Multi-Domain Identification of Functional Network Dynamics at the Neuronal Scale

从多个大脑区域的大量神经元中获取神经数据的最新技术进步，为以前所未有的时空分辨率深入了解神经元之间、神经元群内和大脑区域之间复杂的动态相互作用打开了一扇独特的机会之窗。为了利用这些数据，需要能够同时捕获底层功能网络的动态性、任务依赖性和统计特征的计算效率分析技术。本提案的研究目标是开发一种系统识别方法，以提取神经元尺度上的功能网络动态，并将其应用于大规模记录，以探测和揭示行为背后的神经元机制。研究方法是以鲁棒性和可扩展性的方式识别神经元集合中的时间因果影响，捕获神经元数据协方差结构的突变，并以高分辨率提取网络节点的交叉频率耦合。该项目解决了现有方法面临的几个突出挑战，包括由于使用滑动窗口而导致的光谱时间分辨率的损失，缺乏特定领域的模型来捕获网络动态，以及导致高度偏倚的网络特性的临时估计程序。通过使用来自雪貂和小鼠听觉系统的电生理学和双光子钙成像数据，所提出的建模和估计框架将用于研究与听觉处理相关的网络功能组织有关的几个基本问题。该项目有望通过提供用于神经控制系统的信号处理解决方案来影响技术。该研究还与教育和推广活动相结合，包括高中水平的研讨会，本科生参与和课程开发。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Recent technological advances in neural data acquisition from a large number of neurons across multiple brain regions have opened up a unique window of opportunity to gain insight into the complex dynamic interactions among neurons, within neuronal populations, and across brain regions at unprecedented spatiotemporal resolutions. In order to exploit these data, computationally efficient analysis techniques capable of simultaneously capturing the dynamicity, task-dependence, and statistical characteristics of the underlying functional networks are required. The research objective of this proposal is to develop a system identification methodology to extract functional network dynamics at the neuronal scale, and to apply it to large-scale recordings in order to probe and reveal the neuronal mechanisms that underlie behavior. The research approaches are to identify the temporal causal influences in neuronal ensembles in a robust and scalable fashion, capture abrupt changes in the covariance structure of the neuronal data, and extract the cross-frequency coupling of the network nodes with high resolution. This project addresses several outstanding challenges faced by existing methodologies, including loss of spectrotemporal resolution due to the use of sliding windows, lack of domain-specific models to capture the network dynamics, and ad hoc estimation procedures that result in highly biased network characteristics. By employing electrophysiology and two-photon calcium imaging data from the auditory systems of ferrets and mice, the proposed modeling and estimation framework will be used to investigate several fundamental questions pertaining to the functional organization of the networks involved in auditory processing. The project is expected to impact technology by providing signal processing solutions to be used in neural control systems. The research is also integrated with education and outreach activities including high school level workshops, undergraduate involvement, and course development.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.