Clinical Treatment and Process Quality Improvement Due to Cherenkov Imaging

So I’ll be speaking on clinical treatment and process quality improvement due to Cherenkov imaging. And I thought I’d lighten it up a little bit by quoting Hank Williams Sr., who wrote “I Saw the Light,” and it was released in 1948. Seventy-seven years later, a very little-known physicist wrote another verse to that tune, and this is the world premiere of that verse. So it’s:

“I saw the light, Cherenkov light. There’s no more questions if the beam’s too wide or tight. We’re all so happy when we treat our patients right. So praise the Lord, I saw the light.”

Well, thank you. You’re too kind. So, this is my band and a couple of important people here. If you’re looking at clinical implementation, you want to speak with Rory Rosselot at Dose Optics. She’s probably seen more Cherenkov images than anyone on the Earth and really is an expert on practical implementation, getting things going.

The next most important people are the graduate students. And I just heard this last talk by Josh. It was awesome, about beam sight or beam guide—I forget already what we call it. And so the question is, why would you need to look at the Cherenkov images at all? And the answer is my newest student, Bomi Lee. She’s a sophomore right now. And we’re working on a method to get actual dose from the patient, which has always been a confounding factor because of tissue optical properties, including blood content, melanin content, and that sort of thing. But we think we’ve got our hands wrapped around that right now.

So, I’ll get to physics now. Cherenkov light is part of the megavoltage radiation dose. When X-rays or electrons are impinging on tissue at faster than the speed of light, they polarize that tissue and the depolarization event results in emission of visible light. So here’s a nice example from nuclear fusion in a fuel cell. You see the nice blue glow. For those of you who may have caught me breaking a law of nature or physics, no, I didn’t. $3 \times 10^8$ meters per second is a limit in a vacuum, but we’re in tissue or medium, and so light obviously is scattering and going slower in that tissue. Electrons, however, can still go faster than that speed of light in the tissue.

Monte Carlo calculations show us that the Cherenkov emission should be absolutely proportional to dose. As I’ve already mentioned, people get in the way of everything good and pure, and so tissue optical properties keep us from having a linear response. However, we can irradiate a phantom, and this is a phantom. It’ll show back up at the end of the talk. Rongxiao Zhang put that together. He was the first graduate student to observe Cherenkov emission from a clinical beam in a water tank, and he put this cute little phantom together.

So you all know that the trick is done with cameras in addition to the Vision RT cameras mounted on the ceiling. The actual trick is that our linear accelerators are delivering irradiation in pulses, right? So we’ve got these three or four microsecond pulses that are spaced every ten milliseconds apart. And what we do is use this pulse of irradiation to trigger the image intensifier on the camera, so it’s only acquiring data when there’s actual data to be acquired.

So an example of a live image, you see the beam changing shape. You also see superficial vasculature in that first frame, and maybe it’ll pop up again. But that vasculature can be used to fingerprint a patient, if you will, but it’s also indicative of these tissue optical properties that can get in our way of thinking just about dose proper. Cameras are mounted on the ceiling. These are two of our vaults. You’ll notice probably some extra pods up there because we’ve done some research as well.

Another part of the trick, as you know, is that there’s structured light which is projected by the Vision RT pods, and that same signal that’s used to trigger the image intensifier is also used to mute the projector out of the pods so that we don’t have light interfering with the process. This is just an indication of a patient being treated, and you can see in real time that the field size will change with beam modulation. And at the end, we get this summated image where the bright yellow is higher dose levels and the blue on the outside is lower dose out at the beam penumbra. Here you can clearly see that tissue vasculature as well as a perceived intensity change due to the areola nipple complex.

This is a cute observation that we made. On the left is the cumulative view building up, and on the right is the real-time image. This is a stereotactic radiotherapy for a posterior orbital tumor, and what we see there is light being emitted from the patient’s lens due to transmission through the vitreous humor. So everything was proper. The location of the tumor, as you see in the lower right inset, mandated that we have some dose to the posterior orbit. But the surprise was it was actually shining out of the patient’s eye.

So four key areas where we can see improvement in radiotherapy are patient setup, treatment planning, use of accessories, and patient compliance during treatment. So we had this prospective study of the first 60 patients that Josh referenced, and first case out of the box, and what do we see but chin dose. So, we can move the patient’s chin out of the way or think about replanning the case. Case two was limb monitoring, so we’ve got a little bit of spill going off to the contralateral leg. And one can use the frog leg position for a single leg if that would be a little wiser than just letting it go.

We’ve got these setup and planning real-time discovery of issues. So here we’ve got chin dose before adjustment, and then moving the head out of the way, we can avoid that. And this turns out that it was already in the treatment plan from the beginning, but it wasn’t noticed either by the dosimetrist, the physicist, or the physicians, because people are not in the habit of turning the dose cloud down to the 10% line, right? We’re always focused on what is the treatment dose. Are we covering the tumor? But we’ve got these low-dose areas that can be important as well.

Another case study is bolus alignment. So in the upper panel, that’s a properly aligned piece of bolus, and in the lower panel, it’s too far posterior and we’re missing the medial edge. Another case where the dosimetrist was hand-adjusting the multileaf collimators and left the inferior-most set of collimators in the open position, and it created a stripe on the patient. This, again, was missed through all the treatment plan review and was picked up only during the treatment. Here’s another case of bolus misplacement. You can see light piping to the edge of the bolus, and so it becomes immediately obvious, and I just won’t belabor that.

Here we have patient compliance. They’re supposed to be holding their hand up but drops it during the beam delivery. This is an exit field from a pelvic spine treatment. Probably not much harm done with that exit dose, but nonetheless, it’s not best form. So may as well come up with a better plan of immobilizing the patient.

This is the first patient treated at a site where DoseRT was installed, and the first thing we saw was dose to the contralateral breast, and that was unexpected. Yeah, just the 3D rendering of the same thing. So, this contralateral breast dose turns out to be quite more common than we had imagined, and so we opened a trial internal to Dartmouth called the EDUCATE Trial, evaluating dose using Cherenkov and scintillation technology. This piggybacks on a WE CARE study, which was done a number of years ago by another institution. They found that dose in excess of one gray in young patients results in excess contralateral breast cancers. So this is actually a clinical finding where you want to keep the dose low in order to avoid long-term morbidity, which you probably wouldn’t expect from such a low dose.

Since that’s been published, we’ve seen a lot of changes in techniques, and so especially with the adoption of VMAT technology, and accelerated partial breast irradiation, inclusion of intramammary nodes, one can imagine that there is indeed more spill of dose to the contralateral side. So we went into this thinking that perhaps the incidence of contralateral breast dose is underappreciated.

So in the objectives—Gosh, I stole somebody’s slide, and I don’t know what the animation does, right? So the objectives were to establish what are the incidence of contralateral breast dose in our routine clinical practice. We want to quantify what that dose is using Cherenkov image-guided in vivo dosimetry. So it’s nice to think that you’re going to put a TSLD or a OSLD or a TLD on a patient surface to measure the dose at a spot of interest, but how do you know where you’re interested? So the best way to do that, we think, is using image guidance from the Cherenkov images. And then lastly, we want to determine the root cause of that contralateral breast dose. We need to distinguish between that which was planned and unplanned, and we’d like to come up with techniques that minimize that dose in the future.

So, we ran the Cherenkov imaging on all of our patients at both the core academic medical center as well as one of our community hospitals and measured contralateral breast dose when it was detected. So we imaged 129 unique patients over 1,800 fractions, reviewed those for over six months, and found that 94 of those patients were treated with supine technique. Not surprising, the majority. Contralateral breast dose was identified during delivery in 43% of those patients, and that is a surprising result that it’s that high.

The spread of these observations were that normal tangent treatments, not so bad, six out of 56. Wide tangents, 93% of those cases had contralateral breast dose included, and of course, that’s to get the IMNs when you just open up your tangent fields. Tangents with medial electrons mixed in, two of two. Not strong statistics, but yet we expect beam spread and wide penumbra from an electron beam. And then accelerated partial breast irradiation using VMAT, 100% of the cases had some kind of contralateral breast dose detected.

So it was unplanned in about 10% of the cases. These are cases where planned and unplanned, just two different examples. And a combination of planned and unplanned also occurred, so the patient on the right-hand side. And here you see the marker block that people have talked about today. It’s pretty easy to understand that. I heard a lot of questions about it, but I tried it on myself. So if you lie on the table with the block on your abdomen, you cannot breathe at all and make your abdomen go up and down, right? So belly breathers, chest breathers, people who play trumpet like to breathe from their belly, right? You need a big volume of air. So that’s not a really robust way of tracking breathing motion. Although here you see it moved up superior on the sternum, so it could do a little better job of monitoring the chest wall.

Dosimetric measurements from the three different sort of techniques, so wide tangents, tangents plus electrons, and accelerated partial breast irradiation. The TLDs were showing anywhere between, well, negligible and two gray, actually, in one fraction. And so if you add this up, the contralateral breast dose can vary between negligible in the case of the VMAT and up as high as 40 to 48 gray. If that were to occur in a younger patient, you can be almost assured that it’s going to result in a long-term toxicity. The limits that we talked about were to keep that dose low, and patients under 50 did have contralateral breast dose in excess of our constraint of one or two gray. So future work is going to be monitoring every young patient with the Cherenkov imaging and try to reduce that to zero so that we don’t see those long-term adverse events. So this, again, the animation, 17% of those younger than 50 had these issues.

So to recap, we’ve got suboptimal planning where dose can be originally planned to be hitting the contralateral side but underappreciated. We’ve seen patient motion during treatments, which we pick up with the imaging, and inconsistent bolus placement.

So in summary, we can see real-time delivery of the dose. It’s recorded so that the teams can look at it retrospectively. Images do show daily variations. Non-ideal delivery can result due to these four features that we’ve already talked about. And Dose RT, I claim integrates seamlessly with SGRT, but we should really hear from end users. The therapists at our center actually love it. We’ve installed it now in every bunker across four facilities at Dartmouth. They actually like the Cherenkov imaging monitors better than the CCTV cameras because the images are crisper, clearer, and you get the extra information. So if you want to declutter the machine, it’s my suggestion that you get rid of the normal CCTV, replace it with Dose RT.

Once in a while you can get shown the light. In the strangest of places if you look at it right. Yeah, an actual band.

 

 

*This transcript has been AI-generated. Contact us at secretary@sgrt.org if there are any issues.