Educational Objectives Breast MRI: Image Quality ... · Breast MRI: Image Quality, Artifacts and Quality Control Donna M. Reeve, MS, DABR, DABMP ... • Artifact assessment Breast

D.M.Reeve: AAPM 2011

1

Breast MRI: Image Quality, Artifacts and Quality ControlBreast MRI: Image Quality,

Artifacts and Quality Control

Donna M. Reeve, MS, DABR, DABMPDonna M. Reeve, MS, DABR, DABMPDepartment of Imaging PhysicsDepartment of Imaging PhysicsDepartment of Imaging PhysicsDepartment of Imaging Physics

Educational ObjectivesEducational Objectives

•• Describe common artifacts and image quality issues Describe common artifacts and image quality issues encountered in clinical breast MR images.encountered in clinical breast MR images.

•• Discuss methods for addressing breast MR image quality Discuss methods for addressing breast MR image quality problems.problems.pp

•• Describe quality control procedures accreditation programs Describe quality control procedures accreditation programs specific to breast MRI.specific to breast MRI.

Challenges:

• Adequate SNR ACR: “not too grainy”

• Good spatial resolution

• 1mm x 1mm in-plane resolution

3 li hi k

Breast MR Image QualityBreast MR Image Quality

• ≤ 3mm slice thickness

• Temporal resolution dynamic series (60-90 sec/phase)

• Absence of (or minimal) artifacts

• Effective, uniform fat suppression

SNRSNRSNRSNRPotential causes of low SNR:

• Low field strength• Poor Coil connection• Coil element failure• Incorrect center frequency selectionIncorrect center frequency selection• Protocol parameters:

- Small voxels (large matrix, small FOV, thin slices)

- trade-offs: speed, SNR, resolution

1 0H s ave

samp

FOV FOVSNR N B f

N N

Spatial resolutionSpatial resolutionSpatial resolutionSpatial resolution

High contrast spatial resolution requires small voxels:• Large matrix• Small FOV• Thin slices

= FOV / N Resolution (frequency-encoding direction

= FOV / N Resolution (phase encoding direction)

slice Resolution (slice

Trade-offs: • Longer scan time if phase matrix is increased

• Reduced SNR improve with 3T imaging

direction)

Tscan = TR Nave N Acquisition time

1.5T 1.5T 3T 3T

3T3T–– trade additional SNR for increased spatial trade additional SNR for increased spatial resolution or faster scan time resolution or faster scan time

FSE T2W w/ fat sat,FSE T2W w/ fat sat,

FOV 220mm, 256x192, 4mmFOV 220mm, 256x192, 4mm

FSE T2W w/ fat sat, FSE T2W w/ fat sat,

FOV 200mm, 320x192, 3mmFOV 200mm, 320x192, 3mm


2

Breast MRI ArtifactsBreast MRI Artifacts

Common artifacts in breast MRICommon artifacts in breast MRI

•• MotionMotion

•• Truncation artifactsTruncation artifacts

Out of volume wrapOut of volume wrap•• Out of volume wrapOut of volume wrap

•• SusceptibilitySusceptibility artifactsartifacts

•• Signal nonSignal non--uniformityuniformity

•• Poor or nonPoor or non--uniform fat saturationuniform fat saturation

Motion artifactsMotion artifactsOccur in the phase encoding direction. Caused by cardiac Occur in the phase encoding direction. Caused by cardiac motion, respiration, patient movement. Results in phase motion, respiration, patient movement. Results in phase mismis--mapping in kmapping in k--space due the time delay between space due the time delay between phasephase--encoding and signal readout. encoding and signal readout.

Truncation ArtifactsTruncation ArtifactsTruncation ArtifactsTruncation Artifacts•• Occur at high contrast edges.Occur at high contrast edges.

•• Also Also known known as Gibbs or as Gibbs or “ringing” artifact“ringing” artifact..

•• Can occur in either phase or frequency direction.Can occur in either phase or frequency direction.

•• Minimized Minimized by increasing matrix by increasing matrix sizesizeHi h t t ti l l ti iHi h t t ti l l ti i•• High contrast spatial resolution improvesHigh contrast spatial resolution improves

•• Scan time also increases if phase matrix is increasedScan time also increases if phase matrix is increased

•• SNR reducedSNR reduced

Object profile Measured intensity profile

Truncation ArtifactsTruncation ArtifactsFrequency

Pha

se

Small ACR phantom in 3T GESmall ACR phantom in 3T GE HD Breast array HD Breast array

320x192 matrix320x192 matrix 320x320 matrix320x320 matrix

Aliasing or “WrapAliasing or “Wrap--Around” Around” ArtifactsArtifacts


FOV

Aliased to lower frequencyAliased to

higher

-fmax fmaxf0

gfrequency

Aliased imagef f faliased true Nyquist 2

Phase


•• Increase FOV to include entire Increase FOV to include entire object object -- increase phaseincrease phase--encode encode steps to maintain resolution steps to maintain resolution (trade(trade--off: off: impacts scan time) impacts scan time)

Phase

•• Swap phase and frequencySwap phase and frequency--encoding directions : shorter encoding directions : shorter dimension in phasedimension in phase--encoding encoding directiondirection. (trade-off: motion artifacts)

•• Use “No phase wrap” or “antiUse “No phase wrap” or “anti--aliasing” techniques.aliasing” techniques.


3

Peripheral signal artifact(annefact, star artifact)

Peripheral signal artifact(annefact, star artifact)

e

FSE: Star artifact – bright signal close tocenter of images.

FSE: Spine exam using phased arraysurface coil .

Pha

se

Pha

se

Signal originates in region where gradients are nonlinear. FID from 180 pulses not crushed – aliases back into image.

Magnetic Susceptibility ArtifactsMagnetic Susceptibility ArtifactsMagnetic Susceptibility ArtifactsMagnetic Susceptibility Artifacts

Metallic objects can cause distortions of the static and Metallic objects can cause distortions of the static and gradient fields, RF fields, or bothgradient fields, RF fields, or both Ferromagnetic objects Ferromagnetic objects -- distort Bdistort Boo and Band B1 1 fieldsfields

NonNon--ferromagnetic metal objects ferromagnetic metal objects -- distort Bdistort B11 fieldsfields

Typical effects are Typical effects are signal voids signal voids

and and geometric distortionsgeometric distortions..

Most noticeable on GRE (rather thanMost noticeable on GRE (rather than

SE or FSE). Appearance reduced SE or FSE). Appearance reduced

with wider receive BW, shorter TE.with wider receive BW, shorter TE.

Frequency selective fat satFrequency selective fat sat

•• FrequencyFrequency--selective fat or silicone saturation is routinely selective fat or silicone saturation is routinely used in breast imaging. Frequency of saturation pulse must used in breast imaging. Frequency of saturation pulse must match resonant frequency of fat/silicone. match resonant frequency of fat/silicone.

•• Selection of resonant peak usually automated, but may Selection of resonant peak usually automated, but may require manual adjustment require manual adjustment Technologist training Technologist training essential.essential.

•• Uniform saturation dependent on homogeneity of BUniform saturation dependent on homogeneity of B0 0 field field within the imaged volume:within the imaged volume:

•• challenge (breasts off challenge (breasts off isocenterisocenter))

•• shimming is importantshimming is important

Fat/silicone saturation - peak selectionFat/silicone saturation - peak selection

f

~220 Hz ~100 Hz 1.5T

~200 Hz 3T (fat-water separation 3.5 ppm)

~440 Hz

Effective chemicallyEffective chemically--

waterfat

silicone

Increasing frequency

selective fat or silicone selective fat or silicone saturation depends on saturation depends on accurate peak selection. accurate peak selection.

GE: center on water, saturates GE: center on water, saturates fat signal at fat signal at --220Hz 220Hz

Composition of breast tissueComposition of breast tissue

Glandular tissueGlandular tissue SiliconeSilicone Adipose tissueAdipose tissue

Composition of breast tissue (adipose/glandular/silicone) determines Composition of breast tissue (adipose/glandular/silicone) determines appearance of spectrum. Selecting the correct peak to achieve fat or appearance of spectrum. Selecting the correct peak to achieve fat or

silicone saturation can be challenging.silicone saturation can be challenging.

T2 FSE,T2 FSE, fat satfat sat

Left breast Left breast Right breast Right breast

Difference in center frequency 440 Hz (3.5 Difference in center frequency 440 Hz (3.5 ppmppm) ) = 3T difference in = 3T difference in resonant frequency between fatresonant frequency between fat--water. water. CCentered on entered on fat peak fat peak fat sat fat sat failure.failure.

CF 128,173,640 Hz CF 128,173,200 Hz


4

3D3D T1 postT1 post--contrast dynamic, fat satcontrast dynamic, fat sat

Left breast Left breast Right breast Right breast

Center frequency = 128,173,593Center frequency = 128,173,593

Good fat saturation achieved on both sidesGood fat saturation achieved on both sides

ShimmingShimming•• Shim volume Shim volume –– useruser prescribes graphicallyprescribes graphically

•• Current in shim coils adjusted to optimize BCurrent in shim coils adjusted to optimize B00 field field uniformity within the volume. Improves uniformity of fat uniformity within the volume. Improves uniformity of fat saturation.saturation.

Saturation failureSaturation failure

•• Bandwidth of the Bandwidth of the sat pulse sat pulse centered on fat sufficient to centered on fat sufficient to saturate both fat and siliconesaturate both fat and siliconesignalsignal –– both appearboth appear dark.dark.

T2W fast spin-echoTR =3500ms / TE =86 ms echo train length = 8122 Hz/pixel bandwidth 256x256 matrix, 200 mm FOV1 averagefat sat

•• Incomplete saturation of fat Incomplete saturation of fat and/or silicone can occur in and/or silicone can occur in regions with large static regions with large static magnetic field magnetic field inhomogeneties. inhomogeneties.

AuroraSUPERSHIMTM

One vendor’s shim coil system is designed to improve fat One vendor’s shim coil system is designed to improve fat suppression suppression in breast MR imaging in breast MR imaging

ShimmingShimming

typical typical MRI MRI Aurora Aurora MRI MRI

Oval shape and positioned more anterior so that bilateral breast Oval shape and positioned more anterior so that bilateral breast tissue is centered within shim volume.tissue is centered within shim volume.

http://www.auroramri.com/mri/index.shtml

Signal uniformity and breast coil design

1.5T 1.5T SentinelleSentinelle coil coil --

axial image of small axial image of small ACR phantomACR phantom

3T GE HD array 3T GE HD array --

axial image of small axial image of small ACR phantomACR phantom

Breast MRI Quality ControlBreast MRI Quality Control

Quality control of MRI systems used for diagnostic breast MR imaging and biopsy guidance

• Is important to ensure production of high quality images by evaluating whether MRI scanner and coils used for breast imaging are performing consistently over timebreast imaging are performing consistently over time.

• Should be part of a comprehensive MRI quality control program.

• May be required to satisfy accreditation program requirements


5

ACR Breast MRI Accreditation ProgramACR Breast MRI Accreditation Program

ACR Breast Magnetic Resonance Imaging Accreditation Program ACR Breast Magnetic Resonance Imaging Accreditation Program (BMRAP) launched in May 2010 under breast imaging (BMRAP) launched in May 2010 under breast imaging accreditation programs (mammography, stereotactic breast accreditation programs (mammography, stereotactic breast biopsy, and breast ultrasound).biopsy, and breast ultrasound).

• Separate from the ACR MR Accreditation Program (MRAP)

• Provides accreditation for MR systems used for breast accreditation for MR systems used for breast imagingimaging::

•• Dedicated breast MRI systems orDedicated breast MRI systems or

•• Whole body MRI systems with Whole body MRI systems with

•• detachable tabledetachable table--top breast coiltop breast coil

•• dedicated tables with integrated breast coilsdedicated tables with integrated breast coils

ACR Breast MRI Accreditation ProgramACR Breast MRI Accreditation Program

•• Currently no phantom image submission.Currently no phantom image submission.

•• Clinical cases (bilateral) for each scanner acquired within 2 months* of Clinical cases (bilateral) for each scanner acquired within 2 months* of date on testing memorandum. Cases reviewed by two radiologists.date on testing memorandum. Cases reviewed by two radiologists.

•• BIBI--RADS category 1: negative, or 2: benign findingsRADS category 1: negative, or 2: benign findings

•• BIBI--RADS category 6: known, enhancing, biopsyRADS category 6: known, enhancing, biopsy--proven malignancyproven malignancy

CD/DVD must open in less than 2 minutesCD/DVD must open in less than 2 minutes•• CD/DVD must open in less than 2 minutes.CD/DVD must open in less than 2 minutes.

•• Quality control program and medical physicist involvement essentially the Quality control program and medical physicist involvement essentially the same as MRI Accreditation Program (MRAP)same as MRI Accreditation Program (MRAP)

•• Breast MRIBreast MRI--specific experience/training requirements for technologists specific experience/training requirements for technologists and radiologists.and radiologists.

IntersocietalIntersocietal Accreditation Commission Magnetic Accreditation Commission Magnetic Resonance Lab (ICAMRL) offers a breast MRI Resonance Lab (ICAMRL) offers a breast MRI accreditation option.accreditation option.•• No phantom image review.No phantom image review.

•• Clinical images acquired within the last year submitted for Clinical images acquired within the last year submitted for

ICAMRL Accreditation ProgramICAMRL Accreditation Program

C ca ages acqu ed w t t e ast yea sub tted oC ca ages acqu ed w t t e ast yea sub tted oreview.review.

•• Breast MRIBreast MRI--specific experience/training requirements for specific experience/training requirements for radiologists. radiologists.

•• Cost similar to ACR BMRAP program.Cost similar to ACR BMRAP program.

•• Quality control program established by Quality Assurance Quality control program established by Quality Assurance Committee and/or the Medical Director. Tests performed according Committee and/or the Medical Director. Tests performed according to manufacturer’s performance standards.to manufacturer’s performance standards.

•• Acceptance testing required after installation and major upgrades.Acceptance testing required after installation and major upgrades.

•• Periodic maintenance (PM) requiredPeriodic maintenance (PM) required

ICAMRL Accreditation ProgramICAMRL Accreditation Program

•• QC performed by MR technologist, service engineer, medical QC performed by MR technologist, service engineer, medical physicist or “qualified expert”. physicist or “qualified expert”.

•• Daily and periodic QC required Daily and periodic QC required •• Equipment function and safetyEquipment function and safety

•• Center frequencyCenter frequency

•• SNRSNR

•• UniformityUniformity

•• Artifact assessmentArtifact assessment

Breast MRI QCBreast MRI QC

Physicist:

• MRI system performance evaluation after scanner installation, annually and following major repair or hardware/software upgrade

A l QC f ll RF il (i l di b MRI il )• Annual QC of all RF coils (including breast MRI coils)

Service engineer:

• Periodic/preventative maintenance (PM). Frequency defined in service contract

MRI technologist:

• Daily/weekly phantom scans

Breast RF Coil Quality ControlBreast RF Coil Quality Control

ti ll di lwww.sentinellemedical.com


6


Establish baseline coil performance in order to monitor coil

performance over time.

• Coil inspection

• Signal-to-noise ratio (SNR)

• Signal uniformity• Signal uniformity

• Phased array coils: compare SNR for individual channels

• Artifact evaluation (including ghosting)

• Using QC protocol

• Using clinical protocol

Coil inspection•• Inspect coil, cables, cable insulation, ports and Inspect coil, cables, cable insulation, ports and

connectors for damageconnectors for damage

•• Could present a safety issue or result in low SNR or Could present a safety issue or result in low SNR or image artifactsimage artifacts


image artifacts.image artifacts.

www.invivocorp.com

Coil testing:•• Important to test coils:

• after installation of new scanner or new coils• at least annually• whenever artifacts or coil problems occur


•• Manufacturers provide a coil manual for each coil• includes description of clinical use of the coil• may include detailed description of coil test procedure• may include pass/fail limits• may only say “establish baseline and monitor over time”

Coil testing:• Test all available coil configurations.

• Note any error messages.

• In our experience the vendor is more likely to respond


p y pto coil QC failure when manufacturer’s QC procedure is followed.

Uniformity:• Follow procedure in 2004 ACR MRI QC Manual (min,

max signal intensity within small ROI)

Artifact evaluation•• Evaluate images acquired using QC protocolEvaluate images acquired using QC protocol

•• To troubleshoot artifacts observed on patient images To troubleshoot artifacts observed on patient images may acquire images of homogeneous QC phantom may acquire images of homogeneous QC phantom


using clinical protocol.using clinical protocol.

Consistent scan/measurement methods: Consistent scan/measurement methods: Identical phantom and positioning within coil

•• Homogeneous phantom (sphere, cylinder, custom)Homogeneous phantom (sphere, cylinder, custom)•• ACR or other phantomACR or other phantom

Identical scan parameters:


p•• Pulse sequence, timing parameters, slice thickness and position, Pulse sequence, timing parameters, slice thickness and position,

matrix, FOV, receive bandwidth, etcmatrix, FOV, receive bandwidth, etc•• Record center frequency, transmit gain/attenuation, receiver Record center frequency, transmit gain/attenuation, receiver

gainsgains

Identical measurement methods, ROI positions•• SNR, signal uniformity, ghosting, stability testsSNR, signal uniformity, ghosting, stability tests•• Evaluation of channel performanceEvaluation of channel performance


7

SNR methodsSNR methods

Signal Signal measured in ROI within magnitude imageSignal measured in ROI within magnitude image

Noise Noise measured in background of signal imageNoise measured in background of signal image NEMA approach: Noise measured in subtractionNEMA approach: Noise measured in subtraction NEMA approach: Noise measured in subtraction NEMA approach: Noise measured in subtraction

image: image: •• 2 images acquired with identical protocols and 2 images acquired with identical protocols and

prescanprescan parameters (center frequency, transmit parameters (center frequency, transmit gain/attenuation, receiver gains)gain/attenuation, receiver gains)

Noise measured in “pure noise” image acquired with Noise measured in “pure noise” image acquired with no RF excitation no RF excitation


Coronal Sagittal composite image Noise image

Images acquired with individual coil elements


Unilateral biopsy mode Bilateral imaging mode

• Signal ROI: SE T1, unilateral and bilateral imaging modes

• Noise image (no RF excitation)



The small ACR phantom may The small ACR phantom may be utilized for breast coil be utilized for breast coil quality control. Phantom quality control. Phantom contains objects that enable contains objects that enable evaluation of: evaluation of: geometric accuracygeometric accuracy

41

•• geometric accuracy geometric accuracy

•• high contrast spatial resolution high contrast spatial resolution

•• slice thickness accuracy slice thickness accuracy

•• slice position accuracy slice position accuracy

•• image intensity uniformityimage intensity uniformity

•• ghosting ghosting

•• low contrast detectabilitylow contrast detectability

Small ACR phantomsBilateral mode


SE, Philips Achieva 1.5T, 16 channel breast array

Images courtesy of R. Price, PhD


8

SummarySummary

•• High quality breast MR images exhibit adequate SNR and contrast, High quality breast MR images exhibit adequate SNR and contrast, high resolution, absence of artifacts, and uniform fat/silicone high resolution, absence of artifacts, and uniform fat/silicone saturation. Compromises are often necessary to achieve this in saturation. Compromises are often necessary to achieve this in addition to good temporal resolution of the DCE series.addition to good temporal resolution of the DCE series.

•• Effective and uniform fat saturation can be challenging to achieve Effective and uniform fat saturation can be challenging to achieve and can be more consistent with technologist education and use ofand can be more consistent with technologist education and use ofand can be more consistent with technologist education and use of and can be more consistent with technologist education and use of proper shim techniques.proper shim techniques.

•• Two breast MRI accreditation programs are currently available Two breast MRI accreditation programs are currently available (ACR, ICAMRL)(ACR, ICAMRL)

•• A comprehensive quality control program, including testing of breast A comprehensive quality control program, including testing of breast RF coils, is important to ensure optimal performance and image RF coils, is important to ensure optimal performance and image quality of breast MRI systems.quality of breast MRI systems.

ReferencesReferences1. ACR Magnetic Resonance Imaging (MRI) Quality Control Manual, 2004.

(under revision)

2. ACR Technical Standards for Diagnostic Medical Physics Performance Monitoring of MRI Equipment, revision 2009. www.acr.org

3. Breast MRI Accreditation Program Requirements, 7/27/2011. www.acr.org

4. ICAMRL 2010 Standards for Accreditation in Magnetic Resonance Imaging Parts I and II, 9/10.www.icamrl.org

5. Determination of Signal-to-Noise Radio (SNR) in Diagnostic Magnetic Resonance Imaging, NEMA Standards MS 1-2008. www.nema.org

Documents

Educational Objectives Breast MRI: Image Quality ... · Breast MRI: Image Quality, Artifacts and Quality Control Donna M. Reeve, MS, DABR, DABMP ... • Artifact assessment Breast