Mesoscopic microscopy for ultra-high speed and large-scale volumetric brain imaging

用于超高速和大规模脑体积成像的介观显微镜

• 批准号: 10634911
• 负责人: Ji Yi
• 金额: $ 53.87万
• 依托单位: JOHNS HOPKINS UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-04-01 至 2027-03-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10634911
• 关键词: 3-Dimensional; Action Potentials; Address; Adopted; Algorithms; Animals; Area; Behavior; Biological; Bite; Brain; Brain imaging; Budgets; Central Nervous System; Code; Collection; Communication; Compensation; Data; Data Set; Detection; Development; Dyes; Electronics; Eye; Fishes; Fluorescence; Generations; Goals; Image; Kinetics; Label; Larva; Lasers; Light; Maps; Mechanics; Membrane; Membrane Potentials; Methods; Microscopy; Modeling; Motion; Motor; Mus; Nervous System; Neurons; Neurosciences; Noise; Optics; Pattern; Penetration; Performance; Photobleaching; Photons; Photoreceptors; Posture; Resolution; Retina; Sampling; Scanning; Signal Transduction; Speed; Spinal Cord; Stimulus; Structure; System; Techniques; Tectum Mesencephali; Testing; Three-Dimensional Imaging; Time; Tissues; Training; Variant; Work; Zebrafish; body position; brain volume; computerized data processing; data analysis pipeline; deep learning; deep learning algorithm; deep learning model; denoising; design; detector; flexibility; imaging system; improved; instrumentation; lens; millimeter; motor behavior; nervous system imaging; network architecture; neural; neural circuit; neuroimaging; optic flow; processing speed; prototype; response; spatiotemporal; supervised learning; three-dimensional modeling; transmission process; visual stimulus; voltage

PROJECT SUMMARY The brain is built on billions of neural connections in a highly organized 3D hierarchy. At the same time, neural activity is highly dynamics that requires kilohertz imaging rate to capture action potentials and sub-threshold voltage signals, the fundamental bit for neural communication. While the recent advent of genetically encoded voltage indicators (GEVIs) makes it possible to optically record the neural membrane voltage, the technical challenges are profound in imaging millimeter-scale volumetric voltage imaging at kilohertz with cellular resolution. In this proposal, we aim to address the challenges by developing a one-photon mesoscopic (i.e. millimeter scale field of view, FOV) volumetric voltage imaging, using mesoscopic oblique plane microscopy (Meso-OPM). Our technique will image >1.8 mm2 FOV, >0.1 mm depth penetration at 1 KHz, capable of recording voltage signals across an entire nervous system of a Zebrafish larva. The bright and stable GEVIs Voltron with JF525 dye will be used in our proposed work. Meso-OPM is a variant of light sheet microscopy (LSM), with a single primary objective lens instead of two in conventional LSM. The simplified optical design allows 1) leveraging high photon efficiency in LSM; 2) integrating ultra-fast passive optical scanning to achieve >1 MHz frame rate; and 3) flexible optical designs for millimeter FOV and cellular resolution. In addition to the technical challenges for large-scale ultrafast 3D imaging, the effective data processing pipeline for massive data is also highly desirable. To this end, we propose a robust and efficient deep learning framework to perform self- supervised 4D denoising and neuron segmentation. The pipeline enable massive data processing at 10 volume per second for the downstream neuroscience studies. Finally, to demonstrate the utility of proposed techniques, we will image Zebrafish in response to optic flow by a drifting grating visual stimuli. We will identify neural circuitry responsible to the motion compensation to the optic flow (i.e. maintaining body position when presented drifting grating) from eyes all the way to spinal cord. Altogether, this proposal will greatly improve our capability of dissecting large-scale neural circuitry, and the sub-sequent modeling and creation of artificial neural circuits.

项目摘要 大脑建立在高度有条理的3D层次结构中数十亿个神经联系。同时，神经 活动是高度动态的，需要Kilohertz成像率才能捕获动作电位和亚阈值 电压信号，神经通信的基本位。而最近的基因编码的出现 电压指示器（GEVIS）使得可以光学记录神经膜电压，技术电压 挑战在成像毫米规模的体积电压成像中是深刻的 解决。在此提案中，我们旨在通过开发一个单次介质（即 毫米尺度的视野，FOV）体积电压成像，使用介质倾斜平面显微镜 （Meso-opm）。我们的技术将图像> 1.8 mm2 fov，> 0.1 mm深度穿透在1 kHz处，能够 记录斑马鱼幼虫整个神经系统的电压信号。明亮而稳定的Gevis 带有JF525染料的Voltron将在我们的拟议工作中使用。 Meso-Opm是光片显微镜的变体 （LSM），具有单个主要物镜镜头，而不是传统LSM中的两个。简化的光学设计 允许1）利用LSM中的高光子效率； 2）整合超快速的被动光学扫描以达到> 1 MHz帧速率； 3）用于毫米FOV和细胞分辨率的柔性光学设计。除了 大规模超快3D成像的技术挑战，这是大量数据的有效数据处理管道 也是非常可取的。为此，我们提出了一个强大而有效的深度学习框架来执行自我 监督的4D denoising和神经元细分。该管道以10卷为10的大规模数据处理 下游神经科学研究的每秒。最后，为了证明提议的技术的实用性， 我们将通过漂移的光栅视觉刺激来对斑马鱼进行响应。我们将确定神经电路 负责对视频流动的运动补偿（即在出现漂移时保持身体位置 从眼睛一直到脊髓。总之，该提议将大大提高我们的能力 解剖大规模的神经回路，以及人工神经回路的下一步建模和创建。