Quantifying AS w/o Velocity Information (5 min)

AUDIO TRANSCRIPT: Yesterday, someone asked me, what’s the value of tracing more than just end-diastole and end-systole? It’s a great question. It turns out there are several unique advantages to tracing all the cardiac phases, and this case illustrates one of them.  I’m going to go ahead and kick off the automatic segmentation by clicking on this lightning bolt.

This is a patient who has aortic stenosis that they were having a hard time quantifying on echocardiography. MRI turns out to be a wonderful test for quantifying aortic stenosis because it’s totally not invasive, doesn’t require the use of contrast, and there’s no risk of stroke like there is with cardiac catheterization.

Here are the results of the automatic segmentation. You can see the segmentation is good. I’m going to go ahead and now focus on the long axis image here. I’m going to draw the basal line, which shows where the left ventricle ends. You can see we’re getting a stroke volume of about 75 ml. Sorry, 77 ml. When I compare that to the flow, the pulmonary artery flow was also about 77 ml. Aortic flow was about 75 ml. So we have good agreement between the function and the flow.

Now let’s take a look at the valve morphology. If I come over here, you can see some serial images through the aortic valve. You can see it’s a tricuspid valve. You can see some particular images over here on the right with cross reference lines. I’m going to pause where the valve is most open and I’m going to choose a level where we’re at the leaflet tips, where the valve is most narrowed. I’m now going to go ahead and planimeter the valve by including the area that is bright white and excluding the black. The black includes a lot of calcified material. We don’t want to include that in the valve area. So I’m tracing right along the black. If there’s an area where there is no black, we just connect those areas and we can touch it up a little bit if we want to.  We’re getting a valve area of 1.0 cm2.

What you can see here now is the value of having traced all phases and all slices. We now have a pressure gradient across the valve versus time. Now I want to calculate a peak and a mean gradient. So I’m going to include only the first one, two, three, four, five, six, seven points. Let me go ahead and adjust the time interval here to include the first seven points. What we can see here is we’re now getting a peak pressure gradient of 70 mm Hg and a mean pressure gradient of 31 mm Hg. And, here’s that pressure gradient again, right there.

Now some of you may be asking, how did we do this? How do we calculate the pressure gradient without any velocity information? And you can see that right here. There are equations that are well known and used by the Cath lab community to calculate valve areas. It turns out with MRI, since we can measure the valve area so nicely with planimetry, we can use these equations in reverse, if you will, to calculate what the instantaneous pressure gradient is across the valve. And so what we do is we can get the valve area measurement from planimetry, we can get the cardiac output from the multi-phase data, measuring the left ventricular endocardial contours in all cardiac phases. And, then from that we can calculate a gradient across the valve.

One limitation I should mention is that these equations work provided there isn’t substantial mitral regurgitation or a ventricular septal defect that would artificially elevate the left ventricular stroke volume.

Now some of you may be saying, well, I do that with velocity information from phase contrast images. And that’s true. We sometimes do that also. But, let me show you what it looked like in this particular case. First of all, you can see that we had to acquire a number of slices through that aortic valve. And when we finally did, if you look here on the magnitude and the phase image, you’ll see that we don’t have any good signal because of so much intravoxel dephasing from the severity of the aortic stenosis. We can’t get an accurate velocity in this particular case and therefore we have no way of getting a pressure gradient directly from the velocity data.

To summarize, we can actually calculate the pressure gradient across the valve without spending the time to acquire all these phase contrast images through the valve. And we can do it by using the short axis images, which we acquire anyway, and just segmenting the left ventricular endocardium on all cardiac phases. Thank you.

Back to Videos