CRCNS: Large-scale computational reconstruction of three-dimensional neural

描述（由申请人提供）：神经生理学建模是理解人类大脑的重要工具，然而最先进的模型很少受到解剖学数据的约束。通过提供真实的神经解剖学数据，高倍率、连续切片显微镜图像有可能扩展神经生理学建模领域。该项目解决了从串行切片显微镜中建立神经元三维（3D）连接图的问题。截面数据由一堆高分辨率的二维（2D）图像组成，这些图像旨在捕捉细长神经元过程的横截面。这项工作的重点是两个驱动生物学应用。第一个应用是开发哺乳动物视网膜神经节细胞的完整连接图。二是野生型和突变型斑马鱼视束轴突组织的研究。在这两种应用中，高分辨率串行剖面数据的复杂性和巨大尺寸使得人类无法解释它们。提出了两个研究领域。首先是连续显微数据处理的基础技术发展。这些数据显示出独特的结构和统计特性（例如纹理），这提出了一系列超越2D和3D图像处理最先进技术的独特挑战。第二个领域是实用的计算工具的发展，科学家可以用这些工具定量地分析与这个问题相关的非常大的数据集。为了最大限度地向研究界展示它，这个项目中产生的软件将作为开源工具包ITK （www.itk.org）的一部分公开提供。这项研究将允许科学家通过开发必要的算法和计算工具进行处理和可视化，系统地分析串行剖面数据集。这将使我们更好地理解神经回路是如何在单个细胞的水平上构建的，这对于提高我们对人类大脑基本线路的认识非常重要。显著改善的成像、分析和可视化资源对于表征与大规模神经连接异常相关的神经疾病（如颞叶癫痫、视网膜变性、脊髓损伤和药物依赖）的疾病进展、治疗骤停和再生模式至关重要。

DESCRIPTION (provided by applicant): Neurophysiological modeling is an important tool in understanding the human brain, and yet state-of-the-art models are poorly constrained by anatomical data. High-magnification, serial-section microscopy images have the potential to expand the field of neurophysiological modeling by providing ground-truth neuroanatomical data. This project addresses the problem of building three-dimensional (3D) connectivity maps for neurons from serial-section microscopy. Sectional data consists of stacks of very high-resolution, two-dimensional (2D) images that are oriented to capture cross sections of elongated neuronal processes. The work focuses on two driving biological applications. The first application is the development of complete connectivity maps for ganglion cells in the mammalian retina. The second is the study of the organization of axons in the optic tract of wildtype and mutant zebrafish. In both applications, the complexity and vast size of the high-resolution, serial-section data make them impractical for human interpretation. Two areas of research are proposed. The first is basic technology development for serial microscopy data processing. Such data exhibits unique structural and statistical properties (e.g. textures), which present a distinct set of challenges that surpass the state-of-the-art in 2D and 3D image processing. The second area is the development of practical, computational tools with which scientists can quantitatively analyze the very large data sets associated with this problem. In order to maximize its exposure to the research community, the software produced in this project will be made publicly available as part of the open source toolkit ITK (www.itk.org). This research will allow scientists to systematically analyze serial section data sets by developing the necessary algorithms and computational tools for processing and visualization. This will result in an improved understanding how neural circuits are constructed at the level of single cells, which is important in advancing our knowledge of the basic wiring of the human brain. Significantly improved imaging, analysis and visualization resources will be critical in characterizing disease progressions, therapeutic arrests, and regeneration patterns in neurological diseases, such as temporal lobe epilepsy, retinal degenerations, spinal injury, and drug dependence, which are associated with anomalies in large-scale neural connectivity.