Anonymization

Expert Determination workflow

Expert-guided de-identification with per-field risk assessment.

Problem

The pre-built anonymization profiles (Basic, HIPAA, +Pixel) are opinionated one-size-fits-all approaches. As a domain expert — a radiologist, clinical data manager, or privacy officer — you need to make per-field decisions about which tags to scrub, which to keep, and which to pseudonymize. The Expert Determination method under HIPAA requires you to document that the risk of re-identification is “very small” based on your specific dataset and use case. You need a custom profile builder where you can add, remove, and adjust rules tag-by-tag.

Steps

Create a Custom Profile

Open your DICOM file (⌘O) and switch to Anon mode (⌘2).
Select the Custom profile — click the Custom button in the profile picker. You’ll see a placeholder: “No custom rules configured yet.”
Add your first rule — in the custom rule form:
- Enter a DICOM tag in the tag input, e.g. (0008,103e) for Series Description. You can use hex format (0008,103E) or dotted 0008103E.
- Select an action from the dropdown: keep or remove.
- Click Add.
Add a second rule — enter 00100010 (Patient Name) and set action to keep. Click Add.
Verify the rules appear — each rule is displayed as a row in the custom rules list showing the tag ID and the action (KEEP or REMOVE).

Save and Load Profiles

Export your profile — click 💾 Save Profile. A custom_anonymization_profile.json file downloads containing a JSON array of {id, act} objects.
Remove a rule — click the ✕ button on the (0008,103E) row. The rule disappears from the list, but the (0010,0010) rule remains.
Re-import the saved profile — click 📂 Load Profile, select the JSON file you just downloaded. Both rules (including the one you removed) are restored from the file.

Result

Your custom keep/remove rules show as individual rows, save to and load from a JSON profile, and override the default behavior when applied — the documented, per-field control the HIPAA Expert Determination route calls for.