Transcript
[silence]
[00:00:31]
[music]
[00:00:55]
Bogdan: Hello, everyone, and welcome to a new Novalis Circle webinar. My name is Bogdan Valcu. I'm the director of Novalis Circle. And today I have the pleasure of introducing Prof. Maximilian Niyazi and Philipp Freislederer from LMU in Munich who together will be discussing their initial clinical experience with ExacTrac Dynamic. Prof. Niyazi is the vice-chair in the Department of Radiation Oncology at Ludwig Maximilians University in Munich. He is also the deputy coordinator for the Neuro-Oncology Center. And Philipp Freislederer is the scientific employee in the Department of Radiation Oncology and he serves as a medical physicist. As some of you may remember, the team at LMU was the first one in Europe to utilize the ExacTrac Dynamic for initial treatments back in April, but since then, a more comprehensive work has been put in place to commission, validate, and re-establish ExacTrac for a classical radiosurgical treatments in the brain and spine.
In the webinar, today, we will review the benefits of the new camera system and its advantages that it brings for patient setups. We will touch on the IGRT benefits that ExacTrac Dynamic brings for cranial and spine radiosurgical setups, discuss the clinical implications of our IGRT system in the clinical practice and review relevant commissioning procedures. Ever since the release of ExacTrac Dynamic back in April of 2020, we have begun adding new systems to the more than 1000 install base of previous ExacTrac versions with 17 systems that are currently clinical in Europe and a total of almost 58 systems to be installed by the end of the year in Europe, North America, and Asia Pacific. And as more of you have an opportunity to install the new system, we want to provide you with scientific relevant data, clinical visibility, and commissioning details to aid you in the implementation of the system in your department. Stay tuned for ongoing information on ExacTrac Dynamic clinical applications. And the next webinar that we have planned during ASTRO time will continue to cover not only cranial applications, but also extra cranial indications.
As always, we continue to provide CE credits for those of you that successfully complete the course. And should you have any questions regarding your CAMPEP, MDCB, ASRT credits, please email us at info@novaliscircle.org. And don't forget to sign up for our upcoming webinar on September 30th. That would actually mark the introduction of a case of the month series where we will begin reviewing interesting cases at some of our partner sites. There's Macyszyn, Kaprealian, Agazaryan at UCLA will be reviewing a spine patients during the introductory webinar and we'll hope to see you online.
For today's webinar, please remember to either utilize Google Chrome or Safari. And should you have any connectivity issues with your internet connection, simply refresh the window. Utilize the chat interface to send us questions. Upon completion of the two lectures, we will answer your questions in a live session, monitor the polling interface for questions that we may like to ask you and should you follow us on social media, please utilize #NovalisCircle. And I'd like to turn it over to Dr. Niyazi for the first lecture.
Prof. Niyazi: So, dear ladies and gentlemen, so I'm very grateful to give today's Novalis' webinar on our first clinical experiences with ExacTrac Dynamic. My name is Maximilian Niyazi. I'm the vice chairman of the Department of Radiation Oncology at the LMU University Hospital in Munich. I'm very happy to give this talk together with my dear colleague, Philipp Freislederer, who will stop talk about the physics aspects afterwards. So, these are our disclosures.
So, the LMU Munich and our department has about 2400 patients per year, has an equipment of 6 LINACs, including MRIdian LINAC and 2 ExacTracs at 2 different campuses. We have the staff of around 130 persons. And as you can see, we try to install the ExacTrac Dynamic system to enhance our imaging capabilities. And the nice thing about ExacTrac Dynamic is that you have more or less four dimensions. You have the surface of the patient, the thermal information, and you have X-ray information. That's what you all know, but the thermal component is the new one. And the good thing is that it is more or less everything integrated. And in this sense, you do not need multiple in-room cameras to track the surface. You can do this just by one imaging instance. And the other nice feature about ExacTrac Dynamic is that you have additional real-time tracking using a real SGRT.
And that's actually a nice first starting image. And that's my dear colleague, Philipp, who is actually included here. That's the thermal information. And that's the two-dimensional surface tracking. So, it's a very nice sketch of my colleagues. And the indications from the sketch starting with our program in June of this year was that we introduced treatment types that we have already applied on our old ExacTrac. So, we have included multiple Brainlab SRS and every kind of hyperfractionated SRS in the brain. We introduced fractionated stereotactic radiotherapy with open face masks, for example, in meningioma established for normal glioma. We replaced the cone-beam workflow in palliative settings and for spine SRS, you know, running a prospective study. And of course, you can treat every bone oligometastatic disease anywhere else in the body.
So, I will talk a bit about the workflow of the ExacTracs Dynamic. And the nice thing is it's a fully integrated system. With LINAC, you have the automatic plane loading from the RV system. Pre-positioning itself is done by using the outer contour of the planning CT and it sends shifts to the LINAC. So, that's the situation when the patient is lying on the table with an open face mask. So, there's the thermal information, the pre-positioning that is done using the outer contour as defined. And afterwards, you select the area of interest for your surface tracking. In this case, of course, it's the image that's outside of the thermoplastic mask system. And here we have your classical VRRs with the registration of the X-ray images from orthogonal directions. And you have very nice tools to, for example, have your control over the match. And if you're, for example, satisfied with that, you can send the changes. And then you start with the monitoring workflow. If you enable the X-ray imaging and the surface tracking, you have to choose certain settings for your tolerances. For example, the X-rays shift tolerances and the surface tracking tolerances. And you normally would choose them a bit higher because of higher deviations expected from the surface itself.
In this case, we had an acoustic neuroma and there was a very nice and stable situation and everything was intolerance. And the X-ray imaging in a stereoscopic fashion is done at the 90-degree angles, 180, 270 and zero-degree. And as you can see, as far as you have such an out-of-tolerance information, then you have to check for the next stereoscopic image and then you have to adjust. And that may happen, for example, and for instance, in this case, if something of the gantry, for example, comes into this small field of view. So, our experience up to now is that the first patient was treated on June 2nd of this year. We've treated 20 patients, 16 brain tumor patients, 2 spine, and 2 head and neck cases. So, that's not too bad. And there is a clinical example in this case of correction of non-coplanar fields which is quite important for stereotactic radiosurgery. And as you can see here, you have the situation that your angle is actually approaching the 90 degrees here and they view X-ray imaging. Everything is nice within tolerances and then just continue. And as you can see here for the registration you can swipe out some structures that are more deformable, and in this case, it was very smooth treatment.
Okay. So, something of, let's say, preliminary information on stereotactic radiosurgery offers excellent local control and pain reduction that's a nice overview of Tseng 2017. You have really high rates of local controls irrespective of prospective or retrospective design trials. And you have very fast pain reduction compared to conventional fractionation. So, it's a very nice method being used in oligometastatic disease. And there are ASTRO guidelines as well with very precise information on, for example, support tolerances and other organs at risk. And Sahgal was one of the leaders in this field. And that's actually something that is used in the brain lip software that comes up within consortial target definition of spine volumes, which are, for example, referring to the lesion size where it's located and whether to include the pedicles, process transverse, spine process, or the laminae. So, it's very important that this target volume is actually clearly defined by the guidelines. And the software automatically delineates the vertebral body and the structures of the target volumes according to the lesion that is placed here. And the best thing about software is that you can co-register your MRI in a fashion that the deformation is actually used.
As you can see here, you have your deformation done from the MRI to the CT. And then you have really this kinking of the vertebral column that works pretty nicely. And afterwards, you have your classical overview of the target. You can choose the prescription doses. You can normalize here to the whole PTV and you can use the primer lesion within the vertebral body. And you can choose among different features how to shape the dose with a VMAT technique. So, that's an integrated workflow, nicely fitting for the ExacTrac Dynamic. And that's the first case actually not with the Brainlab software because we're right now installing it. It's uterine leiomyosarcoma metastasis. Initial presentation was in 2015. And we went resection and we went chemo. Had been sacral metastasis, resection, again and we went chemotherapy. And later on three years later we had destructive process of T9 and the humerus. This was resected and we had palliative conventional or moderately fractionated treatment with 3 gray times 12. And seven months later, we had the massive progression of this sarcoma metastasis, and we treated it with the VMAT and the spine with the Elekta software Monaco. And you have a very, very steep gradient to the code because of this previous irradiation just seven months ago with five times four in six gray. And in this case, we use the very tight margins.
And as you can see here, we had, of course, the problem if you have the surface here on the stomach, for example, on the abdomen, then you have high margins that you should choose for the surface tracking because as you detect breathing motion, it doesn't make too much sense if you would like to track the spine. And if we had very tight margins for X-ray surveillance with 0.7 millimeters and 0.5 degrees and use the highest frame rate for X-ray imaging possible in this case, only the stereoscopic views with 4 per arc. And the surveillance was done by X-ray imaging. As you can see, it worked out very nicely. Patient could treat it just some weeks ago.
So, the other experience was on the multiple brain mets. We have a high-resolution contrast-MRI that we're using for target definition. Sample rate sequence, the slice thickness of 1 millimeter with a double-layered mask, planning-CT with contrast, dominates within a week of the MRI with a GTV that is contoured and checked on MRI and planning-CT and 1 millimeter of PTV margin is added and we're treating up to 10 test disease and in some cases even more but on trial with 4 to 10 metastasis and comparing this to historical whole brain [inaudible 00:14:31] from our department. And as you can see here. And now this is a nice software tool of Brainlab that is doing the multiple brain met SRS treatment. You have, again, your nice panel where you can delineate the lesions. And here you can have a nice overview on the prescription dose. And if you click on the single lesions, you will get feedback on the conformity index, the gradient index, the prescription dose, min-max and min values that makes it very easy to assess whether this plan is suitable to be treated.
And concerning another clinical example, we had a patient with a non-small cell lung cancer that was EGFR-mutant with a PD-L1 smaller than 1%. Initially treated by VATS pleurectomy and pleurodesis. Received afatinib and was switched to carboplatin/pemetrexed and liquid biopsy was negative and EGFR mutation had a massive progressive disease. The third line nivo treatment and within the latest MRI she had four metastases and now we're treating all these four metastases with one single isocenter treatment. And now I'd like to hand over to my dear colleague, Philipp, who will explain what the potential sources of errors are in this case and, of course, on all the commissioning data of ExacTrac Dynamic and his experience. So, thank you so much for your attention.
Bogdan: Thank you for your talk, Dr. Niyazi. And I would like to ask Philipp Freislederer to continue with his talk.
Philipp: Hello, everyone, also from my side. Thank you, Max, for the introduction. My name is Philipp Freislederer. I am a medical physicist here at the Department of Radiation Oncology. And I'm now gonna talk a little bit about the error sources in cranial SRS treatments and then go quickly over some issues we potentially avoided in the cases and the clinical examples Prof. Niyazi showed and then talk about commissioning a little bit at the end.
In general, when you do cranial SRS, you have dosimetry-related errors. So, your output factors for small fields below two centimeters in your MSC calibration which affects your output factors. You also have errors related to the imaging. So, for example, geometrical distortions in the MR images. And also, which is quite important, the calibration of the imaging and the radiation isocenter. This is something extremely important for the ExacTrac Dynamic as we, in our case, switched from a cone-beam CT workflow for most of the patients we have there to a full ExacTrac workflow and we placed all the cone-beam CT for these types of patients. And also you do have mechanically-related errors. So, any residual intro-factual motion inside the immobilization masks. You also have runouts of the gantry, runouts of your collimator, and runouts of the couch. And when it comes to SRS treatments cranial, you do have a lot of couch rotations, which leads also to couch runouts. And this is something you could avoid with ExacTrac and ExacTrac Dynamic.
In general, all of the patients we treat for the multi-metastasis cases, we treat with a single isocenter. If you look back from how you might use to have done it, going from an isocentric stereotactic treatment in the head, you will have one isocenter per treatment target. And it means if you have some small translations that might affect your 80% isodose in relationship to the PTV margin, which you see here a little bit. Rotations do not affect for spherical targets. Mutations do not affect those coverage too much. And now if we switch to monocentric treatment, so treatment of multiple metastases with one isocenter. If you, again, have some translations, you might have some effect on your dose on your PTV coverage. But if you have rotations, which is in the right button, small rotations can have a huge impact depending on either the distance between the targets, the rotational error or the target volumes. So, we have a high-risk of compromised coverage.
If you go back to our clinical example Prof. Niyazi showed, in this case, we took one correctional X-ray image before each couch angle. What we already have seen for this one patient we have evaluated the data with the older version of the ExacTrac for multiple patients, but for this one patients, we saw errors in yaw for up to one degree. And all of these errors could have...were compensated using the ExacTrac or the ExacTrac Dynamic in this case. And we might have had a significant deviation from the treatment plan. So, this is the one... This shows the importance of tracking your couch or tracking the movements of the patient.
Now, quickly from this topic to QA and commissioning. As I think everybody knows, no direct guideline or recommendation is available yet for the commissioning of such a system, there are no guidelines available for surface-guided radiation therapy only systems. And this is a hybrid machine with X-rays and surface guidance. It is quite tough to sort of know what you do or what do you need to do at the beginning. When we start commissioning a system like this, we have some initial thoughts when it comes to surface guidance only. So, we want to test multiple things, static accuracy, dynamic accuracy, the impact of different region of interests, also the impact of your field of view. You want to check if there's any blocking of the gantry, blocking of the accessory. But it's quite important for our stereotactic treatments and all other treatments as well is the coincidence of the different isocenters, so the isocenter of your surface guidance system, your X-rays, your cone-beam CT, and most importantly, those have to be coincident with the MV beam.
When you perform treatments with couch rotations, you also have to commission the system or test what happens when you rotate the couch. Another interesting point for us was the relative and absolute error in shifts when you compare cone-beam CTs with the ExacTrac Dynamic. This was important for us because we sort of took away all of the cone-beam CT workflows for our patients which we are treating at the ExacTrac Dynamic. And the impact of system configurations might be also of importance, but this is something which is already done pretty well automatically by the company. But we will test these configurations in the future quite extensively. The so-called warm-up effect of the system or the thermal drift of the system is also part of it. I have some words on this later on.
To start off, we'll be doing for routine QA for the ExacTrac Dynamic. There's a daily check implemented by Brainlab. It takes a maximum of five minutes in total. In the steady checks, two things are tested, one is the deviation between the surface camera and the X-ray positioning system. So, it's the so-called consistency check between the two systems in one. And the other one is the deviation from the radiation isocenter. Once a month you have to do a thermal to 3D surface calibration. This also takes five minutes. There is a dedicated phantom for it. We do recommend a monthly calibration of the radiation isocenter, which is simply the placement of any size of a ball bearing inside the radiations isocenter. You could do that either with Winston-Lutz pointer or with any type of ball bearing inside any anthropomorphic phantom. As we do a lot of stereotactic treatments, we perform weekly SRS checks mostly on the day where we perform the actual treatment. This is done like... It's actually pretty straightforward. You move anthropomorphic phantom into the radiation isocenter using the ExacTrac Dynamic. In the isocenter of this phantom there is, again, a ball bearing and you check the position of the ball hearing using your MV beam.
So, this is at the moment most of the routine QA we're doing. We will add something in the future when we add more types of treatment sites. For commissioning, we wanted to first see how the surface and the X-ray positioning fit together. So, we used a Brainlab phantom called Lola. The phantom has a heat signature inside, so you can actually use the surface and the thermal tracking. We perform 10 different random isocentric couch deviations and translation and rotation for couch zero degrees to simulate isocentric uncertainties. And we perform for only two couch angles. I think a colleague of mine is at the moment using the same test for more couch angles for pretty much the 180-degree couch angle in 10 degrees steps. But for the first commissioning tests, we wanted to stick to two only to simulate intrafactional movement on a different couch angle. And then we had a deviation between the X-ray. So, the delta D position is deviation between the results we got it from the X-ray shift and the surface shift. For couch zero, everything was below one millimeter. One little higher impact had the longitudinal direction. And for the couch 60 and 350 degrees, we wanted to check on intrafractional monitoring. These shifts where all shifts were below 0.5 millimeters. So, the system has an accuracy below or submillimetric accuracy when you compare the two entities of the system together.
The second part we did where we can show some data already is the verification and positioning accuracy. When it comes to ExacTrac Dynamic versus a cone-beam CT workflow. We use two anthropomorphic phantoms with six random isocenter locations each. We set up the phantoms according to ExacTrac Dynamic and according to cone-beam CT and then position the phantom according to ExacTrac Dynamic and afterwards to the verification ExacTrac Dynamic image, X-ray image, and a verification cone-beam CT image. So, the deviation for the initial correction was between the two entities, ExacTrac Dynamic and cone-beam CT, was again very low and maybe a little bit higher in rotation. And when it comes to verification, those two systems agree very well for a very static non-moving phantom. That's what you have to add in this case.
When it comes to all the other initial thoughts on QA and commissioning, I wanna go through the same list I showed you before again to see what our thoughts were and what our results were. So, for static accuracy, we wanted to test it as an SGRT system only. This is not possible using the ExacTrac Dynamic because the baseline for the accuracy or rather the baseline for the positioning result is always the X-rays. So, the static accuracy of the ExacTrac Dynamic surface camera only cannot really be tested. It is also not as important as the accuracy and a static case relies only on the X-rays. So, you rather have to test for the accuracy of the X-ray positioning, which we did when we looked at the...which we did directly when we looked at the comparison between the surface and the X-ray system.
When you want to test the dynamic accuracy, this is something you might want to do with a surface guidance system also. This is possible but in our case, we are kind of hoping that the next version brings a little physics mode because the data, if you want to store the data, you can only store it for Beam ON. This is something that I'm sure it's gonna be made a little bit more open in the future, but you have to know that if you want to store any dynamic accuracy tests, like, the test is quite simply. You put a phantom on the table and then you watch how the surface is drifting.
The impact of the region of interest is always hard to tell on any phantom surface. We will need some patient data to do that. It's not really essential for commissioning, but it's something we would like to know in the future. The field of view is quite easy to test or not to test. It's just the look of the field of view. When it comes to gantry and accessory blocking, there are two parts of it. The surface camera does not have any blocking because the registration is done using one camera only and that's also the reason why you have the thermal camera. The thermal camera is the reason why you only need one camera inside the room and not three. This means that you don't have any gantry blocking off the surface camera which is a good thing. But any accessories of the...in our case, was the Elekta HexaPOD table. Any accessories of the six-dimensional table could block potentially the field of view so you have to keep that in mind and test it for your different locations.
For the X-ray system, there is the potential of parts of the gantry or even of the flat panels that could be in the X-ray field of view for a dual stereoscopic on a dual stereoscopic image acquisition. So, if the isocenter is too caudally or the speed of the X-ray image acquisition is too low, which rather is a thing that the gantry speed is too high, then you could have some parts of the gantry inside your X-rays. It is not a big deal, actually, because the software allows you to subtract some parts of the X-ray image where you don't want any registration to happen. What we saw for Elekta LINACs, the flat panels cannot be completely retracted if the isocenter is too caudal. That's something you have to measure for yourself. And if the flat panels are not retracted, they could be in the field of view of this case.
For the second part, we tested the isocenter coincidence for the surface, the X-ray, the cone-beam CT, and the MV field. That's the preliminary data I showed you before. For couch angles, I don't have the data here right now. We're still evaluating some of the results. But we will perform or we have performed a Winston-Lutz test for each couch angle, the so-called couch runout test. And we will combine it with a measurement of the difference between the surface/thermal camera as one system and the X-rays also. The relative and absolute error in shifts in a comparison between cone-beam CT and ExacTrac Dynamic. This is the second part I have already showed you. The impact of the system configuration is always possible to test, but we don't see the necessity at the moment. When it comes to skin tone, this is something you can do on a patient-specific basis. Also, the settings for the generator, for the X-ray generators is something you could do patient-specific, but there's potential rather not for commissioning, but more for research in the future.
The so-called warm-up effect. For those of you who are not aware of this, any surface guidance system has a warm-up effect. It's the time it needs to generate a static equilibrium inside the system itself. We wanted to test it but it's not so easy as the X-ray tubes needs a warm-up because you don't wanna destroy it with the first chart at the moment. You will have to shut down your whole system for 48 hours and then start right away with a surface camera acquisition. Again, this is not really a big problem because any potential warmers or the warm-ups effects or thermal drift effects of the surface camera could already be done by the time.
So, when it comes to the potential and the already proven benefits of the ExacTrac Dynamics, we have a couple of things we can look into. One of it is the automatic or the manual X-ray triggering. If you see any deviations between the actual surface and the surface you plan on, and if you detect them and you have a direct repositioning if you needed, this is a factor which can increase your time efficiency drastically. You see that here in this image on the lower-left side you have the two blue dots next to each other. These are the two X-ray tubes. When the gantry is in a position when you can make a stereoscopic X-ray shot, so to say. So, these are your planned X-rays for the fraction. And also sort of the green line on the bottom is the patient surface. And if you have a certain tolerance levels and if you exceed these tolerance levels, there's another X-ray to be taken at this exact moment.
A second drastic benefit is that you have surface motion monitoring throughout the entire fraction. This serves as an additional patient safety design because you do have automatic beam hold when your patient is out of tolerance. You always watch the patients on the right-top side at the moment. And as I said before, you monitor the patient throughout the entire fraction. A third point is you could potentially have less cone-beam CTs. If that is possible, less cone-beam CTs means less dose and also a faster workflow. You see that you always when you have potentially a good match according to the bony anatomy and not reliant on too much soft tissue contrast. You can significantly reduce your dose from the cone-beam CTs and increase your time.
Again, you have X-ray repositioning for each couch angle. This is a benefit which is also the older version of the ExacTrac has also proven is benefit. If you compare it to a surface-guided workflow only, you have to...or you monitor your couch kicks or your different couch angles. This would mean if you see a potential intrafractional movement or just any deviation after you turn your couch, you would have to reposition the patient at a couch degree of zero angle, zero degrees using cone-beam CT, and then shift the couch back. This could, again, result in some additional interaction or motion or even motion from residual motion or movement from the couch.
A big point in the surface guidance community is the reduction of the tattoos or skin markers. So, a clinic which can go full to too less or markless, even with a cone-beam CT workflow, that's when you use the pre-positioning workflow of the system, which you see here. You don't have to use any skin marks because the patient or the software allows you to position the patient according to his surface from the planning to and from your treatment planning system. And you automatically shift the patient to the desired very good pre-position. Afterwards you can do either cone-beam workflow or a full ExacTrac X-ray workflow, but you could potentially leave out one additional verification cone-beam CT image because the first cone-beam CT is also quite close to the actual position because you have a good pre-positioning. So, all in all, you can say that the whole dynamic workflow has a potential to increase the accuracy or for the intrafractional motion monitoring using the X-ray tracking and the safety using the X-ray tracking and the surface-guided workflow.
A quick outlook on what we will do next. For some palliative settings, we will reduce our numbers or amount of cone-beam CTs. We will hopefully completely replace our cone-beam city workflow which comes to...which our palliative patients will benefit a lot from. And if the soft tissue contrast actually matters, we will reduce the frequency of the cone of our daily cone-beam CTs at the moment and implement the new system regularly. When it comes to gating and Deep Inspiration Breath Hold, we are waiting on the next software update from Brainlab where we can have our Deep Inspiration Breath Hold left-sided breast cancer treatments. In the future, we will also look at liver SBRT when free-breathing gating or deep inspiration and exhalation breath-hold is feasible. And also lung SBRT in breath-hold is also one of the big parts we will try to look at in the future.
So, thank you, again, to Max Niyazi who did the first part of this presentation. Thank you, everyone. Thanks everyone for your attention. And we will be happy to ask any questions you have.
Bogdan: Thank you for your talk, Philipp. And before we go to questions, I would like to actually show you a complete cranial setup with ExacTrac Dynamic. This is, in effect, the second patient that the team of LMU has treated with the new system. And for those of you who have not had the chance yet to see ExacTrac Dynamic, this is a summary of the workflow from beginning to the end. To start, the patient is loaded on the LINAC which prompts ExacTrac Dynamic to open the same patient. Once the patient has opened an ExacTrac, the confirmation page pops up where the user can check that the correct patient and plan were indeed open. After confirming the patient, ExacTrac will take you to the pre-positioning screen. To begin, ExacTrac will automatically pre-position the patient at the treatment isocenter by aligning a 3D representation of the real-time patient surface picked up by the structural side cameras to the outer contour generated from the treatment planning CT. The calculated shifts are sent to the LINAC and applied by the coach.
Once the shifts are made, the user can proceed with defining an area of interest of the patient that is to be tracked by the thermal camera. Thermal camera tracks the patient's position throughout the entire treatment using the heat signature of the patient's body. After the area of interest is defined, the user will proceed with acquisition of X-ray images to verify the position of the patient. Once the X-ray images are acquired, the software automatically fuses the images to the DRRs generated by the software from the treatment planning CT. ExacTrac Dynamic has some cool tools for the users to analyze the fusion before shifts are made. Here you can see an example of the blending of the X-rays and DRRs using the toggle button.
Next, you can see an example of the rubberband feature where the DRR is slightly displaced from the X-ray and bounces back into position. Another nice feature is the region of interest. Here the user can paint out any anatomy that they don't want the software to consider during fusion. After a region of interest is defined, the user will redo the fusion of the modified X-ray and DRRs and analyze the fusion using the rubber band feature. This can be done in amber blue mode or gray value mode.
Once the user is happy with the fusion, the calculated shifts are once again sent to the LINAC and applied. After the shifts are applied, a set of verification X-ray images are taken to check that the shift was completed correctly and also that the patient did move during the shifts. Here the user will verify the fusion using the rubber band feature. After the patient's position has been verified with X-ray images, monitoring of the patient's treatment can begin. And you can see here the patient's monitoring screen. On the upper left-hand side is where you will see the X-ray images that will be automatically triggered by the software. Automatic X-ray imaging can either be triggered from amount units delivered or gantry position. And these are always user-defined settings.
To the right of the screen you see the life surface tracking of the patient. Once the surface is in the position and the software indicates this in the bottom-right-hand of the screen, the user can beam on and monitor the patient. On the bottom-middle of the screen, you can monitor the patient's real-time motion graph throughout the entire first arc. As long as the patient remains in the treatment position and no deviations are detected on the surface or X-ray images, the beam will remain on. After the first beam is treated, the software will prompt the user to move the couch to the next treatment position and a pair of X-ray verification images will be acquired to check the position of the patient at the new couch angle. The fusion is analyzed and modifications are made to the X-rays using the ROI feature. Then the fusion is completed again automatically with the modified X-rays. After the X-ray position is verified, the user is ready to monitor the patient during treatment and a beam on signal is sent from ExacTrac Dynamic to the LINAC.
This is a short arc treatment, so a set of X-ray images will be taken at the first Cardinal angle. All looks good. Both on surface. And for the internal anatomy, so the beam continues on until the treatment is completed. After all the beams have been treated, the patient is closed and the treatment data is automatically saved. For those of you have been familiar with the previous versions of ExacTrac, you can see that the core principles continue to apply, yet in this new version we are introducing a lot of automation between the console and ExacTrac that is always double-interlocked and improving the therapist's workflow and maintain confidence that the patient is at all times adequately positioned for beam on. More in-depth training is always available through our Brainlab Academy infrastructure. And should you require additional information, please check the brainlab.com page. And now let's go to Dr. Niyazi and Philipp Freislederer for the live question and answer session.
Thank you, both, for your presentations. And Dr. Niyazi, we have some questions and I think I'll start with you. I guess the first question is from our polling section related to immobilization devices. So, with more IGRT these days, we've seen a shift from invasive immobilization to mask systems and an introduction of open face masks in the last few years as well. So, for what kind of cranial indications? Do you use select open face masks? And are you still utilizing stereotactic head rings for anything in your practice? Yes.
[00:45:15]
[silence]
[00:46:21]
Bogdan: Right. Another question is regarding to the type of imaging that you do for your cranial treatment. So, we still see, I would say, 50% of people that responded are utilizing combination of ExacTrac and cone-beam CT. So, what kind of indications, radio surgical indications when you're gonna address point too? Do you still utilize cone-beam CT versus ExacTrac which is a combination of the two and what have you found so far in your practice?
[00:46:51]
[silence]
[00:48:24]
Bogdan: And speaking of intrafractional checks, what is your...what has been your workflow, I should say, for cranial treatments to verify intrafraction motion? And now with the new version of ExacTrac Dynamic that allows you to actually check for intrafraction motion automatically throughout the arc. Have you seen any cases so far where the patient actually moved outside of your tolerance throughout the arc?
[00:48:51]
[silence]
[00:50:00]
Bogdan: Thank you so much. Philipp, we have a few questions for you regarding commissioning and just some overall functionality of the system. So, maybe I'll start with some commissioning questions. And we do have two that are somewhat related pertaining to how you checked for rotational accuracy part of your commissioning. And then related to that, what would be the clinical implications of rotations and differences of almost 0.4 degrees between cone-beam and ExacTrac on rotation? Maybe you can address those.
Philipp: Okay. So, in general, what we did so far was a couch runout test with a simple Winston-Lutz pointer. And we took an MV flat panel image of pretty much every couch angle in a 10-degree steps. And then we even... We redid that for the new ExacTrac and we did some evaluations on the old ExacTrac. And we saw that our couch on another LINAC was quite all right, was not too bad. And most of the motion we have seen so far has come from actual intrafraction motion of the patient and not from the couch. In this case, we have a new couch. We are not 100% sure how good it is yet. If we have such a nice stable couch in the Netherlands, which we have another neck. So, we redo the test, but this is basically... If you have a Winston-Lutz pointer, you make ExacTrac images at the 10-degree steps from the couch and you make an MV image and you monitor your Winston-Lutz pointer.
When it comes to the deviation between cone-beam CT and the ExacTrac, I think I saw the question of 0.4 degrees if that's too big. It depends on what you believe. If you trust your cone-beam CT more or your ExacTrac, both your cone-beam CT and your ExacTrac have small deviation from your MV isocenter. We made a lot of efforts to have a very good radiation isocenter calibration on the ExacTrac system. So, I think on this particular LINAC as we do most of all the low-margin treatments with the ExacTrac, we trust the ExacTrac system or our calibration more in that case. So, the deviation doesn't really say too much. But we still want to see... You could also say the ExacTrac could be the gold standard and you can check your cone-beam CT using the ExacTrac. That's vice versa.
Bogdan: Okay. Thank you for that. And a question that maybe I'll ask both of you regarding the repositioning of the patient and the effects of rotation. So, maybe Dr. Niyazi, you can start with what type of shifts have you seen? What's the amplitude for the corrections that you've applied? And maybe address a little bit what kind of PTV margins you would put on these tumors especially if you're looking at single isocenter multiple PTV treatments?
[00:53:28]
[silence]
[00:56:24]
Philipp: Yeah. From the physical side we see the same, we think the same way. The way we measured our accuracy of the system with the old ExacTrac and with the new one. We see a cool wide tire accuracy is really quite sub-millimetric. You can actually say this might always sound like a marketing thing, but it is sort of submillimetric. And what you can do, you can actually monitor all the intrafractional motion of the patient inside the mask. So, you account for a lot of errors, which you would normally have. Again, you have some uncertainties as steeply said from the MR imaging. But we think with the current setup, at least for what I can say on Versa HD LINAC and with the ExacTrac and very high stability and reproducibility. And we are quite confident.
Bogdan: Dr. Niyazi, since you address the physical characteristics of Versa HD, you have a follow-up question on what is the smallest tumor volume that you treat? And maybe you can just address it in largest diameter, I guess. And link to that, honestly, what's the smallest field molecule that you measured for your treatment planning application?
[00:57:48]
[silence]
[00:59:12]
Philipp: We'll stick to the five-millimeter minimum.
Bogdan: Okay. We have questions, of course, about the new camera system. So, maybe Philipp, I'll start with you. What in your opinion do you consider to be the advantage of the thermal imaging over optical imaging?
Philipp: Well, one of the advantages was actually there's another question related to longitudinal deviation which was seen between the surface in the X-ray. This is the actual problem of surface imaging itself. If you look at it from one side and you see a tube, the tube can...the registration algorithm does not know how the tube moves in the longitudinal direction. And that's when the surface imaging from one side starts to fail or even from three sides starts to fail. If you add thermal imaging to it, the phantom or tube is maybe not the best example for it. But if you have it on a human being, the pelvic area on the patient is also sort of tube-shaped like and any surface imaging has some problems with it. If you have thermal imaging included, you sort of see the whole tube not as a tube, but more as like...which has something which has a lot of peaks and a lot of valleys inside. And that's something a registration algorithm can grab and hold on to. So, you have increased your accuracy for the surface registration. And that is the one main point of it.
The second point is if the thing that was the original Brainlab idea to have only one in-room camera and not two, or three, or four where you have gantry blocking and stuff like that again. So, you have one camera, and thanks to the thermal imaging, a very high position. And the results, which I showed with the 0.9-millimeter or something or 0.7-millimeter in the longitudinal direction, these came from...because that's when you compare the surface imaging only with the ExacTrac. And we have very... In the second data subsets with couch rotation, we compared surface and thermal imaging with X-rays. And that's basically it.
Bogdan: Yeah. Regarding the application of the SGRT solutions of ExacTrac Dynamic, maybe you can speak a little bit about what is the precision and accuracy for the SGRT components? And what kind of tolerances do you utilize in your workflows to trigger either a beam hold or an X-ray image? And maybe you can also address then the linking between the SGRT and the X-ray imaging or whether you can just monitor with SGRT.
Philipp: Yeah. So, we use, for every patient site, a different site we'll use a different subset of templates. So, if you use cranial SRS, we have a 0.5 millimeter and 0.5-degree tolerance level for the X-rays and 0.7 millimeter and 0.7 degrees tolerance level for the surface imaging. A little high on the surface because the ground truth, in this case, would definitely be the X-rays. And that's also what the system makes the ground truth is always the X-rays. So, you cannot yet do a surface-only workflow. You might do it in the future. The system always needs the X-rays to begin with. I think that sort of historically happened because the ExacTrac Dynamic is an X-ray-based system. And I think as far as I've talked to the developers from Brainlab, everyone was implementing the surface workflow. And after testing the system, everyone was quite happy that the surface guidance does indeed have a very high accuracy if you would do it surface only.
So, maybe in the future, there will be a surface-guided workflow that's, I'm sure this might be helpful for some patients. But if you have image guidance at your clinic implemented for all of your patients and you could reduce the number of cone beams or patients end up using MV beams and flat panels. You would not think of... And you can do that with the X-rays. I don't know if there's really a need for a surface guidance only, but it is possible. The accuracy of the system itself is a little hard to test as I already said of the surface system only, the surface and thermal system because a lot of the things are done internally. And so most of the tests are only to be done for the whole system. But as the... What we saw in some tests that the stability is quite good and also that the surface and thermal camera are in sort of the same accuracy levels as the X-rays for the patients in this case.
Bogdan: Thank you for that. And some practical questions related to what you just mentioned. What is the largest deviation that you've seen picked up by the camera system? And how do you utilize the camera setup, so the SGRT setup in conjunction with either close masks or open face masks?
Philipp: So, the largest deviations we see, it depends really on the patient. When it just comes to numbers, the largest deviations were from the service camera were from a stereotactic treatment, the one we saw in the presentation where we actually saw the breathing motion of the patient. So, those are deviations to up to... I don't know. I don't have the value in my head right now. But they were actually the maximum you can use for surveillance. We monitor open face mask and we also monitor the face mask, an enclosed face masks themselves. Just to give an... I have an extra safety level also for the closed face masks. The open face mask would actually be a monitoring of the movement inside the mask. And monitoring the closed face mask also bring some information because the heat signature can be seen through the face mask and that's another good thing for the internal registration algorithm.
In general, I would say we've seen a lot of high deviations when it comes to couch rotations where we correct for but we sometimes also have movement in the mask for stereotypic treatments of a multiple brain met patient. If the patient is on the couch for 40 minutes, 35, 40 minutes, something like that, there will be some motion. And this is something where we start... We have also stopped the treatment and we position the patient according to the X-rays because that's actually what the system does. If the surface goes out of your desired motion range, which you can determine at the beginning. You can choose if you want to stop treatment, if you want to trigger an X-ray, or if you want to do nothing. That's actually up to you. And it also depends on the specific patient site.
Bogdan: And Philipp, a question that we probably get every time we present this hybrid camera system between structural lights and thermal. And let me know if you're comfortable answering this. But the question is on the stability of the thermal signature if you need to reacquire it. So, maybe you can address a little bit how the two camera systems work together in terms of what's absolute versus what's relative.
Philipp: So, as far as I have understood it, because you don't really have inside what...you don't see what the thermal camera or the system does that internally. But as far as we've seen it, the two systems are calibrated together, thermal and the surface camera. And if you take one snapshot at the beginning of the treatment or at each couch rotation of the actual thermal signature, and this signature is quite stable at least the next 10 minutes or something. If you would have a large drift, the system would tell you that you have a large drift and that the monitoring is going in a weird direction. But also if you have a large drift, you take another X-ray system and that's the point where also the thermal and the surface actually are moved to the position of the X-ray. So, the stability of the thermal signature is not a big problem also throughout different fraction. The thermal signature is acquired before each fraction individually.
Bogdan: Right. And again, for everybody, every time you acquired the new set of X-ray images, the surface information is rewritten, so it becomes essentially a relative measure.
Philipp: That would have been the short answer to mine.
Bogdan: Have you had any experience with either skin tones, variations, or room lighting with the new camera system?
Philipp: We didn't have any room light deviations. We didn't see any effects. Although we have some fancy room lights inside, we didn't see that at all. But regarding skin tones, this is something to be tested because structured light or the measurement of structured light always relies on the reflection of the light and the body. And the skin tone does affect the system in general. You have certain settings for your skin tones. You can set it to bright or darker skin tones, which you can...but it doesn't have a big effect because you have the thermal information as well. So, in any surface guidance systems, you as the skin tone has quite high effect on the patient position or on what the camera can actually see because of the reflection settings
Bogdan: Dr. Niyazi, we seem to have quite a few questions regarding margins. And we have addressed this in other webinars, so I would invite all of you to perhaps review some of the other webinars that we had on brain metastases treatments and setups. But perhaps you can address this a little bit with what you do in your practice. And if you do nonlinear margins, margin definitions based on distance to the isocenter, and maybe address what those values are.
[01:10:46]
[silence]
[01:14:31]
Bogdan: Gentlemen, thank you very much for your great presentations. Obviously, they sparked interesting conversations too. Thank you, everybody, for joining the webinar. And if we missed any questions, please direct them on the Novalis Circle forum. We can answer them there for you. And stay safe and we'll see you at the next webinar.
Philipp: Thank you so much.
Bogdan: Thank you. Bye, bye.
Philipp: Bye.
[00:00:31]
[music]
[00:00:55]
Bogdan: Hello, everyone, and welcome to a new Novalis Circle webinar. My name is Bogdan Valcu. I'm the director of Novalis Circle. And today I have the pleasure of introducing Prof. Maximilian Niyazi and Philipp Freislederer from LMU in Munich who together will be discussing their initial clinical experience with ExacTrac Dynamic. Prof. Niyazi is the vice-chair in the Department of Radiation Oncology at Ludwig Maximilians University in Munich. He is also the deputy coordinator for the Neuro-Oncology Center. And Philipp Freislederer is the scientific employee in the Department of Radiation Oncology and he serves as a medical physicist. As some of you may remember, the team at LMU was the first one in Europe to utilize the ExacTrac Dynamic for initial treatments back in April, but since then, a more comprehensive work has been put in place to commission, validate, and re-establish ExacTrac for a classical radiosurgical treatments in the brain and spine.
In the webinar, today, we will review the benefits of the new camera system and its advantages that it brings for patient setups. We will touch on the IGRT benefits that ExacTrac Dynamic brings for cranial and spine radiosurgical setups, discuss the clinical implications of our IGRT system in the clinical practice and review relevant commissioning procedures. Ever since the release of ExacTrac Dynamic back in April of 2020, we have begun adding new systems to the more than 1000 install base of previous ExacTrac versions with 17 systems that are currently clinical in Europe and a total of almost 58 systems to be installed by the end of the year in Europe, North America, and Asia Pacific. And as more of you have an opportunity to install the new system, we want to provide you with scientific relevant data, clinical visibility, and commissioning details to aid you in the implementation of the system in your department. Stay tuned for ongoing information on ExacTrac Dynamic clinical applications. And the next webinar that we have planned during ASTRO time will continue to cover not only cranial applications, but also extra cranial indications.
As always, we continue to provide CE credits for those of you that successfully complete the course. And should you have any questions regarding your CAMPEP, MDCB, ASRT credits, please email us at info@novaliscircle.org. And don't forget to sign up for our upcoming webinar on September 30th. That would actually mark the introduction of a case of the month series where we will begin reviewing interesting cases at some of our partner sites. There's Macyszyn, Kaprealian, Agazaryan at UCLA will be reviewing a spine patients during the introductory webinar and we'll hope to see you online.
For today's webinar, please remember to either utilize Google Chrome or Safari. And should you have any connectivity issues with your internet connection, simply refresh the window. Utilize the chat interface to send us questions. Upon completion of the two lectures, we will answer your questions in a live session, monitor the polling interface for questions that we may like to ask you and should you follow us on social media, please utilize #NovalisCircle. And I'd like to turn it over to Dr. Niyazi for the first lecture.
Prof. Niyazi: So, dear ladies and gentlemen, so I'm very grateful to give today's Novalis' webinar on our first clinical experiences with ExacTrac Dynamic. My name is Maximilian Niyazi. I'm the vice chairman of the Department of Radiation Oncology at the LMU University Hospital in Munich. I'm very happy to give this talk together with my dear colleague, Philipp Freislederer, who will stop talk about the physics aspects afterwards. So, these are our disclosures.
So, the LMU Munich and our department has about 2400 patients per year, has an equipment of 6 LINACs, including MRIdian LINAC and 2 ExacTracs at 2 different campuses. We have the staff of around 130 persons. And as you can see, we try to install the ExacTrac Dynamic system to enhance our imaging capabilities. And the nice thing about ExacTrac Dynamic is that you have more or less four dimensions. You have the surface of the patient, the thermal information, and you have X-ray information. That's what you all know, but the thermal component is the new one. And the good thing is that it is more or less everything integrated. And in this sense, you do not need multiple in-room cameras to track the surface. You can do this just by one imaging instance. And the other nice feature about ExacTrac Dynamic is that you have additional real-time tracking using a real SGRT.
And that's actually a nice first starting image. And that's my dear colleague, Philipp, who is actually included here. That's the thermal information. And that's the two-dimensional surface tracking. So, it's a very nice sketch of my colleagues. And the indications from the sketch starting with our program in June of this year was that we introduced treatment types that we have already applied on our old ExacTrac. So, we have included multiple Brainlab SRS and every kind of hyperfractionated SRS in the brain. We introduced fractionated stereotactic radiotherapy with open face masks, for example, in meningioma established for normal glioma. We replaced the cone-beam workflow in palliative settings and for spine SRS, you know, running a prospective study. And of course, you can treat every bone oligometastatic disease anywhere else in the body.
So, I will talk a bit about the workflow of the ExacTracs Dynamic. And the nice thing is it's a fully integrated system. With LINAC, you have the automatic plane loading from the RV system. Pre-positioning itself is done by using the outer contour of the planning CT and it sends shifts to the LINAC. So, that's the situation when the patient is lying on the table with an open face mask. So, there's the thermal information, the pre-positioning that is done using the outer contour as defined. And afterwards, you select the area of interest for your surface tracking. In this case, of course, it's the image that's outside of the thermoplastic mask system. And here we have your classical VRRs with the registration of the X-ray images from orthogonal directions. And you have very nice tools to, for example, have your control over the match. And if you're, for example, satisfied with that, you can send the changes. And then you start with the monitoring workflow. If you enable the X-ray imaging and the surface tracking, you have to choose certain settings for your tolerances. For example, the X-rays shift tolerances and the surface tracking tolerances. And you normally would choose them a bit higher because of higher deviations expected from the surface itself.
In this case, we had an acoustic neuroma and there was a very nice and stable situation and everything was intolerance. And the X-ray imaging in a stereoscopic fashion is done at the 90-degree angles, 180, 270 and zero-degree. And as you can see, as far as you have such an out-of-tolerance information, then you have to check for the next stereoscopic image and then you have to adjust. And that may happen, for example, and for instance, in this case, if something of the gantry, for example, comes into this small field of view. So, our experience up to now is that the first patient was treated on June 2nd of this year. We've treated 20 patients, 16 brain tumor patients, 2 spine, and 2 head and neck cases. So, that's not too bad. And there is a clinical example in this case of correction of non-coplanar fields which is quite important for stereotactic radiosurgery. And as you can see here, you have the situation that your angle is actually approaching the 90 degrees here and they view X-ray imaging. Everything is nice within tolerances and then just continue. And as you can see here for the registration you can swipe out some structures that are more deformable, and in this case, it was very smooth treatment.
Okay. So, something of, let's say, preliminary information on stereotactic radiosurgery offers excellent local control and pain reduction that's a nice overview of Tseng 2017. You have really high rates of local controls irrespective of prospective or retrospective design trials. And you have very fast pain reduction compared to conventional fractionation. So, it's a very nice method being used in oligometastatic disease. And there are ASTRO guidelines as well with very precise information on, for example, support tolerances and other organs at risk. And Sahgal was one of the leaders in this field. And that's actually something that is used in the brain lip software that comes up within consortial target definition of spine volumes, which are, for example, referring to the lesion size where it's located and whether to include the pedicles, process transverse, spine process, or the laminae. So, it's very important that this target volume is actually clearly defined by the guidelines. And the software automatically delineates the vertebral body and the structures of the target volumes according to the lesion that is placed here. And the best thing about software is that you can co-register your MRI in a fashion that the deformation is actually used.
As you can see here, you have your deformation done from the MRI to the CT. And then you have really this kinking of the vertebral column that works pretty nicely. And afterwards, you have your classical overview of the target. You can choose the prescription doses. You can normalize here to the whole PTV and you can use the primer lesion within the vertebral body. And you can choose among different features how to shape the dose with a VMAT technique. So, that's an integrated workflow, nicely fitting for the ExacTrac Dynamic. And that's the first case actually not with the Brainlab software because we're right now installing it. It's uterine leiomyosarcoma metastasis. Initial presentation was in 2015. And we went resection and we went chemo. Had been sacral metastasis, resection, again and we went chemotherapy. And later on three years later we had destructive process of T9 and the humerus. This was resected and we had palliative conventional or moderately fractionated treatment with 3 gray times 12. And seven months later, we had the massive progression of this sarcoma metastasis, and we treated it with the VMAT and the spine with the Elekta software Monaco. And you have a very, very steep gradient to the code because of this previous irradiation just seven months ago with five times four in six gray. And in this case, we use the very tight margins.
And as you can see here, we had, of course, the problem if you have the surface here on the stomach, for example, on the abdomen, then you have high margins that you should choose for the surface tracking because as you detect breathing motion, it doesn't make too much sense if you would like to track the spine. And if we had very tight margins for X-ray surveillance with 0.7 millimeters and 0.5 degrees and use the highest frame rate for X-ray imaging possible in this case, only the stereoscopic views with 4 per arc. And the surveillance was done by X-ray imaging. As you can see, it worked out very nicely. Patient could treat it just some weeks ago.
So, the other experience was on the multiple brain mets. We have a high-resolution contrast-MRI that we're using for target definition. Sample rate sequence, the slice thickness of 1 millimeter with a double-layered mask, planning-CT with contrast, dominates within a week of the MRI with a GTV that is contoured and checked on MRI and planning-CT and 1 millimeter of PTV margin is added and we're treating up to 10 test disease and in some cases even more but on trial with 4 to 10 metastasis and comparing this to historical whole brain [inaudible 00:14:31] from our department. And as you can see here. And now this is a nice software tool of Brainlab that is doing the multiple brain met SRS treatment. You have, again, your nice panel where you can delineate the lesions. And here you can have a nice overview on the prescription dose. And if you click on the single lesions, you will get feedback on the conformity index, the gradient index, the prescription dose, min-max and min values that makes it very easy to assess whether this plan is suitable to be treated.
And concerning another clinical example, we had a patient with a non-small cell lung cancer that was EGFR-mutant with a PD-L1 smaller than 1%. Initially treated by VATS pleurectomy and pleurodesis. Received afatinib and was switched to carboplatin/pemetrexed and liquid biopsy was negative and EGFR mutation had a massive progressive disease. The third line nivo treatment and within the latest MRI she had four metastases and now we're treating all these four metastases with one single isocenter treatment. And now I'd like to hand over to my dear colleague, Philipp, who will explain what the potential sources of errors are in this case and, of course, on all the commissioning data of ExacTrac Dynamic and his experience. So, thank you so much for your attention.
Bogdan: Thank you for your talk, Dr. Niyazi. And I would like to ask Philipp Freislederer to continue with his talk.
Philipp: Hello, everyone, also from my side. Thank you, Max, for the introduction. My name is Philipp Freislederer. I am a medical physicist here at the Department of Radiation Oncology. And I'm now gonna talk a little bit about the error sources in cranial SRS treatments and then go quickly over some issues we potentially avoided in the cases and the clinical examples Prof. Niyazi showed and then talk about commissioning a little bit at the end.
In general, when you do cranial SRS, you have dosimetry-related errors. So, your output factors for small fields below two centimeters in your MSC calibration which affects your output factors. You also have errors related to the imaging. So, for example, geometrical distortions in the MR images. And also, which is quite important, the calibration of the imaging and the radiation isocenter. This is something extremely important for the ExacTrac Dynamic as we, in our case, switched from a cone-beam CT workflow for most of the patients we have there to a full ExacTrac workflow and we placed all the cone-beam CT for these types of patients. And also you do have mechanically-related errors. So, any residual intro-factual motion inside the immobilization masks. You also have runouts of the gantry, runouts of your collimator, and runouts of the couch. And when it comes to SRS treatments cranial, you do have a lot of couch rotations, which leads also to couch runouts. And this is something you could avoid with ExacTrac and ExacTrac Dynamic.
In general, all of the patients we treat for the multi-metastasis cases, we treat with a single isocenter. If you look back from how you might use to have done it, going from an isocentric stereotactic treatment in the head, you will have one isocenter per treatment target. And it means if you have some small translations that might affect your 80% isodose in relationship to the PTV margin, which you see here a little bit. Rotations do not affect for spherical targets. Mutations do not affect those coverage too much. And now if we switch to monocentric treatment, so treatment of multiple metastases with one isocenter. If you, again, have some translations, you might have some effect on your dose on your PTV coverage. But if you have rotations, which is in the right button, small rotations can have a huge impact depending on either the distance between the targets, the rotational error or the target volumes. So, we have a high-risk of compromised coverage.
If you go back to our clinical example Prof. Niyazi showed, in this case, we took one correctional X-ray image before each couch angle. What we already have seen for this one patient we have evaluated the data with the older version of the ExacTrac for multiple patients, but for this one patients, we saw errors in yaw for up to one degree. And all of these errors could have...were compensated using the ExacTrac or the ExacTrac Dynamic in this case. And we might have had a significant deviation from the treatment plan. So, this is the one... This shows the importance of tracking your couch or tracking the movements of the patient.
Now, quickly from this topic to QA and commissioning. As I think everybody knows, no direct guideline or recommendation is available yet for the commissioning of such a system, there are no guidelines available for surface-guided radiation therapy only systems. And this is a hybrid machine with X-rays and surface guidance. It is quite tough to sort of know what you do or what do you need to do at the beginning. When we start commissioning a system like this, we have some initial thoughts when it comes to surface guidance only. So, we want to test multiple things, static accuracy, dynamic accuracy, the impact of different region of interests, also the impact of your field of view. You want to check if there's any blocking of the gantry, blocking of the accessory. But it's quite important for our stereotactic treatments and all other treatments as well is the coincidence of the different isocenters, so the isocenter of your surface guidance system, your X-rays, your cone-beam CT, and most importantly, those have to be coincident with the MV beam.
When you perform treatments with couch rotations, you also have to commission the system or test what happens when you rotate the couch. Another interesting point for us was the relative and absolute error in shifts when you compare cone-beam CTs with the ExacTrac Dynamic. This was important for us because we sort of took away all of the cone-beam CT workflows for our patients which we are treating at the ExacTrac Dynamic. And the impact of system configurations might be also of importance, but this is something which is already done pretty well automatically by the company. But we will test these configurations in the future quite extensively. The so-called warm-up effect of the system or the thermal drift of the system is also part of it. I have some words on this later on.
To start off, we'll be doing for routine QA for the ExacTrac Dynamic. There's a daily check implemented by Brainlab. It takes a maximum of five minutes in total. In the steady checks, two things are tested, one is the deviation between the surface camera and the X-ray positioning system. So, it's the so-called consistency check between the two systems in one. And the other one is the deviation from the radiation isocenter. Once a month you have to do a thermal to 3D surface calibration. This also takes five minutes. There is a dedicated phantom for it. We do recommend a monthly calibration of the radiation isocenter, which is simply the placement of any size of a ball bearing inside the radiations isocenter. You could do that either with Winston-Lutz pointer or with any type of ball bearing inside any anthropomorphic phantom. As we do a lot of stereotactic treatments, we perform weekly SRS checks mostly on the day where we perform the actual treatment. This is done like... It's actually pretty straightforward. You move anthropomorphic phantom into the radiation isocenter using the ExacTrac Dynamic. In the isocenter of this phantom there is, again, a ball bearing and you check the position of the ball hearing using your MV beam.
So, this is at the moment most of the routine QA we're doing. We will add something in the future when we add more types of treatment sites. For commissioning, we wanted to first see how the surface and the X-ray positioning fit together. So, we used a Brainlab phantom called Lola. The phantom has a heat signature inside, so you can actually use the surface and the thermal tracking. We perform 10 different random isocentric couch deviations and translation and rotation for couch zero degrees to simulate isocentric uncertainties. And we perform for only two couch angles. I think a colleague of mine is at the moment using the same test for more couch angles for pretty much the 180-degree couch angle in 10 degrees steps. But for the first commissioning tests, we wanted to stick to two only to simulate intrafactional movement on a different couch angle. And then we had a deviation between the X-ray. So, the delta D position is deviation between the results we got it from the X-ray shift and the surface shift. For couch zero, everything was below one millimeter. One little higher impact had the longitudinal direction. And for the couch 60 and 350 degrees, we wanted to check on intrafractional monitoring. These shifts where all shifts were below 0.5 millimeters. So, the system has an accuracy below or submillimetric accuracy when you compare the two entities of the system together.
The second part we did where we can show some data already is the verification and positioning accuracy. When it comes to ExacTrac Dynamic versus a cone-beam CT workflow. We use two anthropomorphic phantoms with six random isocenter locations each. We set up the phantoms according to ExacTrac Dynamic and according to cone-beam CT and then position the phantom according to ExacTrac Dynamic and afterwards to the verification ExacTrac Dynamic image, X-ray image, and a verification cone-beam CT image. So, the deviation for the initial correction was between the two entities, ExacTrac Dynamic and cone-beam CT, was again very low and maybe a little bit higher in rotation. And when it comes to verification, those two systems agree very well for a very static non-moving phantom. That's what you have to add in this case.
When it comes to all the other initial thoughts on QA and commissioning, I wanna go through the same list I showed you before again to see what our thoughts were and what our results were. So, for static accuracy, we wanted to test it as an SGRT system only. This is not possible using the ExacTrac Dynamic because the baseline for the accuracy or rather the baseline for the positioning result is always the X-rays. So, the static accuracy of the ExacTrac Dynamic surface camera only cannot really be tested. It is also not as important as the accuracy and a static case relies only on the X-rays. So, you rather have to test for the accuracy of the X-ray positioning, which we did when we looked at the...which we did directly when we looked at the comparison between the surface and the X-ray system.
When you want to test the dynamic accuracy, this is something you might want to do with a surface guidance system also. This is possible but in our case, we are kind of hoping that the next version brings a little physics mode because the data, if you want to store the data, you can only store it for Beam ON. This is something that I'm sure it's gonna be made a little bit more open in the future, but you have to know that if you want to store any dynamic accuracy tests, like, the test is quite simply. You put a phantom on the table and then you watch how the surface is drifting.
The impact of the region of interest is always hard to tell on any phantom surface. We will need some patient data to do that. It's not really essential for commissioning, but it's something we would like to know in the future. The field of view is quite easy to test or not to test. It's just the look of the field of view. When it comes to gantry and accessory blocking, there are two parts of it. The surface camera does not have any blocking because the registration is done using one camera only and that's also the reason why you have the thermal camera. The thermal camera is the reason why you only need one camera inside the room and not three. This means that you don't have any gantry blocking off the surface camera which is a good thing. But any accessories of the...in our case, was the Elekta HexaPOD table. Any accessories of the six-dimensional table could block potentially the field of view so you have to keep that in mind and test it for your different locations.
For the X-ray system, there is the potential of parts of the gantry or even of the flat panels that could be in the X-ray field of view for a dual stereoscopic on a dual stereoscopic image acquisition. So, if the isocenter is too caudally or the speed of the X-ray image acquisition is too low, which rather is a thing that the gantry speed is too high, then you could have some parts of the gantry inside your X-rays. It is not a big deal, actually, because the software allows you to subtract some parts of the X-ray image where you don't want any registration to happen. What we saw for Elekta LINACs, the flat panels cannot be completely retracted if the isocenter is too caudal. That's something you have to measure for yourself. And if the flat panels are not retracted, they could be in the field of view of this case.
For the second part, we tested the isocenter coincidence for the surface, the X-ray, the cone-beam CT, and the MV field. That's the preliminary data I showed you before. For couch angles, I don't have the data here right now. We're still evaluating some of the results. But we will perform or we have performed a Winston-Lutz test for each couch angle, the so-called couch runout test. And we will combine it with a measurement of the difference between the surface/thermal camera as one system and the X-rays also. The relative and absolute error in shifts in a comparison between cone-beam CT and ExacTrac Dynamic. This is the second part I have already showed you. The impact of the system configuration is always possible to test, but we don't see the necessity at the moment. When it comes to skin tone, this is something you can do on a patient-specific basis. Also, the settings for the generator, for the X-ray generators is something you could do patient-specific, but there's potential rather not for commissioning, but more for research in the future.
The so-called warm-up effect. For those of you who are not aware of this, any surface guidance system has a warm-up effect. It's the time it needs to generate a static equilibrium inside the system itself. We wanted to test it but it's not so easy as the X-ray tubes needs a warm-up because you don't wanna destroy it with the first chart at the moment. You will have to shut down your whole system for 48 hours and then start right away with a surface camera acquisition. Again, this is not really a big problem because any potential warmers or the warm-ups effects or thermal drift effects of the surface camera could already be done by the time.
So, when it comes to the potential and the already proven benefits of the ExacTrac Dynamics, we have a couple of things we can look into. One of it is the automatic or the manual X-ray triggering. If you see any deviations between the actual surface and the surface you plan on, and if you detect them and you have a direct repositioning if you needed, this is a factor which can increase your time efficiency drastically. You see that here in this image on the lower-left side you have the two blue dots next to each other. These are the two X-ray tubes. When the gantry is in a position when you can make a stereoscopic X-ray shot, so to say. So, these are your planned X-rays for the fraction. And also sort of the green line on the bottom is the patient surface. And if you have a certain tolerance levels and if you exceed these tolerance levels, there's another X-ray to be taken at this exact moment.
A second drastic benefit is that you have surface motion monitoring throughout the entire fraction. This serves as an additional patient safety design because you do have automatic beam hold when your patient is out of tolerance. You always watch the patients on the right-top side at the moment. And as I said before, you monitor the patient throughout the entire fraction. A third point is you could potentially have less cone-beam CTs. If that is possible, less cone-beam CTs means less dose and also a faster workflow. You see that you always when you have potentially a good match according to the bony anatomy and not reliant on too much soft tissue contrast. You can significantly reduce your dose from the cone-beam CTs and increase your time.
Again, you have X-ray repositioning for each couch angle. This is a benefit which is also the older version of the ExacTrac has also proven is benefit. If you compare it to a surface-guided workflow only, you have to...or you monitor your couch kicks or your different couch angles. This would mean if you see a potential intrafractional movement or just any deviation after you turn your couch, you would have to reposition the patient at a couch degree of zero angle, zero degrees using cone-beam CT, and then shift the couch back. This could, again, result in some additional interaction or motion or even motion from residual motion or movement from the couch.
A big point in the surface guidance community is the reduction of the tattoos or skin markers. So, a clinic which can go full to too less or markless, even with a cone-beam CT workflow, that's when you use the pre-positioning workflow of the system, which you see here. You don't have to use any skin marks because the patient or the software allows you to position the patient according to his surface from the planning to and from your treatment planning system. And you automatically shift the patient to the desired very good pre-position. Afterwards you can do either cone-beam workflow or a full ExacTrac X-ray workflow, but you could potentially leave out one additional verification cone-beam CT image because the first cone-beam CT is also quite close to the actual position because you have a good pre-positioning. So, all in all, you can say that the whole dynamic workflow has a potential to increase the accuracy or for the intrafractional motion monitoring using the X-ray tracking and the safety using the X-ray tracking and the surface-guided workflow.
A quick outlook on what we will do next. For some palliative settings, we will reduce our numbers or amount of cone-beam CTs. We will hopefully completely replace our cone-beam city workflow which comes to...which our palliative patients will benefit a lot from. And if the soft tissue contrast actually matters, we will reduce the frequency of the cone of our daily cone-beam CTs at the moment and implement the new system regularly. When it comes to gating and Deep Inspiration Breath Hold, we are waiting on the next software update from Brainlab where we can have our Deep Inspiration Breath Hold left-sided breast cancer treatments. In the future, we will also look at liver SBRT when free-breathing gating or deep inspiration and exhalation breath-hold is feasible. And also lung SBRT in breath-hold is also one of the big parts we will try to look at in the future.
So, thank you, again, to Max Niyazi who did the first part of this presentation. Thank you, everyone. Thanks everyone for your attention. And we will be happy to ask any questions you have.
Bogdan: Thank you for your talk, Philipp. And before we go to questions, I would like to actually show you a complete cranial setup with ExacTrac Dynamic. This is, in effect, the second patient that the team of LMU has treated with the new system. And for those of you who have not had the chance yet to see ExacTrac Dynamic, this is a summary of the workflow from beginning to the end. To start, the patient is loaded on the LINAC which prompts ExacTrac Dynamic to open the same patient. Once the patient has opened an ExacTrac, the confirmation page pops up where the user can check that the correct patient and plan were indeed open. After confirming the patient, ExacTrac will take you to the pre-positioning screen. To begin, ExacTrac will automatically pre-position the patient at the treatment isocenter by aligning a 3D representation of the real-time patient surface picked up by the structural side cameras to the outer contour generated from the treatment planning CT. The calculated shifts are sent to the LINAC and applied by the coach.
Once the shifts are made, the user can proceed with defining an area of interest of the patient that is to be tracked by the thermal camera. Thermal camera tracks the patient's position throughout the entire treatment using the heat signature of the patient's body. After the area of interest is defined, the user will proceed with acquisition of X-ray images to verify the position of the patient. Once the X-ray images are acquired, the software automatically fuses the images to the DRRs generated by the software from the treatment planning CT. ExacTrac Dynamic has some cool tools for the users to analyze the fusion before shifts are made. Here you can see an example of the blending of the X-rays and DRRs using the toggle button.
Next, you can see an example of the rubberband feature where the DRR is slightly displaced from the X-ray and bounces back into position. Another nice feature is the region of interest. Here the user can paint out any anatomy that they don't want the software to consider during fusion. After a region of interest is defined, the user will redo the fusion of the modified X-ray and DRRs and analyze the fusion using the rubber band feature. This can be done in amber blue mode or gray value mode.
Once the user is happy with the fusion, the calculated shifts are once again sent to the LINAC and applied. After the shifts are applied, a set of verification X-ray images are taken to check that the shift was completed correctly and also that the patient did move during the shifts. Here the user will verify the fusion using the rubber band feature. After the patient's position has been verified with X-ray images, monitoring of the patient's treatment can begin. And you can see here the patient's monitoring screen. On the upper left-hand side is where you will see the X-ray images that will be automatically triggered by the software. Automatic X-ray imaging can either be triggered from amount units delivered or gantry position. And these are always user-defined settings.
To the right of the screen you see the life surface tracking of the patient. Once the surface is in the position and the software indicates this in the bottom-right-hand of the screen, the user can beam on and monitor the patient. On the bottom-middle of the screen, you can monitor the patient's real-time motion graph throughout the entire first arc. As long as the patient remains in the treatment position and no deviations are detected on the surface or X-ray images, the beam will remain on. After the first beam is treated, the software will prompt the user to move the couch to the next treatment position and a pair of X-ray verification images will be acquired to check the position of the patient at the new couch angle. The fusion is analyzed and modifications are made to the X-rays using the ROI feature. Then the fusion is completed again automatically with the modified X-rays. After the X-ray position is verified, the user is ready to monitor the patient during treatment and a beam on signal is sent from ExacTrac Dynamic to the LINAC.
This is a short arc treatment, so a set of X-ray images will be taken at the first Cardinal angle. All looks good. Both on surface. And for the internal anatomy, so the beam continues on until the treatment is completed. After all the beams have been treated, the patient is closed and the treatment data is automatically saved. For those of you have been familiar with the previous versions of ExacTrac, you can see that the core principles continue to apply, yet in this new version we are introducing a lot of automation between the console and ExacTrac that is always double-interlocked and improving the therapist's workflow and maintain confidence that the patient is at all times adequately positioned for beam on. More in-depth training is always available through our Brainlab Academy infrastructure. And should you require additional information, please check the brainlab.com page. And now let's go to Dr. Niyazi and Philipp Freislederer for the live question and answer session.
Thank you, both, for your presentations. And Dr. Niyazi, we have some questions and I think I'll start with you. I guess the first question is from our polling section related to immobilization devices. So, with more IGRT these days, we've seen a shift from invasive immobilization to mask systems and an introduction of open face masks in the last few years as well. So, for what kind of cranial indications? Do you use select open face masks? And are you still utilizing stereotactic head rings for anything in your practice? Yes.
[00:45:15]
[silence]
[00:46:21]
Bogdan: Right. Another question is regarding to the type of imaging that you do for your cranial treatment. So, we still see, I would say, 50% of people that responded are utilizing combination of ExacTrac and cone-beam CT. So, what kind of indications, radio surgical indications when you're gonna address point too? Do you still utilize cone-beam CT versus ExacTrac which is a combination of the two and what have you found so far in your practice?
[00:46:51]
[silence]
[00:48:24]
Bogdan: And speaking of intrafractional checks, what is your...what has been your workflow, I should say, for cranial treatments to verify intrafraction motion? And now with the new version of ExacTrac Dynamic that allows you to actually check for intrafraction motion automatically throughout the arc. Have you seen any cases so far where the patient actually moved outside of your tolerance throughout the arc?
[00:48:51]
[silence]
[00:50:00]
Bogdan: Thank you so much. Philipp, we have a few questions for you regarding commissioning and just some overall functionality of the system. So, maybe I'll start with some commissioning questions. And we do have two that are somewhat related pertaining to how you checked for rotational accuracy part of your commissioning. And then related to that, what would be the clinical implications of rotations and differences of almost 0.4 degrees between cone-beam and ExacTrac on rotation? Maybe you can address those.
Philipp: Okay. So, in general, what we did so far was a couch runout test with a simple Winston-Lutz pointer. And we took an MV flat panel image of pretty much every couch angle in a 10-degree steps. And then we even... We redid that for the new ExacTrac and we did some evaluations on the old ExacTrac. And we saw that our couch on another LINAC was quite all right, was not too bad. And most of the motion we have seen so far has come from actual intrafraction motion of the patient and not from the couch. In this case, we have a new couch. We are not 100% sure how good it is yet. If we have such a nice stable couch in the Netherlands, which we have another neck. So, we redo the test, but this is basically... If you have a Winston-Lutz pointer, you make ExacTrac images at the 10-degree steps from the couch and you make an MV image and you monitor your Winston-Lutz pointer.
When it comes to the deviation between cone-beam CT and the ExacTrac, I think I saw the question of 0.4 degrees if that's too big. It depends on what you believe. If you trust your cone-beam CT more or your ExacTrac, both your cone-beam CT and your ExacTrac have small deviation from your MV isocenter. We made a lot of efforts to have a very good radiation isocenter calibration on the ExacTrac system. So, I think on this particular LINAC as we do most of all the low-margin treatments with the ExacTrac, we trust the ExacTrac system or our calibration more in that case. So, the deviation doesn't really say too much. But we still want to see... You could also say the ExacTrac could be the gold standard and you can check your cone-beam CT using the ExacTrac. That's vice versa.
Bogdan: Okay. Thank you for that. And a question that maybe I'll ask both of you regarding the repositioning of the patient and the effects of rotation. So, maybe Dr. Niyazi, you can start with what type of shifts have you seen? What's the amplitude for the corrections that you've applied? And maybe address a little bit what kind of PTV margins you would put on these tumors especially if you're looking at single isocenter multiple PTV treatments?
[00:53:28]
[silence]
[00:56:24]
Philipp: Yeah. From the physical side we see the same, we think the same way. The way we measured our accuracy of the system with the old ExacTrac and with the new one. We see a cool wide tire accuracy is really quite sub-millimetric. You can actually say this might always sound like a marketing thing, but it is sort of submillimetric. And what you can do, you can actually monitor all the intrafractional motion of the patient inside the mask. So, you account for a lot of errors, which you would normally have. Again, you have some uncertainties as steeply said from the MR imaging. But we think with the current setup, at least for what I can say on Versa HD LINAC and with the ExacTrac and very high stability and reproducibility. And we are quite confident.
Bogdan: Dr. Niyazi, since you address the physical characteristics of Versa HD, you have a follow-up question on what is the smallest tumor volume that you treat? And maybe you can just address it in largest diameter, I guess. And link to that, honestly, what's the smallest field molecule that you measured for your treatment planning application?
[00:57:48]
[silence]
[00:59:12]
Philipp: We'll stick to the five-millimeter minimum.
Bogdan: Okay. We have questions, of course, about the new camera system. So, maybe Philipp, I'll start with you. What in your opinion do you consider to be the advantage of the thermal imaging over optical imaging?
Philipp: Well, one of the advantages was actually there's another question related to longitudinal deviation which was seen between the surface in the X-ray. This is the actual problem of surface imaging itself. If you look at it from one side and you see a tube, the tube can...the registration algorithm does not know how the tube moves in the longitudinal direction. And that's when the surface imaging from one side starts to fail or even from three sides starts to fail. If you add thermal imaging to it, the phantom or tube is maybe not the best example for it. But if you have it on a human being, the pelvic area on the patient is also sort of tube-shaped like and any surface imaging has some problems with it. If you have thermal imaging included, you sort of see the whole tube not as a tube, but more as like...which has something which has a lot of peaks and a lot of valleys inside. And that's something a registration algorithm can grab and hold on to. So, you have increased your accuracy for the surface registration. And that is the one main point of it.
The second point is if the thing that was the original Brainlab idea to have only one in-room camera and not two, or three, or four where you have gantry blocking and stuff like that again. So, you have one camera, and thanks to the thermal imaging, a very high position. And the results, which I showed with the 0.9-millimeter or something or 0.7-millimeter in the longitudinal direction, these came from...because that's when you compare the surface imaging only with the ExacTrac. And we have very... In the second data subsets with couch rotation, we compared surface and thermal imaging with X-rays. And that's basically it.
Bogdan: Yeah. Regarding the application of the SGRT solutions of ExacTrac Dynamic, maybe you can speak a little bit about what is the precision and accuracy for the SGRT components? And what kind of tolerances do you utilize in your workflows to trigger either a beam hold or an X-ray image? And maybe you can also address then the linking between the SGRT and the X-ray imaging or whether you can just monitor with SGRT.
Philipp: Yeah. So, we use, for every patient site, a different site we'll use a different subset of templates. So, if you use cranial SRS, we have a 0.5 millimeter and 0.5-degree tolerance level for the X-rays and 0.7 millimeter and 0.7 degrees tolerance level for the surface imaging. A little high on the surface because the ground truth, in this case, would definitely be the X-rays. And that's also what the system makes the ground truth is always the X-rays. So, you cannot yet do a surface-only workflow. You might do it in the future. The system always needs the X-rays to begin with. I think that sort of historically happened because the ExacTrac Dynamic is an X-ray-based system. And I think as far as I've talked to the developers from Brainlab, everyone was implementing the surface workflow. And after testing the system, everyone was quite happy that the surface guidance does indeed have a very high accuracy if you would do it surface only.
So, maybe in the future, there will be a surface-guided workflow that's, I'm sure this might be helpful for some patients. But if you have image guidance at your clinic implemented for all of your patients and you could reduce the number of cone beams or patients end up using MV beams and flat panels. You would not think of... And you can do that with the X-rays. I don't know if there's really a need for a surface guidance only, but it is possible. The accuracy of the system itself is a little hard to test as I already said of the surface system only, the surface and thermal system because a lot of the things are done internally. And so most of the tests are only to be done for the whole system. But as the... What we saw in some tests that the stability is quite good and also that the surface and thermal camera are in sort of the same accuracy levels as the X-rays for the patients in this case.
Bogdan: Thank you for that. And some practical questions related to what you just mentioned. What is the largest deviation that you've seen picked up by the camera system? And how do you utilize the camera setup, so the SGRT setup in conjunction with either close masks or open face masks?
Philipp: So, the largest deviations we see, it depends really on the patient. When it just comes to numbers, the largest deviations were from the service camera were from a stereotactic treatment, the one we saw in the presentation where we actually saw the breathing motion of the patient. So, those are deviations to up to... I don't know. I don't have the value in my head right now. But they were actually the maximum you can use for surveillance. We monitor open face mask and we also monitor the face mask, an enclosed face masks themselves. Just to give an... I have an extra safety level also for the closed face masks. The open face mask would actually be a monitoring of the movement inside the mask. And monitoring the closed face mask also bring some information because the heat signature can be seen through the face mask and that's another good thing for the internal registration algorithm.
In general, I would say we've seen a lot of high deviations when it comes to couch rotations where we correct for but we sometimes also have movement in the mask for stereotypic treatments of a multiple brain met patient. If the patient is on the couch for 40 minutes, 35, 40 minutes, something like that, there will be some motion. And this is something where we start... We have also stopped the treatment and we position the patient according to the X-rays because that's actually what the system does. If the surface goes out of your desired motion range, which you can determine at the beginning. You can choose if you want to stop treatment, if you want to trigger an X-ray, or if you want to do nothing. That's actually up to you. And it also depends on the specific patient site.
Bogdan: And Philipp, a question that we probably get every time we present this hybrid camera system between structural lights and thermal. And let me know if you're comfortable answering this. But the question is on the stability of the thermal signature if you need to reacquire it. So, maybe you can address a little bit how the two camera systems work together in terms of what's absolute versus what's relative.
Philipp: So, as far as I have understood it, because you don't really have inside what...you don't see what the thermal camera or the system does that internally. But as far as we've seen it, the two systems are calibrated together, thermal and the surface camera. And if you take one snapshot at the beginning of the treatment or at each couch rotation of the actual thermal signature, and this signature is quite stable at least the next 10 minutes or something. If you would have a large drift, the system would tell you that you have a large drift and that the monitoring is going in a weird direction. But also if you have a large drift, you take another X-ray system and that's the point where also the thermal and the surface actually are moved to the position of the X-ray. So, the stability of the thermal signature is not a big problem also throughout different fraction. The thermal signature is acquired before each fraction individually.
Bogdan: Right. And again, for everybody, every time you acquired the new set of X-ray images, the surface information is rewritten, so it becomes essentially a relative measure.
Philipp: That would have been the short answer to mine.
Bogdan: Have you had any experience with either skin tones, variations, or room lighting with the new camera system?
Philipp: We didn't have any room light deviations. We didn't see any effects. Although we have some fancy room lights inside, we didn't see that at all. But regarding skin tones, this is something to be tested because structured light or the measurement of structured light always relies on the reflection of the light and the body. And the skin tone does affect the system in general. You have certain settings for your skin tones. You can set it to bright or darker skin tones, which you can...but it doesn't have a big effect because you have the thermal information as well. So, in any surface guidance systems, you as the skin tone has quite high effect on the patient position or on what the camera can actually see because of the reflection settings
Bogdan: Dr. Niyazi, we seem to have quite a few questions regarding margins. And we have addressed this in other webinars, so I would invite all of you to perhaps review some of the other webinars that we had on brain metastases treatments and setups. But perhaps you can address this a little bit with what you do in your practice. And if you do nonlinear margins, margin definitions based on distance to the isocenter, and maybe address what those values are.
[01:10:46]
[silence]
[01:14:31]
Bogdan: Gentlemen, thank you very much for your great presentations. Obviously, they sparked interesting conversations too. Thank you, everybody, for joining the webinar. And if we missed any questions, please direct them on the Novalis Circle forum. We can answer them there for you. And stay safe and we'll see you at the next webinar.
Philipp: Thank you so much.
Bogdan: Thank you. Bye, bye.
Philipp: Bye.