Sharing Neuroimaging Data with Squirrel - A Relational Data Format to Store Raw to Analyzed Data and Everything in...

Area of Science:

• Neuroimaging research
• Data science
• Scientific data management

Background:

• Reproducibility and data aggregation in human subjects imaging research face challenges due to the lack of a universal, user-friendly data format.
• The widely adopted Brain Imaging Data Structure (BIDS) format is becoming increasingly complex, with extensive documentation, hindering its implementation.
• Existing standards struggle to balance data complexity with format simplicity.

Purpose of the Study:

• To introduce a simple yet generalizable data sharing specification, the squirrel format, for neuroimaging data.
• To address the need for a flexible format that minimizes complexity while handling diverse imaging data and analysis steps.
• To facilitate easy sharing and dissemination of neuroimaging data packages.

Main Methods:

• Developed a relational hierarchy with a structured representation of neuroimaging data collection and analysis.
• Created a generalizable specification to accommodate any neuroimaging modality.
• Classified all research data into ten distinct object types, minimizing redundancy using relational database principles.

Main Results:

• The squirrel format successfully stores subject information, experimental parameters, raw and analyzed data, and analysis methods.
• Data is organized in a human-readable hierarchy, enabling easy sharing and dissemination.
• The format's 'chunking' approach, using ten object types, simplifies the management of complex neuroimaging datasets.

Conclusions:

• The squirrel format offers a simple, flexible, and generalizable solution for sharing neuroimaging data, enhancing reproducibility and data aggregation.
• Its design objectives are met by providing a human-readable, organized structure that accommodates all stages of neuroimaging research.
• The squirrel format can be extended beyond neuroimaging to any imaging research, promoting broader data sharing and collaboration.