Robot assisted brain-wide neural recordings and comprehensive behavioral monitoring in freely behaving mice

总结 大脑处理感官输入，并将这些信息与内部大脑状态联系起来，产生驱动运动和认知行为的信号。底层的计算分布在几个解剖学和功能上不同的大脑区域。因此，神经科学需要工具来同时研究执行自然行为的动物的多个大脑区域的高空间和时间分辨率。虽然在啮齿类动物的大规模神经记录技术方面取得了显着进展，但大多数技术需要限制头部，从而显着限制了行为能力。虽然已经开发了几种头戴式小型化记录设备，但这些设备通常限于小鼠体重的约10%（2-4g），这导致性能和能力的大幅折衷。为了解决这个问题，我们提出了一种根本不同的方法，消除了对用于自由行为动物的神经接口的日益小型化的需要。我们将设计一种外骨骼，它连接到动物的头部，能够操纵复杂的神经记录硬件，重量比老鼠高一个数量级，在6个自由度（DOF）中响应动物的运动。有三个具体目标。在目标1中，我们将设计一个6自由度的颅骨外骨骼，能够操纵一个1 kg的头部平台，对接到一个自由行为的小鼠。外骨骼将具有传感器阵列，以实现5 nM尺度的6 DOF力感测，并使用零阻抗闭环控制。对于目标2，我们将设计和构建一个兼容的像素驱动头台，并实现高效的啮齿动物机器人接口。头台将包括用于神经探针操作的6自由度力传感硬件和用于行为跟踪的摄像头。我们将通过在小鼠导航的开放领域中记录单个Neuropixels探针来优化系统。在Aim 3中，我们将利用颅骨外骨骼和像素驱动的功能，对执行新型物体识别和定位任务的小鼠进行全脑记录。这将包括轨迹规划策略，以及通过聚合物头骨将头台引入大脑的多达6个Neuropixels探针，并建立自由移动小鼠的多部位记录程序。该提案汇集了机器人专家，神经工程师（Kodandaramaiah博士实验室）和系统神经科学家（Ebner博士实验室）的跨学科团队，以解决神经科学中的这一紧迫问题。当完全开发后，拟议的颅骨外骨骼和像素-驱动器将能够记录分布在整个大脑中的1000个接触，显着提高行为的广度和复杂性，可以使用高性能，大规模神经记录和行为跟踪设备进行研究。

SUMMARY The brain processes sensory inputs and contextualizes this information with internal brain states, generating the signals that drive motor and cognitive behaviors. The underlying computations are distributed across several anatomically and functionally distinct brain regions. Therefore, neuroscience requires tools to simultaneously investigate multiple brain regions at high spatial and temporal resolution in animals performing naturalistic behaviors. While remarkable progress has been made in technologies capable of large-scale neural recordings in rodents, most require restraining the head, significantly limiting the behavioral repertoire. Although several head-mounted, miniaturized recording devices have been developed, these devices are typically limited to ~10% (2-4g) of a mouse’s body weight that results in large tradeoffs in performance and capabilities. To solve this problem, we propose a fundamentally different approach that eliminates the need for ever-greater miniaturization of neural interfaces for use in freely behaving animals. We will engineer an exoskeleton that attaches to the animal’s head and is capable of maneuvering sophisticated neural recording hardware, weighing an order of magnitude higher than a mouse, in 6 degrees of freedom (DOF) in response to an animal’s movements. There are three Specific Aims. In Aim 1 we will engineer a 6 DOF Cranial-exoskeleton capable of manipulating a 1kg head stage docked to a freely behaving mouse. The exoskeleton will have an array of sensors to achieve 6 DOF force sensing at the 5 nM scale and use zero-impedance closed-loop control. For Aim 2 we will design and build a compatible Pixel-drive headstage and achieve efficient rodent-robot interfacing. The headstage will incorporate 6 DOF force-sensing hardware for neural probe manipulation and cameras for behavioral tracking. We will optimize the system by recording from a single Neuropixels probe in mice navigating an open field. In Aim 3 we will leverage the capabilities of the Cranial-exoskeleton and Pixel-drive for brain-wide recordings in mice performing novel object recognition and location tasks. This will include strategies for trajectory planning and introducing up to 6 Neuropixels probes with the headstage into the brain via polymer skulls and establish procedures for multi-site recordings in freely locomoting mice. The proposal brings together an interdisciplinary team of roboticists, neural engineers (Dr. Kodandaramaiah lab), and systems neuroscientist (Dr. Ebner lab) to tackle this pressing problem in neuroscience. When fully developed, the proposed Cranio-exoskeleton and Pixel- the drive will enable recordings from 1000s of contacts distributed throughout the brain, significantly enhancing the breadth and complexity of behaviors that can be studied using high performance, large scale neural recording, and behavior tracking devices.