Remove patient data from DICOM files (.dcm) - locally in the browser, the image pixels stay unchanged, with no upload.
Running locally on your device ...
Running locally on your device ...
0%
Your files never left your device
No. Everything runs in your browser - your file never leaves your device. How this is verifiable
DICOM is the standard format of medical imaging (DICOM PS3.10). Besides the image itself, every file carries an extensive header of metadata fields - among them the name, birth date, patient ID, address and phone number of the person, plus the clinic, station, device and the names of the attending physicians. These very fields turn a scan into personal data. Anyone sharing an image for research, teaching, a second opinion or a technical bug report must remove these identifiers first.
This tool reads the DICOM file in the browser and applies the DICOM PS3.15 Basic Application Level Confidentiality Profile: it removes every known identity field, replaces each study, series and instance UID with a newly generated random UID, removes all private manufacturer fields, and sets the markers PatientIdentityRemoved to YES and a DeidentificationMethod. The image pixels themselves are taken over byte for byte unchanged - the scan, its resolution and its grey-scale range stay exactly intact, whether the image is uncompressed or compressed. Nothing is recomputed; only the identity is taken out of the header.
The decisive point is privacy: because the anonymization runs entirely on your device, the original file with all patient data never leaves the browser. Nothing is uploaded, nothing is cached, no server sees the scan - the decisive advantage for clinical and research data. You can drop several .dcm files at once; each is processed individually and saved as a new .dcm with the suffix _anon. You keep the original image unchanged, and you can share the anonymous copy safely.
| Input formats | DCM |
|---|---|
| Output format | DCM |
| Batch processing | Yes |
| Processing | Locally in your browser (JavaScript) |
| File upload | None |
Limitations: Removes identity fields from a DICOM header per the DICOM PS3.15 Basic Profile: name, patient ID, every date and time field, address, phone, clinic, station, device, the names of all physicians, plus ethnicity, occupation and comments. Every study, series and instance UID is replaced by a newly generated random UID that stays consistent within one run and cannot be computed back from the original; all private manufacturer fields are removed. PatientIdentityRemoved is set to YES. The pixels stay unchanged byte for byte; the patient's sex, age, size and weight are kept for clinical assessment. It is a structural helper, not a certified anonymization attestation: text burned into the image (burned-in annotations) is not read, and anyone holding both the original and the copy can still match them by their image content. Check sensitive scans yourself before sharing.
No. Anonymization runs entirely locally in the browser (pure JavaScript, no server); the file with the patient data never leaves your device - the decisive advantage for clinical and research data in particular.
Name, patient ID, every date and time field, address, phone number, clinic and station name, device serial number and the names of all physicians, plus ethnicity, occupation and comments. In addition, all study, series and instance UIDs are reassigned, the curve and overlay planes are removed and all private manufacturer fields are removed. The selection follows the DICOM PS3.15 Basic Profile.
Yes, completely. The pixels are taken over byte for byte - only the metadata in the header are cleared. The resolution, grey-scale and compression of the scan do not change.
It is a structural helper based on the DICOM PS3.15 Basic Profile: identity fields are removed, all UIDs are reassigned and all private manufacturer fields are removed by default. What it does NOT capture is text burned into the image (burned-in annotations). Check sensitive scans yourself before sharing.
Yes. Drop several .dcm files at once; each is processed individually and saved as its own anonymous copy with the suffix _anon.