多通道无线低功耗双向脑机接口关键技术研究

发布时间：2018-05-03 02:29

本文选题：侵入式双向脑机接口 + 电激励　；参考：《武汉大学》2016年博士论文

【摘要】：脑机接口系统依靠传感器提取的神经元活动信息为移动计算机终端和假肢等外部设备提供命令及控制信号。近年来,脑机接口作为非传统的交流渠道为大脑与外部设备建立直接连接。脑机接口的应用包括：(1)通过控制辅助设备帮助患者实现已丧失的交流能力和运动能力；(2)对某些患有特定疾病的患者进行实时身体状态监控；(3)对患者在康复治疗期间以及之后进行实时情感状态的监控；(4)与运动功能相关的大脑功能复健等。脑机接口根据系统位置的不同分为侵入式脑机接口系统和非侵入式脑机接口系统,由于侵入式系统独特的系统环境和人体对外来设备的排异性,我们对侵入式双向脑机接口系统的设计提出了严格的要求,并针对各个模块进行了详细的探讨与分析,提出了具体的设计指标,为后续设计并验证生物信号采集系统的性能提供了理论基础。在脑机接口的设计过程中,我们期望通过一个基础平台来整合不同的硬件系统和软件系统,这种选择的多样性和灵活性有效地降低了脑机接口的开发成本和研究门槛,增强了不同领域的合作研究机会。本文提出的一种模块化无线脑机接口软硬件系统能够采集并传输32通道脑电信号或8通道肌电信号,同时可选择其它生物信号作为辅助输入信号,譬如运动传感器数据和温度传感器数据等。该系统为研究者提供了一个低功耗的通信接口和组件化的软硬件基础框架,使得研究者可以根据自己的实际情况选择最适合的软硬件系统整合到基础框架中。该系统已通过不同的组件配置测试,并取得可与其它医疗级脑电信号、肌电信号采集系统相媲美的数据结果。针对传统的脑机接口系统只包含将采集的信号传输到主机设备端进行信号处理的单方向功能的问题,设计了一个具有无线充电功能的双向脑机接口系统,分析并检测集到的信号特征,进而施加电流刺激信号反向作用于脑部或脊髓神经,用于治疗某些中枢神经系统的疾病,促进神经可塑性。通过实验室台架实验和猴子体内实验进行局部场电位信号采集,分析并验证了系统性能,得出了对采集到的局部场电位信号进行时域分析和时频域分析,验证了电流激励信号对大脑感知运动皮层的影响。对于生物信号应用不同的信号处理技术是揭示生物神经生理背景的重要手段之一。本文设计的可穿戴肌电信号采集系统与市场上已有的高精度肌电信号采集系统相比,可获得更出色的信号质量及更稳定的系统性能。通过提取前臂不同肌肉群的肌电信号对7组不同的前臂和手部动作进行离线信号分类处理,获得了较高的分类准确率。近年来,深度学习算法在分析生物信号特征时也发挥了显著的作用。本文将多个模型的深度信念网络应用在从上臂肱二头肌提取的肌电信号上,分析并对肌肉的疲劳程度进行分类。非侵入式脑电信号提取系统被用于与注意力集中程度相关联的脑电α节律检测。除了使用传统的特征提取方法检测注意力集中程度,本文提出续同源性优化算法脑电信号中α节律代表的周期特征,作为辅助特征对有α节律出现的脑电数据和没有α节律出现的脑电数据进行分类。论文设计的脑机接口系统针对不同的系统设置进行了功耗分析,通过与近年来研发的类似系统进行比较,该系统具有较为先进的系统性能,证明了该系统在促进神经可塑性领域具有很大潜力。
[Abstract]:The brain machine interface system relies on the neuron activity information extracted by the sensor to provide command and control signals for the external devices such as mobile computer terminals and artificial limbs. In recent years, the brain machine interface has been used as a non-traditional communication channel to establish direct connection between the brain and external devices. The application of the brain machine interface includes: (1) help the auxiliary equipment to help. The patient realized lost communication and exercise ability; (2) real-time physical state monitoring for certain patients with specific diseases; (3) monitoring the emotional state of the patients during and after rehabilitation treatment; (4) the brain function related to motor function could be rehabilitate. For the intrusive brain machine interface system and the non-invasive brain machine interface system, due to the unique system environment of the intrusive system and the human body's rejection of the external equipment, we put forward strict requirements for the design of the intrusive bidirectional brain machine interface system, and have carried out a detailed discussion and analysis of each module, and put forward the specific design. It provides a theoretical basis for subsequent design and verification of the performance of the biological signal acquisition system. In the design of the brain machine interface, we expect to integrate different hardware and software systems through a basic platform. The diversity and flexibility of this selection can effectively reduce the cost and threshold of the development of the brain machine interface. A modular wireless brain machine interface software and hardware system can collect and transmit 32 channels of EEG or 8 channel EMG signals, while other biological signals can be selected as auxiliary input signals, such as motion sensor data and temperature sensor data. The researchers provide a low power communication interface and a component-based software and hardware framework, so that the researchers can choose the most suitable software and hardware system to be integrated into the basic framework according to their own actual conditions. The system has been tested by different components, and can obtain the signal acquisition system with other medical level brain signals and electromyography. The traditional brain machine interface system contains only the single directional function of signal processing by transmitting the collected signals to the host device end. A two-way BCI system with wireless charging function is designed to analyze and detect the signal features set, and then apply the current stimulus to the reverse signal. The brain or spinal nerve is used to treat some diseases of the central nervous system and promote the plasticity of the nervous system. The local field potential signal is collected through laboratory bench test and monkey experiment, and the performance of the system is analyzed and verified. The time domain analysis and time frequency analysis of local field potential signal are obtained. The effect of current excitation signals on the cerebral cortex of the brain is verified. The application of different signal processing techniques to biological signals is one of the important means to reveal the biological background of biological neurophysiology. The wearable muscular electrical signal acquisition system designed in this paper can be better than the high-precision signal acquisition system in the market. Signal quality and more stable system performance. By extracting the EMG signals from different forearm muscles, 7 groups of different forearm and hand movements are classified and classified, and a higher classification accuracy is obtained. In recent years, the depth learning algorithm has also played a significant role in the analysis of the characteristics of biological signals. The deep belief network is applied to the electromyographic signal extracted from the biceps brachii muscle of the upper arm to analyze and classify the degree of muscle fatigue. The non invasive EEG extraction system is used to detect the alpha rhythms associated with the concentration of attention. In addition to using the traditional feature extraction method to detect the concentration of attention, this paper This paper presents the periodic characteristics of the alpha rhythm representation in the EEG, which is used as an auxiliary feature to classify the EEG data with alpha rhythm and the EEG data without alpha rhythm. The paper designed the brain machine interface system to analyze the power consumption for different system settings, through similar lines developed in recent years. By comparison, the system has more advanced system performance, which proves that the system has great potential in promoting neural plasticity.

【学位授予单位】：武汉大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN911.7;R318

【参考文献】