using 14 channels and 1105 pulses/sec/ channel, and a pulse duration of 32 microsec/phase. Spectral resolution was measured using broadband noise with a sinusoidally rippled spectral envelope with peaks evenly spaced on a logarithmic frequency scale. Speech perception was measured for monosyllables in quiet and in steady-state and fluctuating noises. Subjective comments on music experience and preferences in everyday use were assessed through questionnaires.

RESULTS

Thresholds and most comfortable levels with Mono and Vchan were both significantly lower than levels with Tri. Spectral resolution was significantly higher with Tri than with Mono; spectral resolution with Vchan did not differ significantly from the other configurations. Moderate but significant correlations between word recognition and spectral resolution were found in speech in quiet and fluctuating noise. For speech in quiet, word recognition was best with Mono and worst with Vchan; Tri did not significantly differ from the other configurations. Pooled across the noise conditions, word recognition was best with Tri and worst with Vchan (Mono did not significantly differ from the other configurations). These differences were small and insufficient to result in a clear increase in performance across subjects if the result from the best configuration per subject was compared with the result from Mono. Across all subjects, music appreciation and satisfaction in everyday use did not clearly differ between configurations.

CONCLUSIONS

(1) Although spectral resolution was improved with the tripolar configuration, differences in speech performance were too small in this limited group of subjects to justify clinical introduction. (2) Overall spectral resolution remained extremely poor compared with normal hearing; it remains to be seen whether further manipulations of the electrical field will be more effective.

Cochlear-implant high pulse rate and narrow electrode configuration impair transmission of temporal information to the auditory cortex

In the most commonly used cochlear prosthesis systems, temporal features of sound are signaled by amplitude modulation of constant-rate pulse trains. Several convincing arguments predict that speech reception should be optimized by use of pulse rates > or approximately 2,000 pulses per second (pps) and by use of intracochlear electrode configurations that produce restricted current spread (e.g., bipolar rather than monopolar configurations). Neither of those predictions has been borne out in consistent improvements in speech reception. Neurons in the auditory cortex of anesthetized guinea pigs phase lock to the envelope of sine-modulated electric pulse trains presented through a cochlear implant. The present study used that animal model to quantify the effects of carrier pulse rate, electrode configuration, current level, and modulator wave shape on transmission of temporal information from a cochlear implant to the auditory cortex. Modulation sensitivity was computed using a signal-detection analysis of cortical phase-locking vector strengths. Increasing carrier pulse rate in 1-octave steps from 254 to 4,069 pps resulted in systematic decreases in sensitivity. Comparison of sine- versus square-wave modulator waveforms demonstrated that some, but not all, of the loss of modulation sensitivity at high pulse rates was a result of the decreasing size of pulse-to-pulse current steps at the higher rates. Use of a narrow bipolar electrode configuration, compared with the monopolar configuration, produced a marked decrease in modulation sensitivity. Results from this animal model suggest explanations for the failure of high pulse rates and/or bipolar electrode configurations to produce hoped-for improvements in speech reception.

Current focusing and steering: modeling, physiology, and psychophysics

Current steering and current focusing are stimulation techniques designed to increase the number of distinct perceptual channels available to cochlear implant (CI) users by adjusting currents applied simultaneously to multiple CI electrodes. Previous studies exploring current steering and current focusing stimulation strategies are reviewed, including results of research using computational models, animal neurophysiology, and human psychophysics. Preliminary results of additional neurophysiological and human psychophysical studies are presented that demonstrate the success of current steering strategies in stimulating auditory nerve regions lying between physical CI electrodes, as well as current focusing strategies that excite regions narrower than those stimulated using monopolar configurations. These results are interpreted in the context of perception and speech reception by CI users. Disparities between results of physiological and psychophysical studies are discussed. The differences in stimulation used for physiological and psychophysical studies are hypothesized to contribute to these disparities. Finally, application of current steering and focusing strategies to other types of auditory prostheses is also discussed.

Penetrating multichannel stimulation and recording electrodes in auditory prosthesis research

Microelectrode arrays offer the auditory systems physiologists many opportunities through a number of electrode technologies. In particular, silicon substrate electrode arrays offer a large design space including choice of layout plan, range of surface areas for active sites, a choice of site materials and high spatial resolution. Further, most designs can double as recording and stimulation electrodes in the same preparation. Scala tympani auditory prosthesis research has been aided by mapping electrodes in the cortex and the inferior colliculus to assess the CNS responses to peripheral stimulation. More recently silicon stimulation electrodes placed in the auditory nerve, cochlear nucleus and the inferior colliculus have advanced the exploration of alternative stimulation sites for auditory prostheses. Multiplication of results from experimental effort by simultaneously stimulating several locations, or by acquiring several streams of data synchronized to the same stimulation event, is a commonly sought after advantage. Examples of inherently multichannel functions which are not possible with single electrode sites include (1) current steering resulting in more focused stimulation, (2) improved signal-to-noise ratio (SNR) for recording when noise and/or neural signals appear on more than one site and (3) current source density (CSD) measurements. Still more powerful are methods that exploit closely-spaced recording and stimulation sites to improve detailed interrogation of the surrounding neural domain. Here, we discuss thin-film recording/stimulation arrays on silicon substrates. These electrode arrays have been shown to be valuable because of their precision coupled with reproducibility in an ever expanding design space. The shape of the electrode substrate can be customized to accommodate use in cortical, deep and peripheral neural structures while flexible cables, fluid delivery and novel coatings have been added to broaden their application. The use of iridium oxide as the neural interface site material has increased the efficiency of charge transfer for stimulation and lowered impedance for recording electrodes.

Cochlear implant electrode configuration effects on activation threshold and tonotopic selectivity

The multichannel design of contemporary cochlear implants (CIs) is predicated on the assumption that each channel activates a relatively restricted and independent sector of the deaf auditory nerve array, just as a sound within a restricted frequency band activates a restricted region of the normal cochlea The independence of CI channels, however, is limited; and the factors that determine their independence, the relative overlap of the activity patterns that they evoke, are poorly understood. In this study, we evaluate the spread of activity evoked by cochlear implant channels by monitoring activity at 16 sites along the tonotopic axis of the guinea pig inferior colliculus (IC). "Spatial tuning curves" (STCs) measured in this way serve as an estimate of activation spread within the cochlea and the ascending auditory pathways. We contrast natural stimulation using acoustic tones with two kinds of electrical stimulation either (1) a loose fitting banded array consisting of a cylindrical silicone elastomer carrier with a linear series of ring contacts; or (2) a space-filling array consisting of a tapered silicone elastomer carrier that is designed to fit snugly into the guinea pig scala tympani with a linear series of ball contacts positioned along it Spatial tuning curves evoked by individual acoustic tones, and by activation of each contact of each array as a monopole, bipole or tripole were recorded. Several channel configurations and a wide range of electrode separations were tested for each array, and their thresholds and selectivity were estimated. The results indicate that the tapered space-filling arrays evoked more restricted activity patterns at lower thresholds than did the banded arrays. Monopolar stimulation (one intracochlear contact activated with an extracochlear return) using either array evoked broad activation patterns that involved the entire recording array at current levels <6dBSL, but at relatively low thresholds. Bi- and tri-polar configurations of both array types evoked more restricted activity patterns, but their thresholds were higher than those of monopolar configurations. Bipolar and tripolar configurations with closely spaced contacts evoked activity patterns that were comparable to those evoked by pure tones. As the spacing of bipolar electrodes was increased (separations >1mm), the activity patterns became broader and evoked patterns with two distinct threshold minima, one associated with each contact.

Evidence of a tonotopic organization of the auditory cortex in cochlear implant users

Deprivation from normal sensory input has been shown to alter tonotopic organization of the human auditory cortex. In this context, cochlear implant subjects provide an interesting model in that profound deafness is made partially reversible by the cochlear implant. In restoring afferent activity, cochlear implantation may also reverse some of the central changes related to deafness. The purpose of the present study was to address whether the auditory cortex of cochlear implant subjects is tonotopically organized. The subjects were thirteen adults with at least 3 months of cochlear implant experience. Auditory event-related potentials were recorded in response to electrical stimulation delivered at different intracochlear electrodes. Topographic analysis of the auditory N1 component (approximately 85 ms latency) showed that the locations on the scalp and the relative amplitudes of the positive/negative extrema differ according to the stimulated electrode, suggesting that distinct sets of neural sources are activated. Dipole modeling confirmed electrode-dependent orientations of these sources in temporal areas, which can be explained by nearby, but distinct sites of activation in the auditory cortex. Although the cortical organization in cochlear implant users is similar to the tonotopy found in normal-hearing subjects, some differences exist. Nevertheless, a correlation was found between the N1 peak amplitude indexing cortical tonotopy and the values given by the subjects for a pitch scaling task. Hence, the pattern of N1 variation likely reflects how frequencies are coded in the brain.

Design and fabrication of multichannel cochlear implants for animal research

The effectiveness of multichannel cochlear implants depends on the activation of perceptually distinct regions of the auditory nerve. Increased information transfer is possible as the number of channels and dynamic range are increased and electrical and neural interaction among channels is reduced. Human and animal studies have demonstrated that specific design features of the intracochlear electrode directly affect these performance factors. These features include the geometry, size, and orientation of the stimulating sites, proximity of the device to spiral ganglion neurons, shape and position of the insulating carrier, and the stimulation mode (monopolar, bipolar, etc.). Animal studies to directly measure the effects of changes in electrode design are currently constrained by the lack of available electrodes that model contemporary clinical devices. This report presents methods to design and fabricate species-specific customizable electrode arrays. We have successfully implanted these arrays in guinea pigs and cats for periods of up to 14 months and have conducted acute electrophysiological experiments in these animals. Modifications enabling long-term intracochlear drug infusion are also described. Studies using these scale model arrays will improve our understanding of how these devices function in human subjects and how we can best optimize future cochlear implants.

Threshold and channel interaction in cochlear implant users: evaluation of the tripolar electrode configuration

The efficacy of cochlear implants is limited by spatial and temporal interactions among channels. This study explores the spatially restricted tripolar electrode configuration and compares it to bipolar and monopolar stimulation. Measures of threshold and channel interaction were obtained from nine subjects implanted with the Clarion HiFocus-I electrode array. Stimuli were biphasic pulses delivered at 1020 pulses/s. Threshold increased from monopolar to bipolar to tripolar stimulation and was most variable across channels with the tripolar configuration. Channel interaction, quantified by the shift in threshold between single- and two-channel stimulation, occurred for all three configurations but was largest for the monopolar and simultaneous conditions. The threshold shifts with simultaneous tripolar stimulation were slightly smaller than with bipolar and were not as strongly affected by the timing of the two channel stimulation as was monopolar. The subjects' performances on clinical speech tests were correlated with channel-to-channel variability in tripolar threshold, such that greater variability was related to poorer performance. The data suggest that tripolar channels with high thresholds may reveal cochlear regions of low neuron survival or poor electrode placement.

Using evoked potentials to match interaural electrode pairs with bilateral cochlear implants

Bilateral cochlear implantation seeks to restore the advantages of binaural hearing to the profoundly deaf by providing binaural cues normally important for accurate sound localization and speech reception in noise. Psychophysical observations suggest that a key issue for the implementation of a successful binaural prosthesis is the ability to match the cochlear positions of stimulation channels in each ear. We used a cat model of bilateral cochlear implants with eight-electrode arrays implanted in each cochlea to develop and test a noninvasive method based on evoked potentials for matching interaural electrodes. The arrays allowed the cochlear location of stimulation to be independently varied in each ear. The binaural interaction component (BIC) of the electrically evoked auditory brainstem response (EABR) was used as an assay of binaural processing. BIC amplitude peaked for interaural electrode pairs at the same relative cochlear position and dropped with increasing cochlear separation in either direction. To test the hypothesis that BIC amplitude peaks when electrodes from the two sides activate maximally overlapping neural populations, we measured multiunit neural activity along the tonotopic gradient of the inferior colliculus (IC) with 16-channel recording probes and determined the spatial pattern of IC activation for each stimulating electrode. We found that the interaural electrode pairings that produced the best aligned IC activation patterns were also those that yielded maximum BIC amplitude. These results suggest that EABR measurements may provide a method for assigning frequency-channel mappings in bilateral implant recipients, such as pediatric patients, for which psychophysical measures of pitch ranking or binaural fusion are unavailable.

Phosphene induction by microstimulation of macaque V1

Non-human primates are being used to develop a cortical visual prosthesis for the blind. We use the properties of electrical microstimulation of striate cortex (area V1) of macaque monkeys to make inferences about phosphene induction. Our analysis is based on well-established properties of V1: retino-cortical magnification factor, receptive-field size, and the characteristics of hypercolumns. We argue that phosphene size is dependent on the amount of current delivered to V1 and on the retino-cortical magnification factor. We suggest that to improve the correspondence between the site of stimulation within V1 and the visual field location of an elicited phosphene both eyes must be put under experimental control given that phosphene location is retinocentric and given that the vergence angle between the eyes might affect the position of a phosphene in depth. Knowing how electrical microstimulation interacts with cortical tissue to evoke percepts in behaving macaque monkeys is fundamental to the establishment of an effective cortical visual prosthesis for the blind.

Direct and indirect activation of cortical neurons by electrical microstimulation

Electrical microstimulation has been used to elucidate cortical function. This review discusses neuronal excitability and effective current spread estimated by using three different methods: 1) single-cell recording, 2) behavioral methods, and 3) functional magnetic resonance imaging (fMRI). The excitability properties of the stimulated elements in neocortex obtained using these methods were found to be comparable. These properties suggested that microstimulation activates the most excitable elements in cortex, that is, by and large the fibers of the pyramidal cells. Effective current spread within neocortex was found to be greater when measured with fMRI compared with measures based on single-cell recording or behavioral methods. The spread of activity based on behavioral methods is in close agreement with the spread based on the direct activation of neurons (as opposed to those activated synaptically). We argue that the greater activation with imaging is attributed to transynaptic spread, which includes subthreshold activation of sites connected to the site of stimulation. The definition of effective current spread therefore depends on the neural event being measured.

Speech perception with mono- and quadrupolar electrode configurations: a crossover study

OBJECTIVE

To study the effect of two multipolar electrode configurations on speech perception, pitch perception, and the intracochlear electrical field.

STUDY DESIGN

Crossover design; within subject.

SETTING

Tertiary referral center.

PATIENTS

Eight experienced adult cochlear implant users.

INTERVENTION

Each subject used each of three experimental processors for 3 weeks. The following processors were compared that differed only in electrode configuration: 1) monopolar; 2) hybrid quadrupolar, in which half of the current returned to the extracochlear reference electrode and half to two electrodes immediately to the left and right of the active electrode; and 3) flat tripolar +2, which directed all the current to four reference electrodes (two on each side), separated from the active electrode by two inactive electrodes. All the processors used the standard Advanced Bionics HiRes speech-processing strategy, 12 channels, 1,220 pulses per second per channel, and with a pulse width of 33 (micros/phase).

RESULTS

The monopolar processors had the largest stimulation efficiency and the smallest dynamic range in linear current units. The reverse was true of flat tripolar +2 processor, whereas the hybrid quadrupolar processor fell in between. Insufficient loudness growth prevented the use of the flat tripolar +2 processor in three subjects. Word recognition did not differ between the clinically used 16-channel monopolar processor and the experimental monopolar processor, regardless of the differences in the number of channels, pulse rate, and duration of experience. Word recognition with the flat tripolar +2 processor was significantly poorer than with the monopolar and hybrid quadrupolar processors; monopolar and quadrupolar processors did not differ. There was no significant interaction between processor type and competing noise type (stationary or fluctuating), but performance at the higher level of fluctuating noise was best with the hybrid quadrupolar processor in almost all the subjects. Pitch scaling showed ceiling performance in five subjects and differed between processors in the two other subjects with imperfect tonotopy. Intracochlear current spread was considerable with the monopolar configuration; it was reduced with the hybrid quadrupolar configuration and virtually absent beyond the active electrodes with the tripolar configuration.

CONCLUSION

More confined configurations reduced the longitudinal width of the electrical field, which was expected to enhance channel separation, but no improvement in word recognition was found. More research is needed to test confined configurations that have enhanced efficiency and to evaluate the fundamental effects of configuration on channel discriminability.

Redundancy and synergy of neuronal ensembles in motor cortex

We examined the ability of neuronal ensembles from rat motor cortex to predict behavioral performance during a reaction time task. We found that neurons that were the best individual predictors of task performance were not necessarily the neurons that contributed the most predictive information to an ensemble of neurons. To understand this result, we applied a framework for quantifying statistical relationships between neurons (Schneidman et al., 2003) to all possible combinations of neurons within our ensembles. We found that almost all neurons (96%) contributed redundant predictive information to the ensembles. This redundancy resulted in the maintenance of predictive information despite the removal of many neurons from each ensemble. Moreover, the balance of synergistic and redundant interactions depended on the number of neurons in the ensemble. Small ensembles could exhibit synergistic interactions (e.g., 23 +/- 9% of ensembles with two neurons were synergistic). In contrast, larger ensembles exhibited mostly redundant interactions (e.g., 99 +/- 0.1% of ensembles with eight neurons were redundant). We discuss these results with regard to constraints on interpreting neuronal ensemble data and with respect to motor cortex involvement in reaction time performance.

Loudness adaptation in acoustic and electric hearing

The present study is aimed to evaluate and compare loudness adaptation between normal hearing and cochlear-implant subjects. Loudness adaptation for 367-s pure tones was measured in five normal-hearing subjects at three frequencies (125, 1,000, and 8,000 Hz) and three levels (30, 60, and 90 dB SPL). In addition, loudness adaptation for 367-s pulse trains was measured in five Clarion cochlear-implant subjects at three stimulation rates (100, 991, and 4,296 Hz), three levels (10, 50, and 90% of the electric dynamic range), three stimulation positions (apical, middle and basal), and two stimulation modes (monopolar and bipolar). The method of successive magnitude estimation was used to quantify loudness adaptation. Similar to the previous results, we found that loudness adaptation in normal-hearing subjects increases with decreasing level and increasing frequency. However, we also found a small but significant loudness enhancement at 90 dB SPL in acoustic hearing. Despite large individual variability, we found that loudness adaptation in cochlear-implant subjects increases with decreasing levels, but is not significantly affected by the rate, place and mode of stimulation. A phenomenological model was proposed to predict loudness adaptation as a function of stimulus frequency and level in acoustic hearing. The present results were not fully compatible with either the restricted excitation hypothesis or the neural adaptation hypothesis. Loudness adaptation may have a central component that is dependent on the peripheral excitation pattern.

Cochlear electrode arrays: past, present and future

Cochlear implants are very successful devices: more than 60000 people use them throughout the world. Key to the success of these prostheses is the development of electrode arrays that place contacts close to the target neurons, survive for decades in the tissues of the inner ear, and that provide reliable and repeatable excitation to the cells of the auditory nerve. This article describes the early electrode arrays and their development into the arrays that are used presently in clinical cochlear prostheses. While integrated circuit techniques were proposed and tested in the laboratory two decades ago, the present clinical devices still are hand built and made of wire-based technologies. Current approaches that seek to automate the construction of cochlear electrode arrays are described and discussed.

Cochlear implants: the view from the brain

The cochlear implant arguably is the most successful neural prosthesis. Studies of the responses of the central auditory system to prosthetic electrical stimulation of the cochlea are revealing the success with which electrical stimulation of a deaf ear can mimic acoustic stimulation of a normal-hearing ear. Understanding of the physiology of central auditory structures can lead to improved restoration of hearing with cochlear implants. In turn, the cochlear implant can be exploited as an experimental tool for examining central hearing mechanisms isolated from the effects of cochlear mechanics and transduction.

Categorical perception of speech stimuli in children at risk for reading difficulty

Children determined to be at risk (n = 24) or not at risk (n = 13) for reading difficulty listened to tokens from a voice onset time (VOT) (/ga/-/ka/) or tone series played in a continuous unbroken rhythm. Changes between tokens occurred at random intervals and children were asked to press a button as soon as they detected a change. For the VOT series, at-risk children were less sensitive than not-at-risk children to changes between tokens that crossed the phonetic boundary. Maps of group stimulus space produced using multidimensional scaling of reaction times for the VOT series indicated that at-risk children may attend less to the phonological information available in the speech stimuli and more to subtle acoustic differences between phonetically similar stimuli than not-at-risk children. Better phonological processing was associated with greater sensitivity to changes between VOT tokens that crossed the phonetic boundary and greater relative weighting of the phonological compared to the acoustic dimension across both groups.

Topographic spread of inferior colliculus activation in response to acoustic and intracochlear electric stimulation

The design of contemporary multichannel cochlear implants is predicated on the presumption that they activate multiple independent sectors of the auditory nerve array. The independence of these channels, however, is limited by the spread of activation from each intracochlear electrode across the auditory nerve array. In this study, we evaluated factors that influence intracochlear spread of activation using two types of intracochlear electrodes: (1) a clinical-type device consisting of a linear series of ring contacts positioned along a silicon elastomer carrier, and (2) a pair of visually placed (VP) ball electrodes that could be positioned independently relative to particular intracochlear structures, e.g., the spiral ganglion. Activation spread was estimated by recording multineuronal evoked activity along the cochleotopic axis of the central nucleus of the inferior colliculus (ICC). This activity was recorded using silicon-based single-shank, 16-site recording probes, which were fixed within the ICC at a depth defined by responses to acoustic tones. After deafening, electric stimuli consisting of single biphasic electric pulses were presented with each electrode type in various stimulation configurations (monopolar, bipolar, tripolar) and/or various electrode orientations (radial, off-radial, longitudinal). The results indicate that monopolar (MP) stimulation with either electrode type produced widepread excitation across the ICC. Bipolar (BP) stimulation with banded pairs of electrodes oriented longitudinally produced activation that was somewhat less broad than MP stimulation, and tripolar (TP) stimulation produced activation that was more restricted than MP or BP stimulation. Bipolar stimulation with radially oriented pairs of VP ball electrodes produced the most restricted activation. The activity patterns evoked by radial VP balls were comparable to those produced by pure tones in normal-hearing animals. Variations in distance between radially oriented VP balls had little effect on activation spread, although increases in interelectrode spacing tended to reduce thresholds. Bipolar stimulation with longitudinally oriented VP electrodes produced broad activation that tended to broaden as the separation between electrodes increased.

Effects of cochlear-implant pulse rate and inter-channel timing on channel interactions and thresholds

Interactions among the multiple channels of a cochlear prosthesis limit the number of channels of information that can be transmitted to the brain. This study explored the influence on channel interactions of electrical pulse rates and temporal offsets between channels. Anesthetized guinea pigs were implanted with 2-channel scala-tympani electrode arrays, and spike activity was recorded from the auditory cortex. Channel interactions were quantified as the reduction of the threshold for pulse-train stimulation of the apical channel by sub-threshold stimulation of the basal channel. Pulse rates were 254 or 4069 pulses per second (pps) per channel. Maximum threshold reductions averaged 9.6 dB when channels were stimulated simultaneously. Among nonsimultaneous conditions, threshold reductions at the 254-pps rate were entirely eliminated by a 1966-micros inter-channel offset. When offsets were only 41 to 123 micros, however, maximum threshold shifts averaged 3.1 dB, which was comparable to the dynamic ranges of cortical neurons in this experimental preparation. Threshold reductions at 4069 pps averaged up to 1.3 dB greater than at 254 pps, which raises some concern in regard to high-pulse-rate speech processors. Thresholds for various paired-pulse stimuli, pulse rates, and pulse-train durations were measured to test possible mechanisms of temporal integration.

Cortical responses to cochlear implant stimulation: channel interactions

This study examined the interactions between electrical stimuli presented through two channels of a cochlear implant. Experiments were conducted in anesthetized guinea pigs. Multiunit spike activity recorded from the auditory cortex reflected the cumulative effects of electric field interactions in the cochlea as well as any neural interactions along the ascending auditory pathway. The cochlea was stimulated electrically through a 6-electrode intracochlear array. The stimulus on each channel was a single 80- micro s/phase biphasic pulse. Channel interactions were quantified as changes in the thresholds for elevation of cortical spike rates. Experimental parameters were interchannel temporal offset (0 to +/-2000 micro s), interelectrode cochlear spacing (1.5 or 2.25 mm), electrode configuration (monopolar, bipolar, or tripolar), and relative polarity between channels (same or inverted). In most conditions, presentation of a subthreshold pulse on one channel reduced the threshold for a pulse on a second channel. Threshold shifts were greatest for simultaneous pulses, but appreciable threshold reductions could persist for temporal offsets up to 640 micro s. Channel interactions varied strongly with electrode configuration: threshold shifts increased in magnitude in the order tripolar, bipolar, monopolar. Channel interactions were greater for closer electrode spacing. The results have implications for design of speech processors for cochlear implants.

Neuronal responses in cat primary auditory cortex to electrical cochlear stimulation: IV. Activation pattern for sinusoidal stimulation

Patterns of threshold distributions for single-cycle sinusoidal electrical stimulation and single pulse electrical stimulation were compared in primary auditory cortex of the adult cat. Furthermore, the effects of auditory deprivation on these distributions were evaluated and compared across three groups of adult cats. Threshold distributions for single and multiple unit responses from the middle cortical layers were obtained on the ectosylvian gyrus in an acutely implanted animal; 2 wk after deafening and implantation (short-term group); and neonatally deafened animals implanted following 2-5 yr of deafness (long-term group). For all three cases, we observed similar patterns of circumscribed regions of low response thresholds in the region of primary auditory cortex (AI). A dorsal and a ventral region of low response thresholds were found separated by a narrow, anterior-posterior strip of elevated thresholds. The ventral low-threshold regions in the short-term group were cochleotopically arranged. By contrast, the dorsal region in the short-term animals and both low-threshold regions in long-term deafened animals maintained only weak cochleotopicity. Analysis of the spatial extent of the low-threshold regions revealed that the activated area for sinusoidal stimulation was smaller and more circumscribed than for pulsatile stimulation for both dorsal and ventral AI. The width of the high-threshold ridge that separated the dorsal and ventral low-threshold regions was greater for sinusoidal stimulation. Sinusoidal and pulsatile threshold behavior differed significantly for electrode configurations with low and high minimum thresholds. Differences in threshold behavior and cortical response distributions between the sinusoidal and pulsatile stimulation suggest that stimulus shape plays a significant role in the activation of cortical activity. Differences in the activation pattern for short-term and long-term deafness reflect deafness-induced reorganizational changes based on factors such as differences in excitatory and inhibitory balance that are affected by the stimulation parameters.

Cortical representation of auditory space: information-bearing features of spike patterns

Previous studies have demonstrated that the spike patterns of cortical neurons vary systematically as a function of sound-source location such that the response of a single neuron can signal the location of a sound source throughout 360 degrees of azimuth. The present study examined specific features of spike patterns that might transmit information related to sound-source location. Analysis was based on responses of well-isolated single units recorded from cortical area A2 in alpha-chloralose-anesthetized cats. Stimuli were 80-ms noise bursts presented from loudspeakers in the horizontal plane; source azimuths ranged through 360 degrees in 20 degrees steps. Spike patterns were averaged across samples of eight trials. A competitive artificial neural network (ANN) identified sound-source locations by recognizing spike patterns; the ANN was trained using the learning vector quantization learning rule. The information about stimulus location that was transmitted by spike patterns was computed from joint stimulus-response probability matrices. Spike patterns were manipulated in various ways to isolate particular features. Full-spike patterns, which contained all spike-count information and spike timing with 100-micros precision, transmitted the most stimulus-related information. Transmitted information was sensitive to disruption of spike timing on a scale of more than approximately 4 ms and was reduced by an average of approximately 35% when spike-timing information was obliterated entirely. In a condition in which all but the first spike in each pattern were eliminated, transmitted information decreased by an average of only approximately 11%. In many cases, that condition showed essentially no loss of transmitted information. Three unidimensional features were extracted from spike patterns. Of those features, spike latency transmitted approximately 60% more information than that transmitted either by spike count or by a measure of latency dispersion. Information transmission by spike patterns recorded on single trials was substantially reduced compared with the information transmitted by averages of eight trials. In a comparison of averaged and nonaveraged responses, however, the information transmitted by latencies was reduced by only approximately 29%, whereas information transmitted by spike counts was reduced by 79%. Spike counts clearly are sensitive to sound-source location and could transmit information about sound-source locations. Nevertheless, the present results demonstrate that the timing of the first poststimulus spike carries a substantial amount, probably the majority, of the location-related information present in spike patterns. The results indicate that any complete model of the cortical representation of auditory space must incorporate the temporal characteristics of neuronal response patterns.

Auditory cortical images of cochlear-implant stimuli: dependence on electrode configuration

This study examines patterns of auditory cortical activity elicited by single-pulse cochlear implant stimuli that vary in electrode configuration, cochlear place of stimulation, and stimulus level. Recordings were made from the primary auditory cortex (area A1) of ketamine-anesthetized guinea pigs. The spatiotemporal pattern of neural spike activity was measured simultaneously across 16 cortical locations spanning approximately 2-3 octaves of the tonotopic axis. Such a pattern, averaged over 40 presentations of any particular stimulus, was defined as the "cortical image" of that stimulus. Acutely deafened guinea pigs were implanted with a 6-electrode animal version of the 22-electrode Nucleus banded electrode array (Cochlear). Cochlear electrode configurations consisted of monopolar (MP), bipolar (BP + N) with N inactive electrodes between the active and return electrodes (0 < or = N < or = 4), tripolar (TP) with one active electrode and two flanking return electrodes, and common ground (CG) with one active electrode and as many as five return electrodes. Cortical images typically showed a focus of maximum spike probability and minimum latency. Spike probabilities tended to decrease, and latencies tended to increase, with increasing cortical distance from that focus. Cortical images of TP stimuli were the most spatially compact, followed by BP + N images, and then MP images, which were the broadest. Images of CG stimuli were rather variable across animals and stimulus channels. The locations of cortical images shifted systematically from caudal to rostral as the cochlear place of stimulation changed from basal to apical. At the most sensitive cortical site for each condition, the dynamic ranges over which spike rates increased with increased current level were restricted to about 1-2 dB, regardless of configuration. Dynamic ranges tended to increase with increasing cortical distance from the most sensitive site. Electrode configurations that produced compact cortical images (e.g., TP and BP + 0) showed the greatest range of thresholds within each cortical image and the largest dynamic range at cortical sites removed from the most sensitive site.