基于多微电极阵列的神经元网络学习模型建立及机制研究

项目名称： 基于多微电极阵列的神经元网络学习模型建立及机制研究

项目编号： No.30800314

项目类型： 青年科学基金项目

立项/批准年度： 2009

项目学科： 化学工业

项目作者： 李向宁

作者单位： 华中科技大学

项目金额： 20万元

中文摘要： 阐明学习和记忆等高级认知过程中神经系统动态获取和储存信息的机制是认知神经科学的主要任务。本课题基于在多电极阵列系统上长时间培养的海马神经元网络，获取了网络自发放电和刺激-响应序列。引入了非线性分析等方法、改进了去除刺激伪迹程序，构建了神经信息海量数据获取及分析平台。通过对神经元网络发育中的复杂放电模式进行连续分析，发现体外培养网络在发育中期存在低频同步振荡。药物调控和多种参数分析结果显示同步振荡是网络稳态可塑性的表征，其中存在混沌等非线性模式，并具有突现等特征。利用低频电刺激对该模式进行了调控，在此基础上通过对多位点刺激-响应信息分析，改进了训练位点选择、刺激时间间隔等参数，建立了基于三位点训练的网络学习模型。通过培养转基因星形胶质细胞-神经元混合网络、构建基于混合神经元网络模型，并结合药物调控，研究了中间神经元、胶质细胞比例对神经元网络同步振荡模式和网络可塑性的影响，结果显示中间神经元参与网络同步振荡的维持和响应信息的精确编码，胶质细胞对网络同步性的产生和可塑性的维持具有重要作用。

中文关键词： 神经元网络;可塑性;同步振荡;多电极阵列

英文摘要： Discovering the mechanism underlying neural information acquisition and storage in learning and memory is the major task in cognitive neuroscience. In this project, with the hippocampal neuronal network cultured on multi-electrode array system, we tracked the network spontaneous firing and stimulus - response sequences from 60 recording sites. To analysis the complicated firing pattern of neural networks, we established a comprehensive neural data acquisition and analysis platform. During development in vitro, neuronal networks showed low-frequency synchronized oscillations. The spontaneous activity was inhibited after tetrodotoxin was added into the medium. Once tetrodotoxin was washed out after a 4 h treatment, spontaneous activities rose significantly with spike rate increasing approximately three times, and synchronized burst oscillations appeared throughout the network, suggesting that the spontaneous synchronized oscillations should be an indicator of homeostatic plasticity in cultured neuronal network. Using multiple nonlinear techniques, we revealed emergent transition between chaotic behavior and superburst occurred periodically in the spontaneous activity of neuronal networks in vitro. Additionally, electrical stimulations could regulate these networkwide oscillatory activities by triggering the next synchronized superbursts prematurely. On this basis, with the stimulus - response information analysis, we improved training parameters and modified the learning model on network in vitro based on three-point training. With co-culture networks including the transgenic astrocyte and neuron, we built a neural network model based on the Izhikevich model under different neuron/astrocyte ratio. Then we studied the effect of glia cells and interneurons on the synchronized oscillation and neuronal network plasticity. The results showed that interneurons effect on the maintenance of synchronization oscillations and accurate coding in neural response, and that the glial cells play an important role in the generation of synchronized firing and the maintenance of plasticity.

英文关键词： Neuronal network; Plasticity; Sychoronized oscillation; Multi-electrodes array