FASTCOUNT Next generation stereology platform for fast, accurate 3D cell counting

描述(申请人提供)：在过去的20年里，立体细胞计数使神经科学领域发生了革命性的变化，许多强调立体细胞计数在基础神经科学以及制药和生物技术研究中的影响的研究继续发表。然而，一项单独的研究可能需要几个月的时间，因为研究人员必须通过肉眼检查来决定是否根据数千个微观视野内的三维(3D)立体计数规则来计数每个细胞。由于这种对人工检查的依赖，体视学细胞计数仍然非常耗费人力和时间。因此，尽管存在已知的缺点和有偏见的结果，但更快的自动化、尽管不太准确的非立体细胞检测方法仍然被广泛使用。为了应对这一挑战，该项目旨在开发一种用于全自动立体细胞计数的创新显微镜系统(“FASTCOUNT”)，以我们的合作者Badri Roysam博士及其团队创建的用于自动化3D细胞检测的Farsight工具包中实施的技术为基础，并将其与为我们的Stereo Investigator(R)软件开发的技术相结合(在全球约1,000个实验室使用)。这个项目完全符合NIMH名为“实验室到市场：大脑和行为研究的工具”的计划。FASTCOUNT将首次允许研究人员进行自动立体细胞计数。对神经科学研究界和整个社会的好处将是BASC神经科学以及制药和生物技术研究的研究吞吐量增加(即至少快十倍)。产量的增加将使目前由于现有方法的劳动密集型性质而不可行的新类型的研究成为可能，从而导致新的发现。我们的项目将分两个阶段实施。在第一阶段，我们将创建一个包含细胞检测、学习和编辑的FASTCOUNT应用程序原型。然后，我们将对其在组织切片上的自动化3D细胞检测方面的性能进行基准测试，以用于物种、组织处理、细胞染色/标记、成像和图像预处理的各种组合。在第二阶段，我们将(I)改进和优化FASTCOUNT使用的自动立体3D细胞检测算法，(Ii)开发一个软件开发工具包，作为FASTCOUNT和第三方软件应用程序之间的接口，以便开发新的自动3D细胞检测方法，(Iii)开发新的图像数据管理功能，用于对经过最近组织清理程序(如Clarity)处理的组织标本进行自动立体细胞计数，以及(Iv)制定FASTCOUNT针对组织准备、标记、成像和前处理的建议操作程序。我们将与经验丰富的学术合作伙伴一起，在整个开发过程中对FASTCOUNT进行广泛的产品验证研究，以证明其优于手动立体细胞计数和有偏轮廓计数。

DESCRIPTION (provided by applicant): Stereologic cell counting has revolutionized the field of neuroscience over the last 20 years, and numerous studies highlighting the impact of stereologic cell counting in basic neuroscience and pharmaceutical and biotechnology research continue to be published. A single study, however, may take months since investigators must decide by visual inspection whether or not to count each cell according to three-dimensional (3D) stereologic counting rules within thousands of microscopic fields-of-view. Because of this reliance on manual inspection, stereologic cell counting has remained very labor intensive and time-consuming to perform. As a result, faster automated, though less accurate non-stereologic cell detection approaches have remained in widespread use despite their known disadvantages and biased results. In response to this challenge, this project aims to develop an innovative microscope system for fully-automated stereologic cell counting ("FASTCOUNT"), building upon technology implemented in the FARSIGHT toolkit for automated 3D cell detection created by our collaborator, Dr. Badri Roysam and his team, and combining it with technology developed for our Stereo Investigator(R) software (used in approximately 1,000 laboratories worldwide). This project is fully in line with the NIMH program titled Lab to Marketplace: Tools for Brain and Behavioral Research. FASTCOUNT will allow investigators, for the first time, to perform automated stereologic cell counting. The benefit for the neuroscience research community - and society in general - will be increased research throughput (i.e., at least ten times faster) in basc neuroscience and pharmaceutical and biotechnology research. Increased throughput will make possible new kinds of studies currently not feasible due to the labor intensive nature of the existing methods, thus leading to new discoveries. Our project will be implemented in two phases. During Phase I, we will create a prototype FASTCOUNT application incorporating cell detection, learning, and editing. We will then benchmark its performance in automated 3D cell detection on tissue sections for various combinations of species, tissue processing, cell staining/labeling, imaging, and image pre-processing. During Phase II, we will (i) improve and optimize the algorithms for automated stereologic 3D cell detection used in FASTCOUNT, (ii) develop a Software Development Kit as interface between FASTCOUNT and third party software applications to allow development of new automated 3D cell detection methods, (iii) develop novel image data management functionality for automated stereologic cell counting on tissue specimens processed with recent tissue clearing procedures such as CLARITY, and (iv) develop recommended operating procedures for FASTCOUNT addressing tissue preparation, labeling, imaging, and pre-processing. Together with experienced academic collaboration partners, we will perform extensive product validation studies of FASTCOUNT throughout development to demonstrate its superiority over both manual stereologic cell counting and biased profile counting.