Groundwork for a Synchrotron MicroCT Imaging Resource for Biology (SMIRB)

拟议项目摘要/摘要 表型为研究形态变化提供了经典的生物学方法。在显微镜下， 成年人和儿童都能区分正常的细胞排列，比如一排柱状的肠道细胞， 并将这种正常表现与一组异常聚集的相同柱状细胞区分开来。 这种区别是通过描述性地比较异常表型和正常表型而发生的。对于一台计算机来说- 基于模型进行相同的区分，必须使用野生型引用进行比较。关于细胞的研究 表型分型广泛使用组织学检查来观察细胞的形态变化；然而，它诱导 分割伪影，从而阻止三维(3D)组织体积的可视化。为了绕过这些 挑战，我们计划利用一种无偏见的组织成像形式，称为X射线计算机 显微断层摄影术(MicroCT)。MicroCT将用于开发正常细胞表型的标准参照物 需要为自动分割和计算表型奠定基础。与传统不同 组织学，微型CT的成像技术，相当于一种非破坏性的组织学形式，用于小金属- 染色的生物样本，将使细胞类型在微米范围内的3D可视化成为可能。小说中的鳞片和 MicroCT的分辨率将从而揭示细胞的三维形态和空间分布，以量化和 描述脊椎动物和无脊椎动物模型的表型差异。结合了以下解决方案 基于机器学习的MicroCT成像，我们将对MicroCT数据集应用有监督的手动分割 创建用于形态评估的计算细胞识别机制。这一整合将包括 开源工具和设备，将支持社区驱动的研究和数据共享。作为一个额外的好处 在这种人工智能集成中，我们的计算方法将进一步帮助增加所需的大量数据 以检验细胞表型的统计关联性。 为了证明我们的方法在任何给定的细胞类型或 生物体，我们的这个项目的试点将集中在单变量方法的“肠道上皮细胞”跨越三个 生物(即斑马鱼、水蚤、小青蛙)。通过优先对这些特化细胞进行表型研究来 哺乳动物肠道细胞的类比关系，这项工作将为增加 我们对野生型细胞的三维生物结构的理解，并推进了形态细胞表型的研究 在整个生物体中进行定量、组织学和疾病识别。

Summary/Abstract of the Proposed Project Phenotypes provide a classical biological approach to studying morphological changes. Under a microscope, adults and children alike can distinguish a normal cell arrangement like a single row of columnar-shaped gut cells, and differentiate this normal presentation from an abnormal clustered group of the same columnar-shaped cells. This distinction occurs by descriptively comparing the abnormal phenotype against normality. For a computer- based model to make the same distinction, a wild-type reference must be used for comparison. Studies on cellular phenotyping widely use histological examination to visualize cell morphological changes; however, it induces sectioning artifacts, thereby preventing the visualization of 3-dimensional (3D) tissue volumes. To circumvent these challenges, we plan to take advantage of an unbiased form of tissue imaging, termed X-ray computed microtomography (microCT). MicroCT will be used to develop a standard reference of normal cellular phenotypes needed to lay a foundation for automated segmentation and computational phenotyping. Unlike conventional histology, the imaging technology of microCT, which amounts to a non-destructive form of histology for small metal- stained biological samples, will enable 3D visualization of cell types in the micrometer range. The novel scale and resolution of microCT will thereby reveal the 3D morphology and spatial distribution of cells to quantify and characterize phenotypic variations across vertebrate and invertebrate models. Combining the resolution of microCT imaging with machine learning, we will apply supervised-manual segmentation to microCT datasets to create computational cell recognition mechanisms for morphological assessment. This integration will include open-source tools and devices that will enable community-driven research and data-sharing. As an added benefit of this AI integration, our computational approach will further assist in augmenting large amounts of data required to examine the statistical association of cellular phenotypes. To demonstrate the applicability of our methodology for phenotypic characterization in any given cell type or organism, our pilot for this project will focus on the single-variable approach of ‘gut epithelial cells’ across three organisms (i.e., zebrafish, daphnia, froglets). By prioritizing phenotypic investigation of these specialized cells to the analogous relationship of mammalian intestinal cells, this work will establish a practical foundation to increase our understanding of the 3D biological structure of wild-type cells, and advance morphological cellular phenotyping in whole-organisms for quantitative, histological, disease recognition.