One-Year Follow Up of Auditory Performance in Post-Lingually Deafened Adults Implanted with the Neurelec Digisonic(®) SP/Saphyr(®) Neo Cochlear Implant System

Longitudinal Speech Recognition Changes After Cochlear Implant: Systematic Review and Meta-analysis

OBJECTIVES

To examine patterns of change and plateau in speech recognition scores in postlingually hearing impaired adult cochlear implant recipients. The study also examines variations in change patterns for different speech materials and testing conditions.

STUDY DESIGN

Used systematic review with meta-analysis.

METHODS

Articles in English reporting speech recognition scores of adults with postlingual hearing loss at pre-implantation and at least two post-implantation time points were included. Statistically significant changes were determined by meta-analysis and the 95% confidence interval.

RESULTS

A total of 22 articles representing 1954 patients were included. Meta-analysis of mean difference demonstrated significant improvements in speech recognition score for words in quiet (37.4%; 95% confidence interval [34.7%, 40.7%]), sentences in quiet (49.4%; 95% confidence interval [44.9%, 53.9%]), and sentences in noise (30.8%; 95% confidence interval [25.2%, 36.4%]) from pre-op to 3 months. Scores continued to increase from 3 to 12 months but did not reach significance. Similarly, significant improvements from pre-op to 3 months were observed for consonant nucleus consonant (CNC) words in quiet (37.1%; 95% confidence interval [33.8%, 40.4%]), hearing in noise test (HINT) sentences in quiet (46.5%; 95% confidence interval [37.0%, 56.0%]), AzBio sentences in quiet (45.9%; 95% confidence interval [44.2%, 47.5%]), and AzBio sentences in noise (26.4%; 95% confidence interval [18.6%, 34.2%]). HINT sentences in noise demonstrated improvement from pre-op to 3 months (35.1%; 95% confidence interval [30.0%, 40.3%]) and from 3 to 12 months (15.5%; 95% confidence interval [7.2%, 23.8%]).

CONCLUSIONS

Mean speech recognition scores demonstrate significant improvement within the first 3 months, with no further statistically significant improvement after 3 months. However, large individual variation should be expected and future research is needed to explain the sources of these individual differences. Laryngoscope, 2022.

A Deep Learning Based Approach to Synthesize Intelligible Speech with Limited Temporal Envelope Information

Envelope waveforms can be extracted from multiple frequency bands of a speech signal, and envelope waveforms carry important intelligibility information for human speech communication. This study aimed to investigate whether a deep learning-based model with features of temporal envelope information could synthesize an intelligible speech, and to study the effect of reducing the number (from 8 to 2 in this work) of temporal envelope information on the intelligibility of the synthesized speech. The objective evaluation metric of short-time objective intelligibility (STOI) showed that, on average, the synthesized speech of the proposed approach provided higher STOI (i.e., 0.8) scores in each test condition; and the human listening test showed that the average word correct rate of eight listeners was higher than 97.5%. These findings indicated that the proposed deep learning-based system can be a potential approach to synthesize a highly intelligible speech with limited envelope information in the future.

A New Method to Test the Efficiency of Cochlear Implant Artifacts Removal From Auditory Evoked Potentials

Auditory evoked potentials are of great interest to objectively evaluate the audition in cochlear implant (CI) recipients. However, these measures are impeded by CI stimulation electrical artifacts present in the EEG. In the first part, this paper investigates the use of a hybrid model approximating CI patient data. This model gives access to both uncontaminated and denoised data, thus allowing for the evaluation of CI artifact removal methods. Here the efficiency of independent component analysis (ICA) is evaluated in the context of auditory steady-state responses (ASSRs). A dedicated experimental setup was developed to simultaneously record EEG data from a normal hearing (NH) participant and CI artifact data from a phantom equipped with a CI. Hybrid data were obtained as a linear mixture of both sources. Amplitude-modulated continuous tones were used as stimuli to elicit ASSRs. After denoising, the comparison of denoised hybrid data and original NH data showed high correlations between the two datasets, demonstrating the efficiency of ICA. In the second part, the ICA was applied to real clinical CI ASSR data. Results support the usefulness of the methodology as regards the performance evaluation of signal processing methods applied to CI patient data prior to clinical application.