以心理音響模型與虛擬通道應用在植入式人工電子耳刺激策略

Abstract

植入式人工電子耳可以幫助嚴重的聽障患者重新再聽到聲音。過去應用於神經刺激器的電極只能刺激部分的聽神經，不僅頻帶解析度低，頻道數目直接由電極個數決定，有些區域的聽神經甚至無法被刺激。傳統的人工電子耳受限於電極數量的問題，因此患者在吵雜環境下語音的辨識以及音樂的旋律變化所聽到的結果並不理想。 近年來有所謂虛擬通道的技術，藉由同時激多個相鄰電極調整刺激範圍，可以增加頻帶解析度。這篇論文提出了一個新型語音機制應用在植入式人工電子耳上，以頻譜上的語音外框為基礎應用虛擬通道的技術發展出多電極混合刺激策略並且根據音響心理學模型來選擇更有效的位置來進行刺激。此混合策略可選擇激化不同個數的電極來呈現各個頻帶，相對於其他現有策略，電極在使用上較具彈性。在刺激時序方面則允許多個頻帶在單一時間週期呈現，同時在數個時間週期內完成所有頻帶刺激動作。 從結果可以發現，這種新型的混合刺激策略與傳統的語音機制 CIS 以及 HiRes120 比較發現在噪音吵雜的環境下的語音及音樂上，都有很大的改善，然而搭配音響心理學模型也証實了除了可以有效的減少電極的選擇外，患者聽到的聲音也不會因為減少了電極的選擇而失真。

Cochlear implants provide an opportunity for profoundly hearing impaired patients to hear sound again. However, the limited number of electrodes is insufficient to provide enough hearing resolution, and speech comprehension for cochlear implant users under noisy environment is far worse than normal hearing people. A vocoder implementation of a novel stimulating strategy for cochlear implants is presented in this study. A multiple-electrodes current steering scheme and psycho-acoustic model were incorporated in this new stimulating strategy for more effective stimulation of the nerve fibers. The primary purpose is to study the hearing performance of a novel hybrid stimulating strategy based on two-electrode or four-electrode current steering scheme. The secondary purpose of this study was to determine if the use of a psychoacoustic model would reduce the number of current steering electrode pair needs to be turned on during a FFT process cycle. Result of this study shows the new stimulating strategy outperforms the HiRes120 stimulating strategy. It also shows that the new stimulating strategy with psycho-acoustic model maintains the spectral resolution while requiring a significantly fewer numbers of current steering electrode pair turn on.