Doppler Angle and Flow Velocity Estimations Using The Classic and Transverse Doppler Effects

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

252 IEEE TRANSACTIONS O N ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 46, NO 1,JANU.

4RY 1999

rres ce -

Doppler Angle and Flow Velocity Estimations the center of the conduit, the mean frequency equals t o the
Using the Classic and Transverse energy peak frequency theoretically The mean frequency, how-
Doppler Effects ever, strongly depends on the velocity gradient of the conduit,
and it is likely to be affected by the wall-thump filter or low
Bor-Ray Lee, Huihua Kenny Chiang", Member, IEEE, frequency disturbance for low-velocity flow
Cheng-Deng Kuo, Win-Li Lin, a n d San-KanLee The classic Doppler equation (1) can be used to estimate
a blood flow velocity, if its Doppler angle 0 is known In clin-
Abstract-Current clinical Doppler ultrasound systems ical applications, the Doppler angle can be determined from
could only measure the flow vector parallel to the ultra- the duplex scan image of ultrasound The scan image shows
sound beam axis, and the knowledge of the Doppler angle the Doppler ultrasound beam axis and the Doppler sample vol-
(beam-to-flow angle) is needed to calculate the real flow ve- ume on the B-scan tissue image, including the image of the
locity. Currently, the Doppler angle is determined visually blood vessel [3] However, many difficulties are involved in de-
by manually aligning a vessel axis marker along the blood termining the Doppler angle For example, it may be difficult to
vessel on the duplex scan image of the ultrasound. The ap- obtain a longitudinal cross-section scan of a blood vessel in the
plication of this procedure is often limited by practical con-
ultrasound scanning plane to determine the Doppler angle In
straints; therefore, measurements are not reliable. In order
to overcome this problem, we developed a simple Doppler addition, it is difficult to determine a three-dimensional (3-D)
angle and flow velocity estimation method using a combi- Doppler angle from a two-dimensional (2-D) ultrasound scan
nation of the classic and transverse Doppler effects. This image.
method uses only a single focused annular array transducer Recently, several Doppler ultrasound methods for 2-D and
to estimate the Doppler angle and the flow velocity. We 3-D flow velocity estimation have been developed These tech-
have verified experimentally that this method is successful niques used two transducers to measure the 2-D flow vector,
for measuring constant flow in a flow phantom between 4 5 O and three transducers for 3-D measurements [4]-[6] Newhouse
and 80' Doppler angle. The standard deviation of the esti- et al [a], [7]and Censor et al [SIused two transducers to esti-
mated Doppler angles is less than 4.5O. This method could
mate the 3-D flow vector by supplementing the classic Doppler
be implemented easily in medical Doppler ultrasound sys-
equation (1) with the transverse Doppler equation (2) The
tems to automatically estimate the Doppler angle and the
flow velocity. need of using at least two Doppler transducers limit its imple-
mentation in medical ultrasound systems Thus, an automatic
flow velocity estimation has not been implemented yet in cur-
rent medical Doppler ultrasound systems
1. INTRODUCTION
Newhouse et a1 [7],Censor et a1 [8], and Newhouse and Reid
[9] have exploited the characteristics of the Doppler ultrasound
EDICAL ULTRASOUND uses the Doppler effect, which as- system using a focused transducer to transmit a focused ultra-
sumes an infinitely wide, single-frequency ultrasound sound beam, and they have established the transverse Doppler
beam crossing a uniform flow to measure the flow velocity. theory describing the relationship between the flow velocity v
The ultrasound beam is scattered from particles carried by the and the bandwidth of the Doppler spectrum Bd
flow, and the ultrasound frequency is changed by the amount of
velocity-induced Doppler shift frequency. The classic Doppler
equation is given by:

where W and F represent, respectively, the effective diameter


and the focal length of a focused ultrasound transducer, and 8
is the Doppler angle Then (2) can be interpreted as a line of
where f d is the energy peak frequency of the Doppler spectrum
scatters intersect the acoustic field of the transducer, and the
[l],8 is the Doppler angle (the beam-to-flow angle), X is the
received bandwidth of the Doppler spectrum broadens as the
wavelength of the ultrasound, and v is the flow velocity. Some-
scatters crossing the focused ultrasound beam [1],[9].
times, f d is referred to as the mean frequency of the Doppler
Tortoli and Newhouse [1] theoretically derived and demon-
spectrum [2] rather than the energy peak frequency used in this
strated that the maximum frequency of the measured Doppler
article. When the center of the sample volume is positioned at
spectrum equals the Doppler shift frequency (fd) plus one-half
Manuscript received February 23, 1998; accepted July 21, 1998. of the Doppler spectrum bandwidth ( B d / 2 ) The maximum fre-
Parts of this work were supported by the National Science Council, quency fmax is
R.O.C., and the Veterans General Hospital and Yang-Ming Univer-
sity joint research projects.
B.-R. Lee and H. K. Chiang are with the Institute of (3)
Biomedical Engineering, National Yang-Min University, Taipei
112, Taiwan, Republic of China (*corresponding author's e-mail: Tortoli and Newhouse [1] also have demonstrated that, for
chiang@bme.ym.edu.tw).
C.-D. Kuo and S.-K. Lee are with the Taichung Veterans General
a known Doppler angle, ( 3 ) can be used to obtain a better esti-
Hospital, Taichung, Taiwan, Republic of China. mation of the flow velocity than the classic Doppler equation.
W.-L. Lin is with the National Taiwan University, Taipei, Taiwan, Equation (3) also has been implemented for blood velocity es-
Republic of China. timation [lo].

0885-3010/99$10.00 @ 1999 IEEE

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 03,2021 at 10:42:58 UTC from IEEE Xplore. Restrictions apply.
LEE et al.: FLOW VELOCITY AND TRANSVERSE DOPPLER EFFECTS 253

111. EXPERIMENTS

A. Doppler System

. A color Doppler ultrasound system (Vingmed CFM-725,


VingMed Sound, Horten, Norway), with a 4-elements annular
e; \
Flow vector array transducer, operating at 2.5 MHz, 1.47 cm in diameter,
and 7.8 cm in focal length was used in our study. The length of
. sample volume is kept at 1.5-mm long and is set at the focus.
:::;2yy Perpendicular
component
The pulsed repetition frequency (PRF) was 4 kHz. The CFM-
725 uses a complex discrete Fourier transform to compute the
Doppler spectrum within 1 ms.
./ Annular array transducers are widely used in medical ul-
Fig. 1. The schematic of the Doppler angle and flow velocity mea- trasound systems. One major advantage of the annular array
surement uses a single focused transducer. transducer is that it forms a symmetrically focused ultrasound
beam and has the same degree of spatial resolution in any plane
perpendicular to the beam axis. Under the assumption that a
flow has a single flow direction, the 3-D flow vector always can
In this paper, we propose and experimentally verify a simple be decomposed into two component vectors, one parallel to the
Doppler angle and flow velocity estimation method using the ultrasound beam axis, the other perpendicular to it, as shown in
combination of the classic and transverse Doppler effects. We Fig. 1. This symmetrically focused property of the transducer
estimate the Doppler angle and the flow velocity by using the is exploited and used to estimate the Doppler angle and flow
combination of fd (1)and fmax (3). Estimating the Doppler an- velocity [8]. The symmetric characteristics of the transducer
gle and the flow velocity (the magnitude) is much simpler than enables the measurement of the two flow vectors (parallel and
estimating the 3-D flow vectors, as proposed in other Doppler perpendicular), and they are inherently included in (1) and (3).
ultrasound flow velocity estimating methods [4]-[8]. The major
advantage of this method is that only a single focused trans-
ducer is needed. Thus, this method can be easily implemented B. Flow Phantom
in medical ultrasound systems to enable an automatic estima-
tion of Doppler angle and flow velocity. A UHDC (Shelley Medical Imaging Technologies, London,
Ontario, Canada) flow phantom, which can be programmed
to generate constant and pulsatile flows, was used in this re-
search. Two constant flows with mean velocities of 18.9 and
11. THEORY 25.2 cm/s (Reynolds number = 1288 and 1717) were generated
in a 9.5 mm internal diameter latex tube. The conduit was
1 meter long and was immersed horizontally in a tank. The
The schematic setup of the flow velocity measurement is UHDC blood mimic fluid was used as the scattering source in
shown in Fig. 1. For measurement of a single flow line inter- the flow. The density of the fluid was 1.01 g/ml, and the dy-
cepted by a symmetrically focused ultrasound beam, the flow namic viscosity was 1.3415 cp (1 cp = lo-' g/sec.cm).
vector can be treated as the combination of two flow vectors; The relationship between the maximum flow velocity (vmax)
one is parallel to the ultrasound beam axis, and the other is per- and the mean flow velocity ( V ) can be represented as a function
pendicular t o the ultrasound beam axis. These two flow vectors of Reynolds number [ll].The value of vmax/V is between 1.5
are represented by v . cos i9 and v . sin 0 in (l),(2), and (3). Al- to 1.8 for the corresponding Reynolds numbers, 1200 and 1700.
though either (1) and (2) or (1) and (3) can be used to derive The estimated vmax/V of the two constant Aows are within this
the Doppler angle, we chose (1) and (3). The lower portion range (1.58 for V = 18.9 cm/s, and 1.55 for i7 = 25.2 cm/s).
of the Doppler spectrum is likely to be affected by the wall-
thump filter, sample volume size, and sample volume position. C. Signal Processing
Thus, (a),the full bandwidth of the Doppler spectrum was not
used; instead, (3), the maximum Doppler frequency was used
to derive the Doppler angle (4): The real-time Doppler spectra were retrieved from the
VingMed CFM-725 Doppler ultrasound system and used in
the analysis. For Doppler spectrum calculation, a duration of
2 seconds Doppler signal, 325 spectra were averaged to reduce
(4) the noise. However, in our preliminary experiment, a spectrum
of 100 times averaged is good enough for Doppler angle and
flow velocity estimation. A typical Doppler spectrum of the
The flow velocity can be estimated by (1) or (3) after 19 is 25.2 cm/s flow measured at 60" Doppler angle is shown in
determined. However, the flow velocity, v , also can be derived Fig. 2. The maximum Doppler frequency fmax is defined as
by v . cos 0 and v . sin 0 in (1) and ( 3 ) : the intersection of the power spectra and an empirically deter-
mined threshold set at -6 dB [12].
The setting of the threshold for fmax should be empirically
determined for each Doppler ultrasound system because each
(5)
system has its own signal processing circuits and Doppler gain

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 03,2021 at 10:42:58 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 46, NO. 1 ~ JANUARY 1999

500
n
2 450
v
+,400
0
5 350
2 300
L4
250
f 200
-0
150
2E 100
\
!
3
200
I
400 600 800
-6dB
1000 1200 1400
50
0
7

45
25.2 cm/s

50 55 60 65 70 75 80 85
Frequency (Hz) Measured Angle

Fig. 2 . A typical Doppler spectrum of a 25.2 cm/s flow measured at


60' Doppler angle.

curve. In this experiment, the threshold is determined by cal- 450 7 I


ibrating a set of estimated Doppler angles to a set of known
angles (e.g., 50", 60", and 70') set in the UHDC flow phantom.
350
D . Sample. Volume h
2 300
In this research, the Doppler sample volume was positioned $!2 250
at the center of the conduit, and the size of the sample volume lG
was kept at the minimum, 1.5 mm, in VingMed CFM-725. This $j 200
0
arrangement could minimize the effect of changing the sample a
Q 150
volume size and position on f d and fmax [13], [14].
2
i-

.- 100
E.Flow Measurement
2 50
Two different flows were generated in the UHDC flow phan- 0- 1 I
tom. The measured maximum Doppler frequency fmax and the 45 50 55 60 65 70 75 80 85
energy peak frequency f d of these two flows in 12 measure- Measured Angle
ments are shown in Figs. 3(a) and (b), respectively. The solid
lines represent the mean value, and the error bars represent
the standard deviations. We observed that fmax and f d de-
crease with the increase of the Doppler angle. After finax and Fig. 3. The measured (a) maximum frequency fmax, and (b) energy
fd were measured, the Doppler angles and flow velocities were peak frequency fd of the 18.9 and 25.2 cm/s flows.
estimated using (4) and ( 5 ) , respectively. The calculated re-
sults are shown in Figs. 4 and 5 . The standard deviation of the
estimated Doppler angles is less than 4.5", and the standard is estimated visually by using a vessel axis marker on the ul-
deviation of the estimated flow velocities is less than 4.7 cm/s. trasound duplex scan image. Therefore, the current methods
are limited because the vessel orientation needs to be manually
set by the operator, and the vessel must lie in the ultrasound
IV. DISCUSSION scanning plane.
Measurements shown in Figs. 4 and 5 demonstrate that the
We have experimentally demonstrated a Doppler angle application of this method for Doppler angle and flow velocity
and flow velocity estimation method by combining the classic estimation is valid for a wide range of angles between 45" and
Doppler equation (I) and transverse Doppler equation (3). This 80". For Doppler angles larger than 80",the lower frequency
method has two major advantages. It is very simple, only one portion of the Doppler shift frequency spectrum might be fil-
focused transducer is needed. It can be implemented directly tered out by the wall-thump filter of the ultrasound receiver
in current medical Doppler ultrasound system with a clinical Thus, the energy peak frequency f d of the filtered Doppler spec-
annular array transducer. The Doppler angle and flow velocity trum may be overestimated; and the Doppler angle may be
estimation method is very important for obtaining quantita- underestimated, especially for flows with a lower velocity such
tive flow velocity measurements. Currently, the Doppler angle as the 18.9 cm/s flow measured at 85", in Fig. 5. However, a

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 03,2021 at 10:42:58 UTC from IEEE Xplore. Restrictions apply.
LEE et al.: FLOW VELOCITY AND TRANSVERSE DOPPLER EFFECTS 255

90 1.0

85
--
80
75 Wall-thump filter
W

?70
$65
I-
(d

.g 60
x
55
-25.2 cm/s
50
45 - 1 - L I .
. ..I
0- I 200 400 600 800 1000 1200 1400
45 50 55 60 65 70 75 80 85 90 Frequency (Hz)
Measured Angle
Fig. 6. The energy peak frequency f d of the Doppler spectrum of the
Fig. 4. The actual (X-axis) and estimated (Y-axis) Doppler angles of 25.2 cm/s flow measured at 60' Doppler angle before and after the
two flows with different flow velocity. wall thump filter.

0.5
n

B 0.45
0.4
*sr
-$
*g 0.35
0.3
0.25
Ag 0.2
E: 0.15
;
t i

i.-o:: -TI- -e-


18.9 cm/s
25.2 cm/s
0 __

45 50 55 60 65 70 75 80 85 90 -200 a 200 400 600 800 1000


Measured Angle Frequency (Hz)

Fig. 5 . The calculated maximum flow velocities of two different flows Fig. 7. Three Doppler spectra of the 25.2 cm/s flow measured at 50°,
measured at different Doppler angles (maximum flow velocity # 70°, and 8 5 O Doppler angle. A represents fmax, and B represents fd.
mean flow velocity).

Doppler angles, respectively. The lower portion of the Doppler


90" Doppler angle can be determined easily by observing the spectrum is filtered by the wall-thump filter at 85". Therefore,
symmetric property of the Doppler spectrum, which is centered the estimated Doppler angle and flow velocity are not accurate
around the zero frequency [15], [16]. Thus, the flow velocity at when the energy peak of the Doppler spectrum is affected by
90' also can be correctly estimated and is shown in Fig. 5 . the wall-thump filter.
The Doppler angle and flow velocity estimation method is not The influence of the size of the sample volume will make the
suitable for Doppler angles less than 42" in this experiment lower portion of the Doppler spectrum broaden [17]. The sam-
because the critical angle between water and the wall of the ple volume could be considered as a cylinder [17]. The height
polyethylene tube is 42" [12]. of the cylinder equals the length of the sample volume, and
The effect of the wall-thump filter on the Doppler shift spec- the width of the cylinder equals the width (beamwidth) of the
trum at a large Doppler angle is schematically shown in Fig. 6. sample volume. If the sample volume is not small enough or
The energy peak frequency fd of the filtered spectrum shifts to not much smaller than the diameter of conduit, the sample
a higher frequency; therefore, the estimated Doppler angle cal- volume will intercept with many flow lines with different veloc-
culated by (4) will be underestimated, and the estimated flow ities. Then, the received Doppler spectrum could be treated as
velocity will be underestimated. Fig. 7 represents the Doppler the result of summation of geometrically broadened line spectra
spectra of the 25.2 cm/s flow measured at 50°, 70°, and 85" associated with each velocity. Law et al. [14]also proved that

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 03,2021 at 10:42:58 UTC from IEEE Xplore. Restrictions apply.
256 IEEE TRANSACTIONS O N ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 46, NO. 1,JANUARY 1999

the Doppler spectrum is affected by the position of sample vol- M. D. Fox and M. Gardiner, “Three-dimensional Doppler ve-
ume. If the sample volume is set at the center of a conduit, the locimetry of flow jets,” IEEE Trans. Biomed. Eng., vol. 35, pp.
effect of changing the size of the sample volume on the vari- 834-841, Oct. 1988.
I. A. Hein, “3-D flow velocity vector estimation with a triple-
ance of fd is less than 8%. Law et al. [14] concluded that fmax beam lens transducer experimental results,” IEEE Trans. Ul-
and f d are affected by the position of the sample volume but trason., Ferroelect., Freq. Contr., vol. 44, pp. 85-89,1997.
to a lesser degree by its size, and the quantification of pulsed V. L. Newhouse, D. Censor, T. Vontz, J. A. Cisneros, and B.B.
Doppler spectrum is feasible if the sample volume is positioned Goldberg, “Ultrasound Doppler probing of flows transverse with
respect to beam axis,” IEEE Trans. Bzomed. Eng., vol. BME-34,
within the central 25% of the tube. pp. 779-789, Oct. 1987.
D. Censor, V. L. Newhouse, and T. Vontz, “Theory of ultra-
sound Doppler-spectra velocimetry for arbitrary beam and flow
V. CONCLUSIONS configurations,” IEEE Duns. Biomed. Eng., vol. BME-35, pp.
740-451, Sep. 1988.
V. L. Newhouse and J. Reid, “Invariance of the Doppler band-
We have demonstrated a Doppler angle and flow velocity width with flow displacement in the illuminating field,” J .
estimation method for constant flow measurement using only Acoust. Soc. Amer., vol. 90, no. 5, pp. 2595-2601, 1991.
one focused annular array transducer. The Doppler angle can Y.Chabria and V. L. Newhouse, “Estimation of axial blood
velocity using the Doppler equation corrected for broadening,”
be computed from the combination of the classic Doppler ef- Ultrasound Med. Biol., vol. 23, pp. 967-968, 1997.
fect (I) and the transverse Doppler effect (3). For pulsatile flow G. Guidi, V. L. Newhouse, and P. Tortoli, “Doppler spectrum
measurements, the Doppler angle and flow velocity need to be shape analysis based on the summation of flow-line spectra,”
estimated in a very short time to obtain accurate results. This IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 42, no. 5,
pp. 907-915, 1995.
problem may be resolved by using fast digital signal process- W. L. McCabe, J. C. Smith, and P. Harriot, Unit Operations
ing techniques, which can be implemented by a customer-built of Chemical Engineering, 5th ed. New York: McGraw-Hill, pp.
circuit to speed up the computation. Further work with more 96-98, 1992.
sophisticated algorithms is needed to apply the Doppler angle J. W. Andrew and W. Junru, “Correction of intrinsic spectral
and flow velocity measurement method for pulsatile flow mea- broadening errors in Doppler peak velocity measurement made
with phased sector and linear array transducer,” Ultrasound
surements. Med. Biol., vol. 21, pp. 1928-1035, 1995.
Y. F. Law, P. A. J. Bascom, E(. W. Johnston, P. Vaitkus, and
R. S.C. Cobbold, “Experimental study of the effects of pulsed
Doppler sample volume size and position on the Doppler spec-
REFERENCES trum,” Ultrasonics, vol. 29, pp. 404-410, Sep. 1991.
G. P. Tortoli, G. Guidi, and P. Pignoli, “Transverse Doppler
P. Tortoli, G. Guidi, and V. L. Newhouse, “Improved blood ve- spectral analysis for a correct interpretation of flow sonograms,”
locity estimation using the maximum Doppler frequency,” Ul- Ultrasound Med. Biol., vol. 19, no. 2, pp. 115-121, 1993.
trasound Med. B i d , vol. 21, pp. 527-532, 1995. S.A. Jones and D. P. Gidedens, “A simulation of transit time
V. L. Newhouse, K. S. Dickerson, D. Cathignol, and J. Y. effects in Doppler ultrasound signals,” Ultrasound Med. Biol.,
Chapelon, “Three-dimensional vector flow estimaiton using two vol. 16, pp. 607-619, 1990.
transducers and spectral width,” IEEE Trans. Ultrason., Ferro- P. Tortoli, G. Guidi, and P. Pignoli, “Transverse Doppler spec-
elect., Freq. Contr., vol. 41, pp. 90-95. tral analysis for a correct interpretation of flow sonograms,” Ul-
P. Fish, Physics And Instrumentation of Diagnostzc Medical Ul- trasound Med. Biol., vol. 19, pp. 115-121, 1993.
trasound. Chichester: Wiley, 1990.
J. R. Overbeck, K. W. Beach, and D. E. Strandness, ‘Vector
Doppler: Accurate measurement of blood velocity in two dimen-
sions,’’ Ultrasound Med. Biol., vol. 18, no. 1, pp. 19-31, 1992.

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 03,2021 at 10:42:58 UTC from IEEE Xplore. Restrictions apply.

You might also like