Guidance

Technical guidelines for MRI for the surveillance of women at higher risk of developing breast cancer

Updated 27 September 2024

Applies to England

1. Introduction

The NHS Breast Screening Programme (NHSBSP) sets and monitors the standards for the imaging surveillance of women who are assessed as being at very high risk (VHR) of developing breast cancer. Screening with Magnetic Resonance Imaging (MRI) has proven to be effective for this cohort of women. See Role of magnetic resonance imaging in breast cancer management (nih.gov)

NHS breast screening services are responsible for ensuring that women at VHR receive consistent and high-quality surveillance with MRI and digital mammography according to the protocols for surveillance of women at very high risk of developing breast cancer.

This guidance sets out the technical standards required for MRI in VHR women, including acquisition of images, reporting and quality control (QC).

Revisions to previous guidance

This guidance has been revised specifically in the following areas:

• there are more details on the image acquisition parameters in the new guidance in section 2.5.
• there are more details on the viewing tools such as colour parametric software in section 2.6.
• all women should have second-look ultrasound at assessment rather than a selected group (section 3.4)
• short term recall should be undertaken between 3 and 6 months instead of between 6 and 12 months (section 3.6)
• the updated report gives more details on undertaking QC using breast coil element check and Suppression effectiveness (section 4.1)
• there are changes to the equipment standards stated in Appendix 2
• standards set out in Appendix 2 of 2012 report on standards for readers and centres have been removed
• the MRI accreditation form has been removed from the updated report

2. Standards for equipment and image acquisition

2.1 MRI system

1.5 and 3 tesla clinical MRI systems are acceptable for acquiring breast MRI for screening women at very high risk. Dedicated bilateral breast coils with at least 8 elements (channels) are required to improve signal to noise ratio and enable the use of parallel imaging techniques. The system should be regularly maintained including quality control (QC) of both the general system and the breast coil. Users of 3 tesla systems (especially older machines) must be aware of causes of poor image quality such as localised shading due to non-uniformity of the transmit field.

2.2 Patient positioning

Women should be positioned prone. Breast movement should be minimised during the procedure in order to obtain the best quality dynamic data. This is best achieved by ensuring patient comfort and moderate breast support or immobilisation.

2.3 Contrast agents administration

A standard dose (typically 0.1mmol/kg) according to the agents Summary of Product Characteristics (SPC) of macrocyclic gadolinium-based contrast agent should be administered as a compact bolus. The contrast agent should be followed with a saline flush.

UK recommendations on administering gadolinium based MRI contrast agents should be followed. Please see advice from the Royal College of Radiologists on Gadolinium based contrast agent administration and Medicines and Healthcare products Regulatory Agency (MHRA) and European Medicines Agency (EMA).

The information leaflet sent to very high risk women prior to their appointment informs them of the benefits and the risks associated with gadolinium-based contrast agents.

2.4 Timing of examinations

The MRI examination should be performed between day 6 to 16 of the menstrual cycle when there is less normal parenchymal tissue enhancement.

Pregnancy and lactation are associated with high background parenchymal enhancement and reduced sensitivity of MRI. MRI should only be carried out during either pregnancy or lactation if there is potential considerable benefit to the patient. RCR Guidance on imaging surveillance in pregnancy and lactation should be followed.

2.5 Scanning parameters

Bilateral imaging is essential. This is most commonly achieved by scanning in the axial plane. Alternatively, a coronal or interleaved sagittal approach can be used. Users should be aware of how sequence parameters can be adapted to optimise image quality and acquisition times utilising MRI acceleration techniques such as parallel imaging.

A description and list of requirements for essential and optional MRI sequences as part of a breast screening protocol are given below.

Scanning Protocol - Mandatory protocol

Sites should acquire as a minimum

1. T2-weighted (T2W) fast / turbo spin echo sequence
2. Dynamic contrast-enhanced (DCE) 3D T1-weighted (T1W) sequence.

Recommendations for the MRI scanner and T2W and DCE-MRI sequences are summarised in Appendix 2.

It is recommended that there should be at least one pre contrast non fat suppressed sequence (T1 or T2W). Where a Dixon technique is utilised it may be possible to reconstruct non fat suppressed (i.e. in-phase images) and fat suppressed images (i.e. water only images) from the same dataset.

Operator intervention may be required if inadequate fat suppression is observed including reshimming and reselection of the centre frequency. Some MR systems may allow the user to assume fat is the dominant signal and this is recommended. When this is set the system should automatically perform this frequency shift, otherwise this shift must be performed manually. Care should be taken to check for artefacts associated with particular reconstruction techniques if used, such as localised fat-water swaps in Dixon-based imaging.

Users should be aware of how sequence parameters can be adapted to optimise image quality and acquisition times utilising MRI acceleration techniques such as parallel imaging.

Essential: T2-weighted acquisition (fast/turbo spin echo)

T2W sequences provide diagnostic information complementary to dynamic contrast enhanced studies which aids in characterising benign (cysts, fibroadenomas) and malignant lesions and intramammary nodes, as well as aiding in the general assessment of breast anatomy and density.

T2W sequence requirements

T2W requirements	Notes
Coverage of both breasts including axillary tail
Acquired* in-plane voxel size ≤1 mm in both frequency encoding and phase encoding directions	The stated voxel sizes relate to the acquired voxel size, before any zero-filling/interpolation
Slice thickness ≤3 mm	Slice thickness and any slice gaps should ideally be chosen so that there is a direct correspondence with the DCE reconstructed images slice positions.
T2 weighted	A long repetition time ( >3000 ms) and long echo time ( >80 ms) should be chosen to achieve adequate T2 weighting
Echo train length	T2 blurring is caused by the signal decaying whilst the echoes are being acquired. If the echo train duration is too long then significant image blurring can occur
Fat Suppression	The sequence may be acquired either with or without fat suppression. The choice of the type of fat suppression (or none) depends on local preference

• It is common for MRI data to be reconstructed with smaller voxels through the use of zero-filling/interpolation, although this does not improve the actual spatial resolution. The stated voxel sizes below relate to the acquired voxel size, before any zero-filling/interpolation

Essential: Dynamic Contrast Enhanced 3D T1W acquisition (spoiled gradient echo)

A 3D (DCE) acquisition is the most sensitive technique for detecting breast cancer. A T1-weighted sequence is repeated multiple times before, during and immediately after the injection of a gadolinium-based contrast agent. This approach maximises the chances of acquiring an image when the lesion is at its maximum enhancement and where there is maximum contrast between the lesion and surrounding breast parenchyma. Dynamic acquisitions also allow curves of the signal intensity - time changes within any lesion - to be produced to aid in characterisation.

At least one pre-contrast imaging volume should be acquired. The temporal resolution requirements specified below should allow the spatial resolution requirements to be met.

To ensure that there is sufficient contrast between an enhancing lesion and the background parenchyma the centre of k-space of one of the DCE-MRI phases should occur no later than 60 seconds after injection.

DCE 3D-T1W Sequence requirements

DCE T1W requirements	Notes
Coverage of both breasts including axillary tail
Fat suppression recommended	Sites may either use a non fat suppressed acquisition or a fat suppressed one (e.g. frequency selective fat suppression, Dixon or water-only excitation but not STIR). A fat suppressed acquisition is recommended but if there are local concerns about the homogeneity and effectiveness of fat suppression then non fat suppressed sequences may be used. If fat suppression is used then a set of pre-contrast images with the same fat-suppression settings must be acquired and checked to ensure fat has been uniformly and correctly suppressed over the imaging volume
Acquired* in-plane voxel size: in both directions: Achievable ≤1 mm, Acceptable ≤1.2 mm	Ensure that this is the acquired in-plane voxel size before any zero-filling/interpolation
Acquired* slice thickness: Achievable <2 mm, Acceptable <3 mm	Ensure that this is the acquired slice thickness not the reconstructed slice thickness
Temporal resolution: Achievable <90 s, Acceptable <120 s	The recommended timings for temporal resolution should allow all sites to achieve the recommended spatial resolution and coverage

T1 weighted	The sequence should be a 3D spoiled gradient echo sequence. A short repetition time (e.g. <10 milliseconds) and a flip angle >10° should be chosen to provide T1 weighting. A short echo time (TE) is recommended to minimise T2* weighting (TE <5 milliseconds)
Post processing	Subtraction is essential to improve the contrast between enhancing and non enhancing regions.

• It is common for MRI data to be reconstructed with smaller voxels through the use of zero-filling/interpolation, although this does not improve the actual spatial resolution. The stated voxel sizes below relate to the acquired voxel size, before any zero-filling/interpolation

Scanning protocol - Optional sequences:

1. Diffusion weighted sequence
2. T1W non fat suppressed sequence
3. High spatial resolution post-contrast T1W with isotropic voxels (spoiled gradient echo with fat suppression)

No sequence specific recommendations are made for these optional sequences.

2.5.2a Optional: pre-contrast T1W acquisition without fat suppression

A pre-contrast T1W image may be useful to aid lesion characterisation.

Additionally, the sensitivity to local changes in magnetic susceptibility and general high signal produced from a T1W spoiled gradient echo may be useful to locate any metal breast marker clips.

2.5.2b Optional: High spatial resolution post-contrast T1W(spoiled gradient echo with fat suppression)

It may be preferable to acquire an additional post-contrast T1W sequence over a longer duration to allow production of a higher spatial resolution image. This may particularly be the case where sites only just meet the spatial resolution requirements for the DCE-MRI sequence. Low grade cancers enhancing slowly may be better demonstrated on this sequence.

2.5.2c Optional: Diffusion weighted MRI

Diffusion weighted MRI (DWI) may be helpful to characterise breast lesions and has been shown to improve specificity and it is frequently included in breast MRI protocols. However, its utility has not been explicitly demonstrated in a screening context.

2.6 Analysis and viewing tools

Interpretation of DCE-MRI requires the reader to identify regions of contrast enhancement (foci, masses or non-mass enhancement) on the Maximum Intensity Projections (MIPs) and images obtained after baseline subtraction,

Regions of interest (ROIs) should be drawn around suspicious lesions. The radiologist should check whether any patient movement is impacting on the accuracy of the curve by scrolling through the dynamics.

Viewing tools

A. Volume subtraction, multiplanar reconstruction (MPR), and MIP

Signal enhancement analysis package, with tools to generate uptake curves from ROIs

B. Colour parametric software to identify lesions with Type 2 and Type 3 enhancement. This allows the reader to identify lesions that require ROI analysis which can be helpful in breasts with moderate to marked background parenchymal enhancement.

The software must allow the user to outline a ROI (containing the enhancing tissue) and should calculate a contrast agent uptake curve by displaying the image intensity as a function of time for the chosen ROI. This will enable classification of the uptake curve. Careful attention should be paid when performing analysis of contrast agent uptake curves that patient movement has not corrupted the curve. The contrast agent uptake curve depends on the ROI outlining, which is a critical part of the process. A smaller ROI may be more accurate.

C. Patient motion between the dynamic phases of a dynamic contrast enhanced acquisition may be rectified in some instances by the use of non-rigid 3D registration software.

3. Reporting MRI examinations

Clinicians responsible for reporting MRI should be aware that image quality may vary between machines.

3.1 Reading requirements

Appropriately trained and qualified breast radiologists, breast clinicians, consultant radiographic practitioners and advanced radiographic practitioners can report breast MRI images. See Standards for interpretation and reporting of imaging investigations

The majority of women in the very high risk group will undergo both X-ray mammography and breast MRI. A few women will have MRI surveillance only (for example because of their young age), while others may have mammography only (for example, if MRI is contraindicated or in women aged over 50 years with low breast density – BI-RADS A).

Reading requirements include:

• for women undergoing MRI and mammography, the most recent mammogram should be available at the time of MRI reporting

• where possible, mammography should be performed on the same day as the MRI

• the minimum standard is that mammography must be performed within two weeks of the MRI

• the mammograms and MRI images must be independently double-read

• one of the MRI readers must be an MRI reader who meets the NHSBSP standards for a breast screening image reader (including the requirement to report a minimum of 5,000 cases per year)

• MRI readers within the NHSBSP should each report a minimum of 100 breast MRI examinations (including screening and symptomatic cases) per year

• the results and images of a client’s previous screening MRI and mammography examinations should be available for comparison

3.2 MRI screening reporting categories

The following standard MRI screening reporting categories must be used:

MRI code	Outcome	Description
MRI 1	Normal	No enhancing lesions
MRI 2	Benign	All non-enhancing lesions that are morphologically benign and have a benign enhancement curve
MRI 3	Indeterminate	Probably benign, including morphologically unclear lesions with benign enhancement curve and also morphologically benign lesions with suspicious enhancement curve
MRI 4	Suspicious	Suspicious morphology and enhancement curve
MRI 5	Malignant	Malignant morphology and enhancement curve

MRI screening and assessment forms have been developed for the national breast screening system (NBSS). These can be used to record the MRI opinion, and the outcome of the assessment of MRI-detected abnormalities. These forms are shown at Appendix 1.

3.3 Assessment of abnormalities

Abnormalities detected by MRI and/or digital mammography should be assessed by the local NHSBSP screening team, which should ideally include a clinician familiar with breast MRI. This assessment need not take place in an NHSBSP screening assessment unit, but the assessing team must satisfy NHSBSP requirements and standards. If assessment takes place outside of the NHSBSP, then details of the examinations performed and of the person(s) performing them, plus the outcome of multidisciplinary team (MDT) discussions, should be returned to the responsible breast screening centre and data entered onto NBSS.

Initial assessment should be with conventional imaging which may include supplementary mammographic imaging such as digital breast tomosynthesis and ultrasound.

The MRI reporting forms at Appendix 1 should be used to record data to be entered on NBSS.

3.4 Selection of cases for second-look using ultrasound

All women should have second-look ultrasound at assessment.

The report should recommend the action to be taken when the second-look ultrasound is normal. For example, ‘No further action/MRI biopsy/MRI follow-up’.

3.5 Selection of cases for MRI-guided biopsy

The following cases would be suitable for MRI-guided biopsy:

• a suspicious abnormality seen on MRI, but second-look ultrasound and mammographic review is normal
• Masses >5mm or focal enhancement >10mm where there is no morphological or ultrasound correlate on T1/T2-weighted scans and second look US

MRI-guided biopsy screening centres

MRI-guided biopsies should only be performed at a designated MRI biopsy screening centre. A service level agreement should be in place for this service. To be accredited to undertake MRI biopsies, centres must:

• perform a minimum of 12 MRI-guided vacuum assisted breast biopsies per year

• demonstrate expertise by doing at least 50 (non-MRI) image-guided vacuum assisted biopsy procedures per year
• audits should be undertaken to ensure that reporting accuracy and biopsy rates are at an acceptable level

It is expected that the great majority of significant abnormalities detected by MRI will be identified at assessment and biopsied using conventional imaging.

Rates and Positive Predictive Value (PPV) of MRI biopsy should be monitored by the screening service and screening quality assurance service. Standards for reporting accuracy and biopsy rates will be developed in the light of operational experience.

3.6 Selection of cases for follow-up MRI

The following cases are considered suitable for follow-up MRI:

• suspicious mass ≤5 mm, or area of enhancement ≤10 mm: follow-up MRI should be undertaken at 3 to 6 months using a short term recall protocol following an MDT discussion
• biopsy performed but not conclusive: follow-up should be undertaken at 3 to 6 months using a short term recall protocol following an MDT discussion

3.7 Recall rate

The number of women recalled for further imaging or biopsy should be kept to a minimum and comply with the targets required as outlined in the consolidated standards.

3.8 Waiting times

Standards for waiting times from screen to results and assessment for women in the very high risk programme are published in the consolidated standards.

4. Quality control

4.1 Quality control (QC)

There must be regular QC checks to ensure that MRI scanners and processes used as part of the breast screening programme meet the NHSBSP acceptable standards for equipment, image acquisition, and data storage. These test results should be reviewed by the MRI lead radiographer locally and the MR clinical scientist supporting MRI where available. A set of recommended tests can be found in Appendix 3. Some tests are mandatory whilst optional tests can be considered best practice.

Most MRI centres will take out comprehensive maintenance contracts on their MRI scanners. This will ensure that a range of manufacturer tests will be performed at regular intervals and that the MRI scanner will be within the manufacturer tolerances for those tests that are carried out. The exact tests that are performed vary between manufacturers. Importantly, breast coils may not be tested by default as part of the MRI manufacturers’ service schedule, and individual sites are encourage to discuss with their MRI vendor to ensure routine checks of their breast coil are established.

Annual, start-up and post major upgrades checks

The following checks should be made when the breast screening service is first established and repeated in accordance with the manufacturers recommendations or annually whichever is more frequent. It is advisable to also perform these checks after major upgrades or major repairs have been made to the MRI scanner.

• confirm that the MR scanner meets all of the manufacturer’s performance specifications.
• where possible, the QC tools provided by the manufacturer should be used as directed. The breast coil should operate to manufacturers specification( eg signal to noise ratio for each RF receive coil element). If necessary the manufacturer may need to run additional tests to ensure consistent high quality is maintained.
• If the manufacturer does not provide user-accessible QC tools for the breast coil, then baseline images should be acquired shortly after the breast coil is confirmed to have passed the manufacturer’s QC
• if a breast coil is repaired or replaced check that it has passed all the manufacturer’s image quality checks before it is used clinically

Evidence of the above should be submitted annually to the breast screening lead radiographer to provide assurance.

Routine regular QC checks

It is important to assess that each element of the breast coil is operating correctly at regular and frequent intervals. Where sites use frequency-selective fat suppression techniques, it is also important to assess magnetic field uniformity (via fat suppression effectiveness) to identify potential causes of systematic fat suppression failure (for example unknown metallic object within the scanner bore).

For sites where breast MRI screening is generally performed each week, checks should be made on a weekly basis. For sites where breast MRI screening is performed less frequently the QC checks should be performed before the equipment is used for breast screening.

The mandatory weekly / regular checks are:  

TEST	ASSESSMENT
Breast coil element checks	Manufacturer QC (preferred) or visual assessment of individual coil element images and comparison with baseline images.
Suppression effectiveness	Visual assessment of suppressed images relative to baseline image set

The additional, non-mandatory tests are:

TEST	ASSESSMENT
Breast coil element checks	Calculate overall SNR from regions
Suppression effectiveness	Calculate suppression effectiveness from regions

Breast coil element check

Images should be obtained that allow the user to verify that each element of the breast coil is functional (no significant change in signal compared to baseline). Ideally this would involve the system measuring the SNR from each coil element (via the manufacturer provided QC) but for systems where that is not easily achievable then a simple visual comparison of images reconstructed from each coil element separately relative to a baseline set will suffice. The baseline set must have been acquired shortly after coil passed manufacturer’s QC.

Sites may also make a global measurement of SNR from the combined image but where this is not easily achieved it may be omitted.

Action levels specified by the manufacturer should be used where available. If a coil fails the manufacturer’s action level it should be taken out of service until the fault is rectified.

If sites do not have access to the manufacturer provided QC for the breast coil, the manufacturer should be asked to investigate and rectify the problem where significant subjective differences in appearance as compared to a baseline image set for individual coil elements images is detected (ideally using similar windowing for the images).

Indirect magnetic field homogeneity test (Suppression Effectiveness)

To ensure optimal image quality, it is advisable to perform an assessment of magnetic field homogeneity, as poor homogeneity can interfere with sequences that suppress the signal of fat in an image.

In cases where frequency-selective fat suppression is utilized in clinical practice, the quality assurance (QA) fat suppression sequence should ideally match this technique. However, modifications that do not affect fat suppression, such as reducing the number of slices and increasing slice thickness, may be made to increase practicality.

If a Dixon fat-water separation sequence is employed in clinical practice, using a similar sequence for QA is less relevant due to the nature of Dixon sequences. Nevertheless, B0 inhomogeneities can still impact the quality of Dixon sequences. Therefore, a frequency-selective sequence should be included as part of routine QA to ensure optimal results.

Suppression effectiveness can be measured by performing a measurement using either:

• a dedicated phantom with separate water and fat compartments and a fat-suppressed acquisition or
• a water/doped phantom and acquiring a water suppressed acquisition and a non water suppressed acquisition

The uniformity and magnitude of fat suppression should be assessed qualitatively. An optional quantitative measurement of the percentage signal loss on the suppressed sequence compared to the non-suppressed sequence can be calculated. The quality of fat suppression should be reviewed on a patient by patient basis and ongoing problems identified and rectified.

Results of the routine QC programme should be stored appropriately. A written record of results should be kept and the following should be recorded as a minimum:

• Who undertook the test
• The date the tests were undertaken
• Which tests were carried out and the results
• Whether the tests were acceptable/in tolerance
• What action was taken if the results were out of tolerance

There should be a process in place for immediate review of the regular QC results and remedial steps which need to be taken if action thresholds are exceeded.

4.2 Accreditation

As part of the routine commissioning process, commissioners should ensure that all MRI centres used for very high risk women within the NHS BSP meet the acceptable standards for:

• equipment and measurement set-up
• performance of routine QC checks
• radiological training and experience

4.3 Audit

It is recommended that sites enlist the services of an MR clinical scientist to take part in an annual review of a site’s QC results and any subsequent actions. (see section 4.1 and appendix 3)

The effectiveness of the surveillance process and protocols will be evaluated, and annual data collection will be undertaken.

Regular audits of fat suppression and general image quality should be undertaken on clinical images by the image readers.

5. Appendix 1: MRI Reporting proformas

FORM A: BSP MRI Screening request

FORM B: BSP MRI Screening request

6. Appendix 2: Equipment and acquisition standards

7. Appendix 3 MRI quality control (QC) guidance

It is recommended that sites enlist the services of an MR clinical scientist to advise on the initial set up of the QC and to provide annual review of a sites QC results.

This document provides a generic overview of testing MRI breast screening equipment and protocols for signal to noise ratio and fat suppression effectiveness. The principles outlined should be applicable to all MR scanner manufacturers. However, vendor specific instructions on implementing this guidance may be required.

Signal to noise ratio and suppression effectiveness

7.1 Signal to noise ratio (SNR)

TEST	MANDATORY / NON-MANDATORY	ASSESSMENT
Breast coil element checks	Mandatory	Manufacturer QC (preferred) or visual assessment of individual coil element images.
Breast coil element checks	Non-mandatory	Calculate overall SNR from regions

Table 1: Signal to noise ratio (SNR) tests

7.2 Overview

1. The aim of the test is to identify individual coil element failures.

2. The manufacturer’s proprietary coil QC should be performed if available.

3. The bespoke QC protocol (section 1.3) should be followed if it is not possible to perform the manufacturer’s QC or if the centre requires more detailed results.

4. It is critical that SNR measurements are made reproducibly. The same phantom position, acquisition (sequence parameters) and analysis should be used each time. The most common causes of SNR failures are incorrect phantom positioning, incorrect ROI positioning and phantom filler movement. If QC fails at first, double check positioning, analysis and repeat the measurement to ensure there is not a problem with phantom filling material moving.

5. If any faults with the coil are suspected then contact either the manufacturer or the MRI Clinical Scientist.

7.3 Manufacturer’s QC

1. Follow the instructions provided by the manufacturer for QC of the breast coil. It is usually important to position the phantoms exactly as described and to allow the phantoms 5 to 10 minutes to settle once positioned at isocentre.

2. Run the QC as per the manufacturer’s instructions and note the results. On systems where it is possible to access the SNR results for each coil element these should be recorded.

Bespoke QC (only required if manufacturer’s breast coil QC cannot be

performed)

1. Use the manufacturer supplied phantoms that are recommended for the breast coil. These may be specialist breast phantoms or bottle phantoms.

2. Acquire a set of images with the parameters below from each individual coil element shortly after the MRI vendor has confirmed that the breast coil has passed its own QC process and is working as intended. These images will then serve as a baseline for future QC measurements.

3. Position the phantom within the breast coil using appropriate phantom supports. The phantom should be positioned so that it is within the sensitive region of all coil elements. The manufacturer should be able to recommend appropriate phantoms and positioning.

4. Set up a spin echo sequence with the following recommended parameters: TR = 500 ms, TE < 20 ms, matrix = 128 x 128, slice thickness = 10 mm, field of view = 300 – 340 mm, no parallel imaging. The sequence should be able to produce separate images for each breast coil element (uncombined images) with or without user intervention.

5. The orientation, field of view and number of slices should be sensitive to any coil element failures and will depend on the coil utilised. Specific advice for each manufacturer should be followed to achieve this. For coils that only distribute the coil elements in the head foot and left right directions a single slice coronal acquisition will usually be most appropriate. If the coil elements are also distributed in the anterior-posterior direction then sites should verify that a single coronal acquisition is sufficient.

7.4 Qualitative Analysis

1. Keep a reference copy of separate (uncombined) coil element images from the first time that the coil is assessed. These are the baseline images and should ideally be acquired immediately after the breast coil has been confirmed to be within specification by the system manufacturer or a medical physicist. Each coil element image should show high signal intensity in the region adjacent to the coil element position.

2. Visually inspect all the uncombined coil element images. If there is any visual change from the baseline acquisition then this should be investigated further and the problem coil element noted. Verify that the correct coil elements are being compared as the images may not always be produced in the same order.

7.5 Quantitative Analysis - OPTIONAL

1. To obtain a quantitative measure of SNR the average signal and average noise require measuring.

2. The signal should be measured by placing a region of interest (ROI) within the centre of the phantom. Typically the ROI should be approximately half the diameter / minimum dimension of the phantom in the plane of acquisition. Record the mean signal intensity from the ROI for both left and right breast phantoms. Care must be taken to ensure that the ROI is in the same location as for previous analyses.

3. The noise can either be measured by drawing ROIs in regions of the image where there is no phantom (ensuring that a mask has not been applied to this region) or by forming a subtraction image from two consecutive image sets acquired with the same imaging parameters. The same ROI (size and position) should be used as for the signal measurement. In either case the noise is proportional to the standard deviation of the signal intensities within these regions.

4. See IPEM report 112 - If the noise is measured in background regions then the SNR (note correction factors have been omitted for ease of calculation) is calculated as:

SNR (Signal to noise ratio) can be calculated by dividing the 'mean signal intensity of the unsubtracted image' by the 'mean standard deviation of signal intensities from background ROIs'

If the noise is measured on the subtracted image then the SNR is calculated as:

If the noise is measured on the subtracted image then the SNR is calculated as: SNR = 'mean signal intensity of the unsubtracted image' divided by 'mean standard deviation of signal intensities from subtracted image'

Action thresholds

To establish appropriate measures for signal-to-noise ratio (SNR), it is advisable for individual sites to define local action thresholds. These thresholds can be determined by examining the inherent variability of a set of baseline images acquired when the coil is known to be operating correctly or by referring to established literature on SNR fluctuations. If the SNR falls below the defined threshold, the following steps should be taken:

• A) confirm the positioning of the region of interest
• B) thoroughly assess image appearance
• C) carefully reposition phantoms and obtain new data
• D) ensure that there is a phantom settling time of no less than 10 minutes
• E) check coil connection
• F) check for any signs of obvious coil damage

7.6 Suppression Effectiveness

TEST	MANDATORY / NON-MANDATORY	ASSESSMENT
Suppression effectiveness	Mandatory	Visual assessment of suppressed images relative to baseline image set
Suppression effectiveness	Non-mandatory	Calculate suppression effectiveness from regions.

Table 2: Suppression effectiveness tests

Overview of methods

1. Possible methods for testing suppression effectiveness are listed in table 3.

2. Option (a) in table 2 above is recommended when available because it uses the standard dynamic diagnostic sequence and can be performed with phantoms that are supplied by the MR manufacturer. However, the user must ensure that the scanner centre frequency is adjusted to 3.5 ppm above the measured frequency of the phantom contents. Some systems have an option to assume the MR signal is dominated by fat. When this is set the system should automatically perform this frequency shift, otherwise this shift must be performed manually.

3. An acceptable alternative is (b) although this requires a change from the standard dynamic diagnostic sequence to water suppression instead of fat suppression. Water suppression may not be available with the standard diagnostic sequence so a slightly different sequence type may need to be selected.

Option	Phantom	Tuned frequency	Suppression pulse	Sequence
a	Water based OR fat based	Must be 3.5 ppm (approx. 220 Hz at 1.5T and 440 Hz at 3T) above measured frequency of phantom contents	Fat suppression pulse	Standard diagnostic sequence with reduced slices
b	Water based OR fat based	Use automatic frequency detection of scanner	Water suppression pulse	May require different sequence to diagnostic one
c	Water and fat	Ensure system has tuned to water and not fat	Fat suppression pulse	Standard diagnostic sequence with reduced slices

Table 3: Methods for testing suppression effectiveness

Acquisition

1. The following instructions refer to option (a) in table 3.

2. Use the same phantom set up as for the SNR measurement.

3. The same phantom, phantom position, acquisition (sequence parameters) and analysis should be used each time.

4. Perform a 3D T1-weighted (spoiled) gradient echo acquisition that provides whole coverage of the phantom. Ideally this should be the fat suppressed sequence that is used clinically. To reduce the acquisition time and the number of images that need to be assessed, the number of slices may be reduced (e.g. to 20) provided that the slice thickness is increased to maintain coverage of the sensitive region of the breast coil.

5. Acquire two sets of images, one with fat suppression and the other without fat suppression. Ideally, the two sequences should have identical timing parameters (TR / TE / flip angle). Identical timing parameters may be difficult to achieve with some manufacturers and specific advice for each scanner should be obtained.

6. Keep a reference copy of the images obtained from the first time the test is run. This should ideally be as soon as practicable after the scanner is verified as performing to specification (in particular with respect to shimming).

Qualitative Analysis

1. Ensure that the fat suppression has a similar appearance on acquired images and reference images when windowed to similar levels.

2. If the suppression appears significantly different to the baseline images then investigate the scanner for obvious causes (e.g. small metallic objects within the bore of the magnet). Inspect any patient images carefully for nonuniformity of the fat suppression.

Quantitative Analysis (Optional)

1. An analysis can be performed by drawing ROIs on the non fat suppressed and fat suppressed sequences and calculating the suppression effectiveness for both breasts:

Where unsuppressed and suppressed are the mean ROI signal intensities from the non fat suppressed and fat suppressed sequences respectively.

1. Sites should monitor the suppression effectiveness and investigate any results that do not fall within the expected range from established baseline measurements.