Quality Assurance of MRI Systems David Hearshen Slide 1 ... · Slide 2 Quality Control of MRI ... subject to variability from: ... Occurs when the RF center frequency is off resonance

Quality Assurance of MRI Systems David Hearshen

1

Slide 1

Technical Aspects ofTechnical Aspects ofQuality Control inQuality Control in

Magnetic Resonance Imaging Magnetic Resonance Imaging

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 2Quality Control of MRIQuality Control of MRI

SystemsSystems

David Hearshen, Ph.D.

Department of Radiology

Henry Ford Hospital, Detroit, MI

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 3

ObjectivesObjectives

To gain a general understanding of MRsystem variability

To be able to set up and manage an MRI QCand Compliance Testing program

To gain a general understanding of MRIArtifacts

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


2

Slide 4Objectives of an MRI QCObjectives of an MRI QCProgramProgramTo implement a system practical enough for

technologists to utilize on a daily basis

To obtain meaningful results which revealsystem changes before patient care isaffected

To document problems and corrective actionin a manner which satisfies accreditationand regulatory requirements

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 5System VariabilitySystem VariabilityMagnetic resonance imaging systems are

subject to variability from:Drift in RF Electronics

RF Coils

RF Transceiver Chain

Magnetic Field Decay

Foreign ferro- or para- magnetic material

Introduction of material producing MRI signal

Gradient system failures

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 6

Effects of System VariabilityEffects of System VariabilityRadio Frequency effects

SNR LossOccurs when the RF center frequency is off resonance

Power is insufficient to produce 90 or 180 deg pulses

If the RF coil is detuned

If the RF coil homogeneity is not optimized

NoiseCan increase with failures in preamplifier or amplifier

Can be introduced with failures in the RF shielding(most often door).

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


3

Slide 7Gradient effectsGradient effects

Geometric DistortionCan occur with when maximum amplitude is not

achieved or gradient is miscalibrated.

Can occur if eddy current compensation changes

Spatially Dependent SNR LossCan occur when local field homogeneity decreases

Can occur when gradient waveforms are notoptimized (eg eddy currents) resulting in incompleterephasing of echo

Phase ErrorsCause misregistration in Phase Encoding direction

Can arise from any of the 3 orthogonal gradients

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 8

Magnetic FieldMagnetic Field Inhomogeneity Inhomogeneity

SNR LossCan occur when local total field homogeneity

decreases ⇒ ↓T2*.

Can occur when center frequency is shifted beyondbandwidth of receiver

In Plane Geometric DistortionCan cause regions of low or high signal when

resonant frequency is shifted beyond bandwidth of apixel

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 9

Magnetic FieldMagnetic Field Inhomogeneity Inhomogeneity

Geometric Distortion in Slice Select DirectionCan cause misregistration of signal as a function of

slice location, or distortion of slice profile causingincreased “crosstalk”.

Chemical Shift Frequency OffsetsFrequency selective pulses (e.g. lipid or water

saturation, mtc) can have a spatial dependence totheir intended effect.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


4

Slide 10

Daily TestingDaily Testing

ACR MRI Accreditation Recommendation

Magnetic Field Stability (Center Frequency)

Signal to Noise Ratio

Artifact Inspection

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 11

Standard Pulse SequencesStandard Pulse Sequences

T1 Weighted T2 Weighted

TR 500 2000

TE min full <20msec 20/80 or 30/90

Matrix 128 or 160 128 or 160

Coil Head Head

#excitations 1 1

FOV 24or 25 24 or 25

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 12

Magnetic Field StabilityMagnetic Field Stability(Center Frequency)

Test Object

Procedure

Sequence Parameters

What to record

Factors influencing measurement

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


5

Slide 13

Center Frequency (Center Frequency (contcont.).)

Typical magnetic field stability is on the order of 0.1ppm per hour or 2.4 ppm per day. At 1T thiscorresponds to approximately 100 Hz decay of thecenter frequency per day and 150 Hz per day at1.5 tesla. A graph of center frequency vs. timeshould indicate steady decay, however, typicalinhomogeneity over a 20 cm DSV is also withinthis range. Therefore, it is important to set upconditions over which reproducible measurementsof the center frequency can be obtained.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 14Center Frequency (Center Frequency (contcont.).)

Additional magnetic fields due to variation inmagnetic susceptibility can distort the line shapeof the resultant water signal and produce an errorin determining the peak position. This isminimized by using a spherically symmetricuniform phantom.

Measurement of center frequency is furthercomplicated by the presence of additional staticmagnetic fields used to adjust the homogeneity forspecific applications (shims). The shim settingscan be modified in two ways.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 15


Routine preventative maintenance for MRI calls forperiodic re-shimming of the magnet using all theavailable set of shim fields. These shims involvecoils designed to vary magnetic field as a functionof spatial coordinates. Usually the set of coilsincludes shims which perturb the magnetic fieldspanning geometry characterized by second orderspherical harmonics. These shim settings areusually not changed by the user.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


6

Slide 16


Many manufacturers provide user controlledshimming using the three linear gradient magneticfields, either manually or with a computercontrolled algorithm (autoshimming). Somemanufacturers utilize autoshimming before eachpulse sequence.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 17

Center Frequency (Center Frequency (contcont))

Unaccounted shimming can add an additionalmagnetic field resulting in an increase in thecenter frequency from day to day. It is importanttherefore to use the same setting of the shimmagnetic field gradients for each measurement.

The graph of center frequency over a period of 15months shows an increase in center frequency atseveral time points. The overall decrease inmagnetic field is less than 160 Hz << less thanmanufacturer’s specification

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 18Center Frequency over 15 Month PeriodCenter Frequency over 15 Month Period

63860600

63860650

63860700

63860750

63860800

63860850

63860900

7/7/9

4

7/18/9

4

8/1/9

4

8/15/9

4

8/29/9

4

1/16/9

5

2/20/9

5

4/25/9

5

5/8/9

5

8/7/9

5

9/11/9

5

9/26/9

5

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


7

Slide 19


ProcedurePrescribe a standard T1 weighted single slice pulse

sequence

Make sure any auto shimming algorithm is turned off.

Set up parameters for the auto pre-scan and load the pre-set values of the shims.

Perform the clinical auto pre-scan procedure.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 20


ProcedureRecord the time of the measurement, resultant center

frequency, and shim file settings.Optionally record the transmitter and receiver gains. If

the same shim file setting was used subtract theprevious days frequency from the current value andcompare with the specified decay.

If the measurements were not acquired at the same timeon successive days, adjust for total decay time.

Action level: decay within manufacturer specification.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 21

Signal to Noise RatioSignal to Noise Ratio

Test ObjectsMaterial

GeometryRF Homogeneity

Magnetic Susceptibility Effects

Positioning

Coil

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


8

Slide 22


In measuring signal to noise ratio, a uniformtest object is necessary in order to minimizeloss of signal due to variation in magneticsusceptibility as well as spatial variation inthe signal intensity due to RFinhomogeneity.

Solutions which are not comparable to tissueconductivity may not load the coil properly.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 23

Signal to Noise Ratio (Signal to Noise Ratio (contcont.).)

The choice of coil is determined by the largestvolume of procedures performed clinically, thebest homogeneity characteristics, and the ease ofboth acquiring and analyzing the data. Though insome instances the head coil is not always used forthe largest number of procedures, it usually hassuperior homogeneity compared to surface coilsused in spine or extremity imaging.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 24


Example Pulse sequenceTR - 500 to 800 milliseconds

TE - 10 to 20 milliseconds

Single 10mm slice at isocenter.

Single excitation, 256 x 160 matrix

FOV - 20 to 24cm.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


9

Slide 25


These parameters are used to generate a T1weighted image. Specific TR, TE etc., canbe taken from a common clinical protocol,provided the T1 of the test object material iswithin the range of normal tissue. Oncechosen, these parameters should beremained fixed as they all affect theresultant signal to noise ratio.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 26


Use of the minimum TE as calculated by theimaging system is not recommended sincethis value will change with changes tosoftware and hardware.

Test object dimension and FOV should bechosen to fill greater than 85% of the usablefield view of the coil in order to adequatelytake into account RF homogeneity effects.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 27


ACR Accreditation “T1” Series

TR 500

TE 20 msec

Matrix 256

Coil Head

#excitations 1

FOV 24or 25

Time 2:16

Use of 256 matrix is notnecessary for SNR,however, if using theACR phantom, othertests may be analyzed.The extra minute is smallcompared to set up time(aligning the phantom).

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


10

Slide 28

Signal to Noise Ratio (Signal to Noise Ratio (contcont.).)Noise Characteristics

Theoretically, noise in magnetic resonance israndom. In practice however, there can besystematic contributions to background noisearising mainly from unwanted phase shiftsacquired in the RF transceiver chain. Sincephase is used in the image reconstructionalgorithm to supply spatial information, thesephase shifts produce spatial variation in thisbackground signal (“ghosts”).

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 29


In measuring signal to noise ratio, a uniformtest object is necessary in order to minimizeloss of signal due to variation in magneticsusceptibility as well as spatial variation inthe signal intensity due to RFinhomogeneity.

Solutions which are not comparable to tissueconductivity may not load the coil properly.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 30

AnalysisAnalysis

If S is the mean of the ROI from the original imageand σN is the standard deviation of an ROI fromthe noise image, then

Record the SNR along with means and standarddeviations.

Action level: SNR within manufacturer specification

SNRS

N

= 2σ

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


11

Slide 31

Signal to Noise Ratio ( Signal to Noise Ratio (contcont.).)

Noise Estimation: Method 1While it may be attractive to measure the noise of

a region of interest from a non signal producingarea of the image field of view outside the testobject using a single image, this practice mayintroduce systematic noise into the calculationof the SNR. If a single acquisition is desired,estimate noise from multiple regions in theimage.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 32

Procedure1. Scan test object once using T1 series.

2. Choose ROI that encompasses at least 75% ofthe test object. Measure and record mean andstandard deviation of the ROI

3. Calculate PSG from ACR accreditation.

4. If PSG acceptable, use average of 4 ROI σ’s toestimate noise.

Benefit is that PSG measurement can be tracked

Signal to Noise Ratio ( Signal to Noise Ratio (contcont.).)___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 33 Signal to Noise Ratio ( Signal to Noise Ratio (contcont.).)

ROIs forsystematic noisemeasurement

Susceptibilityfrom airbubble

ACR T1 Series Slice #7

Signal ROI

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


12

Slide 34


SNR and PSG for three systemsFixed 11.5T Fixed 2 1.5T Mobile 1.0 TMean σ Mean σ Mean σ

L 12.08 5.82 11.22 5.87 20.96 11.77R 11.18 6.87 10.59 6.15 22.59 10.40T 10.42 5.13 9.93 5.07 20.28 10.18B 9.04 4.41 10.47 5.14 19.14 10.04Signal 1168.00 50.35 1407.19 40.32 1451.46 50.00

Noise Ave 10.68 5.56 10.55 5.56 20.74 10.60

PSG 0.001627 0.000501 0.001423

SNR 297.2202 358.0868 193.6942

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 35


Noise Estimation: Method 2An accepted practice is to image the test object

twice, repeating the acquisition within a fewminutes of the first. The noise is estimatedfrom an ROI in an image formed by subtractingthe first acquisition from the second. Theassumption is that the signal (and noise) in thetwo images are uncorrelated.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 36Noise ImageNoise Image

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


13

Slide 37


Procedure1. Scan test object twice using same parameters.

2. Subtract the second image from the first on apixel by pixel basis. (This may require theimages to be downloaded to a work station).

3. Choose ROI that encompasses at least 75% ofthe test object. Measure and record mean andstandard deviation of the ROI in both theoriginal images and the subtracted noise image.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 38SNR Method 2SNR Method 2

P la ne S σ S NRAxia l 1185 11.5 145.7255S a g itta l 1184 11.62 144.0989Co ro na l 1174 11.5 144.3728

1.5T Body Coil

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 39

SNR over an 15 Month PeriodSNR over an 15 Month Period

A graph of SNR in the head coil over 15 months show day to dayvariation of SNR with a mean =104.3 and σ=2.6

90

92

94

96

98

100

102

104

106

108

110

7/7/9

4

7/18/9

4

8/1/9

4

8/15/9

4

8/29/9

4

1/16/9

5

2/20/9

5

4/25/9

5

5/8/9

5

8/7/9

5

9/11/9

5

9/26/9

5

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


14

Slide 40Systematic BackgroundSystematic Background

Axial

Body Coil

FOV 36 cm

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 41Systematic BackgroundSystematic Background

Sagittal Coronal

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 42

Weekly TestingWeekly Testing

ACRProcessor sensitometry

OtherPhase Stability

Magnetic Field Homogeneity

Signal - to - Noise Ratio in orthogonal planes

Image Uniformity

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


15

Slide 43

ProcessorProcessor Sensitometry Sensitometry

Though analogous to other digital imagingmodalities, printing of MR films can beproblematic due to variation in the output ofthe MR system. This variation is due toVariable gain in the transceiver subsystem

Variation in tissue contrast (T1, T2, )

Variation in operator selectable parameters

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 44


Example ProtocolDisplay SMPTE pattern

Inspect SMPTE image on the monitorNote 5% and 95% levels visibility

Note any artifacts

Film image using preset window and levelWindow width 256, Window level 128

Scan output image and recordSpeed, Contrast, Base+Fog

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 45Modified SMPTE ImageModified SMPTE Image

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


16

Slide 46

0.00

0.50

1.00

1.50

2.00

2.50

3.00

1 4 7 10 13 16 19

Dens ity Ste p

Op

tica

l D

en

sit

y

Tone Scale CalibrationTone Scale Calibration___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 47


This procedure allows a standardized way ofcomparing the image on the monitor with theoutput of the laser printer. The film densities maybe scanned, plotted, and interpreted in a mannerconsistent with other processors used in thedepartment, however, this test does not narrow thesource of a potential problem to either themonitor, MRI console output (either video ordigital), laser printer performance, or processorperformance.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 48Processor QCProcessor QC

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


17

Slide 49SNR ResultsSNR Results

75

80

85

90

95

100

105

110

115

120

125

7/7/9

4

7/18/

94

8/1/

94

8/15/9

4

8/29/9

4

1/16/9

5

2/20/

95

4/25/

95

5/8/9

5

8/7/9

5

9/11/9

5

9/26/9

5

+/- 10%

mean =104.3 and σ=2.6

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 50

-200

-150

-100

-50

0

50

100

150

200

7/7/9

4

7/18/

94

8/1/9

4

8/15

/94

8/29/

94

1/16/

95

2/20

/95

4/25

/95

5/8/9

5

8/7/

95

9/11

/95

9/26

/95

∆υ

1.0T

1.5T

Daily Change in FrequencyDaily Change in Frequency___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 51Review of Test ResultsReview of Test Results

Technologist - Daily ChecklistInspect QC phantom for artifacts

Inspect SMPTE Pattern for artifacts (monitor)

Inspect SMPTE Pattern for artifacts (film)

Graph Center Frequency, SNR, Sensitometry

PhysicistReview of Daily QC

Monthly or Quarterly (ACR Semi-Annual)

Review Artifacts, Action Limits -ad hoc

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


18

Slide 1

Artifact Inspection

MRI is susceptible to artifacts:Low SNR (and low energy) compared to other

imaging modalities

Ability to directly modify spatial frequencysampling

Sensitivity to metallic objects

Patient – RF coil interaction is variable

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 2

Artifact Inspection

Artifacts may appear differently in test objectsand patients because of differences in:Magnetic susceptibility

Coil loading

T1, T2 of test object material

High contrast between plastic (no signal) andfluid

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 3EPI 1/2 FOV Phase Artifact

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


19

Slide 4

Artifact Inspection

A single source may cause artifacts to appeardifferentlyUsing different coils

Using different pulse sequence parametersAnatomic plane

pulse sequence type (eg spin echo vs. gradient echo)

timing (TR, TE, gating)

Spatial resolution (FOV, matrix size)

Using different materials

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 5

Gradient Echo vs. SE of Bleed

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 6

RF Artifacts

External RF NoiseSpecific frequency usually one pixel in

width

Phase Incoherent

Independent of anatomic plane

May change position with FOV orbandwidth

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


20

Slide 7External RF Noise

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 8External RF Noise

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 9

RF Artifacts

Transmit Receive switch failureInductive coupling of transmit coil and

surface coil

Spatial variation in RF power

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


21

Slide 10Transmit - Surface Coil Artifact

Phased Array Spine

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 11ADC Overflow (Clipping)

Occurs when the signal is larger thanthe maximum ADC value, usually if gainis set too high.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 12

Non Phase Encoded Signal

Stimulated Echoes

FID (inhomogeneous B1, gradients)

Signal from outside FOVUsually if phase and frequency directions

swapped.

Small FOV in larger body part

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


22

Slide 13Stimulated Echo Interference

Fringe spacing inverselyproportional to differencein echo timing.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 14Axial, Sagittal Lumbar Spine

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 15Homogeneity ArtifactsLocal inhomogeneity

causes shifts in position in frequency encodingdirection

causes dephasing within a voxel

causes frequency selective pulses to be spatiallyshifted

Generally independent of pulse sequenceparameters exceptworse on gradient echo vs spin echo

saturation pulses

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


23

Slide 16Homogeneity Artifacts

Shim Misadjusted

SE GRE

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 17Homogeneity/Susceptibility Artifact

Susceptibility

Homogeneity

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 18 Metallic Artifacts - SE Spine ___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


24

Slide 19Metallic Artifacts - FSE Spine

SpatialSat

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 20Metallic Artifacts - SE Spine

WireRemoved

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 21Metallic Artifacts SE Knee

Coronal

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


25

Slide 22Metallic Artifacts FSE Knee

CoronalFat Sat

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 23Metallic Artifacts FSE Knee

SagittalFat Sat

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 24

Gradient Artifacts

Geometric DistortionSimilar to Metallic Artifact but usually

over FOV

Direction DependentIn plane Frequency vs. Phase

Slice profile distortion

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


26

Slide 25Geometric Distortion

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 26

Phase Artifacts

Gradient effectsSpatially dependent SNR loss

GhostsDifferentiate from motion

Non gradient effectsPhase Modulation Ghosts

DC offset

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 27 Phase Modulation Artifact

Weak Ghost displaced in PE directionPeriodicity dependent upon synchronization with TRLine voltage, harmonicsGradient instability

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


27

Slide 28EPI 1/2 FOV Phase Artifact

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 29Phase Contrast Venogram

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 30 Gradient Instability Artifact

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


28

Slide 31

Data Acquisition Errors

Data corrupted during digitization before FFT(eg electrical transient)

Bad memory locations

Pattern depends on what data points aremodified

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 32Data Acquisition Errors

Modifying a fewpoints during digitization of asingle echo.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 33Signal Corrupted During Scan

Similar to a data acquisition error, the signal may bemodified by discrete events during any part of thepulse sequence. Their effect upon the image dependsupon when and how many events occur.

ν ν

Not atisocenter

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


29


ν

Noise appears to be phase incoherent at specific frequencies onupper right.

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


ννAxial Sag

High frequency or high contrast residual image of legTwo dimensional pattern

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Slide 36Artifact Summary

Artifact vary with imaging conditionsPulse sequence details, coil, patient

RF ArtifactsExternal noise phase incoherent, single pixelOther artifacts not along orthogonal directionsPhase modulation - ghosts in PE direction

Gradient ArtifactsInstabilities - Ghosts in PE directionDistortion - direction dependentSeverity, appearance direction dependent

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________


30

Slide 37

Acknowledgments

Carl Gregory Biomedical Magnetic Resonance Laboratory

University of Illinois, Urbana, Illinoishttp://bmrl.med.uiuc.edu:8080/~cgregory/notebook/artif

acts.html

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

___________________________________

Documents

Quality Assurance of MRI Systems David Hearshen Slide 1 ... · Slide 2 Quality Control of MRI ... subject to variability from: ... Occurs when the RF center frequency is off resonance