Breast Shear Wave Elastorgraphy - HD

Introduction

Hi, I am Richard Barr.

I'm in private practice in Youngstown, Ohio.

Today I'd like to give a follow up lecture on breast elastography, on new developments since our prior talks that we did for Sono World.

Today I'd like to give an update on our work on sheer wave Elastography of the breast.

This is to follow up the other lectures that we have on Sono World.

I do have some disclosures that I work for several companies as here.

Types of Elastography

There are two types of elastography.

One is strain and one is ShearWave imaging strain, sometimes called displacement or compression.

Elastography is based on tissue deformation from an external or patient source.

And this is qualitative, so we do not get an absolute measure of the stiffness of the lesion.

ShearWave imaging applies a push pulse, a result in a ShearWave propagation that can be measured as a velocity.

And this is quantitative so that we get an absolute measurement and know exactly how stiff lesion is.

And both strain and sheer wavy elastography have been shown to improve the characterization of breast masses.

Strain Elastography

On strain Elastography malignant breast lesions appear larger in strain Elastography and benign lesions appear smaller.

The EI to B mode ratio or the length of a lesion on the strain ELA divided by the length of a lesion on B mode, can be used to characterize lesions as benign or malignant with a very high sensitivity, greater than 98% in RS and other people's studies with a specificity of approximately 85%.

Other methods, including a five point scholar scale, have been proposed for using and characterization of breast masses.

But we feel that this ratio measurement gives us better results.

In this example, we have a case of an invasive ductal cancer on the left, which shows the lesion measures approximately six millimeters on the B mode image and on the ELAs gram measure 7.7 millimeters.

And again, we use a ratio of equal to or greater than one is a malignant lesion and a ratio of less than one is benign.

The image on the right shows a benign fibroadenoma, which measures a little bit over one centimeter on the B mode image and measures only approximately eight millimeters on the ELAs so that it has gotten smaller on the ELAs.

And again, consistent with a benign lesion.

ShearWave Elastography

With ShearWave imaging this push pulse is used to generate shear waves, which are perpendicular to the push pulse.

The ShearWave speed or VS is measured with conventional ultrasound vs.

Is proportional to the stiffness of the tissue with the heart of the tissue or the fascia.

The VS one can measure the VS in a very small voxel or a point measurement, or can use a color map of Vs over a larger field of view to obtain the measurements at several pixels throughout the image.

Here are some examples of both a malignant and a benign lesion.

On the left is a mucinous cancer, and you can see that the color coating of this lesion is a lot of reds and yellows, which are very high velocities.

And these velocities can be measured either in meters per second or ki kilopascals, and there's a easy conversion between the two.

On the image on the left, we can see that the mass maximum velocity of the tumor is 6.2 meters per second.

And in our lab we like to use 4.5 meters per second is the cutoff between benign and malignant.

The image on the right is a benign fibroadenoma, and you can see that it color codes blue, which is very low velocities, and in this case it has a velocity of 2.5 meters per second.

And again, consistent with a benign lesion.

Challenges with ShearWave Imaging in Breast Cancers

One of the problems that people have reported with ShearWave imaging is that many cancers may show up as blue cancers or soft cancers.

In our experience, about 50% of invasive ductal cancers code either as soft, meaning they show up as blue on the color code, or they may have no signal.

That is, they show up as black on the ShearWave imaging.

Some of these may have a ring of high velocity surrounding the tumor.

Here's an example of a six millimeter invasive ductal cancer, and you can see on the ShearWave ELAs that it color codes a teal, which is a very low velocity, suggesting it's a benign lesion.

However, you can see on the strain gram that the lesion increases in size from 6.4 millimeters to 10.9 millimeters, suggesting that it is a malignancy.

And this difference in information between ShearWave imaging and strain imaging is somewhat concerning as we're supposedly measuring the same parameters on both of these techniques.

And to show you another case, this is a 67-year-old that had a right mass, a breast mass that was confirmed to be an invasive ductal cancer.

And using both vendors that have ShearWave breast technology present, they both show up as blue or soft cancers.

And you can see that the cancer has a maximum velocity of about 2.2 meters per second on both systems.

If we look at the strain imaging from two, and actually this was done on four systems that the lesion increases in size with an EI to B motor ratio of greater than one on all of the strain systems that we have.

So again, this does not seem to be a vendor specific problem, but a problem that's very specific to breast cancers and the development of ShearWave properties within them.

We wanted to evaluate this a little more.

And we actually have a publication at JUM that shows results that, I'm not gonna go in detail here, but you can see that when we look at the shear waves themselves at the tumor, which is coated blue here, and the surrounding tissue, that the sheer waves are actually all noise within the tumor.

If we look in the peri tumoral tissue, there is some noise in the signals, but we can pick the maximum height of these over time so we can actually get an accurate velocity.

So there's something unique about breast cancers, and in our experience, this phenomena seems to only occur in breast cancers that we get a lot of noise in the sheer wave.

And unfortunately, the algorithms sometimes predict this as blue cancers.

Quality Map Solution

So one of the solutions to this is a modification of the previous algorithms in the system.

And this is an example of one method we can use.

It's the development of what we call a quality map.

So in addition to the velocity map, which is the map we've been showing where the color coding corresponds to the velocity at each pixel.

This quality map actually looks at the shear waves and determines if the quality of the shear wave is good or bad.

And we use a stoplight scale so that green is go, it has very high quality, and red is stop.

The sheer wave quality is very bad to not believe the information.

So here we can see that although this cancer has coated very soft and is suggestive of a benign lesion, that the sheer waves in the region of the cancer are very poor.

And we should now know not to believe this information.

And again, sometimes we can get a black coating, which is the original algorithm which rejects that data and says the shear waves are here so bad, I can't give you a number.

Study on Quality Map Technology

So we wanted to do a study to evaluate this new technology.

Patients that were scheduled for an ultrasound guided biopsy were asked to participate in this IRB HIPAA approved study.

Standard B mode imaging and color doppler was performed.

Strain imaging was performed at 14 megahertz on a Siemens S 2000.

And the sheer wave imaging was performed on a modified Siemens S 2000, which contained SVI, which is now called VTIQ.

And it was performed at the nine megahertz linear probe.

The data was displayed both as the velocity map as well as the quality map for strain imaging.

The EI to B mode ratio was calculated.

A ratio of less than one was considered benign, while a ratio of equal two or greater than one was considered malignant for sheer waves, the highest vs in the lesion or in the surrounding ring was documented.

And note was made if the if there was a ring and of the quality measure, ROC curves were generated.

The results were correlated with the biopsy result.

And if surgery was performed, the surgical results were used as the final pathology.

We had 143 patients that had 165 lesions.

The patient age was 48.5 years with a very wide range from 18 to 81 years.

Lesion size averaged a little bit over 10 millimeters, but ranged from 1.5 millimeters to 86 millimeters.

And on pathology, approximately one third of the lesions are 55 were malignant, and approximately two thirds or 110 were benign.

ROC Curves

This is the ROC curves and let's go through these.

The blue line is the area under the curve, the ROC curve.

When we look at the velocity itself without the use of the quality map, the purple line was mammo birads assuming that BIRADS three was negative.

The green line is our conventional ultrasound, and the yellow line is the ROC curve for our strain ultrasound.

And you can see that in, at least in our hands, the strain elastography provided the best information and as we'll discuss later, the area under the curve here was 0.98.

But we turned out that two of the cancers that we had in the series were actual lymphomas, which are truly soft cancers.

So in those elastography actually was correct, but it counts against us because we were considering only heart lesions as cancers.

And if we would've eliminated the lymphomas and looked at our data as only in breast cancers or primary breast cancers, the area under the curve for the strain would've been 0.99.

I'm sorry, it actually would've been 100.

So based on our ROC curves in terms of the ShearWave imaging we chose a cutoff point to 4.5 or 60 kilopascals as the best cutoff.

And again, without the use of the quality map, that gave us a sensitivity of 50% and the specificity of 95%.

Looking at the ROC curves on strain imaging, again, our results validate our previously reported cutoff of equal to or greater than one as being malignant.

Detailed Study Results

Let's look at our results at a little bit more detail.

If we look at the 110 benign lesions, 16 of those or 15% had no shear waves.

That is the map on the velocity was color coded black.

And all of these had a low quality measure.

And the vast majority of these were cystic lesions as sheer waves do not propagate well in simple cystic fluid.

And therefore these are almost always color coded as black.

If we look at the true negatives with the velocity was less than 4.5 meters per second, which accounted for a little bit over 80% of the benign cases, six or 7% had a low quality measure.

And again, some of these may have been cystic lesions and that gave us the reason for the low quality measure.

If we look at the false positive cases that had a velocity of greater than 4.5 meters per second, there were five were counting for 4.5% of the benign lesions.

Two of these actually had poor quality measure.

If we look at the 55 malignant lesions, there was no sheer wave signal in nine or 16%.

And all of these gave us a low quality measure.

If we look at the false negatives, there were 24 40 4%, and of these 83% had a low quality measure.

So the vast majority of the cases that were false negative actually showed us the we were able to know that based on the low quality measure, and again, we had two lymphomas, which actually had high quality measures, and these are soft lesions.

So the sheer wave velocity was actually correct.

And if we look at all the cancers that had a velocity of greater than 4.5, which accounted for 40% of our cases, all had a very high quality measure.

Summary Table

This table summarizes the results of the study.

Again, comparing our results with mammography, conventional ultrasound strain imaging, sheer wave imaging without the use of the quality measure and the sheer wave of imaging with the use of the quality measure.

And again, what I'd like you to to look at is if we look at our strain imaging, we had a very high sensitivity and specificity.

And again, if we would have eliminated the two lymphomas and looked at only primary best cancers, our sensitivity actually would have been 100%.

And our area under the curve would have been 100%.

If we look at the shear wave imaging without the quality measure that our sensitivity, which is quite low at 48%, but we had a very high specificity at 95%.

But if we look at adding the quality measure and knowing that if the lesion is solid and we have a very poor quality measure, this is because it is a breast cancer, we markedly increase our sensitivity up to 91% in our specificity we lose a little bit dropping to 87%.

But as I said in the beginning of this lecture, one of the things that was concerning to us was that with strain imaging as well as ShearWave imaging, we're looking at the same properties of cancers and we really should get same similar results.

And you can see now by adding the quality measure, we are now getting results both with strain imaging and ShearWave imaging that are very similar.

The strain imaging seems to have a slightly better sensitivity, and the sheer wave imaging appears to have a slightly better specificity.

We have to realize that we will never have a hundred percent sensitivity or specificity because in our lab we have some cases that are often false positive and false negative.

Examples of False Positives and Negatives

And one of the false positive cases that we often see is fat necrosis at some stages, which I think is usually in its early stages, it actually is a stiff lesion.

And here you can see in both ShearWave imaging, we get high velocities here at a little bit over five meters per second.

And then strain imaging, we get a ratio of 1.7.

So here, both strain and shear wave imaging are concordant predicting that this is a hard lesion.

And again, it turned out to be fat necrosis.

And we've had other false positive cases in cases of mastitis as another example of cases that can give us false positive both on shear wave and strain.

And we are probably getting the correct answer.

And again, the problem here is there are some benign lesions that are very hard and there are false negative cases.

And again, in our case we had two lymphomas and these are soft lesions.

And again, here you can see that in both the shear wave imaging where we got a maximum shear wave velocity of 2.2 and a very soft lesion where it's even hard to measure because the lesion is so soft on strain imaging.

So there's concordant information here again, that both shear wave and strain predict this lesion is benign, which it is.

But again, if we use this and assume that this was an a primary breast cancer, these are always gonna show up as false negatives.

Unified Evaluation Method

I'd also like to show a table that I've made.

I think that there are several methods that had been proposed to evaluate both strain and sheer wave of imaging.

And I generated this table as one way of trying to pull all that information together and say that these are all looking at the very similar properties.

And it really shouldn't matter which ones we use 'cause we should get the same results on either.

The first three rows are actually the same technique where we're using the five point color scale.

But what I've done is some manufacturers have blue as hard and some manufacturers have red as hard.

And you're actually on almost all these machines to change to whichever color scale you like.

And again, I really like to use the black and white scale because I believe I can do my measurements much better.

But I will mention that if you are going to use the black and white scale, you need to make sure that you're only showing the elastography data on the ELAs Graham.

So you need to turn off showing the elastography information superimposed on the B mode image.

'cause if you use the black and white scale and you have the B mode image in black and white and the elastography superimposed on it, you really have can interpret those images 'cause you have a gray scale on gray scale image and you really have a hard time distinguishing the two images.

So again, all the manufacturers except one, will allow you to completely eliminate the B mode image from the gram.

So in all the images I showed you, the image of the B mode is by itself and the image of the ELAs Graham is by itself.

But I've color coded these based on kind of a BIRAD score going from one to five.

So on this, using this scale, a birads one would be that the lesion is all very soft.

A BIRADS two would show that you have areas that are soft in some areas that are hard.

A birads three would be the lesion is hard or stiff, but it measures smaller than the lesion on B mode imaging.

If the lesion was stiff and was the same size as the lesion, I would give that a four.

If the lesion was larger, stiff and larger than it was on B mode, I kind of gave that a birads five.

And another thing that we haven't talked about, but I believe I've given a sono world lecture on this, is there's a cyst artifact that we find extremely useful in our practice.

We've been able to eliminate a lot of biopsies based on this artifact.

And this artifact can be the bullseye artifact, which I show here where we have the white in the black background with the white distal spot.

Or on some manufacturers it shows up as this blue green, red, or rg, excuse me, ble BRG signal.

And I kind of gave that as a CYS one and assists two.

And again, that just varies based on the manufacturer, but this is a very useful sign in determining something is cystic with very high sensitivity and specificity.

If we look at the method that I use where we're using the EI to B mode ratio, if the ratio is less than one based on our data, I would give that as a birads two.

Because less than 2% of these lesions turn out to be malignant.

If we have a ratio of equal to one, they gave that a four.

Because some of these lesions are benign, but in our hands some low grade cancer such as DCIS or mucinous cancer, give us a ratio as one.

And if we use the ratio of greater than one, I gave that as a five.

And again, based on our data, we have a sensitivity of greater than 98%, that these are going to be malignant lesions.

Another thing that we haven't talked about on today's lecture, but on some of my older lectures we've talked about is the strain ratio.

Where on EA strain imaging, we can put a region of interest in the mass and we can put a region of interest in the fat.

And although we don't know exactly how stiff each of those are, we can tell how much stiffer the lesion is than fat.

And in the literature, there's a very wide range of what the magic number is.

And I think this is because there's not been a standard protocol for doing this measurement.

But if we look at all the data, everyone that has a strain ratio of less than 2.8 has shown that these are benign lesions.

So we're gonna give that as a a two and everyone that has done this, if their ratio is greater than 4.5, we give that a five.

'cause those cases have all shown that those were malignancy.

And again, there was a wide range between 2.8 and 4.5, and I kind of cheated and gave that as a three four.

If we look at the ShearWave imaging, again, there is different methods of looking at this information.

But if we look all of the papers, let's use kilopascals here, and I give you the meters per second here.

Also, every paper that had a lesion of less than 20 KI kilopascals shown that those were benign lesions who were gonna give that as a two.

And all the papers that showed if it was greater than 80 ki kilopascals have shown those to be cancer.

So we'll give that a five.

In our hands we use 60 kilopascals and other people have come up with 60 PEs scales as their cutoff number also as malignancy.

So I gave that a four.

And then if we go between the 20 and 60, most of those are benign, but there may be some malignancies based on papers in the literature.

So I gave that a three.

And again, I think that now we're able to take all of the methods and come up with a unified method of evaluating breast cancers and showing that these are all somewhat related as we're all measuring the same factor that is the stiffness of the cancer.

Conclusion

So to conclude, I think we wanna say that both strain and ShearWave imaging provide additional information on breast lesion characterization.

We do a elastography on all our breast cases and have done so for many, many years.

Now, technique is very critical and it's very important that you use excellent technique to get appropriate information.

We're very lucky in our lab that we do both strain and ShearWave imaging on our patients.

And it's a very helpful to us that if we get concordant information on both of those, it really increases our confidence.

And if we get discordant information between the two, then we know that something is going on and we do a little bit of extra work to try to figure that out.

And again, we're always going to do elastography in conjunction with B mode imaging.

The B mode imaging is a very critical and the main thing that we should be using to evaluate the breast lesions and the elastography is supplemental to the B mode imaging.

We believe that strain imaging has the highest sensitivity and is greater than 98%.

And like I said, I think we're nearing a hundred percent if we eliminate the false negatives which turned out to be in our hands, mostly lymphomas.

Sheer wave imaging tends to have the highest specificity.

And I think part of this is for us, we have a problem because benign lesions often have the same elasticity properties as normal breast tissue.

And when we try to do measurements, it blends in with the normal breast tissue.

So if we have a fibroadenoma or fibrocystic change that's hypo coic, but it's surrounded by dense breast tissue, oftentimes we have a difficult time doing measurements because it just blends in on the elastography.

And here in these cases we find that ShearWave is often very helpful.

There is a significant improvement in the sensitivity of ShearWave imaging with the addition of a quality measure.

And the combination of strain and sheer wave increases our confidence and accuracy.

Further studies are required for us to exactly determine what we're measuring in breast elastography and why is it so unique compared to other organs.

And again, it seems that breast lesions are the only lesions that show us this change in size on the elastography.

And I think it's still unclear if the elast elastography size is a better indicator of the extent of cancer and should we be using that size to plant surgery.

And with that, I'd like to end and again, there's a lot of additional information in the previous lectures that I've given on breast elastography that we didn't cover in today's lecture.

Thank you.