Department of Mechanical Engineering, University of Minnesota, Twin Cities, Minneapolis, MN, USA.
Department of Neuroscience, University of Minnesota, Twin Cities, Minneapolis, MN, USA.
Nat Methods. 2024 Nov;21(11):2171-2181. doi: 10.1038/s41592-024-02434-z. Epub 2024 Oct 7.
Technologies that can record neural activity at cellular resolution at multiple spatial and temporal scales are typically much larger than the animals that are being recorded from and are thus limited to recording from head-fixed subjects. Here we have engineered robotic neural recording devices-'cranial exoskeletons'-that assist mice in maneuvering recording headstages that are orders of magnitude larger and heavier than the mice, while they navigate physical behavioral environments. We discovered optimal controller parameters that enable mice to locomote at physiologically realistic velocities while maintaining natural walking gaits. We show that mice learn to work with the robot to make turns and perform decision-making tasks. Robotic imaging and electrophysiology headstages were used to record recordings of Ca activity of thousands of neurons distributed across the dorsal cortex and spiking activity of hundreds of neurons across multiple brain regions and multiple days, respectively.
能够以多个时空尺度记录细胞分辨率的神经活动的技术通常比正在记录的动物大得多,因此只能记录固定头部的动物。在这里,我们设计了机器人神经记录设备——“颅外骨骼”——当它们在物理行为环境中导航时,帮助老鼠操纵比老鼠大几个数量级且更重的记录头。我们发现了最佳的控制器参数,使老鼠能够以生理上现实的速度运动,同时保持自然的行走步态。我们发现老鼠学会了与机器人一起工作,以完成转弯和决策任务。机器人成像和电生理记录头分别用于记录分布在背皮层的数千个神经元的 Ca 活动和多个脑区数百个神经元的放电活动,并持续数天。