U.S. patent application number 11/974689 was filed with the patent office on 2008-06-12 for 3d ultrasonic imaging method.
This patent application is currently assigned to WILK ULTRASOUND OF CANADA, INC.. Invention is credited to Richard Bernardi, Timothy J. Nohara, Peter Weber.
Application Number | 20080139937 11/974689 |
Document ID | / |
Family ID | 27663160 |
Filed Date | 2008-06-12 |
United States Patent
Application |
20080139937 |
Kind Code |
A1 |
Nohara; Timothy J. ; et
al. |
June 12, 2008 |
3D Ultrasonic imaging method
Abstract
A probe for electronic volume data acquisition using ultrasound
incorporates a plurality of transducer elements arranged in a two
dimensional array having an azimuth direction and an elevation
direction. The transducer elements have a first element size in the
azimuth dimension and a second element size in the elevation
dimension. At least one of the first and second element sizes is at
least twice a characteristic wavelength of a waveform used to drive
the array of transducer elements, where the characteristic
wavelength is defined as the wavelength corresponding to a center
frequency of the waveform. Image data is generated in a scanning
process using a CAC-BF technique in an azimuth dimension and/or an
elevation dimension, to form an ultrasound image line, image plane,
or image data cube.
Inventors: |
Nohara; Timothy J.;
(Fonthill, CA) ; Weber; Peter; (Dundus, CA)
; Bernardi; Richard; (Wayne, PA) |
Correspondence
Address: |
COLEMAN SUDOL SAPONE, P.C.
714 COLORADO AVENUE
BRIDGE PORT
CT
06605-1601
US
|
Assignee: |
WILK ULTRASOUND OF CANADA,
INC.
St. Catharines
CA
|
Family ID: |
27663160 |
Appl. No.: |
11/974689 |
Filed: |
October 15, 2007 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10353152 |
Jan 28, 2003 |
7285094 |
|
|
11974689 |
|
|
|
|
60352969 |
Jan 30, 2002 |
|
|
|
Current U.S.
Class: |
600/443 ;
382/128 |
Current CPC
Class: |
G01S 7/52047 20130101;
A61B 8/483 20130101; G10K 11/346 20130101; G01S 7/52044 20130101;
G01S 15/8925 20130101; G01S 15/8927 20130101; G01S 15/8993
20130101 |
Class at
Publication: |
600/443 ;
382/128 |
International
Class: |
A61B 8/00 20060101
A61B008/00 |
Claims
1. A method for generating image data in an ultrasound scanning
process, comprising selectively energizing transducer elements in
an array of transducer elements and selectively polling said
transducer elements to effectively divide said array into a
plurality of subapertures transmitting and receiving a plurality of
low-resolution ultrasound beams that span a volume to be imaged;
and coherently combining beamformed signals from said subapertures
to synthesize an aperture larger than any one of said subapertures
and focused at each point of said volume.
2. The method defined in claim 1 wherein receiving a plurality of
low-resolution ultrasound beams includes receiving, from said
transducer elements, return signals encoding reflected ultrasound
waveforms; collecting said return signals to form range lines;
digitizing said return signals; and coarse beamforming the
digitized signals.
3. The method defined in claim 2 the coherent combining includes
subjecting the coarse beamformed digitized signals to a fine
beamforming.
4. The method defined in claim 3 wherein at least one of said
subapertures receives a plurality of said low-resolution ultrasound
beams, the fine beamforming including spatially interpolating
adjacent beams of said one of said subapertures.
5. The method defined in claim 3 wherein the fine beamforming
includes temporally interpolating between samples of said coarse
beamformed digitized signals.
6. The method defined in claim 3 wherein the fine beamforming
includes determining range delays for signal samples.
7. The method defined in claim 3 wherein the fine beamforming
includes scan-converting the range lines to a high-resolution grid
of voxels to form a low-resolution subaperture image for each
subaperture, and coherently combining the low-resolution
subaperture images to synthesize a larger aperture and a final
high-resolution image.
8. The method defined in claim 3 wherein the coherent combining
further includes range filtering the coarse beamformed digitized
signals prior to the subjecting of the coarse beamformed digitized
signals to the fine beamforming.
9. The method defined in claim 1 wherein at least one of said
subapertures receives a plurality of said low-resolution ultrasound
beams, the coherent combining including spatially interpolating
adjacent beams of said one of said subapertures.
10. The method defined in claim 1 wherein the coherent combining
includes temporally interpolating said beamformed signals.
11. The method defined in claim 1 wherein the coherent combining
includes determining range delays for said beamformed signals.
12. The method defined in claim 1 wherein the selectively
energizing and the selective polling of said transducer elements
includes energizing and polling said transducer elements so that
each subaperture transmits and receives a respective sequence of
overlapping phased-array beams each focused at a different
angle.
13. The method defined in claim 12 wherein said overlapping
phased-array beams are spaced so as to cross approximately at
respective -3 dB points.
14. The method defined in claim 12 wherein said transducer elements
have an inter-element spacing, the overlapping beams of any given
one of said subapertures subtending a total angle of less than a
weighted mathematical reciprocal of said inter-element spacing.
15. The method defined in claim 1 wherein each of said subapertures
overlaps an adjacent one of said subapertures, the overlap
including at least 50% of the transducer elements included in said
adjacent one of said subapertures.
16. The method defined in claim 1 wherein each of said subapertures
overlaps an adjacent one of said subapertures, the overlap
including approximately 50% of the transducer elements included in
said adjacent one of said subapertures.
17. The method defined in claim 1 wherein said array is a
two-dimensional array in an image space having a first dimension
and a second dimension, said subapertures extending along at least
one of said first dimension and said second dimension.
18. The method defined in claim 1 wherein the selective energizing
of said transducer elements includes energizing transducer elements
of only one of said subapertures to generate at least one outgoing
waveform.
19. The method defined in claim 1 wherein the selective energizing
of said transducer elements includes energizing transducer elements
of each of said subapertures so that each of said subapertures
generates at least one outgoing waveform.
20. The method defined in claim 1 wherein the selective energizing
of said transducer elements and the selective polling of said
transducer elements are such that each of said subapertures
transmits and receives a plurality of low-resolution ultrasound
beams that span the volume to be imaged.
21. The method defined in claim 1 wherein the selective energizing
of said transducer elements and the selective polling of said
transducer elements are such that each of said subapertures
transmits an outgoing waveform while fewer than all of said
subapertures receive reflected waveforms.
22. The method defined in claim 1 wherein the selective energizing
of said transducer elements and the selective polling of said
transducer elements are such that fewer than all of aid
subapertures transmit an outgoing waveform while all of said
subapertures receive reflected waveforms.
23-43. (canceled)
Description
FIELD OF INVENTION
[0001] This invention relates to ultrasound imaging systems. More
particularly, this invention relates to methods and devices for
three-dimensional image acquisition. The devices and methods are
also suitable for 2D and 4D ultrasound systems. The invention is
particularly, but not exclusively, useful for medical diagnoses and
treatment. The devices and methods of the present invention are
useful components of practical high-quality real-time 3D ultrasound
systems with fully electronic volume data acquisition.
BACKGROUND OF THE INVENTION
[0002] Two-dimensional (2D) ultrasonic probes are necessary to
support three-dimensional (3D) electronic, volume data acquisition
for many clinical applications. State-of-the-art one-dimensional
(1D and 1.5D) probes which electronically scan only in azimuth
provide the 2D ultrasound images (azimuth and range) which are
commonly used today. 2D probes scan electronically in elevation as
well as azimuth, to provide a three dimensional data cube (azimuth,
elevation and range) which can be processed using image processing
software to produce a variety of image formats. These formats
include conventional planar images, planar images at arbitrary scan
planes, as well as representations such as surface rendering and
orthographic presentations. Four-dimensional (4D) representations
include 3D animations where the 3D rendering is updated in
time.
[0003] Two-dimensional sensors are employed in other imaging
modalities such as CT-scanners, and in other fields such as radar;
and hence are well understood conceptually. Practical difficulties
arise with the ultrasound modality due to the small, elemental
feature size (fractions of a mm) and the large number of channels
typically needed. These difficulties have stalled the introduction
of fully electronic, 2D, ultrasonic probes.
[0004] Ultrasound systems today make use of a variety of 1D and
1.5D ultrasonic arrays. A 1D array has a fixed elevation aperture
which is focussed at a fixed range, and is usually realized with a
mechanical lens of sorts. A 1.5D array, on the other hand, has a
variable elevation aperture, shading and focussing, but they are
symmetric about the centerline of the array.
[0005] 1D array transducers contain several tens or even hundreds
of elements typically arranged linearly. The transducer elements
10, 12 may be arranged on a straight line (linear array) or a
curved line (curved linear array or simply curved array) as shown
in FIGS. 1A and 1B, respectively. The operation of a linear array
or curved array are similar, the main difference being that the
image expands with range (depth) for the curved array. A typical
linear or curved array could have anywhere from 64 to 512 (or more)
elements, depending on the cost and the application. The azimuthal
spacing of elements is typically between half a wavelength and one
wavelength. The elemental size in the elevation dimension is much
larger, typically tens of wavelengths. The operating frequency is
typically somewhere between 2 MHz to 20 MHz, depending on the
clinical application.
[0006] Let's consider an example, where a 7.5 MHz curved array of
the type shown in FIG. 1B has 256 transducer elements 12 in azimuth
spaced by one wavelength; and the dimension of an element in
elevation is, say, 40 wavelengths. At 7.5 MHz, the wavelength,
.lamda., in tissue is about 0.2 mm. Therefore, the array spans
about 51 mm in azimuth and 8 mm in elevation.
[0007] A narrow beam is created in the azimuth dimension by
focussing the transmitted and receive energy along a particular
beam or scan line 14, 16, as illustrated in FIG. 2A and FIG. 2B.
Scanning is performed in azimuth (i.e. in a single elevation plane)
using one of two schemes, sequential scaring or phased-array
scanning. With sequential scanning, any given beam line is offset
from all of the other beam lines in the azimuth direction. If the
array is linear (rather than curved), the beam lines 16, 18 are
parallel to one another (FIG. 3A). With reference to FIG. 2B, the
central beam line 16 that is illustrated is shifted (offset) to the
left or right with different offsets 20 to create a set of beam
lines 22 that spans the region or volume to be imaged, as
illustrated in FIG. 3A. Phased-array scanning, on the other hand,
is achieved by rotating the central beam line 24 illustrated in
FIG. 3B in azimuth, to the left and to the right, by a set of
angular offsets 26. The beam lines 28 of the resulting set 30 of
beam lines intersect at a common apex 32 (which may actually occur
behind the array), and separate from each other as a function of
range, as illustrated in FIG. 3B.
[0008] Premium probes generally employ wideband waveforms to
achieve the fine resolutions needed in range. As a result,
beamforming is done by adjusting time delays (in the narrowband
waveform case, phases are adjusted rather than time delays) at each
element used on transmit and receive. For a given pulse, a focal
point is set along the range dimension. Appropriate time delays are
used on the elements involved in transmission, so that their
respective acoustic energy arrives at the specified focal range,
along the specified beam line, at the same time. As a result, the
waveform is said to be focussed at this point. On receive, time
delays are dynamically applied to the elements involved in
reception, to focus the received energy at each range.
[0009] Generally speaking, focussing is needed only in the near
field of the array, where the ultrasound wave cannot be assumed to
be planar, as it is in the far field. If one looks closely at the
effect of this focussing operation in the azimuth and elevation
spatial dimensions, one notices a difference. In azimuth, numerous
transducer elements are available, each with a respective time
delay to adjust dynamically with range on receive. The result is
the azimuth resolution of the beam can be generally maintained
uniformly with range as illustrated in FIG. 2A for a 1D linear
array. There are no delays to adjust in elevation, however. As a
result, a typical, fixed, lens-like beam pattern results in the
elevation dimension, with the best elevation resolution occurring
at the transmit focal point (in range), and with a degradation of
the elevation resolution as one moves away from this focal point in
range. This effect is also illustrated in FIG. 2A. The image plane
thickness (i.e. in the elevation dimension) in effect varies with
range for a 1D linear array.
[0010] The 1.5D array provides a solution to the image thickness
problem, and therefore produces higher-quality, planar images than
the 1D array (Wildes, D. G., et al., "Elevation Performance of
1.25D and 1.5D Transducer Arrays", IEEE Transactions on Ultrasound,
Ferroelectronics and Frequency Control. Vol. 44, No. 5, September
1997, pp. 1027 to 1036). By using multiple rows of elements in the
elevation dimension, as illustrated in FIG. 2B, multiple elevation
lenses can be effected, each focussed at a different focal range.
This is achieved by varying the time delays (through switching or
otherwise) applied to the elevation elements while the acoustic
signals are being received. In addition, a lens is typically used
in the elevation dimension to help control the elevation focus. The
net effect is that the elevation thickness (resolution) is
maintained with range, thereby improving image quality. This is
illustrated in FIG. 2B.
[0011] In a typical 1.5D array, each element might be
.lamda..times.4.lamda. (i.e. azimuth by elevation) in dimension.
For an array with 128 elements per row and 8 rows of elements, the
elevation dimension is 32.lamda. or 6.4 mm and the azimuth
dimension is 128.lamda. or 25.6 mm at 7.5 MHz.
[0012] Consider the linear 1.5D array shown in FIG. 3A, containing
256 elements 34 in azimuth. Now 128 sequential beams 18 are
typically used to form a rectangular, azimuthal image plane by
scanning in azimuth as illustrated in the figure. Typical
transducer dimensions for this state-of-the-art array are also
indicated. (Note: only 16 columns of elements 34 are shown in FIG.
3A, for simplicity, where in fact, 256 elements are represented in
the azimuth dimension).
[0013] A state-of-the-art array with .lamda./2 spacing in azimuth
to support phased-array scanning is illustrated in FIG. 3B . This
type of array produces pie-shaped images in contrast to the
rectangular images produced using sequential arrays.
[0014] Unlike 1.5D arrays which are commonly found in premium
ultrasound systems, 1.75D arrays are not yet in use in commercial
systems (Puyun Guo, Shikui Yan and Quing Zhu, "Elevation
Beamforming Performance of a 1.75D array", IEEE 2001 Ultrasound,
Ferroelectronics and Frequency Control Conference). 1.75D arrays
are like 1.5D arrays, except there is no symmetry constraint. As a
result, it is possible to provide a little bit of elevation
steering. However, due to the large element size in elevation
(several wavelengths), grating lobes become serious if the
electronic scanning is significant (Puyun Guo, Shikui Yan and Quing
Zhu, "Elevation Beamforming Performance of a 1.75D array", IEEE
2001 Ultrasound, Ferroelectronics and Frequency Control
Conference).
[0015] Interest in 3D Ultrasound is growing and all major
ultrasound companies are paying attention. There are two ways that
scanning is currently performed: sequential scanning and phased
array scanning. It is common knowledge to those skilled in the art
that if one conventionally-extends a 1D phased array (typically
with .lamda./2 element spacing) to two dimensions (of equal size),
or a 1D sequential array (typically with .lamda. element spacing)
to two dimensions, then data cubes could be acquired by 2D
scanning, and the fine (e.g. an F number of 2, denoted herein as
F/2) azimuth resolution currently available extends to elevation as
well. Two fundamental difficulties, however, arise:
[0016] 1. the cost is prohibitive;
[0017] 2. the frame-time to acquire a 3D volume is far greater than
the time it takes to acquire a 2D image.
[0018] Consider extending a linear array with 256 elements (maximum
of 128 used on receive) to two dimensions. The number of elements
increases to 256.times.256=65,536. Transducer design/fabrication is
very difficult, if not impossible, today. The number of receiver
channels would also increase by a factor of 128 in order to provide
the same resolution in both dimensions, all else being equal, while
not increasing the number of shots (and hence acquisition time)
needed per vector. Since system cost is proportional to the number
of channels, the resulting cost is unaffordable.
[0019] Finally, it takes longer to acquire the data cube (as
compared to the tens of milliseconds needed to acquire a
conventional 2D image plane) since there are many more beams needed
to interrogate the volume. At least 128.times.128=16,384 beams are
needed, for each transmit focal range, with about 100 .mu.s two-way
time needed for each shot (this assumes a 10 kHz firing rate and a
7 cm depth needed). For two focal ranges, this implies an
acquisition time of 3.2 seconds, assuming the number of channels
available equals the number of elements used in the beamformer.
[0020] The aforementioned difficulties require practical trade-offs
and novel solutions if 2D arrays supporting 3D electronic, volume
data acquisition are to be an affordable reality.
Additional Prior Art
[0021] Many engineers have attempted to solve the aforementioned
difficulties in order to help make 3D ultrasound imaging an
affordable reality. Some of the more relevant approaches with
respect to the current inventions are discussed below.
[0022] As a result of the complexities associated with 2D
electronic scanning, some engineers have proposed the use of
mechanical scanning in the elevation dimension. That is, a
conventional, 1D linear array is used to provide the conventional,
B- or C-mode, range-azimuth, planar image; but it is moved up and
down quickly using mechanical means such as a motor, to acquire
successive planar images at a set of elevation positions. In U.S.
Pat. No. 6,106,471 "Procedure for an Examination of Objects by the
Means of Ultrasound Waves", Wiesauer, Fosodeder and Gritzky
describe such an approach. The mechanical movement in the elevation
dimension is done automatically and continuously using a motor in
the 2D probe housing. As focussing scan lines requires precise, a
priori knowledge of the location of the 1D array elements with
time, the quality of 2D and 3D imagery produced using this approach
is limited by the accuracy of the mechanical movements in the
elevation dimension.
[0023] In Ultrasonic Blanket with CAC and SCA patent application,
U.S. Ser. No. 09/514,928, filed 28 Feb. 2000, 3D volume data
aqcuisition and focussing of beams using active transducers is
described. A singular, rigid carrier structure constructed using
scalar transducer elements arranged in the likeness of an array is
disclosed. Signal transmission apertures and data gathering
apertures are formed and used to electronically scan desired
regions and electronically acquire 3D volumetric data; where
coherent aperture combining (CAC) is used to combine the structural
data from multiple data gathering apertures, thereby increasing the
size of the effective data gathering apertures employed, and
thereby increasing image resolution. Both monostatic (on pulse one,
transmit and receive out of aperture one, on pulse two, transmit
and receive out of aperture two) and bistatic (transmit from one
aperture and receive simultaneously on two or more apertures)
operations are disclosed. Also disclosed is the use of 1.5D and
1.75D array technology to form a 2D array and effectuate volume
data acquisition by scanning in azimuth and elevation.
[0024] In U.S. Pat. No. 6,482,160 "High Resolution 3D Ultrasound
Imaging System Deploying a Multidimensional Array of Sensors and
Method for Multidimensional Beamforming Sensor Signals",
Stergiopoulos and Dhanantwari describe an adaptive beamforming
method which can be used to process sensor signals received on a 2D
array of sensors in order to generate a high resolution, 2D beam
response for each of a set of beam directions defined by
azimuth-elevation angle pairs. The method described in the
specification is applicable to the case where the imaged object is
in the far-field of the 2D array of sensors, and relates only to
processing techniques to be used on receive. It is assumed that a
single, low-gain, transmit sensor (e.g. an omni-directional
transducer element) is located away from the 2D array of sensors
and illuminates the entire region being imaged, causing the
ultrasound energy to reflect from the object towards the receiving
array. The inventors exploit adaptive beamforming algorithms to
increase the spatial resolution otherwise unobtainable from the
receive array, had conventional, linear beamforming techniques such
as the discrete Fourier transform been used. In theory, adaptive
algorithms can estimate the noise process competing with the
desired signal associated with each beam, and use this information
to adapt the receive beam so as to better suppress the noise.
Adaptive and linear beamforming techniques are well known to those
skilled in the art. The inventors acknowledge that if the assumed
noise characteristics are inaccurate, performance of the adaptive
beamformer will degrade significantly and may even result in
cancellation of the desired signal. Furthermore, implementing
adaptive algorithms directly on the full array of sensor
data-requires very significant computational resources; and
convergence of the adaptive solution requires significant training
data. To mitigate these practical difficulties, the inventors
propose a partially adaptive beamformer which reduces the number of
adaptive degrees of freedom (DOF) by preprocessing the array sensor
data using conventional Fourier beamforming. The partially adaptive
beamformer, for the case of a 2D array, begins by dividing the
array into smaller subapertures, each of which is a 2D array. For
each subaperture, a 2D conventional beamformer is implemented,
which, for computational efficiency, is organized as a cascade of
two 1D beamformers. For example, each row of the subaperture's
sensor data would be processed using a 1D azimuth Fourier
beamformer; and then the resulting column of azimuth-processed data
would be operated on by a 1D elevation Fourier beamformer. This
cascaded approach, known to those skilled in the art, only applies
for the case of far-field imaging. Finally,, adaptive beamforming
is performed on the resulting, conventionally-beamformed
subaperture signals, by adaptively processing those respective
subaperture signals which were conventionally-beamformed to the
same azimuth-elevation angle direction.
[0025] In U.S. Pat. No. 6,419,633 "2D Ultrasonic Transducer Array
for Two Dimensional and Three Dimensional. Imaging", A. Robinson,
B. Robinson and Detmer describe a particular implementation of an
electronic 2D array. The inventors disclose an electronic 2D
transducer array that can be configured or switched to provide both
2D arrays and 1D arrays for 3D and 2D imaging, respectively. A
variety of particular element switching and summing circuits are
disclosed to combine rows and columns of elements as needed for
each supported array configuration. By these designs, the inventors
intend to limit the total number of signal leads coming out of
their transducer (and by implication, the number of digital
receiver channels in the ultrasound system). This objective is
achieved primarily by using a sparse 2D array on receive, with
admitted negative implications on array sensitivity and grating
lobes. In the example, 19-by-19 2D transducer array used for
illustration, only 100 signal leads are needed to support the
sparse 2D array configuration, rather than 361 leads if all
elements were used. They also intend that 2D image quality is not
degraded, as compared to conventional 1D arrays which are optimized
for this purpose. This objective is met by configuring the 2D
transducer array as a fully populated, 1D array.
[0026] In U.S. Pat. No. 6,238,346 "System and Method Employing
Two-dimensional Ultrasound Array for Wide Field of View Imaging",
inventor Mason discloses a 2D rectangular transducer array which
scans an elongated sector volume using fewer transducer elements
than in prior art systems, while avoiding sidelobe anomalies. The
switching circuitry forms a number of small subarrays, where each
subarray spans the entire elevation dimension and includes a
contiguous subset of azimuth elements. Each subarray is shifted
from the next in azimuth by one element. The 2D array provides
phased-array scanning in the shorter elevation dimension using all
of the elements in each subarray. Conventional arrays, on the other
hand, perform phased-array scanning on the longer azimuth
dimension. The disclosed 2D array scans in azimuth by stepping
through subarrays, one at a time, and beamforming so as to produce
scanlines normal to the face of the subarray. With this design,
fewer elements are required along the elevation dimension because
it is kept deliberately small, to produce an elongated sector
volume. In azimuth, fewer elements are needed because the elements
are spaced further apart (as much as 2.lamda.) since phased-array
scanning is not used in this dimension. In other words, the 2D
array is sparse in the azimuth dimension. The transducer elements
used are square in shape and less than or equal to 0.75.lamda. in
size.
OBJECTS OF THE INVENTION
[0027] An object of the present invention is to provide ultrasound
imaging technology which may be incorporated in practical,
affordable, high-quality, 3D ultrasound imaging systems which are
clinically useful, and which exploit 3D, electronic, volume data
acquisition.
[0028] Another object of the present invention is to provide 3D
ultrasound systems using state-of-the-art elemental transducer
technology (e.g. using 1.75D transducer technology), switching,
multiplexer and cable technology, and real-time signal processing
technology such as ASIC beamforming and filtering hardware.
[0029] The most significant cost and complexity in premium 2D
ultrasound systems relates to the receive electronics, the cost of
which is proportional to the number of digital receive channels.
For affordability therefore, another object of the present
invention is to keep the number of digital receive channels in the
contemplated 3D ultrasound systems similar to that provided in
current premium 2D ultrasounds.
[0030] A further object of the present invention is that the 2D
probe apparatus and beamforming methods disclosed provide
high-quality volume data which can be used to generate high-quality
3D imagery (e.g. 3D surface rendering or orthographic
presentations) as well as 2D imagery of similar or better quality
than premium, state-of-the-art, 2D ultrasound systems currently
produce. High-quality imagery is characterized by azimuth,
elevation and range resolutions equal to or better than that
provided by premium 2D ultrasound systems, as well as grating lobe
beam responses (or sidelobe responses for narrowband systems)
similar to that provided by premium 2D systems.
[0031] It is another object of the present invention to provide
azimuth resolution that is significantly better than that of a
premium 2D ultrasound system using the same azimuth aperture.
[0032] It is another object of the present invention to provide
means to produce elevation resolution significantly better than
that provided by premium 1.5D arrays in use today.
[0033] It is another object of the present invention to provide
fully-electronic, 3D volume data acquisition to support rapid and
accurate interrogation of volumes combined with highest-quality, 3D
image formation.
[0034] In order to maximize the clinical utility of the 3D
ultrasound system contemplated with the present invention, it is an
object of the invention to minimize the 3D volume acquisition time
to sustain the highest 3D frame rates without significantly
sacrificing affordability or image quality. For example, it is an
object of the present invention is that meaningful, high-quality 3D
data cubes can be electronically acquired in a fraction of one
second.
[0035] Another object of the present invention is to provide 2D
probe and beamforming technology that can be manufactured in a
conformal form factor, and be used as a building block (i.e.
providing transmission apertures and data gathering apertures) in
ultrasound medical imaging systems, exemplarily as described in
U.S. Pat. No. 5,666,953, U.S. Pat. No. 5,871,446, U.S. Pat. No.
6,023,632, U.S. Pat. No. 6,319,201, and U.S. Pat. No.
6,106,463.
[0036] Another object of the present invention is to provide a
compact and deployable 3D ultrasound system of size, weight, power,
and form-factor similar to conventional 2D ultrasound systems.
[0037] Yet another object of the present invention is to increase
the image quality of light-weight, portable, 2D ultrasound imaging
systems employing fewer receive channels than premium 2D
ultrasounds, without appreciably increasing the size or cost of
such improved systems.
[0038] Another object of the present invention is to provide an
ultrasound system for 3D imaging of the carotid artery, providing
improvements in safety and accuracy over current diagnostic
methods.
[0039] A further object of the present invention is to provide
ultrasound technology permitting a standard 2D ultrasound medical
procedure to be carried out more quickly and hence more safely.
[0040] An additional object of the present invention is to provide
ultrasound technology enabling a relatively unskilled medical
practitioner the ability to perform an ultrasound medical
procedure.
[0041] Further objects of the invention will be apparent from the
drawings and descriptions herein. It is to be noted that each
object is achieved by at least one embodiment of the present
invention. However, it is not necessary that any given embodiment
achieve all of the objects of the invention.
SUMMARY OF THE INVENTION
[0042] The present invention addresses the above-mentioned
difficulties by employing innovative approaches which together
provide a practical solution to ultrasound imaging systems
employing 2D ultrasonic arrays which support electronic, 3D volume
data acquisition and beamforming.
[0043] The present invention is directed in part to a probe for
electronic, 3D volume data acquisition using ultrasound, comprising
a plurality of transducer elements arranged in a two dimensional
array having an azimuth dimension and an elevation dimension. The
transducer elements have a first element size in the azimuth
dimension and a second element size in the elevation dimension. In
a preferred embodiment, at least one of the first and second
element sizes is at least twice a characteristic wavelength of a
waveform used to drive the array of transducer elements, where the
characteristic wavelength is defined as the wavelength
corresponding to a center frequency of the waveform.
[0044] In a particular embodiment, an ultrasound imaging transducer
in a system in accordance with the present invention exploits 1.75D
elemental technology (Puyun Guo, Shikui Yan and Quing Zhu,
"Elevation Beamforming Performance of a 1.75D array", IEEE 2001
Ultrasound, Ferroelectronics and Frequency Control Conference).
[0045] An ultrasound imaging transducer in a system in accordance
with the present invention can be manufactured in a conformal form
factor, and be used as a building block (i.e., a 2D transducer
array module) in ultrasound blanket systems as disclosed in U.S.
Pat. No. 5,666,953 and its progeny.
[0046] The present invention is also directed to a method of
generating image data in a scanning process, using a CAC-BF (see
below) technique in at least one of an azimuth dimension and an
elevation dimension, to form an ultrasound image line, image plane,
or image data cube. The CAC-BF method can be applied advantageously
to any 1D or 2D ultrasonic probe or array, and is not restricted to
the preferred embodiments disclosed herein.
[0047] The present invention includes a novel beamforming method
(CAC-BF) that produces high-resolution ultrasound images more
efficiently than conventional methods. CAC-BF divides the
transducer into a number of smaller subapertures, each of which
transmits and receives a number of low-resolution beams that span
the imaged region. High resolution is obtained at each image point
by coherently combining the beamformed signals from the
subapertures, synthesising a large aperture focussed at the
point.
[0048] The present invention provides practical, clinically useful,
high-resolution, 3D ultrasound, electronic, volume data
acquisition.
[0049] An ultrasound imaging system in accordance with the present
invention exhibits 3D imaging with voxel resolution equal to or
better than that of state-of-the-art planar images, in both the
azimuth and elevation dimensions. In one preferred embodiment, the
voxel resolution is twice as good in azimuth and/or elevation as
that in state-of-the-art planar images.
[0050] An ultrasound imaging system in accordance with the present
invention is relatively inexpensive to manufacture. The imaging
system can be implemented as an inexpensive upgrade to existing
premium ultrasound systems, or as a stand-alone solution of
comparable cost to state-of-the-art 2D ultrasound systems.
[0051] In a particular embodiment, an ultrasound system in
accordance with the present invention is able to electronically
acquire the 3D data cube of size 26 mm.times.26 mm by 70 mm spanned
by the transducer in under one second.
[0052] As with any new imaging technology, it's usefulness must be
proven clinically using one or more clinical applications. The
initial target application of the present invention is the carotid
artery, although there is nothing that would restrict its use in
other applications (e.g. obstetrics and gynaecology). This clinical
application involves diagnosing plaque in the carotid artery, which
can be fatal if left untreated. Presently, high frequency (7.5
MHz), wideband linear arrays (that use sequenced azimuth scanning)
are used primarily for ultrasound imaging of the carotid artery.
State-of-the-art, 1.5D probes produce rectangular images that span
about 3 cm (in azimuth) by 7 cm (in depth), and whose image quality
is characterized by an F number of 2 in azimuth and 8 in elevation.
The ability to electronically acquire a data cube (rather than just
a plane) and to improve the elevation resolution to an F number of
4 are highly desirable for this application. The present invention
delivers such improvements affordably.
[0053] It is obvious to one skilled in the art that there is more
to building a 2D or 3D ultrasound imaging system as contemplated
herein than simply employing the disclosed 2D arrays or CAC-BF
method. The design and implementation of a complete probe,
beamformer or ultrasound imaging system assumes a large amount of
hardware, software, and systems engineering and manufacturing
knowledge known to those skilled in the art. For example, the 2D
probe array technology disclosed herein requires array
manufacturing, power, switching electronics, cabling, and housing
considerations to be determined for a particular implementation.
When the CAC-BF method is applied to a 1D probe array or 2D probe
array as disclosed herein, special switching and/or cabling
considerations known to those skilled in the art are needed. For
each desired beam, contiguous sets of elements associated with the
subapertures used by the CAC-BF method must be switched or
electrically connected to the cabling which feeds the received
signals to the receive electronics and beamformers. Particular
probe embodiments, all within the scope of the present invention,
are realized by employing combinations of switching and
multiplexing electronics known to those skilled in the art, to
trade-off cost, performance and complexity of the resulting probe.
Switching and multiplexing electronics may be contained entirely
within the probe housing, or distributed between the probe housing
and the ultrasound engine containing the receive electronics,
without departing from the spirit and scope of the present
invention. While it is a preferred embodiment of the present
invention for the CAC-BF coarse and fine beamforming operations to
be performed digitally in the ultrasound engine, this functionality
can also be distributed throughout the entire ultrasound system,
and be implemented in hardware or software in a variety of ways
known to those skilled in the art, without departing from the scope
of the present invention. In addition, post-beamforming operations
such as vector processing and imaging processing are also known to
those skilled in the art, and any conventional form of these
operations could obviously be used effectively with the disclosed
inventions.
Comparison of Invention with Prior Art
[0054] While U.S. Pat. No. 6,106,471 "Procedure for an Examination
of Objects by the Means of Ultrasound Waves" describes a practical
solution to 3D ultrasound imaging, it is fundamentally different
from the present invention. The present invention uses electronic
scanning in both the azimuth and elevation dimensions, affording
higher 3D image quality over the mechanical, elevation scanning
solution provided in U.S. Pat. No. 6,106,471.
[0055] The probe or imaging transducer of the present invention is
similar to the 2D array disclosed in U.S. Pat. No. 6,238,346 in
that 3D volume acquisition is done electronically, and both attempt
to reduce the number of transducer elements without degrading
sidelobe performance. However, there are several significant
differences. The 2D array in U.S. Pat. No. 6,238,346 uses square
elements which are spaced sparsely in the azimuth dimension which
reduces sensitivity and increases grating lobes. The present
invention does not use a sparse arrangement of elements. Rather, it
uses rectangular elements whose size in azimuth results in a
similar reduction in number of elements, without reducing
sensitivity. The imaging transducer of the present invention has
switching circuitry to form subarrays of transducer elements, each
consisting of a contiguous subset of azimuth elements and a
contiguous subset of elevation elements. Adjacent subarrays in the
azimuth dimension typically require an overlap of about 50% of the
number of azimuth elements in the subarray for optimal performance.
The 2D array in U.S. Pat. No. 6,238,346, on the other hand, forms
subarrays by switching in subsets of contiguous elements in the
azimuth dimension, but using all of the elements in the elevation
dimension. Furthermore, adjacent subarrays in azimuth are only
shifted by a single element (i.e. they require a much greater
overlap than 50%). The imaging transducer array of the present
invention preserves the state-of-the-art, B-mode planar image;
whereas the 2D array in U.S. Pat. No. 6,238,346 does not.
[0056] The present invention, like that described in U.S. Pat. No.
6,419,633 "2D Ultrasonic Transducer Array for Two Dimensional and
Three Dimensional Imaging" is concerned with a 2D electronic
transducer array which can be configured to support both 2D imaging
and 3D imaging. However, the present invention performs differently
in three key ways as a result of its CAC-BF method and its
preferred transducer design. First, it not only preserves 2D planar
image quality (as compared to that afforded by optimized 1D
arrays), it provides an azimuth resolution that is significantly
better. Second, it does not use a 2D sparse array for 3D imaging;
rather, it uses a full array; hence, the present invention does not
suffer reduced sensitivity, and grating lobes are avoided by using
sequential scanning in elevation when larger elements (greater than
.lamda.) are used. Finally, while the invention disclosed in U.S.
Pat. No. 6,419,633 does reduce the number of signal leads (and
hence channels) otherwise required, the reduction is not sufficient
for the objectives of the present invention. For example, an
instantaneous aperture to achieve F/2 in azimuth and F/4 in
elevation with a conventional element size .lamda. requires a fully
populated array of at least 128-by-64 elements. Using the sparse 2D
configuration of U.S. Pat. No. 6,419,633 still requires
64.times.32=2,048 signal leads and channels, which is an order of
magnitude larger than that needed by the present invention. As a
result, the present invention is better suited to premium image
quality applications requiring larger apertures.
[0057] In U.S. Pat. No. 6,482,160 "High Resolution 3D Ultrasound
Imaging System Deploying a Multidimensional Array of Sensors and
Method for Multidimensional Beamforming Sensor Signals",
Stergiopoulos and Dhanantwari describe a method for processing the
signals recevied from a 2D electronic, ultrasonic array. Both the
assumed transducer array, and the processing method employed are
fundamentally different from the present invention. The assumed
transmit sensor is a single, low-gain transducer element which
illuminates the entire region being imaged, causing ultrasound
energy to reflect back towards a receiving array. The region being
imaged is assumed to be in the far-field of the receive array, and
a conventional, 2D electronic scanning receive array is assumed. In
the case of the present invention, the transmit aperture is a
high-gain aperture (made up of several receive elements), and for
the case of the preferred CAC-BF method disclosed herein, the same
aperture is used for each transmit/receive scan-line pair. The
region being imaged can be in the near-field of the receiving array
of the present invention (which is the case for the carotid artery
application disclosed herein). The array element feature sizes are
not conventional for the disclosed, preferred 2D array transducer.
The beamforming method used in U.S. Pat. No. 6,482,160 is very
different that the CAC-BF method of the present invention. First,
the CAC-BF method works very well when the object being imaged is
in the near-field of the receiving array, but the method of U.S.
Pat. No. 6,482,160 only applies to far-field imaging. It employs
adaptive beamforming algorithms to increase spatial resolution over
that obtainable from conventional beamformers; however, the
inventors acknowledge that their beamformer's performance can
degrade significantly if the assumed noise characteristics are
inaccurate. The CAC-BF method of the present invention also
increases spatial resolution; however, it makes no assumptions
about the noise characteristics and hence is more robust.
BRIEF DESCRIPTION OF THE DRAWINGS
[0058] FIG. 1A is a schematic perspective view of a prior-art 1D
linear array of ultrasonic scanning sensing elements.
[0059] FIG. 1B is a schematic perspective view of a prior-art 1D
curved array of ultrasonic scanning sensing elements.
[0060] FIG. 2A is a schematic perspective view of the beam produced
by a prior-art 1D linear array of ultrasonic scanning or sensing
elements.
[0061] FIG. 2B is a schematic perspective view of the beam produced
by a prior-art 1.5D linear array of ultrasonic scanning or sensing
elements.
[0062] FIG. 3A is a schematic perspective view of a prior-art 1.5D
linear array of ultrasonic scanning or sensing elements with
sequential scanning in azimuth.
[0063] FIG. 3B is a schematic perspective view of a prior-art 1.5D
linear array with phased-array scanning in azimuth.
[0064] FIG. 4 is a schematic perspective view of a 2D ultrasonic
transducer array pursuant to the present invention.
[0065] FIGS. 5A and 5B are a diagram illustrating basic CAC-BF
concepts utilized in carrying out the present invention.
[0066] FIG. 6 is a block diagram showing functional components of
an ultrasound scanning system in accordance with the present
invention.
[0067] FIG. 7 is a block diagram showing elements of a fine
beamformer shown in FIG. 6.
[0068] FIG. 8 is a block diagram similar to FIG. 7, showing an
alternative configuration of the fine beamformer of FIG. 6.
[0069] FIG. 9 is a block diagram similar to FIGS. 7 and 8, showing
another alternative configuration of the fine beamformer of FIG.
6.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0070] FIG. 4 shows an ultrasound transducer probe where scanning
occurs in both the azimuth dimension (i.e., horizontally) and in
the elevation dimension (vertically). The probe includes a
rectangular array of transducer elements 36 mounted to a holder or
substrate member 38. Four different beams 40a, 40b, 40c, 40d are
illustrated in FIG. 4, demonstrating that the probe is capable of
illuminating a volume. The transducer elements 36 can be controlled
to effect sequential scanning in azimuth and elevation as
illustrated in FIG. 4; but nothing prevents one from using
phased-array scanning or CAC-BF (coherent aperture combining
beamforming) scanning (see below) in azimuth and/or elevation. If
phased array scanning is used in a given dimension, the acquired
data usually is scan converted (i.e. transformed or mapped onto a
Cartesion grid) in order to present the image on a conventional
display (i.e. a monitor using a cathode ray tube (CRT)). The net
effect of any of these scanning techniques is that a volume of
ultrasound data is ultimately acquired electronically, which
ultimately can be represented on a Cartesian (x-y-z) grid.
[0071] A preferred embodiment of the ultrasound transducer device
or probe of FIG. 4 for 3D imaging of the carotid artery is
characterized by the following baseline parameters:
[0072] 256.times.40 piezoelectric transducer elements 36
[0073] 0.2 mm (.lamda.).times.0.8 mm (4.lamda.) element spacing in
azimuth and elevation, respectively
[0074] Scans in azimuth and elevation
[0075] Uses 128-elements for azimuth instantaneous aperture
[0076] Uses 20% elevation instantaneous aperture (i.e. 8-element
subaperture)
[0077] TX focal depth is 50 mm
[0078] Imaging depth is typically 0 to 7 cm
[0079] Nominal azimuth resolution of F/2
[0080] Nominal elevation resolution (using 8 elevation elements per
beam) of F/8
[0081] High elevation resolution (using 16 elevation elements per
beam) of F/4
[0082] Frequency 7.5 MHz (central wavelength 0.2 mm)
[0083] Pulse length 0.4 mm (0.27 .mu.s) with Hann weighting,
yielding approximately 100% bandwidth
[0084] 128 digital receive channels
[0085] image volume 25.6 mm.times.25.6 mm.times.70 mm
[0086] It should be noted that any of the above parameters,
including the frequency, can be changed for other applications in
order to create other preferred embodiments, and such changes would
not depart from the spirit or scope of the probe in accordance with
the present invention.
[0087] An ultrasound transducer device or probe, as shown in FIG.
4, characterized by these parameters produces an image data cube of
dimension 25.6 mm (elevation) by 25.6 mm (azimuth) by 70 mm
(depth). (The dimension of the data cube can change without
changing the basic design of the 2D ultrasound transducer device or
carotid-scanning probe.) The 2D ultrasound transducer device or
carotid-scanning probe operates at 7.5 MHz so the wavelength,
.lamda., is nominally 0.2 mm. The ultrasonic carotid artery scanner
therefore uses elements that are nominally spaced (and sized,
neglecting kurfs) 4.lamda. (elevation) by .lamda. (azimuth). A key
feature of the ultrasound transducer device or carotid-scanning
probe is the large elemental size in elevation. Sequential scanning
is assumed to scan in the elevation dimension. The .lamda. spacing
in azimuth supports both sequential scanning, as well as
phased-array type scanning. Phased array scanning is usually
limited to about +/-45 deg to avoid grating lobes.
[0088] Consider the case of sequential scanning in azimuth, where
128 elements are used at any one time so the instantaneous azimuth
aperture is 128.times.0.2 mm=25.6 mm. At 50 mm depth, this leads to
an F/2 which is desired. Since it is desired that the image extent
in azimuth is also 25.6 mm, 256 elements are needed in total, where
128 of them are used for any given vector. In elevation, the total
array spans 40*0.8 mm=32.0 mm. However, a sub-aperture of 8
elements (consistent with state-of-the-art 1.5D arrays) spans 6.4
mm. Elevation vectors that are not degraded due to asymmetries must
not be closer than 3.2 mm (6.4/2) from the edge of the ultrasound
transducer device or carotid-scanning probe. Therefore the useable
elevation dimension is 32.0 mm-2*3.2=25.6 mm.
[0089] The azimuth dimension of stated volume can be larger than
25.6 mm because on transmit, one generally only uses an F/5 (i.e.
about 50 elements), instead of the 128 elements used on receive.
One should also note that if phased-array scanning is used in
azimuth, then only 128 elements are needed in azimuth (rather than
256 as for sequential scanning).
[0090] The aforementioned 2D transducer device of FIG. 4 has
several advantages over the conventional 2D array described
earlier. The conventional 2D array requires at least four times the
number of elements to scan the same volume (assuming the azimuth
and elevation element sizes are both .lamda.). If a conventional
probe restricts itself to the same number of elements as the 2D
transducer device disclosed herein, then it will also result in
lower spatial resolution and image quality because the physical
aperture will be smaller. As a result, the 2D transducer device in
accordance with the present invention is more practical, less
complex and less expensive than a similarly performing conventional
2D array.
[0091] For practicality and affordability, the preferred embodiment
of the present invention uses only say 128 receive channels,
consistent with that found in premium 2D ultrasound systems. To
form beams with state-of-the-art resolutions of F/2 in azimuth and
F/8 in elevation using the probe of FIG. 4, then 128.times.8=1024
elements are needed in the formation of each beam. If a higher
elevation resolution is desired, say F/4, then 16 elevation
elements are needed and hence 2,048 elements to form each beam.
This poses a problem, since one is restricted to only 128 receive
channels. Conventionally, synthetic aperture methods would be used
to acquire the received signals, 128 elements at a time, using at
least 8 and 16 shots, respectively, for the F/8 and F/4 beams. The
number of shots would be even larger if a synthetic aperture mode
is also required on transmission. The problem with using the
synthetic aperture approach for the 2D probe of FIG. 4 is that the
acquisition or frame time associated with acquiring the desired 3D
volume increases proportionately to the number of shots required.
As it is an object of the present invention to also reduce the
acquisition or frame time, the CAC-BF method has been developed for
this purpose. This method results in significant reductions in
acquisition time, while not degrading image quality.
[0092] CAC-BF scanning, discussed in detail below. forms part of
the present invention. It is used with the probe of FIG. 4, or any
other 1D or 2D probe, when it is necessary to minimize the
volumetric or image acquisition time. The baseline CAC-BF
configuration is a preferred embodiment of the CAC-BF method which
uses sequential scanning in elevation, combined with CAC-BF in
azimuth. For F/2 in azimuth and F/8 in elevation, the CAC-BF
subapertures are typically of dimension 16.sub.az.times.8.sub.cl
(i.e. each subaperture has 16 contiguous elements in azimuth and 8
in elevation for a total of 128), and twenty (20) subapertures are
used in total to span the 128 element azimuth aperture desired,
where adjacent, azimuth subapertures have an overlap of 10 azimuth
elements. Typically 10 phased-array type beams are formed for each
subaperture, and the collection of resulting beams from all
apertures are coherently combined to produce the final image. If
higher F/4 elevation resolution is desired, then a total of
forty-one (41) 8az .times.16.sub.el subapertures are typically
used, with adjacent subapertures having an overlap of 5 elements in
azimuth. For a given application, the size and number of the
subapertures, their overlap, and the number and spacing of beams
formed per subaperture are optimized to yield the required
performance.
Coherent Aperture Combining Beamforming: In General
[0093] In a preferred embodiment, the 2D ultrasound transducer
device or carotid-scanning probe of FIG. 4 uses a novel beamforming
technique referred to as coherent aperture combining beamforming
(CAC-BF) to achieve substantially reduced volume acquisition times
while maintaining imaging performance.
[0094] The general concept underlying CAC-BF will now be described
with reference to FIG. 5A. A volume 42 to be imaged is divided into
a Cartesian grid of points (voxels), nominally separated by the
achievable resolution in each dimension. A number of subapertures
44.sub.a, 44.sub.b, . . . 44.sub.n are defined within the
ultrasound transducer device or carotid-scanning probe. A number of
beams 46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2,
. . . 46b.sub.j,. . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk from
each subaperture 44.sub.a, 44.sub.b, . . . 44.sub.n are directed
towards different angles spanning the volume of interest 42.
High-resolution voxels are formed by summing signals from a number
of low-resolution (coarse) beams originating from different
subapertures 44.sub.a, 44.sub.b, . . . 44.sub.n. High resolution is
achieved because the summation results in the full aperture (i.e.
the total extent of the subapertures) being synthesised and
focussed at each voxel in the image.
[0095] The differences between a method and apparatus using
conventional techniques and a method and apparatus utilizing CAC-BF
are as follows. Conventional techniques are organized into two
categories: those using synthetic aperture beam formation, and
those not. We consider the latter first. In a conventional
ultrasound (without synthetic aperture beam formation), beams are
fully formed, both on transmit and receive, with a single shot. The
number of elements in the aperture is thus limited by how many
channels are available. The transmit aperture is restricted to be
smaller than the receive subaperture to provide a reasonable depth
of focus. This means that with the receive hardware currently
available (i.e. we do not want to increase the number of receive
channels in premium 2D ultrasound systems), high-resolution 2D
electronic scanning is not practical. The CAC-BF method differs
from high-resolution conventional beamforming in two key ways.
First, conventional beamformers do not use subapertures 44.sub.a,
44.sub.b, . . . 44.sub.n in the beamforming process. Second,
conventional beamformers transmit on a smaller aperture than they
receive on.
[0096] In conventional systems with `synthetic aperture`
beamforming, a number of subapertures (offset in azimuth only) are
focussed along a given range line on consecutive shots, and then
the return signals are summed in order to synthesize a single beam.
If a high-resolution transmit beam is also needed, then multiple
subapertures are used on transmit for each receive subaperture;
shots that have different transmit and receive subapertures are
`cross-terms` in the summation. Interestingly, cross-terms actually
decrease resolution, but do help to reduce sidelobes. Current
premium ultrasounds, when operating in synthetic aperture mode,
typically use only two subapertures on receive (There is usually
one direct transmit/receive and one cross-term transmit/receive.
For example, transmit on a central subaperture and receive on
central subaperture, followed by transmit on central and receive on
outer subaperture.). Several key differences exist between the
CAC-BF method and synthetic aperture beamforming. First, there are
no cross-terms in the baseline CAC-BF concept as used herein,
meaning that resolution is better for CAC-BF. Second, whereas
synthetic aperture beamformers form a single high-resolution beam
for each voxel, the CAC-BF method forms several coarse,
low-resolution beams and combines them at each voxel. Third, in a
preferred embodiment of CAC-BF, each low-resolution beam is formed
using transmit and receive subapertures that are the same size. In
synthetic aperture beamforming, the transmit and receive
subapertures used for beam formation are different in size, the
transmit being smaller.
[0097] In a given region, the CAC-BF concept contemplates that each
low-resolution beam is transmitted and received from the same
(sub)aperture. High-resolution `beams` are not really formed;
rather a high-resolution aperture is synthesized at each voxel.
Multiple low-resolution beams from each subaperture cover a region
spanning many beamwidths. The ultrasound imaging process utilizing
CAC-BF breaks up the existing large aperture.
[0098] A conventional ultrasound may also transmit multiple shots
to get better depth of field (one shot for each range interval).
This is because the transmit beam must be focussed at a particular
range, and only voxels for ranges within its depth of focus can use
the beam. To cover a wide range swath, a sequence of shots are
transmitted along each range line, each focussed at a different
range. This forces the system to take more time to cover the
volume. The higher the resolution, the smaller is the depth of
focus, and hence the longer it takes to image a volume. With the
disclosed CAC-BF method, the depth of focus is defined by the
resolution of the coarse beams, giving it a natural advantage over
state-of-the-art beamforming methods. This is a fundamental
difference between conventional beamforming methods and the CAC-BF
method. The preferred embodiments of the CAC-BF described herein
only require a single shot per range line or image vector.
[0099] The ultrasound transducer device or carotid-scanning probe
(see FIG. 4) is divided or partitioned into overlapping
subapertures 44.sub.a, 44.sub.b, . . . 44.sub.n (FIG. 5A). These
are composed of (say) 64 to 128 elements to match the available
number of signal receive channels. In a preferred embodiment, the
subapertures 44.sub.a, 44.sub.b, . . . 44.sub.n overlap by at least
50% of their width in each dimension (azimuth or elevation) where
CAC-BF is applied. The percentage overlap strongly affects the
impulse response of the resulting, high-resolution, CAC-BF image.
Grating lobes may result if enough overlap is not selected. FIG. 5
shows a 1D array partitioned along the azimuth dimension into
subapertures 44.sub.a, 44.sub.b, . . . 44.sub.n. CAC-BF is
illustrated below for this single azimuth dimension. Extension of
CAC-BF applied to two dimensions (i.e. azimuth and elevation) is
straightforward.
Coherent Aperture Combining Beamforming: Image Partition
[0100] The image space 42 is divided into overlapping beams
46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2, . . .
46b.sub.j, . . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk from each
subaperture 44.sub.a, 44.sub.b, . . . 44.sub.n. FIG. 5A shows the
beam boresights as lines originating from the subapertures
44.sub.a, 44.sub.b, . . . 44.sub.n, and travelling through the
volume 42. Each subaperture 44.sub.a, 44.sub.b, . . . 44.sub.n
transmits and receives a respective sequence of overlapping
(coarse) phased-array beams 46.sub.a1, 46.sub.a2, . . . 46.sub.ai,
46.sub.b1, 46.sub.b2, . . . 46b.sub.j, . . . 46.sub.n1, 46.sub.n2,
. . . 46.sub.nk, each beam being focussed at a different angle. A
pulse (shot) is transmitted and a range line is received for each
beam 46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2, .
. . 46b.sub.j, . . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk from
each subaperture 44.sub.a, 44.sub.b, . . . 44.sub.n. The beams for
each subaperture are normally spaced so that they cross at
approximately their -3 dB points. To avoid grating lobes, the total
angle (volume) spanned by the beams 46.sub.a1, 46.sub.a2, . . .
46.sub.ai, 46.sub.b1, 46.sub.b2, . . . 46b.sub.j, . . . 46.sub.n1,
46.sub.n2, . . . 46.sub.nk from each subaperture 44.sub.a,
44.sub.b, . . . 44.sub.n is usually limited by the reciprocal of
the element spacing weighted by a constant that takes into account
unit conversion. This consequently limits the size of the full
aperture that can be synthesized. Only beams that intersect the
imaged volume 42 need be transmitted, thereby saving additional
acquisition time; thus subapertures at the edges of the volume
transmit fewer beams.
Coherent Aperture Combining Beamforming: Image Formation
[0101] As illustrated in FIG. 6, a two-dimensional array 48 of
transducer elements mounted to a probe (not shown in FIG. 6) is
accessed by switching electronics 50. Switching electronics 50
includes a signal generator 52, a control unit 54 and a switching
network 56. Signal generator 52 produces a waveform having a
characteristic ultrasound frequency that is directed to elements of
transducer array 48 by switching network 56 in response to signals
from control unit 54. Switching electronics 50 selectively
energizies the elements of array 48 and selectively polls those
elements to effectively divide the array, along at least one of two
dimensions, into subapertures 44.sub.a, 44.sub.b, 44.sub.n. As
discussed above, each subaperture 44.sub.a, 44.sub.b, . . .
44.sub.n transmits and receives a respective plurality of
low-resolution ultrasound beams 46.sub.a1, 46.sub.a2, . . .
46.sub.ai, 46.sub.b1, 46.sub.b2, . . . 46b.sub.j, . . . 46.sub.n1,
46.sub.n2, . . . 46.sub.nk that span the volume 42 to be imaged. A
signal processor 58 is operatively coupled to the switching
electronics 50 for coherently combining received beamformed signals
from the subapertures 44.sub.a, 44.sub.b, . . . 44.sub.n and
synthesizing, from the coherent combination, a large aperture
focused at each point of the image volume 42.
[0102] The image formation process loops on beams 46.sub.a1,
46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2, . . . 46b.sub.j,
. . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk and subapertures
44.sub.a, 44.sub.b, . . . 44.sub.n (2 nested loops), collecting
range lines (the sampled signal in range). From each shot, return
signals are received from the transducer elements of the
transmitting/receiving subaperture 44.sub.a, 44.sub.b, . . .
44.sub.n. These signals are digitized by a digitizer 60 (FIG. 6)
and (coarse) beamformed by a module 62, with dynamic focussing
along the radial line from the phase center of the subaperture
through the transmit focal point, as is usually done.
[0103] In this preferred embodiment, the next operation performed
on each line is range filtering, performed by a range filter 64.
This operation is linear for the fine beamforming step to work
optimally, and it retains the complex-valued nature of the signal;
i.e. the output remains complex (I and Q). A conventional bandpass
filter can be applied (matching the waveform bandwidth), or
alternatively, a matched filter can be used and applied to the
ultrasound signals; in this case, a preferred approach is to base
the matched filter on the pulse replica (as the real part of the
kernel) and its Hilbert transform (as the imaginary part). Matched
filtering with the transmit pulse is logically done after coarse
beamforming, but before fine beamforming, since fine beamforming
(module 66) removes the range lines. Range filtering may also be
omitted depending on the waveform used.
[0104] One advantage of the way coarse beamforming and range
filtering is performed is that a 2D ultrasound engine could be used
for these operations, making the preferred 2D probe and the CAC-BF
method amenable for upgrading existing premium ultrasound
systems.
[0105] The resulting coarse beams are transferred to the fine
beamforming module 66. Coarse beamformer module 62 and fine
beamformer module 66 may be realized by generic digital processor
circuits modified by respective programming algorithms to
accomplish the respective beamforming operations.
[0106] A conventional ultrasound also loops on beams in a similar
manner, but our invention uses a unique set of different beams,
differing in both the elements used and the focused directions.
With conventional ultrasound, each image point is typically
generated from the nearest high-resolution beam which is generated
from one or more shots. With the CAC-BF method on the other hand,
each image point is generated from an associated set of nearby
low-resolution beams, each generated from an associated shot.
Coherent Aperture Combining Beamforming: Fine Beamforming
[0107] The image space is divided into a high-resolution grid of
voxels. The voxels are spaced more finely than the coarse beams
46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2, . . .
46b.sub.j, . . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk, nominally
at the achievable resolution in cross-range from the synthesized
apertures. In FIG. 5B two beam boresights (range lines) 46.sub.a1,
46.sub.a2 and 46.sub.b1, 46.sub.b2 from each of two subapertures
44.sub.a and 44.sub.b, along with four grid points 72.sub.a,
72.sub.b, 72.sub.c, 72.sub.d, are shown. The intensity at each
image point (voxel) 72.sub.a, 72.sub.b, 72.sub.c, 72.sub.d is the
coherent sum of signals received from the various nearby
subaperture-beam shots. The sum is from subapertures across the
array, thereby synthesizing a larger aperture. From a given
subaperture, each voxel's sum preferably includes the two nearest
beams that straddle the given voxel 72.sub.a, 72.sub.b, 72.sub.c,
or 72.sub.d. For example, for voxel 72.sub.d and subaperture
44.sub.a, the two beams 46.sub.a1 and 46.sub.a2 are used. More
particularly, signal sample points 73 and 73' are used. The spatial
interpolation weight for each of the two beams is such that the
pattern of the interpolated beam reaches a maximum (peaks) at the
voxel. For image points between the beams, this not only helps the
signal-to-noise ratio, but it also reduces the sidelobe pattern of
the interpolated beam. Time-interpolation of the signal sample from
each shot is also preferrably included in the summation at each
voxel, as illustrated in FIG. 5B. Various methods for spatial and
temporal interpolation known to those skilled in the art all fall
within the scope of the fine beamforming method, as does using a
different number of beams or time samples for interpolation. FIGS.
7 and 8 illustrate possible modular combinations of a spatial
interpolator 74, a temporal interpolator 76, and an adder 78. The
result of applying the aforementioned fine beamforming algorithm is
that the signal samples from a scatterer at the voxel all peak and
add up in phase when summed. The range delay (which determines the
phase) for a given voxel's signal sample is equal to the range
delay between the subaperture phase center and the voxel. Line 68
(FIG. 5B) represents points with approximately the same range delay
to voxel 72.sub.d from subaperture 44.sub.a and line 70 represents
points with approximately the same range delay to voxel 72.sub.d
from subaperture 44.sub.b.
[0108] Another preferred embodiment of the fine beamforming
algorithm is depicted schematically in FIG. 9. The
coarse-resolution range lines from each subaperture 44.sub.a,
44.sub.b, . . . 44.sub.n are scan-converted by a module 80 (using
conventional scan conversion algorithms known to those skilled in
the art) to the high-resolution Cartesian grid of voxels to form a
low-resolution subaperture image. In this operation, the complex
nature of the signal (amplitude and phase) are retained. The
subaperture images are added together by a module 82 to synthesize
a larger aperture, and result in the final, high-resolution image.
This addition operation can be with unity weights, or
alternatively, non-unity weights to effect a taper. One advantage
of this embodiment is that low-resolution images can be created
quickly at a high frame rate. Higher and higher-resolution images
can be obtained by using more and more subapertures and combining
their respective low-resolution images.
[0109] The high-resolution image, once formed using CAC-BF as
described above, can be further processed and/or transformed using
image processing methods known to those skilled in the art. The
image can be rectified (i.e. converted to an amplitude or power) or
its real and/or imaginary parts can be processed.
[0110] A key advantage of using a CAC-BF approach is that
high-resolution beams are obtained in less time for given number of
(element) channels. This advantage is demonstrated in the
discussion that follows.
[0111] A problem with conventional ultrasound beamforming, when
applied to 2D scanning, is that it can take multiple (synthetic
aperture) shots to form a high-resolution beam. This happens when
the beam requires more elements than there are channels. This can
make acquisition time unwieldy, especially for 3D ultrasound. For
example, it takes at least 52 seconds to image a volume of 25.6 mm
by 25.6 mm by 7 cm using .lamda.-spaced elements in azimuth and
elevation; and at least 13 seconds when 4.lamda.-elements rather
than .lamda.-sized are used in the elevation dimension, when only
128 channels are available. This assumes sequential scanning in
both azimuth and elevation with a beam step of .lamda. in azimuth
and 2.lamda. in elevation, and an F/4 elevation resolution.
[0112] Consider first the .lamda.-spaced-element case in azimuth
and elevation. In this case, in order to support the aforementioned
undistorted volume with F/2 in azimuth and F/8 in elevation (F/4 in
elevation results in an undistorted elevation volume dimension that
is less than 25.6 mm), the array is 256.lamda. by 160.lamda. with
256 elements in azimuth and 160 in elevation, for a total of 40,960
elements. For F/4 and F/8 in elevation, the receive subaperture has
128.times.64=8,192 elements and 128.times.32=4,096 elements,
respectively. The number of beams required to sample the volume is
also 128.times.64=8,192 assuming a beam step of .lamda. in azimuth
and 2.lamda. in elevation. Using a 10 kHz firing rate (suitable for
7 cm depth), the minimum acquisition time (i.e. with 1 shot per
beam) is 0.8192 s, assuming that 8,192 receive channels and 4,096
receive channels are available, respectively, for F/4 and F/8
elevation resolution. Of course, building a system with these many
channels is extremely expensive and impractical today. If only 128
channels are available (i.e. as found in modern premium systems),
then at least 8,192 elements/128 channels=64 shots/beam and
4,096/128=32 shots/beam are needed, causing the acquisition times
to increase to 52.4 s and 26.2 s, for F/4 and F/8 respectively.
These numbers assume a single transmit focus. If multiple transmit
focii are used (up to 4 are used in practice), then the acquisition
times increase proportionately.
[0113] Consider next the case where 4.lamda.-elements rather than
.lamda.-sized are used in the elevation dimension. Then the number
of elevation elements per beam reduces to 16 and 8, for F/4 and
F/8, respectively. As a result, the number of elements in each
receive subaperture reduces to 2,048 and 1,024 respectively. If
2,048 and 1,024 receive channels are available for the respective
F/4 and F/8 elevation resolutions, then the acquisition time is
again 0.8192 s. With only 128 available channels, however, this
acquisition time increases to 13.1 s and 6.55 s, respectively. Once
again, these times increase proportionately with the number of
transmit focii used.
[0114] With the present CAC-BF method, low-resolution beams
46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2, . . .
46b.sub.j, . . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk are used,
needing fewer elements per beam, so that potentially only one shot
is needed per beam. The low-resolution beams 46.sub.a1, 46.sub.a2,
. . . 46.sub.ai, 46.sub.b1, 46.sub.b2, . . . 46b.sub.j, . . .
46.sub.n1, 46.sub.n2, . . . 46.sub.nk cover a greater volume, but
beams are needed from more (smaller) subapertures 44.sub.a,
44.sub.b, . . . 44.sub.n in order to get resolution. The volume is
covered by the `product` of subapertures 44.sub.a, 44.sub.b, . . .
44.sub.n and beams 46.sub.a1, 46.sub.a2, . . . 46.sub.ai,
46.sub.b1, 46.sub.b2, . . . 46b.sub.j, . . . 46.sub.n1, 46.sub.n2,
. . . 46.sub.nk. This can be accomplished with many subapertures,
and few beams per subaperture (coarse beams, few elements per
beam), or with just a few subapertures, with many beams per (finer
beams, many elements per beam) subaperture. The product of the two
(the total number of beams) is, to first order, independent of
subaperture size. Thus with the present method, a certain number of
beams 46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2,
. . . 46b.sub.j,. . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk are
required to cover a given volume at a given resolution, and
coverage time is proportional to volume/resolution. Once the beams
46.sub.a1, 46.sub.a2, . . . 46.sub.ai, 46.sub.b1, 46.sub.b2, . . .
46b.sub.j,. . . 46.sub.n1, 46.sub.n2, . . . 46.sub.nk are low
enough resolution (i.e. they have a small enough number of
elements), they only need one shot. Thus one of the key advantages
of our method: when down to 1 shot for each beam, we have minimized
coverage time. The aforementioned volume can now be imaged in under
1.6 seconds for F/4 in elevation, and in under 0.7 seconds for F/8
in elevation. The calculations are illustrated below.
[0115] Consider first the F/4 elevation case with
4.lamda.-elements. With 16 elements needed in elevation and 128
channels available, one preferred CAC-BF solution is to use a
receive subaperture containing 8 elements in azimuth and 16
elements in elevation. Using 41 subapertures 44.sub.a, 44.sub.b, .
. . 44.sub.n to span the 128 azimuth elements with a preferred
overlap of 5 elements between adjacent subapertures (in azimuth), a
preferred 6 beams can be used to cover the 25 mm in azimuth. As a
result, the number of beams (shots) needed to acquire the whole
volume is 6 (per subaperture).times.41 (subapertures).times.64
(elevation beams)=15,744, which translates to an acquisition time
of 1.57 s. Again, we have assumed a beam step of 2.lamda. in
elevation. Furthermore, because of the large depth of focus
associated with the CAC-BF technique, a single transmit focus
should suffice for virtually all imaging applications. This results
in eight times the frame rate over the conventional beamforming
technique, even when only a single transmit focus is employed with
conventional imaging.
[0116] Finally, consider the F/8 elevation case with
4.lamda.-elements and 128 available channels. With only 8 elements
needed in elevation for each receive subaperture, one preferred
CAC-BF solution is to use a receive subaperture containing 16
elements in azimuth and 8 elements in elevation. With 20
subapertures overlapped by a preferred 10 elements in azimuth to
span the entire 128, 10 beams are preferably needed by each
subaperture to cover the 25 mm in azimuth. As a result,
10.times.20.times.64=12,800 shots are needed which translates to
1.28 s for acquisition. The CAC-BF technique results in five times
the frame rate over the conventional beamforming technique, in this
case, assuming the conventional beamformer uses a single transmit
focus. If the elevation beam step was increased to 4.lamda., then
the frame time would reduce to 0.64 s using CAC-BF.
[0117] One advantage of the present methodology is that the
aperture size can be tailored to every grid point 72.sub.a,
72.sub.b, 72.sub.c, 72.sub.d (FIG. 5B). The preferred algorithm
(using all available intersecting beams) naturally uses more
apertures (on both transmit and receive) for grid points at greater
ranges. With this embodiment, azimuth resolution (in mm) is
constant with range. In a conventional ultrasound, the number of
elements making up the aperture is not allowed to grow beyond the
number of available channels. In the case of the present invention,
the effective aperture can grow to the full size of the physical
aperture, exceeding the number of available channels, to the extent
limited by the element directivity and desired grating lobe
performance.
[0118] Another advantage of the CAC beamforming algorithm is that
it can be combined with other conventional scanning approaches so
that certain parts of a volume to be imaged use CAC beamforming
while other parts use conventional beamforming. For example, CAC
beamforming could be applied only at further ranges where
resolution degrades and conventional beamforming used
elsewhere.
[0119] Another advantage of the methodology described herein is
that better depth of field is obtained with lower resolution beams.
This means that an equivalent 3D volume can be covered with fewer
shots, or a greater volume can be covered with the same number of
shots.
[0120] Yet another advantage of the methodology described herein is
that better resolution is achieved with the same size aperture
(because of the lack of cross-terms). Resolution is trio times
better than that of a receive-only aperture (i.e. a system that
uses a significantly lower resolution transmit aperture), and 1.4
times better than that of a conventional beam using full apertures
on transmit and receive. It is to be noted that the full aperture
is not normally used on transmit because of the limited depth of
field, thus CAC-BF gets almost twice the resolution of conventional
systems using the same sized physical aperture. Analyses and
experimental measurements show that CAC-BF with 50% overlap
performs equivalently to a conventional synthetic aperture of twice
its size.
[0121] Other variations to the CAC-BF method are described to
illustrate the scope of the CAC-BF method in accordance with the
present invention.
[0122] CAC-BF can be performed with an arbitrary amount of
subaperture overlap, recognizing that the resulting image (beam)
response characteristics (e.g. the sidelobe behavior including the
presence of grating lobes) at an image point will be affected
accordingly.
[0123] It is to be noted that CAC-BF can be done in two dimensions,
or, alternatively, CAC-BF can be performed in one dimension and
conventional scanning in the other. CAC-BF can also be used with
one-dimensional (1.5D, 1.75D etc.) probes to increase the frame
rate for a given number of channels. The frame rate is further
improved by the fact that the depth of focus is greater, reducing
the number of transmit focii needed. Alternatively, larger
effective apertures (better resolution) can be realized without
reducing the frame rate. CAC-BF can also be used and tailored to
work with systems having virtually any number of receive signal
channels.
[0124] It is to be noted also that the fine grid within the image
space need not be Cartesian and that the coarse beams need not be
spaced equally in angle. For example, the beams could be spaced
equally in sine-space, or spaced equally in the Cartesian grid.
When CAC-BF is applied in both dimensions, the coarse beams could
be placed on a grid that is not the product of an azimuth and an
elevation grid (e.g. hexagonal or cylindrical scanning).
[0125] It is possible to include subaperture cross-terms, in order
to improve sidelobes. Moreover, it is possible to transmit from one
subaperture only, and receive from the rest of the
subapertures.
[0126] This has the disadvantage of only getting half the
resolution per length of aperture, but gets equivalent resolution
to CAC-BF per pulse, because aperture overlap is not needed. A
potential advantage is a reduction in hardware complexity (may not
need transmit multiplexer, or it will be simpler). The reciprocal
arrangement (transmit from all subapertures, only receive from
middle one) may also be attractive.
[0127] Interpolation between beams helps reduce the sidelobes at
grid points where the beams from different subapertures don't line
up, effectively smoothing the addition of the subapertures at these
points. The interpolation can be of any desired amount, using any
algorithm known to those skilled in the art. It is possible not to
use any interpolation but performance will be affected
accordingly.
[0128] Shading (windowing) may be done in the summation across the
aperture (to reduce sidelobes with narrow-band systems). For
similar reasons, or alternatively, the subapertures themselves
could be shaded or windowed.
[0129] Higher resolution coarse beams (requiring multiple shots)
could be utilized, trading off coverage time versus sidelobes.
Dynamic focussing may be unnecessary if beams are of low enough
resolution (i.e. very small subapetures), and this may reduce
complexity.
[0130] The `product` of beams and subapertures need not have each
of the nominally 50% overlapping subapertures transmitting all of
the coarse beam angles. The product could be formed with a greater
number of highly-overlapped subapertures, each transmitting a
smaller number (e.g. one) of the beam angles. This has the
advantage of having smaller blind zones at the close ranges between
the subaperture centres, where no beams are transmitted.
[0131] Non-linear or adaptive, high-resolution beamforming
techniques are computationally expensive, but may be worthwhile in
some applications to combine the multiple subapertures used in the
fine beamforming algorithm. The structure of the CAC-BF method is
appropriate as there is typically a small number of subapertures.
This is not burdensome if it takes a few seconds or minutes to
compute; a physician could look at a linearly beamformed image, and
suggest an area he would like to see better resolved; then the
high-resolution algorithm could be applied.
Coherent Aperture Combining Beamforming: 2D Scanning
[0132] With 2D electronic scanning, the designer has a number of
choices for which methods to use in each dimension. The choices
include element size (n*lambda, where n can be between 0.5 and 4),
what type of scanning (phased, sequenced, CAC-BF), the number of
elements in the aperture, and with CAC-BF, the subaperture sizes
and overlap. As element size goes up, the cost to achieve a certain
level of resolution goes down, but sidelobe performance degrades.
Each element size has a maximum achievable resolution, and larger
elements have poorer performance. For example, 4-lambda elements
cannot do better than about 0.35 mm azimuth resolution, whereas
lambda elements can achieve about 0.2 mm azimuth resolution at 7.5
MHz. Systems using CAC-BF in 2D with lambda or 2-lambda elements in
both dimensions are viable compromises. With CAC-BF, the larger the
subapertures, the costlier (i.e. more channels are needed), but
sidelobe performance is better. Once the 1D performance of each
alternative is established by testing, then 2D
cost/performance/acquisition-time trade-offs can begin. A key
feature of the CAC-BF algorithm is that it naturally provides the
designer with this cost/performance/acquisition time trade-off.
[0133] With 2D CAC-BF, in order to deal with motion within the
imaged volume, one can transmit the beams ordered within the
volume, i.e. all the beams in top left corner first, then each row
left to right, rows ordered top to bottom. In this way, each grid
point is illuminated over a short period of time. This is exactly
true only in focal plane, grid points at longer or shorter ranges
taking somewhat longer to illuminate. The alternative of ordering
by subaperture means that every grid point requires the whole
sequence of pulses to be imaged.
[0134] The CAC-BF method has been simulated extensively and its
performance as described herein validated by these simulations. In
addition, experimental results have been obtained by applying
CAC-BF in the azimuth dimension as described herein using a
64-channel ultrasound system and a 192-element off-the-shelf probe,
suitably programmed to implement the CAC-BF method. The improved
resolution, the high-quality imagery, and the reduction in
acquisition time compared to conventional beamforming have all been
confirmed. Images (2D or 3D) may be generated on a video monitor 84
(FIG. 6) by an image generator 86 in response to image data stored
in a memory 88 connected to output of processor 58, more
particularly to an output of a CAC component 90, and even more
particularly to an output of fine beamformer 66. CAC component 90
includes digitizer 60, coarse beamformer 62, range filter 64 and
fine beamformer 66.
3D Ultrasound Imaging Systems
[0135] An ultrasound imaging system in accordance with the present
invention provides a novel solution for 3D ultrasound imaging that
is affordable, and yet high-performing. Unlike other designs which
degrade state-of-the-art imaging performance in order to reduce
cost, the present solution maintains or exceeds state-of-the-art
imaging performance, while keeping the 3D ultrasound system cost
comparable to that of 2D ultrasound systems.
[0136] State-of-the-art azimuth imaging performance requires an F
number of 2; i.e. an F/2. The F number is the ratio of the focal
range divided by the imaging aperture dimension. For example, an
F/2 at 5 cm depth requires an instantaneous receive aperture of
size 2.5 cm. The present probe is designed to provide an F/2, and
an enhanced resolution of F/1 (i.e. twice as good as
state-of-the-art).
[0137] State-of-the-art-elevation imaging performance requires an
F/8. The present probe provides a standard elevation resolution of
about F/8 and is capable of providing an enhanced elevation
resolution of F/4 (twice as good as state-of-the-art) or
better.
[0138] For state-of-the-art resolutions in both azimuth and
elevation, the 2D probe of the present invention (for 3D imaging)
is able to acquire a 25 mm.times.25 mm.times.70 mm volume
electronically in under 1 second; the same volume is acquired with
enhanced resolutions (azimuth and elevation) in under 2 seconds.
And these acquisition times are achievable when just 128 receive
channels are available, keeping the number of channels (and hence
cost) comparable to state-of-the-art ultrasound imaging systems
(for 2D imaging).
[0139] This win-win (performance-cost) 3D imaging technology is
made possible from the use of the beamforming techniques of the
present invention that reduce acquisition time (i.e. the number of
shots needed) when the number of received channels available is
less than the number of elements in the imaging aperture. The
beamforming techniques are referred to as coherent aperture
combining beamforming (CAC-BF) as discussed earlier.
[0140] In order to reduce the element count and simplify the
transducer design, the 2D probe pursuant to the present invention
is able to exploit 1.75D elemental technology, with .lamda. spacing
in azimuth, and 4.lamda. spacing in elevation in one preferred
embodiment. As a result, its baseline transducer requires only 10k
elements, 256(.lamda.).times.40(4.lamda.)=10,240, in order to
produce an undistorted volume extending 25.6 mm in azimuth and 25.6
mm in elevation. The beamforming is completely electronic (i.e. no
mechanical lenses are used in elevation). Individual receive beams
use just 128.times.8=1,024 elements for standard, state-of-the-art
imaging resolutions, and 128.times.16=2,048 elements for enhanced
resolution imaging.
[0141] The 2D probe can be used with conventional beamforming
algorithms, as well as with CAC-BF, making it very versatile,
saving on the number of probes otherwise required by an ultrasound
system. For example, it can operate as a conventional 1D array,
employing conventional scanning techniques such as sequential or
phased-array scanning in azimuth (or elevation) only. If 2D
scanning is desired, then conventional electronic scanning can be
performed in both azimuth and elevation. For example, if sequential
beamforming is used in both azimuth and elevation, then for
standard imaging resolutions (F/2 in azimuth and F/8 in elevation)
128 beams are needed in azimuth and 32 are needed in elevation to
fill the volume. For the case of 70 cm depth, an acquisition time
of 6.5 seconds results with two transmit focal ranges (128.times.32
vectors.times.8 shots/vector.times.2 focii/10 kHz). For enhanced
imaging resolutions, 64 elevation beams are needed along with 16
shots/beam (2048 elements/128 channels) resulting in an acquisition
time of 26.2 seconds.
[0142] With CAC-BF, the beamforming can be configured in a number
of ways (as described earlier) to reduce the volume acquisition
times required if only conventional beamforming algorithms were
used. For standard imaging resolutions, consider the case where 20
overlapped subapertures, each consisting of 16 (in azimuth) by 8
(in elevation) elements are used with 10 shots (beams) per
subaperture. CAC-BF is performed in azimuth while sequential
beamforming is performed in elevation. The acquisition time for
this configuration has already been shown to be just 0.64 seconds
(10 shots/subap.times.20 subaps.times.32 elev_beams/10 kHz), a
substantial reduction over the 6.5 seconds required using only
sequential beamforming. For enhanced imaging resolutions, recall
the case where 41 overlapped subapertures, each consisting of 8 (in
azumith) by 16 (in elevation) elements are used with 6 shots
(beams) per subaperture. In this case, the volume acquisition has
been shown to take only 1.57 seconds (6 shots/subap.times.41
subaps.times.64 elev_beams/10 kHz), as compared to the 26.2 seconds
needed by the sequential beamformer.
[0143] Although the invention has been described in terms of
particular embodiments and applications, one of ordinary skill in
the art, in light of this teaching, can generate additional
embodiments and modifications without departing from the spirit of
or exceeding the scope of the claimed invention. While the
preferred embodiment described herein represents the form of the
invention currently being developed for its carotid artery
application, the scope of this invention goes far beyond the form
of this preferred embodiment. For example, the CAC-BF solution can
be used in the elevation dimension, instead of azimuth, or it could
be used in both dimensions, and still be within the scope of the
invention. Alternatively, the number and size of elements in each
dimension of the transducer could be different, and still fall
within the scope of the invention. In the limiting case, the 2D
transducer array could collapse to be a 1D array (i.e. designed to
scan in only one dimension), and if CAC-BF is used in that single
dimension, this configuration is still within the scope of the
invention. Accordingly, it is to be understood that the drawings
and descriptions herein are proffered by way of example to
facilitate comprehension of the invention and should not be
construed to limit the scope thereof.
* * * * *