Skip to content
mg/cem

Contrast-Enhanced Mammography

CEM read: enhancing masses and non-mass enhancement, BPE level and BI-RADS by the CEM lexicon.

Coming soonMGMammography & Tomosynthesis

This model is not available to run yet. Vote to show interest and get an email when it opens.

Overview

What it does

Contrast-Enhanced Mammography reads a CEM study (also called contrast-enhanced spectral mammography, CESM). After an iodinated contrast injection, each view is acquired as a low-energy image, which looks like a standard mammogram, and a recombined image that shows iodine uptake. The run reads both images of every view together.

On the recombined images it finds enhancing masses and non-mass enhancement, outlines them, grades how strongly they enhance and scores each for malignancy, using the matching low-energy image to place them in the breast. It grades background parenchymal enhancement (BPE) for each breast with its symmetry, because marked BPE can hide a lesion. A BI-RADS assessment closes the read, using the CEM lexicon of BI-RADS v2025.

Intended use

Decision support for radiologists using CEM to work up a recalled or symptomatic patient, to stage a known cancer, or as an alternative where breast MRI is not available. Each enhancing lesion comes with its mask, enhancement level and malignancy score, and the BI-RADS assessment carries the management that follows.

Who it is for

Breast radiologists, and general radiologists in centres that are adopting CEM without MRI access; breast surgeons and oncologists planning surgery. Developers can add structured CEM reads to a breast imaging PACS or reporting system.

Inputs and protocol

Accepted input

  • DICOM Digital Mammography X-Ray Image objects from a CEM acquisition: low-energy and recombined images.
  • Image types are recognised from ImageType and series descriptions, so the pairs do not need to be labelled.
  • CC and MLO views of both breasts, with any additional diagnostic views.

Requirements

Requirement Why
Paired low-energy and recombined images for each view Enhancement is read on the recombined image and located on the low-energy image
CC and MLO of both breasts Lesions are matched across views, and BPE is graded and compared per breast
ImageLaterality and view codes Each pair is assigned to the right breast and projection
Pixel spacing Lesion size in mm

Vendor differences in recombination and recombination artefacts (rim and ripple) are recognised and flagged where they affect the read.

Optional context

Indication (recall, symptoms, staging of a known cancer), age and prior imaging in clinical_context and prior_files.

Outputs and standards

The result

A JSON result with a BI-RADS assessment and the two CEM sections, with masks and boxes on the recombined images.

Section Content
Enhancing lesions Enhancing masses and non-mass enhancement with boxes, masks, enhancement level and malignancy score
Background parenchymal enhancement BPE level (minimal, mild, moderate or marked) per breast and its symmetry
Draft report Opt-in (options.report): an English narrative that interprets the result for the reader (findings, impression, limitations); every number is checked against the findings

Standards

  • Reporting: ACR BI-RADS v2025 Manual, which adds the CEM chapter: CEM lexicon for enhancing findings, BPE categories and assessment categories 0–6.
  • DICOM: Mammography CAD SR (TID 4000) for lesions, SEG for lesion masks and GSPS overlays on the recombined images; display through the IHE AI Results (AIR) profile.
  • Terminology: RadLex BI-RADS terms and SNOMED CT.

Result sections

One run returns every section its input supports.

  1. Enhancing lesionslesions
  2. Background parenchymal enhancementbpe
  3. Draft reportreport
Explore all Mammography & Tomosynthesis models

Mammogram (2D)

Mammogram read: suspicion score, recall, BI-RADS, density, located lesions and risk.

Breast8 sections
MG· image-2d-multiview

Breast Tomosynthesis

DBT read: lesions located in 3D with their best slice, exam score, recall and BI-RADS.

Breast2 sections
MG· image-2d-multiview