BIDS — Brain Imaging Data Structure

Overview

BIDS (Brain Imaging Data Structure) is the community standard for organising and describing neuroimaging and electrophysiology datasets in a consistent, machine-readable directory and file naming structure. Published in 2016 in Scientific Data (Gorgolewski et al., doi:10.1038/sdata.2016.44), BIDS emerged from an INCF-sponsored data sharing working group at Stanford and addressed the lab-specific, idiosyncratic dataset organisation that made sharing and reuse of neuroimaging data difficult. The project is now governed by the BIDS Steering Group (since October 2019) with INCF as an endorsing and supporting body; it is widely required or recommended by neuroscience data repositories and is a widely adopted format for open neuroimaging data.

What BIDS Defines

BIDS specifies a directory structure that organises files into sub-, ses-, task-, and run- folders, with deterministic key-value file naming (e.g. sub-01_ses-01_task-rest_bold.nii.gz). Each data file is accompanied by a sidecar JSON metadata file with standardised fields describing acquisition parameters. Dataset-level description files (dataset_description.json, participants.tsv, README, CHANGES) sit at the root level. Processed outputs are stored under derivatives/ with provenance metadata using the BIDS Derivatives convention.

Covered Modalities

The core BIDS specification covers MRI (structural T1w/T2w/FLAIR, functional BOLD, diffusion DWI, perfusion ASL, and quantitative MRI), PET brain imaging with blood input function metadata, and intracranial electrophysiology including EEG, MEG, and iEEG (stereo-EEG and ECoG). The following modalities have been added via accepted BEPs: Microscopy (BEP031, merged in BIDS 1.7.0), Genetics (BEP018), and Motion capture (BEP029). MRS (magnetic resonance spectroscopy, BEP038) is under active development and not yet merged into the specification as of 2025.

BEP Extension Process

New modalities and data types are added via BIDS Extension Proposals (BEPs), a community-driven RFC-style process where anyone can propose, discuss, and implement extensions. BEPs are tracked on GitHub with open discussion. Accepted BEPs are merged into the official specification, and the BIDS Steering Group governs the process and ratifies proposals.

BIDS Apps

BIDS Apps is an ecosystem of containerised (Docker/Singularity) data analysis pipelines that accept BIDS-organised datasets as input and produce BIDS derivatives. Notable examples include fMRIPrep (fMRI preprocessing), dMRIPrep (diffusion MRI preprocessing), MRIQC (automated MRI quality metrics), and the BIDS Validator (specification compliance checking).

BIDS Raw vs Derivatives

BIDS distinguishes two dataset layers. Raw BIDS is unprocessed data as acquired in the standard structure. BIDS Derivatives are processed outputs stored under derivatives/ with provenance metadata recording which pipeline, version, and parameters were used.

Connections

• Governed by: BIDS Steering Group
• Endorsed by: INCF
• Requires: NIfTI (mandatory MRI/PET image format within BIDS)
• Compatible with: HED (event annotation), EDF (accepted EEG/iEEG format), BrainVision (accepted EEG format)

Resources

• https://bids.neuroimaging.io
• https://bids-specification.readthedocs.io (full specification)
• https://github.com/bids-standard (GitHub organisation)
• https://bids-apps.neuroimaging.io (BIDS Apps ecosystem)
• https://bids.neuroimaging.io/get_started.html (getting started guide)