NeuroGrid: a scalable system for large-scale recording of action potentials from the brain surface

我们建议开发一种新型的电极阵列(‘NeuroGrid’)，用于大规模记录脉冲和 用于记录自由行为啮齿动物神经活动的改进的小型化、多路复用设备， 同时最小化细胞/亚细胞和时间分辨率的损失。我们将开发出神经元大小 密度(10x10微米，间距30微米)、超适合性(4微米厚)和可扩展的聚对二甲苯 探头(与核磁共振兼容)，可放置在大脑表面以记录个体的尖峰活动 神经元及其聚集活动(局部场电位，LFP)。由于具有可伸缩性， 神经网格，我们可以记录数百个，甚至几千个来自表层的神经元 在实验动物和人类患者中长期存在的皮质层，即收集数量级 更大的样本和更长的时间，比目前的技术所可能的。将使用NeuroGrids 对于发现科学来说，理解作为认知和提取基础的神经元计算 癫痫模式中的生理标记物可能会提高我们理解癫痫的能力 发病机制和预测癫痫的发生。神经网格记录稳定信号的能力 延长的时间也可能支持假肢。我们将测试NeuroGrid的各种配置 到256点到&gt；1000个记录点。将测试多种信号多路复用方法，以实现 最紧凑的前台配置。在实验项目中，我们将同步执行 使用高密度硅探针从皮质深处进行记录和细胞特异性激活 1层和2/3层神经元用光遗传学方法识别表面记录的棘波的来源。我们 将测试和改进现有的各种单元聚类方法，并将探索新的压缩 从神经网格信号中提取棘波数据的传感方法。在该项目的临床部分，我们 在接受诊断性治疗的患者的手术过程中，将同时记录癫痫患者和完整的患者的位置 大脑监测，并检查尖峰信息是否提供了比LFP更可靠的标记。 NeuroGrids和专有技术将向合作者提供并商业化。

We propose to develop a novel electrode array (`NeuroGrid') for large-scale recording of spikes and improved miniaturized, multiplexed devices for recording of neural activity in freely behaving rodents, while minimizing the loss of cellular/sub-cellular and temporal resolution. We will develop neuron-size density (10x10 µm with 30 µm pitch), ultra-conformable (4-µm thick) and scalable parylene-based probes (MRI compatible) that can be placed on the brain's surface to record spiking activity of individual neurons and their aggregate activity (local field potentials, LFP). Owing to the scalable nature of the NeuroGrid, we can record from hundreds, potentially several thousands, of neurons from the superficial cortical layers chronically in experimental animals and human patients, i.e., collect orders of magnitude larger samples and for longer time than is possible with current technologies. NeuroGrids will be used for both discovery science to understand neuronal computation that underlies cognition and extracting physiological markers in epileptic patterns that may improve our ability to understand seizure pathogenesis and predict seizure occurrence. The ability of the NeuroGrid to record stable signals for extended time may also support prosthetics. We will test various configurations of the NeuroGrid from 64 to 256 to >1000 recording sites. Multiple approaches of signal multiplexing will be tested to achieve the most compact headstage configuration. In the experimental project, we will perform simultaneous recordings from the depth of the cortex using high-density silicon probes and cell specific activation of layer 1 and 2/3 neurons with optogenetic methods to identify the origin of surface-recorded spikes. We will test and improve various existing methods for unit clustering and will explore novel compressed sensing methods to extract spike data from the NeuroGrid signals. In the clinical part of the project, we will record from both epileptic and 'intact' locations during surgery in patients undergoing diagnostic brain monitoring and examine whether spiking information provides a more reliable marker than LFP. NeuroGrids and know-how will be made available to collaborators and commercialized.