A Fully Automatic System For Verified Computerized Stereoanalysis

描述（由申请人提供）：用于验证的计算机化立体分析的全自动系统概述对受过训练的用户与组织和图像交互的要求是使用无偏体视学（用于生物结构的精确量化的最先进方法）对生物微观结构进行更高通量分析的长期障碍。第一阶段研究通过验证计算机立体分析（VCS）解决了这一局限性，这是一种自动体视学分析的创新方法，与传统计算机体视学相比，吞吐量效率提高了6-9倍。第二阶段的工作将显微镜整合到“立体镜”中，这是一个集成的硬件-软件-显微镜系统，用于组织切片和存储图像的体视学分析。一阶体视学参数的验证研究。也就是说，体积、表面积、长度、数量的测量结果证实，在高S：N生物微结构的自动体视学分析中，在所述方法中基于颜色的检测方法实现了准确的结果。这些研究表明，可以通过消除两个剩余障碍来实现组织切片和存储图像的全自动体视学分析，这两个障碍将在第二阶段继续竞争更新中解决。在目标1中，部分由海军研究办公室资助开发的特征提取和微观结构分类应用程序将被集成到该计划中。新的应用程序（EMPII）将使用这些方法自动检测和分类生物学感兴趣的多态微结构，使用一系列特征计算，包括大小，颜色，边界，形状和纹理，并支持主动学习和支持向量机。目标2中的工作将通过为Stereologer系统配备由Stereologer系统控制的自动载玻片加载/卸载技术，消除计算机化体视学研究期间对载玻片的物理处理。该技术将使当前VCS计划的吞吐量效率大约增加一倍，并支持两个或多个相邻类别之间边界上的样本微结构的“人在回路”交互。目标3中的研究将严格检验以下假设：全自动体视学可以量化一阶和二阶体视学参数，与当前的黄金标准-非自动计算机化体视学相比，准确度没有损失，例如，手动立体发生器。如果这些研究证实了ESTII的准确性，那么全自动程序的商业化将促进在各种生物医学研究项目中测试科学假设的效率;减少计算机体视学研究的劳动力成本;加速我们对健康，长寿和疾病的生物过程的理解的增长;并加速开发用于人类疾病治疗管理的新方法。SRC及其战略合作伙伴在过去13年中对Stereologer系统的全球销售和支持证明了他们能够有效地将这项技术商业化。主要人员和参与机构：7 Peter R. Mouton博士(PI)，Stereology Resource Center，Chester，MD. 7 Dmitry Goldgof博士，南佛罗里达大学工程学院，坦帕，佛罗里达州7拉里霍尔，博士，南佛罗里达大学工程学院，坦帕，佛罗里达州7乔尔Durgavich，MS，系统规划和分析，亚历山大，弗吉尼亚州。7库尔特克雷默，硕士，计算机程序员，南佛罗里达大学，工程学院，坦帕，佛罗里达州7迈克尔E。卡尔霍恩博士，Sinq Systems，哥伦比亚，MD 公共卫生相关性：许多科学研究领域需要训练有素的专家对动物和人体组织的微观变化进行繁琐和重复的测量。该项目将产生一个计算机程序，执行这些测量与受过训练的专家相同的准确性，但在时间和成本上有显着的节省。允许科学家在更短的时间内完成更多的研究，将加速我们对促进健康和长寿的因素的理解，并加速人类疾病新疗法的开发。

DESCRIPTION (provided by applicant): A Fully Automatic System For Verified Computerized Stereoanalysis SUMMARY The requirement for a trained user to interact with tissue and images is a long-standing impediment to higher throughput analysis of biological microstructures using unbiased stereology, the state-of-the-art method for accurate quantification of biological structure. Phase 1 studies addressed this limitation with Verified Computerized Stereoanalysis (VCS), an innovative approach for automatic stereological analysis that improves throughput efficiency by 6-9 fold compared to conventional computerized stereology. Work in Phase 2 integrated VCS into the Stereologer", an integrated hardware-software-microscopy system for stereological analysis of tissue sections and stored images. Validation studies of first-order stereological parameters. i.e., volume, surface area, length, number, confirmed that the color-based detection methods in the VCS approach achieve accurate results for automatic stereological analysis of high S:N biological microstructures. These studies indicate that fully automatic stereological analysis of tissue sections and stored images can be realized by elimination of two remaining barriers, which will be addressed in this Phase II Continuation Competing Renewal. In Aim 1, applications for feature extraction and microstructure classification, developed in part with funding from the Office of Naval Research, will be integrated into the VCS program. The new application (VCS II) will use these approaches to automatically detect and classify polymorphic microstructures of biological interest using a range of feature calculations, including size, color, border, shape, and texture, with support from active learning and Support Vector Machines. Work in Aim 2 will eliminate physical handling of glass slides during computerized stereology studies by equipping the Stereologer system with automatic slide loading/unloading technology controlled by the Stereologer system. This technology will approximately double the throughput efficiency of the current VCS program and support "human-in-the-loop" interaction for sample microstructures on the border between two or more adjacent classes. The studies in Aim 3 will rigorously test the hypothesis that fully automatic VCS can quantify first- and second-order stereological parameters, without a loss of accuracy compared to the current gold-standard - non-automatic computerized stereology, e.g., manual Stereologer. If these studies validate the accuracy of VCS II, then commercialization of the fully automatic program will facilitate the throughout efficiency for testing scientific hypotheses in a wide variety of biomedical research projects; reduce labor costs for computerized stereology studies; hasten the growth of our understanding of biological processes that underlie health, longevity, and disease; and accelerate the development of novel approaches for the therapeutic management of human disease. Solid evidence that the SRC and its strategic partners can effectively commercialize this technology is demonstrated by their worldwide sales and support of the Stereologer system for the past 13 years. Key personnel and participating institutions: 7 Peter R. Mouton, Ph.D. (PI), Stereology Resource Center, Chester, MD. 7 Dmitry Goldgof, Ph.D., University of South Florida Coll. Engineering, Tampa, Fl. 7 Larry Hall, Ph.D., University of South Florida Coll. Engineering, Tampa, Fl. 7 Joel Durgavich, MS, Systems Planning and Analysis, Alexandria, VA. 7 Kurt Kramer, MS, Computer Programmer, University of South Florida, Coll. Engineering, Tampa, Fl. 7 Michael E. Calhoun, Ph.D., Sinq Systems, Columbia, MD PUBLIC HEALTH RELEVANCE: Many fields of scientific research require a trained expert to make tedious and repetitive measurements of microscopic changes in animal and human tissues. This project will produce a computer program that performs these measurements with equal accuracy to a trained expert, but with dramatic savings in time and costs. Allowing scientists to complete more research in less time will accelerate our understanding of the factors that promote health and longevity, and hasten progress toward the development of new treatments for human diseases.