3-D Ultrasound Imaging Using Forward Viewing
CMUT Ring Arrays for Intravascular and
Intracardiac Applications
David T. Yeh
∗
,
¨
Omer Oralkan
∗
, Ira O. Wygant
∗
, Matthew O’Donnell
†
, and Butrus T. Khuri-Yakub
∗
∗
Edward L. Ginzton Laboratory
Stanford University, Stanford, CA 94305–4088
Email: dtyeh@stanford.edu
†
Biomedical Engineering Department
University of Michigan, Ann Arbor, MI 48019–2099
Abstract— Forward-viewing ring arrays can enable new appli-
cations in intravascular and intracardiac ultrasound. We have
demonstrated full synthetic phased array volumetric ultrasound
imaging using a forward-viewing CMUT ring array with 64
elements, in both the conventional (8 MHz) and collapse (19
MHz) regimes of operation. Measured SNR of an echo from a
plane reflector at 5 mm is 29 dB for 8 MHz and 35 dB for 19
MHz. The 6-dB axial and lateral resolutions for the B-scan of
the wire target is 189 µm and 0.112 radians for 8 MHz, and
78 µm and 0.051 radians for 19 MHz. Rendered 3-D images of
a Palmaz-Schatz stent are also shown, demonstrating that the
imaging quality is sufficient for clinical applications.
I. INTRODUCTION
Forward-viewing intravascular ultrasound enables new pro-
cedures in medicine such as diagnosing severely occluded
blood vessels or guiding the placement of stents. A ring-
shaped forward-viewing transducer provides clearance for the
guidewire in catheter-based applications. Also, the forward-
viewing ring array is capable of volumetric imaging, which
is highly desirable because it reduces operator dependence in
clinical ultrasound.
Previous efforts have found it challenging to design and fab-
ricate ring arrays using piezoelectric transducers with sufficient
performance in the forward-looking mode [1]. Consequently,
there have been several efforts to make ring arrays using
Capacitive Micromachined Ultrasonic Transducers (CMUTs)
with the goal of volumetric intravascular ultrasound [2], [3].
This paper presents the characterization of a CMUT ring array
and its imaging capabilities.
II. CMUT R
ING ARRAYS
CMUTs offer several advantages over piezoelectric trans-
ducers for use in medical imaging [4]. The microlithography
process used to make CMUTs churns out batches of transduc-
ers with the fine dimensions required for high-frequency ring
arrays. The wide bandwidth of CMUTs in immersion improves
the resolution of ultrasound images.
Operating the CMUT in collapse affords particularly good
characteristics for imaging, including higher echo signal levels
and higher frequency [5]. In addition, the CMUT can be
Fig. 1. (a) Ring array wire bonded to electronics; (b) 16-channel trans-
mit/receive circuit; (c) 64-element CMUT ring array.
switched between its two operating modes during the imaging
procedure. The operator can choose conventional mode opera-
tion for lower frequency and better penetration for navigation,
or collapse mode for higher frequency and resolution for
diagnosis.
The parameters for the CMUT ring array presented are
as follows: ring diameter, 2 mm; number of elements, 64;
element pitch, 102 µm; element size, 100×100 µm; cells per
element, 9; cell membrane radius, 13 µm; electrode radius,
9 µm; membrane thickness, 0.4 µm; gap distance, 0.15 µm;
collapse voltage, 50 V. The very small element area required
for this application makes ultrasound imaging difficult, yet
the CMUT provides sufficient performance to produce clear
images.
III. E
XPERIMENTAL SETUP
Four integrated circuits, with 16 independent pulsers and
amplifiers each, were wire bonded to the elements of the
CMUT in a 209 pin grid array (PGA) electronics package, as
shown in Fig. 1. Although this arrangement is capable of full
phased array operation, full synthetic phased array imaging
was performed to simplify the data collection and to acquire
the most general data set for offline reconstruction with various
beamforming schemes. The CMUT was biased at 30 V for
6.5 7 7.5 8 8.5
−2
0
2
Time (µs)
Voltage (mV)
(a)
6.5 7 7.5 8 8.5
−2
0
2
Time (µs)
Voltage (mV)
(b)
Fig. 2. Pulse-echo response: (a) Conventional; (b) Collapse mode;
0 5 10 15 20 25 30 35 40
−40
−30
−20
−10
0
Frequency (MHz)
Normalized Magnitude (dB)
(a)
0 5 10 15 20 25 30 35 40
−40
−30
−20
−10
0
Frequency (MHz)
Normalized Magnitude (dB)
(b)
Fig. 3. FFT of filtered pulse-echo response: (a) Conventional; (b) Collapse
mode.
conventional operation and 100 V for collapse mode. Elements
were excited one at a time with a 25-V pulse, and A-scans
were acquired from the entire array for each transmit element.
Each A-scan was collected at a sampling rate of 500 MS/s for
both conventional and collapse mode operation, and using 16
averages.
IV. R
ESULTS
A. A-Scan Results
Pulse-echo data of a plane reflector (the oil-air interface) at
5 mm was taken from a single element. For the conventional
case, the pulse width was 60 ns; for collapse mode, the pulse
width was 27 ns. The echo, shown in Fig. 2, demonstrates the
wide bandwidth of the CMUT. Fig. 3 shows the FFT of the
pulse-echo signal after it has been filtered using a Gaussian
bandpass filter with a 6-dB band from 5.5 to 13 MHz for
0 16 32 48 64
6
8
10
12
Array Element Index
f
o
(MHz)
−3 σ
+3 σ
µ
µ = 8.3 MHz
σ = 307 kHz
(a)
0 16 32 48 64
0
50
100
Array Element Index
Bandwidth (%)
−3 σ
+3 σ
µ
µ = 69.7%
σ = 9.4 %
(b)
Fig. 4. Uniformity across ring array, conventional mode: (a) Center
frequency; (b) Fractional bandwidth.
0 16 32 48 64
17
18
19
20
21
Array Element Index
f
o
(MHz)
−3 σ
+3 σ
µ
µ = 18.9 MHz
σ = 187 kHz
(a)
0 16 32 48 64
0
50
100
Array Element Index
Bandwidth (%)
−3 σ
+3 σ
µ
µ = 69.0%
σ = 1.57%
(b)
Fig. 5. Uniformity across ring array, collapse mode: (a) Center frequency;
(b) Fractional bandwidth.
conventional, and 10 to 27.5 MHz for collapse. In conventional
mode, the device operates at 8.3 MHz with a 6-dB fractional
bandwidth of 70%, and in collapse, 19 MHz with a fractional
bandwidth of 69%. The SNR of a plane reflector at 5 mm is
29 dB for conventional and 35 dB for collapse.
B. Imaging Results
A conical volume was reconstructed offline using the full
64×64 set of A-scans from an imaging target with weightings
for full-aperture resolution [6] and cosine apodization. All
images are shown with 40 dB of dynamic range. The imaging
phantom is shown in Fig. 6, and consists of three steel wires,
each 0.3 mm in diameter. Fig. 7 shows the Y-Z and X-Z
planes of the conical volume (depicted in Fig. 6). Because of
the higher frequency and reduced acoustic crosstalk, collapse
mode produces images with a narrower main lobe and fewer
x
y
z
y-y
z
-2 0 2
2
4
6
8
10
(mm)
(mm)
d = 0.3 mm
Fig. 6. Phantom of three wires.
Fig. 9. Photograph of spring, 3-D rendered ultrasound of spring, cross
sections with 40 dB dynamic range.
artifacts than conventional mode imaging. The image SNR in
conventional is 50 dB compared to 22 dB for the SNR of the
wire A-scan. In collapse mode, the image SNR is 48 dB and
the A-scan SNR, 24 dB.
The experimental axial and lateral line spread functions
(LSF) of the array are shown in Fig. 8 alongside the simulated
results. The 6-dB axial and lateral resolutions for for the B-
scan of the wire target is 189 µm and 0.112 radians for 8 MHz,
and 78 µm and 0.051 radians for 19 MHz.
Finally, volume images of several targets were generated
and are shown in Figs. 9,10,11 to exhibit the quality of images
produced by this ring array.
Fig. 10. Photograph of Palmaz-Schatz stent, undeployed, 3-D rendered
ultrasound of stent, cross sections with 40 dB dynamic range.
Fig. 11. Photograph of Palmaz-Schatz stent, deployed, 3-D rendered
ultrasound of stent, cross sections with 40 dB dynamic range.
V. C ONCLUSION
We have demonstrated volumetric ultrasound imaging with
a forward-viewing CMUT ring array. Future work involves
developing a fully integrated system with flip-chip bonded
electronics for incorporation into a catheter probe. These
results show that CMUT ring arrays with custom front-end
integrated circuits can produce images with ample quality for
clinical use.
A
CKNOWLEDGMENT
This work was supported by the National Institutes of
Health. Thanks to Bill Broach and the Portable Power group
at National Semiconductor Corporation for supporting us with
assistance in circuit design and for providing the custom inte-
grated circuits. Thanks to Volcano Therapeutics for providing
the stents. David Yeh is supported by a National Defense
Science and Engineering Graduate Fellowship.
R
EFERENCES
[1] Y. Wang, D. Stephens, and M. O’Donnell, “Initial results from a forward-
viewing ring-annular ultrasound array for intravascular imaging,” in Proc.
IEEE Ultrason. Symp., vol. 1, Oct. 2003, pp. 212–215.
[2] U. Demirci, A. S. Ergun,
¨
O. Oralkan, M. Karaman, and B. T. Khuri-
Yakub, “Forward-viewing CMUT arrays for medical imaging.” IEEE
Trans. Ultrason., Ferroelect., Freq. Contr., vol. 51, no. 7, pp. 887–895,
July 2004.
[3] F. L. Degertekin, R. O. Guldiken, and M. Karaman, “Micromachined
capacitive transducer arrays for intravascular ultrasound,” in Proc. SPIE
MOEMS Display and Imaging Systems III, vol. 5721, no. 1, San Jose,
CA, 2005, pp. 104–114.
[4]
¨
O. Oralkan, A. S. Ergun, J. A. Johnson, U. Demirci, M. Karaman,
K. Kaviani, T. H. Lee, and B. T. Khuri-Yakub, “Capacitive micromachined
ultrasonic transducers: Next-generation arrays for acoustic imaging?”
IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 49, no. 11, pp.
1596–1610, Nov. 2002.
[5] B. Bayram, E. Hæggstr
¨
om, G. G. Yaralioglu, and B. T. Khuri-Yakub, “A
new regime for operating capacitive micromachined ultrasonic transduc-
ers,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 50, no. 9, pp.
1184–1190, Sept. 2003.
[6] S. J. Norton, “Annular array imaging with full-aperture resolution,” J.
Acoust. Soc. Am., vol. 92, no. 6, pp. 3202–3206, Dec. 1992.
Y (mm)
Z (mm)
−6 −4 −2 0 2 4 6
2
3
4
5
6
7
8
9
(a)
Y (mm)
Z (mm)
−6 −4 −2 0 2 4 6
2
3
4
5
6
7
8
9
(b)
X (mm)
Z (mm)
−6 −4 −2 0 2 4 6
2
3
4
5
6
7
8
9
(c)
X (mm)
Z (mm)
−6 −4 −2 0 2 4 6
2
3
4
5
6
7
8
9
(d)
Fig. 7. Slices of 3-D volume, 40 dB range: (a) Y-Z plane, conventional; (b) Y-Z plane, collapse mode; (c) X-Z plane, conventional; (d) X-Z plane, collapse
mode.
3 3.5 4 4.5
-40
-20
0
Axial Distance (mm)
N
o
r
m
a
l
i
z
e
d
M
a
g
n
i
t
u
d
e
(
d
B
)
Experiment
Simulation
(a)
3 3.5 4 4.5
-40
-20
0
Axial Distance (mm)
N
o
r
m
a
l
i
z
e
d
M
a
g
n
i
t
u
d
e
(
d
B
)
Experiment
Simulation
(b)
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
-40
-20
0
sin(
θ
)
N
o
r
m
a
l
i
z
e
d
M
a
g
n
i
t
u
d
e
(
d
B
)
Experiment
Simulation
(c)
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
-40
-20
0
sin(
θ
)
N
o
r
m
a
l
i
z
e
d
M
a
g
n
i
t
u
d
e
(
d
B
)
Experiment
Simulation
(d)
Fig. 8. Experimental versus simulated line spread functions: (a) Axial LSF, conventional; (b) Axial LSF, collapse mode; (c) Lateral LSF, conventional; (d)
Lateral LSF, collapse mode.
