NeuroJSON - A Scalable, Searchable and Verifiable Neuroimaging Data Platform

传统的基于文件的神经成像数据管理和集成策略表明， 日益增加的限制，以适应迅速增长的规模和复杂性，今天的 神经成像数据。当今的许多企业都需要复杂的软件和硬件管道， 神经影像学研究已经产生了许多平台特定的数据文件， 更广泛的研究社区进行解析、交流和理解。现代神经影像学研究 不仅受到解析和管理严格且多样的文件类型的挑战的阻碍，而且还缺乏 系统数据验证、查询、操作和集成的统一接口，从而限制了其 处理大型数据集的能力。创建面向未来、高度可扩展且维护成本低的数据存储 和传播平台是广泛和快速增长的神经影像学社区非常需要的。的 未来的神经影像数据管理必须是可扩展的、可搜索的、可验证的，并且能够 适应从涉及多模态的复杂范例生成的高度复杂的分层数据 输入。受到NoSQL数据库平台巨大成功的启发，我们设想统一数据库 用于管理复杂神经成像数据和交换人类可读分层数据记录的接口 将非常适合解决下一代神经影像数据管理的迫切需求。在这 项目，我们的目标是巩固一系列易于采用，易于扩展，人类可读的JSON（http：json.org） 基于数据文件规范，以系统地协助现有和 新兴的神经成像数据集。这些JSON编码的通用数据文件使用户能够轻松地利用 高度可扩展和高性能的NoSQL数据库，如CouchDB和MongoDB， 传播NIH资助的大型神经成像公共数据集，并实现验证和自动化。我们集团 自2011年以来一直是基于JSON的科学数据存储的主要贡献者。我们公开发表了- 源规范（http：openjdata.org），以标准化神经成像数据的交换， 格式，如NIfTI/GIfTI/SNIRF，为特定应用的采用奠定了坚实的基础。在这个项目中， 我们寻求进一步开发、巩固和传播基于JSON的数据交换规范， NoSQL数据库。我们与主要的神经影像数据分析利益相关者建立了合作关系，例如 FreeSurfer，SPM，FieldTrip，HOMER，BrainStorm.在这个项目结束时，我们将能够1）开发一套 稳定的通用文件格式，极大地现代化了神经成像应用程序中的数据共享， 维护和扩展，2）为用户提供开源工具来构建NoSQL数据库后端， 促进公共/私人数据库的集成和自动化，实现查询、验证和扩展。 这个项目的成功将产生一个强大的数据交换平台，以促进方便的数据共享， 促进可重复的研究，并在广泛的神经影像学社区中建立有效的合作。

Traditional file-based neuroimaging data management and integration strategies have shown increasing limitations in accommodating the meteoric growth in both the scale and complexity of today’s neuroimaging data. The sophisticated software and hardware pipelines required in many of today’s neuroimaging studies have produced numerous platform-specific data files that are increasingly difficult to parse, exchange, and understand by the broader research community. Modern neuroimaging studies are hampered by not only the challenge of parsing and managing rigid and diverse file types, but also the lack of a unified interface for systematic data validation, query, manipulation and integration, thereby limiting its ability to handle large datasets. Creating a future-proof, highly scalable, and low-maintenance data storage and dissemination platform is highly desirable for the broad and rapidly growing neuroimaging community. The future of neuroimaging data management must be scalable, searchable, verifiable, and capable of accommodating highly complex hierarchical data generated from complex paradigms involving multi-modal inputs. Inspired by the great success of NoSQL database platforms, we envision that a unified database interface for managing complex neuroimaging data and exchanging human-readable hierarchical data records will be highly suitable to address the urgent needs of next-generation neuroimaging data management. In this project, we aim to solidify a series of easy-to-adopt, easy-to-extend, human-readable JSON (http://json.org) based data file specifications to systematically assist the storage, exchange and integration of existing and emerging neuroimaging datasets. These JSON-encoded universal data files readily enable users to utilize highly scalable and high-performance NoSQL databases, such as CouchDB and MongoDB, to rapidly disseminate large, NIH-funded neuroimaging public datasets, and enable validation and automation. Our group has been a major contributor to JSON-based scientific data storage since 2011. We have published open- source specifications (http://openjdata.org) to standardize the exchange of neuroimaging data for common formats such as NIfTI/GIfTI/SNIRF, building a solid foundation for application-specific adoptions. In this project, we seek to further develop, solidify, and disseminate JSON-based data exchange specifications and NoSQL databases. We have built collaborations to major neuroimaging data analysis stakeholders, such as FreeSurfer, SPM, FieldTrip, HOMER, BrainStorm. At the end of this project, we will be able to 1) develop a set of stable universal file formats that greatly modernize data sharing in neuroimaging applications, easing future maintenance and extension, and 2) provide open-source tools for users to build NoSQL database backends to facilitate integration and automation of public/private databases, enabling query, validation, and scale-up. Success in this project will result in a robust data exchange platform to facilitate convenient data sharing, promote reproducible research, and forge efficient collaborations among a broad neuroimaging community.