DICOM anonymizer
Remove patient names, IDs, dates and other identifying details from DICOM files before you share them for teaching, research or a second opinion. Everything happens in your browser, so the original files never leave your computer.
Drop DICOM files here
or
Files from one study keep their series together.
Want to test first? with made-up patient details.
What to change
What gets removed
| Information | Example tags | What happens |
|---|---|---|
| Patient name and ID | (0010,0010) (0010,0020) | Replaced with the values you choose |
| Birth date, address, phone, other IDs and names | (0010,0030) (0010,1040) (0010,1000) | Emptied or removed |
| Accession number, admission and order numbers | (0008,0050) (0038,0010) (0040,2016) | Emptied |
| Hospital, department, station and device serial | (0008,0080) (0008,1010) (0018,1000) | Emptied |
| Referring, performing and reading physicians, operators | (0008,0090) (0008,1050) (0008,1070) | Emptied |
| Study, series and instance UIDs | (0020,000D) (0020,000E) (0008,0018) | Replaced with new UIDs, consistently across all files in the batch |
| Dates and times | (0008,0020) (0008,0030) | Shifted, removed or kept, as you choose |
| Private tags and overlays | odd groups, (60xx,xxxx) | Removed (optional) |
Burned-in text is not removed. Ultrasound images, screen captures and some fluoroscopy and dose screens have the patient's name drawn into the pixels. The anonymizer warns you about these files. Check them in the DICOM viewer and crop or exclude them before sharing.
The tool follows the spirit of the DICOM Basic Application Level Confidentiality Profile (PS3.15 Annex E) but is not a certified de-identification system. For formal research data sharing, follow your institution's approved process. Read more in how to anonymize DICOM files.
DICOM anonymizer: common questions
Are my files uploaded to a server?
No. The files are read, cleaned and saved by JavaScript in your browser. Neither the originals nor the cleaned copies are sent anywhere.
Will the anonymized files still open in a PACS or viewer?
Yes. The image data is copied exactly and the header stays valid DICOM. New UIDs are generated in a consistent way, so slices from one series stay together and series from one study stay in the same study.
Is this enough for HIPAA or GDPR?
It removes the header fields that most often identify a patient, but no tool can guarantee compliance on its own. Burned-in text, unusual private tags and the image itself (a face in a head CT 3D render, for example) can still identify someone. Check the output and follow your organization's rules.
How do I check what was removed?
Open a cleaned file in the DICOM tag viewer and tick "Only identifying fields" to see what is left.