Qian-Jie Fu - US grants

The long-term goal of this research is to understand the mechanisms involved in speech recognition by the electrically stimulated auditory system, and further, the plasticity of the auditory cortex. The present proposal will address three fundamental questions of speech perception in electric hearing: 1) How are the electrically evoked peripheral neural patterns affected by parametric variations of the speech processor? 2) Ho are the central speech pattern templates reshaped by new peripheral neural patterns? 3) What are the causes of the high variability in speech performance among cochlear implants patients? The hypothesis of this research is that speech recognition in electric hearing is primarily based on a similarity measure between electrically evoked peripheral neural patterns and central speech pattern templates. Based on patients' experience with the implant device, central speech pattern templates can accommodate new peripheral patterns, to some degree. We further hypothesize that a deficit in auditory resolution can remarkably reduce cochlear implant users' capabilities in speech pattern recognition. The significance of this research is not only of theoretical interest in understanding the mechanism involved in speech pattern recognition by electrical stimulation, but also of great clinical importance in designing appropriate speech processors and clinical fitting systems for cochlear implant listeners. It is also possible that principles derived from the present proposals, such as the relation between auditory resolution and speech performance, , are useful to understand the perception problem in hearing-impaired patients.

DESCRIPTION (provided by applicant): The long-term goals of the present proposal are to develop individualized, efficient and effective training protocols and materials to maximize cochlear implant patient performance in real-world listening conditions. We hypothesize that targeted contrast training can greatly improve cochlear implant patients'performance for a variety of simulated real-world listening conditions, and that individualized training protocols and materials may be necessary to maximize training outcomes. We further hypothesize that passive learning may not allow cochlear implant patients to receive the full benefit of advanced coding strategies, and that targeted auditory training may help patients access the additional cues provided by novel processing schemes. There are three specific aims in the proposed research. First, we propose to measure the benefits of auditory training in simulated real-world listening conditions, including speech perception in noise, speech perception via telephone, and music perception. Second, we propose to develop individualized, efficient and effective training protocols and materials to maximize training outcomes. Third, we propose to quantify interactions between auditory training and changes to speech processor parameters. While the proposed experiments will evaluate different aspects of auditory training somewhat independently, there are areas of overlapping consideration that may be combined to advance auditory rehabilitation techniques for cochlear implant patients. These proposed experiments have great theoretical significance, as the results will shed light on auditory plasticity in electric hearing and the importance of training when evaluating experimental speech processor parameters. The experiments have even greater clinical significance, as the results will provide strong evidence of the benefits of auditory training, hopefully leading to affordable, efficient and effective rehabilitation that cochlear implant patients can perform at home, using personal computers. We expect that these training approaches, if successful in real-world simulations, will generalize to improved performance outside the lab. As recent advances in implant technology seem to be reaching the point of diminishing returns, auditory training may provide the most cost-effective approach for cochlear implant patients to maximize the benefit of the implant device.

DESCRIPTION (provided by applicant): The long-term goal of this research is to improve cochlear implant patient performance by maximizing both the transmission and reception of acoustic patterns. We hypothesize that, due to the loss of fine spectral details, cochlear implant patients have great difficulty in challenging listening conditions (e.g., noise, competing speech, reverberation, unfamiliar talkers, etc.). We propose to optimize the input acoustic signal in response to the acoustic environment, or to different speaker characteristics, thereby improving the transmission of speech patterns. We further hypothesize that poor patient performance may be partly due to sub-optimal settings of important speech processor parameters (e.g., stimulation mode, frequency allocation, stimulation rate, etc.). We propose to optimize these parameters according to individual patients'psychophysical capabilities, thereby improving the reception of speech patterns. Combining these two approaches - pre-processing the input signal and optimizing processor parameters - will provide the greatest benefit to patient performance for a variety of listening conditions. There are three specific aims in the proposed research. Specific aim 1 is to improve the transmission of acoustic patterns. We will evaluate novel speech enhancement algorithms that optimize the input acustic patterns in response to the acoustic environment, or to different speaker characteristics. Specific aim 2 is to improve the reception of acoustic patterns. We will explore the perceptual space for important speech processor parameters and optimize these parameters according to individual patients'psychophysical capabilities. Specific aim 3 is to evaluate the long-term effects of changes to speech processing. While we will generally study the effects of each optimization technique independently in each experiment, the techniques can be easily combined to further optimize audio processing for cochlear implants, once the parameter space is defined. Each of the proposed strategies seeks to optimize some aspect of speech processing and, when combined, the benefit from one strategy may be directly enhanced by the benefit from another. This synergy may further improve patient performance for a wide variety of listening conditions. The proposed research is of great clinical importance in terms of maximizing patient performance under a variety of listening conditions. It is also of great theoretical interest in terms of understanding the neural and perceptual mechanisms involved in pattern recognition.

SUMMARY: Bilateral cochlear implant (CI) patients and CI patients with single-sided deafness (SSD) must integrate spectral patterns that might be quite different across ears. Due to interactions between the acoustic- to-electric frequency allocation and the limited extent/insertion depth of the electrode array, CI patients may experience an intra-aural mismatch between the acoustic input frequency and the electrode place within an implanted ear. Bilateral and SSD CI patients may also experience inter-aural mismatch between the frequency-place allocation in each, which may limit binaural benefits for speech and localization. Radiological imaging can estimate electrode positions within the cochlea, but cannot characterize the electrode-neural interface (the proximity of healthy neurons to the electrode), which is the ultimate arbiter of frequency mismatch. Single-channel inter-aural pitch-matching may not fully characterize the effects of mismatch for multi-channel speech perception. Given its additive properties, using band-limited, non-redundant speech (rather than broadband speech) may provide greater insight regarding the effects of frequency mismatch on spectral integration. Optimal bandwidths and frequency ranges for non-redundant speech may be estimated from frequency importance functions. The long-term goal of this proposal is to better understand how frequency mismatch affects spectral integration within and across ears. It is hypothesized that frequency importance functions may be quite different between CI and normal-hearing (NH) listeners, and may be affected by frequency mismatch. It is also hypothesized that the effects of frequency mismatch on spectral integration can be better estimated in noise using band-limited, non-redundant speech. Finally, it is hypothesized that the degree of inter-aural mismatch in CI patients can be accurately and efficiently estimated using complementary, non-redundant speech information presented to each ear. Three aims are proposed to better explore how spectral integration is affected by frequency mismatch. Aim 1 will explore how frequency mismatch affects the frequency importance function for speech intelligibility. Aim 2 will explore how frequency mismatch affects spectral integration within and across ears in NH subjects listening to unilateral, bilateral, and SSD CI simulations, and further evaluate a novel technique to estimate inter-aural mismatch by delivering, complementary, band-limited, non-redundant speech cues to each ear. Aim 3 will use the above novel technique to estimate inter-aural mismatch in real CI patients and further optimize the frequency allocation to reduce inter-aural mismatch. The proposed research is of great theoretical interest, as it will provide greater insights into factors that limit binaural integration in bilateral and SSD CI patients. The proposed research is also of great clinical value, as the results may provide useful clinical tools to accurately and efficiently optimize CI frequency allocations to maximize binaural benefits for bilateral and SSD CI users.

SUMMARY: Increasing numbers of cochlear implant (CI) patients are able to combine residual acoustic hearing (AH) with electric hearing (EH) in the same ear (electric-acoustic stimulation, or EAS), across ears (bimodal), or both (biEAS). The benefits of acoustic-electric hearing (AEH) are variable; some CI users even experience interference between acoustic and electric stimulation patterns. In the clinic, there is little effort to optimize CI and/or hearing aid signal processing with regard to combined acoustic and electric hearing, and it is unclear whether acoustic and electric patterns are combined differently within or across ears. EAS and biEAS patients may experience energetic interference between the current spread from EH and the spread of excitation (SOE) from AH. Bimodal, EAS, and biEAS patients may also experience informational interference because the same acoustic input may be delivered to different cochlear places (tonotopic mismatch) within and/or across ears. To mitigate these adverse effects, the optimal fitting for CI and/or hearing aid devices might be quite different for AEH than for EH and/or AH. The long-term goals of this proposal are to understand the mechanisms that underlie integration of acoustic and electric hearing, which can be used to improve AEH benefits for CI users. We hypothesize that acoustic and electric stimulation patterns are integrated differently within and across ears. Specifically, we hypothesize that integration for EAS listeners is limited by both energetic and informational interference; reducing the current spread/SOE between AH and EH and reducing tonotopic mismatch may improve integration. We also hypothesize that integration for bimodal listeners is limited by informational interference, due to tonotopic mismatch and sound quality differences across ears; improving the sound quality of EH (e.g., focused stimulation) while reducing the tonotopic mismatch between AH and EH may improve integration. Acoustic-electric integration will be evaluated in pairs of parallel experiments with normal-hearing (NH) subjects listening to simulations of AEH and in real EAS, bimodal, and biEAS CI listeners. Speech measures (vowel and sentence recognition in quiet and in noise, vocal emotion recognition) and frequency resolution will be collected with AH, EH, and AEH, and evaluated in terms of overall AEH performance, AEH benefit (performance difference between AEH and AH or EH), and integration efficiency (the ratio between observed and predicted AEH performance). Aim 1 will primarily explore aspects of electric hearing that affect acoustic-electric integration while Aim 2 will focus on aspects of acoustic hearing that affect acoustic-electric integration. The proposed research is of great theoretical interest, as it will provide greater insight into acoustic-electric integration for AEH. The proposed research is also of great clinical value, as the results may provide better guidance for optimizing CI signal processing for AEH.