Frequently Asked Questions

A researcher posts a request describing the data they need; verified hospitals and labs then upload matching de-identified data and expert labels, and every batch is quality-scored before it is accepted.

Post a data request specifying modality, anatomy, label schema, class balance and quantity. Verified providers fulfill it with de-identified, expert-labeled datasets tailored to your model — instead of scraping limited public sets.

GetDATA is a marketplace where you source de-identified imaging datasets — X-ray, CT, MRI, ultrasound, ECG and pathology — directly from verified healthcare providers, with provenance and compliance built in.

Use the guided request form to define the modality (for example chest X-ray or 12-lead ECG), label schema, cohort criteria, quantity and budget. You can also browse category pages such as the ECG or chest X-ray catalogs to see live requests.

X-ray and chest radiography, CT, MRI, ultrasound and echocardiography, ECG, and pathology — across anatomical regions including chest, brain, abdomen and heart.

Standard medical formats including DICOM, NIfTI, PNG/JPG, EDF, CSV and JSON, depending on the modality and what you specify in your request.

Classification, bounding boxes, segmentation masks, measurement tables, and custom JSON schemas, delivered alongside the source data.

Verified radiologists, clinicians, and trained annotators. Low-quality labels are flagged and rejected by reviewer consensus and automated checks.

Verified hospitals, imaging centers and laboratories. Each provider is reviewed before they can fulfill requests, and uploads undergo provenance checks.

Providers remove protected health information (PHI) from DICOM headers and pixel data before upload, automated PHI scanning flags residual identifiers, and brain scans can be defaced or skull-stripped where required.

Yes. De-identification is required before upload, and the platform supports HIPAA/GDPR, Data Processing Agreements (DPAs) and Business Associate Agreements (BAAs), with audit trails on every transfer.

Uploads are transferred securely with provenance checks, automated PHI scanning, staged deliveries and rolling approvals; access is controlled and fully audited.

Both data and labels are quality-scored with reviewer consensus, consistency checks and automated QC for drift, and acceptance gates enforce a minimum quality before any payout.

You set a budget per request and choose pay-per-sample or a fixed dataset price. Final cost depends on modality, labeling depth and quantity — see the pricing page for plan tiers.

Payouts split proportionally by approved volume and quality score across both data providers and annotators.

Typical fulfillment completes in weeks, depending on modality, cohort criteria, labeling depth and quantity.

Yes. Specify inclusion and exclusion criteria, demographics and class balance in your request; the multi-site provider network helps assemble rarer cohorts that a single institution cannot supply alone.

Source from multiple verified institutions and scanners. Models trained on a single site often fail to generalize, so multi-site sourcing reduces domain shift.