Future of SGRT

So we’ll talk about the future of SGRT. I don’t know, so don’t hold me to anything I’m about to say. It’s all going to be speculation. We’ll talk about what’s happening right now in SGRT. We’ve talked about a lot of new products that are coming into the market in this field or in this area, and we’ll talk about these things from a clinical perspective, but also from a perspective of automation, integration of care, and how these things can better be utilized within the clinic. And then we’ll do some speculations and look at some pet projects that we’ve done at our facilities out in Colorado. Some labors of love from one of my physicians as well that has some implications within SGRT, and we’ll just go through that. And hopefully it just sparks some ideas in your mind to take back and have conversations with your staffs, and about how you could be using SGRT potentially in different ways within your clinic.

So again, my disclosures haven’t changed since about an hour ago, but if you want to drop bags of money on me after this, I will add you to this slide with no problem. But when they ask me to talk about the future of SGRT, one, I’m always like, “Huh, what does that mean?” But we’re going to talk about why this conversation, why now. We see a lot of innovation in this field as we’re talking about all these new exciting projects, as we were talking just now about the Cherenkov imaging, about two speakers ago, talking about how much he loves Cherenkov imaging. I am super jazzed about Cherenkov imaging; it’s super cool. If you come out to the site visits this afternoon, we’ll be showing some examples of that and what it looks like clinically on phantoms. Obviously not on people, so don’t worry, we’re not going to radiate anybody when you show up, so you’re going to be totally safe.

But I spend countless hours playing with Cherenkov imaging and trying to find ways that it’s going to be more and more applicable within our clinical setting. In fact, my wife thinks I’m dating a girl named Cherenkov right now and wants to know why she’s ruining such a happy home, because I’m never there. But it’s an addictive tool to play with because there’s so many things that we don’t see in the clinic. So why this conversation, why now? We all know that patient volumes are increasing, at least across our centers right now. Patient volumes are increasing, but staffing capacities tend to stay relatively flat until there’s significant justification. And we need additional throughput, but we don’t want to sacrifice accuracy or that touch that we heard people talk about earlier with our patients and that connection with the patients. And so SGRT is positioned to bridge that gap for everybody.

There’s technology convergences that are happening now within radiation oncology. I’m seeing more and more adaptive planning being introduced in community settings, so like community hospitals putting in Ethos, which a couple of years ago were a university style thing, which needed a lot of additional staffing and a lot of additional time at the machine, and now we’re seeing these things pop up in community cancer centers. Those things are dragging with them AI. We’re using AI for contouring, AI for evaluation of imaging, AI for all the different types of things. I know Anton and I are always talking about AI and where he’s ready to embrace his computer overlords in the future, and all the robots that are going to take over all of our jobs. I don’t think that’s going to be the case, but AI is going to be a tool that we’re going to be using in the clinic and how can we roll these things into these platforms. Protons with their new small form factor, protons that are going in standard vaults are now becoming more attractive and cheaper. And so I’ve even heard community cancer centers are looking at putting in single vault proton facilities, and making a hard go at going bankrupt. And so all of those things are out there. Sorry if there’s any proton vendors in here, I really apologize. They’re super cool too, but I don’t have one.

And then we have tattoo-free types of treatments, maskless head and necks, which I know Dee gave a talk on yesterday, and things that we’re doing as well out in Colorado in AdventHealth. These things are all patient satisfaction type of things that we’re going to provide to our patients as a measure of things they’re going to begin to expect in the future as these become more prevalent as well. So how many times if someone walked in and said, “Do you do protons here?” I anticipate in the future how many people are going to come and say, “Do I really have to wear a mask? I can go across the street and not have to wear the mask.” And so SGRT is fundamentally positioned to address all of these things in the future, and I think that it will in successful ways.

And so we’re going to talk maybe three stages, three parts to this talk, about what’s going on right now. And what’s going on right now, if you come to AdventHealth, this is later on this afternoon, we’ll talk about these workflows, sim, plan, treat, dose. We’ve heard this over and over and over. And there’s products within these spaces. So we have SimRT, MapRT in planning, AlignRT in treatment or AlignRT inborn treatment, and the Cherenkov imaging system called DoseRT at treatment as well, but visualizing dose instead of surface positional information. These are all great products. These are all things that are out there, and we’re all very excited about these things, but they are individualized products right now and things that we integrate within a workflow.

We have four established pillars of clinical use. We all know that SGRT is fantastic for initial setups. We hear about all, I think every other day, there’s some LinkedIn post where someone’s getting some glass trophy for going tattoo-less. The active motion management, we’ve heard about all of the active motion management, the drive towards that with the 2026 billing changes. But this has been a motion management tool for many, many years. I’ve been doing this for 15, 16 years at this point. And it has just always been an active motion management tool. We all know, at least from the published evidence, probably the most evidenced application of SGRT for deep inspiration breath hold and respiratory breath control. We know that it is well established. There’s tons of documentation for that, as well as the precision and accuracy of these SGRT systems for SRS and SBRT. But the nice thing about SGRT is that it’s for every single patient. So if I commission a system as a physicist for SRS, I get that SRS precision on every single patient. I don’t dust off a special set of immobilization gear to treat a patient anymore. I am giving every single patient the same SRS precision, regardless of who’s there, what they’re being treated for, or why they’re under my beam. They all get the same thing, which is a powerful tool that spans that huge patient population.

There’s new applications. So what’s going on right now is we’re seeing an explosion of invention. So we have surface-based planning. This surface-based planning is well evidenced as well. If you look at the literature, this table, I just gave an RSS talk about non-coplanar planning and looking at just publications since 2020, we see things for GBMs and for brain metastases and SRS. We see it for hippocampal sparing whole brain, head, neck, cervical, GYN cases, liver and lung SBRT, prostate, all of these things, and whether or not they’re evidenced, what is the advantage? So scale, I use check marks, you can use whatever you like, but the more check marks, the more advantage that these papers have said that non-coplanar planning brings to these types of cases. These are the references I’ll give you on the next slide in case you get this after the meeting, so you can look them up if you want. And the disadvantages are largely around the complexity of the workflow. So how we plan these people, how do we go through a workflow? We have non-coplanar treatments, and we have to worry about additional collision checks, those types of things. All of these things are being addressed by an SGRT product. And so we can, with high fidelity, set these patients up into the position that they were simmed at, and then we can predict now with MapRT whether they’re going to have a collision issue at the time of treatment. And so we can incorporate this information into our planning so that the complexity argument kind of goes away. It’s not that complex. They’re telling you the answer at the time of planning. You just have to have the faith in the tools and the understanding of how to apply the tools.

And so these are just the references from the previous page. So this is largely if you just get this afterwards. We talked about this in scripting, so if you weren’t in here, the power of surface-based planning is the fact that I can take a CT. This is the standard CT. This is all I need for dose calculation. This is what I can get with optical surface guidance, and then I can get a fusion of these information down here because everything I’m worried about colliding with in non-coplanar planning exists outside of the CT. And so if you weren’t in that talk, we showed an example of using the tools appropriately. So we have a CT, we have an optical surface, we have a fusion, and we can see that these tools can be used incorrectly as well. So there is a need for robust QA programs, an understanding of what is a procedural issue and what is an actual technical issue. If this was to happen, we cannot trust the clearance map, but we can actually use an incorrect clearance map if we don’t understand these things to put a patient in harm’s way, and we don’t want to go that direction as well.

So there’s this tool, again, something that we’re playing with. This is a caching mechanism for caching these clearance maps so that we can hunt for isocenters. So we saw a great talk on lattice or spatially fractionated treatment with these large tumors and showing the challenges that we have with clearance. But if we can actually have a tool that now optimizes the solution space and allows us to virtually hunt within the clearance space in 3D, we can actually find preferential things where this case specifically, I can actually extend this arc all the way around the patient if I just choose a better isocenter location. And so we can start using these tools in the future to actually help us to better integrate the tools that we already have, but also make our plans more effective and much quicker to implement, even though they are complicated as we would call them right now.

There’s the dose visualization. Again, my girlfriend, Cherenkov, my favorite person to hang out with in the evenings. And then now we have BeamGuide. So now we can now project into the future. As we’re sitting there setting up our patients, we can see or have that knowledge of what was intended from our physician and what was approved and what we’re looking at right now. Is that going to be executed? We don’t have to wait until after execution and then kind of triage what happened or what went wrong to make changes for tomorrow. We can make changes right now so that we can prevent the error right now. And what that looks like, this is one of my favorite cases that we’ve had at our site. This is a prone breast, and this was delivered for two days, and we see what it looked like for the first two days on the left. And then on the third fraction, we had this, and this was a kind of an aha moment for us with DoseRT of the power of what this was doing. This young lady had a bad back, had issues with her lower back, had pain, and so she was trying to bear her weight on her elbows. So instead of being really stretched out the way we wanted her to, the way she did the first two days, she was pulling her elbows in to kind of help hunch her shoulders and bear her weight on her arms rather than across her back. And in seeing this, we were able to address her pain issue and stretch her back out, and we were able to identify this very quickly. But that introduced a question of, well, could we have avoided that? Would we have known that when we set her up? Because we set her up with an ROI across her back and everything looked great. Her breast looked great. We just didn’t have an ROI on her arm. So how would we have caught that? Because the postural video kind of gave us that this was going to be okay.

But if we look at that same scenario, what happens if we see these things on day one? So this is an example that Anton gave. This is a prone breast from here in Celebration. This is also a prone breast from here in Celebration. And how do we know that these are the planned exposures? If this is our first day, we don’t have previous days of references to go off of assuming that those days went well and this day is maybe not going well. Well, if this is the first day, we can look at this and say, well, maybe she’s rolled into the hole too far. Maybe she’s rolled away from the hole too far, and these could be an underexposure and overexposure, but we have to have some sort of a priori knowledge of that plan. And we do our plan checks, and we do those things, but these are not traditional things that we’re rolling into our workflows. But now with BeamGuide, as we’re setting up the patient, we are getting the intention that was approved by the physician and seeing that, yes, in fact, this is the intended exposure for this patient for whatever clinical reason, and this is the exposure for this patient, and then the actual delivery conform to that exposure. So we’re now seeing as we have invention, we also have synergy between these tools moving into the future that we can use these tools in conjunction with one another to get answers ahead of time and apply a cheat code while we’re setting up our patients.

And then we have the AI, the advanced workflows. We have advanced workflow integration into OIS. We have the OIS module. We’ve heard a lot of talk about that, where we’re now promoting SGRT into its full capability or its full adoption into the image guidance space, rolling that into the OIS like we would any of our IGRT workflows. But we also have adaptive workflows and proton workflows and those types of things that we have to look at. So we have to look at what’s changing in this space as well. And so for those of you who do adaptive, please don’t judge me. This is the most cartoon adaptive workflow I could come up with. This is adaptive workflow, and right now you have some sort of daily imaging plan of the day. You go in, you look, this doesn’t look right. You’re going to maybe do an adaptation. You’re doing this on-table adaptive approach. You may make the decision to replan. And now the SGRT is basically a reference, and we’re watching our patients while we’re doing 30 minutes worth of work of transferring contours and those types of things. We can take this and shift this mentality of saying, “Okay, we’re going to treat yesterday’s plan. We’re going to set up to yesterday’s plan, and we’re going to do all the on-table adaptive stuff and make these decisions.” But as we move these things into high-volume community cancer centers, this might not be the way to go. We might need to set up to today’s plan with today’s surface and have different today’s tolerances. Maybe the internal targets are in a different orientation, or there’s different filling of different organs, and we might want to have tighter tolerances today. So we have to start thinking, well, how does SGRT play a role other than just eyes in the sky, they’re not moving while I’m doing adaptation, to something maybe in the future where we have a pre-treatment, maybe pre-table adaptation. So I can think of ways to use SGRT all the time. It would be awesome to have a little photo booth of SGRT, so the head and neck patient who often has to have adaptation at least a couple of times during treatment can get surface-guided 3D captures before on table, and we can let the physician know nothing’s changed since yesterday. Don’t even come to the machine. We’re going to treat the normal plan. But as you have this metric that you’re tracking every day because it’s non-ionizing, you can get that surface every day. If you get that surface every day and finally a volume trigger is tripped, then you say, “Okay, today, we would like you to come to the machine because there’s a high probability that we may have to adapt this patient today.” So you’re saving the patient additional time on the table by telling the physician, “Hey, this may be an adaptation day,” versus getting on the table and every day having to be there making the call, “Okay, just use yesterday’s plan and then go move on.” This could be a throughput thing or a workflow thing that could be added into the workflow that’s not on table, it’s pre-table. And so that’s a future adaptation for adaptive radiotherapy.

Protons, as I said, more small form protons are becoming reasonably affordable. Affordable in my wife’s sense of affordable, she saves me money. Only $26 million, but it’s not $80 million, so I saved 60-some million dollars or 50-some million dollars. But SGRT is very important for protons, and as more and more protons are out there, I think SGRT should be the standard of care for protons. That’s my opinion, I guess. I’m a bashed proselytizer for SGRT. But in this case, we all know the Bragg peak, the discrepancies, the small millimeter discrepancies that can lead to centimeters of dose discrepancy. But there are operational problems. Gantries are slow. Treatments are long. We have limited imaging on some of these gantries, so not all proton systems have a three-dimensional imaging system within the vault. They’re relying on planar imaging. And so SGRT can bring some benefits to these workflows that they just don’t have available to them because of the geometry of the room that we’re dealing with. So you can look at continuous gantry motion monitoring. You can look at patient positioning. You can look at collision mapping or clearance mapping, sorry, Thomas, wherever you are. But you also have operational problems with those. And so you have to think about the motion complexity gantry for protons is very different than a CRM Linac or than a Halcyon. You have these complex six-dimensional couches that have deterministic couch positions, but they are very heavily dependent on the starting conditions. So you have a knuckle under the couch, and that knuckle might be in different orientations depending on if you’re moving to position B from start position A or start position C. So as you move that couch, you have deterministic positions, but they’re all nodal dependent. So those models are a little bit more complex than what we’re seeing in MapRT right now.

Automation and real-time safety, we all know about wrong patient prevention. I think everybody should have the facial recognition system. That’s just my opinion. Again, take it for whatever it is. It’s free. It’s usually worth what you’ve paid for it. But the facial recognition system is incredibly robust now. I know it had a predecessor whose name we shall not mention. Joe’s laughing at me right now already. He’s like, “Please, God, don’t say the name.” But the patient ID system is absolutely fantastic. You can see it at a desite this evening. Tighter, smarter tolerances. There’s plenty of papers now about using SGRT data to predict PTV margins, and then therapist in the loop instead of therapist being the loop. So anything we’re doing checks with now that could be automated should be automated. And so the future should be a more automated approach, but it should be things that, not new checks, but automating or continuously running the checks that we already trust. So it’s not something we’re trying to add additional checks, but we’re trying to make the checks that therapists have to do very automated so that they can be focusing on patient care at the machine.

And so planning system and OIS integration, this was a huge wish list of people who have talked to me for years on some of this stuff. I’m just going to leave this here. There’s the OIS integration, I think more and more integration. The one I will harp on is this HIS/RIS integration at the CT sim. We have used DICOM for a very, very long time. CTs will run off of schedules using HIS/RISes and DICOM worklists. The fact that I’m still fat-fingering medical record numbers into my SimRT system makes me want to jump off the parking garage every day. If I cannot get a 4D CT sim VXP file over because the mismatch of the medical record number one more time, I think I will go insane. But there are some low-hanging fruit for the future that we could easily make these systems more integrated together, integrating these systems a little bit tighter. And then what comes next, this is just purely speculation, so we’re just going to get a little crazy on some of these. These are some love projects from some of our folks in AdventHealth in Colorado. Osteoarthritis is becoming a huge thing in our clinics. I don’t know. So I see some people already nodding, yes, we’re seeing more and more of this. I don’t know how you guys feel about your osteoarthritis hand sims. I loathe them with a fiery passion of a thousand suns, because we do silly things. We have these little old women coming in and saying, “I need my hands treated,” and the first thing we’ll do is like, “You know what we’re going to do? We’re going to have you lay Superman on the CT, so we can CT your hands with your hands above your head on this carbon fiber diving board.” She’s got arthritis, guys. It’s kind of a ridiculous thing. So this is how I feel. This is just to end the day on a light mood. This is how our sims go. My fingers hurt. What’s that? My fingers hurt. Oh, well. Now your back’s going to hurt because you just pulled landscaping duty. Anybody else’s fingers hurt? I didn’t think so. Yeah. So if that strikes a chord with you on how you feel when you do osteoarthritis sims, that is osteoarthritis sims at my clinic. So what we would like to do is reimagine how this works in the future. And so what we would like to do is do optical sims. So this is a MapRT capture of a hand, and we’re running in our scripted software. We’re essentially doing an optical sim of this patient. So we’re lining up the hand that we captured optically, using the cameras. We’re setting an isocenter, and we’re going to export this hand as a surface for treatment planning. We can then convert this hand, this three-dimensional mesh, into a synthetic CT. We can then use that CT for planning. So now we have a representation of the hand, but the patient never had to lay on the couch. They never even had to have any radiation exposure. They just had an optical sim. You can also use a much more accurate camera. So this is a very, very small, same type of technology, surface camera. This is my 15-year-old son’s hand. My wife does not let me irradiate him, so don’t call Child Protective Services on me. This is his hand, and then this is his synthetic CT for his osteoarthritis plan. I can see his fingernails. I can block the nail beds. I can do all the same things I would want to do with a CT, but I can do this completely without exposing him to radiation, as far as my wife knows. We can then line him up. We can set him up on the treatment machine just like any other patient now because we have the synthetic CT in the planning system. We can line him up with AlignRT, get him ready for treatment. We can actually take an image. I didn’t do this, but you can take an image of the person and make sure that you agree with the block. You can then use the DoseRT system to make sure that the coverage, now this is a live patient. This is not my son. The coverage over the wrist is what you would prescribe. And the nice thing about this workflow is this is the first time the patient has received radiation. There’s absolutely no radiation to this patient because they have a benign condition. They don’t have cancer. And so we have completely adapted this plan to an individual without ever having to expose them, put them in an uncomfortable situation where they have to lay face down on a carbon fiber plate because her fingers hurt.

You then when you get done, you can see the absolute workflow for treatment of both hands using the OIS module. You can see it all documented out. And this patient has a complete medical record number. You can bill this actually level three if you want because it’s two isocenters, and I was told I can do whatever I want. I was in that seminar this morning. So I’m billing for multi-iso and image tracking. But you can do this, and so I’ve not done anything incredibly earth-shattering here, but I have reimagined how that sim process should look for a benign condition within our clinic. We also have a project of love by one of my physicians is patient lymphedema. This is something that our patients really do struggle with in radiation oncology specifically. If you are irradiating nodal regions and you have prior lymphatic disruption via surgery, resections, dissections, this is a real thing. The detection struggles with lymphedema clinics is they usually don’t have baseline measurements. They lack consistent tools to do so, limited standardization of how these measurements are done. And these patients suffer for that. And the lymphedema clinics do not make money. Unfortunately, they cannot afford a lot of the tools that are offered to them are extremely expensive subscriptions. And so this is something where maybe SGRT can jump in in the future because the main challenges are we have no pre-treatment baseline. We’re seeing these patients every day in radiation oncology. We have measurement methods that are variable, inconsistent protocols for those measurements, and limited access to objective tools to analyze those measurements. And why it matters is because if we don’t catch this early, it’s hard to reverse these types of things, and you subject this patient to a longer-term risk of chronic swelling, infection, and it’s a hard-to-manage thing after it’s kind of gotten out of the gate. And so you’d like to be able to use a tool to actually identify this and refer these people into lymphedema clinics for therapy earlier on. A lot of the tools are very archaic. The circumferential tape method is most common, where you either lay the arm or the limb on a grid and then you’re measuring every four to five centimeters. You’re estimating the volume by assuming there’s a cylinder between each mark. And so you get an overall grid, and if the volume changes by 10% then maybe that lymphedema is setting in and you should refer them out. You can also put indelible marks on their arms so as they come back, you have a little bit more consistent location for each of those measurements. Water displacement is considered the gold standard, but you can imagine how fun that is trying to take a head and neck patient who may have edema in their neck and try to stand them on their head in a bucket of water. Probably doesn’t go super well. The arms, legs, that’s probably also a mess. Periometry is actually super interesting. As a physicist, I could probably dork on that for a while. But it’s an infrared light that allows you to get a contoured estimate volume. The bioimpedance is also another one, but extremely, extremely expensive for these clinics, where it’s measuring the impedance through the limb, using that as a surrogate for the amount of fluid around each cell. And then obviously there’s image testing, which is the most accurate, but unbelievably expensive, and insurance doesn’t like to track lymphedema once a week with an MRI. So, these are the things that are available. These are the things that we have tried. So we took our therapists, and we measured their arms with these scanning cameras, for getting a surface mesh just like you would for SGRT. And then it looks like this is not updating, so I don’t know why the video down here is not running. But essentially, you take measurements on this surface mesh. So you can actually use a geodesic measurement tool and actually measure this limb, and the scan takes maybe two to three minutes. And so if they’re already there for treatment, you can scan them in an exam room, and you can have these measurements repeatably over weeks of time. You can then use the tool like we just looked at, and since we’re looking for volume, why take the measurements and do all the stupid estimation? You could actually convert this into a synthetic CT, actually auto-contour it, auto-fuse it to last week’s, and just check the volume change from week to week. And this is a totally doable thing where you can automate this process probably very easily. This is me playing with Python code in my spare time. So someone much smarter than me could probably make something in the SGRT realm. Joe, I’m already calling it. If you make it, I get 1% of whatever you sell. But if you do something like this, you can imagine where surface guidance is now expanding beyond just on-table things to quality of life metrics for our patients. So, if you take nothing else out of the last 30 minutes, these are the things I wish you’d take away into the future. SGR is becoming a workflow platform. It is a treatment infrastructure. Right now it is kind of siloed tools, but it is becoming a workflow, in reality, in our clinics. To anchor the cases that you want to look at in your clinic, your anchor cases are your test cases. So if you are a clinic that is doing a bunch of maskless head and neck, or you’re a case that does DIBH breast, or you’re doing pediatric SBRT, if those cases are going well, then those become your anchor cases and you move on from there. If they are not going well, the first thing you need to do is use the tools more effectively and address the limitations in your workflow. Once you’ve kind of dialed in your workflows, you’re now ready to go on into the future and have that integrated workflow platform, and integrate more tools into that workflow. And the last one’s mostly for the vendors, but for everyone else too, is that the future is integration, it’s not invention. We keep having more and more invention. We have more and more amazing tools, and I love to play with them, and I will always be there to nerd out with whoever wants to give me a tool to play with. But unfortunately, I also work in clinic—not unfortunately, I shouldn’t say it that way. That sounds terrible, doesn’t it? Unfortunately, I have to deal with cancer patients. Unfortunately, we live in a reality where these things need to be practical. They need to be workflow integrated where I’m not dealing with crashing computers, or I’m not dealing with the same system not talking to itself. And the more and more we can integrate tools within SIM and tools within treatment where they speak with one another, but also kind of integrate with the other systems that we’re working with, that will be the future over, I think, the next decade. So, don’t hold me to it, though, because it could be totally something different. So that whole last 30 minutes could’ve been a waste of time. So but that’ll be it.

 

 

 

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