Deep Learning-Based Speech Enhancement With a Loss Trading Off the Speech Distortion and the Noise Residue for Cochlear Implants

The cochlea plays a key role in the transmission from acoustic vibration to neural stimulation upon which the brain perceives the sound. A cochlear implant (CI) is an auditory prosthesis to replace the damaged cochlear hair cells to achieve acoustic-to-neural conversion. However, the CI is a very coarse bionic imitation of the normal cochlea. The highly resolved time-frequency-intensity information transmitted by the normal cochlea, which is vital to high-quality auditory perception such as speech perception in challenging environments, cannot be guaranteed by CIs. Although CI recipients with state-of-the-art commercial CI devices achieve good speech perception in quiet backgrounds, they usually suffer from poor speech perception in noisy environments. Therefore, noise suppression or speech enhancement (SE) is one of the most important technologies for CI. In this study, we introduce recent progress in deep learning (DL), mostly neural networks (NN)-based SE front ends to CI, and discuss how the hearing properties of the CI recipients could be utilized to optimize the DL-based SE. In particular, different loss functions are introduced to supervise the NN training, and a set of objective and subjective experiments is presented. Results verify that the CI recipients are more sensitive to the residual noise than the SE-induced speech distortion, which has been common knowledge in CI research. Furthermore, speech reception threshold (SRT) in noise tests demonstrates that the intelligibility of the denoised speech can be significantly improved when the NN is trained with a loss function bias to more noise suppression than that with equal attention on noise residue and speech distortion.

Feasibility evaluation of transtympanic laser stimulation of the cochlea from the outer ear

Infrared laser stimulation has been studied as an alternative approach to auditory prostheses. This study evaluated the feasibility of infrared laser stimulation of the cochlea from the outer ear bypassing the middle ear function. An optic fiber was inserted into the ear canal and a laser was used to irradiate the cochlea through the tympanic membrane in Mongolian gerbil. A pulsed infrared laser (10.1 mJ/cm2) and clicking sound (70 peak-to-peak equivalent sound pressure level) were presented to the animals. The amplitude of the laser-evoked cochlear response was systematically decreased following insertion of a filter between the tympanic membrane and cochlea; however, the auditory-evoked cochlear response did not decrease. The filter was removed and the laser-evoked response returned to around the original level. The amplitude ratio and the relative change in response amplitude before and during filter insertion significantly decreased as the absorbance of the infrared filter increased. These results indicate that laser irradiation could bypass the function of the middle ear and directly activate the cochlea. Therefore, an auditory prosthesis based on laser stimulation represents a possible noninvasive alternative to conventional auditory prostheses requiring surgical implants.

Psychophysical measures of tonotopic selectivity in cats and humans

Sound spectra are represented by patterns of activity along the tonotopic axis ofthe cochlea. Cochlear implants can transmit spectra by stimulating tonotopicallyappropriate electrodes, but fidelity is limited by intracochlear spread of excitation. We aim to better evaluate present-day experimental stimulation procedures and, potentially, to improve transmission of spectra with novel stimulation modalities. As a first step, we are developing non-invasive measures of tonotopic spread of excitation that can be compared between normal-hearing cats and humans. These measures include psychophysics in the present study and scalp-recorded electrophysiology in a companion study (Guérit et al., 2021). Cats and humans detected pure-tone probes presented in continuous 1/8- and 1-oct noise-band maskers. Masker bandwidths were readily discernable in both species by the dependence of masked thresholds on probe frequencies. Thresholds were largely constant across the bandwidth of the 1-oct masker, whereas thresholds dropped markedly at frequencies away from the center of the 1/8-oct masker. Cats and humans differed in that the feline auditory filter centered on 8 kHz, which we measured using a notched-noise procedure, was 22% wider than published values for humans at the same center frequency. Also, thresholds for the cats in the 1-octmasker condition consistently were 1.0 to 3.2 dB higher than expected based on the estimated masker power in the feline auditory filter. The present psychophysical results parallel those in our companion electrophysiological study, thereby providing perceptual validation for that study. These psychophysical and electrophysiological methods will be valuable for future investigations of novel approaches for auditory prosthesis.

Auditory Brainstem Implantation: An Overview

An auditory brainstem implant (ABI) is a surgically implanted central neural auditory prosthesis for the treatment of profound sensorineural hearing loss in children and adults who are not cochlear implant candidates due to a lack of anatomically intact cochlear nerves or implantable cochleae. The device consists of a multielectrode surface array which is placed within the lateral recess of the fourth ventricle along the brainstem and directly stimulates the cochlear nucleus, thereby bypassing the peripheral auditory system. In the United States, candidacy criteria for ABI include deaf patients with neurofibromatosis type 2 (NF2) who are 12 years or older undergoing first- or second-side vestibular schwannoma resection. In recent years, several non-NF2 indications for ABI have been explored, including bilateral cochlear nerve avulsion from trauma, complete ossification of the cochlea due to meningitis, or a severe cochlear malformation not amenable to cochlear implantation. In addition, growing experience with ABI in infants and children has been documented with encouraging outcomes. While cochlear implantation generally remains the first-line option for hearing rehabilitation in NF2 patients with stable tumors or post hearing preservation surgery where hearing is lost but a cochlear nerve remains accessible for stimulation, an ABI is the next alternative in cases where the cochlear nerve is absent and/or if the cochlea cannot be implanted. Herein, we review ABI device design, clinical evaluation, indications, operative technique, and outcomes as it relates to lateral skull base pathology.

Electrophysiological Evidence of the Basilar-Membrane Travelling Wave and Frequency Place Coding of Sound in Cochlear Implant Recipients

Aim: To obtain direct evidence for the cochlear travelling wave in humans by performing electrocochleography from within the cochlea in subjects implanted with an auditory prosthesis. Background: Sound induces a travelling wave that propagates along the basilar membrane, exhibiting cochleotopic tuning with a frequency-dependent phase delay. To date, evoked potentials and psychophysical experiments have supported the presence of the travelling wave in humans, but direct measurements have not been made. Methods: Electrical potentials in response to rarefaction and condensation acoustic tone bursts were recorded from multiple sites along the human cochlea, directly from a cochlear implant electrode during, and immediately after, its insertion. These recordings were made from individuals with residual hearing. Results: Electrocochleography was recorded from 11 intracochlear electrodes in 7 ears from 6 subjects, with detectable responses on all electrodes in 5 ears. Cochleotopic tuning and frequency-dependent phase delay of the cochlear microphonic were demonstrated. The response latencies were slightly shorter than those anticipated which we attribute to the subjects' hearing loss. Conclusions: Direct evidence for the travelling wave was observed. Electrocochleography from cochlear implant electrodes provides site-specific information on hair cell and neural function of the cochlea with potential diagnostic value.

Tubular manipulators: a new concept for intracochlear positioning of an auditory prosthesis

The aim of this study was to investigate the applicability of tubular manipulators as an actuator mechanism for intracochlear positioning of the electrode array (EA) of a cochlear implant (CI). This is motivated by the vision of an atraumatic insertion of the EA into the inner ear (cochlea) without any damage to the intracochlear structures in combination with a well-defined final position. To realize this, an actuator mechanism is required which allows consideration of the patient-specific anatomy. We propose a tubular manipulator for this task. It consists of three concentric tubes: A straight outer tube serves as a guiding sleeve to enter the inner ear (cochlea) and two additional telescoping, superelastic, helically precurved tubes. By selecting helical tube parameters of both tubes prior insertion, a patient-specific curling behaviour of the tubular manipulator can be achieved. For preliminary investigation, segmentation and skeletonization of 5 human scala tympani were performed to determine their centrelines. These centrelines were considered as individual ideal insertion paths. An optimization algorithm was developed to identify suitable tube set parameters (curvature, diameter, length, torsion, stiffness) as well as configuration parameters (translation and rotation of the 2 inner tubes). Different error values describing the deviation of the shape of the tubes with respect to the insertion path were used to quantify the optimization results. In all cases tube set parameters for a final position within the cochlea were found, while keeping the maximum error below 1mm. These preliminary results are promising in terms of the potential applicability of tubular manipulators for positioning auditory prosthesis inside the scala tympani of the inner ear.