DOI:10.2214/AJR.04.0857
AJR 2006; 187:1092-1106
© American Roentgen Ray Society
Live 3D Echocardiography: A Replacement for Traditional 2D Echocardiography?
Robin C. Houck1,2,
Jason E. Cooke1,3 and
Edward A. Gill1
1 Department of Medicine, Division of Cardiology, University of Washington
School of Medicine, Harborview Medical Center, Box 359748, 329 Ninth Ave.,
Seattle, WA 98104-2599.
2 Present address: Medicine Department, University of Vermont, Burlington,
VT.
3 Present address: Medtronic, Inc., Minneapolis, MN.
Received June 1, 2004;
accepted after revision April 10, 2006.
Address correspondence to E. A. Gill.
Abstract
OBJECTIVE. We describe the development of real-time 3D imaging and
review the previously used versions of 3D echocardiography so that the reader
will appreciate why current developments truly do represent a quantum leap in
the technology.
CONCLUSION. Three-dimensional echocardiography has now been shown to
have several advantages over 2D echocardiography, particularly for volume
measurements, visualization of septal defects, and whole-valve evaluation.
Given these data, it is clear that 3D echocardiography is here to stay and
soon will become part of routine echocardiographic examinations.
Keywords: cardiovascular imaging • congenital malformations • echocardiography • heart
Introduction
Echocardiography, or sonography of the heart, is the most used imaging test
for diagnostics of the heart. At present, the mainstay of echocardiography is
2D imaging, also known as B-mode imaging. The limitations of 2D
echocardiography, especially from the stand-point of left ventricular
quantification, are that many of the formulas for calculation of the left
ventricular ejection fraction and volume are based on assumptions that do not
necessarily hold true in the setting of dilated, failing ventricles that
become more spherical as the disease process progresses. In addition, one
reason that interobserver variability exists in 2D echocardiography
interpretations is that individual observers interpolate the data between 2D
images in different ways.
Three-dimensional echocardiography was developed partly to address these
limitations. Until recently, however, 3D echocardiography has undergone rather
slow and arduous improvements. In the past, image quality was often marginal,
and a more important limitation was that acquisition of the images was, at
best, cumbersome. Recently, there has been the equivalent of a quantum leap in
3D imaging with the development of a type of real-time 3D imaging called
"live 3D" imaging in the most advanced currently available
version. Here, we describe the development of real-time 3D imaging and review
the previously used versions of 3D echocardiography so that the reader will
appreciate why current developments truly do represent a quantum leap in the
technology.
Three-Dimensional Echocardiography Techniques
Definitions
As 3D echocardiography looms on the horizon ever so close to routine
everyday use in clinical practice, it is important to define the currently
marketed techniques and appreciate the evolution of the previous approaches.
Throughout the past 30 years, attempts to develop 3D echocardiography have
used a transthoracic echocardiography approach with acquisition of a data set
consisting of many (i.e., 15-61) adjacent 2D images and then volume rendering
the 2D images into a 3D image with the aid of computer interpolation. The
requirement for this type of 3D rendering is to be able to align the 2D images
together, and this requires some method for the computer to register all of
the images and ultimately align them. To accomplish this feat, the transducer
has been located or tracked in space or rotated around a fixed axis so that
the assortment of 2D images acquired could then be properly aligned to
reconstruct a 3D image.
Although tracking and locating are virtually synonymous, for purposes of
this discussion, "tracking" the transducer is defined by methods
that determine the transducer's location by mechanical means—that is,
placing the transducer on some type of mechanical arm so the location is
tracked manually by rotating, fanning, or stepping the transducer around a
fixed point. Conversely, "locating" the transducer is defined as
sending a signal (such as audio or magnetic) from the transducer to a
calibrated device that can use the information to determine the transducer's
location on a 3D grid.
View larger version (68K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1 —Diagram shows relationship of transesophageal
echocardiography probe to heart in example of use of rotational sweep to
acquire 2D images for a 3D data set. (Courtesy of Philips Medical Systems
[Andover, MA and Bothell, WA] and TomTec Imaging Systems GmbH [Munich,
Germany])
|
|
Almost all of the tracking and locating techniques have been largely
abandoned, first, because of complexity and, second, because the image quality
was often poor and not acceptable for making diagnostic decisions. The only
surviving remnant of this type of approach was developed by Hewlett Packard
(spun off as Agilent Technologies and purchased by Philips Medical Systems)
and TomTec Imaging Systems GmbH; it uses a transesophageal echocardiography
(TEE) probe for acquisition with transducer tracking accomplished by rotation
around a fixed axis, with that axis being the center of the rotating
transducer on the very tip of the TEE probe
(Fig. 1). Although currently
only Siemens markets this TEE approach as a solution for 3D, this technology
can be adapted to any cardiac ultrasound system but does require partnership
with TomTec Imaging Systems GmbH.
All of the techniques referred to so far require reconstruction of many 2D
images into a 3D image. Reconstruction involves scan conversion, which means
the scanned data must be converted into a 3D cartesian grid followed by volume
rendering. Hence, this technique is far from real-time 3D imaging, defined as
imaging and volume rendering simultaneously because, first, acquisition of
each 2D image requires a separate cardiac cycle and, second, reconstruction
takes additional time. All of these older techniques are referred to from this
point on as "reconstructive 3D imaging."
The quantum leap that has recently happened in the development of 3D
echocardiography is the introduction of real-time 3D imaging using a
transthoracic approach. As the term implies, "real-time" 3D
imaging takes place on the fly, with reconstruction performed simultaneously
with imaging via a PC on the sonography machine. Real-time 3D imaging was
first developed at Duke University, and Volumetrics Corporation attempted to
market the first version. This version of real-time 3D imaging never made an
impact on clinical practice because image quality was limited. Recently, a
more advanced version of real-time 3D echocardiography was introduced by
Philips Medical Systems and is called "live 3D." Because of
improved image quality, this type of 3D imaging is currently making inroads
into both the research and clinical arenas
(Table 1).
Past Techniques
Here, we review briefly the old reconstructive techniques that used
transthoracic echocardiography leading up to the newest form of reconstructive
3D using TEE. Dekker et al. [1]
in 1974 are credited with performing 3D echocardiography first. With that
technique, the patient was scanned using a transducer that was connected to a
mechanical arm that fed quadrant data to a computer mainframe
(Fig. 2).
View larger version (167K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2 —Mechanical arm, first described by Dekker et al.
[1] in 1974, is shown.
Transducer (arrow) is held on mechanical arm. Location of transducer
is known because it is attached to mechanical arm.
|
|
Moritz and Shreve [2]
developed an acoustic locater (the so-called "spark gap") whereby
an acoustic sound emitter produced 60-kHz sound waves that traveled to each of
four overhead microphones. The time taken for sound to travel from each
emitter to each microphone was then measured, corrected for environmental
conditions, and used to calculate a range between the two points. From these
ranges, the x, y, and z cartesian coordinates of the
transducer and, subsequently, its image were computed in a spatial coordinate
system. In Figure 3, the device
that actually was attached to the transducer and emitted sound is shown.
Later, Raab et al. [3]
developed a magnetic locator that attached to the transducer and was attached
via thin cabling to a central computer that located the transducer in space
and subsequently followed its position. This development led to so-called
"freehand" scanning whereby the person doing the scanning did not
have to move the transducer in any certain way but, rather, any movement of
the transducer could be followed by the magnetic locator
[4]
(Fig. 4). Using this method, a
magnetic field was created by a rectangular magnetic box
(Fig. 4) that was typically
placed underneath the bed the patient was lying on. A sensor attached to the
transducer was connected via a microcomputer to the magnet and allowed sensing
of the transducer in space.
View larger version (40K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4 —Diagram illustrates freehand technique. Transducer is tracked
via a magnetic field created by magnet enclosed within cube. See Past
Techniques section for further discussion. (Courtesy of TomTec Imaging Systems
GmbH, Munich, Germany)
|
|
Between the acoustic spark gap technology and freehand methods, the
predominant method for 3D acquisition used an automated rotational device. In
this rendition of reconstructive 3D imaging, the transducer was held in a
cylindric device and rotated in a circular pattern around a fixed axis
(Fig. 5). In this case, the
transducer position was the same throughout the acquisition, so it did not
need to be located, but the axis of rotation could not move. Similar less
popular methods compared with the rotational devices were the fan method, for
which the transducer was moved in a predetermined fan-shaped direction around
a fixed axis, and finally, the linear method, for which the transducer was
moved in a straight line on a track using a stepper motor (Figs.
6A and
6B).
View larger version (84K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5 —Rotational device designed to hold a transducer is shown.
Transducer, an older mechanical type, is shown protruding from bottom of
rotator (arrow). (Courtesy of TomTec Imaging Systems GmbH, Munich,
Germany)
|
|
View larger version (37K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6A —Techniques used for 3D image acquisition protocols. Diagrams
show fanning (A) and linear or stepper (B) motor devices used
for transthoracic 3D image acquisition protocol. Fanning device follows fan
motion, and linear device, by definition, follows linear tracking protocol.
(Courtesy of TomTec Imaging Systems GmbH, Munich, Germany)
|
|
View larger version (30K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6B —Techniques used for 3D image acquisition protocols. Diagrams
show fanning (A) and linear or stepper (B) motor devices used
for transthoracic 3D image acquisition protocol. Fanning device follows fan
motion, and linear device, by definition, follows linear tracking protocol.
(Courtesy of TomTec Imaging Systems GmbH, Munich, Germany)
|
|
In the past few years, attempts were made to bring reconstructive
transthoracic 3D imaging to market; it was the freehand methodology that was
used predominantly for transthoracic imaging, although the rotational method
enjoyed some popularity as well, and these two technologies continued to
develop side by side. In fact, a rotational device was developed by Hewlett
Packard in which the entire rotation apparatus was contained within the
transducer. Hence, an external transducer holder was no longer necessary
(Fig. 7). This led naturally to
using the same approach with the Hewlett Packard TEE probe
(Fig. 1); indeed, in 1995, the
first reconstructive 3D images using TEE were obtained, and in the same year,
our laboratory, then located at the University of Colorado, also performed 3D
TEE, and we began to apply this technology for studying stenotic mitral valves
before percutaneous mitral valvuloplasty (Figs.
8A,
8B,
8C,
8D,
8E,
8F, and
8G).
View larger version (139K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7 —Photograph shows transducer designed with rotation
capabilities contained within housing. End result is similar to
Figure 5, but rather than
having a separate device to hold transducer, transducer itself has electronic
steering that steers in rotational pattern. Design is also similar to
transesophageal echocardiography transducer shown in
Figure 1. (Courtesy of Philips
Medical Systems, Andover, MA and Bothell, WA)
|
|
View larger version (107K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8A —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Three-dimensional image
of stenotic mitral valve (arrows, A) from left ventricular
side of valve (A) and corresponding 2D cut (B) are shown.
|
|
View larger version (68K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8B —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Three-dimensional image
of stenotic mitral valve (arrows, A) from left ventricular
side of valve (A) and corresponding 2D cut (B) are shown.
|
|
View larger version (83K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8C —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Atrial septal
defect—ASD in C and arrow in D—is shown on 3D image
(C) in its entirety as opposed to corresponding 2D cut (D). In
C, AoV refers to aortic valve.
|
|
View larger version (73K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8D —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Atrial septal
defect—ASD in C and arrow in D—is shown on 3D image
(C) in its entirety as opposed to corresponding 2D cut (D). In
C, AoV refers to aortic valve.
|
|
View larger version (130K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8E —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Three-dimensional image
shows aortic valve (arrow) with focal sclerosis of commissure between
the non and right coronary cusp.
|
|
View larger version (73K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8F —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Two-dimensional image
shows flail posterior mitral valve leaflet (arrow).
|
|
View larger version (137K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8G —Three-dimensional images obtained using rotational
transesophageal echocardiography acquisition and subsequent reconstruction.
All are designed to show benefits of en face view of entire structure with
depth as opposed to 2D image showing only single cut. Three-dimensional image
of mitral valve in systole shows middle scallop of posterior leaflet.
|
|
Newest Techniques: Real-Time and Live 3D Imaging
First rendition of real-time 3D imaging— By the end of the
1990s, research at Duke University led by Von Rahm produced a transthoracic
echocardiography probe that could acquire a volume of data rather than the
single slice and ultimately could display a real-time 3D image
[5-9].
This transthoracic probe had its elements configured in a sparse array pattern
(Fig. 9). A transducer is
considered fully sampled if the entire transducer crystal is used to transmit
and receive the ultrasound beam. It is possible to use less than the entire
transducer crystal to perform this function. Hence, such arrays are considered
sparsely sampled arrays or sparse arrays and the entire face of the transducer
is not connected to individual elements; rather, elements are intermittently
dispersed in an organized fashion throughout the face of the transducer. The
result is that only 128 elements are used for transmit and receive
functions.
View larger version (30K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 9 —Diagram illustrates real-time 3D transthoracic
echocardiography probe with sparse array design. Note comparison with dense
array shown in Figures 11A and
11B. See First-Rendition of
Real-Time 3D Imaging section for discussion. (Courtesy of Philips Medical
Systems, Andover, MA and Bothell, WA)
|
|
By distributing these active elements across the 2D face of the transducer,
it is possible to steer and focus the scan lines in both the elevation and
lateral directions. The use of the sparse array limits the beam-formation
process and can lead to significant limitations in image quality. The
advantage, however, is that because only 128 active elements are in use, the
sparse array can be connected directly to a 128-channel system. A fully
sampled array that would use approximately 2,500 elements cannot be directly
connected to a 128-channel ultrasound system. Such a transducer requires the
development of a microbeam former, as described later in this article. In the
end, although the system developed at Duke University and marketed by
Volumetrics could produce 3D images, they were of limited quality because of
the sparse array limitation. In fact, the system was mostly used to produce 2D
images from the 3D data set, thus producing the term and accompanying images
called "C scans" (Fig.
10). C scans are orthogonal images to the long axis of the heart.
Hence, short-axis cuts of the heart from the apex to the base could be
generated from the 3D data set.
View larger version (93K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 10 —Standard 2D echocardiography images of left ventricle
(right) and corresponding cross-sectional views (left), or C
scans. Four-chamber (top) and two-chamber (bottom) views are
shown. This figure is from older version of real-time 3D imaging. (Courtesy of
D. Adams, Durham, NC)
|
|
Live 3D imaging—In November of 2002, live 3D imaging, an
advanced form of real-time 3D imaging, was introduced by Philips Medical
Systems. The major advance that this latest version of real-time 3D imaging
immediately offered was improved image quality due to a fully sampled or dense
array configuration of the transducer. The face of the transducer is
completely sampled (Figs. 11A
and 11B). This dense array
transducer, called a "matrix array," consists of approximately
3,000 elements.
View larger version (41K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 11A —Images show schematic and photograph of dense array real-time
3D transducer. (Courtesy of Philips Medical Systems, Andover, MA and Bothell,
WA) Schematic representation of dense array real-time 3D transducer. Each
square represents an element. See First-Rendition of Real-Time 3D Imaging
section for discussion.
|
|
View larger version (67K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 11B —Images show schematic and photograph of dense array real-time
3D transducer. (Courtesy of Philips Medical Systems, Andover, MA and Bothell,
WA) Photograph shows live 3D matrix transducer (Matrix X4) manufactured by
Philips Medical Systems.
|
|
The major challenge during development of the dense array transducer was
interfacing all of the elements of the probe to the 128-channel ultrasound
system; this required the development of a microbeam former. The microbeam
former is located in the head of the transducer and provides electronic
control of the transmit and receive functions for the approximately 3,000
elements. The elements are divided into subgroups, and each subgroup is
connected to one of the 128 channels of the ultrasound system. Significant
technical advances were required for the electronics to be miniaturized into
the head of the transducer and allow the interconnection between the elements
and the electronics. This miniaturization process was equivalent to placing
150 circuit boards into the head of the transducer.
Live 3D imaging is based on the premise that a 3D image can be created
using ultrasound waves by scanning an ultrasound beam through a 3D volume.
Such scanning can be done mechanically by physically reorienting the
transducer, as described earlier in previous methods. The disadvantage of this
approach, as previously emphasized, is that the mechanical apparatus places a
tremendous limitation on the speed with which the ultrasound beam can be
scanned. An alternative approach is to scan the beam electronically using a
phased-array transducer. Traditional beam steering for 2D echocardiography
scans (or B-mode scans) is currently performed by a one-dimensional (1D)
transducer that is steered by a phased-array scanner
(Fig. 12).
View larger version (30K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 12 —Diagram depicts transition from one-dimensional (1D) to 3D
data set. Note single scan line from single crystal 1D or A-mode scan
transitioning to axial and lateral imaging with 2D phased-array scanning and
to axial, lateral, and elevation scanning with 3D matrix scanning. See further
discussion in Live 3D Imaging section. C scan is essentially an axial cut
through 3D data set. (Illustration by Richard Gersony)
|
|
To scan the beam over the lateral (azimuth) and elevation dimensions that
are required for a 3D scan necessitated the development of a 2D phased-array
transducer [10]. With
electronic steering of such a 2D phased-array transducer, the scanning could
be done in any pattern. A typical pattern, and the one used to obtain live 3D
images currently, is scanning a series of 2D slices that are displaced over a
series of elevation angles. An advantage of this pattern is that spatially
adjacent areas are temporally adjacent. When this complete pattern has been
scanned, the result is a pyramidal shape, called a "frustum"
(Fig. 12). It is not a true
pyramid because its base is curved.
From this point, a number of alternatives about how to use and display this
volume of data exist. One alternative is to extract from the frustum an
arbitrary 2D slice. Such a slice is called a "multiplanar reformatted
slice" or "multiplanar reconstruction slice"
(Fig. 13, images labeled 1a
and 1b). Another alternative is to volume render the data. This is
accomplished by projecting through the data from a selected point of view
(Fig. 13, 2a and 2b).
View larger version (45K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 13 —Progression from 2D imaging to 3D narrow-angle (NA) to 3D
full-volume (FV) imaging is shown with respective representative images. In 2D
imaging, only single slice of data is acquired with result being 2D image with
no depth. In 3D narrow-angle imaging, entire heart volume is not obtained;
rather, 3D slice of varying width, depending on resolution, is obtained and
simultaneous volume rendering adds depth not seen with 2D slice. Finally, for
3D full-volume imaging, example shows that although entire volume of heart is
captured, exact same volume-rendered slice shown in 3D narrow-angle imaging
can be extracted from larger volume data set resulting in identical image.
(Illustration by Richard Gersony)
|
|
To further explain how the live 3D echocardiography transducer works, it is
important to step back and realize that a 1D non-phased-array ultrasound
transducer can create a single scan line, called an A-mode line, that is
essentially a point in space (Fig.
12). A 1D phased-array transducer creates a scan line that can be
steered in the lateral direction. This produces a 2D image called a B-mode
image (Fig. 12). An important
concept is that the B-mode image has an axial (depth) dimension and a lateral
(azimuthal) dimension. Due to the physics of the how the scan line is formed,
the axial dimension typically has much better resolution than the lateral
dimension. A 2D phased-array probe can steer a scan line in both the elevation
and lateral dimensions. This allows a 3D set of data to be formed
(Fig. 12).
Full-volume versus narrow-angle acquisition—Current live 3D
imaging routinely performs imaging in a narrow angle, a 3D volume sector that
cannot span the entire heart; cardiac gating is used only in the sense that
frames are captured throughout the cardiac cycle, the same as with 2D imaging.
A full-volume acquisition requires cardiac gating to acquire four cardiac
cycles and then reconstruct them into an entire cardiac volume. Gating is an
important part of this acquisition for aligning similar frames for each of the
four cycles. The four cardiac cycles are then volume rendered with some
interpolation required between each cycle. This approach is, of course,
subject to the limitations of patient and transducer movement, but because
only four cycles are acquired, it is less of a limitation (Figs.
14A,
14B, and
14C).
View larger version (20K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 14A —Diagrams show concept of full-volume versus narrow-angle
imaging compared with 2D imaging. Note lack of elevation steering for 2D
imaging and that larger volume data set is acquired for full-volume imaging.
(Illustration by Richard Gersony)
|
|
View larger version (24K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 14B —Diagrams show concept of full-volume versus narrow-angle
imaging compared with 2D imaging. Concept of acquisition of full volume is
shown. Full volume is acquired from total of four cardiac cycles. Each
individual cardiac cycle contributes one fourth of thickness of entire data
set as illustrated with individual colors. Cardiac gating is required to meld
four cardiac cycles into one. (Illustration by Richard Gersony)
|
|
View larger version (27K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 14C —Diagrams show concept of full-volume versus narrow-angle
imaging compared with 2D imaging. When more than one cardiac cycle is used, it
becomes important to match individual frames within a cardiac cycle. As
illustrated, when one cardiac cycle is acquired, each cardiac cycle is divided
into multiple frames; total of 11 frames are shown in this example. Number of
frames used for each acquisition depends on heart rate (R-R interval) as well
as frame rate of acquisition machine. It is desirable (but not always
possible) to have same R-R interval for each cardiac cycle so that each frame
can be matched. (Courtesy of TomTec Imaging Systems GmbH, Munich, Germany)
|
|
Each cardiac cycle provides one fourth of the volume data set, as shown in
Figures 14A,
14B, and
14C. The width of the data set
acquired can be increased with the trade-off being reduced resolution for both
narrow-angle and full-volume sets
[11]. Finally, a Zoom mode in
the narrow-angle mode allows focusing in on a particular structure with high
resolution with the trade-off being a narrower data set
(Fig. 15). These adjustments
also affect the pixel size of the image. A standard display has approximately
500 x 500 pixels. Hence, at an average depth of 12 cm, this makes the
pixel size 0.25 x 0.25 mm. Likewise, decreasing the average depth to 6
cm would decrease the pixel size to 0.125 x 0.125 mm. Magnification of
the image using the Zoom function will increase the effective pixel size by a
factor dependent on the percent size of the zoomed image compared with the
total image size. For example, if the area of the image selected to zoom is
one fourth the size of the total image, the resultant pixel size will increase
by roughly a factor of 4 and hence would be in the range of 1 mm in
diameter.
View larger version (27K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 15 —Narrow-angle versus full-volume imaging is shown as well as
different levels of resolution possible (high, medium, low). In this
particular algorithm for 3D imaging, choice was made to keep number of scan
lines; hence, volume rate is constant and results in change in resolution from
narrow angle to full volume. See Full-Volume Versus Narrow-Angle Acquisition
section for further discussion regarding resolution and zoom. (Illustration by
Richard Gersony)
|
|
As previously described, multiplanar reconstruction slices can be taken
from the 3D data set at any orientation. With electronic 3D steering,
essentially any plane is possible. Examples are shown in
Figure 16: Drawings A and C
show lateral (azimuthal) and drawings B and D show elevation steering, but any
combination of angles is possible depending on the desired structure for
imaging. Note the accompaniment of the corresponding 2D multiplanar
reconstruction slice and 3D rendering in opposite directions originating from
the multiplanar reconstruction slice.
View larger version (60K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 16 —Versatility of 3D data set is shown with three consecutive
cropping planes (lateral, elevation, and azimuthal [axial]), each
perpendicular to other, used to create 3D images that are perpendicular to one
another. Three-dimensional rendering can be performed in any direction
perpendicular to reference plane. Note that final image (drawing E) represents
equivalent of C scan with cutting plane being perpendicular to axial plane or
depth of ultrasound beam. (Illustration by Richard Gersony)
|
|
One common orientation is the C slice
(Fig. 16, drawing E shown with
2D and 3D versions, one 3D rendering looking toward the mitral valve and the
other looking toward the apex of the left ventricle). This slice is orthogonal
to the axial direction, and from a cardiac imaging perspective, this slice is
orthogonal to the long axis of the heart. The result is a cross-sectional or
short-axis view of the heart. The two dimensions of the C slice image are
lateral (azimuthal) and elevation. Lateral spatial resolution and elevation
spatial resolution are approximately the same, and both are inferior to axial
resolution. Typical resolution at a 10-cm depth is approximately 0.6 mm for
axial and 1.5 mm for lateral and elevation. C slice images therefore have poor
spatial resolution in both dimensions. From a practical standpoint then, the
3D beam former is steered through multiple axial planes of the left ventricle,
the resolution of the mitral valve moving from slice to slice is superior to
the resolution of the left ventricular endocardium (in the periphery of the
scan), and the scan is generated from the elevation and lateral planes. In
fact, this has been the challenge for wall motion analysis and calculations of
left ventricular volume for 3D echocardiography.
Volume rendering of 3D data is a computationally intensive task. In the
past, this computation required many seconds. Reconstructive 3D rendering
required interpolation between the 2D slices. Live 3D imaging also requires
interpolation because it is also formed from a set of 2D slices. All 3D
approaches require that the scan lines be interpolated into a 3D cartesian
data set before volume rendering. Interpolation and volume rendering both
require intense computer processing; fortunately, advances in PC power have
now made it possible to perform volume rendering in as little as 50
milliseconds. This is one of the many factors that has facilitated 3D imaging
in becoming real-time imaging. However, real-time rendering must be
accompanied by real-time scanning, and this is a challenge given the speed of
the ultrasound beam and the width and depth of tissue penetration required to
create the 3D volume.
Although electronic scanning allows the ultrasound beam to be steered
rapidly, there still must be sufficient time for the ultrasound beam to be
transmitted, propagate through the body, be reflected back to the transducer,
be received, and then be interpreted and displayed as an image. Once again,
this task is a daunting one because sound propagates in the body at
approximately 1,500 m/s, and this limits how rapidly the ultrasound beam can
be transmitted and received. For a typical 16-cm scanning depth, an ultrasound
beam can be transmitted and received in only 220 µsec. For a 2D image that
has 90 ultrasound lines, this would require 19.8 msec. The frame rate for such
an image would be 50 Hz. This is adequate for most clinical scanning,
particularly scanning of adults with heart rates typically in the range of
60-100 beats per minute. By comparison, a 3D image that is 60° x
60° would require up to 3,600 scan lines. For the same depth, this would
require 800 milliseconds. The accompanying frame rate of 1.25 Hz would be
unacceptable for clinical use. Hence, numerous solutions exist for increasing
the frame rate for 3D scanning. The limitations of the solutions are
degradation of image quality, smaller field of view, or triggered (based on
the R wave of the ECG) acquisition of more than one cardiac cycle.
As is true for most sonography solutions, the success or failure of a
product is contingent on choosing the most optimal compromises after
considering the resultant trade-offs. The possible solutions for improving
frame rate while still scanning in real-time 3D are as follows: first, receive
multiple beams for each transmit event (an approach called "parallel
processing"); second, increase the line spacing of each transmit event;
third, decrease the image size either laterally or in elevation (the depth
required cannot be changed any less than roughly 12 cm); and, fourth, acquire
data over multiple cardiac cycles and build a composite volume. Current live
3D imaging uses a combination of these approaches. Parallel processing is
definitely used; the details about the number of scan lines involved in the
parallel processing remain proprietary. The image size can be adjusted, with a
smaller size resulting in improved frame rate and spatial resolution. The
concept of the full-volume acquisition using four cardiac cycles has been
discussed.
Perceived and Proven Clinical Benefits of 3D Echocardiography Compared with 2D Echocardiography
Analysis of Cardiac Volumes and Mass
Three-dimensional echocardiography has been shown to allow more accurate
assessment of ventricular volume and ejection fraction than its 2D
echocardiography counterpart. This has been shown in multiple studies with
comparisons to left ventricular angiography and MRI
[5-34].
These data are largely from the older method of reconstructive 3D
echocardiography. Now several studies show favorable agreement with older
real-time 3D echocardiography (i.e., Volumetrics type) to standard noninvasive
methods
[35-44].
In addition, there are now three studies comparing MRI with live 3D
echocardiography that show very good correlation and agreement
[45,
46]. Correlation and agreement
shown from the study by Mor-Avi et al.
[47] are typical for live 3D
echocardiography (Figs. 17A
and 17B).
View larger version (12K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 17A —Graphs show correlation and degree of agreement between 2D
echocardiography and 3D echocardiography compared with currently perceived
gold standard of cardiac MRI. Note considerable improvement in both
correlation and agreement to MRI provided by live 3D imaging compared with 2D
imaging. (Reprinted with permission from Circulation
[47]). Graphs show correlation
and degree of agreement between 2D echocardiography and live 3D
echocardiography (A) and between each echocardiography technique and
cardiac MRI (B) for determining left ventricular (LV) mass.
|
|
View larger version (15K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 17B —Graphs show correlation and degree of agreement between 2D
echocardiography and 3D echocardiography compared with currently perceived
gold standard of cardiac MRI. Note considerable improvement in both
correlation and agreement to MRI provided by live 3D imaging compared with 2D
imaging. (Reprinted with permission from Circulation
[47]). Graphs show correlation
and degree of agreement between 2D echocardiography and live 3D
echocardiography (A) and between each echocardiography technique and
cardiac MRI (B) for determining left ventricular (LV) mass.
|
|
Congenital Heart Disease
The other major area in which 3D echocardiography has enjoyed particular
success is in the assessment of anatomic relationships in congenital heart
disease
[48-64].
There is considerable evidence of the superiority of 3D echocardiography over
2D echocardiography for evaluation of congenital heart disease, although these
data were, of course, obtained using older techniques. The application of 3D
echocardiography to this specific subset of patients is intuitive. Details of
defects are more readily recognized, and 3D echocardiography provides
additional information that may be clinically useful when compared with 2D
echocardiography. For example, Salustri et al.
[49] undertook an evaluation
of a variety of congenital defects comparing 3D echocardiography with 2D
echocardiography and found that 3D echocardiography provided additional
clinical information in 36% of patients. The clinical advantage to such
additive information is still unclear.
Whereas 2D echocardiography is useful for identifying membranes in patients
with atrial septal defects or ventricular septal defects, 3D echocardiography
can measure diameters of the defects or of closure devices after placement
using en face views from both the left and right sides (Figs.
18A and
18B). Defect sites,
fenestrations, membranes, and particularly relationships to surrounding
structures and tissues are theoretic advantages of 3D echocardiography, all of
which may influence management options, particularly in the era of
minimal-access surgery and percutaneous catheter closure.
View larger version (68K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 18A —Live 3D imaging examples of how 3D imaging is better than 2D
imaging for evaluating structures such as septal defects. Atrial septal defect
(ASD) occluder device (Amplatz Occluder Device, AGA Medical Corporation) is
shown on 3D imaging. Note ability to see device en face in B after
tilting image shown in A. In B, AoV indicates aortic valve.
|
|
View larger version (68K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 18B —Live 3D imaging examples of how 3D imaging is better than 2D
imaging for evaluating structures such as septal defects. Atrial septal defect
(ASD) occluder device (Amplatz Occluder Device, AGA Medical Corporation) is
shown on 3D imaging. Note ability to see device en face in B after
tilting image shown in A. In B, AoV indicates aortic valve.
|
|
Until recently, there were no data about the use of live 3D
echocardiography in patients with congenital disease, only extensive
experience [64,
65]. However, Chan et al.
[66] have shown that live 3D
echocardiography results in additional information compared with 2D
echocardiography in a significant number of cases. In this assessment of live
3D imaging for evaluation of structural heart disease, the authors imaged 106
consecutive patients referred for echocardiography and graded the 3D studies
as A, new finding identified; B, useful anatomic perspective identified; C,
equivalent to 2D studies; or D, missed findings seen on 2D echocardiography.
Live 3D echocardiography was graded A in 7% and B in 18%. The majority of the
A and B cases were congenital or mitral valve disease. Live 3D
echocardiography has recently been used to identify the precise location of a
subaortic membrane [67]. In a
surgical correlation study, live 3D echocardiography revealed the size and
location of both atrial and ventricular septal defects
[68].
Valvular Heart Disease
Echocardiography has traditionally been the undisputed test of choice for
the evaluation of valvular heart disease, particularly those involving masses
and structural abnormalities. Three-dimensional echocardiography in general
and live 3D echocardiography in particular have added the ability to view the
entire valve in one image with the bonus of depth as opposed to requiring
multiple 2D slices and views to show the entire valve. The 3D echocardiography
volume-rendered view of the orifice of the mitral valve has led to several
studies showing that live 3D echocardiography is the best noninvasive method
for determination of mitral valve area, and this technique compares favorably
with catheterization methods of stenosis quantification
[69] (Figs.
19A,
19B,
19C, and
19D). In addition, live 3D
echocardiography has superior ability for evaluating the perivalvular area,
mitral valve commissures, and mitral valve subvalvular apparatus in the
setting of mitral valve stenosis
[70]. For mitral
regurgitation, 3D echocardiography has been used to determine accurately which
division or "scallop" of the mitral valve leaflets is involved in
mitral valve prolapse, hence aiding in the surgical planning for mitral valve
repair [71].
View larger version (59K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 19C —Live 3D transthoracic view of mitral valve repair. Images
show mitral valve and annuloplasty ring with view from left atrial side of
valve during systole (C) and diastole (D). These images were
obtained via reconstructive transesophageal echocardiography technology.
Arrows depict central orifice (thick arrow) and incoming pulmonary
veins (thin arrow) on periphery.
|
|
View larger version (59K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 19D —Live 3D transthoracic view of mitral valve repair. Images
show mitral valve and annuloplasty ring with view from left atrial side of
valve during systole (C) and diastole (D). These images were
obtained via reconstructive transesophageal echocardiography technology.
Arrows depict central orifice (thick arrow) and incoming pulmonary
veins (thin arrow) on periphery.
|
|
Conclusions
Three-dimensional echocardiography has now been shown to have several
advantages over 2D echocardiography, particularly for volume measurements,
visualization of septal defects, and whole-valve evaluation. Given these data,
it is clear that 3D echocardiography is here to stay and soon will become part
of routine echocardiographic examinations. This is particularly true because,
at the time of this writing, a second vendor, GE Healthcare, has developed a
form of real-time 3D echocardiography. If a third major vendor, Siemens
Medical Solutions, follows suit as expected, then real-time 3D
echocardiography will be available on most cardiac ultrasound systems. Given
this development, we conclude that 3D echocardiography will largely displace
2D echocardiography because history tells us that a similar transition
happened as we moved from M-mode echocardiography to 2D echocardiography.
Acknowledgments
We recognize Dave Prater and Gina Kelly of Philips Medical Systems for
aiding with the explanation of the mechanisms for live 3D imaging. We
recognize Richard Gersony forthe development of illustrations that demonstrate
these mechanisms. We also acknowledge the sonographers at Harborview Medical
Center for their aid in acquisition of the images used in this article.
References
- Dekker DL, Piziali RL, Dong E Jr. A system for ultrasonically
imaging the human heart in three dimensions. Comput Biomed
Res 1974; 7:544
-553[CrossRef][Medline]
- Moritz WE, Shreve PL. A microprocessor based spatial locating
system for use with diagnostic ultrasound. IEEE Trans Biomed
Eng 1976; 64:966
-974
- Raab FH, Blood EB, Steiner TO, et al. Magnetic position and
orientation tracking system. IEEE Trans Aerospace Elec
Sys 1979; AES-15:709
-718
- Legget ME, Leotta DF, Bolson EL, et al. System for quantitative
three-dimensional echocardiography of the left ventricle based on a
magnetic-field position and orientation sensing system. IEEE Trans
Biomed Eng 1998; 45:494
-504[CrossRef][Medline]
- Matsumoto M, Inoue M, Tamura S, Tanaka K, Abe H. Three-dimensional
echocardiography for spatial visualization and volume calculation of cardiac
structures. J Clin Ultrasound 1981;9
: 157-165[Medline]
- Miwa H, Hayashi H, Shimura T, Murakami K, Sugiyama Y, Kawabe K. 3-D
envisioning by real time perpendicular bi-plane echocardiography (simultaneous
multifrequency ultrasonography-III). Ultrasound Med
Biol 1983; [suppl 2]:283
-287
- von Ramm OT, Smith SW, Pavy HG Jr. High-speed ultrasound volumetric
imaging system. Part II. Parallel processing and image display.
IEEE Trans Ultrason Ferroelectr Freq Control1991; 38:109
-115[Medline]
- Smith SW, Pravy HG Jr, von Ramm OT. High-speed ultrasound
volumetric imaging system. Part I. Transducer design and beam steering
IEEE Trans Ultrason Ferroelectr Freq Control1991; 38:100
-108[Medline]
- von Ramm OT, Smith SW. Real time volumetric ultrasound imaging
system. J Digit Imaging 1990;3
: 261-266[Medline]
- Ariet M, Geiser EA, Lupkiewicz SM, Conetta DA, Conti CR. Evaluation
of a three-dimensional reconstruction to compute left ventricular volume and
mass. Am J Cardiol 1984;54
: 415-420[CrossRef][Medline]
- Raichlen JS, Trivedi SS, Herman GT, St. John Sutton MG, Reichek N.
Dynamic three-dimensional reconstruction of the left ventricle from
two-dimensional echocardiograms. J Am Coll Cardiol1986; 8:364
-370[Abstract]
- Matsumoto M, Matsuo H, Kitabatake A, et al. Three-dimensional
echocardiograms and two-dimensional echocardiographic images at desired planes
by a computerized system. Ultrasound Med Biol1977; 3:163
-178[CrossRef][Medline]
- Nixon JV, Saffer SI, Lipscomb K, Blomqvist CG. Three-dimensional
echoventriculography. Am Heart J 1983;106
: 435-443[CrossRef][Medline]
- Linker DT, Moritz WE, Pearlman AS. A new three-dimensional
echocardiographic method of right ventricular volume measurement: in vitro
validation. J Am Coll Cardiol 1986;8
: 101-106[Abstract]
- Moritz WE, Pearlman AS, McCabe DH, Medema DK, Ainsworth ME, Boles
MS. An ultrasonic technique for imaging the ventricle in three dimensions and
calculating its volume. IEEE Trans Biomed Eng1983; 30:482
-491[Medline]
- Altmann K, Shen Z, Boxt LM, et al. Comparison of three-dimensional
echocardiographic assessment of volume, mass, and function in children with
functionally single left ventricles with two-dimensional echocardiography and
magnetic resonance imaging. Am J Cardiol1997; 80:1060
-1065[CrossRef][Medline]
- Keller AM. Positional localization: three-dimensional transthoracic
echocardiographic techniques for the measurement of cardiac mass, volume, and
function. Echocardiography 2000;17
: 745-748[Medline]
- Sapin PM, Schroder KM, Gopal AS, Smith MD, DeMaria AN, King DL.
Comparison of two- and three-dimensional echocardiography with
cineventriculography for measurement of left ventricular volume in patients.
J Am Coll Cardiol 1994;24
: 1054-1063[Abstract]
- Sapin PM, Schroeder KD, Smith MD, DeMaria AN, King DL.
Three-dimensional echocardiographic measurement of left ventricular volume in
vitro: comparison with two-dimensional echocardiography and
cineventriculography. J Am Coll Cardiol1993; 22:1530
-1537[Abstract]
- Munt BI, Leotta DF, Bolson EL, et al. Left ventricular shape
analysis from three-dimensional echocardiograms. J Am Soc
Echocardiogr 1998; 11:761
-769[CrossRef][Medline]
- Hubka M, Bolson EL, McDonald JA, Martin RW, Munt B, Sheehan FH.
Three-dimensional echocardiographic measurement of left and right ventricular
mass and volume: in vitro validation. Int J Cardiovasc
Imaging 2002; 18:111
-118[CrossRef][Medline]
- Sheehan FH, Bolson EL, McDonald JA, Reisman M, Koch KC, Poppas A.
Method for three-dimensional data registration from disparate imaging
modalities in the NOGA Myocardial Viability Trial. IEEE Trans Med
Imaging 2002; 21:1264
-1270[CrossRef][Medline]
- Dorosz JL, Bolson EL, Waiss MS, Sheehan FH. Three-dimensional
visual guidance improves the accuracy of calculating right ventricular volume
with two-dimensional echocardiography. J Am Soc
Echocardiogr 2003; 16:675
-681[CrossRef][Medline]
- Chuang ML, Parker RA, Riley MF, et al. Three-dimensional
echocardiography improves accuracy and compensates for sonographer
inexperience in assessment of left ventricular ejection fraction. J
Am Soc Echocardiogr 1999; 12:290
-299[CrossRef][Medline]
- Kasprzak JD, Vletter WB, van Meegen JR, et al. Improved
quantification of myocardial mass by three-dimensional echocardiography using
a deposit contrast agent. Ultrasound Med Biol1998; 24:647
-653[CrossRef][Medline]
- Mele D, Maehle J, Pedini I, Alboni P, Levine RA. Three-dimensional
echocardiographic reconstruction: description and applications of a simplified
technique for quantitative assessment of left ventricular size and function.
Am J Cardiol 1998;81
: 107G-110G[CrossRef][Medline]
- Mele D, Levine RA. Quantitation of ventricular size and function:
principles and accuracy of transthoracic rotational scanning.
Echocardiography 2000;17
: 749-755[Medline]
- Pini R, Giannazzo G, Di Bari M, et al. Transthoracic
three-dimensional echocardiographic reconstruction of left and right
ventricles: in vitro validation and comparison with magnetic resonance
imaging. Am Heart J 1997;133
: 221-229[CrossRef][Medline]
- Papavassiliou DP, Parks WJ, Hopkins KL, Fyfe DA. Three-dimensional
echocardiographic measurement of right ventricular volume in children with
congenital heart disease validated by magnetic resonance imaging. J
Am Soc Echocardiogr 1998; 11:770
-777[CrossRef][Medline]
- Nosir YF, Salustri A, Kasprzak JD, Breburda CS, Ten Cate FJ,
Roelandt JR. Left ventricular ejection fraction in patients with normal and
distorted left ventricular shape by three-dimensional echocardiographic
methods: a comparison with radionuclide angiography. J Am Soc
Echocardiogr 1998; 11:620
-630[CrossRef][Medline]
- Buck T, Hunold P, Wentz KU, Tkalec W, Nesser HJ, Erbel R.
Tomographic three-dimensional echocardiographic determination of chamber size
and systolic function in patients with left ventricular aneurysm: comparison
to magnetic resonance imaging, cineventriculography, and two-dimensional
echocardiography. Circulation 1997;96
: 4286-4297[Abstract/Free Full Text]
- Nosir YF, Fioretti PM, Vletter WB, et al. Accurate measurement of
left ventricular ejection fraction by three-dimensional echocardiography: a
Top
Abstract
Introduction
