Brain Imaging Data Structure

Brain Imaging Data Structure
Name	Brain Imaging Data Structure
Abbreviation	BIDS
Domain	Neuroimaging
First published	2016
Latest release	2024
Website	Brain Imaging Data Structure

Brain Imaging Data Structure Brain Imaging Data Structure is an open specification for organizing and describing neuroimaging and related data. It promotes reproducibility by prescribing file naming, directory layout, and metadata conventions used by researchers working with modalities such as magnetic resonance imaging, electroencephalography, and positron emission tomography. The specification is maintained by an international consortium and is implemented across many software tools, repositories, and laboratories.

Overview

The specification standardizes dataset layout to enable automated processing pipelines developed by groups at institutions such as Massachusetts Institute of Technology, Harvard University, Stanford University, University of Oxford, and Imperial College London. Its design was motivated by challenges identified in projects including the Human Connectome Project, Alzheimer's Disease Neuroimaging Initiative, UK Biobank imaging study, Adolescent Brain Cognitive Development study, and OpenfMRI. Adoption is encouraged by data platforms like OpenNeuro, Dryad, Zenodo, XNAT, and repositories associated with National Institutes of Health. Community governance draws contributors from consortia such as International Neuroinformatics Coordinating Facility, European Research Council grantees, and research centers like Max Planck Society and McGill University.

Specification and File Organization

The layout prescribes a hierarchical directory tree with subjects and sessions, influenced by directory conventions used at Wellcome Trust Centre for Neuroimaging, Johns Hopkins University, University of California, San Diego, University College London, and Karolinska Institutet. Filenames encode entities such as subject, session, task, and acquisition modeled after practices at labs led by investigators associated with Sanes Lab, Smith Lab (FMRIB), Van Essen Lab, Poldrack Lab, and Gorgolewski Lab. The specification details required files and recommended optional files; these choices reflect standards used by projects funded by National Science Foundation, European Commission, and national biobanks like FinnGen and Estonian Biobank.

Data Types and Modalities

The standard covers multiple modalities including structural and functional magnetic resonance imaging adopted in studies from Human Connectome Project, diffusion MRI applied in works from Connectome Coordination Facility, arterial spin labeling used in projects at University of Pennsylvania, electroencephalography and magnetoencephalography common to groups at University of Cambridge, Massachusetts General Hospital, and McLean Hospital, as well as positron emission tomography used by teams at Karolinska University Hospital and Mayo Clinic. Extensions accommodate behavioral logs, physiological recordings, and model outputs used by collaborations such as OpenNeuro children', COINS and domain-driven consortia like ENIGMA and ADNI. The format interoperates with image formats and standards from organizations like Digital Imaging and Communications in Medicine and initiatives such as Neuroimaging Informatics Technology Initiative.

Metadata and Sidecar Files

Metadata is stored in human- and machine-readable sidecar files influenced by metadata practices at National Library of Medicine, Library of Congress, European Bioinformatics Institute, Wellcome Sanger Institute, and Broad Institute. JSON sidecars capture acquisition parameters, provenance, and task descriptions mirroring fields used in datasets from OpenfMRI, MyConnectome Project, Nathan Kline Institute, Kaiser Permanente-linked studies, and clinical trials registered at ClinicalTrials.gov. Event files, stimulus descriptions, and physiological logs align with conventions from experiments at MIT Media Lab, Max Planck Institute for Human Cognitive and Brain Sciences, and labs affiliated with NIH intramural programs.

Tools, Validation, and Software Ecosystem

A validator toolchain developed by contributors from Montreal Neurological Institute, Stanford Center for Reproducible Neuroscience, University of Washington, and University of Melbourne ensures compliance; popular converters and utilities are distributed by teams at NeuroDebian, Anaconda, Python Software Foundation, and package authors connected to GitHub. Processing pipelines such as those from FMRIB Software Library, AFNI, SPM (Software), FreeSurfer, nipype, fMRIPrep, and MRtrix integrate BIDS-aware input handling. Visualization and dataset browsing tools from projects like MRIcroGL, Brainstorm, MNE-Python, and nilearn provide user interfaces for BIDS datasets.

Adoption, Use Cases, and Community

BIDS is used in large-scale studies including Human Connectome Project, ADNI, UK Biobank, ABIDE, and multi-site consortia like ENIGMA and IMAGEN. Clinical research groups at Mount Sinai Health System, Cleveland Clinic, Johns Hopkins Hospital, Massachusetts General Hospital, and academic centers in networks such as European Organization for Research and Treatment of Cancer employ BIDS for harmonization. Educational workshops and hackathons hosted by OHBM, SfN, ISMRM, and regional schools at Cold Spring Harbor Laboratory and Banff International Research Station foster developer and user communities.

Governance and Development History

The specification originated from efforts by researchers affiliated with University of California, Berkeley, University of Texas Health Science Center at Houston, University of Pennsylvania, and McGill University and was publicly released after community review processes involving stakeholders from NIH Office of Data Science Strategy, Wellcome Trust, European Commission Horizon 2020, and philanthropic funders like Gordon and Betty Moore Foundation and Chan Zuckerberg Initiative. Development occurs openly on platforms maintained by organizations such as GitHub, with steering groups including representatives from International Neuroinformatics Coordinating Facility, academic centers, and industry partners like Siemens Healthineers, GE Healthcare, and Philips Healthcare.

Category:Neuroimaging standards