Tissue Elasticity for Daily Practice - HD

Introduction

Hi, I am Richard Barr from Youngstown, Ohio, professor of radiology at Northeastern Ohio Medical College.

Today I'd like to give you an overview of the technical factors and how to do a breast elastography.

Today I'd like to review the initial results of elastography imaging of the breasts and discuss the limitations of and the of the technique.

And basically do an overview of the clinical applications and how you should be using this clinically.

Early Work and Development

Evaluating elasticity of the breast tissue is found to have limited use in the clinical setting.

However, the technique used in these studies suffered from technical problems not being realtime and being suboptimal and resolution.

With the development of realtime elasticity systems, which have realtime dual display of bemo and elastic imaging, imaging, significant improvement in spatial resolution, the technique is now ready for clinical use and has been approved for use for several years worldwide.

System Functionality and Limitations

The compression listy system generates images that show relative stiffness to the rest of the image.

Therefore, given tissue will have a different shade of gray in each ela, gram fat in an entirely fatty breasts may appear dark or hard than in a normal, very dense breast.

This is one of the limitations of this technique, and therefore this technique cannot be used in screening.

This is an example to show you the difference that can happen because of the scaling of this being a qualitative and not a quantitative test.

Here we can see a fat lo in some dense breast tissue, and we've also included some adjacent tissue, and you can see that fat is white.

For my all, my image I'm gonna be showing white is soft and black is hard, and as we would predict, the fat lo is white.

However, in this second patient, you can see where the area in the field of view is all fat, that there is some area of the fat that's coated as black or hard because it is the hardest tissue within the image.

So even though it's soft fat, it does code black.

And this is very important and you need to remember this as you are doing your images.

Technique Factors

Another key factor is the technique to get optimum images different for every manufacturer.

While some machines require little to no compression, others require some manual compression and release.

You really need to work with your applications, people from your vendor to make sure you're using the appropriate technique.

And you need to also be careful when you're reading papers out of the literature that they're using the same system you have, because the technique or interpretation of the results may be different depending on the vendor.

For our system, we select hard lesions or less strain as black and soft lesions, or more strain as white.

This fat, excuse me, this fat is depicted as a white lesion, and cancers are depicted as black lesions.

Color maps can be used, algorithms requiring the strain changes remain in plain.

So if a lesion is moving in and out of the plane, which sometimes happens with things like fibroadenomas, you're going to get inaccurate results.

So you need to be very conscious of the fact that you need to be holding still in one plane and the lesion must remain within the plane as you're doing the image.

Interpretation: Size Changes and Ratios

In the early work, it was noticed that cancers appeared larger on compression elastography than on B mode imaging while benign lesions appeared smaller.

In 2001, Tim Hall published a paper showing that a ratio of the length of the strain that is EI over the B mode area with a ratio of 1.2 had a sensitivity of 100% and a specificity of 75.4%.

So here's a video clip of an example of an invasive ductal cancer, and you can see on the B mode image the hypo coic nodule.

And you can see that on the ELAs gram, the lesion appears much larger than it does on the B mode image.

Here's a still image of that, and again, showing what I've done is the red line and the blue line are the same size.

You can see that obviously the invasive ductal cancer is much larger on the ELAs agram than it is on the beam mode image.

This is a amuse, this cancer that I'm showing a clip of, and it really is about equivalent incised with a ratio of one on the be mode image.

And I show this because we've looked at our results in several hundred malignancies and I've plotted out the ratio of length of the lesion on strained versus that on BM mode image or the EI to B mode ratio.

And you can see that that ratio changes and the more aggressive type cancers have larger ratio.

So more benign type cancers, such as mucinous or CTO carcinomas have a ratio of very close to one, whereas invasive ductal cancers can range anywhere from 1.2 to an average of about 2.5.

And we've actually seen cancers that have had higher ratios of over three.

2006 RSNA Preliminary Results

In 2006 at RSNA, we presented our preliminary results on using this strain elastography.

In breast lesions, we had 123 patients that were sent in for an ultrasound guided biopsy of a lesion.

We did ELAs agram and then correlated the EI to B mode ratio to the pathology.

And out of those 123 patients, 106 had a benign pathology, and 17 had malignancies.

The benign lesions, 105 had a ratio of less than one giving us a specificity of 99%.

And all 17 malignancies had a ratio of equal to or greater than one giving us the sensitivity of 100%.

International Multicenter Trial

With those encouraged results, we did a international multicenter trial with a similar protocol where our patients that were already scheduled for an ultrasound guided biopsy came in.

We had 635 patients.

They were all female with age ranging from 20 to 89, average of age of 56, approximately two thirds of the lesions or 413 were benign, and about one third were malignant 222.

Because we took all comers, we had a very normal distribution of both benign and malignant lesions.

As shown on this slide, these are the results from the six sites, and is the number of cases from that site.

B, is the number of lesions benign on pathology B, less than one, or the number of benign lesions that had a ratio of less than one giving us our specificity.

M is the number of lesions that were malignant on pathology and m uh, greater than one is the number of lesions that had a ratio of equal to or greater than one giving us our sensitivity.

And you can see looking at the malignant lesions that we had a sensitivity of a hundred percent at five of the six sites.

There were three lesions out of 90 that were less than one ratio.

And I think we learned a couple things from this study.

One of those cases was a case of DCIS, where on B mode imaging, the lesion was really not well defined, and we believe the measurements on B mode were wrong.

So using this technique, you really need to be able to see the lesion well on both B mode imaging as well as on the ELAs agram.

Another lesion we believe actually was two lesions, a malignant lesion adjacent to a benign lesion.

And depending on where the measurements were taken, we could get results that were smaller or larger.

And later on in the talk, I'll show you a case showing you an example of that.

And in the third case, while one lesion actually got larger in one dimension and smaller in another, and the site elected to use the smaller measurement.

If we look at the specificities, they range from 67% to 95% with an average of about 87%.

This variability I think was caused because at that time there were some factors we were not aware of.

And I think if we did this study now that our specificities would all increase and we would be above 90% in our specificities.

The benign lesions had a ratio ranging from 0.2 to 1.5, an average of 0.76.

The malignant lesions range from 0.9 to 3.1 with their average of 1.45.

And these results were statistically significant.

Another interesting observation that we made during this study was that both simple cysts and complicated cysts have a characteristic ELAs gram.

And I have left to call this the bullseye artifact.

And I'll show you an example of that shortly.

But the other interesting thing was there were some solid lesions that were referred for biopsy that had this bullseye appearance.

And on biopsy, these turned out to be complicated cysts.

Bullseye Artifact for Cysts

And this is the appearance of that bullseye artifact.

It has three components, and you really need all three components to use this artifact accurately.

There's a black outer ring, a white central dot, and then a white dot posterior to the lesion.

We did a study where we looked at 127 bullseye lesions ranging from two millimeters to four centimeters with an average of nine millimeters.

Of these lesions, 62 or about half were biopsied and pathologically confirmed as simple or benign complicated cysts.

The other half were confirmed as simple cysts on bemo imaging and then were not biopsied.

All of the lesions that had the bullseye artifact turned out to be simple or complicated cysts.

90% of the cases had a BMO image of a simple or complicated cysts, but actually 10% of the lesions that had the bullseye appearance actually had an appearance of a solid lesion on B mode image and had the artifact and were proven to be actually iso dense or solid appearing complicated cysts.

I just want to mark this is a from Phillips actually has several settings, and you need to be aware that if you're using their system, it's their setting Number two, that gives you the bullseye artifact.

They have a setting one that you will lose the central white echo, and they also have something called koic imaging, in which things that are moving fluid show up as yellow on a blue background.

In our experience, we really prefer this EI setting too, because we believe that there are things that can give this appearance that may not be cysts.

We were concerned that would we miss a cystic neoplasm?

And this is a case taken from the multicenter trial.

It is courtesy of Carmel Smith in Brisbane, Australia.

This is a two millimeter papilloma within a cystic leg, cystic area, and you can see that when we do have these cystic lesions, the solid component will appear as a black lesion within the cyst artifact.

So, and we've seen many, many cases now, and we are very confident that if there is a cystic lesion that has a solid component, you will identify the solid component within the cyst artifact.

We have actually published an ultrasound quarterly, the results of the study.

And we showed that by using this artifact in a retrospective study, that we could have reduced our biopsy rate by 25% in our population.

Other Interpretation Methods

There is another way of interpreting images, and it has been promoted by Hitachi.

It's the WAO score.

It's similar to birads.

In their system they use red as soft and blue is hard.

So they give a score of one in a lesion that is green or very soft, a lesion as characterized as two, if it has some soft blue components and is heterogeneous.

If the lesion is hard and smaller than the lesion on B mode, it's given a code of three.

A score of four is given if the lesion is hard and is the same size on B mode imaging.

And a five is given if the lesion is hard and larger than that on B mode imaging.

So in the scheme I've showed you before, we group the WAO scores fours and fives together and say they're malignant.

And we group the one twos and threes together and say they're benign.

On the Hitachi system, they have a different appearance to cysts where you get this blue green red appearance.

And this layered three layered appearance.

In an in order to try to get the this technique to be somewhat quantitative.

People have been looking at using the ratio of the strain measurements and comparing it to fat in the lesion.

So in this technique, you would use a region of interest that you put over the lesion and you put another region of interest over the fat.

And most systems will allow you to give you a number that corresponds to the stiffness.

And although that number has no relationship to a quantitative measurement, the ratio does.

So if you do this using the lesion compared to fat, you can tell how much harder a lesion is to fat.

And we can assume that fat in most patients is gonna be similar.

So there was a paper presented at RSNA in 2008 by sed, and they were able to show that with a ratio of 4.8 or less, a lesion was most likely benign.

And if the ratio was 4.8 or greater, it was malignant.

Pre-Compression Issues

One thing that I think you really need to be aware of is something called pre compression.

In our experience, most people that are having problems or getting poor results, this is the major problem.

The strain image that you generate is affected by how much pre compression or how much pressure you're using when you obtain the image.

This is not the amount of pressure you're using to generate the strain.

This is actually how heavy handed you are when you do the exam.

And that additional pressure that you use causes tissue to be stiffer and the system is so sensitive that it actually picks that up and can give you poor results.

Clinical Use of Strain Imaging

How do we use strain imaging?

In our clinical practice, there are several things we can do.

We can say is the lesion benign or malignant?

In our practice, we really like to use the length measurements.

And again, benign lesions will appear smaller.

Malignant lesions will be the same or larger in size.

You can do the same thing with area measurement.

It works just as well.

It takes a little more time.

And we prefer just using the length measurements because we like to do it while the patient's there and give them the results and it's a little bit quicker.

You can do the relative stiffness to fat.

In our experience, we feel more comfortable that the length measurements may be a little more sensitive in our hands, and it also takes a lot more time to do the ratio.

And at this time, the ratio is actually not approved for use in the United States.

And you can use the color scale that's been promoted by Hitachi.

But in addition to that, we can say is the lesion heart or soft?

If we are often confronted with a lesion in dense breast tissue and we can't tell if it's a fat lo or a a different type of lesion.

And elastography is really great in showing that this is a fat lole because it'll show up as being extremely soft.

You may see an iso coic area and not sure if it's a lesion.

And again, turning on elastography will also help you determine if a lesion is present, and I'll show you some cases of that shortly.

And again, is there a lesion and you're not sure?

Could it be a complicated cyst using that bullseye artifact?

That will be very helpful.

You can use ELASTOGRAPHY to help define where things are better to biopsy.

If you have a very complex lesion, the elastography will show you where the stiffest areas are to guide your biopsy.

And I think we're just beginning to try to look at the differences within the lesion of the strain to see if we can actually generate some additional information about the character of a lesion.

Clinical Cases

So some clinical cases, this patient was referred to us for a biopsy.

This hypoechoic lobular lesion in this dense background was referred to us as a birads for a lesion.

And on elastography you can see that this lesion is extremely soft.

It's much softer than the adjacent normal breast tissue.

And this was a fat lole and we canceled the biopsy.

This is a case of a lipoma, which is iso coic on the be mode image here.

Actually, it shows up as being hard.

And again, if you look, there's only fat in this image.

So again, that's one of the pitfalls.

This fat in the lipoma is slightly harder or about the same as the fat adjacent.

But another interesting thing you see here is there's a white ring around the lesion, and we call this the sliding artifact.

And this occurs when a lesion is moving in and out of the plane as you're doing your strain gram.

So as the patient is breathing in this case and the lesion is moving, the breast is being compressed a little bit between the probe, it's actually sliding in and out of the frame.

And that sliding in and out of the frame leads to this artifact of this white ring.

And this actually has some clinical utility because this is telling you that this lesion is not attached to the adjacent tissue and is moving independently, then the surrounding tissue and therefore is benign.

And you'll see this in not only lipomas, but in fibroadenomas in general, the cancers are attached to the adjacent tissue.

And you're not gonna see the sliding artifact in the malignant lesion.

This was a 36-year-old woman that had a palpable mass.

She went to another place and had a mammogram and an ultrasound that were read as negative.

She saw a surgeon who insisted that there was a lesion there and referred to us.

On our BMO image, we could not identify a lesion.

We put the probe over where we palpated the lesion and turned on strain.

And you can see we've got a perfect bullseye artifact.

And this lesion, which we really didn't realize was a lesion actually is an iso dense, complicated cyst.

We aspirated this with a needle and it completely disappeared, and patient's palpable lesion disappeared.

Large cancer, you can see it's much larger on the ELAs gram Here we can see there's a lot of shadowing on the B mode image.

On most cases, the ELAs, the shadowing does not interfere with the ELAs, and you can see the inferior border of the lesion much better on the ELAs.

This is a nice interesting case to show some principles.

The patient had a screening mammogram.

She had a four B lesion, excuse me.

And I like to say this patient had a lesion that had a body, a head that I circled in red, and it had a tail that I put with the green marker.

You can see that the body got much larger on the ELAs agram as well as the tail.

Very suspicious for a malignancy, but the head actually disappeared on the ELAs gram.

We biopsied the lesion and it came back in the invasive ductal cancer.

We went back to look at the pathology and it turns out that this lesion was actually a fibroadenoma that was attached or adjacent to the malignancy.

So the ELAs Graham was actually correct.

And here again, you can see the pitfall that if you have lesion that is benign adjacent to a malignant lesion and they're very similar in size, and you pick a plane that you're going through the benign lesion, you may get a ratio that's smaller than one.

So now we always look when we do our ELAs to make sure we get similar results, and we try to do the ELAs gram in several planes.

And if we get this current, excuse me, discordant results in different positions, then we're always ly concerned that we're really looking at two lesions that we cannot distinguish on B mode image, and that our ratios may not be accurate, and are a little more concerned about that lesion taller than wider lesion.

We would've classify this as a four B lesion on the strain.

It got much smaller and on biopsy, it was a benign fibroadenoma.

This patient came in and had a mammogram.

She had a lesion that was suspicious.

We went to ultrasound, and here we can see the lesion.

It has a little speculations, it's slightly hypoechoic, it has a very indistinct border.

We classified this as a four C lesion.

We turn on the ELAs and we get a bullseye artifact.

So we put in a needle and aspirated this, it actually disappeared, but because of the appearance on the beam mode image, we were really concerned and actually went in and corded this entire area.

And on pathology, this was indeed just a complicated, benign cyst.

Another very complex lesion on B mode imaging that has some cystic and solid components on the strain.

You can see that of this bi lobe lesion, one half is very hard, the other half is very soft.

This patient did present with bloody discharge.

Our interpretation was that this soft area was hematoma and the hard part was the malignancy.

We used a 20 gauge needle to go in this portion, and we did aspirate liquified hematoma, and then we did a core biopsy of this lesion that showed a benign papillary lesion.

I showed you this case before.

Again, this was a two millimeter papilloma.

And you can see even with that small of a lesion, it's very easy to see within the cyst artifact that this lesion is not a free flowing attachment, but is really a solid lesion.

This is another case just to show you what happens.

Here we can see a complicated cyst with some debris in it.

You can see we have the black circle, the white inner dot, and the white distal dot.

But you could see on the lesion on B mode, this is concave where, and the ELAs, it's convex.

And the reason is that this material in the bottom of the cyst is not moving and is actually this hard area below the white distal dot there.

So again, we feel very confident that we can determine which areas are cystic and which areas are solid within these complicated lesions.

This is one of the first ELAs grams we did.

It was in a patient that had a very large invasive ductal cancer.

We always look at our lymph nodes and patients that we have high suspicions of a lesion gonna be a malignancy to try to determine if we can stage the patient.

And here we have a lymph node that has normal central fat, but when we did the ELAs Graham, there was an area within the lymph node that remained very hard on all the images.

And again, I apologize for the quality of the ELAs.

This is actually over 10 years old now.

But we saw this heart area.

We went and did a core biopsy of this.

You can use do a biopsy with the elastography on.

And this lesion turned out to be a focus of metastatic cancer within the lymph node.

Another extremely suspicious lesion.

It was picked up on a screening mammogram.

We've got multiple spicules, we've got shadowing, we've got extremely hypo coic lesion.

We actually read this out as a birads five.

And then when we do our ELAs gram, we see our black circle, a white dot, and a distal white dot.

And again, this ended up being a complicated cyst that really appeared as a birads five lesion on mammography, both on mammography as well as on BMO imaging.

And again, we stuck a needle in this and aspirated went away.

But again, because of its high suspicion on BMO images, we additionally cord the lesion and again, on pathology.

This was confirmed to be a benign, complicated cyst.

One of the things that we found in our multicenter trial is our specificities are low, and one of the reasons is that many hypoechoic lesions noted on B mode imaging are encased or surrounded by normal dense breast tissue.

And the problem is on the strain ela, the elastic properties of fibroadenomas fibrocystic change are all very similar to normal breast tissue, so that when you do your ELAs, it's very hard to determine the size of the lesion as it tends to blend in with the adjacent tissue.

And this is a case of mine in this study where this red ring is the lesion we were worried about a lobular hypo coic lesion.

The green ring shows that this is surrounded by normal dense breast tissue.

I interpreted this lesion on the study as becoming much larger.

And now in retrospect, this was a fibro glandular change that probably the lesion is actually not this whole black area, but this other smaller area within this area.

And it actually got smaller.

But again, it's very, very difficult to pick out the sizes.

And one of the things we do now is when we look at these kinds of lesions, we look and see if it is adjacent to normal dense breast tissue.

We try to look to see what is the ratio of the darkness to the lesion to the adjacent breast tissue that's relatively far away.

And if they are the same, then we're much more confident that the lesions are benign.

I will say that ShearWave imaging will not have this problem.

And if you use ShearWave imaging, this will show up as a benign lesion.

And you don't have to worry about those size changes.

Two other lesions, again, we thought on B mode imaging, both of these were a solitary lesion, but when you look on the ela, you can see that actually there is a small cyst and then another lesion adjacent to it.

And again, one lesion, we got a cyst sign here, so this is a cyst adjacent to another lesion.

Again, now it becomes very hard for you to actually determine the length or area ratios because you don't see the two lesions here, and you're gonna measure both on this image, and you're only gonna measure one on this image.

So again, these are pitfalls that make it a little bit difficult.

These always, almost always tend to be benign lesions, but it becomes very difficult in interpretation.

ShearWave Imaging

We talked about ShearWave imaging.

That's another way other than strain imaging.

Both techniques have been shown to be useful in characterization of breast lesions.

Each has some advantages and disadvantages.

Luckily the advantages of one are the disadvantages of the other.

So if you have the pleasure of having both systems, it becomes much better for you and increased confidence if you can get the same results in both lesions.

This is a mucinous cancer.

You can see on the sheer wave imaging, we've got very high numbers with in kilopascals ranging from about 80 to 123.

In our practice, we like to use a cutoff of approximately 60.

Other studies have used higher cutoffs about 80.

Work is continuing, and that number will probably be refined in a consensus on those numbers.

You also have to be aware that sometimes on ShearWave imaging, you may see a low kilopascal central area, but a ring around the lesion.

And if you see a high ring around the lesion, you have to be concerned that that's a malignancy.

And again, if we look here, we've got high numbers on the shear ShearWave.

And this lesion increase from 14 to 18 millimeters on strain, again, both consistent with the lesion being malignant.

Conclusion

So to conclude, at this time the techniques and interpretation of strain images vary significantly from each manufacturer.

And you really need to work with your manufacturers applications people to make sure you're using appropriate technique as well as appropriate way of interpreting the images.

The results that we got in our early studies have been reproduced at other sites in the multi-center trial.

Many other sites are now publishing and getting very good results.

Our feeling is that the sensitivity and specificity of strain elastography based on the size comparisons, is that the sensitivity is probably greater than 98% and probably almost a hundred percent.

And the specificity is approximately 90%.

And as we have gotten better in learning more of our interpretation errors, we believe that number is probably now over 90%.

This technique only adds a few minutes to the study and can be interpreted immediately.

And this may have the potential to eliminate a large number of biopsies, maybe better saying that we can better select which lesions really need to be biopsied.

I would recommend that you, once you get your system, that you don't immediately stop doing biopsies.

You need to do some number of cases to make sure you're confident and you're using appropriate technique.

We really rely on the bullseye appearance.

In a lot of cases, it really helps us to determine if something is a benign, complicated cys.

And as we now have this technique, we're really realizing more and more lesions that we thought were solid or actually complicated cys.

And we can either appropriately, if we're gonna biopsy the lesion appropriately, tell the pathologist that it's a complicated cyst.

In general, your pathologist will not tell you that there's a cyst in a lesion.

If you tell 'em it's solid, you'll get a report back that says there's nothing on this biopsy that can explain a solid lesion.

And then you're stuck with the dilemma of what to do.

And then when you go back and look, the lesion's gone because it was a complicated cyst and you biopsied it and ruptured it.

And now you are in this situation that what do you do?

And oftentimes before we had elastography, these patients went and had surgical biopsies of the area to confirm that there was no lesion missed.

So again, with that bullseye artifact, I think we're doing a much better job if you're gonna biopsy and you can let the pathologist know that you think it's a complicated cyst, and they will confirm that on their path report.

And Ashley, now that we've got much more experience with this, we are usually not biopsying these lesions.

Even if we think that they are solid on the beam mode image.

There are some limitations of strain imaging, the size criteria.

You really need to be able to see the lesion well, both on beam mode and on the strain image.

This is not a problem with sheer wave imaging.

And again, the combination of the two in our hands, we think is much better than each individually alone.

At this time, we can't do screening.

Don't forget you have to do this in the same image plane.

So it's not like you can be scanning the breast with the elastography on.

You need to stop at every lesion and do an image.

So at this time, we really don't think screening with this technique is possible.

Mostly because of time constraints.

Hopefully in the future, 3D systems will come out that may help us to be able to use this as screening.

And a 3D ShearWave technique may be extremely helpful to evaluate the entire breast with some form of elastography in a reasonable amount of time to do a screening exam.

Elastography is still in its infancy, even though we've been doing this for 10 years.

I think I learned something new every day.

So this will continue to evolve.

But I think at this point in time we can say that we really believe elastography should be a part of every breast ultrasound examination.

It will be surely added to the Birads in some form.

And I'm sure as time goes on, it'll play a more and more important role in our Bly Rads classification in ultrasound.

And I also want to conclude that work and other organs is going on since I'm asked that question very much.

And you will be seeing that this technique is gonna be helpful in other organs.

I think the breast is very unique.

The size changes that we see in breasts do not occur in any other organ.

And our sensitivities and specificities for characterization of lesions as benign and malignant are gonna be much greater in breast lesions and in any other organ.

The cyst artifact does occur everywhere in the body.

And again, that can be used to decide if something is cystic and solid.