Multi-Wave Imaging A New Paradigm in Ultrasound Imaging - SD

Introduction

Hi, I am Jack Ske.
I'll be talking about new technology related
to the measurement of stiffness on of tissue called transient elastography
that we have developed in a startup company in France.
And what I'll be talking about today relates
to the interaction of two ultrasound wave in tissue.
This is why my talk is entitled Multi Wave Imaging,
a New Paradigm in Ultrasound Imaging.

New Mode of Imaging

This new mode of imaging that we have created
leverages a brand new architectural advancement
for an ultrasound imaging system based on advancement in microprocessors
and what's done in the video game industry.
And with such advancement, we're capable
of acquiring data at ultra fast frame rate.
I'm talking about roughly
around 5,000 frames per second, if no more.
So that's a factor 10 greater than whatever is available
today in conventional ultrasound system.
We need this fast frame rate in order to provide assessment of tissue elasticity in real time
and in a quantifiable manner.
And this tissue elasticity is there
to help characterize palpable masses.
We've had, we have developed our product targeting a first clinical application being imaging
of breast diagnosis, enabling
the move from conventional anatomical morph morphological
images to towards more physio pathological content
in the image.

Conventional Ultrasound Image Formation

If one looks at conventional classical image formation today, an ultrasound image is typically built through the sequential acquisition
of a certain number of lines in in the human body,
roughly 128 lines.
And if one has to take into account the fact
that the ultrasound velocity is roughly
around 1,540 meters per second,
then when looking at the image at 10 centimeter depths,
this directly leads to an acquisition frame rate of roughly
25 to 30 frames per second.
The technology has moved from the mid seventies
of an analog domain into the mid eighties into the realm of digital beam forming.
But the point I wanna stress in here is
that the architecture
of every system since the mid eighties are no different than
the architectures of a system which existed previously.
With the exception that analog component have been replaced
by digital component providing some form of flexibility,

Breaking the Architectural Paradigm

what we have achieved
and what we've done is
is break this architectural paradigm, where now the data
that we acquire from tissue is not processed in hardware,
but entirely in software.
So the ultrasound system only contains one board,
which captures information, send that information
non-processed onto a high speed busts towards the multiprocessor of the shelf board, which
with software provides the total image that is
the outcome of processing data.
So, if one compares an ultrasound system of today with the ultrasound system that we building,
you can see on this diagram that we have replaced lots
of hardware, component hardware, PC boards
with nothing than software adding flexibility
and lots
of reliability in the way an
ultrasound system is being built.
And this is typically the outcome of
what the PC industry
and the video game industry is providing us.

Computational Capabilities

If one looks at an ultrasound image we have estimated that we need a computational load
of roughly 3.2 giga multiplication addition per second in
order to make an image, whether it's a 2D gray scale image
or a colorful image.
And today, we are demonstrating that we can achieve
more than eight giga multiplication addition per second
as far as computing capability.
Now, if one looks also at the other key element,
which is the way the image is transferred onto the micro
core processor, leveraging this high speed bus called the PCI express, we are capable
of transferring data towards a PC at rates of
around 20 gigabits per second.
Just in order of magnitude, this is equivalent
to one full video DVD transfer every two seconds.
So it's fairly fast.

What is Elastography?

So what is elastography?
If one looks at conventional ultrasound image, one image
with typically density of tissue and today there is been no systematic correlation which has been demonstrated between echogenicity and
and pathology palpation.
Manual palpation is a technique which has been known
since thousands of years.
People have discovered even in ancient Egypt
Ely, representing the concept of palpation.
So the goal here is can we combine conventional ultrasound
assessing density with some form
of technology which can assess stiffness of tissue
and combining that onto the same ultrasound system?
Well, naturally the answer is yes, but it's yes
because the human body is helping greatly
in that endeavor.

Physics Behind Shear Waves

And this will be the only equation I will show here in that talk.
If one looks at the human body, at the frequencies
which are the conventional frequencies used in ultrasound
imaging, the body behaves like a fluid.
And the kind of ultrasound wave
that can be sent in the body are called compressional waves
or longitudinal waves.
And the speed at which those wave propagates is
around 1500 meters per second.
However, if one looks at the same body at a lower frequency,
the body behaves no more like a fluid,
but behaves like an elastic solid.
And in that environment,
besides longitudinal compressional wave,
there is another kind of wave that can be sent in the body.
Those are known as sheer waves.
Now, the sheer wave is interesting for two main reasons.
Number one, sheer wave speed is very slow in the body,
few meters per second compared to thousands
of meters per second in the case of a launching wave.
So that's first observation.
Second observation is that the sheer wave velocity
is directly proportional to the square root
of the elasticity of tissue,
which can be directly correlated
to the young's modulus of a tissue.
So the challenge in here is that number one, if one know
how to, if one knows how to launch a sheer wave in tissue
and then can assess its speed of propagation,
we directly have the elastic
coefficient of the tissue.
So this is a challenge in front of us in here,
and I'll tell you this challenge has been addressed
and how we've solved it.

Types of Elastography

Just one more common
before, there's been lots of various work being done in the field of elastography today.
There is a so-called phy called static elastography.
It's been developed by John here in the United States,
where a mechanical displacement is induced to the tissue,
typically by pushing the probe on the tissue to induce
that deformation.
There is also continuous vibration called
dynamic elastography.
But what we are talking about in here in our case is called
transient elastography.
The capability to look at the displacement of the sheer wave
as it propagates
before it bounces back on the various border of
of the tissue, whatever kind of elastography is used.
The methodology is always the same.
One has to induce a mechanical force to the tissue.
This can be done externally.
What static elastography does by pushing the transducer
on the organ and creating that deformation with
that external push,
or by sending a sheer wave inside the tissue, which is
what we've done, and creating an internal force.
Step number two, one has to be able
to measure the displacement
of a tissue which is being induced that way,
this can be done in two ways with ultrasound
or with MRI naturally will follow the ultrasound path.
And step number three, one has then
to estimate the tissue stiffness in the case of
the external deformation,
but estimation is purely qualitative.
We measure strain simply
because we do not know the amount
of stress which has been applied.
This is directly related to how the user pushes
with the probe on the tissue,
or we can estimate tissue stiffness in a
quantitative fashion.
And so this is what we are proposing in our approach of assessing elasticity of tissue.

Process in Supersonic Imaging Mode

So the whole process for looking at elastography in our supersonic imaging mode is very simple.
Step number one, one has to generate a sheer wave in tissue
by sending a long pulse in the tissue, similar to
a color flow pulse, a doppler pulse,
which will create then a sheer displacement
of a tissue of a few micron.
And that sheer displacement will trigger a generation
of the sheer wave perpendicular to the displacement
that has been created.
So step number two, as the sheer wave is created
and is propagating,
as is being shown here in this picture,
the transducer is automatically put into the mode
of sending a compressional wave in the tissue, a flat beam.
And with the interaction of that flat beam,
with the sheer wave propagating, we are capable
of capturing at different instant in time
the behavior of the sheer wave as it propagates
and extract out
of this the velocity at which the sheer wave propagates.
So now we've done step one, generate a sheer wave.
Step two, assess the sheer wave velocity, which means
that we now have directly the elasticity
of the medium in which the sheer wave propagates.

Phantom Study

So we've run a certain number of tests in here which is shown in here in this phantom study that I showing here.
So we've built a phantom with a gel.
And inside that phantom we have inserted an inclusion.
The property of that inclusion is twofold.
It has one the same density
as the surrounding medium of the phantom.
This is why on the picture you see the same gray scale,
the same speckle rendition, but two, it's stiffer.
So if one looks at the movie
which is generated in here, you see from that movie
that the sheer wave propagates.
And as it encounters the lesions that we have created in the phantom, you see
that the sheer wave propagates much faster
in the lesion than in the surrounding tissue.
So out of this
and the estimation of a sheer wave velocity
through the whole image, we are capable of
building an image, which is shown on the right slide,
superimposed over the original gray scale image
of a phantom, which shows the inclusion
coated in color and in some bright red fashion in here
and the rest of the phantom being greenish or
or yellowish.
But all of that we can quantify it.
The red here version is roughly five kilo pascal,
while the surrounding greenish tissue is around two
and a half kilo pascal.
So therefore assessing in here a ratio of two in stiffness
between the lesion and the surrounding medium.
So that's one way to look at transient elastography.

Comparison with Static Elastography

Now, if one compares transient elastography with static
as it's shown is this next slide, you will see number one,
that the scale of static elastography is here,
purely qualitative.
Blue means in French dur, which is hard,
and red is soft,
but absolutely with no numbers related to
how hard it is compared to the soft portion of a tissue.
Well, in our case we know that the blue is roughly
around four kilo pascal while the red is
below one kilo pascal.
You see differences between the two method.
In the case of static elastography, there are inherent
artifact as shown in here
and reported in the literature,
which we do not have in the field of transient elastography.
And number two, the uniformity lack
of variance in the lesion artificial lesion
that we are able to express
with a transient elastography approach as compared
to the static elastography.
So that's point number one.

Quantification

Now, point number two, we claim qualification.
So what we've done is that we've acquired a tissue phantom which had been calibrated in two zones.
A tissue fountain coming from Ernie Matson.
One zone was rated at between five
and seven kilo pascal stiffness.
And the other zone here, as you can see on this picture,
between 10 and 13 kilo pascal,
drawing a region of interest in the less stiffer zone
with our elastography method, we estimate the mean value
of stiffness here between five and six kilo pascal.
And the other zone we measure 10
and a half kilo pascal, so well within the boundary stated by the manufacturer of the phantom.
So our quantification process makes sense.
This in hand. We decided to go to a clinical evaluation at the S two QE in Paris.
At the date of that report,
we only had 50 patient at Cur Purina.
We have more than 220 patients to date.
The population,
which was recruited is a population of women.
We were gonna be sent to biopsies anyway.
We looked at lesions of less than two centimeters,
and the population was a mix
between birad three, four, and five.
In order to acquire that information, we used an old A TL Phillips product, VHDI 1000,
with a conventional Phillips probe of L seven four.
Now I don't want you to look at the quality
of a 2D gray scale image.
What is important in that experimentation is the data
that we have acquired from an elasticity standpoint.

Clinical Results Examples

So here is the first set of data.
The two images that are shown in here show two
patients, one with a palpable lesions of 1.5 centimeter,
which has been rated by rat five, so undoubtedly malignant,
and the other one, an un palpable lesion here
by rat five of eight millimeter.
I'm now showing the outcome
of transient elastography showing the zone of stiffness in here, in in red.
In both cases and biopsy result indicate on the left that we have an invasive ductal carcinoma of grade three,
and on the right, an invasive lobular carcinoma.
So, interesting results, other results obtained
during the course of this trial are being shown in here,
which are interesting.
Also, the first image shows here a Birad four,
suspicious lesion of two centimeter.
Elasticity didn't show any changes in stiffness.
The image elastic image is quite uniform in here.
The outcome of the biopsy in that
situation was benign hematoma.
The case in the center is very striking,
a lesion undoubtedly the physician didn't know
what kind of lesion it was,
and it was not very clear that this also could be a
cyst.
So the lesion was rated by at three, it was
15 millimeters in size.
And when we launched the shear wave, we discovered
that the shear wave did not propagate in the lesion.
You see color everywhere, except in the lesion
where there is absolutely no color.
In physics, sheer wave did not propagate in fluid.
So, which led immediately to our observation
that this would be fluid filled
and the outcome of a biopsy was a viscous cyst.
The next image in here simply describe kind
of the opposite assessment, a small lesion birad three,
a centimeter in size, suspicion of a cyst.
When we started to use the sheer wave, we saw
that the sheer wave did propagate in the lesion,
therefore ruling out the fluid field capacity of
that lesion and biopsy indicated carcinoma in that situation.

Image Quality and Real-Time 3D

So this next slide in here describes now the quality
of the conventional 2D gray scale image that we are
obtaining on this new product, which acquires this data at multiples of thousands
of frames per second.
Just wanted to point out at the number one,
both the excellent contrast resolution,
but also at the excellent spatial resolution
and a very powerful treatment of speckle in in
the image in here inherent
to the way the data was acquired.
The next movie here, the a VI file, shows the concept
of real time graphy leading to
real time 3D of elasticity imaging.
This is a phantom data.
The transducer is manually moved on top of that phantom,
and you see the lesion appear,
the big cross-section diameter of the lesion
and the lesion disappearing afterwards.
So this is simply a 3D acquisition of elasticity
in that breast phantom leading to a real time 3D rendition.

Conclusion

So in conclusion, I just would like to say that the outcome
of what is presented, what I presented, is part
of a huge development effort.
The technology platform benefits roughly from 50 man
year effort, from academic research with a huge significant IP portfolio.
The technology leveraged the shift from hardware
to software as
we see in every high tech development today.
And the challenge confronting us are the extension of
that technology to other application like the liver,
the kidney, the prostate, and the cardiovascular.
Thank you for your attention.