Doppler Features, Pitfalls, and Artifacts - HD

Doppler Optimization

All right.
I'm gonna talk about doppler optimization.
We have a whole schedule here of vascular and doppler topics.
I'll kind of lay the table here a little bit with the physics part of it.
I'm gonna talk about the doppler equation, certain pitfalls, and then maybe we will put a toe into artifacts if we have time.

I should say this ahead of time.
I have nothing to declare except that I wish I kind of did, because that would be cool.
And the other thing I should declare is that I got a lot of my material from Leslie Scout, and she talks after me, and she's awesome.

The Doppler Equation

This is the Doppler equation about, of all the equations that you may have learned in medical school or residency, this is one that actually is worth knowing.
This doppler equation will tell you a lot about how to optimize your image and what factors you can and cannot control in that optimization process.

Just to remind you of it, the change in frequency are, the doppler shift is directly proportional to the fundamental frequency of your transducer.
So that means that if you wanted to detect more doppler shift, you should use a higher frequency transducer.
And it is also directly proportional cosine theta.
And I'll remind you that cosine of zero is one and cosine of 90 degrees is zero.
So, as you know, already, 90 degrees incidents on a vessel will give you very little are and mostly unreliable doppler signals.

So that's two things you can adjust.
And then that's divided by C, which is the speed of sound and soft tissue.
And that's of course, a speed of sound average assumption that we use with all these machines.

This is how it looks, again, this is the angle of incidents or the angle of intonation that you can control, and then the blood flow that you are measuring.
And in fact, what you're really looking at is the reflection off the blood cells.

Improving Doppler Characteristics

What can you do to improve your doppler characteristics?
You can increase your doppler incident frequency, choose a higher frequency transducer.
You can look at something that maybe has a higher reflector speed, and you can decrease your angle of incidents.
So that's basically what you can do.
But of course, there's always fine tuning and tweaking with these things.
And so I'm gonna talk about some of the pitfalls or some of the tweaks that you might do with doppler.
And that has to do with recognizing variability of the Doppler measurements.
Things you might want be able to do with the doppler angle, sene, spectral, broadening helical flow, boundary layer separation, which is worth knowing about, and vessel wall discrimination.

Pitfalls in Doppler Measurements

Doppler measurements are pretty good.
I think the wonder of it is, is that they work as well as they do, but there's fair bit of variability in them.
There's some noise.
There's noise that's introduced by differences in technique in different types of machines.
You won't necessarily get the right or the same value from one machine to another.
Physiologic differences in patients.
And then vessel and plaque morphology will also influence the doppler signal that you will analyze.

Gain Settings

Here's an example of that.
Something as simple as gain setting can make a difference here.
In this case, if you start, you're measuring a waveform here at 70 centimeters per second, if you turn down the gain, you can obscure the higher frequency components of that returning signal, and you will get an artifactually low velocity measurement of 55 centimeters per second.
Not a tremendous amount, but something worth noting.
So under gaining the Doppler signal is a potential hazard.

Well, you can under gain the pulse doppler, but you can under gain the color doppler as well.
So here is a color Doppler signal that I purposely turned the gain down.
And you can see there's gaps in the information here.
I turned it way too high.
And you can see that now we have bled out from the vessel walls and just like the three bears, this one is just right.

So you can adjust the Doppler gains to better or worse effect.
Here's another example of that.
This is the Doppler gain at 44%, 65, and then at a hundred percent.
And you can see that the only one of those that is most useful to you would be here in the middle.
What I tend to do as far as Doppler gain is concerned, is that if I have any doubts about the settings of it, I will turn it all the way up until I get all those flurries, all that flash, and then back it off until I get something that looks reasonable.
So that's typically the way I think a lot of sonographers will do it.

If you over gain on the color doppler, of course, you can overwrite some real pathology.
So here's an example of a carotid you over gain, and you're gonna step all over this plaque because as I will talk about later, color doppler is a bi stable measurement, and you will overwrite the boundary in color.
That's not true of powered doppler or b flow for that matter, but those are different types of ways that the doppler is signal is processed.

All right, so here's another example of spectral gain changes, but this one perhaps more dramatic.
This is obviously very under regained, and we're getting measurements in the range of about 50 turn up the gain, and you can get the true signal and it's 129.
And so maybe a bit artifactual in this case.
But you can cross boundaries on our carotid table.
So this one here would clearly be within a normal range.
129 starts to get you in the range where you are worried about stenosis.
So there is issues with that.

Center Stream Sampling

Center stream sampling, of course, when you do these things, you should be sampling the center stream.
And if you put your cursor too close to the wall, you're gonna get an artifactually low velocity.
Not to mention that you'll probably get bizarre waveforms.
And the reason for that is that with laminar flow, the philosophy of the blood cells is lower at the wall.
And so you will get something quite abnormal in that location.
So to remind you to the center stream is a real issue.

Here's an example where I included not only the center stream, but I opened up the range gate to such an extent that I included both walls.
And so you get both wall motion in this case, and you also get those slower velocities.
And what you get is something that I guess superficially could be seen as spectral broadening.
There's a filling in of that spectral window underneath the envelope of peak velocities.
And so no one would do that, of course, in real life.
But it just illustrates the point that the keeping the cursor in the center of the stream is important.

Wall Filter

You can over wall filter.
This is something I think I got from Leslie.
It's you can see that if here the wall filter is a filter that is supposed to eliminate the low frequencies of the motion of the vessel wall itself here is fairly typical setting here.
But as you rev it as you ratcheted it up, you can see that you can overwrite or suppress, I should say, the signals of these lower frequency blood cells.
So wall filter is not something you probably fiddle with too much, but it is worth the knowing that if you have a big blank space here near the baseline, it could be a wall filter issue.

Doppler Angle

Doppler angle we all know that we need to stay below 60 degrees.
This is this has been driven into us by Ed Grant.
I think this is actually his graft here, which illustrates that don't look me on that one.
That's not you.
In any case, you can see here at 60 degrees, the variability goes up.
The measurement error goes up rapidly.
So it behooves us to stay less than 60 degrees.

Here's an example of that where I've appropriately angle corrected and I get something like 143, and then just even if you I dunno if you can see that cursor, but it's not much off, and you're getting a value of 198.
So even small changes in the angle correction can lead to fairly significant errors.
And here's kind of an illustration of that.
Here is intonation from one side or the other at appropriate angles, and you get something really the mere image, the inverse of that.
And here it is at 90 degrees and you get this sort of diamond shaped thing.
And I'm amazed how many of my residents don't really recognize that when they get that diamond shaped thing waveform that they've got an angle problem.

So if you see this or if your trainees see it just reinforce the fact that it's not a reliable tracing.
You can't measure it reliably.
It's an angle issue and something needs to be adjusted.

So just for fun if you do no angle correction, what can you get?
Well, look at here, we're getting velocities in excess of a thousand centimeters per second, which would be really cool if it was an actual value.
But if you angle correct, then you see that really it's something in the neighborhood of 120 or so.
So angle correction really does matter.
If you ever read a paper that talks about doppler or listen to a talk that talks about doppler and they do not angle correct, you cannot trust those values.
And believe me, I have seen papers are presentations by very esteemed people who haven't done angle correction and are making claims for velocity numbers.
So don't fall for that one.

At 90 degrees, that is the absolute worst way to try to interrogate a vessel.
Unfortunately, the portal vein always seems to be at 90 degrees no matter where you go.
And so you can get the what looks to be like portal vein thrombosis, and it may or may not be thrombose.
As an aside, I had this problem fairly recently, and since the patient had an iv, I gave them contrast and it worked beautifully, you know, so even though I couldn't get a better angle, the contrast kind of saved me.
I couldn't get accurate velocities, I didn't really care about that.
What I cared about was whether this thing was thrombo or not.
And the contrast showed unequivocally that it was not.

Aliasing

Alias sine is an under sampling error.
What to say about this is that you may not be aware, or maybe you are, that the pulse repetition frequency is the crucial factor that determines whether something aliases or not.
And the way that people well, because pulse repetition frequency is in fact the sampling rate.

So this your sampling rate must be it must be greater than the nyquist limit.
The nyquist limit is represented here.
It's on your every doppler image here It is one centimeter per one meter per second.
In this case a ine occurs above the nyquist limit.

So what do we do automatically when we see aliasing?
That is, you have this waveform that comes up and it's truncated, and then it's finished from below and comes up.
It's this is distinctly different than reversal flow, which of course, the base of that would be at the baseline.
Here we are a truncated peak velocity.

Well, automatically what we do is start to change the velocity scale.
What you may not be aware of is that when you're changing that velocity scale, you are changing the pulse repetition frequency.
That is in fact what you're doing.
It's not simply changing a scale up and down.

So to illustrate that point, here's the velocity scale as it was first set, the pulse repetition frequencies at 1221.
So we adjusted the scale here and the pulse repetition frequency was increased to 24 41.
But those numbers aren't important.
What's important to recognize is that every adjustment of your velocity scale is in fact an adjustment of your pulse repetition frequency.
And that's how that is solved.
You increase that rate of sampling to decrease the aliasing.

The other thing you can do, of course, is you can adjust the baseline.
In that circumstance, you're just assigning more of the velocities to the forward flow direction.
It's a simple reassignment rather than a change in the sampling rate.
And the third thing that you can do is you can use a lower ultrasound frequency, which essentially if you use lower frequencies, you're decreasing the higher frequency component of the signal, which is aliasing.
So those are three things you can do.
Usually we don't do more than the first two.

Here's an example of alyne in color.
What's true in spectral is also true in color doppler.
This is sort of almost a barbershop where this is alyne so much, it's going from blue to red to blue to red to blue to red.
And I included it 'cause I thought it looked pretty neat.
But you can see why we're aliasing because the nyquist limit, which is demonstrated here is eight centimeters per second.
It's very, very low.
So I'm gonna alias like crazy and which is what I did on purpose.

And here's an example.
This is the flow on a pulse doppler.
And here it is when I've adjusted the ni aquis limit to 18, at which point it no longer aliases and you get a solid color.

As all of you know, lysine is not necessarily a bad thing.
Aliasing can be useful to identify areas of stenosis like in this tip shunt.
And in fact, I just learned this morning that the way you find the ductus osis, according to Anne Kennedy, is you go look for that area of lysine and go right there.
So lysine is your friend at times.
And so it is an artifact, one that you need to know how to deal with, but also know how to use spectral broadening.

Spectral Broadening

There's a lot of reasons why you might see spectral broadening and in a normal laminar flow, in the ideal circumstance, you only get this narrow envelope of peaks velocities here, and you get very little of the other.
So that's center stream sampling.
It's appropriate size of the range gate, and it's in a vessel that's pretty straight where you get this normal parabolic laminar flow.

Here's a this is not artery, but this is a vein, but I thought it's cool 'cause you could actually see that or something very much like that parabolic flow here in this vein.

Spectral broadening or non laminar flow is often seen in areas of tortuous vessels and also in small vessels.
So that's a very common finding in those circumstances, you might even say a normal finding in those circumstances.
And of course, it's also seen in areas where there is tur disturbed flow or stenosis.
So something causing the broadening of the number of velocities, like a like an area of disturbed flow.

Now, some people will say, oh, that's an area of turbulence, which is fine by me if you call it turbulence.
But there's gonna be some penant some time in your life that will tell you that no turbulence has to have a certain rentals number.
And this is more appropriately called disturb flow.
But you've been warned, if you fall into the hands of a penant, you will know what to say.

Pitfalls, high gain settings, wall motion.
If you start sampling the entire length of the lumen, or you start to pick up part of the wall, you will know that you'll get spectral broadening.
And that in that case, it's because you didn't set things up right.
And it's normal where a vessel abruptly changes caliber.

Here's an example of that as you all probably know velocities as they go around a steep angle tend to slow in the inside of the curve and speed up on the outside of the curve.
Here's an example of that.
If you do those measurements, this was 50 centimeters per second here on the inside of the curve, and 70 centimeters per second on the outside.
And that's just what how things happen at abrupt curves or tortuous vessels.

Here's an example of a very tortuous vessel, and as many of you have already dealt with, it's difficult to do a appropriate angle doppler angle correction for this when it's so tortuous, and that's just life.

Helical Flow

There is a phenomenon known as helical flow.
This actually happens quite a bit and it's really pretty interesting when you see it.
Helical flow is just what it sounds like.
It's kind of a spiraling of the flow.
So it has, instead of going in lam laminar character through the vessel, it will actually spin and also move forward.
So like this, so here's it's illustrated.

You'll see this helical flow and it's often seen around bifurcations, and this is what it looks like in real life.
So you if you're about 90 degrees, since there are now two vectors that flow, the there's a vector of flow that goes down the vessel, but it's also one that's spinning.
So if you go 90 degrees, you can see it actually spin.
It looks like it's spinning here.
And if you angle your transducer to correct the angle so that the flow down the vessel is the dominant vector, you becomes a solid.

The thing about helical flow, of course, is that you don't want to mistake it for a dissection because I mean, that looks for all the world like a pretty good dissection to me anyway.

Interestingly, if you turn on that lengthwise and you're staying in the 90 degree incidents, you could go from side to side and you'll still see this red on one side, blue on another.
So how do you distinguish it from a dissection?
Well, it would be the rare dissect dissection where blood flow is going in opposite directions of the of the membrane of the at.

The other thing is, is that you tend to see the separating membrane, the dissected intima, but also you'll see very radically different flow profiles on either side of that.
So here's a true dissection. Here it is.
You can see there's a separation there, but it looks very much different than helical flow, where you're actually seeing flow that looks like it's going in opposite directions.
This flow is going in the same direction.
And here's another true dissection here it is the false lumen.
And if you sample that, you get a flow profile that looks nothing like it should in the ICA.
So whereas with helical flow, it looks pretty normal boundary layer separation.

Boundary Layer Separation

Remember as I spoke of, there is slower flow at the edges of vessels.
So if you come to a bifurcation you can get areas around area either a bifurcation or an area where the vessel increases size, like the bulb, you can get actual eddy currents there.
So the slower peripheral flow will actually separate in the sense that it will have a different flow character.

Bill Middleton, who we heard from yesterday, describe this phenomenon many years ago at the bulb where he noticed that at the bulb where the vessel now is more capacious you get an a reversal flow at the boundary of that flow because the slow flow has gotten slower, in fact, in turns around.
And he even proposed that if you look at a bulb and there is no flow reversal at the edges, that maybe that indicates disease, that probably is overselling it a bit in my view.
But suffice it to say if you do sample that area, which you would never do on purpose, but if you did do it on purpose like I did here, you'll get this thing that looks nothing normal.

These ed currents, you know, if you're you can get really kind of creative with it, and here you can see that it's going forward and then back and forward.
And so you'll often see these sort of things around ulcerations and bifurcations behind stenosis and that sort of thing, vessel wall discrimination.

Vessel Wall Discrimination

There has been some enthusiasm and it's kind of waxed and waned from actually measuring the lumen and determining degree of stenosis based on those sort of luminal measurements.
I guess most people probably don't do that much anymore, but there has been some brief episodes of enthusiasm for it.
And but measuring the lumen is not an altogether trivial matter for a couple reasons.

One the criteria by which the early angiographers determined degree of stenosis has to do with looking at the stenosis lumen and a downstream normal lumen, which of course goes against every instinct in our body because we can see where the vessel is right at the area of stenosis, and we want to do it differently.
But also the depiction of that boundary of the vessel is influenced very much by what technique you use.

So you can we already talked about wall filter as something that can suppress the signal at the at the at the edges of the vessel.
So if you turn up your color wall filter, you're going to filter out the lower frequency components of the returning signal, and you're gonna get this sort of black space here, and that's gonna under measure, underestimate the true lumen diameter.

Well, it's also true with which type you use.
So gray scale power or doppler.

So power doppler is a continuous boundary discriminator, which by which I mean that the signal intensity decreases as you but decreases progressively as you reach the edge of the vessel.
Color doppler is very different than that.
Color doppler is what they call bi stable.
And by that I mean that if that voxel occurs at the edge of a vessel, and it will represent the average flow in that that pixel or voxel, depending on how you think about it.
And like here, if it was that big, it will color everything, which means that it will color over the wall.
It will cover, it will overestimate, it will necessarily overestimate the lum, it will diameter.

Now we did a study on that.
We looked at color, power, and gray scale when looked at the differences and what the aluminum, the lumen measurement would be.
And you can see it's it might you might be scoffing in your seats.
This looks like statistically significant, but clinically insignificant.
And you would be right, of course.
But nonetheless, it's a real phenomenon.
And color will overwrite the wall of the vessel, even if it's by only a couple, three millimeters.

So that's I'm gonna end there.
Air.