Privacy tool

DICOM Anonymizer (De-identify .dcm)

Remove patient data (PHI) from DICOM files. The image stays identical; the output is a valid .dcm with all identifiers, dates, physician and institution details and private tags stripped by default.

DICOM DICOM
PrivConvert
Privacy-first conversion
No files stored
In-memory processing
Deleted instantly
No sign-up needed

Complete the security check to begin

Every file is de-identified by default. No configuration is required.

The output is a still-valid DICOM file with the image untouched and the identifying header data removed — patient name, ID, birth date, study and acquisition dates, physicians, institution, device serial and free-text comments, plus every private tag. The one exception is text a scanner may have burned directly into the picture, which is part of the image itself and cannot be removed automatically.

De-identify a DICOM scan — keep the image, drop the patient data

Unlike converting to an image, this keeps the file as a valid .dcm so it still opens in any DICOM viewer — it simply no longer carries the patient’s identity. It follows the DICOM PS3.15 Basic Confidentiality Profile and marks the file as de-identified. Everything is processed in memory and deleted the instant your download is ready.

About the DICOM Anonymizer

A DICOM (.dcm) file is not just a medical image — its header stores a large block of Protected Health Information: patient name, ID, birth date, sex, the study and acquisition dates, the referring and performing physicians, the institution, the scanner’s serial number and often free-text comments. Sharing a raw .dcm for teaching, research, a second opinion or a bug report therefore leaks the patient’s identity along with the picture.

This tool returns a new, still-valid DICOM file with the image completely unchanged but the identifying header data removed. It follows the DICOM PS3.15 Basic Confidentiality Profile: direct identifiers are blanked, optional identifying elements are deleted, every date, time and person-name field is cleared, all private (vendor-specific) tags — a notorious hiding place for PHI — are stripped wholesale, and the study/series/instance UIDs are replaced with fresh random values so the file cannot be matched back to the originating system. The raw file preamble is zeroed and the file is flagged with PatientIdentityRemoved = YES.

Common use cases

  • Sharing a scan for a teaching file, conference talk or textbook without exposing the patient
  • Sending a case to a colleague for a second opinion
  • Preparing imaging data for a research dataset or an AI/ML training set
  • Attaching a .dcm to a support ticket or vendor bug report
  • Publishing an interesting case to a medical forum

Tips for best results

  • The image pixels are never altered — de-identification only touches the metadata, so diagnostic quality is preserved exactly.
  • The one thing no tool can remove automatically is text a scanner has "burned into" the image pixels themselves (some ultrasound and secondary-capture images overlay the name on the picture, and a few legacy files hide overlay bits inside the pixel data). If your scan shows a name on the picture, it will remain — crop it out before sharing.
  • Because private tags are removed entirely, some vendor-specific processing hints may be lost; the standard clinical image and its display settings are kept.
  • Output stays a fully valid .dcm, so it opens in any DICOM viewer just like the original.

Why Choose PrivConvert?

Unlike other online converters that upload your files to remote servers, store them for days, and may use them for AI training or sell your data, PrivConvert was built with privacy as the foundation. Here's what makes us different:

Zero data retention

Files are processed in volatile memory (RAM) and immediately purged. Nothing is ever written to disk or backed up.

Encrypted transfers

All uploads and downloads are encrypted with TLS 1.3. Your files cannot be intercepted in transit.

No account required

Use any tool instantly without signing up, providing an email, or creating a profile. No tracking cookies, no user profiling.

Works on any device

Compatible with Windows, macOS, Linux, iOS, and Android. No plugins or desktop software needed - just a modern web browser.

Trusted by Thousands of Users

PrivConvert processes thousands of file conversions every day. Our commitment to privacy and security has made us the go-to choice for professionals, developers, and privacy-conscious users worldwide.

250 MB Max file size
200+ Conversion tools
0 sec File retention
100% Free to use

Frequently Asked Questions

What exactly is removed?
Patient name, ID, birth date, sex, address and contact details; accession and study IDs; all dates and times; referring/performing/operator physician names; institution name and address; station name and device serial number; free-text comments and descriptions; every person-name field anywhere in the file (including inside sequences); and all private vendor tags. The study, series and instance UIDs are replaced with fresh random values, the raw file preamble is zeroed, and identifying software/station fields in the file header are stripped. Required identifier elements are kept but emptied so the file stays standards-valid.
Is the image itself changed?
No. Only the header metadata is modified. The pixel data is copied through untouched, so the de-identified file is diagnostically identical to the original.
Does the output stay a valid DICOM file?
Yes. You get a proper DICOM Part-10 file that opens in any viewer. It is additionally tagged with PatientIdentityRemoved = YES and a DeidentificationMethod note per the DICOM standard.
Can hospitals or clinics use this?
The converter runs entirely in memory and removes DICOM metadata by default, but organizations should always confirm that using any online service complies with their internal policies and applicable privacy regulations before processing real patient data.
Is my file stored?
No. The conversion runs in an isolated, in-memory process. Your file never touches the disk, is never stored or shared, and is wiped from memory the moment your download is ready.