March 25, 2005 —
Researchers at the University of Southern California's Viterbi
School of Engineering have successfully demonstrated a novel
“High-resolution Ultrasonic Transmission Tomography” (HUTT) system
fthat offers 3D images of soft tissue that are superior to
those produced by existing commercial X-ray, ultrasound or MRI units.
Vasilis Marmarelis, a professor of biomedical engineering at the
Viterbi School, presented HUTT images of animal organ tissue in San
Diego at the 28th International Acoustical Imaging Symposium on March 21st.
|
Vasilis Marmarelis and experimental HUTT apparatus |
According to Marmarelis, HUTT offers nearly order-of-magnitude improvement
in resolution of structures in soft tissue (i.e., 0.4 mm, compared to 2 mm for
the best alternatives). Several other features promise to make the technology a scientific and clinical
tool of
great power:
- Robust algorithmic tools enable HUTT to differentiate separate
types of tissue
based on their distinctive “frequency-dependent attenuation” profiles,
that should allow clinicians to distinguish malignant lesions from
benign growths in a
non-invasive and highly reliable manner.
- In addition to improved resolution, the system can locate tissue
features with extreme precision in a objective, fixed-coordinate 3D
grid, crucial for guiding
surgical procedures.
- Scans can be performed in a matter of a few minutes and because they are ultrasonic,
they
do not use potentially harmful ionizing radiation.
- The system requires a minimum of special pre-scan procedures and appears likely,
in clinical use, to be more comfortable for patients than alternatives.
“"The HUTT imaging system is a novel and potentially very useful approach to
diagnostic ultrasound,” said Dr. Phillip W. Ralls, a professor and vice chair
in the USC Keck School of Medicine department of radiology. “The potential clinical
benefits of the superb images obtained by this completely safe, non-invasive technique
are very exciting."
HUTT image of kidney tissue. Click on image to see rotating view, |
According to Marmarelis, the key features distinguishing HUTT from all previous
ultrasound imaging systems is the use of multi-band analysis with sub-millimeter
ultrasonic
transducers in transmission mode, rather than the commonly used echo mode, to
create the 3-D image.
He explains that in traditional hand-held ultrasound systems, sound waves are
broadcast into the tissue, and the echoes produce an image of the reflecting interfaces
– that is, the sound transmitter
and the receiver are both on the same side of the sample.
However, only a tiny fraction of the transmitted sound comes back as echo on
soft tissues, while a much larger fraction (about 2000 times bigger) is transmitted
through the soft tissue.
Using the sound transmitted through tissue allows the formation of better images
with greater clarity and
resolution.
A hand-held apparatus cannot objectively locate objects in 3D space
(in a fixed-coordinate system), but only allows the user to
subjectively
observe where an object is in relation to other observable structures.
Therefore, it is operator-dependent.
The HUTT system transmits an extremely short ultrasonic pulse (about 250 nanosecond)
of 4-12 megahertz frequency (far above human hearing) and picks up the pulse on
the other side after it has traveled through the imaged object.
The transmitted pulses come from an array of very small ultrasonic transducers
of sub-millimeter dimensions. A parallel array of transducers on the other side
receives the pulses after they travel through the imaged tissue.
A sophisticated coding/decoding signal scheme recognizes a small “sweet
spot” of the signal coming from the opposite transducer, and only that transducer,
and ignores all other pulses transmitted by neighboring transducers.
The transducer is able to distinguish the right signal from the
right transducer by using coding that is almost identical to that used
by a cell phone to detect signals sent
to its number -- and its number only -- from the flood of
electronic signals
on the air at any given time.
When the transducer captures the signal, it is processed with advanced signal
processing algorithms, specially developed by Marmarelis’ group, to
form the multi-band images. Different kinds of tissue allow slightly
more, or slightly less of the pulse through -- the loss is called
“attenuation,” and varies according to the type of tissue, and the
frequency of the pulse.
"Typically the resulting images represent minute variations in relative
attenuation over various frequency bands and they define the different sections
of the tissue in the image,” said Marmarelis.
The two arrays, transmitter and receiver, are mounted on opposite sides of a
drum that spins as it rises around the object (which is suspended in water), creating
a stack of tomographic image slices which visualization algorithms turn into 3D
images.
In the first set of experiments using the HUTT system, the Marmarelis team easily
located a set of small metal balls smaller than a millimeter in diameter embedded
in agar medium.
Many critical refinements occurred during the five-year process of development,
as the team gained proficiency in
imaging animal tissue, notably sheep kidneys and bovine liver.
The most critical feature of the HUTT imaging technology is its
potential to reliably differentiate types of tissue based on their
multi-band
signatures caused by their varying attenuation patterns. This promises
to allow non-invasive
detection of lesions in clinical diagnosis, which represents the “holy
grail”
of medical imaging.
The team found it possible to identify various anatomical
structures within the kidney based on their distinctive attenuation characteristics,
so that computerized algorithms could display in color-coded fashion one tissue
in red, another in green, and so forth – thus assisting visualization in 3D.
The technology could also be used to isolate one type of tissue , allowing, for
example,
all the blood vessel structures to be displayed alone and studied.
|
Choosing a target. Different tissues in the same organ (here, a sheep kidney) vary in "attenuation"
characteristics: how much ultrasound of various frequencies they let through.
Algorithms
that incorporate empirical findiings of these differences can quickly discriminate
between tissue types in scans, and even digitally apply "dyes" to the captured
image to make a given type of tissue stand out. From left to right, 1: overall
view. 2. capsule (blue) 3. blood vessels (red) 4. papillary ducts, (magenta)
5. calyces (green). |
“Preliminary results on a sheep kidney show exquisite anatomic and tissue detail,”
commented radiologist Ralls.
Working with Marmarelis on the project are post-doctoral researchers Drs. Dae
C. Shin, Jeong-Won Jeong, Changzheng Huang, and Syn-Ho Do.
Marmarelis is co-director of the Biomedical Simulations
Resource (BMSR), an NIH-funded center for the advancement research in biomedical
modeling; and also holds an appointment in the Viterbi School’s Electrical Engineering
Department.
Marmarelis' work was funded by the Alfred E. Mann Institute for Biomedical Engineering
at the
University of Southern California.
|
HUTT Team: (from left) ChangZheng "Kevin" Huang, Synho Do, Marmarelis, Dae C. "Roy" Shin,
Jeong Jeong-Won |