:
Spectral Doppler
and Color Artifacts1
Myron A. Pozniak, James A. Zagzebski, KatbleenA. Scanlan,
MD PhD MD
Artifacts in spectral and color Doppler imaging can be confusing and lead to misinterpretation of flow information. The authors review these artifacts by considering three main causes: inappropriate equipment settings, anatomic factors, and physical and technical limitations of the modality. Incorrect gain, wall-filter, or velocity scale settings can cause loss of clinically important information or distortion of the tracing. Reflection of the Doppler signal from highly reflective surfaces can create a color Doppler mirror image. Vascular motion can introduce artifactual variation in velocity as the sample volume passes through different velocities in a laminar flow state. Unintentional motion can cause a generalized Doppler shift. Increasing the angle of Doppler interrogation degrades the quality of the tracing and gives the impression of spectral broadening. As angulation approaches 900 directional ambiguity can occur, suggesting bidirectional flow. Grating and side lobes can interrogate areas unrelated to the sample vol. ume and introduce extraneous Doppler information to the apparent area of interrogation. Recognition of these artifacts is essential to proper interpretation of Doppler information and rendering a correct diagnosis. INTRODUCTION
U
Doppler evaluation ence or absence studied with this flow characteristics, interpretation of nical factors that
of hemodynamics provides useful clinical information. The presof flow in a vessel, flow direction, pulsatility, and velocity can all be modality. In addition, color Doppler provides a visual image of the profiling regions of turbulence. With both modalities, correct tracings and flow images requires knowledge of physical and techinfluence Doppler signals. Artifacts caused by physical limitations
of the modality or inappropriate equipment settings can result in displayed flow conditions that may differ considerably from the actual physiologic situation. Understanding limiting factors and appropriately modifying instrument settings should minimize misdiagnosis. In this article, we review artifacts in spectral and color Doppler imaging. We have grouped
Index
them
terms:
I
From
ent ofa requested
1992;
the
RSNA,
(US),
Certificate August
categories:
artifact
those
#{149} Ultrasound
(US),
resulting
Doppler
from
inappropriate
equipment
studies
12:35-44
Department
(M.A.P.,J.A.Z.,
Measurements, ‘
three
Ultrasound
RadloGraphics
53792
into
of Radiology,
K.A.S.), ofMerit 23 and
Middleton,
and the
University Department
ofWisconsin ofMedical
Hospital
and
Physics,
University
for a scientific exhibit at the 1990 RSNA scientific received September 24; accepted September 26. Wisconsin.
Address
reprint
requests
Clinics,
600
Highland
ofWisconsin,
assembly. Supported
Ave. Madison
Madison, (J.A.Z.).
WI Recipi-
Receivedjune 10. 1991; revision in part by a grant from Radiation
to M.A.P.
1992
35
b.
a.
Figure
1.
Images
sonographic
were
gray-scale
generated
of a common
phantom.
Transducer
moved back and forth across the phantom ate a Doppler shift. (a) With proper gain clear
distinct
velocity
manifests
envelope
in a uniform
der the systolic
is displayed.
direction
peak
was
to genersettings, a
with
indicating
Flow
a window
a narrow
un-
range
of
velocities at any point in time. (b) With a moderate increase in gain, the velocity tracing is thickened, that is, the spectrum broadens (*). Systolic window becomes filled in (**). Additionally, a mirror image tracing that suggests reversed flow is also displayed (arrow). (c) Further increase in gain produces
spikes
over
the tracing
is overwhelmed by the meters per second.
settings,
and
those
artifacts
of the
cal examples imaging and U
the detection
higher
velocity
arising
related
limitations
where
from
anatomic
to physical modality.
tive
factors,
and
technical
Phantom
are shown in both spectral Doppler.
DOPPLER
circuitry rn/s =
shifts.
and
color
clini-
flow
SIGNAL
Doppler processing detects phase or frequency shifts in returning echo signals when there is relative motion between scatterers and
the
ultrasound
transducer. The Doppler 2f)Vcos(O)/c, gives the magnitude of the frequency shiftJ) when reflectors are moving with velocity V and angle 0 relative to the ultrasound beam.f, is the ultrasound frequency, and c is the speed of sound in the medium. In addition to an audio presentation, frequency (or spectral) analysis is presented as a tracing of Doppler frequency disequation,f,)
tribution presented
above
36
U
RadioGraphic-s
=
versus
time.
in terms
equation
U
This
display
of velocity
for V. Direction
Pozniak
Ct
a!
by
may
also
solving
the
of flow
rela-
be
to the
transducer,
flow
velocity,
and
im-
portant flow conditions such as turbulence are inferred from this display. Most color flow instruments derive a flow image by applying Doppler signal processing and detection to the echo signal waveform and then estimating the mean Doppler frequency at multiple locations along each beam line (1). Different methods are used by manufacturers for this estimation. All methods require a series ofperhaps eight to 15 pulseecho sequences along each beam line to obtain sufficient data for estimating the mean Doppler frequency within each pixel. This prolonged data acquisition time reduces the frame rate below that of conventional B-mode imaging and introduces trade-offs between color image quality, image size, and temporal resolution. The color image portrays mean flow velocity and direction relative to the transducer for pixels throughout the targeted field.
Volume
12
Number
1
a. Figure the
(a) Color
portal
tained
U
2.
vein with
the
Doppler
represents color
ARTIFACTS
image
the scale
hepatic
adjusted
RELAThD
INAPPROPRIATE
of a portal artery.
b. fails to show
vein False
downward
diagnosis
(arrow)
TO
Doppler Gain Setting Errors Proper gain setting in B mode is critical to an accurate image; it is also critical in Doppler processing for accurate depiction of flow characteristics. If the gain setting is too low, valuable flow information may be lost. The sonographer frequently adjusts the gain to maximize visualization of the spectral tracing. A gain setting that is too high, however, degrades the velocity envelope on the spectral This
mimics
spectral
broadening
At higher
1992
Scale
velocities,
spikes
display
detectability
setting.
tions
of signals.
is
the
ically
of
it
ing
rate
(ie,
the
Doppler
also
increases
veloc-
is set
too
The
phenome-
or
spectral
angle
(which
the pulse
aliased
tracing
effect.
increasing the
This
that spectral Doppler sound beams. If the signal exceeds the onehalf the pulse
a wraparound
shift),
by the
ambiguous
produced.
pler
example,
range
occurs.
frequency), shows
For
velocity
aliasing
sampling
are
ob-
states might not be displayed; a patent vessel may appear (Fig 2). Conversely, when the is set too low for the flow condi.
present,
signals
to
Image
Aliasing
is affected
non is related to the fact and color flow use pulsed frequency of the Doppler Nyquist
and
Errors
When
high, low-flow for example, thrombosed velocity scale
adjacent
(b)
Doppler and color flow imagscale setting is crucial to a
low-flow
repetition
noise project across the tracing and degrade (Fig 1). Gain setting errors in color Doppler are usually obvious. It is appropriate to set the color gain by turning it up until noise is encountered and then backing off until the noise just clears from the image. When the gain setting is too high, the image becomes cluttered with color noise in a random pattern. When the gain is set too low, Doppler shifts are not displayed in vessels, especially those with relatively slow flow.
January
Velocity
Color
considered. flow.
successful ity
and
gives the impression, for example, of poststenotic turbulence. A mirror image spectral tracing can occur if the directional circuitry overwhelmed.
normal
In both spectral ing, the velocity
.
display.
the lumen. was
demonstrates
.
SETfINGS
flow within
of thrombosis
typ-
By increas-
decreases
velocity
Dop-
scale
repetition
(which
frequency),
changing the baseline setting, or using a lower ultrasound frequency, aliasing can be avoided (Fig 3). Color aliasing projects the color of reversed flow within central areas of higher laminar velocity. The key to perceiving color aliasing with no tween
most black
equipment stripe
the
reversed
from
is the
fact
that
the
low-velocity
colors
(Fig
Pozniak
Ct
4)
(2).
a!
U
there
is
filter
be-
RadioGraphic-s
U
37
Figures
and setting, within
3, 4. the higher
(3a) On Doppler image obtained with the velocity scale set too low (arrows), aliasing occurs, velocity systolic peaks wrap around to project below the baseline. (3b) By adjusting the scale systolic component is correctly displayed. rn/s = meters per second. (4a) Blue and green colors
the the central
portion
of the
vessel
suggest
flow
reversal,
but
they
are
actually secondary proper direction.
the higher central velocities. Slower velocity near the wall is displayed in the the wraparound occurs from the top to the bottom of the scale bar and does motion filter (arrow in a); therefore, directions. This is in contrast to the scale, flow direction is now correctly
#{149} Incorrect
Wail-Filter
no black
image
with
stripe
true
U
RadioGraphic-s
U
Pozniak
Setting
Ct
a!
between
flow reversal
the
not go across
display
the (black) wall. different color By adjusting the color
ofthe
(cf Fig 10). (4b)
two
displayed.
Filtration is designed to remove unwanted low-frequency Doppler signals originating from slowly moving soft-tissue reflectors. The cutoff frequency is operator selectable. With filtration set too high, diagnostically significant velocity information can be lost (2). This is especially true in low-flow states and in the measurement of resistive index (Fig 5).
38
is noted
to aliasing of With aliasing,
U
ANATOMICALLY
RELATED
ARTIFACTS #{149} Mirror
Image
Artifact
Mirror image artifacts are commonly observed in conventional B-mode imaging (3). A similar artifact can occur with color Doppler imaging of any vessel adjacent to a highly reflective surface, such as the lung (4). The subdiaphragmatic region of the liver and the supraclavicular region are most notorious for this artifact (Fig 6) (5,6).
Volume
12
Number
1
Figure 5. Spectral Doppler tracing ofa normal arterial waveform. As filtration is increased across this tracing from left to right (arrows), the lower velocity components become obliterated. Increased filtration can eliminate low-velocity diastolic flow and yield an impression
of higher
flow
resistance.
rn/s
= meters
per second.
Figure 6. (a) Image of the supraclavicular region shows two apparent subclavian veins. The more anterior vessel is the true vein. The deeper vessel below the pleura (arrows) is actually an artifactual mirror image reflecting off the lung apex. (b) Longitudinal image of the inferior vena cava. A second vessel is perceived deep to the true inferior vena cava (arrows). Doppler shift is more prominent in the mirrored vessel. As the sound beam reflects from the diaphragm, it interrogates the inferior vena cava flow at less of an angle, thereby undergoing a greater Doppler shift than when it interrogates the true vessel, which is more perpendicular to the ultrasound beam. (c) Schematic shows the course of the true signal (large arrows) and the mirrored signal (small arrows). Increased distance that the mirrored signal travels causes the phantom vessel to project deeper in the image.
January
1992
Pozniak
et a!
U
RadioGraphic-s
U
39
Figure ing
7. was
(a)
Normal
obtained
portal
in a normal
vein
flow
volunteer
is usually during
at a uniform
velocity
suspended
respiration.
toward With
the
liver.
a small
This
sample
portal
vein
volume
trac-
centrally
positioned in the vessel, the spectral Doppler tracing reveals a pulsatile waveform. (b) With Doppler sample volume slightly enlarged, orientation of the transducer relative to the vessel altered, and sample volume positioned more laterally in the vessel, pulsatility has almost completely resolved. This is due to decreased ex cursion between areas of differing velocities within the vessel and a larger cross-sectional sampling. rn/s = meters per second. (c) Pulsatile flow in a flow phantom was generated. Image captures the onset ofthe pulse
waveform across the interrogated region. Propagating wave assumes a parabolic form revealing the laminar flow profile. (d) Positioning a velocity tag (green) in the middle velocity range demonstrates laminar profile of flow in the portal vein. Slowest velocities adjacent to the vessel wall are darker red; middle velocities are green;
and
highest
velocities,
#{149} Vascular-Motion
which
are
artifact
can
pink,
Artifact
When a vessel that is moving relative transducer is interrogated, artifactual ity can be introduced into the spectral This
light
be
perceived
in the
to the pulsatiltracing. portal
vein and its branches. Transmitted cardiac contraction pushes on the liver and tugs on the inferior vena cava. This rocking action of the liver causes slight motion of the vessels with respect to the Doppler sample volume. Even a small degree of motion is sufficient to displace the sample volume from the highervelocity central laminar flow to the slower-
40
U
RadioGraphic-s
U
Pozniak
Ct
a!
are
found
centrally
within
the
vessel.
velocity peripheral flow and back. This can introduce or increase periodicity of the portal venous flow pattern, an appearance suggestive of, for example, tricuspid regurgitation. Increasing the size of the Doppler gate to inelude the entire vessel, imaging other portal branches, or varying the angle may decrease this artifact (Fig 7).
#{149} Color
in Nonvascular
Structures
An area of low echogenicity such as a cyst or a duct is susceptible to color flash artifact. Any motion of a reflector relative to the transducer produces a Doppler shift. Most color flow processors incorporate motion discriminators that separate true flow from random
Volume
12
Number
1
Figure 8. Image of a gallbladder obtained suspended respiration. Transducer motion duced color flash artifact in the image, but tially
Figure
9.
(a) Portal
vein flow 90#{176} perpendicular
dominates
to the Doppler
the
beam.
hypoechoic
Sample
during introit
preferen-
areas.
volume
was placed
over that
portion of the vessel where it passes perpendicular to the interrogating beam. Doppler shift is perceived; however, it is displayed equally above and below the spectral baseline. (b) By altering the angle of insonation away from 90#{176}, the direction of flow is correctly displayed and the velocity envelope is better defined. rn/s = meters per second.
motion lower
of soft-tissue level signals
soft-tissue
regions
reflectors. arising from less
effectively
However, hypoechoic trigger
the the
motion discriminator and do not suppress the resultant color flash. Even minute motion may produce artifactual color signals within dilated bile ducts, cysts, or the gallbladder (Fig 8). This can be erroneously perceived as flow, especially if the color sensitivity settings are high (7). U
INSTRUMENT-
PROCESSOR-RELATED #{149} Directional Directional interrogating 90#{176} angle.
January
AND
ARTIFACTS
they usually and below gain
settings,
manifest the spectral
as a tracing baseline.
directional
both above At higher
ambiguity
is worse
and the velocity envelope on the spectral tracing becomes more indistinct. The ambiguity is easily
corrected
when
the
interrogating
direction is shifted to an angle of 90#{176} (Fig 9). When the sample sitioned
on
a sector
image,
beam
at either volume
care
should
side is pobe
taken to avoid that portion of a vessel that intercepts the interrogating beam at 90#{176}. With a linear transducer, electronic angulation or wedge standoffs can improve the angle of spectral Doppler interrogation (8).
Ambiguity ambiguity can result when the beam intercepts the vessel at a If Doppler signals are detected,
1992
Pozniak
Ct
a!
U
RadioGraphic-s
U
41
.
phantom. minimal
Sector Direction
angulation
color offlow
b
is right
off the
beam
image of a to left. Only axis
relative
to the
direction of flow is required to accurately display the color flow profile. Narrow colorless area in the lumen of the vessel represents absence of Doppler shift
where
flow
is perpendicular
to the
insonating
beam. This is true reversal of flow relative to the interrogating beam. Aliasing does not produce a black area between color display of alternate color direction (cf Fig 4a).
Figure
11.
ing and side main beam.
Schematic lobes
of grat-
relative
to the
(Fig 1 1). Side lobes occur in proximity to the primary beam. Grating lobes, however, can be quite far removed from the central beam (9). The exact configuration of these lobes de-
(Fig 10). With a linear transducer, this becomes a more significant problem, but again, electronic beam steering or wedge standoffs can compensate.
pends on the construction of the transducer, that is, crystal element size and spacing. Tightly curved, convex-array transducers are most susceptible, as are higher frequency linear transducers. If these off-axis lobes of sound strike a highly reflective surface, they can return to the transducer and cause misregistration of an object on the display of the primary beam. This artifact commonly confuses needle biopsy guidance (10, 1 1). Off-axis lobes can interrogate vessels unrelated to the Doppler sample volume. If a strong enough signal is encountered off axis, it may appear on the spectral Doppler tracing,
#{149} Grating-
or Side-Lobe
Electronically
Artifact
focused, phased-array transducers concentrate the primary interrogating beam toward the Doppler sample volume. Because of the spacing of the array elements and the sequence of firing, however, weak secondary lobes of focused sound may interrogate areas unrelated to the primary beam
displaying a Doppler pected (Fig 12).
U
the location sound
When a color Doppler image is produced with a sector transducer, flow perpendicular to the beam usually occurs over a small segment of a vessel parallel to the transducer surface
42
shows
of focused
RadioGraphic-s
U
Poznlak
et a!
shift
where
Volume
none
12
is ex-
Number
1
Figure 12. (a) Transverse trix ofa Doppler phantom, rogating
the
vessel
and
image ofa far removed
displays
Doppler phantom. Cursor is interrogating from the flow in the tube. A grating lobe,
Doppler
shift
on the
spectral
tracing.
(b)
an area
in the gelatin
however, longitudinal
In the
succeeds plane,
ma-
in interthe true
Doppler sample volume interrogates flow traveling toward the interrogating beam. The grating lobe, however, registers flow in the opposite direction as it interrogates an area completely separate from the primary sample volume. (c) High-resistance tracing of the main renal artery in a transplant recipient. (d) This tracing presumably represents a visceral artery interrogated by a side lobe. As the transducer is manipulated to obtain
a better
Doppler
tracing,
sample
a low-resistance
volume
#{149} Spectral-broadening Spectral broadening tics
identified
in stenotic
flow
is squarely
Artifact of the
is one
flow
pattern
in the renal
appears
hilum.
occurs
in pulsed
and
patterns.
There-
continuous-wave
Doppler because of the finite dimensions of the sample volume. A target traveling through the sample volume at a given speed introduces a range of Doppler frequencies centered about the “true” Doppler frequency (ie, the Doppler frequencyf0 calculated with the equation mentioned previously) (12). As the
January
1992
in the background (arrowheads); however, = artery, rn/s = meters per second.
target characteris-
fore, it is an important clinical sign. However, spectral broadening may be introduced artifactually and can cause errors in interpreting peak velocities in spectral waveforms. The artifact
ART
pier
traverses angle
different sition
the
varies.
Doppler of the
target
sample
This
frequency and
volume,
introduces
the
the
Dop-
a slightly
shift
hence
for each
po-
a broadened
spectrum.
Spectral broadening is dependent on the angle of insonation (Fig 1 3) and the velocity of the blood. Spectral broadening generally becomes more severe for angles closer to 90#{176}. Thus, errors can be minimized by using sound-beam pler angle
orientation of less than
that results 60#{176}.
Pozniak
Ct
a!
in a Dop-
U
RadioGraphic-s
U
43
13. (a) Doppler spectral waveform for a flow phantom obtained with the Doppler beam angle at 71#{176} relative to flow direction. (b) Waveform obtained with a Doppler angle of8l#{176}.Note broadening of the spectral tracing, apparent increase in peak velocity, and partial filling of the systolic window for the steeper (8 1#{176}) angle. VEL = velocity, rn/s = meters per second. Figure
Trans
CONCLUSIONS
U
In B-mode imaging, changing the settings affects image quality. Altering the settings may enhance or camouflage important clinical findings. Both spectral and color Doppler imaging are also acutely sensitive to technical settings. Most important, the display of flow information from low-velocity states is dependent on sensitivity,
transmit
focal
scale, and filtration and spectral Doppler must be constantly locities
and
distance,
of insonation
color controls to ye-
if flow
condi-
tions are to be accurately demonstrated. Operators must also be aware of distortions of flow information such as mirroring artifacts that are due to unusual pathways of reflection and transmission in the body. Use of alternative acoustic windows can help identify such conditions. Finally, flow information can be distorted as a result ofequipment limitations. Examples presented include artifacts from transducer side and grating lobes and intrinsic
spectral
broadening.
avoided
when
limiting
factors.
the
Misdiagnosis
operator
can
is aware
be
3.
5.
Color
U
RadioGraphic-s
BJ, Winsberg
princi-
Radiology F.
Echographic
Reading CC, Charboneau JW, Allison Cooperberg PL. Color and spectral mirror-image artifact of the subclavian Radiology 1990; 174:41-42.
7.
Mitchell
DG,
Doppler
artifact
sound Parvey
8.
Burns
P, Needleman
in anechoic
JW, Doppler artery.
L.
regions.
Med 1990; 9:255-260. HR, Eisenberg RL, Giyanani
Color
J UltraV, Krebs
CA. Duplex sonography of the portal venous system: pitfalls and limitations. AJR 1989; 152:765-770.
9.
of these
ZagzebskiJ.
Doppler
Hagen-Ansert
S, ed.
ultrasonography.
Laing sonic
We are grateful to Carrie of the manuscript.
Namekawa
K, Koyano
Real-time
two-dimensional
ing using
an autocorrelation
U
imaging:
5: 227-237.
instrumentation. Textbook
St Louis:
In:
of diagnostic Mosby,
1989;
76-
A, Omoto
blood
flow
technique.
Pozniak
et a!
FC, Kurtz side.lobe
AB. The importance artifacts. Radiology
of ultra1982; 145:
763-768.
R. imagIEEE
12.
.
Laing FC. in clinical
Commonly ultrasound.
1983;
4:27-43.
Evans
D, McDicken
cockJ. mentation York:
44
Doppler
and artifacts.
6.
REFERENCES C,
DG.
Frequency
appearance of the right hemidiaphragm. J Ultrasound Med 1983; 2:243-249. Middleton WD, Melson GL. The carotid ghost: a color Doppler ultrasound duplication artifact. J Ultrasound Med 1990; 9:487493. Kremkau 1W, Taylor KJW. Artifacts in ultrasound imaging. J Ultrasound Med 1986;
4.
11 Kasai
Mitchell
Lewandowski
10.
1.
32:458-464.
92.
Acknowledgment: Poole for preparation
U
Ferroelectrics
1985;
pIes, limitations, 1990; 177: 1-10.
velocity
settings with both imaging. These fine-tuned relative
angles
2.
Ultrasonics
Control
encountered artifacts Semin Ultrasound W, Skidmore
R, Wood-
Doppler ultrasound physics, instruand clinical applications. New
Wiley,
1989;
87-105.
Volume
12
Number
1