Hello, everyone, and welcome to our webinar. We will be highlighting the additions to our newest RayStation release in version 11B. My name is Nina Bodensteiner, and I'm the Director of Product Management for RaySearch in New York City. I have the privilege of introducing our two speakers today. First, we have Emil Ekström, who is an application specialist located in Sweden on our RaySearch Laboratories team. He is a resident expert on many applications, but he's been on our team for almost seven years. Before that, he has extensive clinical experience doing brachytherapy for six years and also worked for four years at the Swedish Radiation Safety Authority. Next, we have Craig McKenzie, who is our Senior Clinical Specialist from the RaySearch Americas team. Many of you probably recognize Craig as a past AAMD president. He was in the leadership at the University of Florida Health Proton Therapy Institute and then came to work for us. He went back to the clinic as the director of the Center at Miami Cancer Institute for three years, helping them to establish with RayStation, and then rejoined our team again in 2021. He has left the paradise of Miami and now resides in Texas and is joining us today. So before I hand it over to Emil and Craig, I'll give a high-level glimpse of our version 11B and some of the options and features that you can find. So excitingly, we have LET evaluation for ions. We have EQD2 reporting for brachytherapy and photon dose. We have the ability to do a cone beam CT conversion, and we've implemented your enhancement requests, a few of which are improving the registration workflow and visualization settings. All exciting features, and these guys are going to dive deeper into all of them throughout the presentation. One last reminder is that our software is subject to regulatory clearance in some markets, and specifically in the U.S., the 510(k) for 11B is currently pending and under the review of the FDA. With that, if you have any questions throughout, please feel free to use the Q&A panel. I'll be monitoring it, and if I see questions come in, I might turn it over to you guys, or I might wait until the end of the presentation. With that, I'll go ahead and turn this over to Emil. Thank you, Nina. Let's get started with the general improvements. All of you are welcome to this webinar at Nina's side. I hope you find it interesting and insightful. So let's start. We have latest release issued with RayStation 11B, and what we have done is the first part, how the LINAC commissioning process, that we can compute all the doses at once before you need to do anything by energy with the subcomponent and also time inefficiency since the Monte Carlo curves take quite a bit of time to calculate one by one and then commission the machine. Now you can just do it with one click. So this will greatly improve the commissioning process, especially if you're using our Monte Carlo dose method. But also one improvement. Before, when you just calculated one curve, for example, one profile, one dose, no one profile, or one dose for the Monte Carlo, the system automatically computed all the fields for that profile. So if you had a 10% profile, the system would calculate all the depth profiles, but only show the one you have selected. But now we are showing them all at the same time. When we're calculating depth in the Monte Carlo dose engine, it will automatically calculate for all depths. So that is information for free. So if you're using 11B, you should not be confused if you see more curves than you perhaps expect when running the Monte Carlo, because we are choosing to give you all the information actually in the system. We also made it able to have use different materials for different patients and clinicians. The full elementary composition of the different materials can now be edited and used for different patients. And the last of this slide is that we have improved our LINAC templates. So if you buy RayStation for the first time, you have an amount of template machines in your database where we have machine specifications according to how we think they should be. And over the years, these are going to change, and we have now improved our templates quite a lot. And we also have the support for different fluence models with analog and energy. So you can, for example, with 6 MV, you can have both a flattening filter, you can have a flattening filter free, and you can also have an SRS quality in the same machine and energy. So now no need to use different machines for the same physical machine. The other thing that we have improved is the ROI visualization of these things. Now it is a three-click toggle, you could say. If you click the header of one ROI type, the first click, you will remove them all, so no other ROIs will be visible. If you click again, you will make them all visible, and if you click a third time, you will go back to the state where you began. Say, for example, you choose to show the CTV together with, let's say, the bladder and all visible structures. If you click once, then they will all be unhighlighted. You will see nothing of the ROIs. If you click again, you will see all the ROIs, and if you click a third time, you will go back to your only CTV, including the bladder, for example. Also the visualization settings will be consistent now. Before, when you close the patient and open it, you simply had to do this all over. Say, for example, back to the bladder and CTV, for example. So if you select both and you close the patient, you will open again. Those settings will not persist between the different opening of the application. It's now different between after you click the patient and open again. You can see it at times. Also, for the clinical goal, if you have a sample, for example, for the clinical goals, and these will now be mapped automatically in the clinical goals. If you, for example, import a patient, say, for example, that you have at a different nomenclature for the naming of the ROIs when doing this patient, you can now associate that ROI with the ROI through the clinical goals without creating new clinical goals for the name of the structure you have in your plan. One small, yes, that was noted. Before, you could remove a full plan without signing it. You got the warning message, of course, but now you need also to sign it with your password. You cannot accidentally remove a plan as easily as you could before. One last thing is that if you go to the main menu, if you scroll to the top left in RayStation and [RaySearch X], you can use the shortcut dialog, and you can now search for the different shortcuts. Because we implement more and more shortcuts all the time. That list was static before, but now it's possible for easy access. You can quickly find the shortcut that you need to use. Nowadays, it's also possible to use the CT density for ROIs besides models for specification in the dose calculation. So there's no need to do it on the material override anymore. You can use the CT data instead. And the material is easy to visualize in that way. And now also, if you go to the 2D view, you can also choose the toggle between the image data and the material data. We have a dropdown, as you can see in this picture in this slide here. Then we did some improvements in the segmentation support and segmentation ROIs, as I mentioned in last slide. And this is RayStation 11B, and now we have better visualization of the material and the CT raw data, so Hounsfield units. And we did some very small improvements in 11B, which is that you can now compute the total dose with the rotation and patient perturbation. And another thing that we implemented was that the nominal distribution will now be rounded to the closest integer or 0.01 degree if you choose to use degrees as the unit. Because sometimes before, you can get some strange decimals that could lead into problems in the system. Now it's rounded to the closest full integer. And you can also export plans within the defined dose specification points or nominal dose distribution. And this can be selected in the export windows, as you see here in the image on the bottom right. And by the way, I believe Craig Mckenzie, who will speak a little bit of the recent modeling improvements that we're implementing in RaySearch 11B. Hello. Thank you, Craig. Thank you, Emil. Okay, so today I'd like to talk about adaptive modeling, some of the improvements, and then our approved image registration workflow. We now support multiple RaySearch image registrations using the same dataset. In particular, this is used for contouring. At the same time, we also can rename them and add comments, as you can see demonstrated here. We can rename and add the description of what the registration was done on, and we now can approve registrations as well. A new feature also, you'll be able to create POI registrations with less than four POIs, even down to one, which I'll try to demonstrate soon. There's a pan and zoom, which stays the same even after you change directions between axial, coronal, and sagittal, and then we have a new triangulation algorithm that's been updated and is now faster. Okay, so with the new improved registration workflow, this is what it'll look like on the toolbar. You'll have this new button here, which creates a new frame of reference, and with that frame of reference, that actually allows us to create more image registrations or multiple using the same datasets. There will be only one frame of reference prepared in a sense. We'll use this frame of reference to compute those on additional datasets, as well as creating the formal registrations, and when we click on new registrations, and you can see here, there's a frame of reference, and here's a head and a lung that was created as used for contouring, so it's going off of the same frame of reference, but yet we can actually go back and view each of these different registrations even in contouring mode. Okay, and just kind of a little bit of re-emphasis here is the two buttons are here, and as we create a frame of reference, you can see it's here. In this particular case, we did a floor of mouth registration. You can see it's kind of centered on the mandible. However, the C-spine is off a little bit. Therefore, we created a second image registration using the same datasets. And now we've re-centered it so we can use this for contouring. Okay, and just a short video here to kind of demonstrate some of the features that are in the GUI. So the two image sets are there. The two buttons are there. We are centered on the floor of the mouth. You can see the mandible. Again, the C-spine is off just a bit. Therefore, we'll come and create a new image registration. This, we can rename it, as well as we can make a description of what the registration or how the registration was done. Okay, and now using the new triangulation fast algorithm to quickly register these two, we're re-centered on the C-spine now. And you can see how it re-centered the new registration. Now we have two. Coming back to the toolbar, we can actually flip between the views for the frame of reference in any other registrations we have. We also now can approve them. As you can see, we can come in and do approval, which will block the registration. If we switch over to structure definition, now we can actually change that view within this module so that we can see the different registrations as we decide the time points. That includes each of the views that we decide to look at with the layouts, whether it's side by side or overlapping. Okay, and like I mentioned before, now we have where you can do less than four POIs per registration. In this particular case, we have one point that's located on two different image sets. It's placed over a bone spur next to the vertebral bodies. And if we just use that one POI, we can do a registration. And you can see if we click on it, we have one POI with different locations for. Okay, now I'll pass it back to Emil. Thank you, Emil. Thank you, Craig. Another thing that we have implemented in RayStation 11B is that it now features the integrated CT from the CBCT. We have two ways of doing this: the CBCT correction and the synthetic CT. This functionality is part of the license that's called Dose Tracking, which is part of our dose tracking functionality. The first one, the corrected CT, is working in the iterative manner with two main parts. Find a conversion from the CBCT intensity unit to the planning CT Hounsfield unit CBCT. Then find a correction map that can remove low-frequency artifacts. [audio distortion] in one of the CT sets and not in another, we can take that. We also have a second method that is called Virtual CT, which is a hybrid between a default CT and the CBCT correction method inside the loop. And this can be used if the first one is giving the result that you wish for. So you can have the option to use both, one or the other, to get the perfect result. It depends a little bit on the device and other changes in the patient and after the treatment and planning CT, of course. And both these methods will also simulate the CBCT if you have a limited field of view of your original CBCT. So let's say you have the smallest part of the patient as you combine CT when treating or just before treating. This will fill in the remaining information from the planning CT if you have the whole image set representative from the synthetic CT that you generate. So you can take the data from the planning CT to fill in the gaps. This may be clinically irrelevant, of course, but it might be of interest to have here. So it's pretty neat. We're also working on our AI functionality of deep learning segmentation. So we hope to release new models. We have released some, and there are some in the pipeline that is both new and also some are improved for old models. You will see a list shortly about the different models. Also learning, we're using deep learning segmentation templates that we'll use to use specific needs. So you don't have to, you can select and click there. And also the segmentation of lymph node level stations. So I will keep this slide here for your reference for a while. I will not read through it, but as you can see, the models in white are already released models, and the models in green are models that we plan to release this spring, 2022. So this list is really starting to get longer and longer, which is very nice. And of course, our aim is to be able to provide models for all relevant feature types that is common and/or become a problem pattern. We want, of course, to have as comprehensive a model library as possible. And also, just because you see the model here in white, it doesn't mean that it's completely released, but it also means that it's continually improved upon as soon as we get access to more data. As you know, the models themselves have not contained any patient-specific information, so these models can be shared by us to you without depriving you of any privacy laws. That's a neat feature of our models. No patient data is in the model file. Yes, as I said, our goal is to have a RayStation standard model available for infrastructure-relevant patient therapy. Another feature that has been implemented in RayStation 11B is that we can do a machine learning planning model at the beam set level. And also, the licensing has been a little bit changed. Now there are only two specific treatment technique licenses: the RAYD planning programs, the RAYD planning protocols. Some of the software planning improvements here, we have improved the standardized planning method list. We now have the ROI protection option. And if you're a user-optimized system, you've got to use the protection when using something like an MCO functionality or a VMAT functionality. This means that if you block, if the MCO is blocked, the structure, the ROI, actually protects so you won't get any leak openings when going around that ROI. And we also entered some beam angle feasibility during optimization. And the reason for that is to have a greater chance of the plan being deliverable. Because back in the day, we could sometimes get a plan that would be very nice, but then it violated some of the constraints of the machine itself, which is, of course, not good. And it would show at the latest stage, but you can't use it in those. But now it's also continuing to work around during the optimization. You can have these violations during the optimization in an efficient amount of time for planning. We could use it, of course, for the jaw-specific Varian machines, so that the jaws will follow the leaves. Another thing, which might not be very obvious in the GUI, is that we tried to, or we did, do a lot of memory improvements to the segmentation of doses. Because if you have a large patient with a lot of leakage in it, sometimes the memory could run out, leading to the crash. But by tweaking and optimizing the algorithms, we are now much more easy on the memory computer, so these crashes will be very seldom to occur. It's also possible now to reverse or make a copy of an arc and reverse it, so the copy becomes a reversal of the original arc. Lastly, on this slide, we did some improvements to the treatment-based VMAT in the multi-criteria optimization. For example, unidirectional motion of the MLC leaves is no longer strictly enforced. This leads to more deliverable planning, also improves the achievable plan quality, and most important of all, it's less likely to violate [audio distortion]. Another thing that we added to 11B is that we now can, as you can see in the left picture, it's a new icon in the optimization panel. You can press fine-tune. So let's say you have a plan that looks quite good, but you're not really happy. And instead of trying to iterate or re-evaluate your predicted constraints for that plan, you can now pinpoint the ROIs that you're interested in by using this global clinical goal fine-tuning. So what you do is that you press this fine-tune icon, and you will be sent to the dialog to the right, and you can select the specific clinical goals that you really want to fine-tune. And what the app will do is try to fulfill these clinical goals and preserve the VH as well as the overall clinical resolution, but still try to work locally on these specific fine-tuning ROIs that you have selected in this dialog. So it is also a time-saving job. So instead of doing the reiteration of your existing constraints in the optimization process, you can just pinpoint the exact ROIs that you want to keep, and you will save some time. And with that, I give the word over to Craig. Thank you, Emil. I just did a short overview of some of the proton planning improvements. We are now going to compute dose average LET as part of the final application using our super-fast Monte Carlo dose engine. The water equivalent depth is computed and displayed in the beam-specific points alongside their physical depth. We also have a range modulation PTV displayed for the ocular plans. We support single scatter delivery systems, and we support non-uniform fluence for all proton broad beam techniques of simulation modeling. We support setup beams for boron neutron capture therapy, which would include DICOM export. We also, just to point something out, can do background dose when we're doing our 4D optimization. We're using the background dose as one of the functions. We do limit this functionality of robust to be just the beam set dose. LET. I'm proud and excited to say that we're the first commercial TPS which supports clinical evaluation of LET. Application of dose average LET for protons and light ion plans. This is fully integrated into the plan evaluation workspace, where you can see it in 2D. You can see the LET volume histograms. You can see line dose, dose statistics, and with that, we'll demonstrate in a video showing how it looks like in RaySearch. We're coming to plan optimization now. I just wanted to demonstrate under the beam dose specification points that we do have water equivalent depths that are displayed, as well as physical depths, and of course, we changed to a different ESP. We'll be able to have those updated as well, and in order to get the LET, it's as simple as checking a box that says compute LET and hitting final dose. It's going to be part of that fast Monte Carlo algorithm and dose engine. I'll switch over now to the plan evaluation, and I will compute it. Over in the left side column, where we have our evaluation doses, you can see it computes it back for us. Now on our display, we actually can put comparison between the actual dose and the LET. Within the LET, we actually have a color table, which we can adjust. We can actually set a threshold, minimal threshold for that dose in order to filter out some of the LETs that are displayed. We can add a line dose. In this case, we want to trace down the axis of the beams. Okay. We do display our LVH. You can see on the right side at the bottom. We'll do that next to our DVH, which is next to the left. If we switch over to line dose, each one is here. We have dose on the left column, and we have LET on the right column. We can make this fit for max. We can go down the graph and look at each one of them and see where the two dose as well as LET overlap. Now I go back, and I can adjust my line dose a little to kind of see what location within the anatomy of the CT where that ends or where the two intersect. Okay. Now I'll just show that we do this for each beam set of LET. We also can compare LETs of different plans if we choose to do that. The difference is on the bottom right. It also will show those statistics. Here I'll turn on the difference between the two. And we also can change that color table as well to be able to show the differences between the two. We just open it up and change our values. Now we'll be able to show the difference between the two. All right. So exciting for LET. A lot to do. I know a lot of people have been waiting on this. So it is in 11B now. And I know Emil had mentioned this earlier, but it's also a very important feature for keeping protons. I just wanted to reemphasize that we now do patient rotation within the compute perturbed dose function. So now we have the ability to add that as part of the evaluation. And again, we do include the statistics for LET as well as RBE. Okay. I'll pass it back to Emil. Thank you, Emil. Thank you, Craig. So as you know, we have released the [audio distortion] strategy as we released it yet, and a lot of things have happened. It's now possible, for example, to lock specific dwell positions in the plan. You see, for example, if you look at the picture to the right, you can see that we're in the ring, and we use line application to do one fixed dwell position, then we can enter them and lock them, and then use the initial needles that is coming with this feature application. We can only use our optimization function on them, and they will take the dose given from the locked position in the ring and the dwell time into account and do the optimization adhering to the specific constraints there. Another thing that we have put in 11B is that we now have direct access to the fusion functionality that you might have seen in patient modeling before, so you don't have to switch between the different workspaces if you want to do something that involves fusion between different beam sets, so it's a workflow improvement. It's the same functionality as you might be used to if you used fusion in the case of [audio distortion] before, the track time buffers and so on. It's more easily to use. Another magic trick is that you can now switch the visualization on and off for both random channel candidates as well as the applicator template, and if you played around with 11A, you might know that we introduced a functionality that's called [SmartPro], and the [SmartPro] is an algorithm that tries to find the channels in the patient. So before 11A, we only could not detect solid markers. But with this [SmartPro] improvement, it's now also possible to have dotted markers inside, and the system or the algorithm will find the correct source path even if you have a dotted marker. And I think, as you might have noticed, it's now possible to define clinical goals in EQD2 in the brachytherapy module based on the linear quadratic model. And as I mentioned before, it's also possible to calculate EQD2 goals for voxel samples as well. One little caveat here is that the clinical goals EQD2 only works with brachytherapy at this time. But if you have the plan evaluation license, you will be able to calculate EQD2 goals for all focal paths. EQD2 goals. So here is the dialogue. You can also, of course, deform the goals. I will show this in the video very shortly. Deform and accumulate the EQD2 for different treatment times, fractions. You can also deform and sum them, so you get a big picture of what we have really given the patient, how much dose. Another minor improvement, which is very good, I believe, is it's possible to rename the different doses now. So you can have a free text mode that you can actually name the dose. So it's really easy to avoid confusion. This is not only for EQD2, it's for evaluation and summed doses. Let's have a look at how this works. First, we need to compute the EQD2 dose. The priority here is if all ROIs overlap, the one with the highest priority will take priority. Then we enter the alpha beta value for this. So the EQD2 gets 10 with the GTV, and the overall risk is at three. And we choose which doses we want to calculate on. And click Complete. So it's quite fast. And if we go to the arrow, which is the evaluation doses, and press it, we will be able to see what we did. So now we have two EQD2 doses. One is the first two fractions, and the one two is for the three and four fractions. And if we hover over the folder, we can see the exact properties of the different templates that we used when calculating the dose. So now we want to sum the doses. And this is called fraction one dash two, you can see fractions given as fractions three and four. It's two implants, two patients have two implants. One two is the two left fractions with the same implants. The EQD2-1 is for the first implant and the first and second fraction. We have summed this as we did with the first one. This is the fraction one and two. We have summed the dose for the two first fractions and the three and four fractions. We have a deformable registration in the background. We deform this dose to the original image set. We get a new dose for these fractions three and four. This is mapped to the first image set, MR5. We sum those doses. We get the total dose in EQD2 for both fractions in the same image set. We're done with this. We can just evaluate our deviations. We can drag and drop the doses in the plan evaluation. We can look at the difference over. We can also evaluate the deviations. The dashed lines shift the EQD2, and the solid lines shift the physical dose. Of course, we can also look at the dose that we took and the physical dose to make sure that they are physical dose. Or if they are not, how we perhaps could take into consideration when we give fraction five and six. So that is about what I have to say today. Thank you very much for listening. And now I give the speech to Nina again. Okay. Thank you, Emil. Great chat from both of you. And I have some great questions coming in. I have been keeping answers to a few of you, but there are some that we'll go ahead and ask live depending on if you guys are in the class meet. The first one's coming in with regards to the material override. I know that there's some excitement and interest in how we change showing that in the software. And so the question is, can material override CT be exported? And also, can it be used for DRR generation? If either of you want to tackle that one. Okay. I don't know. I don't have a straight answer to this. So I will refrain from answering. Okay. Well, I'll jump in here and say that, per se, material overridden CT cannot be exported. But when we do export a CT, the material overrides of the ROI are included with that export of the CT so we can see what the material overrides are. That is included in the export. So it's not, per se, the CT uniquely with the material overrides. And as far as DRR generation, do we know if the material overrides are taken into account there? I believe that they are. That's how some of the MRI we have can use an MRI as our primary, and we can use the material overrides as in the DRR. We can follow up on that to make sure. So I'm going to jump topics here and go to the deformable image registration. So I have a customer asking that when we do a deformable image registration, we choose the CT as the reference and the cone-beam CT as the target. They want to swap that, and they want to do it the other way around, choosing the cone-beam CT as a reference and the CT as the target. Can either one of you comment as to how that process works for them? I said now I'm just going to put you on the spot. Either one of you can answer that correct, right? We want to switch to the registration. That is something we should be able to create a new registration and switch to, yes. So it's as simple as just checking which is the reference and which is the target. And you can reverse it in any order in that window. It's possible, like Craig said, create a new registration and then reverse those. And if you have questions on that, as you know, as a customer, you can always call or support me. This is a question about the generation of the synthetic CTs and generating them from the cone beam CT. What are the requirements or can you maybe reiterate, Emil, when we have a limited field of view between the two images? I know that's an important feature. Sometimes we don't have all the data there in front of us. If you could view-wise, when we are on the field of view to pick from the cone beam CT, we would do either of these two, either the rigid or the deformable. And once the deformable registrations between those are made and we have the matching structure, what is not in the field of view would just be added from the planning CT. It would not fill in the blanks, so to speak. So you have the whole CT set. Even based on the cone beam CT, you will see in the synthetic CT, you will see the whole patient as it is on the planning CT. And the data without the field of view would be copied and pasted from the planning CT because we don't have any information to give information in those areas. Perfect. Okay. This is back to you also, Emil. How about EQD2 calculation? Is it possible for IR? Not at this moment. But that's something that we definitely need to do. But not in 11B. Perfect. And I've been answering some of these questions. I think you guys can all see questions and answers that have been coming in. This next one, I can answer. It says we currently have 11A. Do we need to move to 11B to implement the new deep learning module? So 11A, we do have the released models that you currently have. The models actually are released independent of the software releases. But the new models will coincide probably with the FDA approval with 11B. So 11A, you can have pre-trained models, and more will be coming out. And how that timeline really lays out is yet to be determined. And friends, please call us because we can help you with that registration to get that. But please call us, and we can do so. Did I jump? Yeah. I see any other questions to ask. So maybe some more are coming in. Okay. I think that's all that we have coming in. I'll go ahead and just do a couple more slides here. I really, as part of the product management team, wanted to note that it's really important to us that we do listen to our customers. And I think you just watched through this presentation that many of these features are in there directly because you asked for them. This is a small sampling. Some of these were already spoken about. Maybe I'll reiterate a few. But ability to do an approval before you delete plans so that we don't accidentally delete some. Some of our ROI visibility settings that EQD2 goes through, naming those some doses. That was really asked for by you, etc., so for listening, there are many, many enhancement requests in each of the releases, and we're really proud of that, and lastly, we've really thrown a lot of information at these presentations so fast. I know it's hard to really, when you're a RayStation user, absorb all of it, so I just wanted to reiterate a couple of resources to you. We have this 11B new marketing sheet that is available. Anyone can send it to you, or you can download it from our website. And it really has the highlights. Here's what's in this release so that you can go looking for them. But when you really want to dive in, go ahead and find the release notes because we've meticulously gone through and really typed out so many of the features in there. I think that our saddest moment is when we have a user that, hey, wants to print some doses, and we're like, "Hey, it's already there. We added it." So just make sure you read those release notes and find all that great information. I do have one more question that came in. Would automatic optimization of gantry and table angles for VMAT be possible in 11B? Automatic optimization of gantry and table angles. Does everybody want to tackle that one? Is it needed to gantry and table move during delivery, or just optimize? I think just optimization of those angles. And that is a simple possibility now, right? Yes. O kay. Agreed. Agreed. All right. So with that, I'll do a reminder that this webinar is recorded. It will be available on the RaySearch Laboratories' website. And it also will be available in the community where we want all of our RayStation users to be going. There are all of our past and present information about our webinars are found there. So if you don't have access to the community, please do make sure that you contact us, that you ask about that, and that you find that access today. And we do have another upcoming. This is for RayStation users only. We have a series called Ask the Expert, where we put our users in contact with the developers themselves. This is a full webinar on the synthetic CT and RayStation and its use in deformable and adaptive workflows. So that is upcoming February 17th. And if you're a RayStation user, you can certainly sign up for that. Again, it will be recorded. And all of the Ask the Experts are found in that community. I don't know that I have any. There are more questions coming. Okay. So Emil, which HDR planning commercial system should we plan for today? I think that you should know which system. In the box, we support the [audio distortion]. But we're working really hard on it. It's a big priority for RayStation to be able to deliver plans from other vendors' machines. Perfect answer. Question is here about printing some of the doses. So I said it was now available. Jared, he wants to know if that is printable. Do you guys know if it's possible to do printing of the DVHs? My graphic representation is that that is a printable option. I agree. I cannot give answers in full, but it should be because we have almost all functionality that is not RayStation-related if printable. I would agree with that. I would agree with that. And I think that we've answered all of the questions. I would really love to thank Emil and Craig for your hard work for being engaged with 11B release. Thank you for showing up and presenting for us today. And with that, I will remind everyone that 11B was released in December. It is still going to FDA approval within the U.S., and some of the other markets also require it. We thank you for your time watching today. And with that, we'll go ahead and conclude this webinar. Thanks, everyone. Thank you. Thank you.