Novel EM technologies for imaging neural network anatomy

描述(申请人提供)：我们的大脑包含数十亿个神经元，每个神经元都有数千个突触。它们共同构成了一个拥有数万亿连接的功能神经网络。它的规模和复杂性令人望而生畏，但从这种复杂性中涌现出感知和行为。我们如何理解如此庞大而复杂的网络的组织？一条前进的关键路径是研究神经元回路中结构和功能之间的关系。神经元的功能从根本上取决于它是如何连接的。因此，了解电路结构-连通性-与细胞功能之间的关系将有助于我们理解神经元和网络如何处理信息。不幸的是，详细的连接映射仍然很困难。一个关键障碍是数据吞吐量。最近， 高通量透射电子显微镜提高了成像速度，但仍然依赖人类进行繁重的人工样本采集和处理。目前的自动切片方法可以在胶带衬底上收集数千个电子显微镜(EM)样本，但由于胶带阻止了电子束的传输，因此与快速电子显微镜成像不兼容。以这种方式收集的连续切片目前使用扫描EM进行成像，扫描EM通常较慢。这项提议旨在开发新的技术，协同跨越自动样本采集和高速透射电子显微镜成像，以超越吞吐量障碍。我们将生产一种与透射电子显微镜成像兼容的用于样品采集的新型带状衬底，并使用它来收集数千个连续的薄片。此外，我们将设计和建造一个样品平台，用于对磁带收集的样品进行连续的透射电子显微镜成像。这些方法将允许在~1年内完成对局部哺乳动物皮质回路的高质量电磁成像，而使用传统方法则需要100多年。我们预计，使用这些新技术常规生成更大、更高质量的数据集也将加快其分析的进展。我们将立即使用这种方法来增加我们对皮层处理和组织的基本原理的理解。此外，随着更高吞吐量的电磁成像，我们最终将能够比较患病和健康的大脑，以评估电路连接是如何改变的，从而指导智能治疗策略。

DESCRIPTION (provided by applicant): Our brains contain billions of neurons, each with thousands of synapses. Together, they form a functional neural network with trillions of connections. Its scale and complexity is daunting, but from this complexity emerges perception and behavior. How do we understand the organization of such an immense and complex network? A key path forward is investigating the relationship between structure and function in neuronal circuits. The function of a neuron is fundamentally dependent on how it is connected. Therefore, understanding the relationship between circuit structure - connectivity - and cellular function will help us understand how neurons and networks process information. Unfortunately, detailed connectivity mapping remains difficult. One critical barrier is data throughput. Recently, high-throughput transmission electron microscopy (TEM) has increased the speed of imaging, but continues to rely on humans for laborious manual sample collection and handling. Current methods of automated sectioning can collect thousands of electron microscopy (EM) samples on a tape substrate, but are incompatible with fast TEM imaging because the tape prevents transmission of an electron beam. Serial sections collected in this manner are currently imaged using scanning EM, which is typically slower. This proposal aims to develop novel technologies that synergistically bridge automated sample collection and high-speed TEM imaging to transcend the throughput barrier. We will generate a novel tape substrate for sample collection that is compatible with TEM imaging and use it to collect thousands of serial thin sections. Additionally, we will engineer and build a sample stage for continuous TEM imaging of tape-collected samples. These methods would allow high-quality EM imaging of local mammalian cortical circuits to be completed in ~1 year compared to more than 100 years with conventional approaches. We expect that the routine generation of larger, high-quality datasets using these novel techniques will also accelerate advances in their analyses. We will immediately use this approach to increase our understanding of the fundamental principles underlying cortical processing and organization. Furthermore, with higher- throughput EM imaging, we will finally be poised to compare diseased and healthy brains to assess how circuit connectivity is altered, thereby directing intelligent treatment strategies.