Breast MRI

What it does

Breast MRI reads dynamic contrast-enhanced studies for high-risk screening, preoperative staging, problem solving, monitoring of neoadjuvant chemotherapy and implant assessment. The run orders the dynamic phases by acquisition time, motion-corrects them, computes subtractions and kinetic curves, and reads the breasts the way the BI-RADS MRI lexicon is structured: composition first, then enhancement, then findings.

It grades the amount of fibroglandular tissue (a–d) and the background parenchymal enhancement with its symmetry, detects and segments mass and non-mass enhancement with size and kinetic curve type, and assigns BI-RADS descriptors and a final assessment category 0–6. On a baseline and on-treatment pair, it reports functional tumour volume change and a pCR likelihood; on silicone-sensitive sequences it looks for intracapsular and extracapsular rupture.

Intended use

Decision support for physicians reporting breast MRI: lesion detection across hundreds of dynamic images, kinetic curves that would otherwise be drawn by hand, and a BI-RADS assessment set in the context of FGT and BPE. Background enhancement varies with the menstrual cycle and therapy, which is why both can be passed as context.

Who it is for

Breast imagers and general radiologists covering supplemental screening of dense breasts, breast surgeons planning resection, medical oncologists following neoadjuvant therapy, and plastic surgeons assessing implants. Developers call the run to bring BI-RADS-structured MRI results into screening registries and breast-centre software.

Accepted input

• DICOM breast MRI (MR Image, Enhanced MR or Legacy Converted Enhanced MR), whole study as a zip or by DICOMweb STOW-RS. Breast DCE studies are large 4D series; the full dynamic acquisition is expected.
• Prior study in prior_files for comparison and for neoadjuvant response.
• Dynamic phases are ordered by TemporalPositionIdentifier or acquisition time.

Requirements

Requirement	Why
Non-contrast T1	Fibroglandular tissue segmentation and FGT category
Pre-contrast and at least two post-contrast T1 phases	Lesion detection, initial enhancement and delayed curve type
Subtractions (computed when absent)	Lesion conspicuity and BPE grading
T2 (optional)	Lesion characterisation, cysts and lymph nodes
DWI (optional)	Additional characterisation and treatment response
Silicone-sensitive sequences (implant studies)	Rupture signs

Optional context

Indication (screening, staging, problem solving, neoadjuvant monitoring, implant), risk status, menstrual phase or hormonal therapy, and treatment dates go in clinical_context.

The result

Section	Content
Fibroglandular tissue	Breast and FGT masks, FGT %, category a–d
Background parenchymal enhancement	BPE level (minimal, mild, moderate, marked) and symmetry
Lesions and kinetics	Mass and non-mass enhancement: 3D mask, size, location and kinetic curve type
BI-RADS assessment	BI-RADS lexicon descriptors per lesion and the final assessment category 0–6
Neoadjuvant therapy response	Functional tumour volume change from baseline and pCR likelihood
Implant integrity	Intracapsular and extracapsular rupture signs
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: SEG for breast, FGT and lesion masks; SR TID 1500 per lesion with tracking identifiers across visits; subtraction and kinetic maps as Parametric Maps; KOS key images and Encapsulated PDF.
• Reporting: ACR BI-RADS Atlas 5th edition, MRI section (BI-RADS MRI 2013), for composition, BPE, descriptors and assessment categories.
• Response: functional tumour volume as defined in the I-SPY trials.

Reading order

The result presents FGT and BPE before the lesions so the assessment is read in context: moderate, marked or asymmetric BPE can mask or mimic lesions.