Anonymization

Public dataset publication

Prepare a fully de-identified dataset for public research release.

Problem

You need to publish a research dataset to a public repository (e.g. TCIA, PhysioNet). The dataset must be fully de-identified to comply with HIPAA Safe Harbor rules — all 18 identifiers must be removed from both DICOM headers and pixel data.

Steps

  1. Open the DICOM file — click Open files… (⌘O).
  2. Switch to Anonymization mode — click the Anon tab (⌘3 in the Mac app, ⌘⇧A in the browser).
  3. Select the HIPAA profile in the profile picker. It removes or replaces the header attributes that carry Safe Harbor identifiers, such as names, IDs, dates and addresses. Check the result yourself: a profile covers header tags, not free text inside them or the pixels.
  4. Check for burned-in text. No profile removes text burned into the pixels. Cover it with redaction regions in the Redaction regions section, where Detect text proposes boxes.
  5. Apply and export — click Apply & Export to produce the de-identified DICOM file.
  6. Verify — load the exported file and confirm no PHI values remain: PatientName is blank, PatientID is anonymized, and all date fields are shifted or cleared.

Result

The exported file has all 18 HIPAA Safe Harbor identifiers removed or pseudonymized, its dates shifted by a consistent offset, and zeroed pixels inside any redaction regions you drew.