Anonymization

Limited Dataset (dates shifted)

Create a limited dataset with dates retained for longitudinal analysis.

Problem

You’re preparing a limited dataset for a research collaborator. Under HIPAA, a limited dataset can retain dates and geographic subdivisions (city, state, ZIP) but must still strip the 16 other direct identifiers. You also want to shift the actual dates by a consistent offset so the temporal relationships between studies are preserved (longitudinal integrity) while the absolute dates are no longer identifiable.

Steps

Method A: Manual Date Shift

  1. Open your file (⌘O) and switch to Anon mode (⌘2).

  2. Enter a date shift offset — find the date shift input (labeled “Date shift in days”). Type a number like 10 to shift every date forward by 10 days.

  3. Apply — click the Apply button.

  4. Verify — switch to Edit mode (⌘3) and filter for Study Date. The value should be exactly 10 days later than the original (e.g. original 2005113020051210).

Method B: Deterministic Auto-Shift (Longitudinal Consistency)

  1. Open your file and switch to Anon mode (⌘2).

  2. Click the Auto button — this derives a deterministic offset from the Patient ID using a hash. The same Patient ID always produces the same offset, ensuring consistent date shifts across all studies belonging to that patient.

  3. Apply the anonymization.

  4. Reload the same file (or another study from the same patient ID), run Auto again, and verify the offset is identical — proving longitudinal consistency.

Expected Result

  • Study Date (0008,0020) is shifted by exactly the offset you specified (manual) or the deterministically-derived offset (auto).
  • Date format (YYYYMMDD) is preserved — only the value changes.
  • All other date fields in the DICOM headers are shifted by the same offset.
  • With Auto shift, the same Patient ID always maps to the same offset — load the file twice, click Auto twice, and you get the same number.
  • If Patient ID is missing, Auto shift gracefully derives a fallback offset (the app does not crash).
  • If Study Date is missing entirely (deleted tag), Apply succeeds without error — the tag remains absent rather than corrupting other fields.
  • Invalid date formats (non-YYYYMMDD strings) are rejected with a red validation error in the Edit modal before they can be saved.