Measurements in Cerebrovascular Ultrasound - SD

Introduction

I am Ed Grant and I'm the chairman of the Department
of Radiology at the University
of Southern California and Los Angeles.

And the lecture that I'm gonna be giving today
is entitled Measurements in Cerebral Vascular Ultrasound.

And it's part of the overall course
that was held here in the first morning session
for the SRU meeting in 2007,
which is essentially devoted to the exploration of
measurements in various parts of the body with ultrasound
and how they're performed.

Today I'm gonna be speaking about measurements in
cerebrovascular ultrasound.

Background on Angiography and Ultrasound in Carotid Endarterectomy

Before beginning the actual measurement discussion.
When it comes to ultrasound,
I think it's worth talking about some of the measurements
that were originally used in angiography.

Basically, if you think about this, most patients
who are having an ultrasound are having it
because there's the possibility
of their undergoing carotid endarterectomy
in this particular case.

Then ultrasound really is the decision making modality
to send the majority of patients either to surgery or not.

In the past, this was largely based upon angiographic
measurements and as such, there were large studies
that have been reported in the literature.

Large multicenter studies
that have been reported in the literature determining
what the success rate
or improvement in risk
for stroke is following endarterectomy,
basically validating the surgical technique.

The NASCET Study

The largest of these was the NASCET study,
the North American Symptomatic Carotid Endarterectomy Trial
in which 70%
or greater lesions proven
by angiography were considered the surgical threshold.

Remember, these were all symptomatic patients.

The ACAS Study

The other large study
that was done in North America was the Asymptomatic Carotid
Artery Study, the ACAS study.

And for reasons that are beyond my ability to understand,
the authors of this study chose 60%
or greater lesions to send to surgery.

And again, this dealt exclusively with the benefit
of endarterectomy in asymptomatic patients.

Both of these studies showed that there was improvement
with endarterectomy when it regard to stroke reduction,
but the amount of stroke reduction in the 70%
or greater lesions in symptomatic patients was really quite
dramatic, 17% over two years.

Whereas, and it's worth keeping this in mind,
the ACAS study showed an only about 5% stroke risk reduction
over five years.

So if you think about those two studies, it's important
to remember that the outcome of those patients
or the risk reduction with regard
to stroke is extremely different in those two groups.

Now, the reason I bring these studies up is
that the original measurements by which patients were sent
to surgery were based on angiography.

It's also important to remember that all of the Doppler
thresholds that we used and
continue to use are also based on a comparison
with the gold standard.

And that gold standard in the vast majority
of cases has been the angiogram.

Angiographic Measurement Methods

It's worth pointing out that there are
therefore two different methods
of actually measuring the angiogram.

The older method, and some people have called this the
University of Washington method, is shown on the right.

And in this particular case, you can actually see
that the residual lumen is measured
and compared to the original lumen of the vessel.

Now remember, the problem here is
that the original lumen is unknown to the angiogram
because you really can't see it.

So in this particular case, your neuroradiologist has
to go in and estimate where the original lumen was
to arrive at a percent stenosis.

Now, this may actually more truly reflect the degree
of stenosis, but unfortunately it's fraught
with a huge amount of inter observer error,
as you might imagine, the NASCET method,
which has been applied not only to the NASCET
but also to the ACAS study.

And in fact at this point, even the European study,
this is the study shown here, the ECST study in Europe
has actually gone back and reworked their numbers
as well along the lines of those used in the NASCET trials.

So I think something to keep in mind
that's very important is
that at this particular point in time,
almost all authorities would agree this is the method
by which we should be determining the degree of stenosis.

Assuming that the gold standard
for all doppler measurements is in fact the angiogram
and it is or has been in almost the majority
or all studies that have been done.

It's important to remember that the NASCET method is the one
to which you would have
to compare your doppler values when it comes
to actually choosing thresholds for 70 or 50
or whatever percent lesion.

If you can imagine comparing the NASCET method
with the old method would give you very,
very different degrees of stenosis.

And it's important again in thinking in terms
of your thresholds being based upon a gold standard such
as the one shown here using the NASCET method.

Doppler Parameters in Carotid Ultrasound

Now, in thinking in terms of doppler measurements
or ultrasound measurements in the carotid arteries, one
of the first and foremost things I think
that's worth talking about is the use of doppler parameters.

What parameters do we use to actually try
to estimate the degree of stenosis based on velocity?

And in looking at the literature over the years,
I've been able to find no less than 15
different possible doppler parameters
and ratios that people have used.

But that aside, I think there are probably three
or four that are in common use today.

By far and away the vast majority
of laboratories will measure the peak systolic velocity.

This is very easily reproduced, commonly used,
and in fact is required as a parameter
by all accrediting bodies.

Now a second, and I think very,
very important doppler parameter is the systolic ratio.

In this particular case,
you compare the peak systolic velocity in the ICA
with the peak systolic velocity in the ipsilateral CCA
and arrive at a ratio.

Now, it's very important to point out
that when you do make this ratio
that you take your common carotid artery measurement one
to two centimeters proximal to the bulb.

If you take that measurement more proximally down toward the
clavicle, your ratios will not be correct
because the normal patient actually demonstrates
a decrease in velocity
as you go forward in the carotid artery.

So again, if you take it from the part of the carotid,
that's not where the original numbers were derived from,
you may actually find
that your ratios are way off.

So very, very important that you have consistent
measurements in the CCA that are taken about one
to two centimeters proximal to the bulb.

The ICA. Obviously, you will continue all along the course
of the visualized ICA
and select the highest peak systolic velocity, assuming that
that would be where a stenosis would be if one were present.

Now the beauty of using the ratio is
that essentially you're able to use the patient
as their own gold standard,
and in some cases you may be able to identify patients
who lie outside
of the expected peak systolic velocity values.

Now, a third very commonly used doppler parameter
and one that we do use in all
of our patients is end diastolic velocity.

There is evidence from the older literature
that this may be good
for differentiating particularly high grade
lesions from others.

ROC Analysis of Peak Systolic Velocity and Systolic Ratio

Now, in looking at the ability of the
peak systolic velocity
and the systolic velocity ratio to determine,
greater than or less than 70% lesions, this is a graph,
an ROC curve from our own
experience from our own laboratory.

And what you can see here is that essentially the ROC curves
for these two techniques,
these two parameters are absolutely interchangeable.

And so what this actually tells us is that statistically one
of these parameters is just as good as the other.

But unfortunately, if you think about it, one
of these patients rather are not statistics and
therefore individual patients are best served
by combining peak systolic velocity.

And in my opinion, the velocity ratio.

And again, what you're trying to do as a diagnostician is
to integrate all of the various parts
of the Doppler examination such
that one is not simply a cookbook number
that you plug in based on a chart.

So what you're trying to do here
or what I would recommend
that you do is the two primary parameters
that I would recommend is peak systolic velocity
for the reasons I already gave.

Then look at the gray scale
and the color doppler ultrasound parameters
to get an idea if the plaque burden actually is in sync
with the degree of narrowing estimated
by the peak systolic velocity.

Finally, I would then bring in the ratio
and of course throw in the ICA
and diastolic velocity as well.

The good thing about the ratio is that it serves
as an internal check when patients may have peak systolic
velocity values, as I mentioned,
that are really not representative
of the extent of the disease.

And probably the most common group
of patients in whom you will find this are those in the
carotid group who have poor cardiac output.

Remember that atherosclerosis is a systemic disease.

If you have it in the carotids, likely you're going
to have it in the heart and elsewhere.

So many of the patients that we examine for the carotids
actually will have poor cardiac output
because of their concomitant atherosclerotic
cardiovascular disease.

These patients, for example, you may find a low
or normal peak systolic velocity
and then find that the ratio is actually elevated in
those kinds of patients.

Then your idea is to put all three
of those parameters together
and try to arrive at a final diagnosis.

So the higher ratio versus a low peak systolic velocity,
if you see the low peak systolic velocity diffusely
and then one area where you actually see
that the peak systolic velocity is focally elevated
and that the ratio is focally elevated,
this should give you a good clue as to the fact
that there is significant disease in that area.

There are other reasons why the
ratio may be better than the peak systolic velocity.

Again, statistically it all pretty much evens out in the
wash, but for a given patient, it's important
to pull together all these pieces of important information.

Limitations of Doppler in Estimating Precise Stenosis

Now, one of the things that we are asked if you noticed,
in that first slide, we are often times I think asked
to do the impossible,
and that is to estimate the degree of stenosis using doppler
to a very, very fine amount.

And basically, I've been doing this
for a very long time and I think most people would accept
the fact that
or believe that doppler is actually pretty capable
of estimating the degree of narrowing of a given vessel.

And certainly I think it is true
that the average doppler velocity value peak systolic
velocity value does increase in proportion to the degree
or percentage of angiographically proven stenosis.

Unfortunately, if you look at any one degree of stenosis
as measured by angio
and compare that to the Doppler velocity numbers,
you'll find that for any given degree
of stenosis there's an extremely wide range of
doppler velocity values.

Most labs for that reason don't try to report a single
percent stenosis value,
but in fact report in relatively wide strata.

And actually we went back,
and I'm gonna talk a little bit about this
and the question that had come to my mind is
what is a reasonably wide, if you will,
strata when you report out your doppler percent stenosis
values are very wide ranges more realistic
or can you get down into very tight ranges?

Or in fact, should you not use ranges at all?

Because most of the doppler studies
that have been done simply look at differentiating patients
who lie above or below a certain percentage
of stenosis rather than actually trying to grade them.

So another question then is how well do these typical strata
that we use perform?

Examples of Strata Performance

Well, if you look at these images, I think there's a patient
with a peak systolic velocity of 184 centimeters per second.

You would probably have no problem,
or most people would've no problem estimating
that this would actually be a mid-range
or in the case of the strata used by our lab a
stenosis in the degree of 50 to 69%,

you can see here on the NASCET measured angiogram
that this in fact did equate to a 61% stenosis.

Likewise, here's a different patient.

You can see an area of calcium
with shadowing posterior to it.

You can see intense aliasing suggesting to you
that this actually represents a high velocity
or high grade stenosis.

In this particular case, you can see
that the peak systolic velocity is 377
centimeters per second.

Again, I don't think anyone would have difficulty saying
that this clearly represents a greater than 70%
or high grade or surgical lesion.

And again, we are confirmed here
by the angiogram showing an 82% stenosis using the NASCET
methodology of measurement.

Analysis of Velocity vs. Stenosis Degree

Well, we actually went back
and in a study that we did, looked at a large number
of angiograms
and compared the degree
of stenosis on the X axis with the velocity.

And I think as most people would expect, you can see that
as the degree of angiographically measured stenosis
increases, so does the average doppler velocity.

The thing that I think is very important to point out,
however, if you look at the error bars around that average
or mean velocity, you can see that they're extremely wide.

So again, for a given degree
of angiographically determined stenosis,
there is a wide range of possible doppler values
that will fall in with these patients.

If you look at this, this is one standard deviation you can
see very easily and very immediately
that this overlaps very, very heavily
with its neighbors telling you that there's no way
that doppler can really
differentiate tight degrees of stenosis.

In fact, if you look at this, you can actually see
that not only tight degrees of stenosis,
but actually in this case, pretty wide
variations or strata of stenosis.

So we actually went back upon constructing this chart,
we actually went back and looked at our Doppler values just
to see how well we actually did.

This seems like a very simplistic thing,
but it's something actually very few people had tried to do
or really tried to evaluate.

Seriously, we used less than 50 50 to 69
and greater than 70% stenosis measured angiographically
as our typical reporting strata.

These are the strata that were recommended by the SRU
and are the strata that I would recommend you
use in your laboratory.

Basically we chose peak systolic velocity numbers
and ratio threshold that are really in keeping, I think
with average numbers.

These were actually validated angiographically in our lab,
but they're very similar to the recommended numbers
for the SRU.

And what we discovered was that, again,
this is probably best case scenario,
this stuff was actually all validated against angio.

What we actually found was that the lesser lesions,
those less than 50%,
those without hemodynamically significant lesions,
we did reasonably well.

We identified about 91% of those also patients
with lesions greater than 70%, again, reasonably well,
almost 87%.

But look at the mid range group here,
you could probably flip a coin
and do equally well, if not better in fact,
than at least the peak systolic velocity.

Ratio wasn't a whole heck of a lot better either.

So what this tells us is,
and it's reflected in those very wide standard deviations,
is that breaking down your doppler diagnosis even
into gradations as wide as these is fraught with problems.

So what doppler is very good at is differentiating patients
into two groups above 70 say or below 70.

But in fact, as far as actually grading the degree
of stenosis, we really don't do very well.

And I mean, one could actually say, if those guys knew
how to do a Doppler study, they'd do better.

My bet is very few people could do much better than this.

It's just the nature of the beast.

Now, this is not to say that we should not continue
to stratify, but what this does show is that even the widest
of stratifications is fraught with problems.

I think that for the referring clinicians, it's important
that we, and they as much as possible realize the problems
with doppler as far as the ability to actually
tightly represent
or characterize the degree of stenosis
that there's major limitations,
but that in fact the clinicians really do need
to have some sort of number to hang their hat on.

And I think that this kind of stratification corresponds
to the levels chosen
for operative intervention in the NASCET study,
and it makes a reasonable amount of common sense
to use these kinds of numbers, even though fully knowing
that there are drawbacks to this and
that there is a great deal of lack
of accuracy in that mid range.

Now, this is all based really upon the fact if you look at
this histogram, this shows a very similar thing.

You can see that the majority
of patients luckily fall over here,
and these are basically less than
50% are essentially normal.

But you look at the blue, which is actually those patients
with greater than or equal to 70%, you can see most
of these lie to the right,
but in fact, in the area in between
that there is a very wide range of Doppler values
for those 50 to 69% lesions.

Variations in Doppler Thresholds and Sources of Confusion

Now that said, I think one of the other issues
that's always been the source of a great deal
of confusion in ultrasound is the doppler choice
of the choice of doppler thresholds.

I really think this is probably the biggest source
of confusion in carotid diagnosis.

And if you look at the literature,
you can find almost any number you could imagine has been
applied or recommended as a doppler threshold
for a certain degree of stenosis.

And over the years, people have tried
to figure out why this is,
that there's such variation in the literature.

And early on people talked about inter laboratory variation,
basically saying,
the machines are not consistent,
you scan differently than I do.

So we all have variations
and that's why there's so much difference in the literature.

And I've looked at the literature and
personally think that that's not the reason,
and I think the real main reason,
although there is certainly inter laboratory variation,
but in my opinion, the vast majority of these
apparent differences in threshold recommendation have to do
with the use of differing statistical methods
to arrive at the threshold
and specifically frequently the desire for increased
sensitivity versus specificity.

It's interesting that also I think it's worth pointing out
that most of the commonly used thresholds are actually based
upon pre NASCET studies.

And again, going back to what I said in those first slides,
you really should probably use a threshold
that was actually based upon a study that was performed
after the NASCET trial 1993
because of, again, the gold standards always going
to be angiography.

And you could see comparing those two angiograms
that if you use the older method,
that would probably greatly affect the degree of stenosis
for the same patient estimated by angio.

And if that's your gold standard,
it will also change the Doppler thresholds
that will be found statistically.

Review of Literature Thresholds

Now this is part of a chart from the article
that we published in radiology as part
of the SRU Consensus conference.

And it's very cluttered,
but basically to explain this quickly, if you look at this,
you can see that there is a huge number of articles
that have been written, and this is only part
of what's in the literature, a small part of it,
but there's a large number of articles
that have been written out there
recommending Doppler thresholds.

But what's important is
that the thresholds are simply all over the map.

And rather than trying to crane your eyes into
that rather cluttered slide, and I only show that
because of the clutter,
these are some representative numbers.

One of the early papers post NASCET was one by Greg Moneta,
in which he actually recommended a velocity threshold
of 325 centimeters per second
as your operative number
or your threshold number for a greater than 70% stenosis.

And if you look down the list, you can see
that the numbers are just all over carpenter at 210,
Neel at 270, and this guy named Grant at 175.

So the variation in the literature is just tremendous.

And since most of us don't really have large numbers
of angiograms through which you could come up
with your own thresholds as we did, it's important
for us to know how to choose a number from the literature.

And again, going back, people have said,
these variations have to do
with this so-called inter laboratory variation.

Basically saying your technique, your
equipment just simply may different than mine.

And as I said, doppler is a rough form
of estimation of velocity.

So there is a good bit of inter observer variability.

But in my own opinion,
when you look at the literature closely, I think
what you arrive at is the simple fact that these variations
usually have to do with the statistical methodology
that's been applied to the raw ROC numbers.

Statistical Methods: Peak Accuracy

Now, if you look at the literature closely, again,
you will find that the majority
of authors actually chose peak accuracy to base their,
to determine their threshold number.

And of course, accuracy is we love to be accurate.

Radiologists want their tests to be as accurate as possible.

Unfortunately, accuracy only tells you coldly
how many times you are right
when compared to a gold standard.

It unfortunately does not take into account whether being
right or wrong is important.

So there's really no clinically predictable effect
of only using accuracy.

And what's particularly interesting,
and I think it's best showed in this ROC data, is
that accuracy really starts out if you choose a very low
threshold 125, as you increase your threshold
to ridiculously high amounts, what you see is
that the accuracy really doesn't change a whole lot over
that very, very wide range.

It goes from about 89 peaks here at about 94, 95%,
at about 200, 225 centimeters per second,
and then falls off a little bit as you go
to these ridiculously high thresholds.

Why is that? How can that possibly be?

Well, the reason for this is simply that
if you can simply differentiate patients with less than 50,
from greater than 50% stenosis,
basically normal versus no hemodynamically significant
stenosis, that will comprise the vast majority
of the patients in your lab.

Now in the VA where we did these studies, we had a lot
of severe vascular disease,
and even there the percentage of patients was,
less than 20% who had any abnormality.

For that reason, you automatically start off
with an accuracy of 80%
and you throw a few more percentage points in there
and boom, you're up to 90.

So really accuracy is probably not one of the best ways.

In fact, it is definitely not the best way
to choose a threshold.

But again, as I said, many
of the papers published in the literature did just that.

This actually kind of shows the distribution
of the peak systolic velocity.

And again, if you choose 150 centimeters as your cutoff,
you are gonna pick up the vast majority of these patients
who are essentially normal, less than 50% stenosis,
giving you an automatic accuracy of about 80%
and throw a few other good hits on there.

And your accuracy can easily get up to 80 or 85 or 90%.

So really peak accuracy is not at all a good way
to determine the degree of stenosis,
or recommend a threshold, probably better stated.

Sensitivity and Specificity

Now, sensitivity and specificity, as I mentioned,
do have a direct clinical association Sensitivity,
remember is the number of patients with a disease
as an example, in this case,
lesions greater than 70% stenotic in the ICA
that are correctly identified.

So how many of patients who were proven to have 70%
or greater lesion does a specific threshold yield based on
doppler peak systolic velocity or ratio for that matter?

Specificity is essentially the opposite.

That's one minus the percentage of over calls.

Basically, specificity equates to the number of over calls.

And remember, each of these will have a clinical effect.

So what you're really doing,
and this is a far more reasonable way
to look at choosing a threshold, you're weighing the number
of false positives versus the number of false negatives
that you are actually going
to accept when you choose a threshold.

Unfortunately, as most of us are aware,
there is an inverse relationship between sensitivity
and specificity.

So you simply can't win. Now, how does that work?

Basically, if you think about it,
if you have a procedure such as an endarterectomy
that produces a great benefit,
you probably never wanna miss one of these patients
or you don't wanna miss a lot of patients
who would have a huge benefit, such as patients
with greater than 70% stenosis who are symptomatic.

For that reason, you wanna maximize your sensitivity
and probably want to choose a low threshold.

Conversely, if you have a procedure, in this case, again,
endarterectomy that really only yields a small benefit,
you probably want to be conservative
and raise your threshold, which would equate
to increasing your specificity.

Who would these patients be?

Well, if you think about it, the symptomatic patients
between 50 and 70% stenosis have about a
5% risk reduction in stroke over several years.

Likewise, as I mentioned
before, the ACAS, the asymptomatic patients
who have greater than 60% lesions, again, only a five
to 6% risk reduction in stroke over five years.

These patients minimally,
minimally benefit from the endarterectomy.

And in fact, at the VA, for example, where most
of our patients had bad multi-system disease,
it was very unusual
to actually consider doing an endarterectomy on an
asymptomatic patient.

Again, all you need to do is increase your surgical risk
by one or two percentage points,
and statistically, you're probably not doing your
patients any great favor.

So what I'm saying basically is that in these procedures
where there's minimal benefit, one could be justified in
siding or being on the side of increased specificity.

And what that translates into when it comes
to a threshold is increasing your threshold number.

And again, if you go back to the ROC data,
I think this is very nicely shown here.

As you take a low threshold for differentiating between
patients above or below 70% stenosis, if you take a very low
threshold here, you can see
that our sensitivity is a hundred percent.

So at this rate, any patient who has greater than 125
or less than 125 stenosis, we would pick up all
of the patients who have a 70% or greater lesion.

Well at what cost?

Well, that cost is measured in specificity.

We would overcall
or send about 13% of patients on
to either angiography or surgery incorrectly.

They would have less than a 70% lesion.

So as you increase your peak systolic velocity threshold,
and again, the same applies to the ratio, you can see
that the sensitivity begins to fall
and in fact, quite dramatically, such
that if you have a very high velocity threshold here,
you would only identify 33% of the patients.

Of course, conversely,
you would almost never send a patient onto surgery in
that particular case unnecessarily
or without having a greater than 70% lesion.

Well, unfortunately, I naively thought
that there would be some magical cutoff point,
but if you look at these numbers, you can see
that this absolutely increases gradually,
this absolutely falls gradually.

And so there's no obvious cutoff point here.

So you really have
to be the one clinically to make the decision.

Again, how many patients are you willing to miss?

How many patients are you willing to overcall?

And that's where the clinical trade off comes in.

If you think about those patients in whom there is
significant symptomatic disease, again,
huge benefit from surgery,
you should never miss one of those patients.

And so for that reason, many
of the more recent studies have chosen relatively low
thresholds because again, you do not wanna miss one
of these symptomatic patients
who really will benefit tremendously from the
carotid endarterectomy.

Some people have actually taken this on to
what may be almost an illogical extreme
or to the extreme where, and you can do this,
and we've done it actually in one of our papers.

You can actually calculate the stroke risk
or even cost to society of strokes
and then do a mathematical calculation.

And actually in this way arrive at an optimum threshold.

And again, it's the same story that
because of the huge benefit or decrease in stroke risk
or stroke in the symptomatic population,
you can see here that again,
the optimum threshold here is going to be really, really low
based on, again, cost in the number of strokes
or even cost in dollars for this.

Whereas with the asymptomatic population,
because there's minimal, minimal benefit to endarterectomy,
the actual threshold falls not in one place,
but in fact over a wide range,
but does actually have a nadir here at about 225
centimeters per second.

So we actually looked at a large number of patients
who had angiograms
and who we then looked at the Doppler values
and applied each
of these different statistical methodologies to
that same group of patients.

And as you might imagine, each method
actually did in fact yield a different threshold value.

Peak accuracy actually gave us a single value.

But again, you could change very, very greatly your threshold
with very little difference in the peak accuracy.

Sensitivity and specificity
and outcomes gave us a range.

And remember, each of these methodologies does have an
advantage and some disadvantages.

So if you think about this sensibly,
outcome is the ultimate thing that we're trying to achieve,
good outcome or stroke risk reduction.

So this, those numbers
that we arrive at based on an outcome study
should probably be number one.

If you then apply sensitivity
and specificity of greater than 90%,
that would probably be the most reasonable secondary thing
to keep in mind.

And again, we can adjust among those ranges to try
to maximize accuracy.

And if you look at these numbers, what you see is
that in the symptomatic population, again,
because of the tremendous benefit based on endarterectomy,
the outcome thresholds that you would use are very low
between 125 and 175.

Likewise, the specificity
and sensitivity are in the mid range.

And again, accuracy as we mentioned, is around 200.

So you'd probably want to be at the very upper limit
of the outcome range.

Interestingly enough, if you look at the asymptomatic,
you would logically have to choose a higher threshold
because again, here you want to err on the side of
specificity.

And so in fact, you would logically want to have a higher
stenosis threshold.

What's important to keep in mind here is we are not in any
way trying to tell you
that we are grading the degree of stenosis.

What we're actually doing here is separating patients
who would go on to endarterectomy
or further imaging from those who simply do not need it.

And unfortunately, that's really what all of the papers
recommending thresholds have done.

None of them have actually used a threshold to try
to tell you how to characterize the degree of stenosis.

SRU Consensus Diagnostic Criteria

So what do we do with all this? Do we throw it all out?
'cause it's just too damn confusing? No, probably not.

And in fact, I think our own body, the Society
of Radiologists in Ultrasound
put together a consensus conference, which I was
luckily asked to chair a couple of years ago.

And if you can see the names on this, you will see
that it's a very wide range of experts.

I think we have people from the technology world.

We have people from surgery, vascular surgery,
cardiology, neurology.

It's a very, very wide range of people.

And we all got together for two days and sat in a room
and locked the door and said,
you guys are not getting outta here until we come up
with some consensus agreements on how to use thresholds.

And interestingly enough, we did just that and
left the room in really pretty complete agreement on
what numbers would be reasonable to use.

And these are the SRU consensus diagnostic criteria.

They are in that article published in radiology and are available.

And I think these are very reasonable numbers.

It's interesting in fact that this number here, the
threshold for greater than 70%
is 230 centimeters per second was actually recommended
by Greg Moneta, Greg being the person who published
that article on 325 as a threshold.

And again, if you look at that article,
there was a good reason for choosing 325.

His was the first article that actually recommended
or said that you could safely take a patient
to endarterectomy based on the ultrasound values alone.

For that reason, he chose an extremely high,
unrealistically high, not unrealistic,
but realistic for his purposes,
but a very, very high threshold
because at that time he felt that he never wanted
to send a patient to surgery unnecessarily.

It's obvious over the years with experience now that again,
conversely, you don't wanna miss a lot of patients
with greater than 70% lesions
because of the tremendous lowering of the stroke risk.

So again, all of this is really a trade off.

These are excellent numbers that can be applied
to any laboratory.

And I think one of the things just in closing are the two
things that I think I'd like you
to take away from this chart is that these, again,
are good numbers that could be applied
and reasonably used in anyone's lab.

If you have internally carefully validated numbers,
best thing to do is stick with them.

Don't let anyone try to change you.

But if you need an article
or if you need a reference upon which to hang your hat so
that you have good workable numbers that you might be able
to tweak as you go, these numbers will serve you well.

And they've served us well since we've had the conference,
and I think many people have actually adopted them.

I think the other thing to keep in mind about this is that,
as I mentioned earlier, there are a number of parameters.

Use as much as you can of the Doppler examination
and the real time examination to get to your diagnosis
or estimation of the degree of percent stenosis.

Don't just use a cookbook.

And so the SRU group, again, we thought
that the most important and easiest
and most reproducible number was peak systolic velocity.

And as you saw from my ROC curve, it works quite well.

That should be your primary diagnostic parameter,
but it should be coupled with an estimation
of the plaque burden so that when you see a high number,
does the patient have a concomitant
or a confirmatory amount of visual plaque present?

If so, that's further confirmation that your estimation
of the degree of stenosis is correct, bring into
that then the ratio or even the ratio
and the end diastolic velocity.

And again, the idea being that all
of this stuff brings you unity, gives you a higher degree
of confidence that you are correctly diagnosing
that lesion in one of the recommended strata.

And again, if you see one of these numbers that's out
of whack, it's up to you as the diagnostician
to try to figure out why.

And again, I went into some of those reasons,
but I think this chart has been adopted by many,
many labs and
has served us well over
the last couple of years.

And again, hopefully many of you, if you haven't already,
will consider using it.

And again, I will close
by saying if you have good internally validated numbers,
my congratulations to you continue to use them,
but that these numbers are good
and will work in the general population of both symptomatics
and asymptomatic, given the trade-offs
that we have to deal with.

So thank you.