基于下丘定量化神经响应模型的耳蜗电刺激与声音刺激比较研究

Accurately and objectively testifying the effectiveness of the novel speech encoding strategies is critically important for the development of new strategies and methods in cochlear implant. It also provides a kind of objective and auxiliary tool for mainstream assessment on cochlear implant. The project has two main purposes. The first one is to determine whether the time and spatial model about neuronal response of inferior colliculus (IC) guinea pig for tonic speech stimuli can be quantified. The second one is to judge whether the proposed model can act as a kind of assessment basis and can be compared with the IC neural response evoked by multichannel electrical stimuli to evaluate the value of speech encoding strategy adopted in hearing restoration improvement. In order to accomplish such goals, we conduct a comparative study between the tonal speech stimuli and encoded electrical stimuli during which we quantify the time and spatial model of IC neural response respectively by acoustic parameters and electrical parameters using guinea pigs in narcotism. Using a simulation and animal electrophysiological experiment, we are able to build the quantitative regulation relationship between acoustic stimuli and speech stimuli characteristic according to quantitative model of neural response. The project would help us to understand further the individual weight of every electrical parameters in IC neural response, providing profound scientific evidence for a quicker and more objective assessment of novel speech encoding strategy. The applicant has a multidisciplinary background (neural eletrophysiology, Biomedical electronics, and biomedical engineering) and has published. Furthermore, the substantial supports for the experimental research from the Support Unit and strong supports for the clinical hearing assessment studies from the Cooperation Unit could ensure the accomplishment of this project.

准确、客观地检验评估新型语音编码策略的效果对促进电子耳蜗新策略新方法的发展及其重要，同时也为主流的电子耳蜗评估方法提供一种客观的辅助评价工具。豚鼠下丘对声调语音刺激的神经响应时空模式能否定量化？能否利用该神经响应定量化模型作为一种评估依据，评估电刺激所采用的语音编码策略在提高听力修复的价值？本项目选择麻醉状态下的豚鼠作为研究对象，利用软件仿真和动物电生理实验，以声调语音刺激诱发的下丘神经响应的定量化时空模式为依据，与电刺激诱发的定量化响应比较，建立声学刺激与电刺激的定量调控关系，为客观评估新型编码策略的有效性提供科学依据。申请人具有神经电生理学、生物医学电子学、生物医学工程多学科背景，已以发表SCI/EI论文10篇，第一作者获权发明专利8项。依托单位重庆大学生物流变科学与技术教育部重点实验室可为项目提供坚实的实验保证。合作单位三医大第一医院可提供听觉临床研究支持。

音调在听觉系统的表征与编码是听觉研究的重要内容。从听觉电生理实验角度，本项目围绕音调在下丘的神经表征，（1）开展了在信号处理方面：选取了自相关法和能熵比的主体基频提取方法，来提取刺激语音信号的基频轨迹。提取神经元响应信号的特征ISI并计算其与基频轨迹的相关计算，从时域的角度定量分析了单神经元电活动对刺激语音的基频轨迹的表征程度。提取神经元响应信号的特征PSTH及其时频谱图并计算和基频轨迹的相关计算，从时频结合的角度分析了神经元响应信号表征刺激语音基频轨迹的程度。结果表明，PSTH的时频谱图与刺激语音的语谱图非常相像，皮尔逊相关系数表明两者具有非常高的相关性，即下丘的部分神经元响应的PSTH可以从时频角度几乎完全表征出刺激语音的基频轨迹。将刺激语音基频轨迹和对应的神经元响应信号进行分段计算矢量之和的平均值，得到矢量强度。结合Rayleigh值表明，神经元响应能够动态并显著表征语音基频轨迹。（2）在下丘神经响应分析方面，通过原始汉语音节与编码后汉语音节诱发下丘在体神经电响应研究，构建下丘STRF模型，预测原始汉语音节和编码后汉语音节诱发下丘的神经电响应，最后将该预测模型应用于不同编码策略下的语音信号，并对其响应结果进行分析，实验结果表明下丘神经元能够通过其响应显示出HSSE编码后语音的预测响应结果（Avg=0.68±0.01(cc)）均要优于CIS编码后的语音（Avg=0.63±0.01(cc)）。验证了编码策略中的谐波结构能增强下丘对声调特征神经表征的假设。（3）从听觉康复的测评角度，本项目开展了重庆版林氏六音发音规律及对重庆地区CI儿童用户的指导，重庆普通话版林氏六音测试频率范围较窄，高频音频率较低，450～600hz频率范围缺失。本研究提示重庆地区的CI听力评估时，应注意高频声音的感知，术后调机及言语康复训练也有一定的参考价值。完成了汉语声调版林氏六音的设计与人工耳蜗儿童评估的研究，24个原始测试音的声调识别实验中元音识别率为83.1%，高于辅音识别率79.0%。结果提示汉语声调版林氏六音可以用于测评CI用户的听力频率范围与声调识别能力。