뉴럴 인터페이스 기반 인공와우모델 개발

Abstract

An electro-neural interface model was developed to understand the clinical outcome and prediction of cochlear implant (CI) performance. The neural discharge patterns of electrically stimulated auditory nerve fibers (ANFs) were simulated using the model and graphically represented by a neurogram. The model was validated through comparison with clinical outcomes to obtain the effect of background noise and various stimulation rates on vowel identification. The vowel identification performance evaluated from both the CI subjects and the model decreased with increasing background noise; in contrast, it had no significant effect on the stimulation rate. However, the clinical outcomes (e.g., spectral ripple discrimination (SRD) performance) have large cross-subject variability to the extent of having the same CIs and sound processors. I investigated neural excitations and predicted the SRD performance depending on the distance between the stimulation electrode and auditory nerve fibers (ANFs), which affect the current spread. The SRD performance was predicted by calculating the similarity between the two neurograms in response to standard and inverted stimuli, respectively. The predicted SRD performance decreased with increasing electrode-ANF distance, which is consistent with previous studies. Additionally, the differential channel interaction (i.e., inhibition or facilitation) induced by various electrode-ANF distances influenced the SRD performance. The reducing current spread could be a solution to improving the low CI performance. The neural excitation, in response to the charge-balanced asymmetric pulse, was exhibited to reduce the current spread using the model. The asymmetric pulse provided a greater threshold, comfortable level, and spread of excitation in comparison with the symmetric pulse. The results of this thesis reveal that the electrode-neural interface model developed can predict CI performance and may serve as a useful tool in understanding the neural elements and underlying factors of CI outcomes that cannot be evaluated by behavioral studies alone.