Back to Glossary

What is DICOM annotation?

DICOM (Digital Imaging and Communications in Medicine) is the international standard for medical images and related information. DICOM annotation means adding labels and measurements to medical images so people and AI models can see what matters. For example, outlining a tumor, marking a fracture line, or recording the diameter of a lesion. It turns raw scans into structured data that can be searched, compared, and used to train AI models.

DICOM is used across CT, MRI, X-ray, ultrasound, and more; it has underpinned digital radiology since 1993 and is recognized as ISO 12052, enabling interoperability across thousands of devices and systems.

Types of DICOM annotations

  1. Text-based annotations: Short labels, notes, or comments added to medical images to provide context. Commonly used to name anatomical structures (e.g., left kidney, lung apex) or flag abnormalities (possible tumor, calcified artery).
  2. Region-of-interest (ROI) annotations: Highlight specific areas such as tumors, organs, or fractures using bounding boxes, polygons, or contours to define precise boundaries—useful for localization and segmentation.
  3. Metadata annotations: Link contextual information (patient ID, imaging modality, acquisition time, timestamp) to images. Helps create complete clinical records and supports compliance with regulations like HIPAA/GDPR, as well as research and AI performance analysis.
  4. Measurement annotations: Capture numeric values directly from images—length, area, volume, or angles—to track disease progression, evaluate treatment response, and feed quantitative data to AI systems.
  5. Multi-modality annotations: Combine findings across imaging types (e.g., CT + PET, MRI + CT) to give a fuller clinical picture—used in oncology, neurology, and cardiology for tumor staging, brain mapping, and cardiac assessments.

Example

Scenario: A radiologist reviews a chest CT and related PET scan for a suspected lung nodule.

  1. Adds a text label: “Suspicious nodule—right upper lobe.”
  2. Draws a polygon ROI around the nodule to outline its exact shape.
  3. Records measurements (longest diameter and volume) to track over time.
  4. Confirms metadata (patient ID, study date, series) so the finding ties to the correct exam.
  5. Views the same ROI in multi-modality (CT fused with PET) to assess metabolic activity.

Why it helps: The care team gets clear, structured information for follow-up and treatment planning, and the same annotations can train and evaluate AI models with consistent ground truth.

‍