1-172
State-of-the-Art Lecture
Three-dimensional Echocardiography
Advances for Measurement of Ventricular Volume and Mass
Donald L. King, Aasha S. Gopal, Andrew M. Keller, Peter M. Sapin, Klaus M Schroder
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
Abstract There is a need for more accurate and reproducible serial measurement of left ventricular volume and mass in
individual subjects by echocardiography. Conventional echocardiography has significant measurement variability because of its
use of geometric assumptions and image plane positioning
errors. Guided three-dimensional echocardiography eliminates
geometric assumptions and reduces image plane positioning
errors by using a "line of intersection" display. Use of threedimensional guided imaging for a one-dimensional measurement of the left ventricle resulted in a threefold improvement of
interobserver variability over conventional echocardiographic
measurements. Computer-aided three-dimensional reconstruction of the ventricle for ventricular volume from a series of 8 to
10 short-axis images also achieved more than a threefold
R
ecent advances in technology have made it possible to create a three-dimensional echocardiograph, a scanner that registers image position
and orientation in three-dimensional space.1 This system
is able to guide acquisition of images using a "line of
intersection" display and to perform three-dimensional
reconstructions of the heart in a clinically practical manner.2 The underlying purpose for creating this system has
been to achieve a significant improvement in the accuracy
and reliability of quantitative echocardiographic measurements, such as ventricular volume and mass. Conventional
echocardiographic estimates of ventricular volume and
mass, whether by one- or two-dimensional echocardiography, have significant inherent measurement variability
because of their use of geometric assumptions and image
plane positioning errors.34 Conventional echocardiography requires the use of geometric assumptions because the
position and orientation of its images are not registered in
three-dimensional space and must be assumed. Also, with
conventional echocardiography, image plane positioning
errors occur because the operator is unable to visualize
anatomic landmarks in the nonvisualized dimension orthogonal to the real-time image plane. Visualization of
these landmarks is essential for accurate and reproducible
positioning of image planes. The effect of these limitations
on quantitative estimates by conventional echocardiography may produce a measurement variability in excess of
the anticipated biologic changes being sought in the
measurement. This necessitates the use of relatively large
numbers of patients to detect a therapeutic effect in
clinical trials and renders the conventional methods of
From Columbia University (D.L.K., A.S.G., A.M.K.), New
York, NY; University of Kentucky (P.M.S.), Lexington; and Free
University of Berlin (K.M.S.) (Germany).
Correspondence to Donald L. King, MD, Cardiology Division,
Columbia University, (PH9C-342), 622 West 168th St, New York,
NY 10032-3702.
improvement of interobserver variability compared with twodimensional echocardiography. Three-dimensional echocardiographic computation of ventricular volume and mass in healthy
subjects was achieved with an accuracy comparable to magnetic
resonance imaging and was superior to two-dimensional echocardiography. Three-dimensional echocardiography promises to
be a more accurate method of estimating left ventricular volume
and mass and may be suitable for serial study of individual
subjects because of its improved accuracy and decreased interobserver variability compared with conventional echocardiographic methods. (Hypertension. 1994^3[suppl I]:I-172-I-179.)
Key Words • ultrasonic diagnosis • cardiac volume •
ventricular mass • image processing, computer-assisted
little use for reliable assessment of the effects of therapy in
individual patients.5 To overcome these limitations, we
have developed a three-dimensional echocardiograph.
Three-dimensional Echocardiography
The three-dimensional echocardiograph is composed
of a personal computer, a three-dimensional spatial
locater, and a conventional real-time echocardiographic
scanner. The personal computer (Gateway 2000/
486DX33, Gateway 2000, North Sioux City, SD) is used
to control system operation, acquire data, generate the
line of intersection display, and perform three-dimensional analysis and reconstruction. An acoustical spatial
locater (GP8-3D, Science Accessories, Stratford, Conn)
registers the position and orientation of the real-time
transducer, and thus its image, in three-dimensional
space. The user's real-time echocardiograph (HewlettPackard, Andover, Mass) supplies the video image and
an R-wave trigger signal to the computer. The spatial
locater is composed of a set of three sound emitters
rigidly attached to the imaging transducer and a set of
four point microphone receivers mounted in an overhead array. The emitters are activated in sequence,
generating 60 kHz sound waves that travel to the four
microphones. The time of flight is measured, and distances are computed. From these distances, the X, Y,
and Z coordinates of the images in a microphone-based
coordinate system are computed and associated with
each set of 16 images acquired at the same time. These
data are then used for concurrent generation of the line
of intersection display and subsequent three-dimensional reconstruction.
line of Intersection Display
The line of intersection display is a unique feature of
our three-dimensional echocardiograph. It allows the
operator to see the position and orientation of the
King et al Three-dimensional Echocardiography
1-173
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
FIG 1. A parastemal long-axis reference Image (left) shows the white line of intersection common to the long-axis image and the
intersecting real-time short-axis image (right).
real-time image in relation to anatomic landmarks in its
orthogonal, nonvisualized dimension (Fig 1). This capability allows the operator to accurately guide the positioning of these images. To create this display, a left
ventricular long-axis image is first acquired and saved as
a reference image. Subsequent real-time short-axis images approximately orthogonal to the reference image
intersect with it, defining a line common to both image
planes, the line of intersection. This line is computed
and displayed as a white line in each image—the
reference long-axis image and the short-axis real-time
image. As the real-time image is moved by the operator,
the line of intersection is rapidly recomputed and
redisplayed, showing the changing relation of the realtime image to anatomic landmarks in its orthogonal
dimension displayed in the reference image. The ability
to see this relation is very useful for guiding the
positioning of images for linear measurements such as a
ventricular minor (chordal) dimension. It also is useful
for accurately positioning end-plane images at the aortic
valve and ventricular apex when computing ventricular
volumes. In addition, it is useful for adjusting short-axis
image position to optimally image endocardial boundaries and wall motion. In addition, the line of intersection may be used to instruct technicians as well as
evaluate their performance.
Assessment of Two-dimensional
Standardized Imaging
The limitations of two-dimensional echocardiography
require the use of standardized images defined by
anatomic landmarks and assumptions regarding the
orientation of these images when these landmarks are
visualized. The structure of the heart, however, is a
complex three-dimensional shape. Variations of anatomy and operator technique may invalidate these assumptions, causing variations of image position and
orientation. These variations may limit accuracy and
reproducibility of measurements recorded from these
views. The line of intersection display, when not viewed
by the operator, may serve to assess actual image
orientation achieved by the operator during standardized two-dimensional imaging.6 In this study, 340 stan-
dard images were assessed for optimal positioning from
85 examinations performed by 11 echocardiographers at
three different institutions. The standard images were
the parasteraal left ventricular long-axis, short-axis at
the chordal level, and apical two- and four-chamber
views. Twenty-four percent of the conventional twodimensional unguided images were optimally positioned
within ±5 mm and ±15° of the standard. Two thirds of
these adequately positioned images were the parasternal long-axis image, but only two thirds of the parasternal long-axis images were optimally positioned. It is
evident from these results that twcndimensional echocardiography does not achieve a consistent and reasonable level of optimal image positioning. At one institution a subgroup of three echocardiographers who each
examined 10 subjects was subsequently evaluated using
the line of intersection display to guide image positioning. They followed the same imaging protocol and were
assessed by the same criteria previously established. By
guided three-dimensional imaging, 80% of their standard images were optimally positioned for image quality, displacement, and angulation/rotation. This threefold improvement compared with their performance
using unguided two-dimensional imaging was highly
significant by x2 analysis.
Guided One-dimensional Echocardiography
One third of the parastemal long-axis images obtained in the above assessment study were not optimally
positioned.6 This fact is especially significant because
standard one-dimensional measurements of the left
ventricle for estimating left ventricular volume and mass
are usually obtained from this image. This result suggests that mediolateral variation of the position of the
parastemal long-axis image may be a major source of
interobserver variability for these measurements. To
evaluate this possibility, we compared the interobserver
variability of anteroposterior measurements of the left
atrium and left ventricle made by conventional unguided two-dimensional imaging with those made by
three-dimensional line of intersection-guided imaging.7
Three pairs of operators at three different institutions
performed unguided two-dimensional and guided three-
1-174
Supplement I
Hypertension
Vol 23, No 1 January 1994
TABLE 1. Left Atrial and Left Ventricular Dimension
Variability With Unguided Two-dimensional and Guided
Three-dimensional Echocardiography
Difference Between Two Observers
Left Atrial
Dimension
Group
Left Ventricular
Dimension
Mean±SD, Variability, Mean ± SO, Variability,
%
mm
%
mm
Group A
Unguided
4.1 ±2.5
10.9
2.7±1.6
7.4
Guided
2.2±1.5
6.2
0.9±0.6
2.4
Unguided
5.6±5.3
17.9
3.4±2.8
9.5
Guided
1.1±1.0
3.4
0.8±0.6
2.1
Unguided
4.3±4.2
14.1
3.3±3.0
10.0
Guided
1.7±1.2
5.2
1.5+1.4
4.3
14.6
3.1 ±2.5
9.1
1.1 ±1.0
3.1t
Group B
Group C
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
Groups A, B, and C
Unguided
4.7±4.1
Guided
1.7±1.3
5.0*
2
*x 8tatlstte=9.5 (McNemar's test), P<.005.
t * 2 statistic=13.9 (McNemar's test), P<.005.
dimensional examinations on 10 patients each. In each
examination, the left atrium was measured in a plane
parallel to and through the inferior surface of the aortic
cusps, and the left ventricle was measured in a plane
perpendicular to its long axis 1 cm below the mitral
leaflet tips. For the unguided two-dimensional examination, the measurements were made in the parasteraal
long-axis image. For the guided three-dimensional examination, a reference long-axis image was used to
guide positioning of real-time short-axis images into the
atrial and ventricular measurement planes described
above. The measurements were then made on the
short-axis image through the center of the chamber and
thus were correctly positioned not only superoinferiorly
but also in the mediolateral direction. These results are
shown in Table 1. The standard unguided examination
was associated with an interobserver variability of
14.6% and 9.1% for the atrial and ventricular measurements, respectively. Thus, for the ventricular measurement, the standard error of the difference between two
replicate measurements by independent observers using
unguided two-dimensional echocardiography was 5.6
mm. In other words, in one third of the cases two
unguided examiners would be expected to obtain measurements of the same left ventricle that differ from
each other by more than 5.6 mm. Guided three-dimensional echocardiography, however, significantly reduced
interobserver variability to 5.0% and 3.1%, respectively,
for the same measurements (/><.005 by McNemar's
test). The standard error for the ventricular measurement by three-dimensional guided measurement was 2.0
mm, a nearly threefold reduction. This reduction was
highly significant (P<.005) and independent of differences in operator skill or practices among the three
institutional groups. These results thus demonstrate
that three-dimensional line of intersection guidance can
provide a basis for significantly improved serial evaluation and comparison of left ventricular one-dimensional
measurements by different operators.
Guided Two-dimensional Echocardiography
In a similar manner, line of intersection guidance can
be applied to positioning image planes for determination of left ventricular volume by the apical biplane
summation of disks algorithm. The assessment study
revealed that only 12% of unguided apical two- and
four-chamber views were optimally positioned through
the center of the ventricle and at a 90° angle. In
addition, it showed that two thirds of the four-chamber
views passed through the inferior septum rather than, as
expected, through the middle septum.6 These results
demonstrated that significant variability also occurs in
placement of the apical views and suggested the possibility that three-dimensional line of intersection-guided
positioning and orientation of these views could improve interobserver variability. To evaluate this, we
performed a study in 15 healthy subjects comparing
repeated determinations of ventricular volume by conventional unguided two-dimensional echocardiography,
line of intersection-guided two-dimensional echocardiography, and three-dimensional echocardiography using
polyhedral surface reconstruction. Interobserver variability for conventional unguided echocardiography was
8.5%, and for line of intersection-guided two-dimensional echocardiography it was 3.1%. 8 Thus, line of
intersection guidance of apical image position and orientation achieved a nearly threefold improvement of
reproducibility compared with unguided conventional
two-dimensional echocardiography. In a separate study,
the accuracy of left ventricular volume computation by
unguided apical views was compared with that obtained
by guided apical views using cineventriculography as a
standard. This study showed that there was no significant improvement of volume determination achieved by
use of image guidance.9 Several reasons for this were
apparent. First, guidance was unable to eliminate foreshortening of the ventricular image from the apical
window. The ventricular apex in a significant percentage
of the patients, probably about 50%, lies underneath a
rib, and the transducer cannot be placed directly over it,
thus resulting in some foreshortening of the image and
reduction in the apparent length of the ventricle and,
therefore, the volume. Also for the same reason, attempting to force the image into a correct position
resulted in a degradation of image quality because of rib
interference. This degradation resulted in greater
boundary-tracing errors. Furthermore, it was suggested
by the study that the geometric assumptions inherent in
the apical biplane summation of disks method are a
more significant source of error than image plane
positioning. Guided two-dimensional echocardiography
therefore can be expected to improve reproducibility if
used for measurement of ventricular volume or mass but
cannot be expected to improve overall accuracy. However, both of these —reproducibility and accuracy —
have been demonstrated to be improved by threedimensional echocardiography.
King et al Three-dimensional Echocardiography
1-175
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
Fra 2. A parasternal long-axis reference image over the base of the heart (left). A second parastemal long-axis reference image is
normally also acquired over the apex (right). The white lines are identical in both images and show the position and orientation of the
real-time short-axis images that were acquired to form the three-dimensional data set used for reconstructing the left ventricle.
Three-dimensional Echocardiography
Registration of all images in a three-dimensional
spatial coordinate system makes it possible to measure
not only within image planes but also between image
planes and, in addition, to perform a three-dimensional
reconstruction of the ventricle and compute its volume
without the use of geometric assumptions. We achieve
reconstruction of the ventricular surface using an algorithm called polyhedral surface reconstruction.10 In a
typical examination, 7 to 10 short-axis cross sections of
the ventricle are obtained from the inferior surface of
the aortic valve to the apex (Fig 2). These cross sections
are neither parallel nor evenly spaced but must not
intersect in the region of interest, the ventricle. The line
of intersection display is used to guide them to optimal
position and orientation to best display the endocardial
boundaries and to document the changing curvature of
the ventricular wall. The ventricular boundaries in each
image are traced manually by the operator. The computer then reconstructs the surface using the polyhedral
surface reconstruction algorithm. One hundred eighty
points are placed on each traced boundary, and pairs of
points on adjacent boundaries are connected by three
spans to form two triangles (Fig 3). Thus, 360 triangles
reconstruct the ventricular surface between each pair of
cross sections. To compute ventricular volume, centroids of each cross section are defined and connected
to the 180 boundary points. Each pair of triangles
connected to the centroids forms a wedge- or sectorshaped solid that is decomposed into three tetrahedrons
(Fig 4). The volumes of the tetrahedrons are computed
and summed to obtain ventricular volume. In addition,
total endocardial surface area is computed by summing
the areas of the surface triangles. And, if appropriate,
"infarct" (or wall motion abnormality) surface area and
its subtended volume may also be computed by demarcating the abnormal region on the traced boundaries. Of
course, myocardial volume and, therefore, ventricular
mass are obtained by subtracting endocardial volume
from epicardial volume.
Validation of Three-dimensional
Echocardiography
In vitro studies using a pin model have been carried
out to determine the system measurement accuracy as
nearly independent of operator error as possible. In
these studies, distances between image planes were
measured with a mean error of 0.4% of true value.
Distances within image planes, as would be measured by
conventional echocardiography, were, in contrast, measured with a mean error of 1.7%. This fourfold increase
is the result of the relatively poor lateral resolution of
the ultrasound beam. Three-dimensional volumes defined by the pin model were measured with an overall
mean error of 1.6%, or 0.64±0.72 mL of true value."
From these studies, we concluded that the intrinsic
system error was less than 0.4%, that it did not introduce any significant new errors compared with conven-
Fra 3. Drawing of two triangles used for surface reconstruction
between pairs of traced endocardial or epicardial boundaries.
1-176
Supplement I
Hypertension
Vol 23, No 1
January 1994
FRA 4. Drawings of centroids of each
cross section are computed and define a wedge- or sector-shaped solid
with each pair of surface triangles.
This wedge is then decomposed into
three tetrahedrons as shown, the volumes of which are computed and
summed to obtain total volume.
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
tional echocardiography, and that the principal source
of measurement error is poor lateral resolution resulting from ultrasound beam width. Measurement of the
volume of irregular water-filled balloon phantoms was
then carried out with an accuracy of 2.27%, an interobserver variability of 4.33%, and an SEE of 2.45 mL.10
Similar determinations of the same balloon phantoms
were carried out by magnetic resonance imaging and
found not to be significantly different. In another study,
distorted and nondistorted water-filled balloon phantoms were measured by three-dimensional echocardiography and cineventriculography, and when compared
with true values, the results also were found not to be
significantly different. Fixed, trimmed animal hearts
were used for validation of measurement of total and
infarct surface areas as well as ventricular mass. Accuracies for total and infarct surface areas were 1.61% and
3.48%, and interobserver variability was 1.92% and
2.37%, respectively. Fixed heart myocardial volume
and mass were measured with SEE values of 2.72 mL
and 2.73 g, respectively.
Comparison With Two-dimensional
Echocardiography In Vitro
The superiority of three-dimensional over two-dimensional echocardiography has been demonstrated in vitro.
Using distorted and nondistorted water-filled balloons
imaged by both methods, it was found that three-dimensional echocardiography had significantly smaller limits of
agreement (mean difference, ±2 SD) and higher repeatability than two-dimensional echocardiography.15 With
excised porcine left ventricles imaged by two-dimensional
echocardiography, three-dimensional echocardiography,
and cineventriculography, the measured volumes were
compared, and three-dimensional echocardiography was
found to have significantly less measurement error than
two-dimensional echocardiography and single-plane cineventriculography. Biplane cineventriculography resulted
in greater mean error than three-dimensional echocardiography, however, the difference was not significant by
statistical test16-17
Three-dimensional Echocardiography
in Humans
In a study of 15 healthy volunteers, three-dimensional
echocardiography was compared with magnetic resonance
imaging.18 In addition, in 10 of these subjects, two-dimensional echocardiography was also obtained. Three-dimensional (3D) echocardiographic measurement of left ventricular end-diastolic and end-systolic volumes (EDV and
ESV) compared favorably with magnetic resonance imaging (MRI): r=.92 and .81; SEE=6.99 and 4.01 mL;
regression equations: 3D EDV=0.84 (MRI EDV)+22.0
and 3D ESV=0.51 (MRI ESV)+18.2, respectively (Figs
5 and 6). Two-dimensional echocardiography compared
less favorably with magnetic resonance imaging: r= .48 and
.70; SEE=20.5 and 5.6 mL, respectively. The end-diastolic
volume SEE for two-dimensional echocardiography is
threefold that for three-dimensional echocardiography.
For total end-diastolic and end-systolic endocardial surface areas, three-dimensional echocardiography correlated very well with magnetic resonance imaging (r=.84
and .84; SEE=8.25 and 4.89 cm2). Interobserver variability for three-dimensional echocardiography ranged from
5% to 8%, and for magnetic resonance imaging from 6%
to 9%. In these same 15 subjects, left ventricular mass was
computed by three-dimensional echocardiography, and in
10 subjects by two-dimensional echocardiography, and
then compared with the mass obtained by magnetic resonance imaging. End-diastolic and end-systolic mass
(EDM and ESM) by three-dimensional echocardiography
compared favorably with magnetic resonance imaging:
King et al Three-dimensional Echocardiography
3D ECHO Intwota. V M W L • 4^9%
MRI IIIIMO4JC Vwisb. •
&90%
r - 0.92
, '
SEE-«.99ml
, '
p < 0.001
, "
/
/
'
/
•
•
/
y
y
/
/y-
y•
30 ECHO
Emt-dotofcioo
Vokmn
(mQ
y
Ti* ,-
30 ECHO EDV - O.M (Mni EDV) +
•
•
•
•
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
MRIEnd-cSMMIoVWifflW (ml)
FIG 5. Scatterptot of end-diastolic volume (EDV) determined by
three-dimensional echocardiography (3D ECHO) plotted against
end-dlastollc volume by magnetic resonance imaging (MRI).
Dashed lines represent 95% confidence intervals.
r=.89 and .92; SEE=10.6 and 9.24 g; regression equations:
3D E D M = (0.81)(MRI E D M ) + 11.6 and 3D
ESM=(0.87)(MRI ESM)+10.3, respectively. Mass by
two-dimensional echocardiography compared less favorably. r=.69 and .72; SEE=21.8 and 23.8 g. The SEE for
mass determination by two-dimensional echocardiography
is approximately twofold that by three-dimensional echocardiography. These results are summarized in Table 2.
Comparison of left ventricular volume determination
by three-dimensional echocardiography and two-dimensional echocardiography with cineventriculography was
also carried out in 14 patients undergoing cardiac
catheterization.20 The correlation coefficients were
higher and the SEE values lower for three-dimensional
30 ECHO EBV - 0J1 (Mfd ESV)
3D ECHO Intante. VMab. MHl Marat*. VMab.
r-O.SI
SEE • 4.01 ml
p < 0.001
30 ECHO
Emf-cyttoOc
(ml)
1-177
echocardiography versus cineventriculography end-diastolic volume, end-systolic volume, and ejection fraction
than they were for two-dimensional echocardiography
versus cineventriculography. The end-diastolic volume
SEE for two-dimensional echocardiography is almost
twice that of three-dimensional echocardiography.
These results are summarized in Table 3.
Other Investigations of
Three-dimensional Echocardiography
Other studies have used a similar approach to transthoracic three-dimensional echocardiography. An
acoustic spatial locating system has been used in conjunction with a reconstruction algorithm to compute
ventricular volume in several in vitro investigations.21"24
These studies largely confirm the results reported
above. Siu and coworkers21 investigated in an in vivo
canine model the minimum number of images necessary
to accurately reconstruct the ventricle. They concluded
that in a large percentage of cases, left ventricular
volume could be accurately quantitated with 8 to 12
images. The same authors have compared three-dimensional echocardiographic volume determinations with
two-dimensional determinations in excised left ventricles of normal and abnormal shapes and found that
three-dimensional echocardiogTaphy had the lowest
SEE. 2224 In another study, they compared in excised left
ventricles three-dimensional echocardiographic volume
determination with angiographic volume determination
and found the SEE of three-dimensional echocardiography to be approximately half that of biplane angiocardiography.25 They have also used an in vivo canine
preparation in which the left ventricular volume could
be varied and continuously measured. Volume computation by three-dimensional echocardiography in this
model was also shown to be highly accurate26 and when
compared with conventional two-dimensional echocardiographic methods to be superior with a lower SEE.27
Jiang and coworkers28-30 have shown the usefulness of
three-dimensional echocardiography for evaluation of
right ventricle volume and mass. Mele and coworkers31
have shown in beating and excised dog hearts that
three-dimensional echocardiography is able to estimate
left ventricular mass more accurately than wall thickness methods. Three-dimensional reconstruction of the
heart is also being carried out by other investigators
using transesophageal echocardiography for image acquisition.32J3 However, this approach will be limited to
selected patients, is not as widely applicable as transthoracic three-dimensional reconstruction, and is not
within the scope of this review.
Limitations of
Three-dimensional Echocardiography
MRI End-«y«to«c VWum* (ml)
FIG 6. Scatterptot of end-systolic volume (ESV) determined by
three-dimensional echocardiography (3D ECHO) plotted against
end-systolic volume by magnetic resonance Imaging (MRI).
Dashed lines represent 95% confidence intervals.
Three-dimensional echocardiography adds to the information available for analysis by registering all images
in a single spatial coordinate system. This enables the
line of intersection display to reduce image plane positioning errors, permits measurements between image
planes, and eliminates geometric assumptions for volume and mass computation. Nevertheless, the process
of three-dimensional data reduction and analysis at the
present time requires the ventricular boundaries to be
manually traced by the operator, as does two-dimen-
1-178
Supplement I Hypertension
Vol 23, No 1 January 1994
TABLE 2. Left Ventricular Volume and Mass by Three-dimensional and Two-dimensional
Echocardiography Versus Nuclear Magnetic Resonance Imaging
r
SEE
P
Regression Equation
15 Normal subjects
3D vs MRI EDV
.92
6.99 mL
<.001
3D=0.84 (MRI)+22.0
3D vs MRI ESV
.81
4.01 mL
<.001
3D=0.81 (MRI) + 18.2
3D vs MRI EDM
.89
<.001
3D=0.81 (MRI) + 11.6
<.001
3D=0.87 (MRI)+ 10.3
10.6 g
.92
9.24 g
2D vs MRI EDV
.48
20.5 mL
NS
2D=0.57(MRI) + 17.1
2D vs MRI ESV
.70
5.6 mL
<.O25
2D=0.44(MRI)+3.8
3D vs MRI EDV
.90
7.0 mL
<.001
3D=0.72 (MRI)+29.7
3D vs MRI ESV
.88
3.1 mL
<.001
3D=0.45 (MRI)+20.2
3D vs MRI ESM
10 Normal subjects
10 Normal subjects
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
2D vs MRI EDM
.69
21.8 g
<.O3O
2D=0.76 (MRI)+27.7
2D vs MRI ESM
.72
23.8 g
<.020
2D=0.91 (MRIJ+13.0
3D vs MRI EDM
.88
10.4 g
<.OO1
3D=0.72 (MRI)+32.2
3D vs MRI ESM
.90
10.8 g
<.001
3D=0.86(MRI) + 13.2
3D Indicates three-dimensional echocardiography; 2D, two-dimensional echocardiography; MRI, nuclear magnetic resonance imaging; EDV, end-diastolic volume; ESV, end-systolic volume; EDM, end-dlastollc mass; and
ESM, end-systolic mass.
sional echocardiography. This step creates the potential
for human error and interobserver variability. A number of approaches to automatic or semiautomatic
boundary detection are under development but have
not been established as acceptable for general application.34 Acquisition of a set of images for three-dimensional reconstruction takes only 6 to 8 minutes and may
be easily incorporated into the normal examination
routine. However, the process of manual boundary
tracing adds 1 or 2 minutes to analysis time for each
boundary traced, depending on the image quality. For a
typical ventricular mass determination using eight images, there would be 16 traced boundaries. Each image
sequence is first viewed as a cine loop to display the
boundary in motion, and then the selected frame is
traced. Experience has shown that accuracy and reproducibility are optimized by tracing on the "white side"
of the black-white boundary. The three-dimensional
echocardiographic system is available at reasonable cost
as an add-on to existing echocardiographs and operates
without affecting the host system. Its operation is guided
TABLE 3. Left Ventricular Volume by Three-dimensional
and Two-dimensional Echocardiography
Versus Clneventriculography
3D vs CINE
2D vs CINE
r
SEE
r
SEE
EDV
.88
16 mL
.76
28 mL
ESV
.96
9mL
.72
29 mL
EF
.84
0.07
.55
0.12
3D Indicates three-dimensional echocardiography; CINE, cineventriculography; 2D, two-dimensional echocardiography; EDV,
end-diastollc volume; ESV, end-systotic volume; and EF, ejection traction.
by menus displayed on the computer screen and is easily
learned. Use of the line of intersection display requires
the operator to learn new eye-hand coordination skills
that are generally achieved rather quickly.
Implications for
Quantitative Echocardiography
Conventional echocardiography has been and remains an examination providing primarily qualitative
information despite attempts to introduce standardized
quantitation.35 These attempts have failed because the
underlying technology, and therefore the data it produced, was imprecise and poorly reproducible in routine
practice. Three-dimensional echocardiography is a technological step forward addressing some of the underlying causes of that imprecision and poor reproducibility,
namely, the lack of registration of image position and
orientation, the need to use geometric assumptions, and
the inability to visualize anatomic landmarks in directions orthogonal to the real-time image. Although these
do not account for all variability in conventional echocardiography, they have been major contributors, and
investigations using three-dimensional echocardiography have suggested that their elimination leads to
substantial improvement in quantitative results. A
threefold improvement in interobserver variability for
simple measurements and for three-dimensional volume computation has been achieved along with a twofold to threefold improvement of accuracy for volume
computation by three-dimensional reconstruction.
Comparison with magnetic resonance imaging and cineventriculography suggests that three-dimensional echocardiography provides comparable data noninvasively,
repeatedly, and at lower expense with greater patient
acceptance. Additional studies will be required for
validation, but the available data strongly suggest that
King et al Three-dimensional Echocardiography
three-dimensional echocardiography may be sufficiently
accurate and reproducible to monitor small biologic
changes, such as the regression of left ventricular mass,
in individual subjects. We expect that by the year 2000
all echocardiographic imaging will incorporate, where
clinically indicated, three-dimensional scanning, image
reconstruction, and quantitative analysis.35
18.
19.
References
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
1. King DL, Al-Banna SJ, Larach DR. A new three-dimensional
random scanner for ultrasonic/computer graphic imaging of the
heart. In: White DN, Barnes R, eds. Ultrasound in Medicine, Vol 2.
New York, NY: Plenum Publishing Corp; 1976:363-372.
2. King DL, King DL Jr, Shao MYC. Three-dimensional spatial
registration and interactive display of position and orientation of
real-time ultrasound images. J Ultrasound Med. 1990,9:525-532.
3. Schiller NB, Acquatella H, Ports TA. Left ventricular volume from
paired biplane two-dimensional echocardiography. Circulation.
1979;60:547-555.
4. Gueret P, Meerbaum S, Wyatt HL, Uchiyama T, Lang T, Corday
E. Two-dimensional echocardiographic quantitation of left ventricular volumes and ejection fraction. Circulation. 1980;62:
1308-1318.
5. Gottdiener JS, Livengood SV, Meyer PS. Should echocardiography
be performed in individual patients to assess LVH regression?
Test-retest reliability of echocardiography for assessment of LV
mass, ./vim Coll CardioL 1993;21:180A. Abstract.
6. King DL, Harrison MR, King DL Jr, Gopal AS, Kwan OL,
DeMaria AN. Ultrasound beam orientation during standard twodimensional imaging: assessment by three-dimensional echocardiography. J Am Soc Echocardiog. 1992^:569-576.
7. King DL, Harrison MR, King DL Jr, Gopal AS, Martin RP,
DeMaria AN. Improved reproducibility of left atrial and left ventricular measurements by guided three-dimensional echocardiography. J Am CoU CardioL 1992;20:1238-1245.
8. Schroeder KM, Sapin PM, King DL, Smith MD, Kwan OL,
DeMaria AN. Improved reproducibility of left ventricular volume
measurements using guided two dimensional and three dimensional echocardiography. Circulation. 1992;86(suppl I):I-271.
Abstract.
9. Sapin PM, Schroeder KM, King DL, Kwan OL, Xie G, DeMaria
AN. Three dimensional guiding of apical biplane views fails to
improve the accuracy of echocardiographic left ventricular volume
determinations. J Am Soc Echocardiog. 1993;6(pt 2):S37. Abstract.
10. Gopal AS, King DL, Katz J, Boxt LM, King DL Jr, Shao MYC.
Three-dimensional echocardiographic volume computation by
polyhedral surface reconstruction: in vitro validation and comparison to magnetic resonance imaging. J Am Soc Echocardiog.
1992^:115-124.
11. King DL, King DL Jr, Shao MYC. Evaluation of in vitro measurement accuracy of a three-dimensional ultrasound scanner.
J Ultrasound Med. 1991;10:77-82.
12. Schroeder KM, Sapin PM, King DL, DeMaria AN. Threedimensional echocardiographic volume computation: an in vitro comparison with cineventriculography. Circulation. 1992;86(suppl I):
1-271. Abstract.
13. King DL, Gopal AS, King DL Jr, Shao MYC. Three-dimensional
echocardiography: in vitro validation for quantitative measurement
of total and 'infarct' surface area. J Am Soc Echocardiog. 1993;6:
69-76.
14. Gopal AS, King DL, Bielefeld MR, Cabreriza S, Spotnitz HM.
Left ventricular mass and volume of fixed hearts by threedimensional echocardiography. Circulation. 1991;84(suppl II):
11-684. Abstract.
15. Schroeder K, Sapin P, King DL, Smith M, DeMaria AN. Is threedimensional echocardiography superior to two-dimensional echocardiography for volume determination? / Am Soc Echocardiog.
1992;5:315. Abstract.
16. Sapin P, Schroeder K, King DL, Smith M, DeMaria AN. Comparison of three-dimensional echocardiography, two-dimensional
echocardiography and biplane angiography for left ventricular volume determination. J Am Soc Echocardiog. 1992;5:316.
Abstract.
17. Sapin PM, Schroeder KM, Smith MD, DeMaria AN, King DL.
Three dimensional echocardiographic measurement of left ventricular volume in vitro: comparison to two-dimensional echocar-
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
1-179
diography and cineventriculography. J Am Coll CardioL 1993;22:
1530-1537.
Gopal AS, Keller AM, Rigling R, King DL Jr, King DL. Left
ventricular volume and endocardial surface area by threedimensional echocardiography: comparison to two-dimensional
echocardiography and magnetic resonance imaging in normal
subjects. JAm CoU CardioL 1993;22:258-270.
Keller AM, Gopal AS, Gerasimou E, Rigling R, King DL Threedimensional echocardiography: in vivo measurement of left ventricular mass in humans. J Am Coll CardioL 1993;21:180A.
Abstract.
Sapin PM, Schroeder KM, King DL, Kwan OL, Xie G, DeMaria
AN. Is three-dimensional echo superior to two-dimensional echo
for LV volume determination in patients? Comparisons to cineventriculography. Circulation. 1992;86(suppl I):I-270. Abstract.
Siu SC, Rivera JM, Handschumacher MD, Weyman AE, Picard MH.
Left ventricular volume quantitation by three-dimensional: simplification without loss of accuracy. Circulation. 1992;86(suppl I):I-270.
Abstract.
Siu SC, Rivera JM, Xie SW, Lethor JP, Thomas JD, Handschumacher MD, Weyman AE, Levine RA, Picard MH. Threedimensional echocardiography improves noninvasrve left ventricular volume quantitation. J Am Coll CardioL 1992;19:18A.
Abstract.
Siu SC, Rivera JM, Xie SW, Lethor JP, Vlahakes GJ, Thomas JD,
Handschumacher MD, Weyman AE, Levine RA, Picard MH. The
utility of three-dimensional echocardiography in assessing left ventricular volume. J Am Soc Echocardiog. 1992^5:315. Abstract.
Handschumacher MD, Lethor JP, Siu SC, Mele D, Rivera JM,
Picard MH, Weyman AE, Levine RA. A new integrated system for
three-dimensional echocardiographic reconstruction: development
and validation for ventricular volume with application in human
subjects. JAm CoU CardioL 1993;21:743-753.
Hibberd MG, Siu SC, Handschumacher MD, Jiang L, Newell JB,
Levine RA. Three-dimensional echocardiography: increased
accuracy for left ventricular volume compared with angiography.
Circulation. 1992;86(suppl I):I-270. Abstract.
Siu SC, Rivera JM, Guerrero JL, Handschumacher MD, Xie SW,
Weyman AE, Levine RA, Picard MH. Three-dimensional echocardiography: in vivo validation for left ventricular volume and
function. JAm CoU CardioL 1992;19:18A. Abstract.
Siu SC, Levine RA, Xie SW, Handschumacher M, Weyman AE,
Picard MH. Three-dimensional echocardiography improves in vivo
quantitation of left ventricular volume. J Am CoU CardioL 1993;
21:236A. Abstract.
Jiang L, Handschumacher MD, Hibberd MG, Siu SC, King ME,
Weyman AE, Levine RA. Three-dimensional reconstruction of
right ventricular volume: in vitro comparison to two-dimensional
methods. J Am Soc Echocardiog. 1992;5:316. Abstract.
Jiang L, Siu SC, Handschumacher MD, Guerrero JL, Prada JV, King
ME, Picard MH, Weyman AE, Levine RA. Three-dimensional echocardiography: in vivo validation for right ventricular volume and
function. Circulation. 1992;86(suppl I):I-272. Abstract
Jiang L, Prada JAV, Handschumacher MD, Guerrero JL, Vlahakes
GJ, King ME, Weyman AE, Levine RA. Three-dimensional echocardiography: in vivo validation for right ventricular free wall mass as an
index of hypertrophy. J Am CoU CardioL 199321368A. Abstract.
Mele D, Jiang L, Handschumacher MD, Siu SC, Vandervoort PM,
Vlahakes GJ, Guerrero JL, Picard MH, Svizzero T, Weyman AE,
Levine RA. Three-dimensional echocardiography can accurately
measure left ventricular mass: in vivo and in vitro validation. J Am
Coll CardioL 1993;21:332A. Abstract.
Gerber TC, Foley DA, Belohlavek M, Seward JB, Greenleaf JF.
Three-dimensional echocardiographic determination of ventricular volume: comparison of three imaging techniques. JAm CoU
CardioL 1993;21:180A. Abstract.
Azevedo J, Cao QL, Pandian N, Schwartz S, Klas B. Quantification
of left ventricular volumes by automated three-dimensional echocardiography using a computer-controlled tomographic ultrasound
probe: in vitro validation. J Am CoU CardioL 1993;21:180A.
Abstract.
Sapin PM, Xie G, Smith MD, Kwan OL, Cater AC, Souther SK,
Reed SK, Shahab T, DeMaria AN. Peak emptying and peak filling
rates derived from automatic border detection echocardiography:
comparison with cineventriculography. J Am CoU CardioL 1993;
21:297A. Abstract.
DeMaria AN. Future developments in cardiac ultrasound: possibilities and challenges. Am J CardioL 1992;69:2H-5H.
Three-dimensional echocardiography. Advances for measurement of ventricular volume and
mass.
D L King, A S Gopal, A M Keller, P M Sapin and K M Schröder
Hypertension. 1994;23:I172
doi: 10.1161/01.HYP.23.1_Suppl.I172
Downloaded from http://hyper.ahajournals.org/ by guest on June 16, 2017
Hypertension is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1994 American Heart Association, Inc. All rights reserved.
Print ISSN: 0194-911X. Online ISSN: 1524-4563
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://hyper.ahajournals.org/content/23/1_Suppl/I172
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Hypertension 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 Hypertension is online at:
http://hyper.ahajournals.org//subscriptions/