HITI Lab OpenData

Advancing medical imaging research through rich, diverse data.

countries
59countries
cities
177cities

Mission

We curate medical imaging datasets with the same rigor researchers bring to their models.

Every collection is de-identified, reviewed and documented before it is published, so the work you do with it starts from data you can account for.

Process

Three steps from request to download.

Each grant is reviewed by the lab, scoped to a single dataset, and issued with an expiry that AWS enforces on every request.

  1. 01

    Request access

    Submit your research purpose and institutional affiliation for review.

  2. 02

    Secure credentials

    Receive time-limited AWS credentials on approval, encrypted and tracked.

  3. 03

    Download data

    Pull curated radiology datasets straight from secure S3 storage.

What you receive

[emory-embed]
aws_access_key_id = AKIA••••••••••••
aws_secret_access_key = ••••••••••••••••
aws s3 sync s3://embed-dataset-open ./embed \
  --profile emory-embed --region us-west-2

Read-only keys, scoped to the dataset you were approved for, valid for a fixed window. Expiry is enforced by AWS itself.

Datasets

Curated collections for machine learning and clinical research.

Full-resolution imaging with linked clinical labels, available to academic researchers under each dataset's licence.

EMory BrEast Imaging Dataset

3.4M
Images

Racially diverse mammography dataset containing 3.4M screening and diagnostic images from 110,000 patients with an equal representation of black and white women

  • Mammography
  • 110,000 patients
  • Racially diverse

Emory Knee Radiograph Dataset

503K+
Images

Knee radiographs with paired reports and structured labels

  • Knee radiographs
  • 83,011 patients
  • Diverse demographics

Contact

Get in touch.

Questions about a dataset, an application, or access you already hold — one address reaches the whole team.

hiti-lab@emory.edu