IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control , vol. 55, no.

9, September 2008 2109

Correspondence
Design and Implementation of High Frequency this approach, the Doppler processing becomes program-
Ultrasound Pulsed-Wave Doppler Using FPGA mable and can be easily integrated with the beamforming
function in ultrasound imaging. Furthermore, this design
Chang-hong Hu, Qifa Zhou, Member, IEEE, and shows an alternative way for implementing the ultrasound
K. Kirk Shung, Fellow, IEEE pulsed-wave Doppler function digitally.

Abstract—The development of a field-programmable gate

array (FPGA)-based pulsed-wave Doppler processing ap- II. FPGA-Based Pulsed-Wave Doppler Design
proach in pure digital domain is reported in this paper. Af-
ter the ultrasound signals are digitized, directional Doppler
frequency shifts are obtained with a digital-down converter The diagram of the proposed pulsed-wave Doppler sys-
followed by a low-pass filter. A Doppler spectrum is then cal- tem is shown in Fig. 1. The single element transducer used
culated using the complex fast Fourier transform core inside in this study is driven by a 10-cycle burst sinusoid signal
the FPGA. In this approach, a pulsed-wave Doppler imple- with a Vpeak-to-peak of 30V. The returned echoes are
mentation core with reconfigurable and real-time processing
amplified with a variable gain amplifier (AD8332, Analog
capability is achieved.
Devices Corp., Norwood, MA) and then sampled with an
ADC (AD9433, Analog Devices Corp.) at the sampling
frequency of 120 MHz. The sampled data are then fed
I. Introduction
directly to the FPGA (Virtex II Pro, Xilinx, Inc., San
Jose, CA). The reason for using the Virtex II Pro is that
H igh frequency ultrasound (>20 MHz) probes and B-
mode systems with a spatial resolution on the order of
a few microns have been widely used [1], [2]. Accompanied
a beamformer for a high frequency linear array at 30 MHz
was designed on this chip [8]. In FPGA, the digitized data
are first sent to the digital-down converter (DDC) block
with the development of the high frequency B-mode imag-
where the demodulated samples are obtained. The de-
ing system, the pulsed-wave Doppler (PD) designs have
modulated in-phase (I) and quadrature (Q) data then go
been proposed and implemented [3]–[6]. High frequency
to the range gate block which determines the location
ultrasound (>20 MHz) with improved spatial resolution
where the Doppler frequency shift should be detected.
with Doppler capabilities can be used to detect lower ve-
This range gate block can set the window in the axial
locities and smaller vessels. Pulsed-wave Doppler systems
direction in 1–2048 samples or any location less than 13
at higher frequencies are generally implemented by either
mm. Finally, I/Q signals with Doppler directional infor-
using analog circuitry such as a mixer to implement de-
mation are used to generate the Doppler spectrum.
modulation and a sample-and-hold (SH) to acquire Dop-
pler shift signals [3]–[5] or, in digital domain, to estimate
the velocity directly after the analog-digital converters
III. Digital-Down Converter
(ADCs) without using mixers and SHs [6]. Although the
work reported in [6] obtained the blood velocity after the
Generally, analog demodulators (mixers) are used to
echo signals were digitized, it emphasized exclusively the
generate the in-phase/quadrature signals. A digital quadra-
algorithm for blood velocity estimation.
ture demodulation method based on direct IF (intermedi-
Since the digital beamformers for high frequency ultra-
ate frequency) sampling has advantages over conventional
sound arrays have been implemented in a field program-
analog quadrature demodulation methods in match ac-
mable gate array (FPGA) chip [7], [8], it is worthwhile
curacy between in-phase and quadrature channels. The
to also implement pulsed-wave Doppler processing func-
other advantages of using DDC are digital stability, con-
tionalities on the same chip to facilitate duplex scanning.
trollability, and small size. However, the main limitation
Therefore, this study investigates the design and imple-
of DDC is that the digital demodulation effect is mainly
mentation of the pulsed-wave Doppler function using a
determined by the ADC dynamic range when the signal’s
FPGA. The motivation of this work is to eliminate any
amplitude varies greatly. In this study, a 12-bit ADC was
need of using analog circuits (mixers and SHs) by process-
used with an effective dynamic range of 70 dB. To achieve
ing Doppler spectrum using a single FPGA chip. Using
higher dynamic range, the high speed ADCs with 14 or 16
bits, if available, should be used.
Manuscript received December 17, 2007; accepted March 17, 2008. The DDC block has a configurable data path compris-
This work was supported by NIH grant # P41-EB2182. ing a digital-down synthesizer (DDS), a digital mixer, a
The authors are with the Biomedical Engineering Department and the filter, and a series cascade of 2 optional polyphase deci-
NIH Transducer Resource Center, University of Southern California, Los
Angeles, CA 90089 (e-mail: [email protected]). mators. The DDS generates the digitized sine and cosine
Digital Object Identifier 10.1109/TUFFC.904 signals required to mix with the digitized echo signals

0885–3010/$25.00 © 2008 IEEE

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 05,2021 at 16:33:09 UTC from IEEE Xplore. Restrictions apply.
2110 IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control , vol. 55, no. 9, September 2008

Fig. 1. Diagram of the FPGA-based pulsed-wave Doppler system.

Fig. 3. Results from digital-down conversion block. (a) raw echo train
from the Field II simulation; (b) modulated signals with I and Q compo-
nents from the FPGA; (c) FPGA-demodulated data.

ter. Finally, the Doppler spectrum is obtained after I/Q

output from cFFT are squared and summed. The entire
spectrogram calculation module is shown in Fig. 2. The
implementation results showed that this design uses 29%
of total slices, 22% block rams, and 28% multiplexers in
the FPGA.

V. Simulation and Experimental Results

Fig. 2. Block diagram of the spectrogram calculation. The performance of the DDC was tested using the
FPGA DSP simulation software (Xilinx, Inc.) and Field II
giving both sum and difference frequencies. The filter is a software [9]. The raw ultrasound data were first generated
low-pass filter with filter taps and configurable coefficients using Field II with a transducer at the frequency of 30
used to realize the sample rate changes. This quadrature MHz (50% bandwidth). In this simulation, a phantom was
demodulator yields the outputs of I and Q signals, which used for calculating RF data as measured from the femo-
are used to obtain directional flow infomation. ral artery in the upper leg. The 250 samples which contain
the echoes were selected to form an echo train and were
then fed into the DDC module. Fig. 3 presents the results
IV. Spectrum Calculation of the digital signal demodulated from the DDC. Fig. 3(a)
shows the raw echo train from the Field II simulation, Fig.
The spectrogram block is used to process demodulated 3(b) shows the modulated signals with I and Q compo-
data. In this block, the Doppler frequency shift data are nents from the FPGA. The FPGA-demodulated data are
selected, weighted, and processed using the complex fast presented in Fig. 3(c). Fig. 4 shows the spectrograms from
Fourier transform (cFFT) core of the FPGA. The win- the numerical simulation and the FPGA data. Fig. 4(a) is
dowed data are also weighted by means of a Hamming the result using Matlab (MathWorks, Inc., Natick, MA),
function to improve the signal-to-noise ratio (SNR) for and Fig. 4(b) is the result from the FPGA design. Both
the spectrogram display before they are used in the cFFT. spectrograms are displayed at a dynamic range of 40 dB.
In this block, 2 dual-port block random access memories The mean squared difference between the 2 spectrograms
(BRAMs) store the Doppler frequency shifts. A counter is 0.42.
is used to generate the reading address for both BRAMs. Measurements of blood velocity from an artery in a
The windowed I/Q signals are read from memory to be human hand were carried out with the implemented high
processed using cFFT. The overlap window between 2 frequency pulsed-wave Doppler function. An angled PMN-
consecutive processing is also controlled by another coun- PT (HC Materials Corp., Urbana, IL) single crystal needle

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 05,2021 at 16:33:09 UTC from IEEE Xplore. Restrictions apply.
Hu et al. : fpga design of high frequency ultrasound pulsed-wave doppler 2111

transducer was used. This angled transducer has a center

frequency of 30 MHz with a 50% bandwidth. The Doppler
angle is 45°. For the measurement, the pulse repetition
rate generated from FPGA is 20 kHz. Fig. 5 shows the
spectrogram of the Doppler signal from the artery. A max-
imum axial flow velocity of about 19.5 cm/s is measured.

VI. Conclusions

A pulsed-wave Doppler function for high frequency ul-

trasound was designed and implemented based on a FPGA
chip. Since the FPGA technique has been adopted in med-
ical ultrasound especially for high frequency ultrasound,
this design can operate in duplex mode by integrating
pulsed-wave function in the same chip with a beamformer
in future work.

References

[1] F. S. Foster, C. J. Pavlin, G. R. Lockwood, L. K. Ryan, K. A. Ha-

rasiewicz, L. Berube, and A. M. Rauth, “Principles and application
of ultrasound backscatter microscopy,” IEEE Trans. Ultrason., Fer-
roelect., Freq. Contr., vol. 40, pp. 608–616, 1993.
[2] T. A. Ritter, T. R. Shrout, R. Tutwiler, and K. K. Shung, “A 30-
Fig. 4. Spectrograms. Upper: simulation result from Matlab; lower: simu- MHz piezo-composite ultrasound array for medical imaging applica-
lation result from the FPGA. tions,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 49, pp.
217–230, 2002.
[3] D. A. Christopher, P. N. Burns, B. G. Starkoski, and F. S. Foster,
“A high-frequency pulsed-wave Doppler ultrasound system for the
detection and imaging of blood flow in the microcirculation,” Ultra-
sound Med. Biol., vol. 23, pp. 997–1015, 1997.
[4] M. Vogt, “Direct sampling and baseband conversion in Doppler
systems for high-frequency ultrasound blood flow measurements,”
Electron. Lett., vol. 41, no. 14, pp. 789–790, July 2005.
[5] Q. Zhou, X. Xu, E. J. Gottlieb, L. Sun, J. M. Cannata, H. Ameri,
M. S. Humayun, P. Han, and K. K. Shung, “PMN-PT single crystal,
high-frequency ultrasonic needle transducers for pulsed-wave Dop-
pler application,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr.,
vol. 54, no. 3, pp. 668–675, 2007.
[6] K. W. Ferrara, B. G. Zagar, J. B. Sokil-Melgar, R. H. Silverman, and
I. M. Aslanidis, “Estimation of blood velocity with high frequency
ultrasound,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol.
43, no. 1, pp. 149–157, 1996.
[7] C.-H. Hu, K. A. Snook, P.-J. Cao, and K. K. Shung, “High-frequency
ultrasound annular array imaging. Part II: Digital beamformer de-
sign and imaging,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr.,
vol. 53, no. 2, pp. 309–316, 2006.
[8] C.-H. Hu, X.-C. Xu, J. M. Cannata, J. T. Yen, and K. K. Shung,
“Development of a real-time high-frequency ultrasound digital
beamformer for high-frequency linear array transducers,” IEEE
Trans. Ultrason., Ferroelect., Freq. Contr., vol. 53, no. 2, pp. 317–
323, 2006.
[9] J. A. Jensen, Estimation of Blood Velocities Using Ultrasound, A
Signal Processing Approach. New York: Cambridge University Press,
Fig. 5. Spectrogram of an artery in a human hand. 1996.

Authorized licensed use limited to: Bursa Uludag Universitesi. Downloaded on September 05,2021 at 16:33:09 UTC from IEEE Xplore. Restrictions apply.