瞬態誘發耳聲傳射為當耳蝸受到短暫聲音刺激時所產生的寬頻響應，而瞬態誘發耳聲傳射當中特定頻率分量之時間延遲被稱為群延遲，其數值大小能夠反映耳蝸之外毛細胞狀態的相關訊息。本篇論文嘗試提出了一套估計方法，以在不同大小的背景噪聲下可靠地估計瞬態誘發耳聲傳射訊號及其群延遲的大小，並同時檢驗方法的實用性與限制性。由於瞬態誘發耳聲傳射訊號的估計需要通過數千次重複量測的平均值獲得，訊號在整個測量過程中難免會受到噪音的污染，這些噪音可能為來自環境周圍的說話聲音或者是來自人體的肌肉動作，因此在對所有重複量測的訊號取平均值前，必須應用偽影抑制的方法進行降噪。相較於較為保守的偽影抑制是基於訊號的最大值進行整個幀的排除稱之為frame-exclusion (FE)，我們提出了逐點的偽影去除方法稱之為noise-shaping (NS)，該方法以逐點的方式排除數值較大的噪音樣本，使得整體噪音數值的分佈盡可能的近似於高斯分佈，並在保有更多量測訊息的前提下產生了更低的噪音功率。接著，為了從降噪後的瞬態誘發耳聲傳射之相位譜估算群延遲的大小，我們進一步的採用分段計算斜率的方式，以解決因為在某些頻段瞬態誘發耳聲傳射訊號強度較弱而引起的相位頻譜波動現象，此方法會選擇訊號強度高的頻段來獲取相位訊息，而不會考慮其他訊號強度低的頻段之資訊。最後，我們將8種不同的噪聲逐次地調整聲量大小並添加到來自11個相異耳的乾淨數據中，以產生88個實例來模擬上述方法之實際應用效果，結果表明了以NS法處理之實例所產生的TEOAE訊號，其方均根誤差平均值相比於使用FE最多可減少60.3%，且由NS產生的群延遲之絕對誤差平均值則相較於FE最多降低了34.4%；此外，同時使用NS及分段計算斜率法可以將群延遲的絕對誤差平均值最多降低55.4%。

Transient-evoked otoacoustic emissions (TEOAEs) is the wideband response of cochlea when it is stimulated by an acoustic pulse. The time latency of the certain frequency component known as the group delay (GD) from TEOAE is promising to provide useful information about the condition of outer hair cells (OHCs). The goal of this thesis is to develop a reliable estimation method for the TEOAE signal and its time latency values in the presence of different background noise levels. Since the estimated TEOAE signal may be contaminated by environment sound, e.g., speech from the background or muscle movements from the human body, an artifact rejection (AR) method must be applied to denoise before the average over all repetitions’ frames. The so-called frame-exclusion (FE) traditional method often delete an entire frame if its maximum signal value exceeds a pre-defined threshold. Here, we propose a point-wise exclusion AR called noise-shaping (NS) which rejects large noise elements to shape the noise value distribution as similar to Gaussian distribution as possible. Thus, it keeps more frames in average while producing lower noise power than the FE. For estimating GD from denoised TEOAE phase spectrum, a segmented-slope method (SSM) is employed to deal with fluctuations of phase spectrum caused by weak TEOAE signal magnitude. The SSM selects the frequency bands with high TEOAE strength while ignoring low-energy bands. Finally, we add 8 kinds of noise with different levels to relatively clean TEOAE data from 11 healthy ears to simulate noisy environments in real life and assess the efficacy of different TEOAE and group delay estimation methods. The results reveal the average of root mean square error from the TEOAE signal generated by NS can be lower by as much as 60.3%, and the mean absolute deviation of the GD can be reduced by up to 34.4%, respectively, compared to FE. Furthermore, the usage of SSM can reduce the mean absolute deviation of GD by up to 55.4%.

[1] Kemp, D.T., Stimulated acoustic emissions from within the human auditory system. The Journal of the Acoustical Society of America, 1978. 64(5): p. 1386-1391.
[2] Keefe, D.H., et al., Comparisons of transient evoked otoacoustic emissions using chirp and click stimuli. The Journal of the Acoustical Society of America, 2016. 140(3): p. 1949-1973.
[3] Norton, S.J., Application of transient evoked otoacoustic emissions to pediatric populations. Ear and hearing, 1993. 14(1): p. 64-73.
[4] Withnell, R.H., G.K. Yates, and D.L. Kirk, Changes to low-frequency components of the TEOAE following acoustic trauma to the base of the cochlea. Hearing research, 2000. 139(1-2): p. 1-12.
[5] Avan, P., Bonfils, P., Loth, D., & Wit, H. P. (1993). Temporal patterns of transient-evoked otoacoustic emissions in normal and impaired cochleae. Hearing research, 70(1), 109-120.
[6] Sisto, R. and A. Moleti, On the frequency dependence of the otoacoustic emission latency in hypoacoustic and normal ears. The Journal of the Acoustical Society of America, 2002. 111(1): p. 297-308.
[7] Kalluri, R. and C.A. Shera, Near equivalence of human click-evoked and stimulus-frequency otoacoustic emissions. The Journal of the Acoustical Society of America, 2007. 121(4): p. 2097-2110.
[8] Brownell, W.E., Outer hair cell electromotility and otoacoustic emissions. Ear and hearing, 1990. 11(2): p. 82.
[9] van Huffelen, W.M., N.J. Mateijsen, and H.P. Wit, Classification of patients with Meniere’s disease using otoacoustic emissions. Audiology and Neurotology, 1998. 3(6): p. 419-430.
[10] de Kleine, E., et al., Evoked otoacoustic emissions in patients with Ménière's disease. Otology & neurotology, 2002. 23(4): p. 510-516.
[11] Bonfils, P., A. Uziel, and P. Narcy, The properties of spontaneous and evoked acoustic emissions in neonates and children: a preliminary report. Archives of Oto-rhino-laryngology, 1989. 246(5): p. 249-251.
[12] Jacobson, J.T. and C.A. Jacobson, The effects of noise in transient EOAE newborn hearing screening. International journal of pediatric otorhinolaryngology, 1994. 29(3): p. 235-248.
[13] Komeili, M., et al., Feature selection for nonstationary data: Application to human recognition using medical biometrics. IEEE transactions on cybernetics, 2017. 48(5): p. 1446-1459.
[14] Liu, Y. and D. Hatzinakos. Biometric identification based on transient evoked otoacoustic emission. in IEEE International Symposium on Signal Processing and Information Technology. 2013. IEEE.
[15] Goodman, S.S., et al., High-frequency click-evoked otoacoustic emissions and behavioral thresholds in humans. The Journal of the Acoustical Society of America, 2009. 125(2): p. 1014-1032.
[16] Withnell, R.H., C. Hazlewood, and A. Knowlton, Reconciling the origin of the transient evoked ototacoustic emission in humans. The Journal of the Acoustical Society of America, 2008. 123(1): p. 212-221.
[17] Whitehead, M., et al., Measurement of otoacoustic emissions for hearing assessment. IEEE Engineering in Medicine and Biology magazine, 1994. 13(2): p. 210-226.
[18] Shera, C.A. and J.J. Guinan Jr, Evoked otoacoustic emissions arise by two fundamentally different mechanisms: a taxonomy for mammalian OAEs. The Journal of the Acoustical Society of America, 1999. 105(2): p. 782-798.
[19] Zweig, G. and C.A. Shera, The origin of periodicity in the spectrum of evoked otoacoustic emissions. The Journal of the Acoustical Society of America, 1995. 98(4): p. 2018-2047.
[20] Shera, C.A. and C. Bergevin, Obtaining reliable phase-gradient delays from otoacoustic emission data. The Journal of the Acoustical Society of America, 2012. 132(2): p. 927-943.
[21] Keefe, D.H., Moments of click-evoked otoacoustic emissions in human ears: Group delay and spread, instantaneous frequency and bandwidth. The Journal of the Acoustical Society of America, 2012. 132(5): p. 3319-3350.
[22] Siegel, J.H., et al., Delays of stimulus-frequency otoacoustic emissions and cochlear vibrations contradict the theory of coherent reflection filtering. The Journal of the Acoustical Society of America, 2005. 118(4): p. 2434-2443.
[23] Shera, C.A. and J.J. Guinan Jr, Stimulus-frequency-emission group delay: A test of coherent reflection filtering and a window on cochlear tuning. The Journal of the Acoustical Society of America, 2003. 113(5): p. 2762-2772.
[24] Bergevin, C., Comparison of otoacoustic emissions within gecko subfamilies: morphological implications for auditory function in lizards. Journal of the Association for Research in Otolaryngology, 2011. 12(2): p. 203-217.
[25] Shlens, J., A tutorial on principal component analysis. arXiv preprint arXiv:1404.1100, 2014.
[26] Liu, Y.-W., et al., Transient-evoked otoacoustic emission signals predicting outcomes of acute sensorineural hearing loss in patients with Ménière’s disease. Acta oto-laryngologica, 2020. 140(3): p. 230-235.
[27] Sajjadi, H. and M.M. Paparella, Meniere's disease. The Lancet, 2008. 372(9636): p. 406-414.
[28] De Luca, P., et al., Dietary restriction for the treatment of Meniere’s disease. Translational Medicine@ UniSa, 2020. 22: p. 5.
[29] Jedrzejczak, W.W., et al., Time–frequency analysis of linear and nonlinear otoacoustic emissions and removal of a short-latency stimulus artifact. The Journal of the Acoustical Society of America, 2012. 131(3): p. 2200-2208.
[30] Kemp, D.T. and S. Ryan, Otoacoustic emission tests in neonatal screening programmes. Acta Oto-Laryngologica, 1991. 111(sup482): p. 73-84.
[31] Kim, T.K., T test as a parametric statistic. Korean journal of anesthesiology, 2015. 68(6): p. 540.
[32] Altman, D.G. and J.M. Bland, Standard deviations and standard errors. Bmj, 2005. 331(7521): p. 903.
[33] Box, G.E., Non-normality and tests on variances. Biometrika, 1953. 40(3/4): p. 318-335.
[34] Cortes, C. and V. Vapnik, Support-vector networks. Machine learning, 1995. 20(3): p. 273-297.
[35] Wong, T.-T., Performance evaluation of classification algorithms by k-fold and leave-one-out cross validation. Pattern Recognition, 2015. 48(9): p. 2839-2846.