0:00

Now, in terms of lesion localization, the first thing

0:03

to realize or mention is that DBT is non-isotropic,

0:08

meaning that it's not a truly 3D exam, and you can't

0:12

construct the images in multiple different

0:15

planes as you can in CT or MRI, for example, and

0:19

this is actually why most breast imaging societies

0:25

do not recommend that we call DBT 3D mammography.

0:30

Now, out there in the world, most

0:32

patients and most marketing

0:35

folks have called this a 3D mammography.

0:37

It's easy to say, it's easy to tell

0:39

patients that, but it's not truly 3D.

0:43

So what that means for us is that this

0:45

is why we still need two views for

0:48

lesion localization and triangulation.

0:49

We can't get away with just doing one

0:51

view — for example, let's say the CC view —

0:53

and then reconstructing the ML view.

0:55

For example, the principles of triangulation

0:58

still apply, meaning that medial

1:00

lesions move superiorly on an ML view,

1:02

lateral lesions lower on the ML view.

1:05

And we still have to use that kind of thing

1:09

to determine where a lesion is in the breast.

1:13

Now, we do have some additional information

1:15

with DBT, and we have the scroll bar.

1:18

And the scroll bar can help us determine where

1:22

approximately something is within the breast.

1:25

The way that we do this is first to

1:27

identify the location of the nipple.

1:29

This, with all parts of breast imaging,

1:31

location of the nipple determines where

1:34

we are in the breast and where our lesion is.

1:36

We determine the location of the

1:38

finding relative to the nipple.

1:39

So let's say upper outer quadrant, or let's

1:42

say, anterior per in the right breast —

1:44

that would be 10 o'clock.

1:45

And then we can use principles of triangulation

1:47

in the scroll bar for single-view findings.

1:49

I'll show and demonstrate this in a minute.

1:53

What we can do — for example, let's say

1:55

you see a finding on the CC view.

1:57

You can scroll through to that finding and

2:00

then look up at the scroll bar and determine.

2:03

It'll tell you whether you're sort of more in

2:06

let's say, the superior breast or inferior breast.

2:10

You can then go over to the MLO view and

2:12

look in the appropriate location based

2:14

on your assessment on the CC view.

2:16

Now, it's not perfect.

2:17

You need to make sure that you actually

2:19

understand where the nipple is, which can

2:20

be more difficult than you might imagine.

2:22

And then you can't just use

2:24

the scroll bar just by itself.

2:28

So in our PACS system, we have a few

2:30

different pieces of information up

2:32

here at the top of the screen, right?

2:34

We have our thumbnail views.

2:35

These are all the different images that were obtained.

2:37

We have,

2:38

SMG, which means the standard synthesized mammography.

2:42

And then we have corresponding tomosynthesis

2:44

views for each one of those views.

2:46

We can toggle between tomo and

2:48

SM by clicking this button here.

2:50

We can go into, so call it slab mode,

2:52

which helps us determine a little bit

2:54

better where we are in the breast.

2:56

Um, but really I frequently just use the slice position.

3:00

And so, for example, in this case, we're

3:02

looking at a CC view, as noted here.

3:05

We have I for the inferior part of the breast.

3:08

Here's our scroll bar.

3:09

Basically goes all the way up to S,

3:11

which is the superior part of the breast.

3:12

So in this particular case, if I was showing

3:15

you the images below here, our placement

3:18

in the breast currently — our slice in

3:20

the breast currently — is at the inferior-most

3:22

portion of the breast, as the scroll bar

3:24

marker is all the way down here by the I.

3:30

There are some pitfalls associated

3:31

with using the scroll bar.

3:32

Um, first is that if you're assuming that the

3:35

nipple localizes to the center of the scroll

3:36

bar — I'll go back here for just a minute —

3:38

so for example, let's say this little piece here

3:41

was directly in the middle of this scroll bar.

3:44

And you might say, well, that's

3:45

gotta be exactly where the nipple is,

3:47

right?

3:47

Should be appropriately positioned

3:49

right in the middle of the image stack.

3:51

But it may be that during positioning,

3:54

the nipple ended up down more in the

3:55

inferior part of the image stack.

3:57

So if you're looking for something that was in the

3:59

inferior part of the breast, it would only be within

4:01

this range here, and something in the superior part

4:03

of the breast would be all within this range here.

4:06

Um, that can lead you to some localization problems.

4:09

Um, if you're not aware of that, the

4:11

scroll bar can also, um, be a little

4:13

confusing, um, when you are looking at image

4:17

slices that are closer to the detector.

4:19

Most imaging systems will include an

4:22

additional five to six slices, uh, on

4:26

the imaging detector side of the stack.

4:29

Um, and if you're using something like a rule of thumb

4:32

to say whether something is in the skin, let's say,

4:35

it's in the first three tomo slices, that would hold

4:38

true, then, for let's say, the superior part of the

4:41

breast, which is up against the compression paddle,

4:44

but would not hold true for the inferior

4:46

part of the breast, which is up against the detector,

4:48

because for something to be in the skin, it

4:51

would necessarily be probably in the first,

4:53

let's say, eight slices or something. That may

4:55

lead you astray to, um, thinking about

4:57

something that's a skin-based finding or not.

5:00

In addition, there's some

5:01

paddle flex anteriorly, right?

5:03

So when we compress the breast, we get

5:05

the breast is thickest at the posterior,

5:07

closer to the chest wall, and there's some

5:09

flexing of the paddle, uh, anteriorly.

5:12

And this can lead to different assumptions that you're

5:14

making about where the lesion is within the breast.

5:18

Um, because when the system presents the image

5:23

stack to you,

5:25

what it assumes is the paddle is straight, right?

5:28

Uh, is perfectly level.

5:30

Uh, and so that may not always be the case.

5:32

Um, and like I mentioned, superficial

5:34

lesions can be a little confusing.

5:35

And then of course there's always

5:37

potential rolling of tissue between images.

5:39

And this can sort of lead you astray.

5:42

So, uh, for example, here we have a screening exam.

5:46

This is a cleavage view.

5:48

We see this asymmetry here in the, uh,

5:51

medial aspect of the right breast, uh, on

5:54

this MLO view. Uh, we can see very vaguely

5:58

this same asymmetry making a focal asymmetry.

6:02

We know we're at the margin of the image stack

6:06

because we can see these low-density circles nearby.

6:11

Those are sort of cutaneous fat

6:13

deposits, uh, within the dermis.

6:15

And so we know that this lesion is at least

6:17

very close to the dermis or potentially in it.

6:21

So this patient got called back for

6:22

some additional views, and we got a spot

6:26

compression, uh, CC view of the right breast.

6:31

And we can see the, uh, lesion here.

6:34

This is what we're looking at.

6:36

And if we look up at our scroll bar, um, we can

6:39

see that it is stressing to us that this lesion

6:43

is in sort of the middle part of the breast

6:44

or maybe a little bit superior, um, in our

6:49

ML, full-field ML view.

6:52

We again see our focal asymmetry here.

6:55

Um, and you might say, well, I don't know, maybe

6:58

this looks like it's middle of the breast, right?

7:00

But in reality, nipple down here.

7:02

And so if you draw a line that sort of approximates

7:04

where the nipple would be, this would kind of be a

7:06

little bit superior, which our scroll bar suggested.

7:09

On this side, again, we see, uh, some of

7:12

these cutaneous, uh, fat deposits suggesting

7:15

this is probably very close to the skin.

7:16

We see them here.

7:18

Um, but this is sort of basically how to do,

7:21

uh, lesion localization, uh, with tomosynthesis.

7:25

I just wanted to show you, um, one of the particular

7:27

challenges with identifying where the nipple is.

7:31

So on this MLO view, it's difficult to

7:34

see, but we sort of can barely see or sort

7:36

of would assume that the nipple is, uh,

7:39

probably about right here in this position.

7:41

We see this projects to the middle

7:43

of the, uh, tomo slices on this view.

7:47

If we move over to the CC view and find

7:50

the nipple again, you can see that it

7:51

looks like it projects more inferiorly.

7:54

That might be potentially confusing, but if you think

7:57

about this breast being compressed, you might get

8:00

the compression paddle starting maybe up here, right?

8:03

Or in this portion of the breast tissue, and

8:05

then catching all the inferior tissue as well.

8:07

So if you take that whole image stack, really

8:10

the nipple is actually more inferior placed

8:14

within the image stack, um, which is what

8:17

the scroll bar would tell us here.

8:19

And we can confirm that by seeing the nipple, um,

8:22

we can scroll to the part where we actually see the

8:24

nipple and then realize that it's more inferior.

8:26

So if we were describing something, um, we would

8:28

wanna make sure, like, let's say we scrolled through

8:30

this and we scrolled up from here just a little bit.

8:33

We wanna make sure that we know that that's in

8:34

the superior breast and not the inferior breast.

8:38

I just wanted to make note of, um, this concept of

8:42

additional, uh, slices added to the imaging stack.

8:46

So this is from our PACS system.

8:48

This mm marker here denotes the overall

8:51

breast thickness of 77 millimeters.

8:55

But interestingly, right, we have 83

8:59

total slices in our imaging stack.

9:01

So if you'd say, well, usually, you know,

9:03

tomosynthesis slices are one millimeter thick,

9:05

how did we get 83 slices out

9:07

of a 77 millimeter, uh, breast?

9:10

And so the reason that we have

9:11

that is that those, uh, additional

9:15

image slices are added to the tomo stack, giving

9:19

us, in this case, six more slices to evaluate.

9:21

And those would be, again, on

9:23

the detector side of the image.