Development And Applications Of The Open Microscopy Envi

开放显微镜环境（http：openmicroscopy.org）是六年前由麻省理工学院的Ilya Goldberg博士在Peter Sorger博士的小组中发起的一个项目。其目的是为计算细胞生物学开发一个信息框架-生物信息学的一个子专业，称为“图像信息学”。这个开源框架由一个数据库、几个分析模块和一个将模块与数据库联系起来的应用程序接口（API）组成。该数据库为通过分析图像获得的生物信息提供语义框架和数据模型。它还跟踪图像本身以及对它们执行的所有分析。数据库也是分析模块之间的通信链路，允许多路复用分析算法。最后，整个系统提供基于Web的Java用户界面，允许远程交互。 该软件作为一个图像存储库和管理系统的科学图像数据的非常大的集合。该软件目前部署在几个研究实验室中，具有相当大的成像需求，包括几个TB大小的存储库。 在过去的一年里，我们已经针对最终用户发布了一个公开版本（版本2.6）。该版本的重点是增强最终用户的可用性，并改进分析系统。去年，我们开发了一种机制，让用户可以在他们的图像中添加自定义手动注释，今年我们发布了对这种机制及其使用的描述，并将其包含在OME软件的最新公开版本中。自定义半结构化注释正在成为形式化本体和自由文本注释之间的一种有价值的折衷。最终用户通常是知识最渊博的，因此是定义本体论术语的最合适的人，但通常缺乏定义本体论的形式主义的专业知识。允许最终用户定义他们自己的术语，而不需要增加一个完全结构化的本体的形式化，实现了建立正式本体的最重要目标：本体术语的识别和定义。一旦术语被定义并且开始在该领域中使用，它们就可以被用来构建更正式的本体。我们已经使用这种确切的机制来定义描述原位实验的本体论术语。 在过去的一年中，MATLAB接口的实验性实现在我们的软件分发网站上提供，并吸引了来自学术界和工业界的一些用户。在本年度，我们改进了这个系统，使其能够在集群和网络计算机上工作，整合了我们的图像分类算法，并在几个大型图像数据集上验证了该系统。正在进行的工作包括完善安装程序，为公开发布做准备。 为OME开发的通用科学图像格式（OME XML）继续获得仪器制造商的认可。今年，我们在威斯康星州大学的合作者开发了这种格式的高性能版本，称为OME-TIFF，并向公众发布。OME-XML正在进行的工作包括根据显微镜制造商的要求对格式进行改进和添加。 该项目的第二个主要工作涉及为OME平台构建信息可视化工具。Harry Hochheiser博士为该系统开发了一个交互式浏览器，该系统基于同时遍历多个组织层次结构。今年也出版了这部作品。此外，Hochheiser博士还帮助修改了基于网络的用户界面，使其能够由最终用户进行定制。多亏了他的努力和Josiah约翰斯顿的帮助，该系统可以支持用户可选择的几种并发信息显示风格。我们使用这种机制来开发一个外部的“客人”用户的接口，以便向公众提供科学的图像数据。在接下来的一年里，我们将填充并发布一个公共数据库，其中包含来自Minoru Ko博士小组的密度杂交数据的图像。 作为增加OME功能和可用性的持续努力的一部分，Arpun Nagaraja和Josiah约翰斯顿开发了一种机制，允许OME分析系统使用预编译的MATLAB脚本，从而提高了用户获得MATLAB许可证以执行大多数图像分析的要求。此外，我们已经开始解决长期以来对OME的批评，主要是最终用户文档。与邓迪大学的细胞生物学家大卫·希夫曼博士合作，我们简化了定量显微镜中越来越常见的任务：在三维荧光显微镜中发现和测量亚细胞结构。我们在Bio Techniques上发表了该方案的描述，以及安装OME、运行FIndSpots算法和以图形方式显示结果以及使用Excel等外部应用程序进行进一步显示和分析的详细在线手册。由于Excel是一种广泛用于操作和可视化生物学定量数据的工具，因此我们开发了一种用于Excel的集成OME连接器，允许用户查询实时OME数据库，以将分析结果直接收集到Excel中。

The Open Microscopy Environment (http://openmicroscopy.org) is a project started by Dr. Ilya Goldberg at MIT six years ago in the group of Dr. Peter Sorger. The purpose is to develop an information framework for computational cell biology - a sub-specialty of bioinformatics called 'Image Informatics'. This open-source framework consists of a database, several analytic modules, and an application program interface (API) that ties the modules to the database. The database provides a semantic framework and a data model for biological information obtained by analyzing images. It also keeps track of the images themselves, and of all the analyses performed on them. The database is also the communication link between analysis modules permitting the multiplexing of analysis algorithms. Finally, the entire system provides web-based and Java user interfaces allowing for remote interaction. This software functions as an image repository and management system for very large collections of scientific image data. This software is currently deployed in several research labs with considerable imaging needs, and includes repositories several terabytes in size. Over the past year we have made one public release of this software targeted at end-users (version 2.6). This release focused on enhancing usability for end-users, and refinement of the analysis system. In the previous year, we developed a mechanism for users to add custom manual annotations to their images, and this year published a description of this mechanism and its use, as well as included it in the latest public release of OME software. Custom semi-structured annotations are emerging as a valuable compromise between formalized ontologies and annotations in free text. The end-user is most often the most knowledgeable and therefore the most appropriate person to define ontological terms, but is usually lacking expertise in the formalism of defining ontologies. Allowing the end-user to define their own terms without the added formalism of a fully structured ontology accomplishes the most important goal of establishing a formal ontology: Identification and definition of ontological terms. Once the terms are defined and begin to see use in the field, they can then be used to construct more formal ontologies. We have used this exact mechanism to define ontological terms for describing in-situ experiments. In the previous year, an experimental implementation of the MATLAB interface was made available on our software distribution site, and has attracted several users from academia and industry. In the current year, we have refined this system to work on clustered and networked computers, integrated our image classification algorithms and validated the system on several large image data sets. Ongoing work involves refinement of the installation procedure in preparation for a public release. The universal scientific image format developed for OME (OME XML) continues to gain acceptance with instrument manufacturers. This year, our collaborators at University of Wisconsin developed a high performance version of this format called OME-TIFF, and released it to the public. Ongoing efforts with OME-XML include refinements and additions to the format based on requests from microscope manufacturers. The second major effort in this project involves building information visualization tools for the OME platform. Dr. Harry Hochheiser has developed an interactive browser for this system based on simultaneous traversal through several organizational hierarchies. This year also saw the publication of this work. Additionally, Dr. Hochheiser was instrumental in modifying the web-based user interface to allow it to be customized by the end-user. Thanks to his efforts with assistance by Josiah Johnston, the system can support several concurrent styles of information display selectable by the user. We used this mechanism to develop an interface for external "guest" users in order to make available scientific image data to the general public. Over the coming year, we will populate and publish a public database containing images of in-sity hybridization data from Dr. Minoru Ko's group. As part of the continuing effort in adding functionality and usability to OME, Arpun Nagaraja and Josiah Johnston have developed a mechanism to allow the OME analysis system to use pre-compiled MATLAB scripts, thereby lifting the requirement that the user obtain a MATLAB license in order to perform most image analysis. Additionally, we have begun to address a long-standing criticism of OME, mainly end-user documentation. Working with Dr. David Schiffmann, a cell biologist at the University of Dundee, we streamlined an increasingly common task in quantitative microscopy: finding and measuring sub-cellular structures in 3-D fluorescence microscopy. We published a description of this protocol in Bio Techniques, and detailed on-line manuals for installing OME, running the FIndSpots algorithm, and displaying the results graphically as well as using external applications such as Excel for further display and analysis. Because Excel is a widely used tool for manipulating and visualizing quantitative data in biology, we have developed an integrated OME connector for Excel which allows the user to query a live OME database to collect analysis results directly into Excel worksheets.