Transcript
Bogdan: Hello, everyone. My name is Bogdan Valcu and I'm the director of Novalis Circle. I would like to welcome you to our first webinar. And thank you very much for joining. This is the first event in a long series to come, comprise of both clinical webinars where we'll invite Novalis Circle experts to lecture on topics of general interest for the community, and technical webinars that will be held by Brainlab, a good service continuing education initiatives. We have received continuing education accreditation for these events. We are happy to provide credits for all webinars via CAMPEP, MDCB and ASRT. Please check on the webinar's follow-up emails for information on how to submit for credits. But if you have any questions, please follow up with us directly via info@novaliscircle.com.
Today I had the privilege to welcome Prof. Agazaryan from UCLA as our first clinical speaker. But before I turn the microphone over, just a few organizational items. All our webinars are recorded and will be available as video files on the novaliscircle.org website for later review. About an hour upon completion of the webinars, you should receive an email from Novalis Circle with a direct link for the recording. Also, for today, questions may be submitted in writing via the chat interface of the webinar. And once the lecture concludes, we will address all your additional requests for information.
This topic highlights the UCLA experience with commissioning and patient-specific QA for the Brainlab multiple brain mets SRS element. Dr. Agazaryan and his team have performed extensive work reviewing multiple QA solutions available in the market. And I believe that their clinical experience will be of great value to us all. Dr. Agazaryan is a professor of radiation oncology and the chief of clinical medical physics of the UCLA Department of Radiation Oncology. He's also a professor of the physics and biology medicine interdepartmental graduate program within the UCLA School of Medicine. Dr. Agazaryan is also an expert panel member of our international Novalis Circle community. Dr. Agazaryan.
Dr. Agazaryan: Thank you. Hello and welcome. I truly appreciate you taking time to attend this webinar regarding the UCLA experience with commissioning and patient-specific QA for single isocenter treatment technique for multiple cranial targets. Our webinar's time yesterday coincided with the new iPhone announcements from Apple. So, maybe we were competing with Apple for bandwidth. We are excited to present our findings and learn from your questions and comments. We have an international audience today with many participants from Latin America, including Mexico and Colombia, Africa, many participants from Russia and numerous people from Europe and Asia. We are honored by your participation and excited to present our findings. There are also numerous people in the room. Everyone is smiling and ready. So, let's get started.
As the title of the presentation suggests, the presentation will include our commissioning and validation efforts as well as our results and recommendations for patient-specific QA methods. To start with, I include my disclosures and credits. Numerous radiation oncologist, neurosurgeons, and physicists have contributed to this project. I would like to specifically thank my colleague, Dr. Steve Tenn for his major contribution to the project. He will be present during the Q&A session. Our group also thanks all vendors who diligently worked with our group on implementing methods with their products. The presentation objectives have been well described in the invitation sent to you. I will, however, mention that towards the second part of the talk, we will discuss specific patient treatment plans and quality assurance examples as well as recommendations for planning margins and image guidance tolerances for treatment.
Let me start by mentioning that Brainlab's implementation of multiple brain metastases SRS treatment with a single isocenter uses enhanced dynamic conformal arcs. Per vendors that recommendation, we will refer to this product as MBM SRS. There are competing products on the market that have not been included in this work. Some of these use volumetric arc modulating therapy, VMAT. So, I should mention that depending on your state regulations and institutional policies, the delivery technique may have an impact on your patient-specific way requirements. At our institution, we have decided to intensively test and QA these patient-specific plans, although they're not VMAT plans until we have collected enough experience to revisit this decision later on.
So, on this slide, I summarized our validation and commissioning efforts. We have worked with multiple vendors to evaluate and compare their products, results, and workflows for commissioning purposes. We will make some recommendation in this regards. Many of the methods were utilized for initial validation of the software prior to even commissioning. So, realistically, typical users could choose a few of the methods presented here for their commissioning purposes. By the way, many of these methods could also be used for patient-specific QA. Just in general, we have used the race as you can see on the slide, MapCheck and ArcCheck. We have used different softwares for secondary calculation, some nucleus PerFRACTION, Eclipse, ModusQA, Mobius3D, MU Check. We have used jails RTsafe and ModusQA.
So, let me start with the very first case that we have evaluated prior to our clinical implementation. This patient had 10 brain metastases. The prescription for these 10 targets ranged between 14 gray and 18 gray based on the target size and the location. Multiple brain metastases elements uses the pencil beam dose calculation algorithm. So, the calculation method uses 1.25-millimeter adaptive grid size. The dynamic and has to follow all calculations use five degrees arc steps. And this will come into play in a moment and during the presentation. We will discuss later in the talk the consequences of this specific parameter on the QA methods.
So, to start with, we have compared the MBM calculation, those distributions, with the AAA forward calculations, the Eclipse AAA calculations. I should mention maybe that the variant accuracy algorithm was also used for comparison. However, we faced some interesting challenges for off-center targets with this algorithm that are still being investigated. And what I mean is with the accuracy algorithm. So, our results shown here are with AAA. So, the plans were transferred from the original planning system to Eclipse and the original monitor units from the MBM were utilized to forward calculate the dose distribution in Eclipse. We have investigated the effects of different calculation grid sizes for AAA forward calculation as shown in this slide. We have used many different sizes.
To keep this discussion brief, we present a comparison of MBM to AAA calculations with only 2-millimeter grid size. Please note that on the X-axis is the target volume, hence, the data is sorted in terms of target size from left to right. The Y-axis is the percent difference between MBM and AAA calculations. In this example, the maximum difference in mean dose is less than 5% between these two calculations. And as you can see, it gets larger when the target size gets smaller. This data gives us confidence that we're dealing with clinically small differences that are likely due to AAA models in our system for small targets. With that said, some institutions create separate AAA model for small fields. We believe with this type of approach even better agreements can be achieved and what we show here. We can Eclipse one dose profiles, 1D dose profiles can be visualized and evaluated.
As it can be seen in this example, there is an excellent agreement between two profiles, the red and blue. However, we further quantify disagreement. What we did is we took the MBM and AAA distributions and export it to PTW very soft software. This is a software that allows us to do more quantitative analysis of comparison. 2D gamma index analysis are performed with acceptable clinical results as shown here. Also, 2D dose distributions and 1D dose profiles are also clinically acceptable. And we're just showing you a few examples in this slide. Next, we measure the dose distributions utilizing EBT3 films in QuiCk Phantom. Measurements were compared to distributions obtained from MBM as well as AAA. I should mention that later in this presentation, we will also discuss measurements with a new EBT extended dose films, EBT-XD. And the results are, by the way, even more impressive with that one.
The agreements between measurements and calculations were clinically acceptable shown in this slide. We have observed discrepancies when ASCII export feature is utilized within MBM software and imported into FilmQA Pro. We found a workaround for this to export the 3D dose to third-party application and then they can export planar those. This issue with the ASCII export and import are still being investigated and we don't have much more details about that. But the point is, you can use third-party software to do this comparison. 2D and 1D analysis of the measured dose distributions compared to calculated dose distributions showed clinically-acceptable results with gamma index calculations in high 90s using 2 millimeters and 2% criteria.
So, now let's transition to using other types of dosimeters. Here we made an attempt to measure 3D dose distributions instead of 2D. Using single use radiochromic polymer gel from ModusQA. We use the ClearView gel shown in this image in a second with an optical CT scanner. So, I just wanna be clear, the imaging is with an optical CT scanner. ClearView comes with the jars shown in this image, different sizes or shapes. And what I'm showing you in this slide are two irradiated gels, one for multiple brain metastases and the other one is validated spinal radiosurgery case. We did that for a reference. Optical changes can be seen in these images. I should mention the best results are obtained when the gel resists around 40 gray dose, hence, often the plans may need to be re-delivered or scaled up. We choose to re-deliver plans to preserve the dynamics, and I'm not saying gantry movements. As you know, if you scale up the plan instead re-delivering, then you are changing the MLC and gantry movements. So, in other words, we wanted to QA and measure the plan that will actually be delivered to the patient. And scaled-up plan is not the same.
This is the analysis of that multiple met MBM case in a gel. Generally good agreement is observed in terms of profile shapes shown on the bottom, the red line and the blue line. There is an agreement between those lines. However, we have difficulties associated with alignment and scanning artifacts because of impurities in the gel. And you see some of these artifacts in the images on the second row on the bottom. Large absolute dose discrepancies were also observed. With more experience, care, and product improvements, better results may be possible. So, that was one type of a gel, which was scanned with an optical CT scanner.
Now, I wanna transition to, again, polymer gel that is not scanned with optical CT scanner. It's scanned with MR. In our commissioning work, we have utilized RTsafe polymer gel dosimeter. We obtain impressive results with this gel dosimetry. Patient-specific 3D phantoms can be printed by RTsafe based on patient's CT scans then you transfer to that. However, this step does not need to be repeated for future patients as the same phantom can be refilled with a gel for future use. So, we took this process a bit slow. First we said, let's validate the dosimetry method itself prior to using it to validate the multiple brain metastasis software. So, we ordered a simple cube phantom shown in this image that comes with calibration vials shown on the left. This particular gel can accurately measure doses in the range of 3 to 35 gray. There is another gel available from the same vendor that can accurately measure those in the range of 0 to 12 gray. We will come back to this. We have utilized both gel types. There's particular one I'm showing you here, uses range 3 to 35 gray. So, you are able to measure higher doses, but you are using accuracy below 3 gray.
Calibration gel tubes were irradiated to doses ranging from 0 to 25 gray with 5 gray increments. And I really like this slide. Here, the calibration vials were sort of calibrated with a 15 MED electron beam at around 4 cm depth because this is where the distribution is relatively uniform using a setup shown in the picture. So, it was somewhat the improvisation of how to uniformly radiate these gels. There may be other solutions that you can come up with. Although the dosimetry is achieved with MR scanning of these vials, I wanna bring to your attention that irradiation of gels creates this polymerization that can be seen in the pictures going from 0 to 20 gray, even visually, you can tell that they are polymerization changes happening. By the way, the irradiation of these gels establishes the sensitometric curve for dosimetry. So, we delivered a very simple 3D conformal plan to this phantom. And I wanna bring to attention again that even in this picture on the cube phantom, you can see the polymerization changes in the phantom. That cloud area where the dose distribution went. First, we tap it to scan the cube gel with our 0.345 Tesla ViewRay, MR-guided radiotherapy system. However, the signal-to-noise ratio was prohibitive to obtaining a clinically meaningful analysis.
Here's the phantom and the calibration vials during that attempt of scanning with the ViewRay phantom...scanner. Then, we MR scan the phantom using 1.5 Tesla MRI scanner. This give a good clinical results. Our MR [inaudible 00:19:11] had to work with the radiology department to install specific QT sequences provided by the vendor. So, I want the users to be aware of the fact that you have to use this specific sequences and work with either your department or radiology department, whoever owns the MRI scanner. I do wanna mention that the second type of the gel which is coming later in the presentation will not necessarily require using calibration vials.
Subsequently, once we've establish the ground truth that we are pretty good agreement between the calculation and measurements with the gel, the gel validity dose established, we went ahead and ordered 3D-printed patient-specific phantom for one of our patients. In fact, we took this phantom to the physician and said, "Oh, by the way, do you recognize the patient? This is the 3D-printed patient." We did this by transferring the CT scan to the vendor. As it can be seen in this image, the phantom has two compartments. The central compartment is within the skull, closed up with a larger cap, and contains the gel dosimetry inside. The outer compartment, closed up with a smaller cap, is filled with water. The gel is temperature-sensitive and it's shipped with ice packs and thermometer.
So, what we see here is the phantom during the end-to-end testing. So, first, the phantom was masked and CT scanned. That's one of our therapists training the mask for the phantom, pretending it's the patient. Then, based on the treatment plan, the phantom was aligned at the treatment position in the treatment room using ExacTrac cranial array. The array is shown on the right side. And then we employ the ExacTrac IGRT process. In this picture, some example images from ExacTrac IGRT process. And I just wanna bring your attention that you do get contrast differences with the phantom. Enough information to position the phantom just like you would do with your patient. So, here we go. What you see here is the video captures. We show the delivery process on that phantom that I just showed from two camera viewpoints in the room.
If I zoom into the MLC movements, and what you're seeing here now is the MLC movements, there are many things that you can observe. The delivery method is a conformal arc. There is really not that much modulation happening here, right? You see these open fields moving around between the jaws. It's not a volumetric modulated arc therapy. Additionally, 6 out of 10 targets are covered. You only see six openings in the capture. The system selects the best combination of targets for each of the arcs. This is because two targets aligned among the same sort of MLC leaf pairs cannot be effectively treated with the same arc. Yeah. If necessary, the arc will be repeated in the reverse orientation, the same arc will be repeated, to cover another subsets of targets. So, if in the planning system that you see the same arc is going clockwise and then counterclockwise, those may not be covering the same sets of targets. Strategically, they're programmed to cover different targets and provide good coverage for all of them.
Now, that's the delivery and then the gel is analyzed. And what I'm showing you here are the results. I should mention that the company, in this case, RTsafe, provides a comprehensive PDF report that includes many aspects of those dosimetric analysis. In this image, what you see here, by the way, pay attention you see the calibration vials around the CT of the phantom. Around the skull, you see those different intensity circles. Those are the calibration vials being scanned by a scanner. You also see long D profiles through some of the targets as examples. Some pretty impressive results. Each upper one is the profile and the lower one is the gamma index calculation of that 1D profile. And other sets of examples, similarly, excellent results are displayed here. The reason I included this slide is because I'm showing profiles through one target, profiles through two targets, and then on the right bottom, I'm showing profiles through three targets. The red color is the TPS, the planning system, and the blue color is the measurement with depressive results. I should mention, again, 2D distributions can also be visualized and then 2D gamma index calculations are also provided by the vendor in this report.
We worked with a vendor and we made a decision that we actually want to measure distributions with an ion chamber. We wanna validate absolute dose measurements with an ion chamber. So, we asked them to 3D-print three ion chamber phantom for most reliable dosimetry measurement with ion chambers. So, we designed this phantom and they printed in 3D. In this case, there is only one for the phantom since there is no gel dosimetry involved. The user fills the phantom with water and the phantom is reusable, making it a good candidate for periodic system QA. We are thinking of using this phantom on a monthly and yearly basis. This phantom is manufactured to hold three PTW 3D pinpoint ion chambers. The volume of design chambers is 0.016 cubic centimeters. So, it's pretty small volume, sensitive volume, allowing us to measure 3D small dose distributions. What I'm showing here is the CT scan of the phantom. Again, you can observe that there is no gel, it's uniformly sort of water all around the skull and inside the skull. And you can also observe the three locations of the ion chambers. We went ahead and created a plan that delivers 14 gray, 16 gray, and 18 gray to each of those ion chambers. And I'm showing that in that green box on the left. Those are the prescriptions. Here, what I'm showing is the distribution, the dose distributions around those ion chambers.
I have an important sort of point to make with this slide here. While the prescription doses are 14 gray, 16 gray, and 18 gray, it's important to note that the mean and max doses are delivered to the cavity are much larger. And I'm highlighting those in the green boxes here. In fact, the mean doses are 25% higher than the prescribed doses. And the maximum doses are 37% higher than the prescribed doses. There's really nothing wrong with this per se. I just wanna bring to your attention that this speaks of the fact that the distributions generated with MBM are more heterogeneous than distributions we have typically simply by plan. You could simulate this by-plan as well in homogeneous distributions. But typically, you see more homogeneous distributions. You can argue either way. I mean, this could be better or, in some cases, worse for clinical cases.
So, what I'm showing you here on this graph is the ion chamber statistics from the planning system in terms of minimum, mean, the maximum doses versus measured dose. The measure dose is shown in red. Maximum is in blue, minimum is in green, and the mean is in yellow. Yeah. So, the most relevant statistics is the mean dose since the ion chamber measures approximately the mean dose. What I did is I took other statistics out here and it showed mean dose verses, the measured dose on this slide just to sort of reduce the clutter on the graph. Again, I'm gonna bring to your attention one more time that measurements are in the range of 19 to 25 gray while the prescribed doses are in the range of 14 to 18 gray. Speaking of the denominator of the distribution. And then I converted this to percent differences between calculation and measurement, mean dose from the planning system versus measurement just to show you that there's pretty good agreement between both calculation and measurement. The maximum deviation from the measurement is less than 2%, 1.7 for the chamber two.
We did not stop there with the three ion chamber plan. We delivered the plan on to a gel phantom for two purposes. So, let's see. This particular gel is a new version of a gel from RTsafe. I mentioned this previously. With a 0 to 12 gray range, and we wanted to validate the gel dosimetry itself because we just measured that plan, we can deliver it on a gel and validated dosimeter method. The vendor offers a relative dosimeter method that does not require irradiation of vials. As you saw in a previous example with the other gel, irradiating those vials was not the most straightforward thing to do. We were interested to see the results of this method and compare it to the other one. Yeah. And here are some results of that ion chamber plan delivered on that gel that I just described. On the left is the MR scan converted to dose, and then you see profiles 1D, 2D, 2D gamma analysis, 1D gamma analysis. All of this is provided in the report to you.
Here's another target. This is a more superior target. The 1D profile agreement is so good that it is difficult to distinguish the two profiles. If you look at those 1D profiles on the top, the TPS in red and RTsafe in blue, really difficult to distinguish those two. Again, we are also showing 1D, 2D gamma analysis as well as 2D distributions on the right upper corner. I wanted also to show you two profiles going through two targets in a coronal plane in this case just to give you different perspectives. And also I'm showing you a plane, coronal plane, that goes through two targets with the 2D distributions on the right upper corner so that we're not selecting one or two just to show it. So, overall, I've seen pretty good results with this dosimeter. And this is a very important slide that I wanna concentrate on. So, let me think about this for a second. This particular gel is a new version of a gel that goes from 0 to 12 gray. We've talked about this.
One of the reasons why the company created this gel with lower range is to be able to accurately measure lower doses for providing DVH analysis for objects. Remember, these are mostly for radiosurgical cases. So, it would have been great to use the other gel because your doses typically are higher than 12 gray. But the advantage of this gel is that it goes from 0 to 12 gray. You get very accurate results for very low doses. When you get DVH analysis like I'm showing you on the left between TPS and RTsafe, it is very important to bring to your attention that these 3D measurements and ExacTrac positioning, they are simulating our patient positioning process. So, the DVH results speak of our...or more scientifically our confirmation of rotational and translational alignment accuracy in 3D versus when you use a race, there is less of an evidence of that. So, 0 to 12 gray gels provide measurements 0-100% range, so you get pretty nice deviation results. The other gel which is 3 to 35 gray also like that gel, provide accurate measurements from 10% to 100%. And the beauty of that one is you can deliver 20, 25, 30 gray doses without scaling down. So, to measure higher than 12 gray with the second gel that goes to lower doses. Your user is limited by the accuracy of the measurements below three.
All right. Now, I'm transitioning to now from physical measurements. We've talked about film, 2D measurements. We've talked about 3D measurements. Now we're going into calculations, Mobius3D, in this example. Mobius Medical Systems has a product named Mobius3D for secondary 3D calculation. This product helps with many QA activities, one of them is 3D dose calculation with an independently verified model and collapse cone algorithm. The software provides an excessively detailed PDF report, comparing the original plan to your calculated dose distribution. Initially, we observed results with good profile agreements, except at the peak of the small targets. These results, they're based on machine data provided by Mobius3D. So, it was not our machine data. It was totally independent calculation. The user can provide small field data with Eclipse commissioning conditions to the vendor and hopefully, that will sort of improve the results. But we have decided to do in our case, the vendor adjusted the factor in their calculation algorithm to match our results. I think this factor is similar to Dynamic Lift Gap or DLG factor in Eclipse. The adjustment greatly improved our future results as you can see in this slide.
It is reasonable to question these types of factor adjustments, but we suggest that, in this case, it is acceptable practice since the algorithm has been well validated with measurements and other methods. In our clinic, we use MUCheck for secondary MU calculations. And we investigated the use of MUCheck for MBM. Our original version of the software did not support this type of calculations, but the vendor provided that capability in their next and current version of the software. The actual... I showed the results of MUCheck for three-ion chamber plan. Some beams and points readily show agreements between 5% for first calculations and other beams like shown here require small volume averaging distance, typically 1 millimeter to get acceptable results.
We have also evaluated some nuclear products or secondary calculation in QAs. This web-based SunCHECK QA platform offers impressive combination of tools for numerous QA methods. What we utilize is the DoseCHECK and PerFRACTION. Some of the options we have explored are included in this slide. We have performed secondary 3D dose calculation using DoseCHECK. We have done 3D dose reconstruction using DynaLog files after the delivery of the final phantom. So, while we are measuring with the film and ion chamber and other things, the machine was generating MLC lock files that are later utilized for 3D calculation. We did also investigate the use of 3D QA with EPID measurements. However, the specific EPID in use in our clinic has been having hardware issues and that investigation is still being addressed. Using 0-millimeter search radius, the agreement in this particular case for all points is less than 6%, shown on the left grid box. The gamma passing rate with 2 millimeters and 2% is 100%, shown on the right. With only 1-millimeter search radius, the limit in this particular case or all points is better than 0.1%. It's listed as 0.0 on the left here. The gamma passing rate with 1 millimeter and 1% is 98%. In fact, our clinical examples that I will show later show even better results.
One item of notes. The reference points exported from Brainlab software are at the prescription line. So, these points are at the prescription line except the isocenter, which means these are typically at the peripheral of the target, making the dose calculation matching more difficult. So, more to come about this. Now, I wanna transition to patient-specific QA methods. We've talked about validation of the software, the commissioning of the software, and the delivery method. I shouldn't say software for the whole package, right? Many of the tools discussed in a previous section are directly applicable as a patient-specific QA tool. I have a quick summary here in this slide. Yeah? Many of the methods were investigated for academic and investigational purposes. I just wanna reiterate again that typical users need to choose just one or two of these for their clinical practice. Our very first case was a patient with two arteriovenous malformations, AVMs, as you can see in this picture on the left. The MR is on the bottom row. It is important to mention that Brainlab's intended clinical use for the product is for brain metastases. With that said, we have utilized this novel delivery method for treating multiple AVMs.
The superior AVM located on the left frontal region is twice as large as the inferior arteriovenous malformation located on the left temporal region, 1.8 cc and 0.8 cc correspondingly. The prescribed dose to both targets were 20 gray. This is with an understanding that the dose distribution within the target is harder than our experience with iPlan. Also, the 20 gray position covers the module plus 1-millimeter margin in all directions. The secondary calculation with Eclipse planning system showed excellent agreements with original plan. Again, the monitor units from the original plan were utilized. Reference point doses were less than 4% different between two planning systems. So, to do reference point analysis, we have created this Excel sheet. The matrix gets really complicated as the number of targets and beams increase. For example, if there are 10 targets and 5 beams, then 50 numbers need to be analyzed realistically.
We went ahead and made an EBT-XD, extended dose film, and ion chamber measurements with the Ashland QuiCk Phantom. We have calibrated the EBT-extended dose film to 32 grays in our clinic. I think it can measure up to 40. It's calibrated to 32. Correct? Yep. The phantom was scanned with infrared markers, highlighting the infrared markers with red circles. This allows us to automatically position the isocenter at locations where the ion chamber measures the dose intended for each of the targets. Based on the coordinates of each of the targets, we calculate the location of an isocenter that will place the ion chamber at each of the targets. So, we create as many plans as the number of targets for ExacTrac. Here's one example where the isocenter was placed in such a location that the distribution intended for the first target covers or falls over the ion chamber. Each of these plans are separately transferred to ExacTrac, and then ExacTrac infrared guidance allows for quick positioning of the QuiCk Phantom at each of those locations. On the left picture, I'll show you the phantom with infrared markers. This particular example is the delivery of data to an implant I showed you. The ion chamber measurements showed an agreement with calculation within 3%. Yep.
So, we went ahead and also analyzed the field measurements in a similar manner. And the 2% and 2-millimeter gamma index analysis yielded 97.5% results. This is the plane center to the first frontal lobe target. And this is the plane center to the second temporal lobe target. Here we tighten up with 1% and 1-millimeter gamma index analysis with 97.4% gamma index. I mean, pretty impressive results as you can see there. Next, we employ the PerFRACTION software. Actually, it's the dose check calculation. The 3D gamma analysis with 1 millimeter and 1% or 98.67%. [inaudible 00:43:35] dose with 0 millimeter search radius, so, 4 and 9.5% deviation. Just a reminder, these are points at the periphery of the target shown in the red box. When we added second sets of points shown in the green box that are centered in the target, even with 0.0-millimeter search radius, these results are better than 3%. And note that they actually said that dose discrepancy is about 19%, which is at a low dose region, but I'll come back to this in the next slide. When we have 1-millimeter search radius, it significantly improves the results. And as you can see, the numbers went significantly down including the isocenter dose that went from 19% to 0%, which is somewhat surprising.
Then this is very different. This is DynaLog file. When we delivered on a film, DynaLog file was created of the MLC movements. And that was used for reconstructing dose distribution. And with 1-millimeter search radius, the point dose analysis are better than 0.1% and 3D gamma passing rate was better than 98% with 1 millimeter of opposite criteria. The MUCheck for this case was also impressive, with 1% with 1-millimeter volume averaging and 3% for the other target. This one is interesting. So, one of the fascinating results we obtained was the Sun Nuclear ArcCHECK. So, we initially didn't predict what will happen with this. Sun Nuclear ArcCHECK dosimeters are located at the distance from the center of the phantom. So, the aramid, those calculation exhibits a streaking artifact because of the five-degree calculation steps. You can see that streaking artifact there now. Even then, there is a surprisingly good agreement at the measurement points shown on that profile that I'm showing on the bottom with the gamma analysis.
One could experiment with placing the isocenter positions where the high dose region falls on the diodes instead of at the center. I suspect there may be problems with that as well relating to angular dependence of diode up because many of the beams will now graze the diodes of large angles. So, while this device can be used for QA purposes, it has the limitations discussed earlier for this specific use. Our second clinical example is a patient with two metastasis, left thalamic target to step 16 gray to cover 99% of the target, right thalamic target, prescribed 18 gray to cover 99% of the target. The Eclipse calculations were between 30%. MUCheck calculations were 2% for all beams except with one that was 5.5% deviation. As you can see the DoseCHECK results for this case were better than 1% for both targets. And the isocenter, I should mention. This calculation is with 1-millimeter search radius. The 2% and 2-millimeter criteria gamma index is 99.5%.
We went ahead and measured the right frontal target prescribed to 18 gray and the ion chamber measurements showed 2.5% deviation and the left alembic target prescribed 16 gray and the measurement was 1.25% difference. So, well within acceptable criterias. The extended dose EBT measurements were made for both targets. And both targets show impressive 98.4% and 96% gamma passing rate with two and two criteria. I'm going to the third clinical example. This is the last clinical example. Our third clinical example was five metastasis, four targets on the left and one target on the right. Two-millimeter margin was used for all targets for PTV creation. All targets were prescribed to 18 gray to cover the 99% of the volume. The left superior frontal target was larger. It was 2.6 cubic centimeters. The rest are small targets, less than 0.5 cubic centimeters. Eclipse's calculations were between 7%. MUChecks were within 6% for all beams. The DoseCHECK results or in this case they're better than 5% for all targets on the isocenter. The search radius is 1 millimeter. The gamma passing rate 99.9% with 2 millimeter and 2% criteria.
So, we selected two targets to measure with an ion chamber out of those five because the others are small. Left superior frontal target, 18 great prescription showed agreement within 3%. Left anterior frontal target prescribed to 18 gray showed an agreement between 3%. So, both were less than 3%. Yes, there may be challenges with much smaller target measurements. So, what we did for that, we went ahead and measured all five targets with extended dose film. We're showing you absolute dose comparisons with excellent 1D profile results here. So, we have confirmed with my colleagues who have specifically made these measurements that there were no adjustments made. These are absolute. All five targets pass 2D gamma with 2 millimeters and 2% criteria better than 98%. So, here the point I'm making is even the small targets were measured perhaps.
It was reported to us the Sun Nuclear ArcCHECK can also be utilized for patient-specific QA. Our colleagues, Dr. Javad Rahimian, and others from Kaiser Permanente in Los Angeles provided the following data from one of their patients. They collapsed all the table angles to zero for these measurements. The beauty of this method is that a single isocenter location or like the ones that we made with other phantom can measure most if not all targets. Their results are shown here. Again, this is from Kaiser. Gamma passing rate is 3% with a 3% and 3 millimeters is 99.5%. I'm guessing their criteria is three and three. In our case in our clinic for radiosurgical cases, two and two and one and one is evaluated. They do show excellent profile agreements with this device.
So, in closing, I wanna transition to an extremely important point that needs to be considered when treating multiple brain metastases with single isocenter. Traditionally, we have treated each of these target with an individual isocenter at each of the targets. If I simulate a 3D degree rotation for each of the targets, this is what will happen, right? You'll see the movements of this sort of radiation fields with a 3D degree rotation. Really not that big of an impact. With a single isocenter treatment, the effects of these rotations become much more pronounced. Now, I'm showing you one isocenter in the center. And if you simulate the motion with some targets, almost completely can be missed with 3D degree rotation. Look up a green one on the left and the yellow one on the right.
In this graph, I display the translational misalignment that can occur with rotations as a function of distance from isocenter. So, the X-axis is the distance from isocenter. The orange curve is the 0.5-degree rotation. And a target that 6 cm away from isocenter 0.5-degree rotation can result to 0.5-millimeter translation. That's the point I'm trying to make here. So, it set ExacTrac thresholds at 0.5 millimeter, 0.5 degrees. The approximate allowable of misalignment will be around 1 millimeter. The overall accuracy may be in the order of one to 2 millimeters. We're not buying, but we suggest that 2-millimeter margins for PTV creations should be used for targets that are far away from the isocenter. So, just to summarize, we recommend 0.5 millimeters and 0.5 degrees tolerances or thresholds as it's worded in ExacTrac. We also recommend 1 millimeter and 2-millimeter margins based on the distances from the isocenter. Smaller margins can be considered for benign conditions. Yeah. So, also, we could consider two isocenter treatments for multiple targets that are distributed in clusters at large distances. So, you could divide them to two groups and treat with two isocenters, then you can utilize smaller margins.
So, let me get back to credits. I want to thank, again, all my colleagues for their contribution to the project. This work was not done alone. And thank all the vendors who worked with us on implementing methods with their products. It was my great pleasure to present the UCLA experience with commissioning and patient-specific QA for single isocenter treatment technique for multiple cleanup targets. We do appreciate that you took time out of your busy schedules today and attended this webinar. We will now open the floor for questions and comments. Correct?
So, the first question, in the iPlan, for multiple mets, where do we normalize at isocenter or target mean? Do we do it for target mean or in Eclipse? Steve?
Dr. Tenn: So, this is a question about iPlan for multiple mets. We normalize the target. The way we normalize targets prescription in the iPlan system is a volumetric normalization. So, we typically try to achieve at least 99.8% coverage on the PTV for medicine. So, that would be essentially on the periphery of that target.
Dr. Agazaryan: Thank you for that. Second question. Also, where should we keep the isocenter in brain mets with more than two metastases? I think I can take this one. So, one feature of the software is that the isocenter is automatically placed at the central volume of all mets, all selected mets, I should say, for that treatment. The third question. Two-millimeter is a lot for less than 1 cm measurement, isn't it?
Dr. Tenn: This is a question about the PTV margin that we're using for a lesions smaller than 1 cm. And I think as it was showed in the slides regarding uncertainty for the rotation and translation and that we use currently 0.5-millimeter and 0.5-degree rotation tolerance in ExacTrac. Based on the slide that you were showing, I think 2 millimeters is appropriate.
Dr. Agazaryan: Yeah. But there's a good point here, right? It's a large margin. It's a clinical decision to [inaudible 00:55:33]
Dr. Tenn: Yeah. Having said that, if you have a cluster of lesions that are very close together, you can consider using smaller margins because your uncertainty will also shrink.
Dr. Agazaryan: Okay. It seems like 2%, 2-millimeter gamma is a poor metric for this geometry, sensitive treatment technique. What other metrics were considered? We utilized 1%, 1-millimeter criteria as well. What other metrics were considered?
Dr. Tenn: Clinically, this is...the gamma index is what we use to review these plans. So, that's why we went with this particular metric to look at the accuracy, the spatial accuracy for these. There may be other metrics that you're interested in and we can definitely consider those.
Dr. Agazaryan: Yep. How high is the phantom setup uncertainties? Maybe we should mention that the thresholds... It's not necessarily the uncertainty, but the thresholds that we're utilizing is 0.5 and 0.5 millimeters and degrees. The uncertainty, as I mentioned in one of my slides, with just these parameters is up to 1 millimeter. And overall, our uncertainty based on the data that we have that was not shown here probably between 1 and 2 millimeters
Dr. Tenn: Yes, I agree.
Dr. Agazaryan: How was the gel dosimetry specially registered to the plan?
Dr. Tenn: The gel dosimetry... So, this is a question about the RTsafe phantom I believe. What happens is once we acquire the MR image of the phantom, following a radiation, we send that MR image off to the RTsafe company and they do in-house registration of that Phantom to the plant. So, yeah.
Dr. Agazaryan: So, let's look at this one here. What PTV margin do you recommend to accommodate the apparent 2-millimeter position of discrepancies? We should mention that up to 2 millimeters not...
Dr. Tenn: Yes.
Dr. Agazaryan: So, it's up to 2 millimeters. If you have a cluster that is really not more than five CMFA or so, we really don't have to utilize 2 millimeters. But if they're spread around more than that, you either can utilize 2 millimeters or you could split them up two clusters and treat them with a couple of isocenters. It's still significant improvement. One of the examples...the five multiple met target when we were treating that, typically we go to the machine, we meet the physician five times to verify each of the isocenters. We only have to go there once. So, if you have 12 targets, for example, and you have to be there for 12 individual treatments, in this case, you will be there only twice. So, it's still a significant improvement even if you use two isocenters. Yeah.
So, let me see. Did you try to let Eclipse AAA algorithm calculate the new values? I think so. Based on the same prescription as Brainlab utilizes. I think that refers to the discrepancy, right? So, if you do the reverse instead of forcing the MUs, you use the dose to calculate the MUs, but discrepancy will be within 5%. So, the answer to your question that's coming from one of the attendees is it will be within 5% at least in the examples shown here. Looking at the controlling tools of Brainlab and Eclipse, what according to your [inaudible 00:59:43] better Eclipse or iPlan? That I will defer to the expert here. Go ahead.
Dr. Tenn: We tend to use the Brainlab. I'm not gonna say one is better than the other, but it really depends a lot of what you're used to. We typically use Brainlab contouring tools here. We're very familiar with them. The neurosurgeons, the radiation oncologist are all very familiar with them and are very confident in them. So, we almost always, I would say, better than 99% of the time will do the contouring and Brainlab rather than Eclipse.
Dr. Agazaryan: I think I can address this one. What is the minimum field size in your beam commissioning measurement for Eclipse? So, this is an interesting question. Ours is 10-millimeter by 10-millimeter. But I should mention that Eclipse does not directly use measurements. It creates a model based on your measurements. So, when you create a model, you adjust the model to have a good agreement for small fields as well as for large fields. And you have to make a compromise in that sense. If you get a good agreement with lush fields, maybe there will be small discrepancies with the small ones. That's why some clinics create two different models. So, the point I'm making is it's not only the measurement, but also how you adjusted your AAA to match the small fields. And if you do match small fields, you may be off for the large fields. So, that's why some clinics have two different AAA models in their software. Yeah?
Dr. Tenn: Yeah.
Dr. Agazaryan: So, should we conclude here or do you have questions? You're okay?
Dr. Tenn: Yeah.
Dr. Agazaryan: So, how multi-mets compare with individual isocenter for low doses? That's an important question. I think outside of the target and the low dose [crosstalk 01:01:46]
Dr. Tenn: Yeah. It's probably similar, possibly better. It really depends. For the single isocenter targets, we've been using no margin and here we're using 2-millimeter margin. So, there's some differences between the two. But yeah, I would say they're similar in the low dose region.
Dr. Agazaryan: So, I'll take the other one. Can you tell, sir, about pencil beam and Monte Carlo calculations in the software? So, let me think. The Monte Carlo is available in the next element that will come out. It's called cranial SRS element. This software here which is the multiple brain metastases element uses pencil beam algorithm. That's the only option you have for now until Monte Carlo will be implemented in the next release. So, Monte Carlo is coming out. Currently, the pencil beam is the only option for that.
We're waiting for questions here. So, between arc treatments, which couch keep arcs? Do you verify the setup? Again, that's an excellent question. Yes, we do. So, we verify... So, let step back. First of all, when ExacTrac is done, we do Cone Beam CT to independently verify the positioning of the patient, and then we repeat the ExacTrac prior to delivering the first arc, and then we will repeat the ExacTrac prior to each arc. Yes. The answer to your question is yes.
Question number 22. Did you use Eclipse for a second check? And if so, what is the small field size in your Eclipse B model? We examined the MU discrepancies found out to be different in very small targets from 2% to 25%. So, I think we already addressed this question to some degree. Depending how your model in Eclipse is adjusted, some clinics use DLG to adjust the model. Your discrepancies can be smaller or larger. It's a model-based planning system. So, when you're adjusting, you're compromising between small fields and large fields. If you adjust for smalls, you will have good agreements for the small ones. So, I think we already answered that. Which custom checks did you do mostly in the software and how did you do it? And what is your protocol? This is for the software, I think, they're asking. Correct?
Dr. Tenn: So, we don't do monthly checks for the software. There's an annual QA that we do for the software and it's pretty much to run a single plan that we've used from the beginning over to check the constancy. It's a typical annual QA for planning system.
Dr. Agazaryan: If you would allow me, Steve. I just wanna make sure we are not misunderstanding. If what you mean is the end-to-end testing or anything like that, with ExacTrac we do that on a monthly basis.
Dr. Tenn: Yes. If the question is about ExacTrac, for sure, it's a monthly test.
Dr. Agazaryan: I just wanna make sure there's no misunderstanding here.
Dr. Tenn: Yes, yes.
Dr. Agazaryan: Is film your preferred patient-specific way for cases moving forward? In our clinic, that's correct, but in clinics where you have the MapCheck like the Kaiser Permanente group, you could utilize that. So, it's actually clinic-dependent how you would do this. And if you do have access to gel dosimetry, maybe that would be...
Dr. Tenn: Yeah. We've gotten really good results with our Gafchromic film QA process. It's really good and high resolution for the small targets. So, we'll be using it going forward. Not to say, again, that you couldn't use another method, but for our clinic, that's what we'll be using for these small targets.
Dr. Agazaryan: So, let me mention that these are dynamic conformal arcs. And depending on your institutional policies, state, country, you may have different requirements for QA, right? Some institutions, regardless of what type of treatment it is, if it's a major surgery, you have to measure. Some institutions if it's not an IMRT or VMAT, they don't necessarily measure. Yeah? We took the approach of measuring all plans up until we have enough data to revisit that decision. Okay. Any more? Oh, how far have you gone with EPID-based QA investigation? We have done this in the past. We had pretty good results. But with this specific project, when we tried EPID-based QA, unfortunately, the machine...we have a dedicated machine for this types of treatments. And on that dedicated machine, we have been having EPID issues. So, I'm sorry to say that we haven't gone too far with this. But when we have results, we'll be happy to share with you.
After ExacTrac, do you use Cone Beam to do make-shift based on CBCT? Excellent question. So, no. The CBCT is not utilized to do any adjustments. The CBCT is used for validation. We look for up to 1 millimeter discrepancy between ExacTrac and Cone Beam CT. And if it's any more than that, then we investigate why we have that discrepancy. The positioning is done with ExacTrac. And intra-fraction motion management is done with ExacTrac while the patient is on a table getting the treatment. The beauty of ExacTrac is that the table kicks, table rotations, you can get the images and you can verify the patient position. Not only it monitors the patient motion, but it also corrects for table uncertainties. Correct? Okay.
So, this one is a comment. I just want to clarify. The margins you use are either 1 millimeter or 2 millimeters or do you use other margins in between the two based on distance from isocenter? So, so far, we have been utilizing either one or 2-millimeter margins. Just to clarify. And thank you for that question.
Dr. Tenn: Yeah. I believe the software currently only allows you to have 1 millimeter step sizes in this PTV margin. I guess you could manually try and contour in intermediate PTV margin, but currently, we're just using either one or two.
Dr. Agazaryan: This is a very good question. In fact, we did all of this. The cavity counters to validate those to the ion chambers. Are they fit to the cavity or do you include the chamber wall?
Dr. Tenn: It's including the chamber wall. That's a good question.
Dr. Agazaryan: So, it does. Yep. Did you beam profile in RTdose for multiple brain mets beam profile? You need to clarify the question for me. I'm looking around. We're not clear about the question.
Dr. Tenn: So, the ques...
Dr. Agazaryan: Question first.
Dr. Tenn: Yeah.
Dr. Agazaryan: Which RTOG protocol was used for the prescription?
Dr. Tenn: So, the physicians here are selecting doses. I don't believe they're using a specific RTOG protocol. We can get back to you on that. I would have to ask one of our physicians.
Dr. Agazaryan: Oh. Did you use the same profile in RTdose for multiple brain mets? Oh, I see. The way I'm interpreting this question. Did you use the same beam data from iPlan to multiple mets? And the answer is yes to that. If that's the question, I don't see a name, the answer is yes. The same data set was used for MBM, Multiple Brain Metastases, as for iPlan. Can you describe a step-by-step process to recalculating with Eclipse? Please, Dr. Tenn.
Dr. Tenn: Sure. Once the plan has been imported into Eclipse, you can select the Eclipse beam model that you wanna use, calculation model. We can...in our system, we have commissioned both the AAA and Acuros. For these cases that [inaudible 01:11:14] has been showing you, I just selected the AAA algorithm with a 2-millimeter grid size, and then recalculate. Because it's a conformal plan, once you recalculate, you can manually enter the monetary units per field. So, once the calculation was completed, I just manually entered the same monitor units as we're coming from MBM, and then you can do the dose comparison from there.
Dr. Agazaryan: The next question is a very good one and a tougher for me to answer, but I'll take it. I'll try. Do you perform [inaudible 01:11:50] QA for multiple met cases? So, let me see. This is done with an LLC. I'm assuming the question is per patient. There are institutions where it's required that you do per patient [inaudible 01:12:10] test prior to treatments. We don't do that per patient. Maybe we should revisit this idea. But I should mention, just to clarify, there is also [inaudible 01:12:22] test within the ExacTrac software that we do. If the user needs or the attendee needs [inaudible 01:12:30] test with the radiation field with an MLC prior to treatment, that's a good idea to do. Our current base decisional protocol doesn't require that. But we do [inaudible 01:12:44] test within ExacTrac. Unfortunately, the same nomenclature is used, but they mean two different things.
In MUCheck, do you check out the central PTV? I think I can take that one. So, we've tried both. Good results come out even at the ones at the periphery, but you could also manually enter at the central PTV and you will get even better results. So, yes, you can do that.
Let's see. The question 33. Did you use the ArcCheck plus 3D VH for single isocenter multiple mets dose verifications? I'm assuming this is an option between their software, right? 3D DBA. So, the answer is no. But as you saw in that one slide, there is this very interesting results with the ArcCheck. What ion chamber did use in the cases that are shown here? Oh, that was the PTW 3D pinpoint ion chamber that has 0.016 cubic centimeter volume. They have two models. This is the 3D pinpoint. By the way, you can order with RTsafe whatever SRT you would like. Our specific one is this one, but you can order for other chambers.
Bogdan: Great. Well, Dr. Agazaryan and Dr. Tenn, thank you very much for your lecture and for answering questions. And thank you all for attending the webinar. As I mentioned, the webinar recording will be available to you about an hour after we conclude today. You should receive an email from Novalis Circle with the link. And you can always re-listen to the webinar and also check on credits if you require them.
I would also like to introduce the upcoming webinars moving forward. Brainlab will try to do about two webinars each month, one will be a clinical and one will be a technical one. The next technical webinar will be on teaching how to plan for simple targets with multiple brain mets SRS. The next clinical webinar will discuss Novalis certified and the audit process. And we have Dr. Tim Solberg from UCSF and Jonathan Howe from Norton presenting that. And the last one, if you are attending ASTRO, we are having a Novalis Circle symposium Sunday night with the clinical focus, again, being on brain metastasis. Thank you very much for all your attendance. Thank you very much for our lectures and have a great day.
[01:15:29]
[silence]
Today I had the privilege to welcome Prof. Agazaryan from UCLA as our first clinical speaker. But before I turn the microphone over, just a few organizational items. All our webinars are recorded and will be available as video files on the novaliscircle.org website for later review. About an hour upon completion of the webinars, you should receive an email from Novalis Circle with a direct link for the recording. Also, for today, questions may be submitted in writing via the chat interface of the webinar. And once the lecture concludes, we will address all your additional requests for information.
This topic highlights the UCLA experience with commissioning and patient-specific QA for the Brainlab multiple brain mets SRS element. Dr. Agazaryan and his team have performed extensive work reviewing multiple QA solutions available in the market. And I believe that their clinical experience will be of great value to us all. Dr. Agazaryan is a professor of radiation oncology and the chief of clinical medical physics of the UCLA Department of Radiation Oncology. He's also a professor of the physics and biology medicine interdepartmental graduate program within the UCLA School of Medicine. Dr. Agazaryan is also an expert panel member of our international Novalis Circle community. Dr. Agazaryan.
Dr. Agazaryan: Thank you. Hello and welcome. I truly appreciate you taking time to attend this webinar regarding the UCLA experience with commissioning and patient-specific QA for single isocenter treatment technique for multiple cranial targets. Our webinar's time yesterday coincided with the new iPhone announcements from Apple. So, maybe we were competing with Apple for bandwidth. We are excited to present our findings and learn from your questions and comments. We have an international audience today with many participants from Latin America, including Mexico and Colombia, Africa, many participants from Russia and numerous people from Europe and Asia. We are honored by your participation and excited to present our findings. There are also numerous people in the room. Everyone is smiling and ready. So, let's get started.
As the title of the presentation suggests, the presentation will include our commissioning and validation efforts as well as our results and recommendations for patient-specific QA methods. To start with, I include my disclosures and credits. Numerous radiation oncologist, neurosurgeons, and physicists have contributed to this project. I would like to specifically thank my colleague, Dr. Steve Tenn for his major contribution to the project. He will be present during the Q&A session. Our group also thanks all vendors who diligently worked with our group on implementing methods with their products. The presentation objectives have been well described in the invitation sent to you. I will, however, mention that towards the second part of the talk, we will discuss specific patient treatment plans and quality assurance examples as well as recommendations for planning margins and image guidance tolerances for treatment.
Let me start by mentioning that Brainlab's implementation of multiple brain metastases SRS treatment with a single isocenter uses enhanced dynamic conformal arcs. Per vendors that recommendation, we will refer to this product as MBM SRS. There are competing products on the market that have not been included in this work. Some of these use volumetric arc modulating therapy, VMAT. So, I should mention that depending on your state regulations and institutional policies, the delivery technique may have an impact on your patient-specific way requirements. At our institution, we have decided to intensively test and QA these patient-specific plans, although they're not VMAT plans until we have collected enough experience to revisit this decision later on.
So, on this slide, I summarized our validation and commissioning efforts. We have worked with multiple vendors to evaluate and compare their products, results, and workflows for commissioning purposes. We will make some recommendation in this regards. Many of the methods were utilized for initial validation of the software prior to even commissioning. So, realistically, typical users could choose a few of the methods presented here for their commissioning purposes. By the way, many of these methods could also be used for patient-specific QA. Just in general, we have used the race as you can see on the slide, MapCheck and ArcCheck. We have used different softwares for secondary calculation, some nucleus PerFRACTION, Eclipse, ModusQA, Mobius3D, MU Check. We have used jails RTsafe and ModusQA.
So, let me start with the very first case that we have evaluated prior to our clinical implementation. This patient had 10 brain metastases. The prescription for these 10 targets ranged between 14 gray and 18 gray based on the target size and the location. Multiple brain metastases elements uses the pencil beam dose calculation algorithm. So, the calculation method uses 1.25-millimeter adaptive grid size. The dynamic and has to follow all calculations use five degrees arc steps. And this will come into play in a moment and during the presentation. We will discuss later in the talk the consequences of this specific parameter on the QA methods.
So, to start with, we have compared the MBM calculation, those distributions, with the AAA forward calculations, the Eclipse AAA calculations. I should mention maybe that the variant accuracy algorithm was also used for comparison. However, we faced some interesting challenges for off-center targets with this algorithm that are still being investigated. And what I mean is with the accuracy algorithm. So, our results shown here are with AAA. So, the plans were transferred from the original planning system to Eclipse and the original monitor units from the MBM were utilized to forward calculate the dose distribution in Eclipse. We have investigated the effects of different calculation grid sizes for AAA forward calculation as shown in this slide. We have used many different sizes.
To keep this discussion brief, we present a comparison of MBM to AAA calculations with only 2-millimeter grid size. Please note that on the X-axis is the target volume, hence, the data is sorted in terms of target size from left to right. The Y-axis is the percent difference between MBM and AAA calculations. In this example, the maximum difference in mean dose is less than 5% between these two calculations. And as you can see, it gets larger when the target size gets smaller. This data gives us confidence that we're dealing with clinically small differences that are likely due to AAA models in our system for small targets. With that said, some institutions create separate AAA model for small fields. We believe with this type of approach even better agreements can be achieved and what we show here. We can Eclipse one dose profiles, 1D dose profiles can be visualized and evaluated.
As it can be seen in this example, there is an excellent agreement between two profiles, the red and blue. However, we further quantify disagreement. What we did is we took the MBM and AAA distributions and export it to PTW very soft software. This is a software that allows us to do more quantitative analysis of comparison. 2D gamma index analysis are performed with acceptable clinical results as shown here. Also, 2D dose distributions and 1D dose profiles are also clinically acceptable. And we're just showing you a few examples in this slide. Next, we measure the dose distributions utilizing EBT3 films in QuiCk Phantom. Measurements were compared to distributions obtained from MBM as well as AAA. I should mention that later in this presentation, we will also discuss measurements with a new EBT extended dose films, EBT-XD. And the results are, by the way, even more impressive with that one.
The agreements between measurements and calculations were clinically acceptable shown in this slide. We have observed discrepancies when ASCII export feature is utilized within MBM software and imported into FilmQA Pro. We found a workaround for this to export the 3D dose to third-party application and then they can export planar those. This issue with the ASCII export and import are still being investigated and we don't have much more details about that. But the point is, you can use third-party software to do this comparison. 2D and 1D analysis of the measured dose distributions compared to calculated dose distributions showed clinically-acceptable results with gamma index calculations in high 90s using 2 millimeters and 2% criteria.
So, now let's transition to using other types of dosimeters. Here we made an attempt to measure 3D dose distributions instead of 2D. Using single use radiochromic polymer gel from ModusQA. We use the ClearView gel shown in this image in a second with an optical CT scanner. So, I just wanna be clear, the imaging is with an optical CT scanner. ClearView comes with the jars shown in this image, different sizes or shapes. And what I'm showing you in this slide are two irradiated gels, one for multiple brain metastases and the other one is validated spinal radiosurgery case. We did that for a reference. Optical changes can be seen in these images. I should mention the best results are obtained when the gel resists around 40 gray dose, hence, often the plans may need to be re-delivered or scaled up. We choose to re-deliver plans to preserve the dynamics, and I'm not saying gantry movements. As you know, if you scale up the plan instead re-delivering, then you are changing the MLC and gantry movements. So, in other words, we wanted to QA and measure the plan that will actually be delivered to the patient. And scaled-up plan is not the same.
This is the analysis of that multiple met MBM case in a gel. Generally good agreement is observed in terms of profile shapes shown on the bottom, the red line and the blue line. There is an agreement between those lines. However, we have difficulties associated with alignment and scanning artifacts because of impurities in the gel. And you see some of these artifacts in the images on the second row on the bottom. Large absolute dose discrepancies were also observed. With more experience, care, and product improvements, better results may be possible. So, that was one type of a gel, which was scanned with an optical CT scanner.
Now, I wanna transition to, again, polymer gel that is not scanned with optical CT scanner. It's scanned with MR. In our commissioning work, we have utilized RTsafe polymer gel dosimeter. We obtain impressive results with this gel dosimetry. Patient-specific 3D phantoms can be printed by RTsafe based on patient's CT scans then you transfer to that. However, this step does not need to be repeated for future patients as the same phantom can be refilled with a gel for future use. So, we took this process a bit slow. First we said, let's validate the dosimetry method itself prior to using it to validate the multiple brain metastasis software. So, we ordered a simple cube phantom shown in this image that comes with calibration vials shown on the left. This particular gel can accurately measure doses in the range of 3 to 35 gray. There is another gel available from the same vendor that can accurately measure those in the range of 0 to 12 gray. We will come back to this. We have utilized both gel types. There's particular one I'm showing you here, uses range 3 to 35 gray. So, you are able to measure higher doses, but you are using accuracy below 3 gray.
Calibration gel tubes were irradiated to doses ranging from 0 to 25 gray with 5 gray increments. And I really like this slide. Here, the calibration vials were sort of calibrated with a 15 MED electron beam at around 4 cm depth because this is where the distribution is relatively uniform using a setup shown in the picture. So, it was somewhat the improvisation of how to uniformly radiate these gels. There may be other solutions that you can come up with. Although the dosimetry is achieved with MR scanning of these vials, I wanna bring to your attention that irradiation of gels creates this polymerization that can be seen in the pictures going from 0 to 20 gray, even visually, you can tell that they are polymerization changes happening. By the way, the irradiation of these gels establishes the sensitometric curve for dosimetry. So, we delivered a very simple 3D conformal plan to this phantom. And I wanna bring to attention again that even in this picture on the cube phantom, you can see the polymerization changes in the phantom. That cloud area where the dose distribution went. First, we tap it to scan the cube gel with our 0.345 Tesla ViewRay, MR-guided radiotherapy system. However, the signal-to-noise ratio was prohibitive to obtaining a clinically meaningful analysis.
Here's the phantom and the calibration vials during that attempt of scanning with the ViewRay phantom...scanner. Then, we MR scan the phantom using 1.5 Tesla MRI scanner. This give a good clinical results. Our MR [inaudible 00:19:11] had to work with the radiology department to install specific QT sequences provided by the vendor. So, I want the users to be aware of the fact that you have to use this specific sequences and work with either your department or radiology department, whoever owns the MRI scanner. I do wanna mention that the second type of the gel which is coming later in the presentation will not necessarily require using calibration vials.
Subsequently, once we've establish the ground truth that we are pretty good agreement between the calculation and measurements with the gel, the gel validity dose established, we went ahead and ordered 3D-printed patient-specific phantom for one of our patients. In fact, we took this phantom to the physician and said, "Oh, by the way, do you recognize the patient? This is the 3D-printed patient." We did this by transferring the CT scan to the vendor. As it can be seen in this image, the phantom has two compartments. The central compartment is within the skull, closed up with a larger cap, and contains the gel dosimetry inside. The outer compartment, closed up with a smaller cap, is filled with water. The gel is temperature-sensitive and it's shipped with ice packs and thermometer.
So, what we see here is the phantom during the end-to-end testing. So, first, the phantom was masked and CT scanned. That's one of our therapists training the mask for the phantom, pretending it's the patient. Then, based on the treatment plan, the phantom was aligned at the treatment position in the treatment room using ExacTrac cranial array. The array is shown on the right side. And then we employ the ExacTrac IGRT process. In this picture, some example images from ExacTrac IGRT process. And I just wanna bring your attention that you do get contrast differences with the phantom. Enough information to position the phantom just like you would do with your patient. So, here we go. What you see here is the video captures. We show the delivery process on that phantom that I just showed from two camera viewpoints in the room.
If I zoom into the MLC movements, and what you're seeing here now is the MLC movements, there are many things that you can observe. The delivery method is a conformal arc. There is really not that much modulation happening here, right? You see these open fields moving around between the jaws. It's not a volumetric modulated arc therapy. Additionally, 6 out of 10 targets are covered. You only see six openings in the capture. The system selects the best combination of targets for each of the arcs. This is because two targets aligned among the same sort of MLC leaf pairs cannot be effectively treated with the same arc. Yeah. If necessary, the arc will be repeated in the reverse orientation, the same arc will be repeated, to cover another subsets of targets. So, if in the planning system that you see the same arc is going clockwise and then counterclockwise, those may not be covering the same sets of targets. Strategically, they're programmed to cover different targets and provide good coverage for all of them.
Now, that's the delivery and then the gel is analyzed. And what I'm showing you here are the results. I should mention that the company, in this case, RTsafe, provides a comprehensive PDF report that includes many aspects of those dosimetric analysis. In this image, what you see here, by the way, pay attention you see the calibration vials around the CT of the phantom. Around the skull, you see those different intensity circles. Those are the calibration vials being scanned by a scanner. You also see long D profiles through some of the targets as examples. Some pretty impressive results. Each upper one is the profile and the lower one is the gamma index calculation of that 1D profile. And other sets of examples, similarly, excellent results are displayed here. The reason I included this slide is because I'm showing profiles through one target, profiles through two targets, and then on the right bottom, I'm showing profiles through three targets. The red color is the TPS, the planning system, and the blue color is the measurement with depressive results. I should mention, again, 2D distributions can also be visualized and then 2D gamma index calculations are also provided by the vendor in this report.
We worked with a vendor and we made a decision that we actually want to measure distributions with an ion chamber. We wanna validate absolute dose measurements with an ion chamber. So, we asked them to 3D-print three ion chamber phantom for most reliable dosimetry measurement with ion chambers. So, we designed this phantom and they printed in 3D. In this case, there is only one for the phantom since there is no gel dosimetry involved. The user fills the phantom with water and the phantom is reusable, making it a good candidate for periodic system QA. We are thinking of using this phantom on a monthly and yearly basis. This phantom is manufactured to hold three PTW 3D pinpoint ion chambers. The volume of design chambers is 0.016 cubic centimeters. So, it's pretty small volume, sensitive volume, allowing us to measure 3D small dose distributions. What I'm showing here is the CT scan of the phantom. Again, you can observe that there is no gel, it's uniformly sort of water all around the skull and inside the skull. And you can also observe the three locations of the ion chambers. We went ahead and created a plan that delivers 14 gray, 16 gray, and 18 gray to each of those ion chambers. And I'm showing that in that green box on the left. Those are the prescriptions. Here, what I'm showing is the distribution, the dose distributions around those ion chambers.
I have an important sort of point to make with this slide here. While the prescription doses are 14 gray, 16 gray, and 18 gray, it's important to note that the mean and max doses are delivered to the cavity are much larger. And I'm highlighting those in the green boxes here. In fact, the mean doses are 25% higher than the prescribed doses. And the maximum doses are 37% higher than the prescribed doses. There's really nothing wrong with this per se. I just wanna bring to your attention that this speaks of the fact that the distributions generated with MBM are more heterogeneous than distributions we have typically simply by plan. You could simulate this by-plan as well in homogeneous distributions. But typically, you see more homogeneous distributions. You can argue either way. I mean, this could be better or, in some cases, worse for clinical cases.
So, what I'm showing you here on this graph is the ion chamber statistics from the planning system in terms of minimum, mean, the maximum doses versus measured dose. The measure dose is shown in red. Maximum is in blue, minimum is in green, and the mean is in yellow. Yeah. So, the most relevant statistics is the mean dose since the ion chamber measures approximately the mean dose. What I did is I took other statistics out here and it showed mean dose verses, the measured dose on this slide just to sort of reduce the clutter on the graph. Again, I'm gonna bring to your attention one more time that measurements are in the range of 19 to 25 gray while the prescribed doses are in the range of 14 to 18 gray. Speaking of the denominator of the distribution. And then I converted this to percent differences between calculation and measurement, mean dose from the planning system versus measurement just to show you that there's pretty good agreement between both calculation and measurement. The maximum deviation from the measurement is less than 2%, 1.7 for the chamber two.
We did not stop there with the three ion chamber plan. We delivered the plan on to a gel phantom for two purposes. So, let's see. This particular gel is a new version of a gel from RTsafe. I mentioned this previously. With a 0 to 12 gray range, and we wanted to validate the gel dosimetry itself because we just measured that plan, we can deliver it on a gel and validated dosimeter method. The vendor offers a relative dosimeter method that does not require irradiation of vials. As you saw in a previous example with the other gel, irradiating those vials was not the most straightforward thing to do. We were interested to see the results of this method and compare it to the other one. Yeah. And here are some results of that ion chamber plan delivered on that gel that I just described. On the left is the MR scan converted to dose, and then you see profiles 1D, 2D, 2D gamma analysis, 1D gamma analysis. All of this is provided in the report to you.
Here's another target. This is a more superior target. The 1D profile agreement is so good that it is difficult to distinguish the two profiles. If you look at those 1D profiles on the top, the TPS in red and RTsafe in blue, really difficult to distinguish those two. Again, we are also showing 1D, 2D gamma analysis as well as 2D distributions on the right upper corner. I wanted also to show you two profiles going through two targets in a coronal plane in this case just to give you different perspectives. And also I'm showing you a plane, coronal plane, that goes through two targets with the 2D distributions on the right upper corner so that we're not selecting one or two just to show it. So, overall, I've seen pretty good results with this dosimeter. And this is a very important slide that I wanna concentrate on. So, let me think about this for a second. This particular gel is a new version of a gel that goes from 0 to 12 gray. We've talked about this.
One of the reasons why the company created this gel with lower range is to be able to accurately measure lower doses for providing DVH analysis for objects. Remember, these are mostly for radiosurgical cases. So, it would have been great to use the other gel because your doses typically are higher than 12 gray. But the advantage of this gel is that it goes from 0 to 12 gray. You get very accurate results for very low doses. When you get DVH analysis like I'm showing you on the left between TPS and RTsafe, it is very important to bring to your attention that these 3D measurements and ExacTrac positioning, they are simulating our patient positioning process. So, the DVH results speak of our...or more scientifically our confirmation of rotational and translational alignment accuracy in 3D versus when you use a race, there is less of an evidence of that. So, 0 to 12 gray gels provide measurements 0-100% range, so you get pretty nice deviation results. The other gel which is 3 to 35 gray also like that gel, provide accurate measurements from 10% to 100%. And the beauty of that one is you can deliver 20, 25, 30 gray doses without scaling down. So, to measure higher than 12 gray with the second gel that goes to lower doses. Your user is limited by the accuracy of the measurements below three.
All right. Now, I'm transitioning to now from physical measurements. We've talked about film, 2D measurements. We've talked about 3D measurements. Now we're going into calculations, Mobius3D, in this example. Mobius Medical Systems has a product named Mobius3D for secondary 3D calculation. This product helps with many QA activities, one of them is 3D dose calculation with an independently verified model and collapse cone algorithm. The software provides an excessively detailed PDF report, comparing the original plan to your calculated dose distribution. Initially, we observed results with good profile agreements, except at the peak of the small targets. These results, they're based on machine data provided by Mobius3D. So, it was not our machine data. It was totally independent calculation. The user can provide small field data with Eclipse commissioning conditions to the vendor and hopefully, that will sort of improve the results. But we have decided to do in our case, the vendor adjusted the factor in their calculation algorithm to match our results. I think this factor is similar to Dynamic Lift Gap or DLG factor in Eclipse. The adjustment greatly improved our future results as you can see in this slide.
It is reasonable to question these types of factor adjustments, but we suggest that, in this case, it is acceptable practice since the algorithm has been well validated with measurements and other methods. In our clinic, we use MUCheck for secondary MU calculations. And we investigated the use of MUCheck for MBM. Our original version of the software did not support this type of calculations, but the vendor provided that capability in their next and current version of the software. The actual... I showed the results of MUCheck for three-ion chamber plan. Some beams and points readily show agreements between 5% for first calculations and other beams like shown here require small volume averaging distance, typically 1 millimeter to get acceptable results.
We have also evaluated some nuclear products or secondary calculation in QAs. This web-based SunCHECK QA platform offers impressive combination of tools for numerous QA methods. What we utilize is the DoseCHECK and PerFRACTION. Some of the options we have explored are included in this slide. We have performed secondary 3D dose calculation using DoseCHECK. We have done 3D dose reconstruction using DynaLog files after the delivery of the final phantom. So, while we are measuring with the film and ion chamber and other things, the machine was generating MLC lock files that are later utilized for 3D calculation. We did also investigate the use of 3D QA with EPID measurements. However, the specific EPID in use in our clinic has been having hardware issues and that investigation is still being addressed. Using 0-millimeter search radius, the agreement in this particular case for all points is less than 6%, shown on the left grid box. The gamma passing rate with 2 millimeters and 2% is 100%, shown on the right. With only 1-millimeter search radius, the limit in this particular case or all points is better than 0.1%. It's listed as 0.0 on the left here. The gamma passing rate with 1 millimeter and 1% is 98%. In fact, our clinical examples that I will show later show even better results.
One item of notes. The reference points exported from Brainlab software are at the prescription line. So, these points are at the prescription line except the isocenter, which means these are typically at the peripheral of the target, making the dose calculation matching more difficult. So, more to come about this. Now, I wanna transition to patient-specific QA methods. We've talked about validation of the software, the commissioning of the software, and the delivery method. I shouldn't say software for the whole package, right? Many of the tools discussed in a previous section are directly applicable as a patient-specific QA tool. I have a quick summary here in this slide. Yeah? Many of the methods were investigated for academic and investigational purposes. I just wanna reiterate again that typical users need to choose just one or two of these for their clinical practice. Our very first case was a patient with two arteriovenous malformations, AVMs, as you can see in this picture on the left. The MR is on the bottom row. It is important to mention that Brainlab's intended clinical use for the product is for brain metastases. With that said, we have utilized this novel delivery method for treating multiple AVMs.
The superior AVM located on the left frontal region is twice as large as the inferior arteriovenous malformation located on the left temporal region, 1.8 cc and 0.8 cc correspondingly. The prescribed dose to both targets were 20 gray. This is with an understanding that the dose distribution within the target is harder than our experience with iPlan. Also, the 20 gray position covers the module plus 1-millimeter margin in all directions. The secondary calculation with Eclipse planning system showed excellent agreements with original plan. Again, the monitor units from the original plan were utilized. Reference point doses were less than 4% different between two planning systems. So, to do reference point analysis, we have created this Excel sheet. The matrix gets really complicated as the number of targets and beams increase. For example, if there are 10 targets and 5 beams, then 50 numbers need to be analyzed realistically.
We went ahead and made an EBT-XD, extended dose film, and ion chamber measurements with the Ashland QuiCk Phantom. We have calibrated the EBT-extended dose film to 32 grays in our clinic. I think it can measure up to 40. It's calibrated to 32. Correct? Yep. The phantom was scanned with infrared markers, highlighting the infrared markers with red circles. This allows us to automatically position the isocenter at locations where the ion chamber measures the dose intended for each of the targets. Based on the coordinates of each of the targets, we calculate the location of an isocenter that will place the ion chamber at each of the targets. So, we create as many plans as the number of targets for ExacTrac. Here's one example where the isocenter was placed in such a location that the distribution intended for the first target covers or falls over the ion chamber. Each of these plans are separately transferred to ExacTrac, and then ExacTrac infrared guidance allows for quick positioning of the QuiCk Phantom at each of those locations. On the left picture, I'll show you the phantom with infrared markers. This particular example is the delivery of data to an implant I showed you. The ion chamber measurements showed an agreement with calculation within 3%. Yep.
So, we went ahead and also analyzed the field measurements in a similar manner. And the 2% and 2-millimeter gamma index analysis yielded 97.5% results. This is the plane center to the first frontal lobe target. And this is the plane center to the second temporal lobe target. Here we tighten up with 1% and 1-millimeter gamma index analysis with 97.4% gamma index. I mean, pretty impressive results as you can see there. Next, we employ the PerFRACTION software. Actually, it's the dose check calculation. The 3D gamma analysis with 1 millimeter and 1% or 98.67%. [inaudible 00:43:35] dose with 0 millimeter search radius, so, 4 and 9.5% deviation. Just a reminder, these are points at the periphery of the target shown in the red box. When we added second sets of points shown in the green box that are centered in the target, even with 0.0-millimeter search radius, these results are better than 3%. And note that they actually said that dose discrepancy is about 19%, which is at a low dose region, but I'll come back to this in the next slide. When we have 1-millimeter search radius, it significantly improves the results. And as you can see, the numbers went significantly down including the isocenter dose that went from 19% to 0%, which is somewhat surprising.
Then this is very different. This is DynaLog file. When we delivered on a film, DynaLog file was created of the MLC movements. And that was used for reconstructing dose distribution. And with 1-millimeter search radius, the point dose analysis are better than 0.1% and 3D gamma passing rate was better than 98% with 1 millimeter of opposite criteria. The MUCheck for this case was also impressive, with 1% with 1-millimeter volume averaging and 3% for the other target. This one is interesting. So, one of the fascinating results we obtained was the Sun Nuclear ArcCHECK. So, we initially didn't predict what will happen with this. Sun Nuclear ArcCHECK dosimeters are located at the distance from the center of the phantom. So, the aramid, those calculation exhibits a streaking artifact because of the five-degree calculation steps. You can see that streaking artifact there now. Even then, there is a surprisingly good agreement at the measurement points shown on that profile that I'm showing on the bottom with the gamma analysis.
One could experiment with placing the isocenter positions where the high dose region falls on the diodes instead of at the center. I suspect there may be problems with that as well relating to angular dependence of diode up because many of the beams will now graze the diodes of large angles. So, while this device can be used for QA purposes, it has the limitations discussed earlier for this specific use. Our second clinical example is a patient with two metastasis, left thalamic target to step 16 gray to cover 99% of the target, right thalamic target, prescribed 18 gray to cover 99% of the target. The Eclipse calculations were between 30%. MUCheck calculations were 2% for all beams except with one that was 5.5% deviation. As you can see the DoseCHECK results for this case were better than 1% for both targets. And the isocenter, I should mention. This calculation is with 1-millimeter search radius. The 2% and 2-millimeter criteria gamma index is 99.5%.
We went ahead and measured the right frontal target prescribed to 18 gray and the ion chamber measurements showed 2.5% deviation and the left alembic target prescribed 16 gray and the measurement was 1.25% difference. So, well within acceptable criterias. The extended dose EBT measurements were made for both targets. And both targets show impressive 98.4% and 96% gamma passing rate with two and two criteria. I'm going to the third clinical example. This is the last clinical example. Our third clinical example was five metastasis, four targets on the left and one target on the right. Two-millimeter margin was used for all targets for PTV creation. All targets were prescribed to 18 gray to cover the 99% of the volume. The left superior frontal target was larger. It was 2.6 cubic centimeters. The rest are small targets, less than 0.5 cubic centimeters. Eclipse's calculations were between 7%. MUChecks were within 6% for all beams. The DoseCHECK results or in this case they're better than 5% for all targets on the isocenter. The search radius is 1 millimeter. The gamma passing rate 99.9% with 2 millimeter and 2% criteria.
So, we selected two targets to measure with an ion chamber out of those five because the others are small. Left superior frontal target, 18 great prescription showed agreement within 3%. Left anterior frontal target prescribed to 18 gray showed an agreement between 3%. So, both were less than 3%. Yes, there may be challenges with much smaller target measurements. So, what we did for that, we went ahead and measured all five targets with extended dose film. We're showing you absolute dose comparisons with excellent 1D profile results here. So, we have confirmed with my colleagues who have specifically made these measurements that there were no adjustments made. These are absolute. All five targets pass 2D gamma with 2 millimeters and 2% criteria better than 98%. So, here the point I'm making is even the small targets were measured perhaps.
It was reported to us the Sun Nuclear ArcCHECK can also be utilized for patient-specific QA. Our colleagues, Dr. Javad Rahimian, and others from Kaiser Permanente in Los Angeles provided the following data from one of their patients. They collapsed all the table angles to zero for these measurements. The beauty of this method is that a single isocenter location or like the ones that we made with other phantom can measure most if not all targets. Their results are shown here. Again, this is from Kaiser. Gamma passing rate is 3% with a 3% and 3 millimeters is 99.5%. I'm guessing their criteria is three and three. In our case in our clinic for radiosurgical cases, two and two and one and one is evaluated. They do show excellent profile agreements with this device.
So, in closing, I wanna transition to an extremely important point that needs to be considered when treating multiple brain metastases with single isocenter. Traditionally, we have treated each of these target with an individual isocenter at each of the targets. If I simulate a 3D degree rotation for each of the targets, this is what will happen, right? You'll see the movements of this sort of radiation fields with a 3D degree rotation. Really not that big of an impact. With a single isocenter treatment, the effects of these rotations become much more pronounced. Now, I'm showing you one isocenter in the center. And if you simulate the motion with some targets, almost completely can be missed with 3D degree rotation. Look up a green one on the left and the yellow one on the right.
In this graph, I display the translational misalignment that can occur with rotations as a function of distance from isocenter. So, the X-axis is the distance from isocenter. The orange curve is the 0.5-degree rotation. And a target that 6 cm away from isocenter 0.5-degree rotation can result to 0.5-millimeter translation. That's the point I'm trying to make here. So, it set ExacTrac thresholds at 0.5 millimeter, 0.5 degrees. The approximate allowable of misalignment will be around 1 millimeter. The overall accuracy may be in the order of one to 2 millimeters. We're not buying, but we suggest that 2-millimeter margins for PTV creations should be used for targets that are far away from the isocenter. So, just to summarize, we recommend 0.5 millimeters and 0.5 degrees tolerances or thresholds as it's worded in ExacTrac. We also recommend 1 millimeter and 2-millimeter margins based on the distances from the isocenter. Smaller margins can be considered for benign conditions. Yeah. So, also, we could consider two isocenter treatments for multiple targets that are distributed in clusters at large distances. So, you could divide them to two groups and treat with two isocenters, then you can utilize smaller margins.
So, let me get back to credits. I want to thank, again, all my colleagues for their contribution to the project. This work was not done alone. And thank all the vendors who worked with us on implementing methods with their products. It was my great pleasure to present the UCLA experience with commissioning and patient-specific QA for single isocenter treatment technique for multiple cleanup targets. We do appreciate that you took time out of your busy schedules today and attended this webinar. We will now open the floor for questions and comments. Correct?
So, the first question, in the iPlan, for multiple mets, where do we normalize at isocenter or target mean? Do we do it for target mean or in Eclipse? Steve?
Dr. Tenn: So, this is a question about iPlan for multiple mets. We normalize the target. The way we normalize targets prescription in the iPlan system is a volumetric normalization. So, we typically try to achieve at least 99.8% coverage on the PTV for medicine. So, that would be essentially on the periphery of that target.
Dr. Agazaryan: Thank you for that. Second question. Also, where should we keep the isocenter in brain mets with more than two metastases? I think I can take this one. So, one feature of the software is that the isocenter is automatically placed at the central volume of all mets, all selected mets, I should say, for that treatment. The third question. Two-millimeter is a lot for less than 1 cm measurement, isn't it?
Dr. Tenn: This is a question about the PTV margin that we're using for a lesions smaller than 1 cm. And I think as it was showed in the slides regarding uncertainty for the rotation and translation and that we use currently 0.5-millimeter and 0.5-degree rotation tolerance in ExacTrac. Based on the slide that you were showing, I think 2 millimeters is appropriate.
Dr. Agazaryan: Yeah. But there's a good point here, right? It's a large margin. It's a clinical decision to [inaudible 00:55:33]
Dr. Tenn: Yeah. Having said that, if you have a cluster of lesions that are very close together, you can consider using smaller margins because your uncertainty will also shrink.
Dr. Agazaryan: Okay. It seems like 2%, 2-millimeter gamma is a poor metric for this geometry, sensitive treatment technique. What other metrics were considered? We utilized 1%, 1-millimeter criteria as well. What other metrics were considered?
Dr. Tenn: Clinically, this is...the gamma index is what we use to review these plans. So, that's why we went with this particular metric to look at the accuracy, the spatial accuracy for these. There may be other metrics that you're interested in and we can definitely consider those.
Dr. Agazaryan: Yep. How high is the phantom setup uncertainties? Maybe we should mention that the thresholds... It's not necessarily the uncertainty, but the thresholds that we're utilizing is 0.5 and 0.5 millimeters and degrees. The uncertainty, as I mentioned in one of my slides, with just these parameters is up to 1 millimeter. And overall, our uncertainty based on the data that we have that was not shown here probably between 1 and 2 millimeters
Dr. Tenn: Yes, I agree.
Dr. Agazaryan: How was the gel dosimetry specially registered to the plan?
Dr. Tenn: The gel dosimetry... So, this is a question about the RTsafe phantom I believe. What happens is once we acquire the MR image of the phantom, following a radiation, we send that MR image off to the RTsafe company and they do in-house registration of that Phantom to the plant. So, yeah.
Dr. Agazaryan: So, let's look at this one here. What PTV margin do you recommend to accommodate the apparent 2-millimeter position of discrepancies? We should mention that up to 2 millimeters not...
Dr. Tenn: Yes.
Dr. Agazaryan: So, it's up to 2 millimeters. If you have a cluster that is really not more than five CMFA or so, we really don't have to utilize 2 millimeters. But if they're spread around more than that, you either can utilize 2 millimeters or you could split them up two clusters and treat them with a couple of isocenters. It's still significant improvement. One of the examples...the five multiple met target when we were treating that, typically we go to the machine, we meet the physician five times to verify each of the isocenters. We only have to go there once. So, if you have 12 targets, for example, and you have to be there for 12 individual treatments, in this case, you will be there only twice. So, it's still a significant improvement even if you use two isocenters. Yeah.
So, let me see. Did you try to let Eclipse AAA algorithm calculate the new values? I think so. Based on the same prescription as Brainlab utilizes. I think that refers to the discrepancy, right? So, if you do the reverse instead of forcing the MUs, you use the dose to calculate the MUs, but discrepancy will be within 5%. So, the answer to your question that's coming from one of the attendees is it will be within 5% at least in the examples shown here. Looking at the controlling tools of Brainlab and Eclipse, what according to your [inaudible 00:59:43] better Eclipse or iPlan? That I will defer to the expert here. Go ahead.
Dr. Tenn: We tend to use the Brainlab. I'm not gonna say one is better than the other, but it really depends a lot of what you're used to. We typically use Brainlab contouring tools here. We're very familiar with them. The neurosurgeons, the radiation oncologist are all very familiar with them and are very confident in them. So, we almost always, I would say, better than 99% of the time will do the contouring and Brainlab rather than Eclipse.
Dr. Agazaryan: I think I can address this one. What is the minimum field size in your beam commissioning measurement for Eclipse? So, this is an interesting question. Ours is 10-millimeter by 10-millimeter. But I should mention that Eclipse does not directly use measurements. It creates a model based on your measurements. So, when you create a model, you adjust the model to have a good agreement for small fields as well as for large fields. And you have to make a compromise in that sense. If you get a good agreement with lush fields, maybe there will be small discrepancies with the small ones. That's why some clinics create two different models. So, the point I'm making is it's not only the measurement, but also how you adjusted your AAA to match the small fields. And if you do match small fields, you may be off for the large fields. So, that's why some clinics have two different AAA models in their software. Yeah?
Dr. Tenn: Yeah.
Dr. Agazaryan: So, should we conclude here or do you have questions? You're okay?
Dr. Tenn: Yeah.
Dr. Agazaryan: So, how multi-mets compare with individual isocenter for low doses? That's an important question. I think outside of the target and the low dose [crosstalk 01:01:46]
Dr. Tenn: Yeah. It's probably similar, possibly better. It really depends. For the single isocenter targets, we've been using no margin and here we're using 2-millimeter margin. So, there's some differences between the two. But yeah, I would say they're similar in the low dose region.
Dr. Agazaryan: So, I'll take the other one. Can you tell, sir, about pencil beam and Monte Carlo calculations in the software? So, let me think. The Monte Carlo is available in the next element that will come out. It's called cranial SRS element. This software here which is the multiple brain metastases element uses pencil beam algorithm. That's the only option you have for now until Monte Carlo will be implemented in the next release. So, Monte Carlo is coming out. Currently, the pencil beam is the only option for that.
We're waiting for questions here. So, between arc treatments, which couch keep arcs? Do you verify the setup? Again, that's an excellent question. Yes, we do. So, we verify... So, let step back. First of all, when ExacTrac is done, we do Cone Beam CT to independently verify the positioning of the patient, and then we repeat the ExacTrac prior to delivering the first arc, and then we will repeat the ExacTrac prior to each arc. Yes. The answer to your question is yes.
Question number 22. Did you use Eclipse for a second check? And if so, what is the small field size in your Eclipse B model? We examined the MU discrepancies found out to be different in very small targets from 2% to 25%. So, I think we already addressed this question to some degree. Depending how your model in Eclipse is adjusted, some clinics use DLG to adjust the model. Your discrepancies can be smaller or larger. It's a model-based planning system. So, when you're adjusting, you're compromising between small fields and large fields. If you adjust for smalls, you will have good agreements for the small ones. So, I think we already answered that. Which custom checks did you do mostly in the software and how did you do it? And what is your protocol? This is for the software, I think, they're asking. Correct?
Dr. Tenn: So, we don't do monthly checks for the software. There's an annual QA that we do for the software and it's pretty much to run a single plan that we've used from the beginning over to check the constancy. It's a typical annual QA for planning system.
Dr. Agazaryan: If you would allow me, Steve. I just wanna make sure we are not misunderstanding. If what you mean is the end-to-end testing or anything like that, with ExacTrac we do that on a monthly basis.
Dr. Tenn: Yes. If the question is about ExacTrac, for sure, it's a monthly test.
Dr. Agazaryan: I just wanna make sure there's no misunderstanding here.
Dr. Tenn: Yes, yes.
Dr. Agazaryan: Is film your preferred patient-specific way for cases moving forward? In our clinic, that's correct, but in clinics where you have the MapCheck like the Kaiser Permanente group, you could utilize that. So, it's actually clinic-dependent how you would do this. And if you do have access to gel dosimetry, maybe that would be...
Dr. Tenn: Yeah. We've gotten really good results with our Gafchromic film QA process. It's really good and high resolution for the small targets. So, we'll be using it going forward. Not to say, again, that you couldn't use another method, but for our clinic, that's what we'll be using for these small targets.
Dr. Agazaryan: So, let me mention that these are dynamic conformal arcs. And depending on your institutional policies, state, country, you may have different requirements for QA, right? Some institutions, regardless of what type of treatment it is, if it's a major surgery, you have to measure. Some institutions if it's not an IMRT or VMAT, they don't necessarily measure. Yeah? We took the approach of measuring all plans up until we have enough data to revisit that decision. Okay. Any more? Oh, how far have you gone with EPID-based QA investigation? We have done this in the past. We had pretty good results. But with this specific project, when we tried EPID-based QA, unfortunately, the machine...we have a dedicated machine for this types of treatments. And on that dedicated machine, we have been having EPID issues. So, I'm sorry to say that we haven't gone too far with this. But when we have results, we'll be happy to share with you.
After ExacTrac, do you use Cone Beam to do make-shift based on CBCT? Excellent question. So, no. The CBCT is not utilized to do any adjustments. The CBCT is used for validation. We look for up to 1 millimeter discrepancy between ExacTrac and Cone Beam CT. And if it's any more than that, then we investigate why we have that discrepancy. The positioning is done with ExacTrac. And intra-fraction motion management is done with ExacTrac while the patient is on a table getting the treatment. The beauty of ExacTrac is that the table kicks, table rotations, you can get the images and you can verify the patient position. Not only it monitors the patient motion, but it also corrects for table uncertainties. Correct? Okay.
So, this one is a comment. I just want to clarify. The margins you use are either 1 millimeter or 2 millimeters or do you use other margins in between the two based on distance from isocenter? So, so far, we have been utilizing either one or 2-millimeter margins. Just to clarify. And thank you for that question.
Dr. Tenn: Yeah. I believe the software currently only allows you to have 1 millimeter step sizes in this PTV margin. I guess you could manually try and contour in intermediate PTV margin, but currently, we're just using either one or two.
Dr. Agazaryan: This is a very good question. In fact, we did all of this. The cavity counters to validate those to the ion chambers. Are they fit to the cavity or do you include the chamber wall?
Dr. Tenn: It's including the chamber wall. That's a good question.
Dr. Agazaryan: So, it does. Yep. Did you beam profile in RTdose for multiple brain mets beam profile? You need to clarify the question for me. I'm looking around. We're not clear about the question.
Dr. Tenn: So, the ques...
Dr. Agazaryan: Question first.
Dr. Tenn: Yeah.
Dr. Agazaryan: Which RTOG protocol was used for the prescription?
Dr. Tenn: So, the physicians here are selecting doses. I don't believe they're using a specific RTOG protocol. We can get back to you on that. I would have to ask one of our physicians.
Dr. Agazaryan: Oh. Did you use the same profile in RTdose for multiple brain mets? Oh, I see. The way I'm interpreting this question. Did you use the same beam data from iPlan to multiple mets? And the answer is yes to that. If that's the question, I don't see a name, the answer is yes. The same data set was used for MBM, Multiple Brain Metastases, as for iPlan. Can you describe a step-by-step process to recalculating with Eclipse? Please, Dr. Tenn.
Dr. Tenn: Sure. Once the plan has been imported into Eclipse, you can select the Eclipse beam model that you wanna use, calculation model. We can...in our system, we have commissioned both the AAA and Acuros. For these cases that [inaudible 01:11:14] has been showing you, I just selected the AAA algorithm with a 2-millimeter grid size, and then recalculate. Because it's a conformal plan, once you recalculate, you can manually enter the monetary units per field. So, once the calculation was completed, I just manually entered the same monitor units as we're coming from MBM, and then you can do the dose comparison from there.
Dr. Agazaryan: The next question is a very good one and a tougher for me to answer, but I'll take it. I'll try. Do you perform [inaudible 01:11:50] QA for multiple met cases? So, let me see. This is done with an LLC. I'm assuming the question is per patient. There are institutions where it's required that you do per patient [inaudible 01:12:10] test prior to treatments. We don't do that per patient. Maybe we should revisit this idea. But I should mention, just to clarify, there is also [inaudible 01:12:22] test within the ExacTrac software that we do. If the user needs or the attendee needs [inaudible 01:12:30] test with the radiation field with an MLC prior to treatment, that's a good idea to do. Our current base decisional protocol doesn't require that. But we do [inaudible 01:12:44] test within ExacTrac. Unfortunately, the same nomenclature is used, but they mean two different things.
In MUCheck, do you check out the central PTV? I think I can take that one. So, we've tried both. Good results come out even at the ones at the periphery, but you could also manually enter at the central PTV and you will get even better results. So, yes, you can do that.
Let's see. The question 33. Did you use the ArcCheck plus 3D VH for single isocenter multiple mets dose verifications? I'm assuming this is an option between their software, right? 3D DBA. So, the answer is no. But as you saw in that one slide, there is this very interesting results with the ArcCheck. What ion chamber did use in the cases that are shown here? Oh, that was the PTW 3D pinpoint ion chamber that has 0.016 cubic centimeter volume. They have two models. This is the 3D pinpoint. By the way, you can order with RTsafe whatever SRT you would like. Our specific one is this one, but you can order for other chambers.
Bogdan: Great. Well, Dr. Agazaryan and Dr. Tenn, thank you very much for your lecture and for answering questions. And thank you all for attending the webinar. As I mentioned, the webinar recording will be available to you about an hour after we conclude today. You should receive an email from Novalis Circle with the link. And you can always re-listen to the webinar and also check on credits if you require them.
I would also like to introduce the upcoming webinars moving forward. Brainlab will try to do about two webinars each month, one will be a clinical and one will be a technical one. The next technical webinar will be on teaching how to plan for simple targets with multiple brain mets SRS. The next clinical webinar will discuss Novalis certified and the audit process. And we have Dr. Tim Solberg from UCSF and Jonathan Howe from Norton presenting that. And the last one, if you are attending ASTRO, we are having a Novalis Circle symposium Sunday night with the clinical focus, again, being on brain metastasis. Thank you very much for all your attendance. Thank you very much for our lectures and have a great day.
[01:15:29]
[silence]