脑干的声音精细结构处理机制及其应用

Hearing is one of the most important perception channels for human. Auditory brainstem implant (ABI) is an effective way to help patients to regain the auditory perception ability, and is regarded as a great challenge, a number of basic problems have not yet been satisfactorily resolved..Facing the currently challenging problems of ABI, such as the lack of deep research on auditory processing mechanism of brainstem, the lack of auditory brainstem computational model and the limited benefits from ABI, this project aims to explore the mechanism of the audio fine structure perception at brainstem and develop corresponding computational model.The pitch and timbre of sounds are selected as perception cues. The main aim is to improve the patients' perception for auditory fine structure after the ABI operation, and the research content includes the following issues: 1) Biological and medical experiments of pitch and timbre perception: cognitive task design, experimental paradigm design, brainstem auditory information processing mechanism exploration, etc. 2) Acoustic signal processing and neural signal analysis: the rapid and efficient timbre extraction methods, the pitch and timbre feature modification techniques, auditory nerve signal analysis techniques, analysis methods and computational model of the brainstem activity, novel stimulus encoding method for pitch and timbre, dynamic network analysis methods for brainstem and cortical activity, etc. 3) Functional imaging for the Central auditory neural system: fMRI signal analysis method for brainstem, standardized anatomical template of brainstem, etc. 4) Auditory mechanism and computational model of the brainstem auditory perception..In this project, the computational model of brainstem auditory processing will be deeply and meticulously studied and a novel stimulus encoding method of pitch and timbre will be proposed for ABI devices. The achievements will greatly deepen the research on central auditory system, that will have important academic value, as well as large economic and social benefits.

听觉是人类的重要感觉通路，近年兴起的听觉脑干植入术(ABI)是恢复听觉的最后努力和巨大挑战。针对当前脑干听觉机制研究不深、计算模型缺乏和ABI术后效果不佳的问题，拟以音高和音质为声音精细结构的感知线索，面向探索脑干听觉机制和提升ABI术后声音精细结构感知能力的目标，重点开展以下研究：1）音高音质感知的生物与医学实验：认知任务设计、实验范式设计、脑干处理机制等；2）声信号处理与神经信号解析：快速高效的音质提取技术、音高音质特征的修饰技术、听神经信号解析技术、脑干活动解析与建模技术、音高音质脑干听觉刺激编码方法和脑干与皮层活动的动态脑网络分析方法等；3）听觉中枢成像信号解析：脑干fMRI信号分析处理技术和标准脑干模型等；4）脑干听觉神经机制探索和脑干听觉计算模型建立等。本项目力图揭示脑干听觉规律，提出脑干听觉计算模型和新的ABI系统音高音质编码方法，其成果将具有重要学术价值、经济和社会意义。

听觉脑干植入术ABI是目前治疗重度蜗后耳聋患者最有效的技术，其原理为通过直接刺激脑干蜗神经核，使患者获得开放式言语识别能力。本课题针对ABI术后效果提升的问题，以音高和音质为线索，研究脑干听觉机制和声音精细结构感知能力，重点开展了：.1）ABI需要将装置植入到人的神经中枢，对电生理技术和植入技术均有较高的要求。本课题由哈尔滨工业大学和北京市神经外科研究所共同完成，哈工大团队完成了脑干的声音精细结构处理机制研究，神经外科研究所团队将其应用于蜗后耳聋患者，完成了5例患者的植入，提高了患者术后听力恢复效果。.2）认知实验研究：分别针对动物、听神经瘤患者、和正常人设计了音高、音质、节拍，以及听觉阈值认知实验，采集了80余人的脑活动数据，总结了一系列听觉脑认知规律与信息处理机制；.3）声音信号编码：研究了听觉网络改变机制，听觉信息特征提取方法与编码优化方法，提出了声信号谐波结构的提取方法，以及基于音素空间的声音编码方法；.4）脑活动成像解析：提出了基于稀疏约束的ERP提取方法提取分析方法，和基于动态脑网络的分析与建模方法，并通过深度神经网络对脑活动成像进行解析，揭示脑干和大脑皮层的音高音质感知规律；.5）最后开展了脑干听觉计算模型探索研究：提出了基于多粒度融合特征融合的计算模型，以及基于C-LSTM的多粒度分析模型。.本项目执行过程中共发表论文22篇，其中SCI期刊论文7篇，申请了发明专利5项，授权发明专利2项，授权软件著作权1项；研究人员获得国家级项目7项，培养博士后1人，毕业博士生6人、硕士生17人。主办国际研讨会2次，派出学生国际交流1人，教师国际交流1人。..本课题揭示了脑干听觉认知规律，提出了脑干听觉计算模型和新的ABI音高音质编码方法，为患者重新融入正常社会生活提供了条件，也为后续研究与提高提供理论基础与技术依据。