Data Access

How to Access the LAION-fMRI Dataset

The LAION-fMRI dataset can be accessed in two ways:

1. Direct AWS Access

  • Direct download from AWS S3 buckets

  • Suitable for downloading entire dataset or specific subjects

  • Provides full control over data storage and management

  • Requires AWS CLI or similar tools

2. Python Package (laion_fmri_dataloader)

  • Streamlined Python interface for data access

  • Built-in data validation and integrity checks

  • Automatic handling of file paths and metadata

  • Convenient for programmatic access and integration into analysis pipelines

Access Requirements

Some parts of the dataset require a Data Use Agreement (DUA).

By accessing the LAION-fMRI dataset, you agree to:

  • Use the data for research purposes only

  • Properly cite the dataset in publications

  • Not redistribute the raw data without permission

  • Follow ethical guidelines for human subjects data

Registration

For restricted data requiring a DUA:

  1. Visit the dataset homepage

  2. Complete the registration form

  3. Receive access credentials

  4. Configure your AWS credentials or Python package with appropriate permissions

Download Instructions

Choose your preferred method for accessing the dataset:

Prerequisites:

  • AWS CLI installed: pip install awscli

  • AWS credentials (for restricted data only)

Download entire dataset:

aws s3 sync s3://laion-fmri/ ./laion-fmri-data/ --no-sign-request

Download specific subject:

aws s3 sync s3://laion-fmri/sub-01/ ./laion-fmri-data/sub-01/ --no-sign-request

Download derivatives only:

aws s3 sync s3://laion-fmri/derivatives/ ./laion-fmri-data/derivatives/ --no-sign-request

Download stimuli:

aws s3 sync s3://laion-fmri/stimuli/ ./laion-fmri-data/stimuli/ --no-sign-request

Note

For restricted data requiring DUA, remove --no-sign-request and configure AWS credentials with appropriate access permissions.

Installation:

pip install laion-fmri-dataloader

Basic Usage:

from laion_fmri_dataloader import LAIONfMRIDataset

# Initialize dataset
dataset = LAIONfMRIDataset(data_dir='./laion-fmri-data')

# Download specific subject
dataset.download_subject('sub-01')

# Load functional data
func_data = dataset.load_functional('sub-01', task='experiment', run=1)

# Load stimulus metadata
stimuli = dataset.load_stimuli()

print(f"Functional data shape: {func_data.shape}")
print(f"Number of stimuli: {len(stimuli)}")

With authentication (restricted data):

# Configure credentials
dataset = LAIONfMRIDataset(
    data_dir='./laion-fmri-data',
    aws_access_key='YOUR_ACCESS_KEY',
    aws_secret_key='YOUR_SECRET_KEY'
)

# Download restricted components
dataset.download_subject('sub-01', include_restricted=True)

Dataset Components

The dataset is organized into several components:

  • Raw data: Unprocessed fMRI and anatomical scans

  • Derivatives: Preprocessed data and analysis results

  • Stimuli: Visual stimuli used in experiments

  • Behavioral: Behavioral response data

  • Metadata: Experimental parameters and subject information

File Organization

After downloading, the dataset structure will be:

laion-fmri/
├── dataset_description.json
├── README
├── CHANGES
├── participants.tsv
├── participants.json
├── stimuli/
├── sub-01/
├── sub-02/
├── ...
└── derivatives/

Data Size

Storage Recommendations

Data Formats

File Formats Used

Neuroimaging Data:

  • NIfTI format (.nii.gz): Functional and anatomical MRI data

  • JSON sidecar files: Metadata and acquisition parameters

Behavioral Data:

  • TSV format (.tsv): Event timing and responses

  • JSON format (.json): Metadata and experimental parameters

Stimulus Data:

  • PNG/JPEG: Image files

  • TSV/JSON: Stimulus metadata

Software Requirements

Data Verification

Support

For data access issues or questions:

Citation

If you use this dataset, please cite:

LAION-fMRI Dataset (2026). ViCCo-Group.
Available at: https://github.com/ViCCo-Group/LAION-fMRI

BibTeX entry:

@dataset{laion_fmri_2026,
  title={LAION-fMRI: Open fMRI Dataset},
  author={ViCCo-Group},
  year={2026},
  publisher={ViCCo-Group},
  url={https://github.com/ViCCo-Group/LAION-fMRI}
}