Skip to main content
Version: V12

Understanding DICOM Processing in VIDIZMO

DICOM (Digital Imaging and Communications in Medicine) is the standard format for medical images such as X-rays, CT and MRI scans, and ultrasounds. A single DICOM file carries protected health information (PHI) in two places: the header tags that describe the study, and text burned directly into the image. VIDIZMO detects and redacts both, so you can share medical imaging without exposing patient identity.

Overview​

VIDIZMO treats a DICOM file as a document. You upload it to the Library, run AI detection and redaction on it through the VIDIZMO Indexer, and download a redacted copy — the same automatic and on-demand processing used for PDFs and Word files. What's different is where the sensitive data hides and how it's removed.

A DICOM file stores patient information in its header as named tags, for example Patient Name, Patient ID, and Patient Birth Date. Many scans also have text printed onto the image itself, such as a patient label in the corner of an X-ray. VIDIZMO reads both layers, finds the PHI, and removes it in a single pass.

Note: DICOM redaction runs through the VIDIZMO Indexer's automatic or on-demand processing. There is no Studio Space editor step for DICOM files. Redaction also requires the Redaction feature — contact VIDIZMO Support if you need it enabled in your package.

How DICOM Processing Works​

Detection​

When a DICOM file is processed, the VIDIZMO Indexer reads two sources:

  • Header tags are scanned for PII such as patient names, identifiers, dates, and physician names.
  • Image frames are run through optical character recognition (OCR) to find burned-in text.

A DICOM file can hold a single frame, such as one X-ray, or many frames, such as a CT series or a multi-frame ultrasound. Each frame is treated as its own page, so detection covers every image in the file.

Redaction​

VIDIZMO redacts both layers and saves one redacted file:

  • Pixel redaction masks the regions of each frame that contain burned-in PHI with a solid color box.
  • Metadata redaction anonymizes or removes the PHI tags in the header. Patient names and identifiers are replaced with a mask, and dates such as birth date are replaced with a placeholder. All private, vendor-specific tags are removed in line with the DICOM Basic Application Confidentiality Profile.

Structural and measurement data the file needs to display correctly — such as ultrasound calibration values — is kept. The result is a redacted .dcm file that still opens in standard DICOM viewers.

Key Characteristics​

AspectDetail
Supported file types.dcm, .dicom, .dic, .ima
Content typeTreated as a document
How redaction runsVIDIZMO Indexer (automatic or on-demand), not Studio Space
What's detectedHeader PII and burned-in text (OCR)
What's redactedImage pixel regions and DICOM header tags
Preview and playbackNot available — download only
OutputA redacted DICOM file that remains valid

What Metadata Is Redacted​

VIDIZMO redacts a standard set of header tags that identify a patient or provider, including:

  • Patient name, patient ID, and address
  • Patient birth date and other identifying dates
  • Referring and performing physician names
  • All private, vendor-specific tags (every private-group tag is stripped)

Turning on Enable AI Metadata Redaction in the VIDIZMO Indexer adds a second layer: VIDIZMO scans the header tags with AI to catch PII that isn't part of the standard identifying set, and redacts that too.

Use Cases​

Healthcare and Research​

De-identify scans before using them in teaching files, case studies, or research datasets. Removing both the header tags and burned-in labels keeps studies usable while protecting patient identity under HIPAA.

When a medical image is submitted as evidence, redact PHI before sharing it with external parties or releasing it under a Freedom of Information Act (FOIA) request, while keeping the diagnostic content intact.

Key Considerations​

  • DICOM files have no in-browser preview or playback. Download the file and open it in a DICOM viewer to review the result.
  • Burned-in text detection through OCR is supported in English.
  • Large multi-frame studies take longer to process, since every frame is scanned and redacted.
  • AI scanning of header tags for additional PII depends on Enable AI Metadata Redaction in the VIDIZMO Indexer.
  • All private, vendor-specific tags are removed during metadata redaction. Any viewer feature that relies on vendor-private data will no longer work after redaction.
  • When burned-in text is pixel-redacted, the file is saved with uncompressed pixel data. A compressed study (JPEG or JPEG 2000) can be several times larger after redaction. Metadata-only redaction keeps the original encoding and file size.
  • Metadata redaction replaces the study, series, and SOP instance UIDs with the same placeholder value. Each redacted file opens on its own, but the files no longer group as a study or series in a viewer, and importing several into a PACS can collide on duplicate UIDs. Share redacted files as individual images, not as a re-importable study.
Did you find this article helpful?