Volume Ultrasound Core Principles for Clinical Practice - SD

Introduction to 3D/4D Ultrasound Principles

Hello, I'm Dr. George Bega.

My lecturer will cover principles of volume imaging for clinical practice.

There will be around 40, 45 minutes of a lecture covering the main principles of 3D ultrasound, in obstetrics and gynecology.

Hello, this is a presentation regarding the volume ultrasound principles.

In this presentation, we're gonna go over a, an overview of the main principles of the use of 3D four D ultrasound techniques, mainly in obstetrics and gynecology.

I would like to first recognize the assistance and the expertise of some of my colleagues from the industry, Ms. Amy Lex, Patty Hartman, Dennis Wisher, and Richard begging for providing advice in some of the materials that I'm gonna share with you.

Background on Ultrasound Technologies

Everybody has been, familiar with the fact that 3D ultrasound, is a new, relatively new technology.

We've had 2D ultrasound for years now.

In the eighties, color and power doppler, as well as pulse doppler were, were added to our normal, clinical tools.

Then recently, 3D four D ultrasound have been increasingly, used and, and, contemplated to be used in clinical practice.

But the fact is there is really no clear understanding of this technology.

There is no clear, there is no clear, picture here as to why we would need this new technology.

Limitations of 2D Ultrasound

What do we normally do in to D with 2D ultrasound?

Well, typically we get the normal standard planes.

We're all told to get this standard planes.

The standard planes are planes that have been over the years found to be accessible.

We can have access to these planes by 2D ultrasound as the 2D beam can, cross-sectionally cut parts of, of the anatomy we can obtain these planes.

So, again, the, the so-called 2D standard planes are planes that 2D ultrasound is good at.

It has been seen that over the years we are, most of the time capable of obtaining them, and no doubt about it, we've been doing a great job using 2D ultrasound for clinical, reasons.

Why Use 3D Ultrasound?

The, so the next question would be, well, why 3D?

What, why would be the need that we would actually have to contemplate first learning how to use this technology and then, obviously purchasing this new equipment?

My personal experience over the years has been that, there is a, a definite lack of understanding and a skepticism into its use.

Both sonographers and physicians feel that this is a new toy that, they're not really clear at times as to what would be their true role clinically.

My own experience has taught me with this technology, this, that, we really cannot expect this technology to spit out a diagnosis for us.

We have to know what do we need, what are the clinical scenarios where 2D ultrasound simply cannot solve our questions.

There is a number of, of, of instances where 3D ultrasound can help.

So basically, I have boiled it down into two main reasons why 3D for the ultrasound helps in clinical practice.

Number one, reason why 3D ultrasound help is this one, it provides for us image planes that are simply impossible or very difficult.

With 2D ultrasound, this is the real big one.

This is the big reason why we need to use 3D ultrasound.

We, in clinical practice have those standard planes with 2D ultrasound, but there are certain planes that are simply not possible with 2D ultrasound.

A a classic example would be the coronal plane of the uterus.

Also, the standard planes, the ones we normally get with 3D ultrasound may be very difficult, at times, depending on, on fetal position.

And 3D ultrasound can give us access to those planes.

The second reason why we need to do a 3D ultrasound in clinical practice is it may provide for us depth.

Depth is something that we simply cannot obtain with 2D ultrasound.

The reason for this is very simple, 2D ultrasound scans with a thin 2D beam.

So here is an, an overview, a diagram that basically goes over this main concepts 3D ultrasound or a volume ultrasound.

And by the way, in this presentation, I'll be use using both of these terms interchangeably, either 3D four D or volume ultrasound, and we're gonna go in detail a bit later on.

This can provide us impossible planes or difficult planes with depth.

Now, these planes can be the conventional planes.

The planes that we are typically doing right now, were obtaining right now with 2D ultrasound, but may be difficult because of the fetal position or can be unconventional planes.

So a con unconventional plane would be a plane that pretty much we're not taught to get because it was not possible or very difficult to obtain.

A typical unconventional plane would be a midsagittal plane off the fetal head to obtain corpus callosum, septa lucida and so forth.

So this would be a typical plane that is possible otherwise, but very difficult to get.

Another unconventional plane is the secondary pallet, plane.

Secondary pallet, typically is very difficult and it needs depth.

It needs depth that only a volume ultrasound acquisition can provide.

So here is a diagram just to go over, the, the capabilities of volume imaging to obtain image planes that are impossible with 2D ultrasound.

So here we have a 2D ultrasound probe that is trying to acquire the fetal spine.

Let's say my options typically with 2D ultrasound or be only two, I would either get a such or I'd get a transverse if I would have a need to obtain a coronal plane.

That is simply impossible.

A coronal plane is not possible because of a very simple reason, and this is a fundamental concept in 3D ultrasound.

The coronal plane is perpendicular to the axis of the beam.

No matter how I twist, I turn rotate angulate.

This probe, I can only get transverse and sag.

I can never really obtain this.

It is perpendicular to the axis of the ultrasound beam.

So how is the ultrasound being obtained? Well, simple.

Basically, a transducer element will send a, a pulse down, and it comes back.

Obviously in the Doppler, we, it'll pick up the frequency shift, but as you can see, the beam is emitted from the, the, the transducer elements down vertically and up.

So here is a diagram of, this is a, an animation of a probe, as you can see here, that the transducer elements will send an, an ultrasound in the periphery, and of course, it receives it back, and it measures the roundtrip time.

Now at the end of a, of a 2D, of a 2D beam, basically what happens is we, have a number of scan line produced.

A a number of scan lines are going to make up a frame.

A frame is pretty much what we see in the ultrasound screen.

Now with 3D ultrasound, the transducer doesn't stop there.

It keeps acquiring these frames at a predetermined speed, at a predetermined angle of acquisition.

This is really the foundation of how you'd grab a volume.

Now as we grab a volume, then the next, question would become, what do we do?

And typically we have access to image planes.

One of the simple examples would be I grab a volume of a uterus, and then I obtain either sage, transverse, and coronal.

Now, the coronals are very interesting.

Coronal planes are planes that are impossible with to the ultrasound and in the uterus, considering that it is verted and inflexed, we do not have one simple coronal plane.

We have more than one coronal plane, basically a coronal plane in this case of the fundus, and another one, at the level of the lower trans segment, and cervix.

And this is really the end result.

This is another diagram showing, nicely the, coronal plane of a uterus.

So this is really the impossible plane. Okay.

A huge confusion over the years has been over a simple fact.

The fact is that, 3D for d ultrasound per se, are marketing terms.

And when people are using these technologies beginners, there is a confusion there as to what they represent.

Basically, what they represent is volume acquisition.

3D for d ultrasound in clinical practice really becomes an issue of acquiring a volume.

If we acquire one single volume, we typically would call that a 3D acquisition of one single volume.

Or if that volume is acquired in continuity, that would be called a so-called four D, four D acquisition.

Four D supposedly is the fourth dimension.

Since we will be able to pick up a motion of the fetal anatomy, typically faster acquisitions or four d acquisitions are needed for moving targets.

These data, upon the acquisition are stored digitally in a hard drive of a, of the machine.

The current machines are pretty much like modern workstations.

They have, huge hard drives where this information is stored either temporarily or permanently, and then, subsequently can be displayed in a monitor interactively, in a multiplanar or rendered view.

Workflow for 3D Ultrasound

So a, a a normal workflow when we use 3D ultrasound is pretty much this one.

We start with 2D ultrasound.

We go over the area of interest.

We typically, over the years know what are some of the best acquisition planes to acquire, a volume.

So we target and optimize our image plane to those standard planes.

Then we drop a volume box, which is the same, box, similar to a color or powered Doppler box.

We adjust width and height, and we adjust the depth.

So this is really the reason we call it a volume box.

It's got a depth. Then we acquire a volume, typically with a lowest speed of acquisition, possible.

Then this information, by default, most, manufacturers will display it as a multiplanar display.

We would explore this volume to see if there is any artifacts, inside this volume, as well as whether we capture the anatomical area we were after.

Then we may do a number of either qualitative or quantitative, post-processing steps.

A qualitative will be pretty much either displaying the surface anatomy or the inside anatomy, bony tissue or, or fluid field structures and quantitative analysis will pretty much, go, around the volume measurements.

Then later, some of these steps of post-process steps can be saved in a video sequence, especially if I want to explore this volume left and right.

I can save this, this, steps as an a VI loop, as an a VI video file and save it in the hard drive.

Then you just, go real time again in your normal 2D mode and start targeting the next area of your, clinical protocol.

That would be a normal, a normal, workflow, diagram.

Now important to understand here is that, there is a confusion here that, new users of 3D ultrasound expect to just find where that 3D four D button is on the machine.

And somehow the diagnosis might be in the screen, right after fact of the matter is this is, a lot of, a lot of information that would need to be acquired, and this information needs to be processed.

There are many, many pre and post steps in order to acquire a diagnostic clinical, clinically useful 3D ultrasound.

It has to do a lot with knowing the right optimization.

The way we do ultrasound acquisition, in a volume today is we need, we do it based 2D ultrasound.

So we need to optimize at maximum our 2D, 2D ultrasound, settings.

Typical, first question would be, am I in the right acquisition play?

Do I have enough, aperture, typically, lower apertures will improve lateral resolution depth will affect, our frame rate.

Persistence is important, especially if I'm, imaging the fetal heart persistence is a frame averaging algorithm, which pretty much will smooth out the image.

We may not want that when I image the, the fetal heart, but we may want some persistence, some smoothing, or frame averaging if we are, visualizing the fetal face, let's say.

Then important here to optimize are the harmonic, the, compound resolution imaging, the speckle reduction code excitation.

These are all very important and useful, pre-processing tools that improve, contrast resolution, but they may have an impact on the frame rate.

Also, they, spec reduction and compound resolution also may slow down your acquisition speed.

So there is always a trade off between using them or avoiding them.

PRF angle gain and TGC are also important, especially if you're using color and power doppler.

As, as, as these, settings are so important in, first, having the right sensitivity to get the vessels.

And number two, they will actually have a problem in a volume, because, especially angle, I may actually have the right angle for one vessel, but if I'm after a number of vessels, I would have to count their orientation in a volume.

So this actually will, require some thinking and some analysis before the acquisition.

Of course, focal zones are important because they do impact, the frame rate as well as a general rule.

Frame rate is an important variable that has to be studied.

Every ultrasound machine has, an indicator of the current frame rate you're scanning, and all of these factors, as well as others will impact it.

The higher the frame rate, typically, the higher the quality of this volume.

From Pixels to Voxels in 3D Imaging

At the fundamental level, A pixel is the smallest unit of a 2D ultrasound image.

We judge image quality based on the number of pixels.

Now, in ultrasound, we have been told that the higher the frequency, the higher the resolution.

But in 3D ultrasound, this image, this information is digitized, and it means that we need to really obtain good images with a higher number of pixels.

Now, pixels by their definition are flat imaging units.

So if a manufacturer gives me the option of putting a pivot point somewhere here in this, in this grid, and I can flip this fact of the matter, is this is a flat to the image, whatever image I got in ultrasound here.

If I flip here and I wanna see on the side or on top or on the back of this image, there is no information to be obtained.

Because this is a flat 2D slice, I only get information in this grid of pixels.

When I acquire a 3D ultrasound volume, I grab a number of 2D slices, typically in the order of 500 to a thousand slices, if not more.

And each individual pixel is converted by software into a voxel.

So a normal question would be, what's a voxel?

A voxel is a pixel in volume.

The pixel information is put into a, a volume matrix.

And in this volume, each one of these voxels, will represent a specific point in the anatomy.

In the background of this information, typically there is an X, y, Z coordinate system that will quantify, the position of every single one of these voxels.

This is very important for quantification purposes, because for the first time with 3D ultrasound, we can accurately quantify volumes.

A very important, implication here also is the fact that this voxels now they really float in space.

Now, I have a true volumetric imaging, object that I can manipulate.

I can put a pivot point at any single point of these voxels.

I can twist, I can turn these voxels, I can look on top of them, I can look, be below them.

I can shine, shine a light source.

Typically, most ultrasound systems have a, by default, a one single central, light source.

But, some of the newer ones are contemplating and adding additional light sources, not only in in image planes that are perpendicular to the ultrasound beam, which is what works right now.

As you can see, these, these V here are, more illuminated in the center, and there are different shades of gray in the periphery.

And I'll show a diagram of how that works.

So here, I can either have the baby rotate, which would be, the, A camera would be our eye right now.

So the object can be rotated or the object can be maintained stationary, and I can move the camera around.

And of course, light sources can be added, one, two, or, a number of them, again, at planes that don't necessarily have to be perpendicular to the beam.

And here is an example of, a couple of, of, of this, concept.

On the left, you see a field face that can move.

Basically, a stationary camera is kept and the object is being rotated.

Otherwise, you can have a, camera that is moving and the object is, can be maintained.

Stationary, in this case, camera is put inside a a uterine cavity dis standard with fluid, in, and this is work done in in university, of California at Stanford.

This is at Stanford Radiology.

This is an important, work because it shows that you can actually simulate a hysteroscopic procedure by putting a virtual camera in an ultrasound volume, of an organ, of a hollow organ, distended with fluid.

Of course, you can change the, the, the light sources.

This is, an interface by Siemens.

As you can see here, by a cursor, you can, go over the area of interest and you can illuminate parts of the anatomy that you feel you have, more interest on.

So as you can see, the light sources now are becoming, much more flexible to the user.

This is a fetal face at 22 weeks, and as you can see here, the light sources are being changed, from one angle to the other.

So this is pretty much, what a a 3D ultrasound set up would look like.

Basically, you have a camera, you have an object to be visualized, and you have a light source, and you have a number of software algorithms to highlight either surface features or inside features of this, of this object.

Terminology and Buzzwords in 3D Ultrasound

Terminology and buzzwords are one of the most confusing parts of this new technology.

This technology is being introduced by many different manufacturers.

They all, all develop it, in their own, at competition with each other.

And we have an issue here because sometimes for the same processes, they come up with different terminology, and of course, they keep throwing at us.

The industry keeps introducing new concepts, and keeping up with this, has become an issue, especially for, beginners.

But typically, the buzzwords, are divided, by the main processes of this technology.

I'm gonna go over the main ones.

I'm gonna divide them by acquisition, display, and post processing.

Acquisition Methods

In acquisition, we are talking here about the acquisition, methods.

We pretty much have three main ways.

There are some sub, divisions on this, but main three main ways to acquire volume R.

The first one is a freehand acquisition.

The second one is a mechanical, automated, acquisition.

And the third one is the matrix array.

This is what we are going to evaluate in detail in the next, couple of minutes.

Of course, you can acquire volume in the once or in continuity.

Now, some of the variations of this, of the acquisition of the volumes are stick stick stands for spatial temporal image correlation.

It is a dedicated package to acquire volumes of the fetal heart display.

Display Modes

In the display modes.

The most, used one is the multiplanar dis uh, display.

But in the meantime, there are these extended displays or, or, or, image displays that simulate a CT or Mr.

Are increasingly being used, in everyday practice.

But as a matter of, of fact, the multiplanar display and the render display are the most commonly used ones.

Post-Processing Techniques

Post-processing, reformatting is a fancy word for saying we are able to, reprocess the volume data and, visualize parts of the anatomy that we couldn't see otherwise.

Reformatting really means the process of exploring different image planes and visualizing them at different, with different settings.

We can have a number of post-processing tools surfaces to visualize the surface anatomy.

Maximum IP would be to look at the maximum echoes of volume.

Typically, the fetal spine, the minimum IP is for fluid field structures like the, the bladder or the stomach.

X-rays really a medium intensity projection, and it's a little bit less useful.

Inverse mode is an, is a mode that inverses all fluid field structures into an opaque, visualization.

Very useful. I'll share examples on that.

Electronic scalpel is a, is a editing tool where parts of the anatomy that were superimposed to the area of interest can be cut out.

And as I mentioned before, we can create small mini movies, small a VI files with the rotation of these files.

So this is a diagram just to show you the main modes.

In a, you have a normal, free hand acquisition of a volume.

In B, there is a mechanical acquisition of A volume.

In C is the matrix array, and d, e, and F are all matrix array, capabilities.

In D, you can appreciate the fact that a matrix array can focus in a 360 degree fashion.

For the first time, we, can focus in the elevational plane or otherwise along the depth, something that the mechanical, displays in B cannot do.

Now, the probes that we are mostly using now, nowadays in clinical practice are the, the B ones they mechanically swept, transducers, because this is really where the technology is at this point.

But the newer technology as the matrix array, which is c, can really allow us to get, a very fast acquisition of volumes, that is focused.

And, these probes have a very high number of 2D elements I'm gonna go over in detail on them.

And in ENF, you can see that one of the most useful, displays of this type of probes is the so-called biplanar.

They, these probes can focus, and can have very high resolution, in both planes, at real time, scanning.

Freehand Acquisition

Now, the free hand acquisition have been the, has been the first way to do 3D ultrasound.

Free hand acquisitions typically use a normal 2D probe, and the, the operator will, will have to physically move the transducer elements over the area of interest.

While this seems like a nice and economical and easy way to do 3D ultrasound, it is prone to artifacts.

The end result is pretty much this.

We would ideally expect any acquisition method to give us a nice, nice registration of these 2D slices.

But this process is really not easy and not as straightforward, especially if it has to be done with a free hand acquisition.

And that is the main reason that the industry, has been, coming up with new acquisition, methods.

So this is an, an ultrasound transducer with a position sensing device where the operator is moving over the area of interest and is acquiring a number of 2D slices.

This is a, a classic 2D probe.

And while it is a, an easy way to introduce GD ultrasound, it is prone to a number of artifacts, especially of, of, of moving targets.

Mechanical Acquisition

The mechanical transducer, the, the dedicated mechanical transducer is a transducer that is slightly bulkier, because it has to, house a small, a small mechanical device that would have to sweep the transducer elements at an operator.

Selected speed at an operator, selected speed of, operator selected angle of acquisition.

This is the way we do 3D ultrasound today.

This is the most commonly used acquisition method to do 3D ultrasound today.

And this is how pretty much it works.

You scan into the ultrasound and you grab one single volume at a pre-selected speed at a pre-selected angle of acquisition.

The slowest, the acquisition speed, the higher the, the number of 2D slices in this volume, the better the resolution.

So again, here, by scanning with these mechanical, transducers, you do not have to physically move the transducer elements.

The motor will do that for you at a preselected speed.

You just have to scan over the area of interest and go into an acquisition plan that has been proven to yield the highest number of 2D uh, slices into this volume.

Here's how a 3D acquisition at the level of the four chamber view would work.

This is how you would acquire a volume at that level.

And here is an acquisition for the acquisition where the transducer keeps acquiring, for these slices.

For the acquisitions are important because they will enable the operator to acquire a number of volumes in continuity and on the fly.

These volumes are displayed into a monitor.

Here in this diagram in this animation, you see that the transducer is moving, over the area of interest, and it keeps acquiring, several, volumes in succession, from each other at an operator, selected speed and angle.

We would not ideally use this as our default acquisition mode.

We would only reserve this for moving targets.

And here are two examples of this.

On the left, you see an 11 week, 10 weeker.

The, the acquisition is not too fast, but it is fast enough to actually, keep up with the slight motions of this, PHUs this early.

The PHUs on the right is a 12 weaker, and there, there are some very fine small movements of the hands, that you'd appreciate, by having this acquisition, which is not really that fast, around five volumes per second.

Again, the volume acquisition angle will impact the number of, of, the acquisition of the volume acquisition number.

Typically the wider the angle, the slower the acquisition.

Matrix Array Acquisition

The new and the most advanced acquisition mode is the matrix acquisition.

This is the future of 3D ultrasound with a matrix ac, matrix probe.

We would acquire volumes with, probes that have a very high number of transducer elements.

As we speak. They go anywhere from 2,004,000 to 8,000 elements soon.

This is the future of 3D ultrasound.

This is a, a way to acquire volumes without mechanically sweeping the transducer elements.

Basically, we have scanning here, orthogonally, the beam is basically acquired electronically.

The, the beam is swept electronically, both in the, in the x and in the Y axis.

Typically with a mechanical devices.

We have electronic beam in one direction and mechanical sweeping in the orthogonal plane of that, volume.

Here we have both steps of the acquisition in X and y axis accomplished by a, a matrix, probe that electronically, sweeps the beam.

As I mentioned before, for the first time, we can, we can, have a very fast, very fast frame rate.

We can have a block of tissue and we can scan in real time with this, large amount of volume.

Typically the volumes are faster, and, faster when we have a, a smaller aperture, and then we can really have a 30 frame per second.

But otherwise, if we want, and this is one of the weaknesses right now of matrix arrays.

The aperture is still narrow.

In order to go fast, we'll have to really keep a narrow aperture, up to now.

And, and as we know in obstetrical scanning, ideally we would prefer wider apertures.

So here are some of the features.

You can actually electronically, only have half of this volume, depending on the area of interest.

Or you can have, a biplanar or a rendered display.

Typically either biplanar or rendered displays, are generated real time here is, as you can see, a, a MultiPlan, I'm sorry, a rendered display.

And you see the biplanar, displays underneath them at preselected, scanning planes that can be moved, real time.

As you can see here, these transducers have a relatively small footprint.

These are not necessarily big transducers, and they can focus in a 360 degree fashion.

This is a very important feature.

This is the real strength of using this technology.

You can electronically focus at any single, direction.

This is, truly something that we, will, use, increasingly, in improving our understanding of, of, abnormal and, and normal anatomy.

And, definitely this is the real improvement over the current acquisition, methods that are not able to, to focus in any direction.

Now, as I mentioned before, this plan, these transducers can have biplanar real time scanning, biplanar, real time scannings, that are real time, and they are focused all over.

This is, again, the future of all 3D ultrasound, because now we are not gonna necessarily have to scan with one beam.

We'll scan with more than one beam, we'll generate at least biplanar displays.

Although more displays are possible, or more image planes are possible.

But important here to understand is that two orthogonal images can be simultaneously displayed and being in focus.

Of course, these, images can also be tilted, up and down, or they can be tilted left and right.

So you can have more than one image plane, scanning.

One stationary in the other can be moving.

It can be up and down, or it can be left and right, or it can be in rotation.

So also you can rotate this, image plane at any angle and at, at any point in this, in this aperture level.

Spatial-Temporal Image Correlation (STIC) for Fetal Heart

Of course, some of the main technologies, that involve cardiac scanning has have had a hard time being introduced.

In clinical practice, one of the earliest ways to get fetal hearts used to be, a static 3D volume acquisition.

That was hard. It was hard because as the heart beats grabbing a, a single volume of the heart, you're basically getting an averaging of AP ole and an diastole.

Nevertheless, the orientation of the heart use could be, could be, analyzed this way.

Now, with stick technologies or spatial temporal image correlation, a volume can be acquired and systole and, and diastole can be displayed at the same time.

This is really the, the real advantage.

Of course, the, the stick technology right now works with mechanical transducers, and it is not a, it's a temporary really solution until the matrix arrays become, real, everyday, reality.

The basics of stick, acquisition of spatial temporal image correlation is the image correlation technology.

As we acquire 2D slices, the kid can move, the fetus can have gross body movements, but the, the software, the algorithm will line them up, for us up to a certain level.

If these movements are really great, it may not, be possible to, to correlate them.

The acquisition is a typical slow, in fact, by 3D ultrasound standards, a super slow acquisition.

The data are, are acquired, from 10 to 15 seconds.

So it is a little bit slow acquisition.

The reason for that is the slower the acquisition, the higher the number of 2D slices acquired.

But obviously, the higher also the exposure time, for this kid to move.

So, this is really a trade off.

Stick acquisitions are not easy, especially when we have an overactive fetus.

The data are displayed in a sta in a standard multiplanar display or in a render displays, and then the user can interactively, reevaluate res slice.

These, image planes to the desired, position here is an example of a, of a four chamber view of, of a fetus acquired with a stick.

As you can see from the four chamber view going up, getting the LVOT, there is the aortic valve.

Then if you go slightly higher, you get the pulmonary, as we can very nicely see here.

Both of them are crisscrossing or ruling out a transposition.

This, manipulation is done offline, which really means the volume has been acquired.

The patient may not even have to be in the room.

You may discharge the patient after a, a reliable and, and artifact free acquisition of, of a fetal heart has been displayed.

And as you can see here, again, the information is not static.

We have a ole and end diastole displayed.

All of these hard views can be, not only obtained, but, the way you obtain.

And there are a couple of described methods in the literature of how to do this.

They can actually be automated, and there is a, a, an algorithm right now called, SONO V ad that, can, can automatically generate all of these hard views, simply because the relationship between these, these views in a volume is known.

Guidelines for Acquisition Methods

So, a normal question here would be, okay, great.

We have all of this, this acquisition methods introduced.

When do we use what and for what reason?

So, as the general rule, we, we would ideally prefer acquired fetus.

If we have acquired fetus or, or otherwise a stationary structure, we acquire a 3D volume with the lowest speed of acquisition.

If the fetus is active, we would acquire the fetus faster, faster, so we can avoid the, the motion.

Ideally, the next question would be, do I need a dynamic information?

If the answer is yes, then I'll go with a four D acquisition at the lowest speed.

The, the typical example here would be, for example, a fetal heart or, motion of the extremities.

If I do not need dynamic information, let's say I need to get a fetal hand, but I don't necessarily need to see this hand moving.

I can go with a 3D acquisition, but with the highest speed.

So I can try to avoid the, the movement, of this hand and artifact generation.

So this is a di this is a picture, a composed image of a number of, of, of, equipment here in a, you see that there is a, this is a free hand acquisition, set up.

So here is a, a normal 2D conventional ultrasound machine.

Here is a computer workstation.

They are connected to each other from the video output of the ultrasound machine to a to a PCI video frame grabber on the workstation.

And as the operator acquires 2D slices with the 2D machine, a volume is acquired in the workstation.

This small box here is a position sensing device that helps register the 2D slice.

This setup works very well in stationary structures, in musculoskeletal abdomen, or even gynecology.

But in obstetrics, this is really not the ideal setup because we are after a moving target.

So, in B, this is a, a machine that has the capability of acquiring, volumes with a mechanical transducer.

In this case, here, the probe can be, held stationary.

These are a variety of probes. This is a matrix probe.

This is a vaginal, mechanical transducer.

And as you can see, even vaginal volume probes can be of, similar, footprint as a normal vaginal probe.

This is a neonatal head probe.

This is a transabdominal volume probe.

This is a breast, volume probe.

So as you can see, the footprint of the transabdominal ones are slightly larger, but they are increasingly being reduced both in weight and footprint, and they're very easy to use.

Display Methods in 3D Ultrasound

So that was pretty much the acquisition.

The display is, the next step.

Typically one of the oldest, display methods is the so-called niche mode, where both three planes can be intersected into a so-called origin or center, and any image plane can be picked up and, scrolled through the anatomy.

This is a, a good way for beginners to understand what 3D ultrasound can offer in terms of, volume and image, acquisitions.

Eventually, most users will need to grow out of this mode, into a, a normal multiplanar display, which is what we would normally use.

Now, a tomographic display, simulating CT and Mr have been, have been used, have been introduced by many manufacturers.

Here, a volume is being displayed in a layout that, will display several 2D slices in a grid that is operated selectable.

The distance is all also operated selectable.

You can either fan this grid into a, a coronal, or you can have it in a s or you can have it in a transverse and interactively obtain every single image plane displayed.

As such, you can have a number of them.

You can have only three, four.

You don't have to have necessarily 16 of them.

This is a stick acquisition with collar doppler of a fetal heart, where you can see that, in relationship to the DR large.

All the heart views required for a standard fetal echo are obtained.

Here are the, the, this is the three vessel view, the pulmonary, the aorta, and the SVC.

This is also the pulmonary artery and the right pulmonary branch.

This is the level where both, both the bifurcation of the, of the, of the, bronchi and the kina level.

This is an LVOT, and this is a four chamber view.

So basically, from the four chamber view, you're going up and you start picking up the LVOT, then the RVOT, and then the roof of both pulmonary and the aorta, the so-called the three VT view, or the trachea view.

Now, the, displays don't have to necessarily be orthogonal and straight.

Manufacturers are increasingly, putting efforts to introduce tools that can adjust and conform to the curvature of the anatomy, which can be irregular shaped.

So in this particular case, as you can see, the green line represents a rendering line along the, the curvature of the spine.

And here is the corresponding, display.

Of course, volumes can be obtained with color or power doppler.

These acquisitions are slow.

Typically, these are acquisitions that, are not in four D modes, can only be done in a 3D mode, and of course, are subject to artifacts.

This is the whole cardiovascular system, as you see here, the umbilical arteries.

Here's the umbilical vein, left portal vein, DRS Osis, left and right, hepatic veins, and here's the descending aorta.

So this very otherwise very complex anatomy of the venous return can be, displayed, spatially, which really helps, the user, judge, especially some of the, very, complex malformations that can happen in this particular anatomic region.

This is the, the vi ERUs of VS.

Acquired with a 3D probe.

With, this is acquired with a transvaginal probe, in a head down fetus.

Here. In this, in this slide, you have a number of, of technologies offered by 3D ultrasound that can visualize the cardiovascular system.

Now, here we have both doppler and non Doppler based technologies.

To visualize the, the, cardiovascular system in a, we have a volume acquired with, a normal single one 3D, volume.

And the rendering is minimum intensity projection.

This is the setting where the minimum amount of echoes, are highlighted more.

As you can see, you have this nice angiogram looking, visualization of the cardiovascular system.

This is the, the aortic arch and descending aorta.

Here is the IVC. This is another very interesting, display in B.

This is a B flow.

B flow is a, is a gray scale technology that displays the, the vessels, the same as an angiogram, but again, a, a volume angiogram, a three dimensional angiogram, which means this can be rotated.

In C, we have a par Doppler, of a normal case at 22 weeks of the cardiovascular system.

Here is the portal, sinus umbilical vein, doctor's, osis, right and left hepatic.

Indeed, we have an agenesis of doctor's osis into a nunan syndrome baby.

As you can see here, umbilical vein enters the abdomen and it bypasses the liver, and it shoots straight into the right atrium.

Here in e is the same, case and, by B flow.

So this is rendered in B flow.

And as you can see, the umbilical vein, sorry, the, the E is actually a persistent right umbilical vein.

E and F are the same patient.

So as you can see here, the left umbilical vein is missing.

And the right, the, the one that normally has to obliterate is persistent.

But still, we have a, a doctors osis present.

These conditions actually may end up in, in AVMs, the duct osis is not always present.

And, and we need a thorough fetal echo evaluation.

And here is a power doppler of the same, persistent of the right, umbilical vein case.

Now, here is a, a heart at 24 weeks that has been acquired with stick and, and rendered with a, with an inverse rendering.

As I mentioned before, inverse rendering, allows us to have visualization of the hypoechoic structures.

It makes them hypoechoic.

So, as you can see here, this is the left ventricle.

This is the right ventricle.

From the right ventricle, we have the pulmonary artery and the doctor's continuation here from the left ventricle underneath the pulmonary, we have the aorta aortic arch.

Here's the aortic isus, and both of them drain to the descending aorta.

This is actually a very useful, view.

Some of the manufacturers are trying to improve on this view that by superimposing color or power doppler information, color doper information would be very useful in this particular, setup, because it'll also show the direction of flow.

Inverse rendering by itself will not be able to show the direction of flow.

Adding color doppler will, enable the operator to see the direction of flow.

So important in fetal echo examination, here is a 24 weeker of a cardiovascular system obtained with stick, and this, and rendered with maximum intensity projection.

This is a stick B flow.

So this is a, a visualization of the cardiovascular system without color or par Doppler.

The user here has to really keep in mind, this is a, a very interesting display because this technology does not depend on Doppler shift as such, it is not angle dependent.

The angle dependency makes it very appealing.

Also, as you can see here, there is no blushing artifacts.

We don't have to adjust PRF, for higher or lower velocities.

Both higher and lower velocities are, picked.

So both high and low PRF areas can be displayed in a non angle dependent mode.

Here is the umbilical vein duct, left hepatic, descending aorta, umbilical arteries.

Here's another case O of a B flow.

Again, you can see here that the, the B flow, image, it can be rotated in space.

So this is really the, the value of having a, a volume.

So you can 360 degree rotate this volume and evaluate that, interactively.

Applications in Interventional Guidance and Quantification

Of course, we can use 3D ultrasound for assisted, to assist interventional guidance.

This is still in its infancy.

Much more work will need to be done, but both multiplanar and render displays can assist and are great, especially in, in visualizing the target or otherwise the coronal plane.

This is what we normally miss.

We don't have access to the target or to the coronal plane.

Far more work will need to actually be done by both the industry and the academia to come up with some, standard, forms of, of interventional guidance.

But otherwise, in our own experience, this is really a very exciting and useful tool.

Quantification in 3D Ultrasound

Quantification has always been a problem with 2D ultrasound, especially when it comes to volumes with 3D.

These volumes can be, acquired and quantified.

Any number of voxels can be traced and, can be appropriately and accurately quantified.

There are a number o of reports in the literature that have come up with nomograms for gestational sac, for lo lung, for liver, heart placenta up to 14 weeks, fetal bladder.

And probably the most, most appealing one is the study of the fetal growth.

One problem here is still volume measurements are still mainly manual.

There are some new semi-automatic, or automatic techniques that try to actually improve this process and speed it up.

But as a, as a main concept here, still, the, the process is manual fetal growth, as a concept has been, has been introduced.

The volume as a concept has been introduced to study the, the fetal growth.

Here. The, the idea here is to, to replace fetal weight 2D variables with 3D volume derived variables.

So instead of using a femur length, you'll use a of, you'd use a fetal thigh volume, and you replace this into the fetal weight formula.

There are technical problems, especially in the, in the, distal parts of a bone, you can have shadowing.

That's why one of the newer, developments is to use the midsection of the thigh only otherwise called the, the fractional thigh volume or the fractional arm volume.

And it has been seen that the, the, the, the, percentage of error has been less than half by using volume derived techniques.

This is another very exciting part of studying the fetal growth, but much more work would need to be done to validate this data, in a wide variety of clinical settings.

Of course, volume, lungs, lung volumes can be, quantified.

The problem here is that the lung volume is not necessarily connected to the lung function.

We have assumed that would be the case before, but nobody as of now, has been able to prove in the literature that we have a direct correlation between lung volumes and lung maturity.

So much more work needs to be done.

This is a, a corona sub, tracing, and the volume, has been, quantified.

And this is a, a graphic representation of that volume.

These are fairly re fairly straightforward techniques.

The, the software in this particular case is the vocal.

The vocal, is a, is, manual ac a manual tracing technique.

The big problem here is when you want to only trace the ambient amnion is tricky, especially in the first trimester.

You'd have to see its borders 360 degrees, and it's not always easy.

But if you can see the borders, then they can be traced.

So not only the corium, but the Amon also.

And there is a paper in the literature that has been able to actually create a nomogram, for the first trimester.

Artifacts in 3D Ultrasound

Artifacts are a big deal in 3D ultrasound.

They are of, different sources. They are multiple.

It's probably one of the weakest points.

In 3D ultrasound, artifacts are really a problem.

There are a number of them related to acquisition, display rendering, and post-processing and editing.

Probably the most common ones have to do with the fact that we have motion artifacts in slow acquisitions.

Another very interesting artifact is the enhancement artifact, especially in the, in the uterus.

You see it in the coronal plane here in the Cornwall.

Near the right side, you see that there is a, an enhancement artifact.

Otherwise, this is the so-called coronal plane.

This is a unique plane because it's perpendicular to the image, to the image acquired by the beam.

And this is as such, this is a transverse, and this is a coronal.

Here we have the external contour of the uterus, and this is the shape of the endometrial canal.

Very useful in ruling out and differentiate in between bi corn versus septate versus arc with uterus.

And this is really the only way to do it.

And to the ultrasound, we cannot reliably differentiate between these entities.

Here is another enhancement artifact in the cervix at 21, weeks gestation.

As you can see here, the cross planar point is not put at the, at the cervical canal.

This is at the cervical canal.

This is a coronal of the cervical canal.

A normal one in this case here is in the posterior lip.

We have put the, the cross planar point, and we still see here what looks like a cervical canal.

This is not the cervical canal. This is the enhancement.

This is the enhancement artifact, in this case.

So this really have to be known.

This is a color, a power Doppler acquired slowly, for five seconds.

And as you can see in B and C, you may not see them in the acquisition plane.

You may see this movement and this motion artifacts, they have to be known because they may be problematic.

This is a rendering artifact, and this is basically a, so, a surface rendering.

This is a maximum intensity projection to look at the, at the bonus structures.

But the threshold, threshold is a tool that removes low intensity echos here has been a little bit exaggerated.

The more you increase the threshold, it starts taking off not only low intensity echos, but soft tissue as well.

And so this may become a problem if you don't know it.

This is an acquisition related artifact.

What looks here in sea is like a, the metal, like a narrowing of the atopic suture.

And normally, a clinician would have to suspect here of a, of a cranial synosis in a, in additional volume.

You see, the atopic suture is normal, no problem.

So as a rule, you should not, base your, your anomalies in one single volume.

It is likely can be an artifact.

As, as a recommendation, more than one volume have to be acquired to really, crosscheck the presence of an anomaly.

Here is a placenta that was acquired, and during the acquisition of around five to six seconds, the mother breathed and she had a deep breathing.

And as you can see here, we have motion artifacts.

Again, keep in mind that they may not be present in the acquisition plane, and the operator has to look for them in the additional planes, typically in the coronal or the, the transverse planes.

Future Developments in 3D Ultrasound

There is a number of new technologies that I, as I mentioned, that have improved 3D ultrasound, harmonic pulse inversion, compound imaging, and thin slice or VCI all improve in, image, image contrast, but, will have an impact on the frame rate.

So that has to be a trade off, and, and something to be known.

Image correlation has made a huge impact on our imaging of the fetal heart matrix.

Arrays are really the new, the new thing.

This is where 3D ultrasound will have a significant improvement over the current tools.

And this is what, we all have to keep an eye on on the new developments of this technology.

Last but not least, both the input devices and the display devices of 3D ultrasound have to be improved.

Our input devices are still very poor.

Force feedback and haptics will improve our capability to manipulate volumes, in space.

And, this is actually one of the earlier contraptions.

As you can see, a force feedback device can go over the surface of the anatomy.

In this particular one, image, you'll see, true volumetric displays.

We are not capable yet visualizing our volumes into volume displays.

Our displays, even though we're dealing with volumetric, objects are still flat.

So we are looking at flat displays.

True volumetric displays are available, in medical imaging, but they're very expensive.

And hopefully, with mass production, we'll see low cost and introduction to our, our, machines.

Last but not least, screenless or monitor less has displays of the, of the anatomy can be, expected soon to evolve, to really be in our, in our, machines.

These are, very impressive displays that will completely change the user interface.

Conclusion: Advantages and Pitfalls

Of course, there are pitfalls.

As I mentioned before, artifacts are a problem.

But the most important thing here is this.

This is still, as we do it right now, 3D.

Ultrasound is a 2D based technique.

As such, it suffers from all the problems that 2D ultrasound has.

Penetration, resolution, frame rate, shadows, artifacts are all, important factors that may impact, 3D ultrasound.

Of course, it'll a 3D ultrasound will not make up for a poor imaging technique.

So scanning in 3D ultrasound, will have to be, kept in mind, would have to really be by, by trained, sonographers or sonologist 3D ultrasound, needs an extra time and extra training.

This is not something that, normally a, a used or an experience to the 2D, sonographer or sonologist can immediately pick up.

You'd need additional time to train yourself and additional time to understand, the relationship between, planes in an interactive display.

We still need faster acquisition methods and more automatic ways to acquire volumes.

But otherwise, it's clear to us that there are major advantages.

We are able to get depth.

We are able to get image planes that are not possible, and we can get both of these at the same time.

We can better improve our, our assessment of complex anatomical relationships.

We can standardize our procedures and protocols much better.

And last but not least, this information is digital.

It can be compressed for security.

It can, I'm sorry, it can be compressed for file size.

It can be encrypted for security, and it can be sent over networks for consultation with the best, pediatric cardiologies in town, in the country, or in the world.

Obviously, the HIPAA and, and, and security of the patient information, aspects would have to be considered.

But otherwise, this is a great way to actually do teleconferencing or, or provide expert, expert analysis of patient information at remote areas.

Thanks for your attention, and I hope that, this presentation gave you an overview of the main concepts of 3D ultrasound.

Thank you.