BIDS

BIDS
Name	BIDS
Abbreviation	BIDS
Launched	2015
Latest release	1.8.0
Field	Neuroimaging, neuroinformatics

BIDS

BIDS is a community-driven specification for organizing, describing, and sharing neuroimaging and associated behavioral datasets. It provides a standardized directory layout, file naming conventions, and metadata schemas intended to improve reproducibility for projects involving modalities such as Magnetic Resonance Imaging, Electroencephalography, Magnetoencephalography, and behavioral testing used by research groups at institutions like Massachusetts General Hospital, McGill University, and University College London. The initiative has influenced data sharing practices across projects at consortia including the Human Connectome Project and the Alzheimer's Disease Neuroimaging Initiative.

Overview

BIDS defines a formalized structure for dataset files and metadata, enabling interoperable workflows across analysis tools developed by teams from Massachusetts Institute of Technology, Stanford University, Harvard University, and University of Oxford. Using sidecar metadata files such as JSON and TSV, BIDS connects raw imaging data with provenance information used by pipelines like FMRIPREP, FreeSurfer, SPM, and AFNI. The specification emphasizes machine-readability and minimal human intervention to facilitate sharing with repositories such as the OpenNeuro platform and institutional archives at European Organization for Nuclear Research collaborations.

History and Development

The specification originated from working groups composed of researchers affiliated with University of Wisconsin–Madison, McGill University, University of Pennsylvania, and University of Cambridge to address reproducibility concerns following high-profile projects such as the Human Brain Project and analysis challenges in studies by labs at Yale University and Columbia University. Early drafts were discussed at meetings hosted by organizations including the Organization for Human Brain Mapping and the International Neuroinformatics Coordinating Facility. BIDS has evolved through community input, with milestones marked by releases that added support for modalities following requests from teams at Max Planck Institute for Human Cognitive and Brain Sciences and clinical consortia such as ENIGMA.

Specification and Components

The core components include a hierarchical directory layout, standardized filename entities (e.g., sub-, ses-, task-), and metadata files in JSON, TSV, and NIfTI formats used by projects at Johns Hopkins University, King's College London, and University of California, Berkeley. Extensions have been formalized for modalities and data types supported by contributors from University of Oxford, University of Zurich, and University of Toronto; examples include extensions for diffusion MRI requested by groups at McGill University and for EEG advanced by researchers at Dutch National Center for EEG Research. BIDS prescribes validators and example datasets to ensure compliance, allowing automated checks within pipelines like Nipype and integration with workflow managers such as Snakemake and Nextflow.

Adoption and Use Cases

Adoption spans academic laboratories and multi-center consortia; notable adopters include the Human Connectome Project, UK Biobank, and clinical studies conducted at Massachusetts General Hospital and Mayo Clinic. Use cases include data sharing for meta-analyses led by ENIGMA, teaching datasets for courses at Stanford University and University College London, and large-scale preprocessing in cloud environments such as those provided by Amazon Web Services collaborations and Google Cloud Platform grants used by research teams at University of California, Los Angeles. BIDS supports reproducible publication pipelines in journals like Nature Neuroscience and NeuroImage and underpins benchmarking challenges organized by Organization for Human Brain Mapping.

Tools and Software Ecosystem

An ecosystem of tools and converters has grown around the specification, developed by teams at Donders Institute, Montreal Neurological Institute, McGill University, and Massachusetts Institute of Technology. Tools include the BIDS Validator, dataset converters from proprietary formats supported by vendors such as Philips and Siemens Healthineers, and utilities for dataset curation integrated into platforms like DataLad and GitHub workflows. Analysis packages with native BIDS support include FMRIPREP, MRIQC, EEGLAB, and MNE-Python, while containerization and reproducible execution use standards promoted by Docker and Singularity in pipelines orchestrated by groups at University of Oxford and University of Pennsylvania.

Governance and Community

Development is coordinated by a steering group and open community contributors spanning universities, hospitals, and institutes such as McGill University, Massachusetts General Hospital, University of Oxford, and the International Neuroinformatics Coordinating Facility. Governance proceeds through public issue trackers, GitHub pull requests, and community meetings held in association with conferences like Organization for Human Brain Mapping and Society for Neuroscience. Extensions and proposals are reviewed by domain experts from centers including Max Planck Institute, King's College London, and University College London to ensure broad applicability.

Criticisms and Limitations

Critiques include the learning curve for small labs without informatics support at institutions like Community hospitals and challenges converting legacy datasets created by teams at older centers such as some units within National Institutes of Health archives. Some stakeholders note limited coverage for niche modalities used by specialized labs at Lawrence Berkeley National Laboratory and complexities when integrating proprietary formats from manufacturers like GE Healthcare. Interoperability gaps remain for multimodal integration across platforms used by consortia such as ENIGMA and for certain clinical trial workflows regulated by agencies like the U.S. Food and Drug Administration.

Category:Neuroinformatics