Next

1 of 50

Breast Elastography in Routine Clinical Practice - Richard G. Barr, MD

Index

Prev

2 of 50

3 of 50

Figure 1

Objectives• Provide a brief historical background of elasticity imaging. • Discuss the principles of elasticity imaging. • Explain the difference between strain elastography(SE) and shear wave elastography (SWE). • Review the use of both SE and SWE in detection and characterization of breast lesions in routine clinical practice. • Discuss the limitations of the technique. IntroductionPalpation has been used for thousands of years to evaluate the human body for pathology. Elastography is the imaging equivalent of a physical examination, or clinical palpation. Our images are now going to be based on the stiffness of tissue and not the anatomy. (Figure 1)

Types of ElastographyThere are two types of Elastography now available. Strain Elastography (SE): where the stiffness is calculated based on the displacement of tissue from a compressive force. This is a qualitative method, so we do not get an absolute number of how stiff something is - we only get a ratio of how stiff it is compared to other tissues within the same field of view. The other technique is Shear Wave Elastography (SWE). In this technique, an ultrasound "push pulse" is applied to the tissues, and conventional ultrasound is used to measure the speed of the shear wave through the tissues. From the speed of sound through the tissues, the strain modulus can be calculated. This is a quantitative technique, so we're going to get an absolute number that we can use to tell exactly how stiff the tissues are. 

4 of 50

Why would we want to use elastography inbreast imaging?Elastography "contrast" for cancerous and non-cancerous breast lesions is between 1000 to 5000%. This was evaluated by Dr. Krouskop in-vitro and shows that there was very little overlap between benign and malignant stiffness of tissues in breast lesions. Therefore elastography should be a highly specific and sensitive way of characterizing breast lesions as benign or malignant.

So with the addition of elastography, we now have three ultrasound modes. B-Mode: where we're looking at the acoustical impedance of tissue which gives us anatomy. Doppler which looks at motion and gives us vascular flow. Elastography which looks at the mechanical properties of tissues and gives us tissue stiffness. (Figure 2)

Figure 2

Figure 4

Strain Elastography of the breast is performed utilizing a conventional ultrasound unit and standard ultrasound breast probe. Software analysis of frame-to-frame difference in deformation in tissue with “mild” compression allows for the display of the “softness” or “stiffness” of a lesion – the strain image. Stress and StrainL0 is the tissue before we applied the stress, L1 is when we applied the stress and, depending on how stiff the tissue is, the strain, or L1 minus L0 over L0, is going to change. (Figure 3)

5 of 50

Figure 3

Another way of describing this is if we had an almond in a bowl of jello and we used a spoon to apply the force, you can see that the jello changes shape, and is therefore soft. The almond did not change shape and is therefore hard. That’s all the system does in this type of imaging. (Figure 4)

So the strain is used to calculate the stiffness between the frame-to-frame differences, and this is all done by software. This algorithm varies by the manufacturer, and the amount of displacement for optimal images varies by vendor. This is qualitative not quantitative.

6 of 50

In the column on the left side, you can see that there are five boxes. That’s what the tissues looked like before we applied the stress. The second column is where we applied the stress, and you can see the boxes change shape based on how stiff they are. So if we look at the box, in the middle of the second column, it does not change shape at all and is therefore the hardest. If we use a grayscale map where black is the hardest, that box would be color-coded black. If you look at the box at the bottom of the second column. It has changed the most, so it’s therefore the softest lesion, and that would be color-coded white. The other three grayscale boxes are varying differences, and they would be colored a shade of gray based on the amount of stiffness they have. (Figure 5) The dynamic range of the coloring, either in black or white or in color, is going to be based on what is the stiffest and what is the softest tissue in the field of view. The Compression Elastogram generated displays the relative stiffness to the rest of the image. Therefore a given tissue will have a different shade of gray in each elastogram. Fat in an entirely fatty breast will appear darker than in a very dense breast. This limits the use of this technique in screening.

7 of 50

Figure 5

In this example the image on the top is a fat lobule with surrounding dense breast tissue. On the elastogram you can see that fat color-codes as white, which is soft, and the dense breast tissues surrounding it color-code as black, which is hard. (Figure 6a) This is what we would expect, however in the bottom image we have a field of view in which the image is almost entirely fat, and you can see in the red circle that a portion of that fat, which is coded as black, appears to be very stiff. It’s not actually that that fat is very stiff, but it is the stiffest fat in the image. We know it’s soft, but in this image it’s going to code as being very stiff compared to the other tissues. (Figure 6b)

Figure 6a

8 of 50

Figure 6b

Figure 7b

In this example, in the thyroid in the left image you can see that in the elastogram the thyroid nodule is dark and the normal thyroid tissue is gray. (Figure 7a) In the right image I've changed the field of view so we're only taking the elastogram in the thyroid nodule. In this case, because the thyroid nodule is relatively homogeneous, all we get is noise because there is very little difference between the strains. (Figure 7b) Looking at the left image, where we had the wide field of view with multiple tissues, we get appropriate color-coding. When we make the ROI box very small and have uniform tissues, all we get is noise. This is a very important point that really needs to be understood when we do breast imagining.

Figure 7a

9 of 50

10 of 50

Invasive Ductal Carcinoma

TechniqueThe technique required to obtain an optimal elastogram varies based on manufacture. Some manufactures require minimal displacement while others require a manual compression-and-release cycle to obtain accurate elastograms. After training and practice it is possible to obtain optimal elastograms on any vendor’s system. A variety of display maps can be used on the elastogram, including gray scale and multiple color maps. The "data" is the same regardless of which map you use. The visual display of the data is a matter of personal preference. My personal preference, unless otherwise stated, is the use of the grayscale map with black as “hard” and white as “soft.” I do this because I rely on measurements and I believe I can do these measurements better on a black and white display as opposed to a color display. 

Figure 9

The algorithm requires that strain changes occur in the plane of imaging. If the lesion is moving in and out of the imaging field of view, (FOV) the change in shape will be interpreted as a stiffness change, and  inaccurate results will be displayed. (Figure 8)

Figure 8

When using this technique we want to make sure the object stays within the field of view, and ideally we want the patient’s breathing to be pushing the lesions, That is going to be our motion or strain that will generate the  elastogram. What we want to do is try to keep the probe perpendicular to the ground and not on an angle, and move patients so that when they’re breathing the lesion will stay in the field of view. (Figure 9)

11 of 50

It is important to have in plane motion, Out of plane motion will give inaccurat results.

Technique

12 of 50

Technique TipsTo get a consistently good Elastogram, it is important to keep the field of view large. We want to include fat, normal breast tissue and the lesion, and to include the pectoralis muscle if possible. The reason is that we now have a wide range of tissue stiffnesses, and that will help keep our dynamic range of the scale very constant. This technique will get a little bit easier to interrupt (interpret?) and images will appear more constant-from patient to patient, or even within the same patient. Using this technique, fat is going to be the softest tissue and will always be displayed as white. If you can include pectoralis muscle, if the lesion is not a cancer, the muscle will be the stiffest tissue and color-coded as black. If the lesion is a cancer, it will be the stiffest lesion, and color-coded black. Another technique is to always compare the stiffness to other tissues, so what we want to do is compare the stiffness of the lesion to normal breast tissues. If the lesion has very similar color on the elastogram as normal breast tissue, it’s probably benign because it has a very similar stiffness. Cancers are almost always much different than normal breast tissue.  You can use the B-mode image to determine the amount of displacement of tissues. I would tend to focus on looking at the B-mode image to make sure you’re scanning things appropriately, and not look at the elastogram, but to save a video clip and then go back and review the elastogram from the video clip. 

13 of 50

PrecompressionThe most important factor in breast imaging that will cause difficulty in getting a good elastogram is precompression. Precompression is the amount of pressure you are using to obtainthe image. i.e. using a heavy hand when your obtaining the images.

The reason you apply pressure to the breast is to increase the stiffness. We are looking at the changes of the tissues with and without that force. If you apply the pre-compression you've made them stiffer so that any given change will cause a less change.

14 of 50

(Barr and Zheng.   J Ultrasound Med 2012; 31:895–902)

15 of 50

How we estimate the amount of pre-compressionThis is the method that I use to get very consistent amounts of pre-compression. I look for an object in the far field - it could be a Cooper's ligament or it could be a rib. As I'm scanning, I find that object, I locate the position where I want to do my elastogram, and I start lifting up the probe. As I lift up the probe, the object drops farther and farther in the far field. When I reach a point where that object is as far in the far field as possible, and we still get a good elastogram, I consider that to be a compression of zero and that's where I want to do my elastogram.

The % compression using our method is 1- (3cm/4cm) =25%.

16 of 50

Pre-compression

17 of 50

Position of Scanning does not appear to change results

Same Invasive ductal carcinoma imaged in different planes and positions.

18 of 50

Bull's Eye ArtifactAnother observation made during the study was that both simple cysts and complicated cysts have a characteristic Elastogram.  Both simple cysts and complicated cysts had a “bull’s eye” appearance with a posterior white spot.  There were 4 solid lesions which were referred for biopsy which had this “bulls eye” appearance, all of which were complicated cysts on pathology, and were aspirated completely.

The Elasticity image has a “bull’s eye” appearance with the lesion black (yellow arrows) with a bright spot in the center (green arrow) and a bright spot behind the cyst (red arrow).  We have found that both simple cysts and benign complex cysts have this appearance.  Also note the size of the cyst is smaller on the elasticity image which in our results suggest this is a benign lesion.

19 of 50

Conventional Ultrasound                         Elasticity Image

We did a study where we looked at 127 consecutive bull's eye lesions ranging from 2mm to 40mm with an average about 9mm. Of these lesions, approximately half were biopsied and pathologically confirmed as simple or benign cysts. The other half met all criteria to be simple cysts on B-mode imaging and therefore were not biopsied. 90% of the cases had a B-mode appearance of a complicated cyst, but 6 of those, or 10%, had the appearance of a solid lesion. What we found by both aspirating the lesion and doing cores was that it confirmed that there was no solid component. So we feel that the bull's eye artifact characterizes lesions as simple, benign, or as a complicated benign cyst.  (Barr RG  UQ 2010)

20 of 50

Shear Wave ElastographyInstead of applying a stress, compression, decompression or vibration, one applies a push pulse creating waves within the tissues, like dropping a stone into a pond. If we measure the speed of the sound in the tissues, which is perpendicular to that push pulse, it is proportional to the stiffness of the tissue. Therefore the harder the tissue the faster the speed of the shear wave sound.  AdvantageThe advantage of this technique is that we can calculate an absolute value corresponding to the stiffness. There is a question that this can be less user-dependent, as we don't have to learn compression and decompression as we do with strain imaging, but again, pre-compression is going to affect these factors both on screen and shear wave imaging significantly. 

Mucinous Carcinoma

21 of 50

37 yo female, recent delivery of baby with palpable mass 

Acute and chronic inflamation

58 yo with abnormal mammogram 

Kilopascals to meters/second conversion chart

Strain vs Shear WaveSo the question that’s often asked of me is should I buy strain or should I buy shear wave imaging. They’re complementary tests and they’re measuring the same thing, but each has its advantages and disadvantages. One of the problems with shear wave imaging is that it uses a push pulse to generate the shear wave, and it’s attenuated, like all other ultrasound beams, by tissue so it has a problem at depth. Usually between 4 to 4 1/2 centimeters we can get accurate measurements. However, usually we're not able to generate shear waves much deeper than 4 1/2 centimeters, and therefore we don't get any information beyond that depth. On strain imaging, as long as you can get a B-mode image, you can get an accurate strain elastogram. So with strain imaging, the thickness of the breast is not a factor. Pre-compression is critical in both of these techniques and significantly affects both In our experience, what we have found is that the sensitivity of strain imaging, in our hands, is a little bit better than it is in shear wave imaging.One of the problems that we noticed with shear wave imaging is that the shear waves sometimes do not generate very well in cancers and you won’t get any signals.

In terms of specificity we find that, in strain imaging, we often have a problem because the lesion may blend in, if it’s benign, with the normal breast tissue, and it becomes very hard to get an accurate measurement to determine a ratio which we need in order to decide if something is benign or malignant. In shear wave imaging, we don't have much problem with benign lesions to color-code soft, and it’s more consistent in bringing us information that we can interpret. Strain imaging is not quantitative, where shear wave is very quantitative. In our practice we like using the cyst characterization or the bull's eye characterization. We've been able to eliminate a large number of biopsies using this technique and we only see that with strain imaging. On shear wave imaging, simple cysts will show up as black because they do not generate shear waves. However if the cysts are complicated and have some debris, they will generate shear waves and will show up as blue. You would not be able to distinguish a complicated cyst from a fibroadenoma because they will have exactly the same color code on shear wave imaging. (Figure 10)

Figure 10

22 of 50

Invasive Ductal Cancer                                Fibroadenoma

23 of 50

Interpretation of Strain ElastographyBecause of the relative scale in strain imaging, evaluation based on the color of a lesion is not accurate.  In other words if you use a color scale where red is ‘hard’ and blue is ‘soft,’ you can't assume every red lesion is going to be a cancer. This is because the scale is not fixed -  it’s relative to what is in the field of view. We already talked about one way of getting more consistent color coding of your images - to have a very large field of view that includes fat, glandular tissue and pectoral muscle as well as the lesion. What we have found that seems to be relatively unique to breast cancer is that on a strain elastogram the cancer appears larger than on the B-mode image. We have found that this ratio, the EI/B-mode ratio, can be used with high sensitivity and specificity in characterizing a lesion as benign or malignant.  Strain ElastographyMalignant breast lesions appear larger in strain elastogram, and benign lesions appear smaller. We can use the EI/B-mode ratio to characterize lesions as benign or malignant with very high specificity and sensitivity.

Barr RG et al, J Ultrasound Med 2012: 31:281-287/0278-4297

Figure 11

Clinical ResultsIn a pilot study that we presented at RSNA in 2006, we demonstrated a sensitivity of 100% and a specificity of 99%. In our initial series  we used the cut off of any EI/B ratio of equal to or greater than 1 as malignant and a ratio of less than one as benign. These were all biopsy-proven cases.  Given the good results in our single site center, we performed a multi-center trial with biopsy-proven pathology using the ‘equal to or greater than one’ ratio as ‘malignant’ and ‘less than one’ as ‘benign.’ In 635 patients, approximately one-third were malignant and two-thirds were benign.  At six international sites we had a sensitivity of 99% and a specificity of 87%.  There were three false positive cases, and in a further review of these cases we noticed that either the B-mode measurements or the elastogram measurements were probably done inaccurately due to the poor conspicuity of the lesion. This is one of the factors, when using this ratio. You have to be able to see the lesion well on both B-mode as well as the elastogram to get accurate measurements. 

24 of 50

Results When looking at the results of our multi-center trial, we noticed that for benign lesions the EI/B ratio ranged from .2 to 1.5 with an average of .76  and a standard deviation of .23. The malignant lesions range was a ratio of .9-3.1, having an average of 1.45 and a standard deviation of .41. The P-value was less than .001. (Figure 11)

25 of 50

Figure 13

In this gauge graph there was good differentiation between the benign and malignant lesions using this technique. (Figure 12) Mucinous CancerAnother interesting observation we made was that this ratio is actually somewhat predictive of tumor grades. What we noticed was that low grade cancers such as mucinous or (callaway???)(spindle-cell?)  cancers or DCIS have a ratio of one or slightly greater than one while the invasive ductal cancers tend to have a ratio significantly greater. In our series that we did, the ratio increased with increasing grade. (Figure 13)

Figure 12

Said NH et al RSNA 2008)

Compression Elastography Ratio of Lesion-to-FatAnother technique that has been proposed to evaluate and characterize lesions is the strain ratio of the lesion to fat. Several others have looked at this, and one study was presented at the RSNA 2008. In 242 biopsied lesions, what they found was that by placing an ROI on the lesion and placing another ROI on the fat, most systems can calculate how much stiffer the lesion is than the fat. What these investigators found was that if the lesion was 4.8 times stiffer than the fat, it was most likely to be malignant, and if it was less than 4.8, it was likely to be benign.

Ueno scale Other techniques have been proposed to characterize the elastograms. One is the scale promoted by ( Dr. Wano???   or the Tasiko  ???? ) score. This scale is a 5-point scale. 1. Lesion is very soft. 2. Lesion has a mixed soft/hard pattern. 3. Lesion is hard and appears smaller than on B-mode imaging.  4. Lesion is hard and appears equal in size to B-mode imaging. 5. Lesion is hard and appears larger than on B-mode imaging.

26 of 50

27 of 50

Strain RatioThis allows us to semi-quantify the data using the ratio of the lesion to background fat.

Courtesy of Dr. Kazutaka Nakajima

The ratio of the lesion to fat is 8.6.  That is the lesion is 8.6 times stiffer than fat.  Early results suggest a ratio of approximately 4.5 is the cut off between benign and malignant.

28 of 50

How Elastography Can Be Used The first major factor is to decide if a lesion is benign or malignant.  You can base this on: - length or area measurements by obtaining that EI/B ratio.  - the relative stiffness-to-fat ratio. - the color scale. You can also decide if the lesion is hard or soft by asking questions like: - Is the lesion a fat lobule? This is sometimes very difficult to decide if something is a hypoechoic or an isoechoic lesion verses a fat lobule. When using elastography, it is very easy to distinguish fat lobules from non- fat lobules. - Is an isoechoic area a lesion? We have found that even in complicated cysts there are many that are isoechoic and are not perceived well on B-mode imaging, but on the elastography they are very conspicuous. - Is a lesion an isoechoic complicated cyst? - Define where to biopsy We can use this technique to find what is the stiffest portion of the lesion, which we assume would allow us to get that highest tumor grade. We can also try to characterize the lesion based on the appearance on how stiff or hard it is within the lesion.

29 of 50

Fat Lobule

Sliding Artifact

If the lesion is moving in and out of the plane during strain elastography a ring of white is displayed around the lesion.  This artifact implies the lesion is not attached to adjacent tissue and therefore most likely benign.

Patient referred for biopsy of a solid lesion. Elastogram demonstrates the lesion is “soft” and similar to fat. Core biopsy confirms the lesion is a fat lobule.

30 of 50

31 of 50

Palpable Mass Identified on Ultrasound at Outside Institution

Posterior shadowing fem malignancies

Patient presents with a palpable mass not identified on mammography or ultrasound.  When an elastogram is obtained at the site of the palpable abnormality the Bull’s Eye artifact is identified confirming the lesion is a complex cyst which is isoechoic to adjacent fat tissue.  Lesion resolved on aspiration.

32 of 50

Lesion identified on mammography confirmed as a BI-RADS 4C lesion (red arrow).  The lesion has a Bull’s Eye artifact suggesting a complicated cyst.  Biopsy changed from core biopsy to FNA – lesion completely aspirated.  Core biopsy then performed confirmed lesion was complicated cyst.

33 of 50

Biopsy Proven Benign Fibroadenoma. The size of the fibroadenoma is smaller on the elasticity image when compared to the conventional ultrasound image.  Our initial results suggest this decrease in size on the elasticity image occurs in benign lesions.

Courtesy of Carmel Smith – Queenlands Diagnostic Imaging

Intracystic Papilloma

We also use this technique to look at lymph nodes. This is one of the first cases about 10 years ago. When we looked at this lymph node on B-mode it did not look very abnormal. When we did the elastogram, the area circled in yellow was very stiff, and persistently stiff on multiple repeat imaging. We went ahead and used the elastography to guide us during the biopsy of this area, and we found that this was a focus of metastatic cancer. Elastography is very helpful in looking at lymph nodes and picking out metastatic foci. (Figure 14)

Figure 14

34 of 50

One of the problems that we found in interpretation of strain images in the multi-center trial was that unfortunately the strained properties of fibrocystic change and even fibroadenomas are very similar to those of normal dense breast tissues. So when we do our elastogram, using strain imaging it tends just to blend in with the background dense breast tissue.

35 of 50

One problem with intrepretation of strain images is that if the lesion has similar elastic properties of the adjacent tissue it “blends in” with the surrounding tissue and therefore is difficult to measure.  In the case above the lesion in the red circle is fibrocystic change which has similar elascity as the adjacent fibroglandular tissue and on the strain image can be intrepreted as the lesion appearing larger and therefore coded as malignant.

36 of 50

A lesion on B-mode may actually be two lesions adjacent to each other.  These can be identified on elastography.  These two cases demonstrate that the “one lesion” on B-mode is actually two lesions, one a cyst (bull’s eye artifact) and the other a solid lesion.

Strain using ARFI I want to mention that we can do strain imaging using ARFI or a push pulse, which is different than shear waving imaging. In this technique, what we do is use the ARFI pulse to generate the compression of the tissues, using an algorithm similar to that of strain. We do not get the absolute value. We are only getting a relative value, but instead of using the motion of moving or breathing to cause the strain, we’re using the ultrasound pulse. 

Prev Step

Step 1YWe first place our region of interest over the are we are concerned about.

Step 3You can see that in the jello the push pulse is slower and when it goes through the harder almond it speed up.

Shear Wave imagingWe use the push pulse or ARFI pulse to generate shear waves perpendicular to the push pulse. The shear wave speed or the Vs is measured with conventional ultrasound B-mode imaging. The shear wave speed is proportional to the stiffness of the tissue: the harder the tissue the faster the speed of the shear wave. You can measure the shear wave speed in a small voxel (point of measurement) or use a color map of the velocities over a wider field of view. Also called 2D shear wave imaging. 

Step 6 (final)We first place our region of interest over the are we are concerned about.

37 of 50

Step 4Conventional B-mode pulses are used to measure the tissue displacement caused by the shear wave. The shear wave speed is calculated based in the tissue displacement.

Step 2We then apply our push pulse.

Next Step

Step 5As you can see the velocity of the wave is faster in stiffer tissue than in soft tissue.

38 of 50

One of the issues with shear wave elastography is that we need to generate shear waves to be able to get an accurate measurement.

Shear Wave Elastography

40 yo WF with a palpable breast mass.  The mass on SWE has a stiffness of 17 kPa indicative of a benign lesion.  The lesion was a benign fibroadenoma on biopsy.

53 yo WF with an abnormal screening mammogram.  The mass on SWE has a stiffness of 79 to 123 kPa indicative of a malignant lesion.  On US guided biopsy the lesion was a mucinous cancer.

39 of 50

BE1 TrialThere has been a large study looking at shear wave elastography in the breast. Called the BE1 trial, it was published in Radiology in 2012. What they found was that for BI-RADS 3 and BI-RADS 4a lesions, if shear waves suggested a benign etiology, you can decrease the BI-RADS score by one, and if the shear wave imaging was suggestive of a malignancy, you can raise the BI-RADS score by 1. Therefore, if you had a BI-RADS 3 or BI-RADS 4a lesion, and the elastography was negative, you can downgrade it to BI-RADS 2 and BI-RADS 3 lesions respectively and not do a biopsy on the BI-RADS 4a lesion. However, if the shear wave imaging is suggestive of a malignancy, that would merit a biopsy even if it was a BI-RADS 3 case. Using those criteria they were able to show that shear wave increased the specificity over B-mode from 61.1% to 78.5% with a p value of less than .001. If they did not find a change in sensitivity, it increased the positive predictive value from 53.6% to 67.1% with a p value of less than .001% Difficulty with Shear WaveOne problem with shear wave imaging is that sometimes cancers may code as “blue” cancers or soft cancers. In our experience, about half of the invasive ductal cancers either code as soft or they may not color-code at all. There is a problem, which is still not clearly understood, that we don't generate good shear waves within a breast cancer. It is often that we get a black - no color-coding -of cancers. Occasionally we'll have a ring of high velocity surrounding a tumor, which is another flag that this may represent a tumor, even though you don't get a shear wave within the lesion. If the shear wave is not generated, the area is not color-coded. In our experience, as long as the lesion is solid it is probably going to be a cancer, and that's a flag that we use. Just remember that if the lesion is a simple cyst, you should be able to tell on B-mode image. It may also not color.

40 of 50

"Soft Cancers" on Shear Wave imagingWe need to more accurately look at the shear waves, and if the quality of the shear waves is not accurate, we need to have some method of discarding what the data is telling us.

6 mm Invasive Ductal Cancer

41 of 50

(Barr   J Ultrasound Med 2012; 31:347–350)

“Soft Cancers” on Shear Wave Imaging

Invasive Ductal Cancer

42 of 50

Quality Map codes that area of the IDC as Red - poor shear wave.  Thus the values on the quality map should not be used to determine if lesion is benign or malignant.

Velocity Map coded IDC (red arrow) as soft

43 of 50

The effect of pre-compression on a cyst.  Note that with minimal pre-compression the cyst has no signal as shear waves do not propagate in simple cysts.  Note that as pre-compression is applied the stiffness of the cyst and surrounding tissues significantly increases.  With moderate and marked compression the cyst has kPa values of a malignant lesion.

In my practice we use a combination of both strain and shear wave elastography. Our sensitivity is almost 100% and our specificity is around 90%. It’s very rewarding to use both shear wave and strain imaging because if we get the same result on both, our confidence in that finding, is extremely high. If we get discordant results then we are concerned something is going on and we do a little bit more work. Most of those cases have been where we have had the shear wave not generating well within the cancer. The development of the quality map has been very helpful because it’s now flagging to us to not use that data. The other manufactures, using shear wave imaging, instead of generating a quality map are increasing the ability of the algorithm to select that they are only looking at very good shear wave imaging. Instead of having two maps, a velocity map and a quality map, they are just not color coding the areas that have the poor shear waves. I think those that have had experience with shear wave and had some difficulties before will notice that in newer software systems some of the issues that you could have are going to be solved because we now understand this process a little bit better. I do want to say that we are never going to be 100% specific and sensitive with this technique, and we do get false positives. 

44 of 50

45 of 50

False positive reuslts on strain and shear wave imaging in fat necrosis.

We can get false negative results. We have had several cases of breast lymphoma, which are soft, and the elastography is correct in coding these as soft. However when using our criteria for breast cancers, saying that only hard lesions are cancers, we might call these benign.

Lesion which is biopsy proven an IDC with central necrosis has a “soft area” (red arrow) on strain corresponding to the necrosis.

46 of 50

Lesion with benign findings on both strain and shear wave imaging which was lymphoma on biopsy.  Although the elastography finding are accurate (the lesion is soft) one must remember that although breast cancers are stiff other cancers which can occur in the breast may not.

Conclusions• Both strain and shear wave imaging provide additional information on breast lesion characterization. • Elastography should be part of a routine breast exam. Every breast mass evaluation in our center, over the last 5 years, has had an elastogram and our positive biopsy rate has increased significantly since we have implemented this procedure. • Both strain and shear wave images have advantages and disadvantages, and often they are diametric to each other. So having the ability of doing both is sometime advantageous as it increases your confidence if both give you the same results.  - The combination of strain and shear wave should be more useful than each individually. In our practice we have found that approach to be extremely helpful. We like to say we have an elastography tool box and that we can use whichever method is going to provide us the best results. • Results previously presented could be reproduced at other sites with appropriate training. If you’re not getting good results, then there must be some technique problem, and we recommend you talk to your vendor about some additional training. • Our sensitivity and specificity are greater than 98% and 87%, and actually they increased beyond that over the last year as more our sonographers gained more experience • The technique adds only a few minutes to the study and can be interpreted immediately. We try to do everything in one setting, so when we do a diagnostic ultrasound and look at the elastogram, if we think it is a cancer we tell the patient that we think they need a biopsy and we do it at that time. • We think that elastography has the potential to eliminate a large number of biopsies because “we can better select which lesions need to be biopsied”. • I recommend that you do not stop doing biopsies until you are able to get good results and validate them at your center. We have found that the ratio of the strain to B-mode image has been the best indicator for malignant vs benign in our practice. Others have used different systems, and they all seem to work. • In our experience the bull's eye appearance has been critical in eliminating a large number of biopsies and being able to tell patients that they have a cyst. We often avoid biopsying complicated cysts because we have found that we can characterize these as benign. 

Thank You - Audio Message

Click to Play

48 of 50

PLAY

Thank You!

TITLE   - Page 1 VIDEO INTRODUCTION IRSHAD   - Page 2 VIDEO INTRODUCTION WEIGERT   - Page 3 BACKGROUND   - Page 5 Density and Increased Risk   - Page 6 WHAT IS "BREAST DENSITY"?   - Page 9 Histology of “Density”   - Page 9 Density Variation   - Page 9 Percent Mammographic Density   - Page 10 Mammographic Methods for Density Assessment   - Page 11 Problems with Wolfe Classification   - Page 12 ACR - BI-RADS   - Page 13 PROBLEMS WITH DENSITY EVALUATION METHODS   - Page 14 Inconsistency in Methods Used   - Page 14 Problems with BI-RADS Density Evaluation   - Page 15 Problems with Visual Density Assessment   - Page 15 DENSITY ASSESSMENT IS NOT A PERFECT SCIENCE   - Page 15 Interobserver   - Page 16 Intraobserver   - Page 16 Quantitative Methods   - Page 18

Spectral Mammography   - Page 19 DENSITY ESTIMATION ON OTHER MODALITIES   - Page 19 What Can We Do For Patients With Dense Breasts?   - Page 20 Breast - Specific Gamma Imaging   - Page 22 Problems in the Density Literature   - Page 22 SCREENING BREAST ULTRASOUND IN CLINICAL PRACTICE   - Page 23 CONCERNS   - Page 26 Office Concerns   - Page 26 Technical Concerns   - Page 27 Referring Physical Concerns   - Page 29 Patient Concerns   - Page 30 DATA COLLECTION PROJECT: YEAR 1 and 2   - Page 32 Results: Year 1   - Page 33 Analysis: Year 1   - Page 34 Analysis: Year 2   - Page 35 Yale University Study   - Page 36 WHAT HAVE WE LEARNED?   - Page 36 State Initiatives   - Page 37 Additional Activity   - Page 39 Step Animation   - Page 37 Thank You!   - Page 46