See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277263800

Real-Time Embedded Digital Signal Processing for Ultrasound Imaging

Conference Paper · April 2011

CITATIONS READS

0 106

4 authors:

Meena M. Makary Asaad M. Anam

Harvard University Cairo University
21 PUBLICATIONS   46 CITATIONS    4 PUBLICATIONS   6 CITATIONS   

SEE PROFILE SEE PROFILE

Hani M. Bu-Omer Yasser Kadah

Cairo University King Abdulaziz University
5 PUBLICATIONS   5 CITATIONS    227 PUBLICATIONS   2,233 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Mentoring and guidelines for Early Career Researchers in the natural sciences View project

FINITE ELEMENT MODELS FOR COMPUTER SIMULATION OF INTRASTROMAL PHOTOREFRACTIVE KERATECTOMY View project

All content following this page was uploaded by Meena M. Makary on 27 May 2015.

The user has requested enhancement of the downloaded file.

 28th NATIONAL RADIO SCIENCE CONFERENCE
(NRSC 2011)
April 26‐28, 2011, National Telecommunication Institute, Egypt

Real-Time Embedded Digital Signal Processing for Ultrasound Imaging

Meena M. Beshara1, Asaad M. Anam2, Hani M. Bu-Omer3, and Yasser M. Kadah4
1
Faculty of Engineering, Cairo University, Giza, Egypt, meena.magharious@hotmail.com
2
Faculty of Engineering, Cairo University, Giza, Egypt, eng.asaadanam@yahoo.com
3
Faculty of Engineering, Cairo University, Giza, Egypt, hani2011ye@yahoo.com
4
Faculty of Engineering, Cairo University, Giza, Egypt, ymk@k-space.org

ABSTRACT
Digital ultrasound diagnostic devices have gained broad acceptance in a wide range of diagnostic fields
because they are able to make observations inside the body safely and in real-time. For a real-time acquisition of
ultrasound signal, the data rate entering the digital beam-former may reach up few Gigabytes. Therefore, it is
desired to somehow reduce such amount of data without compromising the performance of the system. Hilbert
transform can be used to derive the analytical signal from its real part, which can be downsampled to its
bandwidth without aliasing while keeping the phase information intact.
Distinction between different digital ultrasound imaging systems is determined almost exclusively by their
software component, with the presence of the growing availability of high-end integrated analog front-end
circuits. Efficient implementations of digital ultrasound systems rely on embedded digital signal processing on
FPGA with data conversion from oversampled 1-bit delta-sigma analog-to-digital converters (ADC) to minimize
the number of lines going into the FPGA. It is not a trivial task to handle the huge amounts of data collected in the
real-time acquisition of ultrasound system. Therefore, it is desired to somehow reduce such data without
compromising the performance of the system.
The aim of our work is to build high efficient software to implement a digital signal processing algorithms for
the reduction of the processed data size using one-sided Hilbert transform technique using a low cost Xilinx
Spartan 3E FPGA (Field Programmable Gate Array) chip.
In this paragraph, we describe the method used to reduce the required raw ultrasound data transfer bandwidth
while maintaining the phase information. The processing steps are shown in Fig. 1 for single channel data (with
other channels using the same architecture). Assuming that the sampling rate of each channel is N Sa/s, the
acquired samples represent an oversampled version of the real part of the signal. Using a discrete FIR Hilbert
transform filter, the analytic signal can be computed with only the positive side of the original signal spectrum.
Hence, such analytic signal can be downsampled to only the bandwidth of the analytic signal, which is much
smaller than that of the real signal. In the Hilbert transformation exact implementation, it acts like an ideal filter
that removes all the negative frequencies and leaves all positive frequencies untouched. This ideal implementation
requires high computational complexity (or FPGA resources in embedded implementations). Therefore, a number
of authors suggested the use of digital finite impulse response (FIR) filter approximations to implement the
Hilbert transformation. The computational complexity of working on the original high sampling rate is still not
practical for direct implementation. Therefore, we can significantly reduce this complexity by combining the
downsampling and filter implementation together. So, we designed the Hilbert FIR filters as multirate filters in
such a way to do both the filtration and downsampling at the same time. The analytic signal consists of the
original real signal decimated by the factor of choice and the output of the multirate Hilbert filter that output a
decimated version of the imaginary part of the
signal. The Hilbert filter is implemented in the n-channel
Input
common optimized form whereby the zero tap
coefficients are not computed and therefore an
order L filter uses only L/2 multiplications. Pre-amplifier
By implementing these techniques, we can TGC

minimize the total processing time and enhance the Output

system performance. Anti-aliasing
ADC Hilbert Band-pass Down-sampling
Filter Filter Filter

Keywords: Utrasound; Real-Time DSP; FPGA; Figure1

 28th NATIONAL RADIO SCIENCE CONFERENCE
(NRSC 2011)
April 26‐28, 2011, National Telecommunication Institute, Egypt

REFERENCES

[1] A. M. Hendy et al., “PC-Based Modular Digital Ultrasound Imaging System”, Proc. IEEE Int’l Ultrason.
Symp., pp. 1330-33, 2009.
[2] S. Sukittanon et al., “FIR Filtering in PSoC TM with Application to Fast Hilbert Transform”, Application
Note2328, Cypres Semiconductor Corp., 2005.

View publication stats