3D Ultrasound of the Liver: When and Where is it Useful? - HD

Introduction to 3D Ultrasound of the Liver

I am Dr. Franklin Tesler from the University of Alabama at Birmingham.

I'm going to be speaking about 3D ultrasound of the liver.

When and where is it useful?

As I contemplated this presentation several months ago, I looked at the title and realized that it would be easy to rearrange the words like this and retitle it.

3D, ultrasound of the liver, what is it good for?

And if you channel Edwin Starr, you could come up with a phrase like this, absolutely nothing.

But what I hope to convince you of in the next 15 minutes or so is that 3D ultrasound of the liver does have a future with a few caveats.

How Technologies Enter the Mainstream of Imaging

I think when evaluating any new technology, it's interesting and instructive to assess how technologies enter the mainstream of imaging.

So for example, CT was originated in the 1970s, MR in the 1980s static ultrasound back in the seventies, real-time ultrasound came of age in the eighties, color ultrasound in the nineties, and now 3D ultrasound in the two thousands to 2000 and tens.

And subsequently, most of those technologies remain in the mainstream of imaging except for of course, static imaging, which is gone by the wayside.

You'll notice that I show 3D ultrasound somewhat dimmer than the others, and that links back to the caveats that I mentioned earlier that I'll be talking about today.

Keys to Success for Adoption of an Imaging Modality

What are the keys to success for adoption of an imaging modality such that it enters the mainstream of imaging and becomes widely adopted and used?

The Technology Must Be Compelling

The first is that the technology has to be compelling.

And compelling is more of an emotional state.

It refers to the fact that the technology somehow grabs you.

It's intriguing, it's exciting, and yes, those are all emotional rather than objective, but it's important and it's important not just to the medical imaging community, it's also important to the public.

Nowhere is this more evident than in 3D ultrasound in obstetric imaging.

This is a Google image search from just a few days ago for 3D ultrasound and this produced 1.74 million hits.

All you see are babies, babies and more babies.

Even though I didn't mention fetus or obstetrics or any related term, it's all about babies in 3D ultrasound at least to the public.

So there is a compelling element that applies not only to the medical community but to the public at large.

So there's demand for 3D in so-called four D ultrasound amongst the public, and this is a website for a business that happens to be located near where I live in Mountain Brook, Alabama that does 3D slash four D imaging of pregnancies for a fee.

This isn't medical diagnosis, it's responding to a desire from pregnant women to see what their babies look like in utero.

Again, it's very compelling to them and this business and others have responded to that need.

The Technology Must Be Effective

The second element is effectiveness.

Any technology that we adopt has to be proven effective in one way or another, and there are a couple ways to do that.

One of course is anecdotal.

Somebody who uses that technology is an early adopter, just tells you that it works well.

But more importantly, new technologies have to be verified in the literature.

And this is an example of just one article from the Journal of Ultrasound in Medicine in 2006 looking at two dimensional imaging versus three and four dimensional imaging in ob.

And as you can see from the summary statement at the bottom, they found that 3D slash four D volume data is consistent with 2D imaging and there are many other articles like this.

Another area I'll mention parenthetically, where 3D ultrasound has had a great impact is gynecologic imaging as Dr. Baral RAF and others have shown in evaluation of the uterus.

The Technology Must Be Accessible

The next feature is that the technology has to be accessible and accessibility is really about two different things.

The first is cost.

Cost unlike the old days when it was important but not as important, you have to be able to show a return on investment in adopting any new technology or even upgrading an existing technology.

And this is something we deal with day in and day out as we seek to keep our ultrasound equipment at UAB at the forefront of ultrasound imaging.

It's not enough to say this is something we want, this is something we need.

We have to show why we need it.

The second is ease of use and to me, ease of use all comes down to user interface design.

I'll have more to say about both of these a little bit later in the talk.

Defining 3D Ultrasound

Now this may seem like a strange question to ask, but I'll ask it anyway because I can.

I'm going to define what 3D ultrasound is because it's not always evident that 3D ultrasound is more than just volume imaging.

So we always start off with acquisition of a volume data set as shown diagrammatically.

Here we end up with a set of volume elements or voxels that make up a volume acquired by the ultrasound machine.

Those voxels can then be processed in either of two ways.

The most familiar as I just mentioned, is volumetric imaging where you end up with a shaded display showing various structures.

In this case, as I rotate this volume around, we can see the aorta with a dissection flap and the superior mesentary mesenteric artery with a dissection flap as well with the liver in front.

This is what most people think of when they think of 3D imaging.

But the second aspect of 3D imaging, which is illustrated here on the right is multiplanar imaging and I believe that the latter is actually more useful in evaluating the liver than the former.

Uses of Volumetric Imaging

So what are some of the uses of volumetric imaging in general?

Well, the first thing we have to realize is that the bar is set really high and that bar has been set primarily by ct.

This is a surface shaded volume rendered display of the aorta and some of the abdominal organs and the underlying skeletal structures.

That's pretty much routine nowadays.

This is from a CT angiogram.

These used to be time consuming and difficult to produce, but these days they're pretty much routine and easy to get.

However, despite their easy availability, I have to say that diagnostically, I rarely rely on these images to decide whether the scan is normal or abnormal or if there's an abnormality.

What it is, what we'd really like to do of course in the abdomen is produce a liver that looks like a liver and this is an image of the liver, a drawing of it that I got from a website that is actually called where is the liver located.com.

It's a real website but I use that to show you what the picture is in everybody's mind of where they'd like 3D imaging of the liver to go.

Now the truth is you can actually do this in ct.

This image that I'm showing is a volume rendered image of the liver that was segmented automatically using some prototype Philip software that one of my colleagues and I were evaluating several years ago.

And it looks pretty much like the liver drawing that I just showed you.

In fact, this software was able to go beyond just rendering the liver volume and actually was able to segment out the hepatic vessels from the portal vessels.

A useful thing that has now been incorporated into clinical workstations.

Contrast that to this 3D rendered image of the liver from a three dimensional ultrasound.

It's definitely less compelling and doesn't look anywhere like the surface shaded display images that I just showed you, but as I just said, I don't think that there is really much use with rare exceptions for this type of display in liver sonography.

The reason is illustrated again by the same set of images of babies and 3D ultrasounds of babies.

The reason these look so good is that the babies are bathed in a natural fluid, amniotic fluid that provides contrast between the soft tissue structures so you can get these beautifully rendered surface displays that applies to some extent in the uterus as well, even though there's no fluid around it.

The reason we can see the endometrium well on 3D images of the uterus is that it stands out.

It's more echogenic than the adjacent myometrium.

The exception in the liver is cases such as this.

This is a hepatic cyst and as I move the volume around, you can see the inside of the cyst.

However, I don't think these sort of images, although they're fun to look at, really provide much more diagnostic information than we already have from 2D images.

Where I think 3D ultrasound or volume imaging of the liver does have some role to play is in multiplanar imaging.

Multiplanar Imaging in Liver Evaluation

Those of us who've been doing CT for many years, as I have remember back in the days when all images were obtained in the axial plane, and you might look at an image like this and wonder if this centrally located structure shaded and dark brown is shaped round or whether it's elongated.

And if we looked at several slices, either cranial or coddle to the first slice and saw the same structure, we could build that picture mentally in our mind and realize that we're looking at an elongated structure.

But that was done internally.

We would have to figure that out.

Well, with multiplanar reformatting, we don't have to think as hard because we can look at that structure in a plane that wouldn't be normally accessible to us with so-called isotropic CT imaging where every voxel has the same or nearly the same resolution.

We can produce images in any arbitrary plane we want with the same resolution as the originally acquired plane and that's a very helpful thing to do.

So we produce these beautiful images such as the sagittal reconstruction from a CT angiogram showing the aorta all the way from the thorax through the abdomen into the pelvis, seeing the skeletal structures and other structures exquisitely.

And the same thing applies to this coronal reconstruction.

Now I have to say, when we began doing routine coronal reconstructions on CT about five years ago now, I was quite skeptical about it and didn't think that these images were going to add any useful information.

I was wrong.

I routinely look at these images now and often find things or clarify relationships that wouldn't be apparent from just the axial images.

So looking at images in various other planes that are different from the acquisition plane is all about clarifying relationships between one normal anatomic structure in another or between an anatomic structure and something that's abnormal.

And here's an example from ct.

This is an axial CT image of the hepatic dome and you can see several hepatic metastases in the dome, but there's also an ovoid structure here.

And just looking at this image or even scrolling through the dataset, it would be impossible to tell where this was located.

But reconstructing this in the con coronal plane you can see is pointed out by the arrow.

And as I'm outlining here, this is actually located above the diaphragm and that's a pretty routine occurrence with the multiplanar reformatting these days for me doing ct.

The other thing that all CT packs pretty much let you do these days with multiplanar reformats is easily link the reformatted planes from the acquisition plane.

So for example, on this reformatted coronal image, we can see this white line corresponds to the location of the axial image and that's a feature I use often as well.

So how does this apply to ultrasound?

Well, ultrasound's a little bit different unlike CT where we typically acquire in the axial plane and then perhaps reconstruct in other planes in ultrasound, we're used to doing whatever we need by moving the transducer into any plane we want.

However, it's not always possible to acquire with ultrasound even being clever about it.

Every plane we want, and that's where Multiplanar reformatting comes in.

This is an ultrasound example that is similar to the co-location that I just showed you on CT where that white line showed on the one image where the other image was located and this is the ultrasound equivalent of a liver.

You'll see the cross hairs in each of these images if you look very carefully, is pointing to this small echogenic structure.

And there it is in the coronal plane, which was impossible to acquire directly.

And you can see that this is pointing to the same structure.

So we know what it looks like in all three planes.

This happens to be a hemangioma.

In fact, this is my own liver and from these images were acquired about five or six years ago and I know that it hasn't changed since.

Here's another example of multiplanar imaging and its value in liver.

And again, this is a 3D process because it's essentially taking the 3D volume data set and looking at dissecting it in different planes as they put this into motion.

You can see an area in the gallbladder here.

Here is the acquisition plane that's thickened.

This represents gallbladder tumor.

Now next to it you can see this area that's hypo coic within the liver and you can sort of tell that this may be close closely located to this area of thickening.

But as this goes back and forth in the reconstructed so-called seaplane, you can see that this area of decreased echogenicity is contiguous with the area of thickening and this represents local extension of gallbladder carcinoma.

So again, it's all about clarifying relationships as I said earlier.

Literature on 3D Ultrasound in the Liver

However, despite anecdotal accounts of how effective a technology is, it's important to validate that in controlled studies and therefore it's important to look at what the literature says.

Unfortunately, when you look at 3D ultrasound in the liver, there's really not very much there.

There's really a relative handful of articles of 3D in the liver, in fact that applies to the overall applicability of 3D in the abdomen.

Here's one study from 2009 from a group in Italy and they used conventional and volume ultrasound to look at focal liver lesions and found that the two techniques were roughly comparable.

Most of the few articles that do appear in the literature talking about 3D in the liver really refer to contrast and unfortunately in this country contrast ultrasound is still not FDA approved.

My hope is that within the next year or two that'll change and I'll have more to say about contrast momentarily.

But here's one such article from 2010 where they talked about 3D ultrasound with contrast and talked about how this technique could help differentiate focal liver lesions by looking them at them in three dimensions and looking at their contrast enhancement patterns in three dimensions rather than two dimensionally as we're used to.

Accessibility: Cost and Ease of Use

Now I mentioned before two elements of accessibility and I'd like to return to this for just a moment or two.

The first was cost.

The second was ease of use.

Return on investment is extremely important these days and to hospital administrators who control the purse strings when you're trying to upgrade or buy new equipment, they're going to ask two questions.

The first is, can you be more efficient?

Can you do the same number of patients or even more patients in the same amount or less time with 3D?

And the jury is still out on this even in fetal imaging.

I don't think it's been conclusively proven that you save time.

There's certainly some controversy about this.

Some people say you do and some say you don't.

But certainly in the abdomen there's very little to show that 3D makes you more efficient.

There have been a few time motion studies that suggest that it may be by allowing you to acquire volumes and then process them offline and get the next patient into the room to scan, but it certainly hasn't been conclusively shown yet.

The second is can you reduce FTEs?

If you can be more efficient, perhaps you can do so with fewer sonographers and other personnel.

And clearly that has not been shown yet either.

The second aspect is ease of use and as I mentioned previously, it's all about the user interface.

This is one of our 3D ultrasound workstations that you see with the diagnostic imaging panel here on the right and the 3D panel here on the left.

I think that the ultrasound and PAX vendors have a way to go before this technology will be as easy to use and as accessible from a user interface perspective as it needs to be.

The Lack of Killer Applications

The other element that I've talked about in previous 3D presentations relates to the lack of what has been called killer apps in other areas, a killer app is generally defined as an application and this was a term that was famously used by Bill Gates at one point.

So it, it applies primarily to the computer industry, but I'm using it to apply to ultrasound as well.

A killer app is an application that is so intriguing, so compelling, so great that it basically drives adoption of the underlying technology platform.

And I'd like to conclude by talking about two potential killer apps for 3D ultrasound in the liver.

Multimodality Image Fusion

The first is multimodality image fusion, which as you can see from these images from the literature refers to fusion of ultrasound images with ct and it could also be with MR images in real time.

Now other institutions have had a fair bit of experience with this and are finding this to be extremely helpful, particularly in intervention as shown here, we're just getting started with this.

At UAB, this is a shot of what it looks like and the way it works is you scan and you're looking at the ultrasound image overlaid on the CT or MR image.

So they're coregistered what you see in one image matches what you see in the other.

Where I think that's going to be helpful is what I've termed here finding things.

And here's an example of that.

This is an axial scan of the liver and It has contains a small Hypodense area here, that measures less than the centimeter.

We euphemistically call these ditzels because we don't know what else to call them, but we're routinely asked to evaluate these sonographic.

The challenge is being able to find these lesions on the ultrasound and know that you're looking at the same lesion as we've seen on ct.

Typically what we do is our sonographers will review the CT or if it happens to be an mr, we'll review the c mr with them and try to relate the abnormality to anatomic landmarks around it.

So maybe blood vessels, the gallbladder, bile ducts and so on, and hope that they can use those landmarks to know that they're looking in precisely the same location as as the lesion.

The multimodality image fusion method that I just talked about I think will make this much easier because they'll be able to actually look at the CT and see the ultrasound overlaid on top of it and know conclusively they're looking at the same place.

Here's another example of an axial image of the upper to mid abdomen in a patient.

And you'll notice that there is a small enhancing lesion here in the liver that I've outlined And the question often comes up whether this is real or a perfusion abnormality.

This is something we encounter very commonly at UAB evaluating multiphase CT with arterial phase studies.

Now in this case, you may notice that there is an enhancing pancreatic mass as well in this patient with a neuroendocrine tumor and this is a hypervascular metastasis, but more often than not, or certainly quite commonly we see these and do follow-up cts and they turn out to be not real.

How great it would be if combining contrast ultrasound with multimodality image fusion.

We can quickly decide using just ultrasound that something is not real and obviate the need for as much CT follow-up as we do.

Volume Flow Imaging

Now. The second potential killer application for 3D imaging is volume flow.

In many areas of ultrasound imaging on a daily basis, we try to estimate volume flow through vessels, in this case a dialysis access graft with an estimation of volume flow.

And we do that by looking at the cross-sectional area and multiplying that by the velocity to get an estimated at least of volume flow, but it's only an estimate and is subject to errors.

It would be wonderful if we could do this quickly and easily in other locations as well, such as the carotid arteries here or moving on to the liver in the portal venous system.

We see flow in the portal vein would be great to be able to estimate what the volume flow in that vessel is or in this trans hepatic shunt being able to know what the volume flow is or in this hepatic artery in a transplant patient.

Well, it turns out that using 3D techniques you can actually estimate volume flow.

As was beautifully illustrated in this article from Journal of Ultrasound and Medicine in 2006 by the group from the University of Michigan.

And I won't go into all the technical details here.

You can read the article for those details if you wish.

But the basic point is that you can use 3D imaging to directly measure volume flow, which is something we really desperately want to do.

And in this article they ended with this statement which was both brilliant and probably a little bit ahead of its time.

In terms of availability of this technology.

I've underlined the word easy in two locations.

They said this would be easy to implement and medical diagnosis would be easy to achieve.

Unfortunately, it hasn't been as easy to implement as they would've hoped, but my expectation is sometime in the next few years this will become available from ultrasound vendors.

They followed up with another article in 2009 looking at volume flow techniques using 3D in a phantom and looked at, in this case, at Pulsitile flow and showed a good correlation between what they were measuring and the actual flow setting.

But as I said, this is not yet available clinically, but my hope is that it will be available in the next couple of years and I think that's going to revolutionize many areas of imaging including evaluation of the liver in patients both prior to and following transplant or following shunt placement.

Current Practice While Waiting for Killer Applications

So what do I do while I'm waiting for the killer applications?

I start by scanning in two dimensions to get the lay of the land as it were.

And in most cases, certainly if it's normal or near normal scan, we just scan in 2D, maybe 3D to get some nice pictures if we want to, but we generally don't need it.

It if it's abnormal, but straightforward, pretty much the same thing.

Stick with 2D, maybe add 3D if we really want to, but it's not necessary, for example to use 3D ultrasound to diagnose gallstones.

Where I'm tending to use 3D these days is for the abnormal complex cases that I want to analyze offline later on.

And the advantage of being able to do that is you don't have to work everything out while the patient is still in the scanning suite.

You've acquired the volume and you can play around with that volume on the workstation later.

As I mentioned earlier, primarily using multiplanar imaging, interrogate that volume in multiple different planes and hopefully come out with a bit better idea of the relationships between normal and abnormal structures or between normal structures that are encompassed within that volume.

Conclusion

So what I hope I've convinced you of in this presentation is that 3D sonography of the liver will be good for absolutely something if not immediately in the next few years, especially with the adoption of ultrasound contrast agents, the approval of ultrasound contrast agents in the United States and the water availability and accessibility of this technology along with volume flow imaging.

Thank you very much for your attention.