UK MRI Breast screening protocol

A Magnetic Resonance Imaging (MRI) breast screening examination is a complex examination that needs to be performed to a specified minimum standard. Centres providing MRI breast screening should meet specified standards relating to equipment, protocol and interpretation, and should participate in audit. Minimum and expected standards are recommended below for equipment, sequences, contrast and reading of examinations. It is anticipated that symptomatic examinations will also be performed to the same high quality.

A UK working party met to form a consensus on what might be expected in a UK MRI unit. The multidisciplinary group comprised members of the Magnetic Resonance Imaging breast study (MARIBS) steering group, representatives from industry including MR manufacturers and software companies producing tools for use in breast MRI analysis. The members and affiliations are listed. This document is on the Royal College Radiologists Breast Group web site for consultation. The manufacturers comments have been included at the end for interested parties. These comments were made in 2006 and in view of the rapid change in hardware and software may now be superseded. It is recommended that questions related to specific MRI machines and software release be directed to the manufacturer.

Equipment

MRI system

High field modern MRI machine should be used when undertaking breast MR

Minimum standard 1T
Expected standard > or = 1.5T

Comment – evidence from 3T should be assessed to ensure same quality is achieved with higher field system

Breast coil

Dedicated bilateral breast coil to be used for examinations (either open or closed)
Uniform signal homogeneity across coil
Number of elements

Minimum standard: 2 channel
Expected standard: > or = 4 channel

MRI sequences

T1W dynamic set. Fat saturation is preferred to image subtraction if good fat saturation can be achieved within the recommended acquisition time. > or = one pre-contrast. Acquire for >6 minutes post contrast

T2W fat suppressed high resolution or STIR (Short Tau Inversion recovery)
Higher resolution pre and post contrast T1W with fat saturation 0.6mm minimum in plane-resolution. 50% improvement in voxel size compared with dynamic unless already achieved in the dynamic series. These are not required if the dynamic series achieves 0.6mm resolution or better.

In a surveillance exam, need:

Voxel size ≤ 2mm
Dynamic sequence ≤ 60 sec
Total exam time ideally <30 mins
Minimal motion
Integrated fat suppression – desirable in dynamic series where achievable
Examine both breasts – ideally with isotropic voxels
Uniform signal homogeneity across image

Slice thickness

Bilateral breast examination should be undertaken with high resolution to achieve small lesion detection

Minimum standard: ≤ 2.5mm
Expected standard: ≤ 2mm

In plane resolution

Expect high resolution in plane resolution

Minimum Standard <1.3 mm
Expected standard <1.0 mm

Acquisition time

Dynamic contrast examination is required to detect abnormal enhancing lesions to improve specificity

Minimum standard: ≤ 60 secs dynamic acquisition
Expected standard: ≤ 45 secs acquisition

Fat suppression

A fat suppression technique should be used to improve lesion conspicuity. This can either be an integral aspect of the contrast enhanced sequence or obtained by using a subtraction technique

Minimum standard: subtraction
Expected standard: integrated fat suppression

Breast movement

Breast movement should be minimised during the procedure in order to obtain best quality and dynamic data

Minimum standard: support of breasts to minimise motion
Expected standard: as above

Contrast

Pump injection of at least 0.1mmol/kg contrast is recommended with ³ 3cc/sec flow rate with 20ml bolus saline
Initiate start of post contrast exam to time centre of k space at 20 secs post injection
UK recommendations on avoidance of contrast reactions and Nephrogenic Systemic Fibrosis should be followed.

Image registration

An image registration technique should be used if motion artefact is a problem during the examination

Hormonal Factors

Timing of examination

The examination should be carried in the mid portion of the menstrual cycle to reduce normal parenchymal tissue enhancement.
- Minimum standard: Time examination to day 6-16 of the menstrual cycle

Hormone Replacement Therapy (HRT)

There is some evidence that HRT increases parenchymal enhancement. However there is no evidence to suggest that stopping HRT reduces normal glandular tissue enhancement.

Breast Biopsy

Where possible repeat targeted breast Ultrasound (US) and review of mammograms is recommended. Where lesion is not seen on conventional imaging referral to a recognised MR breast biopsy centre is advised.

Reading recommendations

When reporting MRI it is recommended the following information is included in the report: type of image acquisition, breast density, lesion type – morphology, size, enhancement pattern, size of lesion in 3 dimensions, assign score ( score 1-5: 1 normal, 2 benign, 3 probably benign, 4 suspicious of malignancy, 5 malignant).

- Minimum standard – include information as listed above
- Expected standard - standardized report ( to be developed with UK lexicon, and linked to BIRADS)

Mammograms – It is recommended that all relevant imaging is available when reporting MRI , especially for symptomatic referrals.

Audit should be undertaken to ensure that reporting accuracy and biopsy rates are acceptable. As yet no standards have been agreed.

Double reporting –where possible reports should be made by two radiologists who are familiar with breast MRI. There is little evidence for reporting experience but the following is suggested:

Minimum standard – 30 cases/year, second opinion sought where necessary
Expected standard - > or = 50 cases/year, double reporting where possible.

Screening recall standards

Recall rate

A minimum number of women should be recalled for further imaging or biopsy

Minimum standard <10%
Expected standard <7%

Data Import

Minimum standard is uncompressed DICOM 3.0
Data should be written to a CD directly from the scanner/workstation and not via PACS (some centres have reported that data has been rescaled by PACS).
If using PACS then a QA process should be in place to ensure that the source data has not been modified

DICOM compliance

It is expected that all modern MRI machines and workstations will be DICOM compliant
- Minimum and Expected standard

PACS Integration

Minimum standard – QA process should be in place to ensure no alteration of the source data.

Viewing Tools

It is important to have good analysis tools available on the reporting workstation. These will include - Subtraction/MPR/MIP software, signal enhancement analysis package with ROI tools. Hotspot analysis for diagnosis – max SI, Guidance on scaling - air/fat?), software package to support image analysis, image registration features, ability to change display of images
- Minimum standard – as above
- Expected standard – as above together with pharmacokinetic modelling, ROI - Pixel by pixel analysis with wash-in and wash-out maps.

Manufacturers comments

Equipment - Siemens are unable to positively verify compliance on 1T systems but believe they would comply with at least the minimum standards. Philips anticipate that 1T open systems will become an important part of breast imaging. The SNR is as good as a cylindrical 1.5T and the open architecture makes it valuable for cancer centres doing MR RT planning. Therefore they support the minimum standard field remaining at 1T. GE state that the minimum requirements are achievable on their systems.

Breast Coil - Siemens Breast Array is 4 element 2 channel coil and Breast Matrix coil is a 4 channel coil when operated in Dual mode and 2 channel coil when operated in CP mode. Siemens also currently supply 3rd party coils of 4 channel and 7 channel configuration. Siemens have stabilising mechanism and pads integrated into coil design to reduce breast movement. GE state that all their sites have a 4 or 8 channel breast coil.

MR Sequences – Siemens state both hardware and software enable all expected standards to be met using Magnetom Tim systems and the Symphony 1.5T systems installed with the latest version of operating software. However, a number of the requirements might only be met if our customer has purchased certain software/hardware options eg: BRACE (3D non-rigid registration tool).
GE state that any site that any Signa HD or HDx scanner has the option of VIBRANT. The 3D T1 FS sequence which allows shimming of both breasts and ensures a robust fat saturation.

Image registration – Siemens state dynamic ROI is available as standard in the Mean Curve tool used for calculating time/intensity curves. This allows manual adjustment of the ROIs should the breast have moved in between dynamic measurements. A 3D non-rigid algorithm registration software option (BRACE) is available to Siemens customers operating at the latest software level, B13. GE do not have correction for movement on the postprocessing ADW software FUNCTOOL but is able to supply CadSTREAM from Confirma which can correct for motion.

Viewing Tools - Siemens state Syngo multi modality workstations have the majority of these tools/applications as standard.

Working party members

Fiona J Gilbert, Professor of Radiology, University of Aberdeen
Martin Leach, Professor of Imaging, Institute of Cancer research
Anwar Padhani, Consultant Radiologist, Mount Vernon, London
Lindsay Turnbull, Professor of Radiology, University of Hull
Ruth Warren, Consultant Radiologist, Addenbrookes, Cambridge
Emma Hurley, Consultant Radiologist, South Manchester University Hospitals
Preminda Kessar, Consultant Radiologist, Bromley Hospitals
Will Teh, Consultant Radiologist, North West London Hospitals
Erica Scurr, Superintendent Radiographer, The Royal Marsden NHS Foundation Trust
David Collins, MR Physicist, Institute of Cancer research
Martin Graves, MR Physicist, University of Cambridge
Gary Liney, MRI Physicist, Hull Royal Infirmary
Geoff Parker, Senior Research Fellow, University of Manchester
Linda Pointon, MARIBS Co-ordinator, Institute of Cancer Research
Gek Kwan-Lim, MARIBS Research Manager, Institute of Cancer Research
Emillie Bryant, Scientific Officer, Institute of Cancer Research
Andreas Muehler, President, CAD Sciences
Henry Wyszomierski, Chief Technology Officer, CAD Sciences
Raymond Joslin, Chairman, CAD Sciences
Mary Gatewood, Confirma
Daniel White, Confirma
Bart Maertens, Confirma
Elga Grimes, MR Application Specialist, GE Healthcare
Dylan Pritchard, GE Healthcare
Trevor Furniss, Sales Manager UK & Eire, Invivo
Elizabeth Moore, MR Applications Specialist, Philips Medical Systems
David Clark, MRI Applications Specialist, Siemens Medical Solutions
Patrick Revell, Siemens Medical Solutions