Integration of 3D Slicer and BioImage Suite

描述（由申请人提供）：虽然越来越多的医学图像分析开源软件的可用性为集成来自不同来源的复杂任务的算法开辟了新的途径，但在源代码级别集成是一个重要的过程。例如，各种算法可能用不同的编程语言编写，并且/或者需要大量的底层库。这些问题使得这些算法的源集成通常是一个不可能的复杂和容易出错的任务。我们为所建议的工作采用的另一种策略是使用二进制积分。在这种情况下，算法已经以二进制形式预编译，并使用标准接口以类似于web浏览器中插件的方式集成在一起。该策略具有主要的潜在优势，包括代码许可和开发框架中立性，因为需要标准化的只是各种算法的调用。在这项工作中，我们建议推广和标准化3D切片器团队开发的现有二进制接口，并使用它使耶鲁BioImage套件软件包和3D切片器的功能无缝地提供给“其他”软件包。为此，我们提出了两个具体目标：(1)扩展和推广切片器执行层，以纳入额外的文件格式和对象，以实现集成；(2)扩展耶鲁BioImage套件软件包，以完全实现这一层。随着这个框架的开发，我们希望为其他图像分析包之间进一步的互操作性铺平道路。这种集成最终将为用户提供大量数据分析所需的工具，并允许开发人员专注于开发新的图像分析算法。我们预计，通过采用这种类型的接口，像NITRC这样的门户最终可以转换为可互操作组件的存储库，这些组件可以轻松集成以执行复杂的图像分析任务，而不需要所有开发人员使用完全相同的底层软件工程框架来实现他们的算法。相反，开发人员只需要标准化他们的算法如何被外部应用程序调用，这是一个更容易处理的任务。在未来，这种类型的接口还可以使这种算法作为互联网服务的使用成为可能，因此甚至有可能消除对本地软件安装的需求。

DESCRIPTION (provided by applicant): While the increasing availability of open source software for medical image analysis has opened up new avenues for the integration of algorithms from different sources for complex tasks, integration at the source code level is a non-trivial process. For example, the various algorithms may be written in different programming languages and/or require a large assortment of underlying libraries. These issues make the source integration of such algorithms often an impossibly complex and error-prone task. An alternative strategy, which we adopt for the proposed work is the use of binary integration. In this scenario, the algorithms are already precompiled in binary form and are integrated together using a standard interface in a manner similar to the use of plugins in web browsers. This strategy has major potential advantages including code license and development framework neutrality, as all that needs to be standardized is the invocation of the various algorithms. In this work, we propose to generalize and standardize an existing binary interface developed by the 3D Slicer team and use this to make functionality from the Yale BioImage Suite software package and 3D Slicer seamlessly available to the "other" package. To this end, we propose two specific aims: (1) Extend and generalize the Slicer Execution Layer to incorporate additional file formats and objects to enable the integration and (2) Extend the Yale BioImage Suite software package to fully implement this layer. With the development of this framework, we hope to pave the way for further interoperability among other image analysis packages. This integration will eventually provide users with a large collection of tools that they need for data analysis and allow developers to focus on developing novel image analysis algorithms. We anticipate that, with the adoption of this type of interface, a portal such as NITRC could thus be eventually transformed into a repository for interoperable components that could easily be integrated to perform complex image analysis tasks without requiring all developers to implement their algorithms using the exact same underlying software engineering framework. Instead, developers would only need to standardize how their algorithms are invoked by external applications which is a much more tractable task. This type of interface could also enable, in the future, the use of such algorithms as internet services, thus potentially even eliminating the need for local software installation. PUBLIC HEALTH RELEVANCE: Relevance to Public Health The development of standardized integration interfaces for medical image analysis will enable researchers to easily and seamlessly combine image analysis algorithms from a variety of sources to accomplish complex anal- ysis of their data. This integration will, in particular, enable newer and more effective algorithms to be adopted by the medical imaging user community without them having to be officially reimplemented and integrated into larger software packages. Rather, such algorithms could be designed to be available as standard plugins, in the same way that plugins are used in web browsers, and integrated by the users themselves in their customized workflows.