Mammogram (2D)

What it does

Mammogram (2D) reads a full-field digital or synthesized 2D mammogram as a second reader. Upload the four standard views (left and right CC and MLO), any additional diagnostic views and the prior examinations. Set the exam_context option to screening or diagnostic: a screening read returns recall or no recall with BI-RADS 0, 1 or 2, and a diagnostic read returns a BI-RADS 1–6 assessment for each lesion.

Every read carries an exam suspicion score from 0 to 100, a BI-RADS assessment per breast and for the exam, and the density category. Masses, calcification clusters, architectural distortion and asymmetries are boxed on each view, matched between CC and MLO, and located by laterality, clock face, depth and distance from the nipple, each with BI-RADS lexicon descriptors and its own score. Positioning quality, change against priors, image-based risk and arterial calcification complete the result.

Intended use

Decision support for the radiologist reading screening or diagnostic mammograms, including general radiologists without breast fellowship training in high-volume programmes: a second read that marks and scores every lesion, a density category for notification and supplemental-screening decisions, and positioning feedback for technologists.

Who it is for

Breast and general radiologists, screening programmes, mammography technologists, breast surgeons and high-risk clinics. Developers can add a second-reader score, density and positioning checks to a PACS, screening worklist or reporting system.

Accepted input

• DICOM Digital Mammography X-Ray Image, For Presentation (preferred) or For Processing, and synthesized 2D images.
• Standard screening: four views, L-CC, L-MLO, R-CC and R-MLO.
• Diagnostic additions: spot-compression, magnification and rolled or exaggerated views.
• Priors: earlier mammograms in prior_files; strongly recommended.

Requirements

Requirement	Why
All four standard views	Lesions are matched across CC and MLO, and the risk section needs the complete set
For Presentation images for the risk section	The risk estimate is computed on For Presentation FFDM
ImageLaterality, ViewPosition and view code sequences	Views and laterality are verified before reading
BreastImplantPresent	Implant and implant-displaced views are read differently
Pixel spacing	Lesion size and distance from the nipple in mm

Incomplete studies, implants and differences between synthesized 2D and FFDM appearance are flagged in the result.

Optional context

Age, risk factors and the clinical indication (lump, nipple discharge, follow-up) in clinical_context; the exam_context option selects screening or diagnostic logic.

The result

A JSON result with the exam score, recall decision, BI-RADS assessments and the sections below.

Section	Content
Positioning quality	PGMI grade per view, posterior nipple line CC versus MLO, pectoral muscle, nipple in profile, inframammary fold, skin folds, motion and retake advice
Breast density	BI-RADS density a–d per breast and exam, dense-tissue percentage and a dense or not-dense flag
Lesions	Masses, calcification clusters, architectural distortion and asymmetries per view, with malignancy score, CC–MLO match and clock-face location
Calcifications	Microcalcification clusters with masks, BI-RADS morphology and distribution, and a suspicion score
Comparison with priors	New or enlarging masses, new calcifications, developing asymmetry, density change and stability over two years or more
Breast cancer risk	1- to 5-year risk, a risk band and a supplemental-screening suggestion
Breast arterial calcification	Presence and severity grade, with a cardiovascular-risk flag
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

• DICOM: Mammography CAD SR (TID 4000, composite features per TID 4005), Breast Imaging Report SR (TID 4200), TID 1500 for density, risk and arterial calcification scores, TID 4019 Algorithm Identification, SEG for lesion and calcification masks, GSPS overlays; display through the IHE AI Results (AIR) profile.
• Reporting: ACR BI-RADS v2025 Manual (an extension of the 5th edition): assessment categories 0–6, density a–d, the mass, calcification, asymmetry and architectural-distortion lexicon, and management.
• Density notification: the US MQSA categories, dense or not dense.
• Positioning: PGMI grades (UK NHSBSP) and the posterior nipple line rule (CC within 10 mm of MLO).
• Risk: compared with Tyrer–Cuzick (IBIS); ACR criteria for high risk (20 % or more lifetime) and supplemental screening.