A robust, low-cost platform for EM connectomics

项目摘要/摘要 在过去的十年里，连续切片电子显微镜已经成为一种研究 神经回路的连接，从哺乳动物的局部回路到整个无脊椎动物的大脑。最近，重点是 该领域一直在创建越来越大的数据集，而在以下方面花费的努力相对较少 将EM连接学的工具提供给大量的电路神经学家。障碍存在于 多个级别。手工进行连续切片的方法非常困难，而自动方法 需要复杂、昂贵、难以部署的设备。高通量扫描EM仅限于多个 波束方法是极其昂贵的。变速器EM的成本要低得多，但是自动化的 切片的方法仍然具有挑战性，需要昂贵的衬底，很难制造和 很难使用。 我们建议开发一种新的方法，该方法已经由我们的集团和我们的行业合作伙伴建立原型，以建立 一个强大的平台，经过优化以实现最广泛的采用。该系统将以开放源码为中心 串联式分段机器人实现了一种新颖的采集方法。我们的目标是创建一个可以使用的系统 从目前的技术水平(1mm3或更大)到可分割的小体积，有各种规模 并照例进行成像。到目前为止，每一本出版的EM连接学卷都需要多年的努力。 相反，我们的目标是在以下背景下将体积重建用作一种分析，而不是其本身的目的 其他实验。在最后一年，我们将创建数据集来测试该方法的灵活性和健壮性 通过创建从一侧50微米到包含1 mm3的非常大的体积的EM体积。

Project Summary/Abstract Over the past decade, serial-section electron microscopy has come into its own as a method to study the connectivity of neural circuits, from local circuits in mammals to entire invertebrate brains. Recently, the emphasis in the field has been to create increasingly large data sets, while comparatively little effort has been spent on making the tools of EM connectomics available to a large number of circuit neuroscientists. Obstacles exist at multiple levels. Manual approaches to serial sectioning are prohibitively difficult, while automated approaches require complex, expensive equipment that is difficult to deploy. High throughput scanning EM is limited to multi- beam approaches that are extremely expensive. Transmission EM is far less expensive, but automated approaches to sectioning remain challenging and require expensive substrates that are hard to manufacture and difficult to use. We propose to develop a new approach, already prototyped by our group and our industry partner, to establish a robust platform optimized to achieve the widest possible adoption. The system will center on an open source serial sectioning robot implementing a novel collection approach. The goal is to create a system that can be used at a variety of scales, from the current state of the art (1 mm3 or greater), to small volumes that can be sectioned and imaged routinely. Up to now, each published EM volume for connectomics has required a multi-year effort. Instead, our goal is to use volume reconstruction as an assay, rather than an end unto itself, in the context of other experiments. In the final year, we will create data sets that test the flexibility and robustness of the approach by creating EM volumes ranging from 50µm on a side to very large volumes encompassing >1 mm3.