Donald G. Mitchell, MD
of the
Art
C
the
age
tissue
flow. CD information
and
superimposed
on a real-time
B-
mode gray-scale image (1-5). In response to reports that advocate the use of CD images to quantify cardiac disease (6-9), some technical limitations of CD imaging have been re-
ported
in the
cardiology
literature
(10-13). There has been a paucity of reports regarding technical issues of CD imaging outside the heart, how-
While color saturation, pixel and color distal to plaque have used recently to grade carotid
(14), findings dependent such as these on the angle of
flow
as well
relative
to the and
ultrasound
other
beam,
operatorof this
dependent
dent
operator-indepenpurpose
report
ness
is to raise
regarding
the
the
level
of awareof
determinants
pro-
(transducer), the frequency of the reflected sound will be increased, whereas if the cells are moving away from the source, the frequency will decrease. The Doppler shift frequency is the difference between the frequency of the transmitted ultrasound and the frequency of the reflected ultrasound. This frequency is proportional to the velocity component aligned with the axis of the ultrasound beam (15-17). Since the strength of the Doppler signal is much weaker than that of the low frequency shifts from slowly moving tissue, a high-pass filter (eg, wall filter) is often used to allow only the higher frequency shifts to pass. If this is not done, the resulting clutter might overwhelm the receiver, preventing proper analysis of the Doppler spectrum (5,15,17). With pulsed Doppler, a discrete burst of ultrasound is emitted. After a specified delay, Doppler information is sampled for a brief interval of time. Because the speed of sound is relatively constant in the human body, the location of Doppler shifts can be determined from the delay be-
Figure
Doppler
1.
data
Diagram
for CD
of the acquisition
of
boxes Doppler. each line by the and the is the
imaging. Solid indicate range gates for multigate Several pulses are transmitted for of Doppler data. The rate is defined pulse repetition frequency (PRF), duration of sampling for each line
dwell
time.
With
some
CD systems,
a sepato sysare
rate sweep of one pulse per line is used acquire B-mode information. With other tems, Doppler and B-mode information obtained line by line.
tween
production
of the
sound
and
Index
trasound (US), Ultrasound
terms:
(US), physics (US),
Ultrasound
Doppler
#{149} Ultrasound
(US),
studies (US), characterization
artifact
#{149} Ultrasound
#{149} Ul-
technology
tissue 177:1-10
Radiology
1990;
its detection. In clinical practice, a Bmode US image is usually used to specify the location of Doppler interrogation, the range gate. The size of the range gate is determined axially on the basis of the duration of the ultrasound burst and the sampling interval, and laterally on the basis of the thickness of the ultrasound beam (15-17).
Doppler consists of several range gates along a line. This information can be color encoded and superimposed on a real-time M-mode image, which can depict Doppler shift profiles across vessels or through valves
(4).
Two-dimensional Doppler mation can be obtained by multigate Doppler technique inforusing a to
an area of interest (Fig fast sampling is necesThe most of rapid commonly sampling for
adaptmov-
From Main
Real-Time
Two-dimensional
Floor Hospital
Doppler
ReMay 7.
and
ceived
Sansom
revision
Sts,
27, received
Philadelphia,
1990; requests
PA
4; accepted
19107.
June
If Doppler
information
is sampled
Abbreviations:
pulse repetition
requested author.
23;
during several intervals following each ultrasound pulse, multigate Doppler can be performed. Multigate
CD
frequency.
color
Doppler,
PRF
time-shifted
versions
of itself.
By so
to be considerably
less
than
the
peak
doing, motion can be detected, and an estimate of the mean frequency shift is obtained. The amount of fluctuation is often referred to as vanance and is color encoded on many systems (4,5). There are alternatives to autocornelation for creation of a CD image, but discussion of these
frequency.
The
difference
between
the mean and peak frequencies depends on the range of Doppler shifts within the sample volume. While some investigators recommend measuning peak frequency directly from color images (14), errors are likely to
across the mitral anulus during le) (5). Line-by-line interleaving possible but requires more rapid switching between Doppler and
systois B-
mode
pulses.
It is also
possible
to
methods is beyond the scope of this review. It must be emphasized that the data acquired for most CD images is different from that acquired for Doppler spectral analysis. For CD imaging, a single representative numben, usually the mean or a similar quantity, is assigned to a pixel, while the full spectrum of frequency shifts
is displayed in a Doppler spectrum.
occur if this is done. For most systems, B-mode and CD components of the image are acquired separately, during separate sweeps (5). This allows the use of dedicated ultrasound pulses for Doppler and B mode. It is also possible to steer the ultrasound beam so
that the angle of insonation is differ-
ent
for B-mode
of the three
and
image. to 32,
Doppler
compofor
nents usually
The peak frequency shift at a given time is not directly related to the mean frequency shift or to the hue
intensity of color on CD images. In
or
each line of Doppler information, while one pulse per line is sufficient for B-mode imaging. Therefore, most of the acquisition time is devoted to
constructing may be offset the Doppler image. partially by using This a interof Figure 2. Longitudinal CD image of a tortuous common flow toward the transducer. carotid artery. the transducer; High Doppler Blue indicates red, away from frequencies
the
center
of large
vessels
with
plug
flow, where most flow is at a uniform velocity, the mean frequency is close to the peak. Nonuniform velocity
lower
component polating
line
density
of the the space
for the
image, between
Doppler
thus lines
are indicated
shade). The
by decreased
highest frequencies
saturation
in this
(pale
im-
within areas
cations,
as in bifunvessel
Doppler
Separate color can
data.
sweeps for B mode and yield anomalous images in
age (*) correspond to regions where the angle between the ultrasound beam and the
blood flow is lowest, whereas lower frequencies correspond to flow that is nearly perpendicular to the beam. Where flow is perpendicular (arrows), a complex mixture
miat a
flow
of red and
blue
is observed.
a. Figure 3. relationship
carotid
b. Longitudinal between
the large
C.
duplex CD images of the proximal internal carotid color-encoded peak velocity, highlighted by a green
angle between the common and internal carotid
two
different
patients;
the peak
caused
Doppler
internal
these frequency.
artery
disease
carotid
the ultrasound beam. This results in peak Doppler shifts as high as 10 kHz, as depicted in the Doppler spectral display at bottom. Note that flow velocity within the sample volume is relatively uniform, depicted as a clear window beneath the peak Doppler shift in the spectral display. Thus, peak and mean Doppler shifts are similar. In the image, Doppler shifts at or above 9.7 kHz have been encoded as green (arrow). The baseline has been shifted to the bottom to prevent aliasing. Note that a mechanical wedge has been interposed between the skin
and the just transducer below this to decrease value are the angle between the ultrasound beam and the common carotid indicated artery. (b) In a patient with wrapping angiographic eval-
idence
shifts
of a 90% stenosis
of the proximal
encoded
internal
as pale
carotid
red.
artery,
spectral
Doppler
display
shifts
at or above
site
6.4 kHz
have
been
encoded
at 9.8 kHz,
as green.
Doppler
around
(c) The
at the
in b aliases
most to the baseline. The peak Doppler frequency is thus nearly 16 kHz. Note that the flow disturbance has broadened the spectrum, in the window beneath the peak frequency. For this reason, the mean Doppler shift is significantly less than the peak. 2
#{149} Radiology
filling
October
1990
b.
duplex CD images of a normal brachial artery, acquired with excessive gain and power, show a color mirror-image deep to the brachial artery is encoded red. A spectral waveform is obtained from this region. There is no spurious color elsewhere in the image. (b) With power and gain reduced, color is confined to the vessel. The spectral waveform has higher amplitude than that in a but is identical in other respects.
range
sample
shifts
or variance,
within
a
which
helps across
disturbed or abnormal
with
(4,5). Green may also be used to tag frequency shifts above a designated threshold, highlighting the areas of
most rapid flow. Again, the highest frequency as measured by means of green tagging is less than, and bears an inconsistent relationship to, the peak frequency shift (Fig 3).
a.
Figure 5. Longitudinal CD images of a brachial
b.
artery show the differential features of
flow separation and aliasing. V brachial vein. (a) During early diastole, some reversed flow is noted (arrows). Note that the sequence of color changes at the transition between antegrade and retrograde flow (pale red, deep red, deep blue, pale blue) matches the changes at the center of the color bar at the right. Thus, the center of the blue component of the image has the palest color. (b) During peak systole there is no retrograde flow. Decreasing the PRF (note the decreased Doppler shift range as indicated at the top and bottom of the color bar) has produced aliasing. Note that the sequence of color changes is now pale red, pale blue, and deep blue, indicating wraparound from the top to the bottom of the color bar. Thus, the center of the blue component of the image has the deepest color. With this instrument, the transition between pale red and pale blue at the alias frequency is highlighted by a red line (arrows) that is thinner than the color pixels.
CD Manifestation Mirror-Image
Reflection two strong
a delay,
of the Artifact
B-Mode
which
age Mirror
of the
space
situated
between
beneath
the
them
two
re-
flectors,
(18).
process each echo for B-mode and Doppler information (3), allowing the use of a higher frame rate. This may lead to compromise in the optimum pulse length or beam characteristics for B-mode or CD imaging,
one direction and blue, the other direction. The Doppler shift, which indicates the component of velocity aligned with the ultrasound beam, is usually encoded as saturation of color. Decreased saturation (increased
whiteness) typically indicates higher
tery tion
arreflec(20)
however. Color
There
Maps
are numerous strategies for
Doppler shading
as red
shifts toward
toward
such
to
and brachial artery mirror images can also occur (Fig 4). As this is a fundamental artifact of pulsed ultrasound, Doppler shifts from the mirror image are depicted by spectral analysis. Operators should suspect this artifact when Doppler shifts deep to a vessel
mirror those within the vessel. Scan-
facilitate
rapid Some
differentiation
flow in opposite instruments can
between
directions. also depict the
encoding the frequency, variance, and amplitude of Doppler shifts into color. Typically, red indicates flow in
Vtliim 177
.
amplitude
of the
Doppler
shift
M..ml..r
r__1_1,..__.
Figure
liver the
6.
Oblique a deep
vena
duplex
the cava depth
CD images
ambiguity
of the
arti-
demonstrate inferior
field
of view,
flow
in red,
in
and
is depicted
waveform
is obtained.
(b) With
shorter depth (higher PRF), Doppler shifts from the inferior vena cava are mapped in a more superficial location, as if they originated from the liver. Notice that attenuation has produced a weaker Doppler signal, as depicted in the CD image and the spectral
waveform.
shown)
Decreasing
eliminated the
(not
DOPPLER Doppler
(Ahasing)
LIMITATIONS Ambiguity
upper shifts by because it is not posunambiguously a is faster than one-half forms Doppler an
Shift
pulsed
Doppler
of the
Thus,
rate
Doppler
at which
shifts
it is sampled.
greater than
PRF/2 are wrapped around and depicted as if they were in the reverse direction (15-17). Most systems
allow the user to change the position
of the
baseline
in the
Doppler
spec-
tral display so that Doppler shifts less than the PRF can be depicted in the correct direction (Fig 3). Aliasing is fundamental to pulsed Doppler and therefore affects CD images. With most color-mapping schemes, aliasing is depicted on CD images as an abrupt change between pale (highfrequency) shades of color, such as from pale red to pale blue. This can usually be distinguished from flow separation, in which deep shades of color are typically adjacent to each other (Fig 5). Use of the mean velocity for color encoding can result in less aliasing than might occur if peak frequency
where
there
is no flow,
if PRF
is high
enough (22). This artifact may occur with low-frequency transducers, high far gain, and high PRF (short depth) (Fig 6), but it is rarely a clinical problem because intervening tissue usually attenuates the echo amplitude sufficiently, causing these shifts to fall below the threshold for depiction.
(Fig
7). This
latter
appearance
is also
each
other
as in a braid.
Helical
flow
Angle
of Insonation
With B-mode imaging the optimum angle of insonation is typically near 90#{176}, because reflections are strongest when the beam is perpendicular to the reflecting surface. This does not apply to Doppler US. This is because the Doppler shift is greatest when flow is aligned parallel to the ultrasound beam. As the angle of insonation relative to the axis of flow approaches 90#{176}, Doppler shifts decrease and may fall beneath the threshold for detection. On the other hand, a larger angle of insonation may be appropriate in some instances to reduce aliasing. With a large angle of insonation relative to the vessel, flow that is not parallel to the vessel walls will have more effect on CD images. Adjacent areas may be encoded with both red and blue. This may take the form of a diffuse mixture of red and blue, mimicking disorganized flow. Alternatively, the lumen of a vessel may be
is common distal to venous confluences such as the portal vein, in the proximal internal carotid artery, or in dilated tortuous vessels, because flow separation may result when the diameter of a vessel increases. The normal area of reversed flow in the carotid bifurcation (23) appears exaggerated at larger angles of insonation
(Fig
8).
it is likely that CD images
While
than the
has not
B-mode imangle of
been proved.
versed
components
of flow.
If the
One
must
be aware
that
CD imaging
peak velocity is used to encode color, high-velocity components will result in aliasing, and assignment of color will indicate flow in the opposite direction. The mean frequency, however, will be considerably less than the peak, so aliasing occurs less frequently.
does not eliminate errors in angle estimation. It is not known whether streamlines of flow as depicted on a velocity map correlate precisely with the true direction of flow. In addition, the angle as estimated from CD images represents a two-dimensional approximation of a three-dimension-
Depth
Ambiguity
The not
of blood within
the
of the
Reflections that originate deep to the field of view may arrive at the transducer after the next pulse has been emitted. These reflections may
Transducer
The
Geometry
beam
has skin
ultrasound
from
a lin-
be depicted
cation. 4 For
in a more
this reason,
superficial
deep Doppler
lo-
#{149} Radiology
and
on
between sector
the radius
ods. For the user, electronic beam steering is the most convenient, in that the angle of insonation for the Doppler component of the image can
be changed by than simply pushing a but-
ton
gles
(Fig
other
9). Steering
the
beam
at an-
90#{176} to the
transducer
surface
decreases the effective aperture and increases the beam thickness. This can degrade sensitivity and lateral resolution and increase artifacts from grating lobes (5). An alternative is to insert an anechoic wedge between the transducer and the skin to incline the whole linear array (Fig 3a). Disadvantages of a wedge include reverberations, attenuation, and increased distance between the target and the transducer,
a. which requires reduced PRF.
and
Frequency
temporal
resolu-
tion of the flow map likely to be less than of the B-mode image
superimposed.
quency (Doppler)
Temporal
resolution
and
frein a CD
image are also likely to be less than those of the corresponding Doppler spectrum. The axial resolution of Doppler is defined by the size of the range gate.
Improving by decreasing axial the Doppler resolution size of the range
shift
can
CD image
of the portal
1 kHz patient
One reason for this is because the signal-to-noise ratio is lower when a smaller volume is sampled. Reducing the range gate also reduces the time during which blood crosses the ultrasound beam. As red blood cells cross the beam, the amplitude of the echo increases and then decreases, broadening the Doppler spectrum. Because of this transit time effect, even a sin-
gle particle With a sector transducer, however, the angle is greatest at the center the image and least at the edges. ing of interpretation of flow direction,
moving
at a constant
ye-
Thus, a vessel that is parallel to the skin will appear blue at one end and red at the other and have no color or
a complex
mixture
in between
(Fig
transducregard-
it may increase the likelihood of detecting Doppler shifts in vessels perpendicular to the skin. This is especially true in the abdomen, where it is often difficult to change the angle of the transducer relative to the area of interest. The geometry of flow mapping with a curved-array trans-
locity through an ultrasound beam produces a spectrum of Doppler shift frequencies. For B-mode imaging, resolution is improved by using a beam as thin as possible, to reduce beam-width effects. The thinner the beam, however, the more rapidly the echo amplitude changes as a particle crosses it. This results in a fundamental uncerRadinlnev
#{149} c
tainty
relationship between spatial shift resolution. Efforts to improve the spatial resolution of
and Doppler
Doppler
therefore
tend
to reduce
the
accuracy
can be
with
measured.
which
The
Doppler
uncertain
shifts
rela-
between spatial and Doppresolution is one of several why optimal beam characterdifferent for B mode and
Doppler. The low a. b. 8. Transverse with different
(a) With
the
Figure quired
transducer.
CD images of the carotid bifurcation in an asymptomatic degrees of cephalic angulation. Red indicates flow away
the transducer inclined at approximately 600 relative to the
ac-
accounting CD imaging
a small
area of normal
overestimation of the size of the cardiac jets relative to estimates with angiography (10). In general, the jet area depicted by CD imaging is likely to be more dependent on instrument parameters than on vascular anatomy
reversed flow blue. (b) With the transducer indicated by the arrows.
is depicted predominates,
in as
spectral
quency frame
lines
per
second,
of 40-80 imaging
with
a fre-
resolution rate of CD
32 frames resolution
variable
of Doppler volume.
analysis
of a
a.
b.
POSTPROCESSING Color
The
versus
method
Gray
Scale
between
of deciding
color and gray scale for encoding each pixel is the source of some unique artifacts. There is no perfect method for this decision. The most
Figure 9. Longitudinal CD images of the common carotid artery (deep) and internal jugular vein (superficial) in an asymptomatic volunteer demonstrate the effects of transducer geometry on color mapping. (a) With a linear-array transducer, beam steering has been used to improve the angle between the vessels and the ultrasound beam. Flow velocity relative to the beam is similar at either end of the insonated segments of each vessel. (b) With a sector transducer, the direction of flow relative to the ultrasound beam is different at each end of the artery and vein. Furthermore, the color is palest at the ends of each vessel, where the angle relative to the ultrasound beam is smallest. Note that extension of color encoding beyond vessel walls, presumably caused by interpolation, does not occur at the center of the vessels, probably because the low Doppler shifts at the periphery of this portion of the yessels are not color encoded.
successful
strategies
use
echo
ampli-
tude, as well as Doppler frequency, to differentiate flowing blood from stationary fluid and moving solid tis-
sue. Myocardium,
vessel
walls,
perivas-
cular tissue, and moving viscera typically move slower than flowing blood and produce low-frequency Doppler shifts. These frequencies can be removed by means of wall filters,
moving
both
target
(2,4,24).
indicator
Unfortunately,
filters, information
blood and
or
these are Dopp-
filters
from
can
slowly
also
remove effective
flowing but
not
very
completely
strong
in removing
low-frequency
ler shifts. Various frequency-response curves for moving target mdicator filters can be used to vary the relationship of color encoding to
Doppler shift on the clinical High-amplitude frequency, application Doppler depending (2). shifts
a. Figure 10.
b.
CD images of a tissue-equivalent phantom obtained with high gain demonstrate suppression of color noise over echogenic regions by transducer motion. (Reprinted, with permission, from reference 25.) (a) With the transducer held stationary, color noise fills the far field of the image. (b) With slight transducer vibration, color is still present in the anechoic regions but is suppressed over the echogenic material. October 1990
#{149} Radiology
be rethat asre-
transducer
(Fig and
10) (25).
nous
shunt
but
can
also
obscure
vas-
Depending
orities
on how
and
when
gray
priscale
to strong
flections but allows color signed to weak reflections some instruments, motion
be used to suppress especially color. high-amplitude
cess
where
involves shifts
noise in
color Thus,
areas may
the Doppler gain is too high or the Doppler reject too low. If echo intensity is used to suppress color, noise
frame rate is maintained by using a low line density, gaps between lines may be filled with interpolated data.
This may be useful for imaging cardi-
Doppler
color
ac chambers
or large
vessels
but
inyesfill-
actually
be suppressed
by tissue
or
terferes with depiction sels. Spatial persistence ing in the small gaps
thrombosed
vessels
from
containing
color from of realnoise
adjacent
more press
pixels
cohesive color noise
of color
appearance.
to create can
a suponly
A similar
technique, color
color
spatial that
(Fig
filtering,
by 14). displaying Temporal
pixels
pixels
are adjacent
to other
persis-
Pulsations that are transmitted hypoechoic tissue can be especially difficult to distinguish from flow means of CD imaging (Fig 12). Pulsed Doppler spectral analysis,
however, will reveal a Doppler trum indicative of transmitted tions rather than of flow (25).
tence involves averaging B-mode or Doppler information over several frames. The extent to which postprocessing affects the plex hemodynamics plored extensively. depiction has not of combeen ex-
Figure
11. noise.
tissue
a surgically
color
solid
Transverse duplex CD image of proved hematoma filled with Color has been suppressed over
because the reflections from it
specpulsa-
CD
PARAMETERS
(transducer) in several re-
Moving
with color
tissue
adjacent
is often
to areas
encoded
of rapid
are stronger. Thus, setting color gain and reject to avoid noise over solid tissue did not prevent color from filling anechoic regions. Note flow in the deep femoral artery (A). Lack of flow in the hematoma was confirmed by means of spectral analysis. (Reprinted, with permission, from reference 25.)
flow
associated
shunts
with
stenoses
This
or artephe-
riovenous
(27,28).
nomenon
is manifested
as a diffuse
spects. As with B-mode US, penetration with CD imaging decreases with higher frequency. Thus, deep flow is
mixture of color, the extent of which is greatest during systole (Fig 13). This artifact can alert the diagnostician to the presence of an arteriove-
difficult to detect with high-frequency transducers because of limited penetration. On the other hand, Doppler shifts increase with higher carrier frequency. Thus, slow flow is more difficult to detect with low-frequency transducers because the
Doppler
shifts
are
lower.
Additional-
ly, flow aliases at a lower velocity with a high carrier frequency if PRF is kept constant. Jet areas from regur-
gitant
with plitude
or stenotic
low frequency,
valves
are
larger
because periph-
possibly at the
of increased
sensitivity
to lower-am-
ery of the
With
controlled
most
instruments,
separately for
gain
B mode
is
and
Doppler. suppressed
Because over
echoes, use of slightly high B-mode gain may decrease the presence of color noise over solid tissues. Exces-
B-mode color
and color
gain, within
reject noise
however, vessels.
should be ad-
gain
so that
but
depiction
decreases
of flow
as gain
the PRF
is opis
is not
is minimized.
b. CD images of an abdominal aortic aneurysm surrounded by with permission, from reference 25.) (a) Single-gate pulsed from the low velocity of the echogenic material adjacent to A similar signal is obtained from the peripheral hypoechoic pulsations. These echoes are weaker, so color has not been
Cardiac
jet size
directly
controlled
by the
operator
but is adjusted automatically so that it is as high as possible for the depth of interrogation. PRF is also comD.J...1......-.
_
Volume
177
#{149} Number
tied
to the
frequency
or corof
scale,
Doppler shifts. Thus, the scale by which Doppler shifts are color encoded changes with PRF and therefore with image depth (Figs 15, 16).
With many PRF, instruments, the wall fil-
color noise should be used. Users must be aware that this does not prevent mirror-image artifact (Fig 4) or localization of color noise within anechoic areas without flow (Fig 11), and they must be prepared to reduce
the gain further if these artifacts are suspected. For specific applications, additional adjustments may be necessary. The frame rate is affected by many factors. An increase in the field of
ter increases
creasing low-velocity
automatically
thereby components
with
preventing of a jet
in-
from being depicted (21). This may account for the observation that the apparent area of jets through stenotic or regurgitant valves is decreased when higher PRF is used (10,11). The dwell time, also referred to as ensemble length or packet size, is the
duration of Doppler sampling for
each
times
line
of sight.
Longer
dwell
to slow
improve
sensitivity
flow and accuracy of Doppler measurement and may thus improve the color component of a CD image
(5,21). The frame rate, however, be-
comes creases,
CD ing
slower thus
as the limiting
dwell the
time ability
inof
chang-
rapidly
CD
Users
IMAGING
should learn
TECHNIQUE
how to opti-
a.
Figure common 13. Tissue vibration from femoral vein. (Reprinted,
b.
an iatrogenic femoral arteriovenous with permission, from reference 28.) shifts fistula adjacent (a) Longitudinal in both directions. to the CD
mize
Doppler
B-mode
reject
gain,
on
Doppler
their system.
gain,
In
and
image
Spectral
id
shows
analysis is not
a diffuse
reveals obscured
mixture
blue more
(arrows) clearly
adjacent
Doppler
low-frequency
gain cause
of the Doppler
on the B-mode
a.
Figure 14. Oblique CD has been set to maximize images flow of a hepatic vein depiction without
b.
demonstrate color noise. the Note
C.
mapping. of the
with
color.
(b) Same
image
of printing but gaps
as a but without
from between
spatial
filtering.
Color
as b but
noise
with
appears
gain
in the image.
to eliminate
The
uneven
color noise.
appearance
Color
of the images
is now restricted
in a
to the
reduced
#{149} Radiology
October
1990
view will decrease the frame rate because a deeper field of view requires a slower PRF, and a wider field of
view (or wider sector angle) requires
dominal plications,
and
peripheral however,
vascular it is often
ap-
value
assigned
to a pixel
necessary to detect slow flow or resolve several small vessels. To accomplish this, frame rates tend to be slower than for cardiac applications.
Deep
flow,
such
as in the
abdo-
Slow
flow
must
be detected
for ef-
tect
dwell
slow
time
flow,
lower
PRF
both
and
longer
of which
decrease the frame rate. Thus, a rapid frame rate can be maintained only at the expense of spatial resolution (line
density), time), and Doppler field accuracy (dwell of view (depth and
deterare
most important for diagnosis. For cardiac applications, temporal resolution of flow tends to be more important than spatial resolution. For ab-
fective CD imaging of small arteries and most veins. High-frequency transducers should be used. Use of the wall filter should be minimized. In many cases this involves decreasing the PRF. Increasing the dwell time also improves sensitivity to slow flow. These changes also increase the sensitivity of the image to transducer and tissue motion and decrease the frame rate. Thus, slow scanning is usually necessary for satisfactory imaging of slow flow. Increased sensitivity to slow flow may decrease the
men, usually requires low-frequency transducers for detection. Unfortunately, this limits the sensitivity to slow flow. Increasing the dwell time may help. Unfortunately, long dwell time and low PRF (necessary to im-
age
deep
structures)
both
limit
the
frame rate, which makes scanning the abdomen difficult. For this reason, the narrowest possible window for color encoding should be used for deep abdominal CD imaging. Small vessels, such as in the kidneys and scrotum, can be detected best by increasing the slow flow, as described filters and persistence sensitivity above. should to Spatial not be
used,
but
temporal
persistence
might
help. High spatial resolution is not necessary to detect flow in small yessels, but vessels beneath the resolution limits of the system will be color
encoded
as if they
were
larger,
and
small vessels near each other will not be separated. One must be aware that the diameter of vessels cannot be measured accurately by noting the
encoding, limitations
because of color
and slow
of
interpolation,
flow at the periphery of vessels. The presence of flow within small vessels should be confirmed by means of
spectral analysis, as depiction by imaging may be unconvincing. While CD imaging can facilitate detection
a. b. Figure 15. Longitudinal CD images of a brachial artery during peak ual as in Fig 5) demonstrate the effects of PRF on color encoding and tings are similar to those in Figure Sb, except that increasing PRF by medium flow setting has eliminated aliasing. Note that the Doppler dicated at the top and bottom of the color bar, is higher than in Figure a except that PRF has been reduced by increasing the depth, causing
systole (same individaliasing. (a) The setchanging from low to frequency range, as in5b. (b) Same image as
aliasing.
of small
likely
vessels,
to be more
spectral
sensitive
analysis
once
is
the
proper
range
gate
is chosen.
Aliasing can be reduced by increasing the PRF (by reducing the depth or expanding the velocity or Doppler scale), by increasing the angle of insonation, or by reducing the transducer frequency. Users should be-
come
familiar
with
the
appearance
of
aliasing on their system. Although aliasing complicates CD images, it does not usually limit image interpretation by experienced users. In
can be used to identify Doppler shifts in an imbe aware, at which dependent however, aliasing on scan problem ocpa-
rameters
and
angle
of insonation.
important
for interpretation
forms, timation termine depicted
of spectral
as it may velocity.
waveprevent es-
In conclusion,
several
factors
deare of these
are related
principles,
to fundamental
while others
Doppler
result Radiology from
#{149} 9
color
saturation 177
in the image.
#{149} Number
Volume
efforts
of manufacturers
or users
to
8.
Helmcke
F, Nanda
NC,
Hsiung planes.
MC,
et al.
19.
improve
ages. likely
the
appearance
of the
im9.
Color
gitation
Doppler
with
assessment
orthogonal
of mitral
regurCircula-
and
JW,
sub-
Appreciation of these factors is to enhance the appropriate clinical use of CD imaging. #{149} Acknowledgments:
I thank the following individuals for substantive feedback during the preparation of this manuscript: Raymond Powis, PhD, Richard K. Johnson, PhD, Paul Magnin, PhD, Jeff Powers, PhD, Andrew Hayes, PARVT, Peter N. Burns, PhD, and Barry B. Goldberg, MD. I also thank Daniel A. Mer-
artifact
tion 1987; 75:175-183. Hoit B, Jones M, Eidbo E, Elias B, Sahn DJ. Sources of variability for Doppler color flow mapping imaging of regurgitant jets
in an animal model of mitral regurgita-
20.
10.
assistance.
11.
tion. J Am Coil Cardiol 1989; 13:16311636. Klewer SE, Lloyd TR, Goldberg SJ. In vivo relation between cineangiographic jet width and jet width imaged by colorcoded Doppler. Am J Cardiol 1989; 64:1399-1401. Stevenson JG. Two-dimensional color Doppler estimation of the severity of
atrioventricular valve frequency, regurgitation: im-
21.
Middleton WD, Melson GL. The carotid ghost: a color Doppler ultrasound mirror image artifact. J Ultrasound Med 1990; 9:S1. Daigle R. Quantitative aspects of color flow imaging. In: Nanda NC, ed. Textbook of color Doppler echocardiography.
Philadelphia: Lea Kossoff & Febiger, MB, Kossoff 1989; 283-
291.
22. Gill RW, G, Griffiths
KA.
US am1989;
TL. bifur-
References
1. Kisslo J, Adams DB, Belkin RN. Doppler color flow imaging. New York: Churchill Livingstone, 1988. Omoto R, Kasai C. Physics and instrumentation of Doppler color flow mapping. Echocardiography 1987; 4:467-482. Powis RL. Color flow imaging: understanding its science and technology. Diagn Med Sonography 1988; 4:236-245. Kasai C. Principles of Doppler color flow mapping. In: Nanda NC, ed. Textbook of
color Doppler echocardiography. Phila-
portant
pulse frequency. 1989;
effects
repetition 2:1-10.
of instrument
gain
and
setting,
carrier
Flow reversal
cation:
sis.
color JM,
Doppler
1988;
flow GW.
and
analyimag-
J Am Soc Echocardiography
24. A, Sahn DJ. the accuraand spatial
flow map25.
Radiology display
2.
Gessert
ing
Moore
modes
12.
3.
parameters.
4:375-381. Mitchell DG,
Echocardiography
Burns P. Needleman
Col-
J 1987;
M, Suzuki
114:1S2K, Omoto
26.
or Doppler
Ultrasound Mitchell DG,
artifact
4.
13.
158. Wong
ability
Med
artery
conventional
regions.
M, color GL. et al.
M, Matsumura
in the mapping
Needleman
R. Technical
and biologic
flow
sources
jets. Am
Femoral
sis with
pseudoaneurysm:
and
diagno165:687-690. sigDopp-
delphia:
5.
1989; 14-17.
of flow mapNC, ed. Text1989; 18-49. 14.
of aortic,
book of color
Philadelphia: 6.
Doppler
echocardiography.
Doppler
Middleton
US.
Radiology
Erickson
1987;
WD,
5, Melson
C, Hennerici
assessed by
M.
color
Perivascular
nificance 652. Igidbashian and
color
artifact:
pathologic
on color
Carotid Doppler
standard ography.
Burns
artery
disease
appearance
flow imaging:
correlation
with angi28. of
Icr US images.
VN,
Radiology
Mitchell
1989;
DC,
171:647Middleton
PN.
7.
16.
Doppler and spectral analysis. JCU 1987;15:567-590. Atkinson P, Woodcock JP. Doppler ultrasound and its use in clinical measurement.
170:749-752.
Press,
S.
1982.
US. I.
Doppler
Basic
18.
principles,
instrumentation,
174:297-307. KJW. Artifacts
and
in Med
pit-
imaging.
Ultrasound
1986;
5:227-237.
10
#{149} Radiology
October
1990