For each, included:

- Name of Artifact
- Appearance (+/- image)
- Explanation
- How to fix it OR clinical importance

Ultrasound:

Beam Related Artifacts (assumes all from main beam)

Side Lobe Artifact Incorrectly placed echoes Off axis low energy Recognize
overlap anechoic beams (side lobes), “pseudosludge” / artifact,
(From side lobes OR structures received by transducer -- seen more with linear
grating lobes - which are assumed to be from main transducers
another form of off axis beam (when strong
energy) reflector outside of main
beam).

Side lobe energy

generated from radial
expansion of peizoelectric
crystals, worse with linear
transducers

Beam Width Artifact Classic is peripheral In far field, diverging Adjust the focal zone to
echoes in bladder beam beyond margins of area of interest, place
transducer -- return and transducer at centre of
assumed to be from main image
beam

Artifacts Associated with Multiple Echoes (assumes 1 reflection)

Reverberation Artifact Multiple equidistant linear Two closely placed

reflections parallel reflectors -
repeatedly reflect off each
other → multiple
sequential echoes placed
at increased distance
from transducer

Comet Tail Artifact Tapering Triangle Form of reverb, with Small crystalline / calcific
objects placed very close -- adenomyomatosis,
 together (space less than small calculi (renal, cbd,
½ SPL, i.e. less than the pancreatic, testicular
axial resolution) → microlithiasis, surgical
displays as triangle rather clips), biliary hamartoma,
than linear lines due to thyroid colloid nodules
attenuation of later
echoes

Ring Down Artifact Linear, continuous deep Tetrahedron of bubbles Due to GAS
to structure with small fluid trapped
within → nearly
continuous sound wave
deep

Mirror Image Artifact Mirror image at liver lung Reflected between Classically shown at liver
most commonly reflector and adjacent / lung interface
structure -- incorrectly
thought to be another
copy deep to the reflector

Artifacts related to Velocity Errors (assumes 1540 m/s through everything)

Speed Displacement Discontinuous displaced Speed of sound slower in

Artifact diaphragm fat relative to liver -- so
diaphragm appears
deeper when behind fat

Can also occur through

fluid (will be closer than
normal diaphragm)

Refraction Artifact Duplication of SMA, IVC. Refraction at tissue

OR can be displaced or interface results in
appear wider than incorrectly placed beam
normal. Classically shown
deep to rectus
 Attenuation Errors

Shadowing Weaker beams distal to a Material attenuates Gas (dirty), Calcium

structure sound to larger degree (clean), Fibrous tissue,
than surrounding tissue Fat
(distal appears weaker)

Increased through Stronger beams distal to Structure attenuates Often cystic, but some
transmission a structure beam less than solid (i.e. lymphoma, FA)
surrounding structures can demonstrate
posterior acoustic
enhancement

Doppler Related

Aliasing High velocities displayed Doppler shift frequency Increase PRF, increase
as negative on spectral, greater than NYQUIST scale, decrease doppler
blue and red adjacent to FREQUENCY. shift (i.e. lower freq
each other on colour (NYQUIST LIMIT in khz transducer, increase
= ½ x PRF). Aliasing angle closer to 90
occurs if v > ½ PRF degrees), lower baseline
(pulse repetition
frequency)

I.e. wheel example (PRF

= freq of viewing. If PRF
is lower then can’t tell
direction (wheel looks like
it’s going backwards)

Tissue VIbration Perivascular soft tissue Classically with AV fistula

colour of kidney or
pseudoaneurysm

Twinkle Rapid fluctuation of red Multiple strongly Higher sensitivity for

and blue colours (noisy reflecting surfaces like small stones than
spectrum) calcs -- noisy spectrum shadowing, but highly
dependent on machine
settings. Not SP though.
Can increase FL deep to
it to increase the twinkle
artifact

Color Bleed Color extends beyond Decrease colour gain --

vessel wall can miss non-occlusive
thrombi
 Others: Flash Artifact, Flash - burst of colour Mirror image = same as
Mirror Image, Color from motion (i.e. fetal earlier described but for
comet tail kick) -- no spectral colour. colour as well

CT Artifacts:
Types of artifacts include: Streaking (inconsistent single measurement), Shading (group of views deviating
gradually from the true measurement), Rings (detector malfunction), Distortion (helical reconstruction).

Physics Based Artifacts

Beam Hardening Cupping (darker Xray passes through an *Filters - pre-harden

centrally, since the beams object, lower energy beam +/- add bow-tie
that reach the centre are photons removed and left filter to pre-harden edges
harder than those that with ‘harder beam’ with of beam @ thinner parts
reached edge) increased average Calibration Correction -
energy ⇒ Cupping, Dark detector compensates for
Bands and Streak hardening effects (use
phantom to calibrate)
Correction Software
(iterative correction
algorithm)
Dark Bands/Streak (dark Avoidance (tilt gantry to
bands between two avoid areas that cause
dense objects - i.e. hardening)
posterior fossa) -- Xrays First three via
pass through 1 less atten manufacturer, last is
than pass through both operator

Partial Volume Partial volume effect Two densities in a voxel Decrease voxel size (i.e.
(intermediate density in a → averaged to an thinner slices, larger
voxel) is basic. But can intermediate density. Can matrix, smaller FOV). If
also affect in other ways also cause shading noise is a problem can
artifacts adjacent to acquire thin slices and
i.e. dense object lying off dense objects generate thicker by
axis --> shading artifact adding together.
Dense Object off axis
protrudes into the
diverging beam on one
side (but not the other) --
will result in inconsistent
measurements → artifact
Photon Starvation Streaks between dense High attenuation areas → Automatic Tube Current
objects and near metal photon starvation/ Modulation (increased
insufficient photons deep dose through dense
(very noisy due to photon areas -- by increased mA
starvation → magnified at appropriate times only)
on reconstruction → Adaptive Filtration -
(Aside: Adaptive horizontal streaks) smooth the data --
Filtration) software correction to
reduce streaking (see L)

Undersampling View Aliasing = Undersample (too large View Aliasing - increase

(Aliasing) undersampling between interval between projections per rotation
projections = misregister projections) → (slow rotation speed)
= fine stripes from edge misregistration
of dense object, but Ray Aliasing -
distant from it Specialized high
resolution techniques
(increase samples per
projection)

Ray Aliasing = under

sampling within a
projection = stripes
appear close to a
structure

Patient Based Artifacts

Metal Artifact Severe Streaking artifact Several mechanisms: Remove metal (or avoid
beam hardening, partial on imaging if possible)
volume, aliasing, too high
density to be handled. Increase kVP (sometimes
works), Use thinner slices
Metals with high Z (Fe, (to reduce volume
Pl) have more than lower averaging)
Z (Ti)
Interpolation software
(metal artifact reduction
software, beam
hardening reduction)

Patient Motion Misregistration → Reduce voluntary

 shading, streaking motion (tie them down,
GA for kids)
Faster scanner
Align scanner in primary
direction of motion - i.e.
beam is parallel to motion
rather than perpendicular
Overscanning (most
variation btw beginning
and end of scan,
overscan 10% -- last 10%
is averaged to give less
motion artifact)
Gating, Software
Correction

Incomplete Projection Streaking/Shading Parts of patient hanging Position patient correctly

outside of field but still (i.e. arms above head for
attenuating Xrays → chest CT)
messes with computer
math

Scanner Based Artifacts

Ring Artifact Repeated ring Calibration error or Recalibrate detector, or

appearance defective detector replace detector

Helical and Multisection CT Artifacts

Helical Artifact in Axial Distortion @ skull, For helical scanners, if Reduce variation in Z
Plane: Single Section diaphragm (where scanned a cone would be direction (low pitch)
anatomy changing rapidly distorted and not perfect
in Z direction) + worse at circles given path of 180 degree interpolator if
higher pitches helical scanner. See this this is an option
at top of skull and
diaphragm/liver as well Thin sections
Axial scanning (head
CTs often done this way
still)

Helical Artifact in Windmill appearance Due to interpolation Z filter to reduce severity

Multi-Section where several rows of process. Worse with
detectors intersect higher pitch. Complicated
multisection artifact
Cone Beam Effect Streak/distortion? Wider collimation → cone Cone Beam
beam rather than a fan / Reconstruction
flat beam. Artifacts
greater at outer detectors,
similar concept to partial
Outer v inner det. row volume off axis

Multiplanar: Stair Step Stair step at edges of With wide collimation of Thinner slices! (mostly
reformats (i.e. skull) non-overlapping intervals. gone in modern
Less severe with helical scanners), less with
scanner with overlap helical

Multiplanar: Zebra Faint stripes most seen Helical interpolation

Artifact on 3D reformat process (increases noise
along Z axis) -- due to
noise inhomogeneity
along Z axis. Worst off
axis because noise is
worst there (peripherally)

MRI Artifacts:

Image Process Artifacts

Aliasing (wrap-around) Wrapping of anatomy Occurs in phase Make FOV bigger (cost =
from under-sampled encoding direction (note poor spatial resolution)
portions if 3D, there are 2 phase Flip PE/FE dir (cost =
encoding directions) increased table time)
Surface coils (minimizes
Question may ask - which signal from tissue
direction is phase outside)
encoding? Look for Sat bands (eliminate
 aliasing as a clue signal from outside FOV)

Chemical Shift Bright on one side, dark Different precessional Increased bandwidth
on the other (@ Fluid-Fat freq of fat and water is (rBW) (Decreases shift)
Interface) 220 Hz at 1.5T → Swap PE/FE if helps
misregistration in the Fat Suppression (STIR)
“Type 2” is india ink frequency encoding
artifact -- only occurs with direction
GRE -- to fix just change
TE to in phase

Partial Volume Volume averaging Overlap in signal voxel Make voxels smaller
(decrease FOV,
increased matrix,
decrease slice thickness)

Truncation/Gibbs Repetition of edges / Truncation of amount Decrease transmit BW

ringing @ edges sampled in K space (not or pixel/voxel size
enough of peripheral K (increased matrix,
CSF-cord interface is space) → decreased decreased FOV) -- i.e.
most classic (mimics resolution AND gibb’s increase the resolution by
syrinx) - on sag spine ringing artifact. Sharp sampling outer K space!
edges are
Seen in both FE and PE misrepresented Penalty = longer
direction, but more (RINGING) = ‘ripples in acquisition time, lower
common in PE direction data’ SNR

Patient Related Artifacts

Motion Artifact Ghosting, Smearing (in Voluntary and involuntary Stop voluntary motion
phase encoding direction movements Breath Holding
most frequently) Cardiac or Resp Gating
Phase encoding Fat Sat Band (over abdo)
Pulsation artifact (phase direction (most Switch PE/FE if possible
encoding direction) - see frequently but can occur BLADE/PROPELLER -
next as well (flow) in both) radial sampling methods
like helicopter to
Pic shows ghosting oversample centre of k
artifact with head space
movement Faster acquisition (i.e.
FSE)

Flow Ghosting in phase On SE - no or slow flow Apply saturation band

encoding direction will produce an echo -- adjacent to imaging
(pulsation). On SE looks fast flow will be signal section (90 degree →
dark (from 90 → 180 void spoiler / crusher gradient)
degree pulse it moves so = eliminate signal from
no signal), on GRE looks above and below section
 bright (inflow along Z axis
while others recovering
towards)

Magic Angle Signal on T1 and PD, @ 55 degrees with main Higher field strength
GRE at 55 degrees along magnetic field on SHORT (greater shortening of T2
tendons TE sequences (T1, PD) -- times)
often with tendons
Long TE sequences (T2)

RF Related

Cross Talk Band of signal loss Pulses are not Increase gap btw
horizontal across the rectangular (RF, Fourier sections
image! (i.e. axial lower Transform/FT pulses) → Interleave slices (odds
lumbar spine!) can get excitation of then evens)
neighboring sections 3D images don’t have this
leading to partial - entire volume excited
saturation and lower
signal

Zipper Artifact Zipper of high signal Random/Stray RF signals Closing door if open
across phase encoding → zipper of high signal Remove all electronic
direction (1-2 pixels in across image in phase devices from room (pulse
width) encoding direction. ox left on is classic
scenario)
Often hardware or Repair faulty RF
software issue, seen in all shielding (call tech
image series people)

External Field Related (Inhomogeneity) / Susceptibility

Inhomogeneous Fat Inhomogeneities -- failure Field inhomogeneities → Use STIR, especially with
Suppression in fat sat mimics edema fat precesses at different metal
frequencies → areas can
resist chemical fat Aside: Shimming is
suppression process to improve field
homogeneity in a magnet
when buy it and btw pts
(p 302 physics - passive
when buy and active btw
patients)

Susceptibility (including SIGNAL LOSS Substances become Use SE/FSE

metal) Bright-Dark region due magnetized by external Decreased field
to misregistration field → affects local strength
Geometric distortion magnetic field Higher rBW (can
decrease TE this way)
Aside: Longer TE, more Thin slices
DIAmagnetic = slightly susceptibility (T2*) Align long axis of metal
negative susc (most, implant with axis of
including water) main field
PARA = s+ (blood, gad,
etc)

Gradient Related

Eddy Current Distortion (contraction or When gradients are Optimize the sequence of
dilation of image, rapidly turned on and off gradient pulses
stretch/smear) (i.e. EPI) → eddy currents
are generated ⇒
Most severe on DWI distortion (contraction or
(EPI) - i.e. stretch and dilation of image,
smear at brain-bone shift/smear)

Errors in Data

Dielectric Effects / Dark signal central RF pulse wavelength in Worse with stronger
Standing Wave Effects abdomen over left lobe of the body is reduced magnet (reduce magnet
liver (classic appearance) (tissues have a strength to help)
dielectric constant that
Usually @ 3T and in the reduces wavelength). Worse when wider body
abdomen This is worst at higher (i.e. with ascites, large
magnet strengths (i.e. RF bellies)
pulse @ 3T → 200
something cm in air → 26 Apply dielectric pads
cm wavelength in the Parallel RF
body). If the length of the transmission (gives you
body > wavelength in longer pulse)
body → constructive and
destructive interference
result in signal loss
(DARK IN MIDDLE).
Standing waves created
as radio wave
approaches length of
body part.

Herringbone / Obliquely oriented stripes Over-represented areas Re-scan (typically a

CrissCross throughout image in K space -- larger than spurious artifact)
 expected contribution Hardware issue -- call
field engineer

BANDING ARTIFACTS
- With balanced SSFP sequences (now used for abdominal imaging due to fast GRE acquisition)
- Get DEPHASING of Bo from gradients during each TR -- susceptible due to non-uniformity of Bo →
bandlike losses of signal (usually peripherally or near areas with inhomogeneity - but bands are separate
from susceptibility artifact)
- Banding occurs at intervals of 1/TR -- so HIGHER TR = occurs at shorter intervals = MORE BANDING
- Also occurs more at higher field strengths
- To FIX = LOWER TR (half fourier transform), LOWER MAGNETIC FIELD