247

Quantification of Jet Flow by

Momentum Analysis
An In Vitro Color Doppler Flow Study
James D. Thomas, MD, Chun-Ming Liu, MD, Frank A. Flachskampf, MD,
John P. O'Shea, MB, BS, Ravin Davidoff, MB, BCh, and Arthur E. Weyman, MD

Previous investigations have shown that the size of a regurgitant jet as assessed by color
Doppler flow mapping is independently affected by the flow rate and velocity (or driving
pressure) of the jet. Fluid dynamics theory predicts that jet momentum (given by the orifice flow
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

rate multiplied by velocity) should best predict the appearance of the jet in the receiving
chamber and also that this momentum should remain constant throughout the jet. To test this
hypothesis, we measured jet area versus driving pressure, flow rate, velocity, orifice area, and
momentum and showed that momentum is the optimal jet parameter: jet area=1.25
(momentum)28, r=0.989, p<0.0001. However, the very curvilinear nature of this function
indicated that chamber constraint strongly aflected jet area, which limited the ability to predict
jet momentum from observed jet area. To circumvent this limitation, we analyzed the velocities
per se within the Doppler flow map. For jets formed by 1-81-mm Hg driving pressure through
0.005-0.5-cm2 orifices, the velocity distribution confirmed the fluid dynamic prediction:
Gaussian (bell-shaped) profiles across the jet at each level with the centerline velocity decaying
inversely with distance from the orifice. Furthermore, momentum was calculated directly from
the flow maps, which was relatively constant within the jet and in good agreement with the
known jet momentum at the orifice (r=0.99). Finally, the measured momentum was divided by
orifice velocity to yield an accurate estimate of the orifice flow rate (r=0.99). Momentum was
also divided by the square of velocity to yield effective orifice area (r=0.84). We conclude that
momentum is the single jet parameter that best predicts the color area displayed by Doppler
flow mapping. Momentum can be measured directly from the velocities within the flow map,
and when combined with orifice velocity, momentum provides an accurate estimate of flow rate
and orifice area. (Circulation 1990;81:247-259)

C olor Doppler echocardiography flow mapping on physical jet factors other than the regurgitant flow
represents a major technologic break- rate, per se. In particular, apparent jet size has been
through in displaying the spatial distribution observed in vitro to increase with the driving pressure
of abnormal blood velocities that characterize valvu- across the orifice, independent of the flow rate.3-5
lar regurgitation.12 However, the early hope for easy Thus, any theoretical framework for analysis of Dopp-
quantification of regurgitant flow with this technique ler flow maps must characterize observed jets by a
has yet to be realized. One reason for this difficulty is parameter that combines regurgitant flow rate and
that the appearance of the Doppler flow map depends driving pressure.
From the Noninvasive Cardiac Laboratory, Massachusetts Gen-
A second major limitation of most current methods
eral Hospital, Harvard Medical School, Boston, Massachusetts. for quantifying regurgitation with Doppler color flow
Presented in part at the 61st Annual Scientific Sessions of the mapping is that they make only limited use of the
American Heart Association, November 1988, Washington, DC. data available within the flow map, typically using
Supported by grant 13-532-867 from the American Heart simple measurements of the width, length, and area
Association, Massachusetts Affiliate; J.D.T. was supported by the of the regurgitant jet. By their nature, these Doppler
National Heart, Lung, and Blood Institute grant HL-07535; J.P.O.
is a recipient of an Overseas Clinical Fellowship of the National analysis methods treat the color flow map as a binary
Heart Foundation of Australia and the Athelston and Amy Saw quantity: flow is either present or absent, and no use
Research Fellowship of the University of Western Australia. is made of the actual velocities represented in the jet.
Address for correspondence: James D. Thomas, MD, Noninva- Although these methods have been correlated with
sive Cardiac Laboratory, Massachusetts General Hospital, Zero
Emerson Place, Suite 2F, Boston, MA 02114. angiographic assessments of regurgitation,6-9 neither
Received January 9, 1989; revision accepted August 21, 1989. approach provides a truly quantitative measure of
 248 Circulation Vol 81, No 1, January 1990

FIGURE 1. Schematic diagram of axisym-

metric, turbulent jet with momentum calcula-
tion. At left is the jet itself, with the x axis lying
Velocity Momentum along the jet axis and the r coordinate speci-
Profile Calculation fying the radial distance from this central axis.
v(x,r) The jet is formed by a plug of flow with
velocity uo entering through a round orifice
with area A0 at bottom left; within the free jet,
local velocity has an axial component u(xr)
and radial component v(xr). In the center
4 -m1 column are the velocity profiles transverse to
the jet axis (u plotted against r) at the orifice
Within free jet
level (bottom, plug flow) and within the free
jet (top, Gaussian or bell-shaped curve). At
right is shown the momentum calculation for
these two levels. In general, the momentum
flux crossing a given plane transverse to the jet
axis is calculated by integrating u2 across the
particular plane: IA uS dA. At the jet origin,
the velocity profile is flat, and this integral is
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

given simply by Equation ]B, A0uj2 (equiva-

lent to flow xvelocity, Equation 1A). Within
the free jet, the velocityprofile is no longerflat;
however, for a round onfice, it is axisymmet-
ric, so velocity is a function of radial distance
from the axis, and the momentum integral is
2nrLfO u2 r ar.

regurgitant flow. More sophisticated measurement 1) The jet parameter that best predicts the area
would be possible if the velocity map were analyzed displayed by color Doppler flow mapping is momen-
as a continuous quantity, with high-velocity regions tum, which combines flow rate, orifice area, driving
given more weight than low-velocity regions. To be pressure, and velocity into a single number and is
effective, however, any such analysis must use a fixed at the orifice.
theoretical framework based on fluid dynamics prin- 2) Jet momentum may be calculated from the
ciples. actual velocities within a color Doppler flow map and
Fortunately, the general field of fluid dynamics and may be shown to be constant throughout the free
the specific topic of turbulent jet flow have received portion of the jet.
intense theoretical and experimental study during 3) If momentum can be quantified anywhere within
the past two centuries because of their importance in the jet and if orifice velocity can be obtained by
hydraulics, jet propulsion, and pollution control.1011 continuous wave Doppler, then the clinically relevant
These studies have shown that the velocity distribu-
flow rate can be calculated as flow=momentum/
velocity. The effective area of the jet orifice can be
tion of the jet in the receiving chamber is best calculated as area=momentum/velocity.2
characterized by its momentum, a parameter that
combines jet flow rate, driving pressure, orifice veloc- Theoretical Background
ity, and orifice area into a single number.12-15 In the The jet to be studied theoretically is formed by the
absence of external pressure gradients, the momen- constant discharge of blood with velocity uo through a
tum crossing a plane perpendicular to the jet axis circular orifice with an effective area of A0 and flow
should be the same at any point along the jet axis rate Q0=A=u0 (Figure 1; all mathematical symbols
where it is measured.16 Thus, if momentum can be are defined in Table 1). If the Doppler transducer is
quantified anywhere in the jet, it must be the same as assumed to align parallel to the jet axis, then the
the momentum entering the jet at its orifice. At the appearance of the jet will be determined by this axial
orifice, momentum is given by the product of jet velocity as a function of x and r: u(x,r). Radial
velocity and the clinically relevant quantity of flow velocity [v(x,r)] will be ignored because it is orthog-
rate. Thus, by using jet momentum (measured any- onal to the Doppler beam.
where within the Doppler flow map) and orifice
velocity (measured by continuous wave Doppler), it Conservation Law Applied to Jets
may be possible to quantify the orifice flow rate. The jet may be thought of as a source for the
The major hypotheses tested experimentally in this transfer of mass, energy, and momentum into the
study were the following: receiving chamber, and conservation of these entities
 Thomas et al Momentum Analysis of Jet Flow 249

TABLE 1. Mathematical Symbols

A0 Effective jet orifice area (cm2)
JA Displayed color area of jet (cm2)
L Displayed length of free jet (cm)
M Jet momentum flux (cm4/sec2)
MSQ Square root of jet momentum (used in fitting Gaussian profiles to observed velocities,
cm2/sec)
p Pressure (mm Hg or dynes/cm2)
Pressure gradient producing jet (mm Hg or dynes/cm2)
Qo Orifice flow rate (cm3/sec)
r Radial distance from the jet axis (cm)
Effective jet orifice radius (cm)
r, Local jet axis (used in fitting Gaussian profiles to jets, cm)
Time (sec)
u Axial component of jet velocity (cm/sec)
U0 Jet velocity at the orifice (cm/sec)
U,
Low-velocity cutoff for display of color (cm/sec)
Urn Axial jet velocity along the central jet axis (cm/sec)
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

UN Nyquist limit for velocity display (cm/sec)

v Radial component of jet velocity (cm/sec)
x Axial distance from jet orifice (cm)
xc Length of receiving chamber, axial cutoff (cm)
Coefficients for curve fitting
Turbulent viscosity (relative to water= 1)
p Blood density (1.05 g/cm3)
e Base of natural logarithm, 2.71828

largely governs the behavior of the jet. Conservation equation (Ap = ½/2pu02, with pressure expressed in
of mass and energy are familiar from the continuity metric units, 1 mm Hg= 1,333 dynes/cm2):
and Bernoulli equations, respectively, but momen-
tum may be less so. M=Qouo (1A)
In rigid body mechanics (e.g., the flight of a
baseball), momentum is defined as mass multiplied (1B)
by velocity, and its change must precisely reflect
applied forces (such as gravity) as specified by New- (1C)
ton's second law of motion:
force = mass x acceleration_dM/dt (rate of change of =2A0Ap/p (1D)
momentum).
=Q002Ap/p (1E)
In fluid dynamics, we typically speak of momentum
flux, the amount of momentum passing through a Equations 1A and 1B are especially interesting.
plane per unit time. (In this paper, "momentum" and They imply that if the momentum of the jet and its
"momentum flux" are used synonymously unless velocity at the origin are known (obtained, for
stated otherwise.) For the jet in Figure 1, the axial instance, by continuous wave Doppler), one should
be able to derive the jet flow rate (Equation 1A) and
component of momentum is given by flow multiplied effective regurgitant orifice area (Equation 1B).
by axial velocity, or Q0u0.* Conservation of momen- Note that Equations 1A-E are valid only for plug
tum dictates that the total momentum flux crossing flow and so do not apply beyond the jet orifice where
any transverse plane orthogonal to the jet axis must the velocity profile is not flat. Therefore, to calculate
be constant throughout the full extent of the jet. the momentum passing through an arbitrary plane
Calculation of momentum. At the jet origin, momen- orthogonal to the jet axis, we must generalize Equa-
tum is expressible in many different ways, by combin- tion 1B and integrate u2 across the face of the jet:
ing the fundamental definition (M=Q0uo) with the
continuity equation (Qo=Akuo) and the Bernoulli
M= U2dA
JA
*Alternatively, fluid density (p) may be included as M=pQu0.
Because blood is incompressible (therefore, p is constant), either For the special case of an axisymmetric jet, we may
definition is acceptable. replace dA as shown with the circular rim 2mTr dr and
 250 Circulation Vol 81, No 1, January 1990

integrate from the center of the jet to the periphery ity. We have previously shown21,22 that the pressure
as shown in Figure 1: decay follows a predictable parabolic curve, whereas
the flow and velocity decay curves are linear. Thus, by
simply knowing the initial pressure gradient and the
M=27 u2 r dr (2) time elapsed to measurement, we can accurately
predict the instantaneous flow rate, pressure, and
Jet velocity distribution. In the absence of an exter- momentum flux at any time during the decay.
nal pressure gradient, momentum flux throughout For this study, we used circular orifices with areas
the jet retains its orifice value. Applying this conser- of 0.1, 0.2, 0.3, and 0.5 cm2. Data analysis for all
vation principle to simplified versions of the Navier- experimental runs was performed from an initial
Stokes equations permits an approximate description pressure gradient of 10 mm Hg. This combination
of the velocity distribution for a turbulent jet15: allowed the investigation of jets with the following
features: pressure gradient, 0-10 mm Hg; orifice
velocity, 0-158 cm/sec; orifice flow rate 0-60 cm3/sec;
u(x,r)= x
x
e-94(rX)2* (3) and momentum, 0-9,400 cm4/sec2. Heparinized canine
blood with a measured viscosity of 1.8 cP was used,
Thus, the velocity along the center axis where r is yielding peak jet Reynolds numbers of 3,134, 4,433,
equal to 0 (ur) decreases inversely with distance from 5,425, and 7,004 for the four orifices, respectively. We
the jet origin (um=7.8\AM/x), and the velocity pro- have reported that flows above Re= 300 in this model
file across the jet is a Gaussian (bell-shaped) curve: have low-frequency velocity fluctuations about 20%
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

of the mean velocity with full turbulence noted above

u(r) = u e- 94(r/x)' (4) Re=600,23 supporting our use of Equation 3 in
where um is the centerline velocity at a given distance, analyzing these data.
x, from the orifice. In vitro model 2: To examine jets with higher orifice
Equation 3 holds for turbulent rather than laminar velocities, we used a gravity-fed, constant flow model
jets, a transition governed by the jet Reynolds num- with pressure gradients as high as 81 mm Hg. Data
ber, Re=uOd/v, where do is the orifice diameter, and were obtained for flow through 0.005-cm2 and 0.1-
cm orifices at pressures between 18 and 81 mm Hg.
v is the kinematic viscosity. In constrained pipe flow, Maximal jet velocity was 4.6 m/sec with momentum
turbulence begins at about Re=2,300. For free jets 18,900 cm4/sec2. Water (with cornstarch added for
formed by abrupt stenoses, turbulence occurs much Doppler reflectivity) was used, yielding a peak Rey-
sooner, with instabilities noted as low as Re=25017 nolds number of 14,416.
and turbulence at Re=450.18 A recent study shows Echocardiographic examination. Doppler flow images
abrupt onset of turbulence at Re=500.19 were obtained with a Hewlett-Packard 77090 echocar-
Importantly, Equation 3 contains no reference to diograph and recorded onto 1/2 in. videotape for
fluid viscosity or the orifice velocity and flow rate. Far subsequent analysis. Acquisition parameters included
from the jet origin, there is no influence of the carrier frequencies of 3.5 and 5 MHz; depths between
specific orifice shape, only the impact of turbulent 6 and 20 cm, yielding Nyquist limits from 37 to 106
properties and total jet momentum.20 cm/sec; pure velocity display mode; large packet size;
Methods and gain optimized and left constant for the study. To
The three theoretical predictions were tested using examine the velocity decay along the jet centerline,
Doppler color flow mapping of in vitro models of pulsed Doppler recordings were obtained along the
valvular regurgitation. jet axis (guided by flow mapping) at distances of 1.5,
2.5, 3.5, 4.5, 5.5, and 6.5 cm from the orifice. Contin-
Data Acquisition and Initial Processing uous wave Doppler was used to obtain the peak
In vitro model 1. For this study, we primarily used velocity across the orifice.
a previously described 57 cm (H) x 8 cm(W) x 14 cm Color Doppler processing. The Doppler flow data
(L) Plexiglas model of the heart to simulate regur- from each recorded experimental run in model 1 was
gitant flow.21,22 This model is divided into two cham- analyzed from the time of maximal momentum, flow,
bers by a vertical septum, at the bottom of which is a and pressure until equilibration between the two
mount for valvular orifices. After a pressure gradient chambers occurred. Steady-state conditions were ana-
was established between the two chambers, blood lyzed for data from model 2.
was allowed to flow by gravity into the receiving Jet area was measured in model 1 using an offline
chamber. color analyzer (Sony). At 12-20 time points evenly
A very useful aspect of this model is that within a spaced within the jet decay for each orifice, the
single experimental run, it is possible to simulate a visible jet was hand traced and the area was mea-
wide range of flow rates, driving pressure, and veloc- sured and recorded. The time interval from the start
of each experimental run was also recorded to allow
*This formula differs from others that have been published15 in calculation of the instantaneous pressure gradient,
that we have substituted M from Equation 1B and have used the orifice flow, and jet momentum corresponding to that
area of the vena contracta instead of the anatomic area. jet area.
 Thomas et al Momentum Analysis of Jet Flow 251

The actual velocities within the jet were quantified Optimizing the combination of flow rate, velocity,
using a D-200 Off-line Analysis System (Dextra Med- pressure, and onifice area to predict jet area. The final
ical, Long Beach, California). For each experimental analysis of the jet area data was to determine whether
run in model 1, six to eight video frames were the combinations of flow rate, velocity, pressure
digitized, and a rectangular region of interest was gradient, and orifice area that best predicted jet area
marked off that enclosed the jet. In model 2, repre- were the same as those in Equation 1 used to define
sentative frames were analyzed for each level of jet momentum. For instance, the mathematical model
driving pressure. The color values for each pixel of jet area (JA) as a function of velocity and orifice
within the region of interest were then compared area was JA=aA013uj. Obtaining the logarithm of
with a look-up table derived from the color calibra- both sides produced an expression solvable by mul-
tion bar on the videotape,24 and the calculated veloc- tilinear regression: ln(JA) =ln(a) +/31n(Ak) + yln(u0)).
ities were written to a computer disk in ASCII If the ratio between the exponents 13 and y were not
format. Subsequent analysis of the velocity data was significantly different from the ratio used to define
momentum in Equation 1 (in this case, y/IP should be
performed using customized software written with 2 according to Equation 1B), then the momentum
the ASYST Scientific Analysis Package (Macmillan combination was taken to be the optimal one. This
Software Publishing, New York). test was conducted for each of the five combinations
Hypothesis 1. Analysis of Jet Area Versus Flow Rate, used in Equations 1A-E.
Orifice Area, Pressure, and Momentum Hypotheses 2 and 3. Analysis of Actual Velocities
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

Data from experimental runs with the four orifice Within Flow Maps
sizes in model 1 were pooled, producing a data set General velocity analysis. Each velocity map output
with measured jet area as the dependent variable and from the jet region of interest was a rectangular array
known flow rate, orifice velocity, driving pressure, of numbers corresponding to the axial velocities, uij,
orifice area, and momentum as independent (predic- with the rows (i index) arranged perpendicular to the
tor) variables. jet axis and the columns (j index) arranged parallel to
Univariate analysis of jet area versus flow rate, pres- the axis. By knowing the pixel calibration from the
sure, and momentum. Previous work in this video screen (horizontally and vertically) and the
laboratory425 showed a nonlinear relation between position of the region of interest relative to the jet
observed jet area and orifice flow. It is also axiomatic orifice, we could consider uij as discrete samples from
that with no flow, no color should be seen. Accord- the velocity distribution of the jet, u(x,r). These
ingly a power law function passing through the origin ASCII files were read by the ASYST customized
was chosen as the mathematical modeling function: software for further analysis.
y= axo, where y is observed jet area, x is the indepen- Because of the low Nyquist velocities (UN) present
dent variable (flow, pressure, and momentum, in turn), in the Doppler flow maps, many of the jets had
and a and ,B are fitting parameters to be determined. aliased velocities within them. Because bulk flow was
This fitting operation was performed using Marquardt always toward the transducer, any negative velocities
nonlinear least-squares approximation26 with the RS-1 observed within the jet were assumed to be aliased
data analysis program (Bolt, Beranek, and Newman, and were corrected by adding their observed negative
Cambridge, Massachusetts). The optimal jet parameter values to 2UN. No attempt was made to analyze
(of flow rate, pressure, and momentum) was chosen as velocities very close to the orifice where multiple
the one that yielded the highest correlation with the aliasing occurred. Finally, low-speed negative veloc-
observed jet area. ities seen outside the jet were believed to be true
Analysis of covariance. Again based on prior counterflow and were not included in the jet analysis.
observation,4,5 we postulated that distinct functional Overall jet structure. The accuracy of Equation 3 in
relations between jet area and flow would be observed describing jet velocity distribution was tested in two
for each orifice area. Similarly the relation between ways. First, the centerline velocity (um) was measured
jet area and driving pressure was expected to depend by pulsed Doppler at 1-cm intervals and compared
on the orifice area. However, if momentum were with the orifice velocity (uo, obtained by continuous
indeed the optimal jet parameter to predict its appear- wave Doppler) for 20 levels of jet momentum. These
ance, then orifice area should not effect the relation. data were fit to an inverse function predicted by
Accordingly, analysis of covariance was performed Equation 3: um/uo=al(x+13). The offset parameter, 13,
with jet area as the dependent variable; orifice area was included because turbulent jets typically behave
as the grouping variable; and flow rate, pressure as if they originate from a pointlike "virtual orifice"
gradient, and momentum, in turn, as the covariate slightly inside the proximal chamber.15 Second, we
(BMDP program PlV). The dependent variable and tested velocity profiles transverse to the jet axis for
each of the covariates were log-transformed to better their expected Gaussian shape. Data from single
model the power-law function used above. Orifice rows of the color velocity map (corresponding to a
area was considered to have a significant impact on a plane transverse to the jet axis at a distance xo from
given jet area-covariate relation if the adjusted group the orifice) were fit to a formula suggested by Equa-
means differed with a significance of p<0.05. tion 3: u(xo,r)=7.8MsQ/exo e-94[(r rC)IX]2 where the
 252 Circulation Vol 81, No 1, January 1990

parameters to be fit are MS, the square root of the jet any could improve on the predictions provided by
momentum; rc, the location of the jet axis within the momentum alone.
velocity map (to allow for slight random deviations of To analyze the jet velocity as a continuous variable,
the local jet axis from the overall axis; and e, a term the observed jet velocity was assessed for compatibil-
related to the "turbulent viscosity" of blood, expressed ity with the predicted inverse decay along the jet axis
as a ratio relative to that of water. Turbulent jet theory and Gaussian profile across the axis. The transverse
predicts that this term should be independent of the profiles were also used to calculate momentum
actual fluid, and thus, it was expected that E would be directly, which was compared with the known momen-
approximately 1. For each digitized jet, a number of tum of the jet. Finally, these momenta were divided
profiles from different values of x were fit by this by jet velocity to yield estimates of orifice flow rate,
procedure. The fitted MSQ was compared with the which was compared with the known flow rate and
square root of the known momentum, and c was were divided by velocity squared to give an estimate
compared with the expected value of 1. for the effective orifice area.
Hypothesis 2. Calculating momentum directly from
observed velocity. Momentum was calculated directly Results
from the observed velocity with Equation 2, with two Hypothesis 1. Analysis of Jet Area
minor modifications. First, the local jet axis swings Figure 2 shows measured jet areas plotted as
left to right randomly, so the local jet centroid (as a functions of flow rate (Figure 2A), pressure gradient
function of axial distance from the orifice) was cal- (Figure 2B), and momentum (Figure 2C). The same
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

culated to establish the center around which to data are plotted in each graph; only the parameter
assume axial symmetry in calculating momentum: chosen for the x axis is different. In each case, the
data are stratified by the orifice area that produced
r,(x) r u(x,r) dr/ u(x,r) dr. the jet. Figure 2A shows that for a given flow rate,
jets issuing from small orifice areas were considerably
larger than those from the larger orifices (because
The second modification to Equation 2 was to integrate the small orifice jets had higher velocity). Similarly,
the profile on both sides of the jet axis to use all of the Figure 2B shows that for the same driving pressure,
available data: M(x) = 7 f -' u2(x,r) [r-rc(x)] dr. M(x) jets from larger orifices were larger (because the
was calculated for each row in the jet region of interest. associated flow rate was larger).
From a 2-3-cm segment along the jet axis (40-100 rows Optimal prediction for jet area using single variables.
of data, depending on depth), mean momentum (M) Power-law fits derived with flow rate and driving
was calculated for comparison (by linear regression) pressure both show good correlation with the overall
with true momentum, and standard deviation was cal- observed jet areas. With flow rate: JA=3.12 Q 40,
culated as a measure of jet turbulence. r=0.947, SD regression= 1.17. With driving pressure:
Hypothesis 3. Use of momentum to calculate onifice JA=6.92 P 30, r=0.900, SD regression=1.59. With
flow rate and effective orifice area. Orifice velocity (u0) momentum as the independent variable, however, a
corresponding to the momertum calculations above better fit was obtained: JA=1.25 M 28, r=0.989, SD
was measured with continuous wave Doppler in regression=0.53, indicating that momentum is supe-
model 2. For model 1, orifice velocity was recorded rior to either flow or pressure in predicting jet
during repeated experimental runs identical to those appearance.
used for color Doppler acquisition. These velocity Analysis of covariance. Analysis of covariance on
data were then combined with the corresponding these data showed a very significant effect of orifice
mean jet momenta to yield orifice flow rate (QO) as area (p<0.0001) on jet size when the covariate was flow
suggested by Equation 1A: Qo=M/uO. The estimated rate (Figure 2A) or pressure (Figure 2B). However,
orifice flow rates from all of the jets analyzed were when jet area was adjusted by jet momentum, the data
then compared with the known flow rates by linear from the four orifice sizes were superimposable (Figure
regression. Similarly, effective orifice area (A) was 2C), and analysis of covariance showed no effect of
calculated using Equation 1B: A.=M/uo2. These effec- orifice size independent of momentum (p=NS).
tive areas were compared with the physical orifice Optimal bivariate prediction ofjet area. The jet area
area by linear regression. and all of the independent variables were log-
transformed to convert the power-law fits into a
Analytical Summary multilinear regression problem. Again, momentum
The jets generated for this study were analyzed alone predicted the appearance of the jet as well as
both as binary and continuous flow maps. For binary any combination of the other variables. Furthermore,
analysis, jet size was measured, and the best univari- when momentum was excluded from the analysis, the
ate predictor was selected from among flow rate, optimal combinations of the other variables were not
driving pressure, and momentum. Bivariate models found to be statistically different from Equations
with various combinations of flow, pressure, velocity, 1A-E and thus were not different from analysis with
and orifice area were also tested to determine whether momentum alone. This is further evidence that
 Thomas et al Momentum Analysis of Jet Flow 253

Figures 3B, 3C, and 3D display transverse velocity

profiles from the color flow maps of three jets. Also
shown are the best-fitting Gaussian curves of the
form predicted by Equation 4. Where the velocity
was completely nonaliased (as in Figures 3B and 3C),
the Gaussian fit in general was good, within the
limitations of turbulent flow and the coarse velocity
resolution of contemporary echocardiographs.
However, when velocity aliasing occurred (as in
Figure 3D), attempts at unwrapping the velocity
usually resulted in a marked discontinuity at the
aliasing boundary. Though this artifact is certainly
multifactorial in origin, one contributing cause
appeared to be "leakage" between the red and blue
video signals such that an individual pixel could have
forward and reverse velocity data written in it. Despite
this discontinuity, the Gaussian fits to unwrapped,
aliased flow was reasonably good. Throughout a wide
range of jet momenta, this Gaussian form fit the
observed velocity profiles with an average correlation
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

of r=0.85. Of note, the average turbulent viscosity, c,

was 1.46, indicating that the jets spread about 46%
faster than would have been expected for a free water
jet. This may be due to some intrinsic difference
between water and blood or due to the backflow of
B Pressure gradient (mmHg) blood hitting the end of the chamber and increasing
the shear rate of the periphery of the jet.
Hypothesis 2. Direct calculation of momentum. The
digitized regions of interest were used to calculate
momentum directly using Equation 2. To establish
the proper local axis location around which to base
the axisymmetric momentum calculation, a velocity
centroid was calculated for each cross-sectional pro-
file (with each jet having, typically, 80-150 profiles).
On average, this centroid swung randomly 0.9 mm
about the average axis as a function of axial distance,
which was about 5-10% of the total jet width. This
local centroid was then used to calculate the momen-
C
0 2
Momentum
4 6
(cm Is
_3B
X1io
10 tum. Figure 4 shows the typical appearance of jet
momentum as a function of axial distance from the
FIGURE 2. Plots of measured jet area displayed as a function orifice along with the fluctuation in jet axis. For this
of jet flowrate (Panel A), driving pressure (Panel B), and jet, the mean axis lay along r=0. Although large scale
momentum (Panel C). Each is stratified by the orifice area fluctuations in measured jet momentum are evident,
producing the jet. These show that both flow rate and pressure on average the momentum remained relatively con-
affect the jet appearance independently, but combining them stant, independent of distance from the orifice. Only
into momentum, optimally characterizes the jet. A, measured jet within about 2 cm of the chamber wall did the
area; Q, jet flow rate; P, orifice driving pressure; M, jet momentum consistently decay, reaching 0 just at the
momentum;ANCOVA, analysis of covarance with orifice area chamber wall. Figure 5 compares measured mean
as the grouping variable and flow rate (Panel A), pressure momentum with the true momentum for 50 jets (33
gradient (Panel B), and momentum (Panel C) as the covariate. from model 1, 17 from model 2), displaying good
agreement. These data are pooled from both in vitro
momentum is indeed the jet parameter that best models because analysis of covariance disclosed no
predicts the area displayed by Doppler flow mapping. difference in relations between true and observed
momentum. The major difficulty arose where multiple
Hypotheses 2 and 3. Analysis of Jet Velocities aliasing occurred in the core of some high-velocity jets,
General appearance of the jets. Figure 3A shows the making accurate unwrapping impossible.
decay in centerline velocity along the jet axis pooled Hypothesis 3. Calculating flow rate and effective
from 20 levels of jet momentum through the 0.1-cm2 orifice area from observed momentum. Figure 6 dis-
orifice. As predicted by Equation 3, um decays inversely plays the results of dividing the measured jet momenta
with x except for a plateau near the orifice where the by orifice velocity to derive the orifice flow rate (an
jet core has not yet been obliterated. inversion of Equation 1A). Jet velocity was measured
 254 Circulation Vol 81, No 1, January 1990

.-c
1-2
Centerline
. velocity (relative to uo) A
W-V, ., 35
Velocity (cm/s) B
y = 2.31/(x 0.4) _ .-0bserved velocity
r= 0.95 30 Momentum: 129
\E 25 r = 0.96
0.8 p < 0.0001
20
0.6
15
0.4
10
0.2 [ Predicted velocity
5
1.
v. \
0 2 4 6 8 -0.6 -0.4 -0.2 0 0.2 0.4 C07.6
Distance from orifice (cm) Radial Distance (cm)

60 Velocity (cm/s) Cl 120 Velocity

---. . ..
(cm/s) .. ._
D
,,-,*Observed velocity Observed velocity Momentum: 4272
5o Momentum: 434 100 "II*~NffV< p <
r
=
0.83
0.0001
40
\ffipU \yS ~p ( 0.0001
80 Nyquist limit L/.vr \ |
30 60 -
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

20 40
gXPredicted velocity
10 20 -

2 Predicted vefocity
n)_
U' .. a
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
Radial Distance (cm) Radial Distance (cm)
FIGURE 3. Plots of observed spatial distribution of axial velocity within jets from the 0. ]-cm2 orifice. Panel A: Decay of centerline
velocity as a function of axial distance. Data are pooledfrom 20 levels ofdriving pressure and are normalized to the respective onifice
velocity. Panels B, C, and D: Jet cross-sectional profiles with theoretical Gaussian fits. For low-momentum jets (Panels B and C),
the overall fit to the form of Equation 4 is good given the known jet turbulence and the coarse Doppler velocity measurements. In
Panel D, velocities above 68 cm/sec were aliased, and attempts at "unwrapping" them resulted in a marked discontinuity at the
aliasing boundary, reflecting leakage between the red and blue signals, in part due to limited color bandwidth of NTSC video.

by continuous wave Doppler and by application of and the effective orifice area calculated from the
the Bernoulli relation to the pressure gradient mea- observed jet momenta and velocities: y=0.82x+0.01,
sured in the flow models under conditions correspond- n=50, r=0.84, p<0.0001. The slope of the line (0.82)
ing to the color jet analyzed. Again, good agreement corresponds approximately to the coefficient of dis-
is observed with the known orifice flow rate for both charge for the orifice.
in vitro models. Discussion
A strong linear relation was observed between the
true orifice area (x in the following regression equation) We have shown that momentum flux, which com-
bines orifice flow rate, velocity, driving pressure, and
orifice area into a single number, is the best jet
parameter for predicting its appearance by color Dopp-
ler flow mapping. We have also shown in vitro that it
is feasible to measure momentum directly from Dopp-
ler flow maps and combine this with orifice velocity
(measured by continuous wave Doppler) to estimate
the orifice flow rate and the effective orifice area.
These facts may have implications for the quantifica-
tion of valvular regurgitation by Doppler mapping.
Analysis of Jet Area
Most of the early clinical studies of color Doppler
3 3.5 4 4.5
flow mapping correlated the length, width, or area of
the color jet with semiquantitative measures of regur-
Axial distance (cm) gitant severity such as angiography.6-9 Although some
FIGURE 4. Plot ofjet momentum flux (left axis) and local clinically useful algorithms have been produced by
axis location (right axis), each plotted against axial distance this approach, subsequent in vitro work has shown
from the orifice for a single Doppler flow map frame. Random that jet area is independently affected by jet flow rate
fluctuations in both are evident, but on average, momentum and orifice velocity.3-5 We have shown, however, that
remained constant throughout the jet. characterizing jets by their momentum (the product
 Thomas et al Momentum Analysis of Jet Flow 255

Observed momentum Observed flow rate

10 100 1000 10000 0 10 20 30 40 50

True momentum (cm4/s2) True flow rate (cm 3/s)
FIGURE 5. Regression plot of mean jet momentum calcu- FIGURE 6. Regression plot of jet flow rate calculated from
lated fiom Doppler flow maps compared with true momen- Doppler flow maps compared with known orifice flow rate. Jet
tum. Equation 2 was applied to the jet region of interest and momentum was calculated by Equation 2, then divided by
averaged along the jet axis. The line of identity is shown. orifice velocity to yield flow rate (Equation JA). Regression
line, solid line; line of identity, dashed line.
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

of flow rate and velocity) produced the optimal

correlation with the observed jet area. Indeed, when of pulsed Doppler. However, the gain setting will
jet area was plotted against momentum, data from a change the sensitivity of the instrument to low-
fivefold range of orifice sizes were superimposable. velocity signals and thus has much the same effect.27
As shown in this study, the relation between momen- Marked variation in displayed area (due presumably
tum and jet area is a highly nonlinear power law to changes in uc) have been observed in vitro not only
function: JA=1.25M 28. with variations in gain but also in carrier frequency,
Thus, momentum is the best jet parameter to pulse repetition frequency, and frame rate.27-29
predict jet area by color flow mapping. Unfortu- As shown in Appendix 1, the displayed length (and
nately, other factors such as chamber constraint and approximately the width) of an unconstrained jet
machine gain significantly influence the jet appear- varies directly with the square root of the jet momen-
ance and may have more impact on jet area than on tum and inversely with u,. Its area varies linearly with
the actual momentum of the jet. This may be M and with the inverse square of uc (Equation A2).
approached theoretically by considering the expected Thus, doubling the low-velocity cutoff decreases the
color area for jets of given momentum (M) entering displayed area by a factor of four. However, this
a receiving chamber of length xc (the "cutoff" in the dramatic dependence on u, is seen only for com-
axial direction). The third parameter entering into pletely free jets. When the length of the receiving
this analysis (detailed in Appendix 1) is the lowest chamber is less than the potential length of the free
velocity for which the Doppler machine will show jet, then the jet is constrained, and we must use
color, which is termed the low-velocity cutoff (u,). If Equation A3 to calculate its area. Figure 7A shows
we assume the velocity within the receiving chamber the effect of chamber constraint on displayed jet
is given by Equation 3, then color should be displayed area. Here, the jet is constrained at 11 cm, similar to
wherever the velocity is greater than uc. In fact, the in vitro model 1, and the overall relation between
current generation of color flow mappers do not have momentum and displayed area is distinctly nonlinear.
a sharp low-velocity cutoff analogous to the wall filter In fact, the shape of this curve is quite similar to that

An
Jet area (cm2)
.----- A -- Jet area (cm2) C
u,= 20 cm/sec -A 11 cm
40 ..........
25 U0.
uc= 2 cm/sec
5 cm/se
jFreeet 20 1 m
30 .C......... .....j ........e...................e...... ......... t....
20 cm/s
15
20
40 cm/se
10
10 5
nV0 10 8 0 6

Momentum (cm 5sY X10 3) Momentum (cm Vs, X10 3) Momentum (cM4/s, XiJ )
FIGURE 7. Plots of effect ofjet momentum, chamber constraint (xc), and low-velocity cutoff (us) on displayed jet area. Panel A:
For an unconstrained jet (xc= 0), area increases linearly with momentum. However, if the jet is truncated atxxc=11 cm, jet area rises
very nonlinearly and is similar to the curve in Figure 2C. Panel B: Effect of receiving chamber length on displayed area. The top
curve is the same as the constrained curve in Panel A. Panel C: Effect of low-velocity cutoff on displayed jet area for an 11-cm
receiving chamber. The u,=20 cm/sec curve is the same as in Panel A.
 256 Circulation Vol 81, No 1, January 1990

in Figure 2C. Figure 7B shows the effect of changes Application of the method to quantification of valvu-
in receiving chamber length (x,=3, 5, 7, 9, and 11 cm) lar regurgitation. The findings of this study have
on the displayed area, whereas Figure 7C shows the implications for the use of Doppler flow mapping in
influence of machine gain (u,=2, 5, 10, 20, and 40 the assessment of valvular regurgitation. If it were
cm/sec) on the area of a jet constrained at 11 cm. possible to quantify momentum within a clinical jet
Implications for using jet area to grade regurgitant and divide this by the orifice velocity from the same
severity. The lesson from this is twofold: 1) jet momen- time in the cardiac cycle, this should yield the instan-
tum is the best independent jet variable to use in taneous regurgitant flow rate. Repeating this process
predicting color flow area; 2) however, the precise throughout the time of regurgitation would then yield
functional relation between momentum and jet area flow as a function of time [Q(t)], which could be
is unpredictable and depends heavily on the size of integrated to give regurgitant stroke volume: fQ(t)
the receiving chamber and machine-dependent fac- dt. This approach could be simplified if it were
tors such as gain. Thus, no special import should be
possible to assume the regurgitant orifice area to be
constant throughout the regurgitant time period.
placed on the particular functional form found in our Then, the effective area could be found at a single
in vitro model, JA=1.25 M 28 because changes in x, point in the cardiac cycle (usually at the time of the
and uc may be expected to affect both the exponent largest jet) and multiplied by the orifice time velocity
and multiplicative constant in this expression, and integral to give regurgitant stroke volume: (M/u02)
these precise interactions would have to be deter- fuo(t) dt. Furthermore, the effective regurgitant ori-
mined empirically for a given geometry-machine com- fice area itself would be of clinical relevance because
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

bination. Thus, it is unlikely that any relation will be it may represent a load-independent measure of the
found to relate simple jet area to true jet momentum fundamental valve disease process.
(and by Equations 1A and 1B to jet flow rate and Study limitations. It must be emphasized that these
orifice area). This fundamental uncertainty in using techniques have been validated for a very artificial in
simple jet area is what makes analyzing the velocities vitro situation; clinical applicability has not yet been
themselves within the jet such an attractive approach. shown. Furthermore, even in this idealized setting,
several potential limitations were encountered because
Analysis of Actual Velocities Within the of the turbulent nature of the jets and the limitations
Color Flow Map of current color Doppler flow mapping technology.
Analysis of simple jet area treats the color flow The velocity variance within turbulent jets has been
map as binary data: flow is either present (color observed to average from 20% to 25% of the local
shown) or not (no color shown). All of the actual mean velocity.11,23 Thus, one may expect a similar
velocity data within the color map are discarded in degree of variation in measured momentum crossing
this analysis, data that we have examined by compar- individual planes within the jet. Beyond this true
ing the displayed color to the machine-generated variance, we observed further variation in measured
color bar. Although this approach is indirect (far momentum due to the random instantaneous swings
better to analyze the digital velocity data before color in the local jet axis that significantly altered the
encoding), we confirmed that the color-derived veloc- presumed axisymmetric geometry used in calculating
ities within turbulent jets were consistent with the the momentum. In particular, when the local jet axis
predictions of Equation 3: the centerline velocity deviated from the ultrasound plane, the central core
decays inversely with axial distance, and the velocity of the jet was not imaged, and calculated momentum
profile across the jet axis is Gaussian in shape. thus fell. Mathematical simulation of this problem
Advantages of momentum analysis in Doppler jet reveals the following reduction in measured momen-
analysis. Our overall strategy for analyzing these jets tum (expressed as percent of the true momentum)
was based on quantifying momentum flux within the for a given amount of imaging deviation from the jet
jet. Momentum flux is especially attractive for ana- axis (expressed as percent of the half velocity radius
lyzing Doppler flow maps of jets for several reasons. of the jet): 20% deviation, 5% error; 40% deviation,
First, momentum exists only in one form, not like 20% error; and 60% deviation, 40% error. In prac-
energy that is convertible from kinetic energy (mea- tice, it was generally straightforward to identify regions
surable by Doppler) to heat (unmeasurable by Dopp- within the jet where significant axial deviation
ler). Second, momentum is a vector quantity. That is, occurred and to exclude these areas from analysis.
it has components along each of the three coordinate In addition to these physical reasons for variance
directions, and each of these component momenta in measured momentum, there are a number of
must be conserved. This is important because Dopp- technical limitations in the current generation of
ler measures only the component of velocity parallel to color Doppler flow mappers, which have recently
the ultrasound beam and because the momentum been summarized.2 Among these are the relatively
measured with this component must remain constant coarse spatial, temporal, and velocity resolution of
throughout the jet. Finally, we can combine momen- the instrument. In particular, the finite lateral reso-
tum with orifice velocity (readily measured by contin- lution of contemporary echocardiographic equip-
uous wave Doppler) to estimate regurgitant flow rate ment may lead to apparent broadening of the jet with
(Q0=M/uJ) and effective orifice area (Ak=M/u02). consequent overestimation of momentum when Equa-
 Thomas et al Momentum Analysis of Jet Flow 257

tion 2 is applied. Mitigating this effect partially is the axial velocity urn with pulsed Doppler at a distance x
fact that this broadened region is at the jet periphery from the orifice, it should be possible to quantify QO. In
where velocity is the lowest and contributes little to vitro validation of an equation of this form was recently
the overall calculation. Velocity aliasing also limits reported.31 However, the considerable instantaneous
the analysis of high-velocity jets. We attempted only variance in both the velocity and position of the jet axis
a single unwrapping of the observed data, and so our noted in the present study emphasizes the need for
analysis was limited to regions where velocity was less careful data averaging with this simplified equation.
than twice the Nyquist limit. This limitation may be Quantifyingjet kinetic energy. Another recent report
minimized by using the lowest possible depth setting describes analysis of jet kinetic energy,32 showing
and carrier frequency and by not analyzing the improved characterization of jet severity over simple
proximal jet. The future development of a high-pulse jet area for situations of changing orifice area, driving
repetition Doppler velocity map would help further. pressure, flow rate, size, and compliance of the
Finally, at the aliasing boundaries, there were pixels
receiving chamber. Interestingly, the actual calcula-
tion performed was to sum the square of the pixel
that contained red and blue information because of velocities throughout the observed jet, which is sim-
either the limited color bandwidth of the NTSC video ilar to the method used in the present study. Velocity
standard30 or to signal degradation from the video- squared is the appropriate weighting factor for cal-
tape. This red-blue "leakage" led to a velocity dis- culating kinetic energy when the jet is analyzed as a
continuity at the aliasing boundary as shown in rigid body in classic mechanics (termed "a Lagran-
Figure 3D. Analysis of direct digital velocities should gian reference frame"); however, it also is the correct
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

eliminate this problem. weighting factor when momentum flux is calculated

Extension of momentum analysis to the clinical as we have done for the more conventional fluid
situation may encounter further difficulties. Cardiac dynamics analysis with the coordinate system fixed
flow is pulsatile, and Equation 3 has been validated within the moving stream (termed "an Eulerian
only in steady flow. There is a time delay between the frame").33 Two differences, however, should be noted.
onset of flow and full jet development and some In the present study, the known axisymmetric geom-
persistence of the jet after flow has stopped.25 These etry of the jet was used to calculate momentum flux
effects appear partially to offset each other, but this within the physical three-dimensional jet, whereas
must be investigated further. The turbulent nature of Bolger et a132 simply summed squared velocity within
these jets will require careful data averaging. Aver- the observed Doppler flow map without geometric
aging along the jet axis (as was done for the present weighting. Theoretical arguments exist in favor of
study) may not be entirely satisfactory in pulsatile analyzing the jet as a three-dimensional entity as we
flow; gated averaging throughout several cardiac have done, but the additional computational demands
cycles is another approach. The computer memory may in part offset this advantage. In addition, we
and processing power necessary for such an algo- have averaged together the momentum flux at vari-
rithm should be routinely available in clinical echo- ous planes perpendicular to the jet axis, whereas the
cardiographic instruments within a few years. kinetic energy was summed throughout the jet.
The present study used round orifices, whereas Proximalflow convergence. A final recently described
regurgitant lesions are usually irregular. If the jet is approach to Doppler flow quantification involves
not axisymmetric, then quantification of momentum analysis of the convergence zone proximal to the jet
from a single coaxial plane will not be possible. orifice.34 As blood accelerates toward an obstruction,
Fortunately, we have shown that jets originating from aliasing has been frequently noted to occur within the
elliptical orifices become axisymmetric within 2 cm.20 proximal chamber, which indicates that at the alias-
More troublesome will be mixing of flows (e.g., mitral ing point, blood is traveling at the Nyquist velocity,
regurgitation with pulmonic inflow) and eccentrically UN. This aliasing surface is in fact a hemisphere
directed jets that impinge on adjacent walls (and surrounding the orifice, with area 2rr2, where r is the
therefore impart momentum to the wall). Thus, as radius of the alias line. By the continuity principle,
encouraging as our in vitro results have been, exten- the flow through this hemisphere, 2ruNr2, must be
sive in vivo and clinical validation remain. the flow through the orifice. This method is funda-
mentally different from the above ones, being based
Other Quantitative Doppler Techniques for on conservation of mass rather than on momentum
Regurgitant Flow and is thus a potentially complementary technique.
Momentum analysis based on centerline velocity. If For instance, the convergence zone may be unreli-
jet velocity is given by Equation 3, simplifying momen- able for small jets where the aliasing radius cannot be
tum calculation may be possible by measuring only resolved; however, we have shown that momentum
the centerline velocity. Along the jet axis, r=O and flux quantification retains accuracy almost to zero
e-94(r/x)2il,so Um(x)=7.8x,/'M!x. Squaring this and flow. Conversely, high-flow and high-velocity jets,
substituting M=Q0uo from Equation 1A yields whose momentum may be difficult to quantify because
um2=60.8QuOJx2, which leads to the simple expres- of multiple aliasing, may be more easily quantified by
sion for Q: QO=umQx2/60.8u0. Thus, by measuring u0 using the continuity equation through the conver-
at the orifice with continuous wave Doppler and the gence zone.
 258 Circulation Vol 81, No 1, January 1990

Summary L=7.8\/K/uc
We have shown that the best single parameter to
predict the appearance of an axisymmetric jet by The total area of a free jet is obtained by integrating
color Doppler flow mapping is momentum flux, which Equation Al from 0 to L:
combines jet flow rate, velocity, driving pressure, and 7.8\Ai 1/2
orifice area into a single number. Unfortunately, JA(M,uc)=2 fJ 9L7 (n dx
although momentum is the optimal jet parameter, ucx
the actual observed area of color flow is more influ- This may be integrated using the substitution (=ucx/
enced by nonjet factors such as chamber constraint
and machine gain. Thus, analysis of jet area alone is 7.8N/M with the definite integral relation f01 5
[ln(1/)]l/2 d =Fr(1.5)/21 5 to yield*35
unlikely to yield quantitative data about jet flow rate.
We have shown, however, that an analysis based on JA=3.93 M/uc2 (A2)
the actual velocities within the jet is more promising. Jet area within a constrained chamber of length xc
In particular, we showed in vitro that momentum can can be estimated most simply by truncating the jet at
be accurately measured from the flow map for a wide xC cm from the orifice or, in other words, integrating
range of jet severity. Because momentum is also Equation Al from 0 to xc rather than 0 to L:
given by the product of orifice velocity and flow rate,
we divided the measured momentum within the jet by xcx 7.8ViM 1/2
the orifice velocity and obtained an accurate estimate JA(M,xc,uc)=2 J ln dx
ucx )
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

of the clinically relevant quantity of orifice flow rate.

Similarly, momentum was divided by the square of This integral is expressible in closed form through a
the velocity to yield an estimate for the effective jet series of substitutions. First, let (=ucx/7.8\M yielding
orifice area. Although extensive in vivo validation
remains, analysis of Doppler flow maps based on
conservation of momentum appears to be a theoret- JA=K fc 5 l/n(1/) d5
ically sound and potentially practical method for
evaluating the severity of valvular regurgitation. where K= (2 x 7.82M)/(9.7u,2) = 12.5M/uC2 and 4,=ug/
Appendix 1 7.8 that ranges from 0 to 1. Next, let v=
We seek an expression for the area of a color flow ln(1/5) yielding
jet as a function of jet momentum (M), the low
velocity cutoff of the Doppler instrument (uj), and JA=2K V2e-2vdv
the length of the receiving chamber (the axial cutoff
distance, xc). The jet velocity distribution within the
receiving chamber will be given by Equation 3: where V'=n(114.), which ranges from 0 for uncon-

strained jets to X for infinitesimally small receiving

chambers. This expression may be integrated by parts
u(x,r) = e -94(r/x)2
(fu dv=uv-fv du) with u=v and dv= -2`2 dv:
We make the simple assumption that color will be
displayed wherever the velocity exceeds uc; that is,
JA=K[1/2p,e-2vc+(Vrr/4V2) erfc(<V2PJ*
inside the contour where u(xr)=u,. Backsubstituting for P, e, and K yields
To describe the jet profile at a given velocity
contour, we must invert Equation 3 to express the
radius (r) of displayed color as a function of axial JA(Mxcuc)=~ /(1 7.8\/Ki
uckc
1/2

distance from the jet origin (x) and uc. After some
algebraic manipulation, we obtain + 3.93M erfc[(n 7.8v)
(A3)
r(x,u)=x In 7.8V (A1) This function is displayed graphically in Figure 7 and
( )t94 ucxJ
discussed in the text.
It should be remembered that this mathematical
x / 7.8_ M development assumes that the only effect of cham-
= In
9.7 ucx ber constraint is truncation of the jet at the cham-
Equation Al looks rather unwieldy, but it is in fact a
smooth, well-behaved function. *F(x) is the gamma function defined by the integral F(x)= -17
t'-1e-'dt, which is related to the factorial function as n!=F(n+1).
The displayed length (L) of the color jet (in the 2
absence of chamber constraint) is determined by *erfc is the complementary error function, erfc(x)= If x e-t 2
where the centerline velocity falls below u, or dt, familiar from normal (Gaussian) statistics.
x
 Thomas et al Momentum Analysis of Jet Flow 259

ber wall. Modeling the gradual transfer of momen- 17. Johansen FC: Flow through pipe orifices at low Reynolds
tum to the chamber wall would require complex number. Proc R Soc London 1929;126:231-245
18. Smith RL, Blick EF, Coalson I, Stein PD: Thrombus produc-
finite element calculations and should not be greatly tion by turbulence. JAppl Physiol 1972;32:261-264
different from Equation A3. 19. Krabill KA, Sung H-W, Tamura T, Chung KJ, Yoganathan
AP, Sahn DJ: Factors influencing the structure and shape of
References stenotic and regurgitant jets: An in vitro investigation using
Doppler color flow mapping and optical flow visualization. J
1. Omoto R, Yokote Y, Takamoto S, Kyo S, Ueda K, Asano H, Am Coll Cardiol 1989;13:1672-1681
Namekawa K, Kasai C, Kondo Y, Koyano A: The develop- 20. O'Shea JP, Thomas JD, Popovic AD, Svizzero T, Weyman
ment of real-time two-dimensional Doppler echocardiography AE: The profile of a regurgitant color flow jet is independent
and its clinical significance in acquired valvular diseases: With of orifice shape (abstract). JAm Col] Cardiol 1989;13:23A
21. Thomas JD, Weyman AE: A fluid dynamics model of mitral
specific reference to the evaluation of valvular regurgitation. valve flow: Description with in vitro validation. J Am Coll
Jpn Heart J 1984;25:325-340 Cardiol 1989;13:221-233
2. Sahn DJ: Instrumentation and physical factors related to 22. Thomas JD, Wilkins GT, Choong CY, Abascal VM, Palacios
visualization of stenotic and regurgitant jets by Doppler color IF, Block PC, Weyman AE: Inaccuracy of the mitral pressure
flow mapping. JAm Coll Cardiol 1988;12:1354-1365 half-time immediately following percutaneous mitral valvot-
3. Switzer DF, Yoganathan AP, Nanda NC, Woo YR, Ridgway omy: Dependence on transmitral gradient and left atrial and
AJ: Calibration of color Doppler flow mapping during extreme ventricular compliance. Circulation 1988;78:980-993
hemodynamic conditions in vitro: A foundation for a reliable 23. Thomas JD, Liu C-M, O'Shea JP, Davidoff R, McGlew S,
quantitative grading system for aortic incompetence. Circula- Weyman AE: How turbulent is a turbulent jet? An in vitro
tion 1987;75:837-846 color flow Doppler study (abstract). J Am Coll Cardiol 1989;
4. Thomas JD, Davidoff R, Wilkins GT, Choong CY, Svizzero T, 13:22A
24. Lobodzinski SM, Ginzton LE, Laks MM: Quantitation of
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

Weyman AE: The volume of a color flow jet varies directly

with flow rate and inversely with orifice size: A hydrodynamic color Doppler images with the color image processor (abstract).
in vitro assessment (abstract). JAm Coll Cardiol 1988;11:19A JAm Coll Cardiol 1988;11:99A
5. Simpson IA, Valdes-Cruz LM, Sahn DJ, Murillo A, Tamura T, 25. Davidoff R, Wilkins GT, Thomas JD, Achorn DM, Weyman
Chung KY: Doppler color flow mapping of simulated in vitro AE: Regurgitant volumes by color flow over estimate injected
regurgitant jets: Evaluation of the effects of orifice size and volumes in an in vitro model (abstract). J Am Coll Cardiol
hemodynamic variables. JAm Coll Cardiol 1989;13:1195-1207 1987;9:110A
6. Miyatake K, Izumi S, Okamoto M, Kinoshita N, Asonuma H, 26. Press WH, Flannery BP, Teukolsky SA, Vetterling WT: Numer-
Nakagawa H, Yamamoto K, Takamiya M, Sakakibara H, ical Recipes. New York, Cambridge University Press, 1986, pp
Nimura Y: Semiquantitative grading of severity of mitral 521-529
regurgitation by real-time two-dimensional Doppler flow imag- 27. Mohr-Kahaly S, Lotter R, Brennecke R: Influence of color
ing technique. JAm Coll Cardiol 1986;7:82-88 Doppler instrument setup on the minimal encoded velocity:
7. Helmcke F, Nanda NC, Hsiung MC, Soto B, Adey CK, Goyal An in vitro study (abstract). Circulation 1988;78(suppl II):II-12
RG, Gatewood RP: Color Doppler assessment of mitral 28. Stevenson JG: Critical importance of gain, pulse repetition
regurgitation with orthogonal planes. Circulation 1987; frequency and carrier frequency upon apparent 2D color
75:175-183 Doppler jet size (abstract). Circulation 1988;78(suppl II):II-12
8. Perry GJ, Helmcke F, Nanda NC, Byard C, Soto B: Evaluation 29. Utsunomiya T, Ogawa T, King SW, Moore GW, Gardin JM:
of aortic insufficiency by Doppler color flow mapping. J Am Effect of machine parameters on variance image display in
Coll Cardiol 1987;9:952-959 Doppler color flow mapping (abstract). Circulation 1988;
9. Suzuki Y, Kambara H, Kadota K, Tamaki S, Yamazato A, 78(suppl II):II-12
Nohara R, Osakada G, Kawai C: Detection and evaluation of 30. Pratt WK: Digital Image Processing. New York, John Wiley &
tricuspid regurgitation using a real-time, two-dimensional, Sons, 1978, pp 593-598
color-coded, Doppler flow imaging system: Comparison with 31. Cape EG, Yoganathan AP, Levine RA: A new method for
contrast two-dimensional echocardiography and right ventric- noninvasive quantification of valvular regurgitation based on
ulography. Am J Cardiol 1986;57:811-815 conservation of momentum: An in vitro validation. Circulation
10. Howarth L: Concerning the velocity and temperature distri- 1989;79:1343-1353
butions in plane and axially symmetrical jets. Proc Cambr Phil 32. Bolger AF, Eigler NL, Pfaff JM, Resser KJ, Maurer G:
Soc 1938;34:185-203 Computer analysis of Doppler color flow mapping images for
11. Wygnanski I, Fiedler H: Some measurements in the self- quantitative assessment of in vitro fluid jets. JAm Coll Cardiol
preserving jet. J Fluid Mechanics 1969;38:577-612 1988;12:450-457
12. Pai S-I: Fluid Dynamics of Jets. New York, McGraw-Hill Book 33. Hughes WF, Brighton JA: Fluid Dynamics. New York, McGraw-
Co, 1979 Hill, 1967, p 35
13. Abramovich GN: The Theory of Turbulent Jets. Cambridge, 34. Bargiggia G, Recusani F, Yoganathan AP, Valdes-Cruz L,
Mass, MIT Press, 1963 Raisaro A, Simpson IA, Sung HW, Tronconi L, Sahn DJ:
14. Schlichting H: Boundary Layer Theory, ed 7. New York, Color flow Doppler quantitation of regurgitant flow rate using
McGraw-Hill, 1979 the flow convergence region proximal to the orifice of a
15. Blevins RD: Applied Fluid Dynamics Handbook. New York, regurgitant jet (abstract). Circulation 1988;78(suppl II):II- 609
Van Nostrand Reinhold, 1984, pp 229-247 35. Weast RC (ed): Handbook of Chemistry and Physics. Cleve-
16. Yoganathan AP, Cape EG, Sung H-W, Williams FP, Jimoh A: land, Ohio, The Chemical Rubber Company, p A-209
Review of hydrodynamic principles for the cardiologist: Appli-
cations to the study of blood flow and jets by imaging tech- KEY WORDS * valvular regurgitation * echocardiography,
niques. JAm Coll Cardiol 1988;12:1344-1353 Doppler * fluid dynamics * computer methods
 Quantification of jet flow by momentum analysis. An in vitro color Doppler flow study.
J D Thomas, C M Liu, F A Flachskampf, J P O'Shea, R Davidoff and A E Weyman

Circulation. 1990;81:247-259
doi: 10.1161/01.CIR.81.1.247
Downloaded from http://circ.ahajournals.org/ by guest on January 18, 2018

Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1990 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on
the World Wide Web at:
http://circ.ahajournals.org/content/81/1/247

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation is online at:

http://circ.ahajournals.org//subscriptions/