Transcript
Well, I wanna thank all of you for sticking around for the afternoon sessions today, and most importantly, to Brainlab and the organizers for inviting me to present some of our results that we've been working on for some time.
So, when we first started with the ExacTrac, we were concerned about the accuracy and precision of the system. And we did some testing and we were pleasantly surprised with our results. And then when we started moving into the gating domain, we said, "Well, what happens if the target is moving? What are the consequences to the accuracy and precision of the system?" So, that's the goal of my talk.
So, as you all know, Novalis IGRT system is basically a 6D Patient Positioning system. And it consists of infrared markers and stereoscopic kV X-ray images. And they both add up to give precise positioning of the patient. It also uses image fusion that can be either bony fusion, or it can be based on implanted markers in the target or in the tumor. And finally, this ExacTrac can be used with either gated treatments or ungated treatments, as, you know, a lot of you have used for the ungated treatments.
So, here's a picture of the ExacTrac system. You're all familiar with that. There are two kV X-ray sources on the floor and the image receptors in the ceiling. And they give high-contrast resolution and spatial resolution for the images that are taken. So, our goal, as I said earlier, was to evaluate the accuracy and precision of the system. And we also wanted to evaluate or study the dependence on breathing patterns, for example.
So, the methods that we used, the materials that we used is the Brainlab respiratory motion phantom, as you see here. And this is a working image of this phantom, and I wanna thank Dr. Wurm and his colleague, to lend me the slide. And this phantom is really quite versatile. It allows you to do a lot of things. For example, we can put a film in it. There's a fiducial markers embedded in it. There's also a small tungsten ball that you can use to do Winston Lutz test. There's also a position to place iron chamber to measure the exact dose that has been delivered in the moving phantom.
And then you can, of course, get on different windows and levels and get the desired, you know, point that you wanna measure your dose on. And obviously, some of you may have used this, or I know you're quite pleased with this.
The other phantom that I used for my results is the Quasar phantom that is manufactured by a company in Canada called the Modus Medical Devices. And it comes equipped with, basically, it comes equipped with a film insert that is shown here. And you have to put a fiducial marker at the end because, as we'll see in our later work, that this fusion is basically based on fusing the implanted markers. So, you have to have a good implanted marker set and the film is placed right here. And then, well, we'll see.
So, the respiratory motion phantom and the Brainlab motion phantoms are the two phantoms that are used for the study, and the system accuracy was evaluated via end-to-end hidden target test, okay. And basically, the hidden target test is following the phantom through the chain of CT scanning, planning, and the IGRT delivery on a hidden target inside the phantom. So, we basically used protocols that are conventionally used. A 2-millimeter diameter "hidden" marker was placed on the film cassette. We used the Gafchromic film, 1-millimeter CT scan spacing. We used a 10-millimeter cone static field, right? And we measured the difference between the centers of the irradiated field and the marker location to find the error in our setup.
And we repeated this in both static and dynamic modes. And we did it multiple times to see how reproducible these results are. So, basically, for the static treatments, as you all know, you can use the ExacTrac system. And we used that to establish the baseline, what kind of results we get if we just use the static mode. And we used two methods, again, there. We used the hidden target test to get the accuracy of the system and then we also tried doing some simple QA testing, like moving a known amount of the phantom and then trying to find out how much the...what's the result calculated by the system.
And, in the dynamic mode, we had basically an amplitude of 1 centimeter. We didn't play too much with that. And we used three different treatment geometries. We used the gated phantom setup and the gated treatment. Now, that's the true gated treatment where you're setting up the patient in a gated mode, and you're also treating the patient in a gated. And clinically, this mimics the gated beam delivery for small target margins, right. So, that's the state of the art.
And then the second method we used is the gated phantom setup, but free-breathing. Basically, ungated treatments, all right. So we wanted to see how the system responds under this condition. Clinically, this corresponds to improve patient setup, but you're using larger target margins because you're doing free-breathing. So, this is kind of in-between. And then finally, the worst-case scenario is when you have an ungated setup and an ungated delivery. So, you're not using gating at all. You're just trying to match your X-rays and DRRs, which may have been taken in different phases, right. And so, this protocol is used in clinics that do not use any gating.
So, all fusion was based on implanted markers, as is the right way to go. Except for one case, and we'll talk about that. And we studied the system accuracy dependent on the frequency of breathing cycle. We went from 1.5 seconds per cycle to all the way to 12 seconds per cycle. So, basically, it's very slow and fast breathing patterns. We also looked at the gate level from 0% to 50%, and the window from 10% to 30%.
And basically, after we did the hidden target tasks, we looked at the film, and we scanned it at high resolution with the RIT Film Scanning Software. And we looked at the difference between the marker position and the center of the radiated field to get the error, right.
So, here's a picture of the four different geometries that we used. Here's the static geometry. The motion is supposed to be in this direction. Here is the gated treatment and gated setup, right. And so you can see there is some diffusion in the, or diffuseness in the motion direction, and then you have a gated setup but ungated delivery, right. And because of that, you have tremendously large margins. And finally, you have ungated setup and ungated delivery, right.
So, our procedure is basically scanning this field along the longitudinal direction, and finding out the differences between the peaks. And that'll give us the error.
So, for the phantom in the static mode, not surprisingly, many people have done the studies, and the number is about 0.3 millimeters or about that, and our numbers are very consistent. We got 0.22 millimeters plus or minus 0.11 millimeters for the 10 times that we repeated this static test. And the maximum was 0.44, so very reproducible. Gave us confidence in what we were doing at least was, you know, good compared to some of the other investigators.
So here are the results, right? Right, so the mean value is 0.22 and the standard deviation is 0.11. Here I'm plotting the error in millimeters and these are the repeat measurements, right? So, if I go to the gated setup and the gated delivery, all right, so the nice thing is that the system accuracy does not degrade significantly. We were at 0.22 and now we are at 0.14 but there's a larger standard deviation than before. So, the good news is even in the gated delivery and, of course the gated setup, the system is performing quite well, all right. So those are the results that we got.
There's a little typo here, when we go to gated setup and ungated treatment, all right. So, the treatment is ungated, but you're using gated setup. There's a small typo here, this should be 0.67. So, the mean value for the setup accuracy, or the treatment accuracy is degraded now. However, the overall is still well within acceptable limit, I think. So, even for a gated setup and ungated treatments, you're not doing too shabby if you use the ExacTrac system.
But if you go to ungated setup and ungated delivery, you run into a lot of problems, all right? One of the problems you run into is the fact that your two ExacTrac X-rays are taken at two different phases of the breathing cycle which may be completely inconsistent with how the CT scan was done, right. And it's almost difficult, if not impossible, to fuse them. And that's why we ran into a lot of issues.
So, we could not really fuse the DRRs to the X-rays without doing some manipulations of the system. And the way we were able to do this is by using bony landmarks or the bony fusion, and that's what I was talking about earlier that there's one case you cannot use markers all the time. And when that happens, clearly, the mean accuracy of the system is degraded, right? Because you basically are not matching apples to apples, so to speak. And those results are far, far worse. So, we also love that.
Finally, what happens when we change the speed of motion or the breathing pattern is changed, right? So, here is the low speed and here's the high speed. This is kind of, like, a 5 cycles per second, and here is approximately 30 cycles per second. So, you see that for the gated setup and gated treatment, we are just examining what happens to the setup error if we do that.
And you can look at, looking at this graph, there is no pattern that emerges that tells you that there's any significant dependence on that. So that gives us confidence that for the clinically relevant breathing patterns, the ExacTrac system is responding quite well if you use the gated deliveries.
Here's just an example of the fusion that we did on our phantom. Here are the markers placed in the end, all right. So we found that for ungated setups, basically reiterating I said earlier, it's sometimes impossible to fuse the internal marker so you have to find the best compromise manual fusion and that may not always be the most relevant or most, the correct fusion to accept. So you have to be really careful. And if you're doing any gated deliveries, you must use some kind of implanted marker.
We did not find any systematic dependence, as I said earlier, on any of the breathing patterns like the frequency or the window level, or the gate. I had some data that I just don't have time to show you on that.
In conclusion, ExacTrac system uses kV images with high spatial and contrast resolution, right? It's very quick, simple, and robust operation. And Novalis IGRT provides highly precise and accurate method of SBRT delivery. For treatments of mobile targets, we strongly recommend doing some kind of gating with internal markers, not just the bony fusion, because the ungated treatments, all treatments involving bony fusion may completely miss the target. Thank you very much.
So, when we first started with the ExacTrac, we were concerned about the accuracy and precision of the system. And we did some testing and we were pleasantly surprised with our results. And then when we started moving into the gating domain, we said, "Well, what happens if the target is moving? What are the consequences to the accuracy and precision of the system?" So, that's the goal of my talk.
So, as you all know, Novalis IGRT system is basically a 6D Patient Positioning system. And it consists of infrared markers and stereoscopic kV X-ray images. And they both add up to give precise positioning of the patient. It also uses image fusion that can be either bony fusion, or it can be based on implanted markers in the target or in the tumor. And finally, this ExacTrac can be used with either gated treatments or ungated treatments, as, you know, a lot of you have used for the ungated treatments.
So, here's a picture of the ExacTrac system. You're all familiar with that. There are two kV X-ray sources on the floor and the image receptors in the ceiling. And they give high-contrast resolution and spatial resolution for the images that are taken. So, our goal, as I said earlier, was to evaluate the accuracy and precision of the system. And we also wanted to evaluate or study the dependence on breathing patterns, for example.
So, the methods that we used, the materials that we used is the Brainlab respiratory motion phantom, as you see here. And this is a working image of this phantom, and I wanna thank Dr. Wurm and his colleague, to lend me the slide. And this phantom is really quite versatile. It allows you to do a lot of things. For example, we can put a film in it. There's a fiducial markers embedded in it. There's also a small tungsten ball that you can use to do Winston Lutz test. There's also a position to place iron chamber to measure the exact dose that has been delivered in the moving phantom.
And then you can, of course, get on different windows and levels and get the desired, you know, point that you wanna measure your dose on. And obviously, some of you may have used this, or I know you're quite pleased with this.
The other phantom that I used for my results is the Quasar phantom that is manufactured by a company in Canada called the Modus Medical Devices. And it comes equipped with, basically, it comes equipped with a film insert that is shown here. And you have to put a fiducial marker at the end because, as we'll see in our later work, that this fusion is basically based on fusing the implanted markers. So, you have to have a good implanted marker set and the film is placed right here. And then, well, we'll see.
So, the respiratory motion phantom and the Brainlab motion phantoms are the two phantoms that are used for the study, and the system accuracy was evaluated via end-to-end hidden target test, okay. And basically, the hidden target test is following the phantom through the chain of CT scanning, planning, and the IGRT delivery on a hidden target inside the phantom. So, we basically used protocols that are conventionally used. A 2-millimeter diameter "hidden" marker was placed on the film cassette. We used the Gafchromic film, 1-millimeter CT scan spacing. We used a 10-millimeter cone static field, right? And we measured the difference between the centers of the irradiated field and the marker location to find the error in our setup.
And we repeated this in both static and dynamic modes. And we did it multiple times to see how reproducible these results are. So, basically, for the static treatments, as you all know, you can use the ExacTrac system. And we used that to establish the baseline, what kind of results we get if we just use the static mode. And we used two methods, again, there. We used the hidden target test to get the accuracy of the system and then we also tried doing some simple QA testing, like moving a known amount of the phantom and then trying to find out how much the...what's the result calculated by the system.
And, in the dynamic mode, we had basically an amplitude of 1 centimeter. We didn't play too much with that. And we used three different treatment geometries. We used the gated phantom setup and the gated treatment. Now, that's the true gated treatment where you're setting up the patient in a gated mode, and you're also treating the patient in a gated. And clinically, this mimics the gated beam delivery for small target margins, right. So, that's the state of the art.
And then the second method we used is the gated phantom setup, but free-breathing. Basically, ungated treatments, all right. So we wanted to see how the system responds under this condition. Clinically, this corresponds to improve patient setup, but you're using larger target margins because you're doing free-breathing. So, this is kind of in-between. And then finally, the worst-case scenario is when you have an ungated setup and an ungated delivery. So, you're not using gating at all. You're just trying to match your X-rays and DRRs, which may have been taken in different phases, right. And so, this protocol is used in clinics that do not use any gating.
So, all fusion was based on implanted markers, as is the right way to go. Except for one case, and we'll talk about that. And we studied the system accuracy dependent on the frequency of breathing cycle. We went from 1.5 seconds per cycle to all the way to 12 seconds per cycle. So, basically, it's very slow and fast breathing patterns. We also looked at the gate level from 0% to 50%, and the window from 10% to 30%.
And basically, after we did the hidden target tasks, we looked at the film, and we scanned it at high resolution with the RIT Film Scanning Software. And we looked at the difference between the marker position and the center of the radiated field to get the error, right.
So, here's a picture of the four different geometries that we used. Here's the static geometry. The motion is supposed to be in this direction. Here is the gated treatment and gated setup, right. And so you can see there is some diffusion in the, or diffuseness in the motion direction, and then you have a gated setup but ungated delivery, right. And because of that, you have tremendously large margins. And finally, you have ungated setup and ungated delivery, right.
So, our procedure is basically scanning this field along the longitudinal direction, and finding out the differences between the peaks. And that'll give us the error.
So, for the phantom in the static mode, not surprisingly, many people have done the studies, and the number is about 0.3 millimeters or about that, and our numbers are very consistent. We got 0.22 millimeters plus or minus 0.11 millimeters for the 10 times that we repeated this static test. And the maximum was 0.44, so very reproducible. Gave us confidence in what we were doing at least was, you know, good compared to some of the other investigators.
So here are the results, right? Right, so the mean value is 0.22 and the standard deviation is 0.11. Here I'm plotting the error in millimeters and these are the repeat measurements, right? So, if I go to the gated setup and the gated delivery, all right, so the nice thing is that the system accuracy does not degrade significantly. We were at 0.22 and now we are at 0.14 but there's a larger standard deviation than before. So, the good news is even in the gated delivery and, of course the gated setup, the system is performing quite well, all right. So those are the results that we got.
There's a little typo here, when we go to gated setup and ungated treatment, all right. So, the treatment is ungated, but you're using gated setup. There's a small typo here, this should be 0.67. So, the mean value for the setup accuracy, or the treatment accuracy is degraded now. However, the overall is still well within acceptable limit, I think. So, even for a gated setup and ungated treatments, you're not doing too shabby if you use the ExacTrac system.
But if you go to ungated setup and ungated delivery, you run into a lot of problems, all right? One of the problems you run into is the fact that your two ExacTrac X-rays are taken at two different phases of the breathing cycle which may be completely inconsistent with how the CT scan was done, right. And it's almost difficult, if not impossible, to fuse them. And that's why we ran into a lot of issues.
So, we could not really fuse the DRRs to the X-rays without doing some manipulations of the system. And the way we were able to do this is by using bony landmarks or the bony fusion, and that's what I was talking about earlier that there's one case you cannot use markers all the time. And when that happens, clearly, the mean accuracy of the system is degraded, right? Because you basically are not matching apples to apples, so to speak. And those results are far, far worse. So, we also love that.
Finally, what happens when we change the speed of motion or the breathing pattern is changed, right? So, here is the low speed and here's the high speed. This is kind of, like, a 5 cycles per second, and here is approximately 30 cycles per second. So, you see that for the gated setup and gated treatment, we are just examining what happens to the setup error if we do that.
And you can look at, looking at this graph, there is no pattern that emerges that tells you that there's any significant dependence on that. So that gives us confidence that for the clinically relevant breathing patterns, the ExacTrac system is responding quite well if you use the gated deliveries.
Here's just an example of the fusion that we did on our phantom. Here are the markers placed in the end, all right. So we found that for ungated setups, basically reiterating I said earlier, it's sometimes impossible to fuse the internal marker so you have to find the best compromise manual fusion and that may not always be the most relevant or most, the correct fusion to accept. So you have to be really careful. And if you're doing any gated deliveries, you must use some kind of implanted marker.
We did not find any systematic dependence, as I said earlier, on any of the breathing patterns like the frequency or the window level, or the gate. I had some data that I just don't have time to show you on that.
In conclusion, ExacTrac system uses kV images with high spatial and contrast resolution, right? It's very quick, simple, and robust operation. And Novalis IGRT provides highly precise and accurate method of SBRT delivery. For treatments of mobile targets, we strongly recommend doing some kind of gating with internal markers, not just the bony fusion, because the ungated treatments, all treatments involving bony fusion may completely miss the target. Thank you very much.