 Okay, I'm going to give you a little bit of history of my personal experience. Maybe it could be interesting to you. I know a lot of people come from different places. You come from maybe different hospitals. You maybe just start your career and you think you're coming here. You can learn everything. So when you go home, you have everything you can do. Then for the rest of your medical professional, medical physics professional, I'm going to give you some history. So that's what you can see. Okay, I'm just talking about head and neck because I'm only in charge of the neck. Head and M.D. Anderson, okay? So we first started our head and neck MRT on 1998, relatively late. A lot of people has been doing that already. We first started 1998, that's the number one cases. And we have to spend a lot of time, yesterday I mentioned, to convince our doctor, say, this is what we're going to treat. They scratch their head, say, no way, because it feels too small. And then later, we gradually, because they don't have confidence, we have to increase the, you know, treat the patient gradually. So from 1999 to 2000, that's the number 20 cases for two years. Almost one and a half years, see? But then, after they gain their experience, they have the confidence, we increase very fast. Year 2001, within one year, we reached 100. And then 2000, we get 1,000. See the difference? Maybe show the curve, maybe even easier. This is the curve. We started in 1998, and suddenly we increased. But at the beginning, very slow. I think in this population, you need to do also, when you go back, you need to start slowly. And once you get the confidence, then you increase the number. You don't want to make a mistake. At the beginning, if you make a mistake, you're corrected. So you can get more experience. When you feel comfortable, you increase the number. Okay, that's a take-home message. Okay, for all physicians, that's the same story. Okay, this gives you a different way to explain this. Okay, this is a different curve. This is a physician say, for all advanced oral friends cases, when we do like year 2000, 2001, most of them using conventional, meaning 3D conformal therapy, okay? But then after they have this, before that actually nothing, okay? Very small number. But then after they gain experience, by year 2006, 2007, most of them MRT. So they have to have this experience to build up before you do all the MRTs, okay? So that's my story. What I'm trying to say is, even for the physician, for the physicists, at this point, we don't have an idea of what the MRT is. And then at this point, I try to explain. You think the MRT is a good thing knowledge, but I can tell you after maybe five years, 10 years, people will laugh at us. Say, why are you doing this? You are stupid. Just like we are laughing at the people using also voltage machine. Why are you still using water machine? We use the electron or the cobalt machine. In United States, if you use cobalt machine, people think you are stupid, but actually still works, same as saying. But when we first doing this, we're still trying to do it right, doing like a sophisticated way. So that we deliver the right dose. Even though we don't have a good detector, we don't have a good technology, but we need to use the most current technology and the knowledge to do what it is right. But you need to know the knowledge keep changing, keep adding up. See what I'm saying? Right now, okay, maybe use the next slide. This is another, keep the story, okay? So the number increase very fast, okay? Right now, just for the head and neck, every year, we're perpetrating 630 patient just for head and neck, okay? This is just the story. My personal story, okay? When I first practiced, we are using cerebellum. I was cutting the cerebellum. I was, physicists was cutting the cerebellum, okay? When we first practiced. We just serve them to, to get a treatment. And then very soon, we don't, at that time, we have a very big, big mode room. We make the mask, make the cerebellum, make everything very dirty. But after this MRI, all this disappeared. It's not saying when I first practice, I learned everything I know about how to cut the blocks and I stopped there. You know, I have to keep going, keep going. You see what I'm saying? Same thing for everybody. Even though we think the MRT is the best thing to do, by the time we use this MRT, we are crazy, okay? So, but this is the way it is. Things keep changing, okay? That's the rule. One thing's never changes, everything's going to change. Going to be changed, okay? Okay, so give you some software. Okay, that's the hardware things, okay? Using the, from the servant to the, to the VMAG. When I first tried to do that, computer training for the MRT, I was using normal cover three. That's what we use for our first cases. And then that time, we think that's the best in the world. We're using normal. But every time we set things up, everything's fine. We turn the keys and start optimization, take 14 minutes. So I go out to take a coffee break. When I come back, still running. When it's come out, it's not good. I have to change some parameters to rerun it, okay? And once I change and then turn it again, another 14 minutes disappeared. See what I'm saying? So that's the time. And then, but pretty soon, we change everything. We change it from version three to version, software version four, which we can do a lot more, a lot more things. We have to learn very quick. And by the time of December 2003, we decided to do another version. Every time you change a different version, you need to learn new things, okay? But when we wanted to upgrade to version three, we decided not to upgrade to version three. We decided to change it totally, different treatment plan system. We decided to change to pinnacle treatment plan system. At that time, when we do 3D treatment planning, we use a pinnacle. When we do MRT, we use normal. Then when we have like a half pin matching, so on the upper neck, we do MRT, no neck. We do superclap code. We send the two different plans to the treatment plan, to the treatment machine, so that we can deliver the treatment. When physician wants to see this composite dose, that's our physician always want to see. We cannot show them. That's why our physician at the beginning really reluctant to do MRT treatment. That's another reason, see what I'm saying? But instead of using normal, version five, we decided to use pinnacle. Why? Because at that time, before that version, version six come up, 40 minutes is the fast at that time. I said 40 minutes for the break, but at that time, the version five, pinnacle can do MRT treatment planning. But when you turn in the job, you say, oh, everything set up, parameter set up, objective function set up, you say, stop. It takes six, seven hours. So it's no way for us to use the pinnacle to do the MRT plan. We have to make sure everything is set up, and then you go home. Before you finish your work, you go home. Hopefully after the sleep, you come back next day, you have a good plan. But when it's not happening, you lose one day, basically. So at that time, we cannot do the treatment plan using the pinnacle for MRT. But for the version six, then we do the test, they can do like in 40 minutes, and we were very happy. So we quickly change to pinnacle so that we can see composite plan, combo plan, or in eclipse, we call the sum plan. We can sum the plan. So since we changed to pinnacle, we have to relearn everything. We have to unlearn everything from the anomalous, because that's a totally different concept. Then we need to do this now. After we know all this difference, the take home messages, no matter how much you learn over here, I know for sure you learn a lot. But when you go back home, or you practice, you need to know, you need to keep learning, okay? Keep learning things. I'm going to talk today about why we needed to do a MRT QA, okay? Very easy. I think Dr. Pippen has discussed already, because we do have very complex dose distribution. And what the problem we have is this, we call the steep dose gradient because of this MLLC, they are moving, okay? And how precise they are moving to a certain location, you need to do these tests. So this is called the mechanic test. This is like a peak offense. This is not related to a specific patient. You have to have a rigorous QA system like we talk about for these all accessories, make sure they work as it is predicted, okay? So that you can have a good dosimetry with each patient. Well, I'm not talking about this now, but I'm going to talk about how you verify your plan, okay? So when we see the plan before the MRT, see that for the breast, you want to treat maybe 180 a day or 200 a day, maybe 200 a day or tangential or lateral, maybe give 100 a day. And you know roughly 120, 130 MU is probably good enough, okay? If a sicker patient maybe 140, 150. If a wedge maybe 200, you know roughly where the mountain units should be. So you know what I'm saying? You have some quick way to say or this is the way, this number is supposed to be correct or not too far away, at least within 10%. But when you see the MRT, the MU come out, it doesn't make sense at all. Certain field like 100, certain field maybe just like a 50. Another field suddenly 200. Your overall, your MU is about three or four times whatever you had before. And each field has so many variations that you don't really have a clue why we have these different numbers. You have to go back there, see each field, switch subfield, switch subsegments no way you can calculate that. No way you can do that. So it's not intuitive. So, and in order to calculate those for each individual patient, we have to call the modeling so that we can have this modeling to calculate the dose. And this modeling could be very, I think the last talk, talking about all this modeling could be very tricky. Sometimes you messed up. Like when we do the normal streaming plan system, no matter how good we are, we make all the measurements and the calculation and the measurement just doesn't match. Okay, no matter how you do it, the measurement and the calculation does not agree with each other. All we can do, we call this the farge factor. This is like artificial correction factor. Whatever the number we have, give like a ratio 1.05. Whatever we have, we need to give this number so that our measurement agree with our calculation, which is very dangerous. So I'm saying because this just for the orders or the standard plan or standard patient, this probably is a good number. But for each individual, some patient may be too small or too big. This farge factor may not be applied. That's why you have to do the test. But the point I'm trying to say is, sometimes this modeling could be very, very hard to do. Actually for each term machine, there's at least we call the tweaking parameters. You need to know where are they. So sometimes it's just like a leakage. You need to give a little more transmission or get a little less transmission. Then you have a different way to calculate those so your measurement and your calculation will agree. You need to know how to tweak. And then we now go to the VMA, that's even harder. Actually, at least at the beginning, our physician, we discussed with our physician, our physician does not want to use the sliding window. So from the beginning, we use a step and shoot because when it's moving, during the treatment, our physician just crazy. I say it's no way. When I train the patient, I don't know what we are treating because during the treatment, since moving, I don't want to see that. Okay, so we are never using sliding window for MRT until we have this VMA. That's why at the beginning when we do the VMA, it's very slow process also. But basically we have a big jump. We found step and shoot, we jumped to VMA. But we have to spend a lot of time to demonstrate how it works anyway. Okay, so we need to do this MRT QA. That's the reason, okay? We need to verify for each individual patient now, after you commissioned your treatment plan system, you calculate this plan for this particular patient, then you need to verify this particular calculation, this particular modeling, whether it is applied for this particular, because different patient may have a different complication. You know, if some patient is complicated, some patients may be easier. Some patients are sicker, some patients are thinner. So everyone's different. I think my throat is bad. But at this point, we have some material, I can only show the homogeneous phantom. At this point, ideally, you need to do this in homogeneity correction. But hopefully when you do the three-dimensional plan system, we have verified that at least those kernel or this convolution, whatever, or this like the way you correct this in homogeneity has been verified, okay? Now, basically what we're going to show you is how we can make sure our approved plan by physician is why we're going to deliver those to the patient, okay? We just can do quick, okay? At the beginning, at the beginning, we are using this, we call the traditional protocol. This has been used for a long time. This we call the arc phantom. This is arc phantom. This is originally made because, you know, our department used to use, we call the electron arc to treat chest wall. And this is arc phantom made to verify that the electron arc works. But when we started to do this MRT, we realized this is like a circle we treat in like nine angles. And we realized, oh, this is got phantom to verify our MRT plans. So we take this phantom. Basically, this is where you put on chamber here. This is like a holder. This can have a screw over here, can move this at a different depth, different location. So it can have a six or seven location to move up and down to check the different depth. When we do the arc check, this is what may be the isocenter to check the brainstorm, okay? Anyway, that's for the electron arc. And then you can move to the surface to check what's the surface does, okay? So we use this and the iron chamber to measure the absolute dose. So we have a waterproofed iron chamber. We try to see the dose difference for at least different location, okay? And then this is at the point dose in the homogeneous phantom in water. Yes, the point. The measure. The measure. There are four things. Uh-huh. Basically, we just made, yes. That kind of point for you. Okay, for this phantom, the purpose to check one point, dose difference at one point. If you are not sure, as I say, you can move up and down to check only several points. But usually you pick a point. This point is to measure, we call the dose saturation. Usually it's high dose region. So that you know your dose, total dose delivered at the least target is what you prescribed, physician prescribed. If you want to see this different location, then you need a solid water phantom to do film measurement, okay? To see the whole picture, okay? That's why we need, you need the two, for each patient, you need to do two measurement. One measurement, point measurement using iron chamber. Another measurement using film to check the dose distribution, okay? For each patient, which is time consuming, correct? Showing you, okay, this is the way, origin, this is like a prostate patient, correct? This is the plan, okay? Then you, you, okay. I was going to show you the arc not towards the end. I'm going to show how to commission this. This just give you the result, okay? Basically this is for a different point I'm showing you, different point. See this is different iron chamber locations. You scan, you specify in the, in your phantom, okay? So you can calculate any point you want to. Usually you don't want to put your iron chamber here. This is called the dose gradient. You don't put over here. You put the chamber, iron chamber measurement somewhere over there called the dose flat region, okay? You find this patient plan, put this plan to calculate dose on this phantom. This is called the hybrid plan, okay? And then you, you calculate how much dose you have over there. Then you put this phantom. You cannot put the patient on, okay? You put this phantom on the treatment machine and to do the measurement, okay? See it? You need to choose the right location, okay? For prostate, it is right in the middle. But for the head and neck, you have to move. Usually we have this happy matching. You have to move at least two centimeters away from the center. That's why that phantom is useful anyway. But then after making this measurement, you do phantom, field measurement. Okay, I think it's time to talk about. Okay, this is what we're going to do. I don't even have time. But the basic list is you need to do trans, dose transfer factor, meaning you need the commissioning system so that you know for this setup, actually we do 270 and 90 degree to make sure that the setup is precise and that you can average our list number. Then you measure, give different beans from individual angle and then you deliver the treatment to the phantom and then you give the measurement and to calculate the dose, okay? I have to bypass this. Basically you need to make sure what's the number, the ratings you have on this particular point on that date corresponding to what centigrade, okay? For the end chamber, we know it's really like insensitive to the any spectrum, insensitive to the some other factors. Basically it is a good chamber, it's that golden standard. Only thing we need to do is the temperature pressure. Instead of we do temperature pressure, we ask our physics assistant to do this setup check, 90 and 270 and then we just take this dose transfer factor and say they don't have to take temperature pressure. That's introduce another error. This is kind of a double check also for the setup. By the way, okay, for all amnesia, physics is not doing the MRT measurement, QA measurement. There's like a group called the physics assistant, they do this all measurement. Why? Because this has to be done after treatment finished. Usually that's a late evening, all the physicists are lazy and go home. They just come back next morning to verify everything's fine, okay? But point is you need to set up, you need to establish this program so that whoever is making this measurement for you do it right, okay? So next day you just sign off, okay? Anyway, and the result, if the measurement and the calculation for this particular patient is fine, but first remember when we set up this QC, QA, we need to set up this tolerance level or action level. Remember? So what is the number of, that's the problem. I think at the beginning we can have a very good setup for all the prostate, this is 3.5 for us. Why? Because this picture, okay? We did a lot of tests. This is traditional, this augmented measurement. Like I said, we have a different services. Different services, we try different way of a plan. Different services, we have to tweak the number so that our baseline is somewhere meet our criteria, see what I'm saying? Why we use 3.5 in this? Pediatric services always cause the problem. Head and neck is fine. Nymphoma is fine. For different services have a different type of a plan. So you need to know where is that. This is like one sigma, okay? So most likely if you set this window like 2% or 3%, you're going to have this kind of force alarm to trigger you so that you cannot send off your QA result. You don't want this window too big. That sometimes is not good. But we, pretty soon we realize this 3.5% is not enough. We always have this trigger alarm. We have to go back to do what to do. So for the most services right now, our absolute measurement, 5% is our window. But for head and neck, 3% is the window. Because our head and neck doctor, they won't sign off if the absolute is more than 3%. If that's the case, we have to go back, repeat the measurement, find a different point, see whether we can get it within 3%. If within 3%, we sign off. What happened to that 5% difference? We just say, you can all. We just need to do everything we can so that patient can pass. But that's a joke. Anyway, okay, but then you see, we spent a lot of time to do the iron chamber measurement and then we have to tear up the setup, put out another, I didn't show another film at that time, bigger piece, okay? You need to do another film measurement, two step. So it's no good, okay? So quickly we change to this IBA IMRT phantom. Sometimes because our PA said, I am really tired to repeat the measurement, okay? So we do this one phantom to do two measurement at one time, okay? So you can put an iron chamber in there, you can put a film over there, film there, iron chamber over there, okay? So use the iron chamber to do a single point. The best thing is you can only do one point. You cannot move to anywhere. If this point no good, you're in trouble, okay? You have to repeat next day, okay? For the film, we do relative those measurements, those ratio usually is in a corona plan. And then we do gamma analysis. Anyway, have to be real quick. And then we do this traditional QA, we realize we have to use 5%, okay? And so this is 5% over there to do gamma analysis, so that we can pass actual asking for 90%. And this is the film measurement, film measurement. All this film measurement, film, I don't know if somebody talk about it, it's really tricky, I don't really have time. Because this really depends on film processor, that the film itself, by itself, it's just enriching, you know, it's a server, whatever, it's not reproducible. By itself, you have that. And also another thing is once you expose this film, and you have to develop, once you develop, you have to have another film radar to get at this dose and to the analysis. When you see this analysis, but sometimes the setup's wrong or something's wrong, and then the period of this film measure is not good. But which is very late. Or by the time it's doing the analysis the next day, and then you have to come back the next day. See what I'm saying? By the time you get the data, and by the time you analysis the data, usually it takes a long time for the film, that's not good. And for the film, in order to do the analysis, you need to do all this calibration, okay? It's good since you have a spatial resolution, but it is, the film is like a spatial dependent, and also any dependent, you need to do this calibration. For each day, this process could be too warm, could be too cold, the dose is different. You need to change this density to dose correction. This is very, very tedious process. Then you renormalize to get the number right. So it's tough. And then by the time then we do this MRT, we deliver it even harder, you know? We have this kind of sliding window, or we do the VMAT, that the radiation being on during the treatment, then the step shield you can trust each day, but when you do the sliding window, you do need the kind of more information, correct? You need to make sure you get all the information, because all the country angle, dose rate, MLC, they are moving all at the same time. And which one plan may not necessarily the same plan. It's one point may not be enough. So usually you need the more measurement, which is really make people kind of nervous. We have to show our physician, yes, not just one point is enough. We have all these volunteers acceptable, okay? Anyway, so all these end up, we have to have figure out some new detector, new measurement device to figure out VMAT for our physician. Then we just, okay, show you a different way to do this. We have a lot of things, okay? So you need just, you cannot just do one D now. For all the VMAT, all the sliding window, you're supposed to use at least 2D dosimetry. Like this, we have an MRT, matrix X, you have a map check, you have used the iPad. I think the last talk we also discussed a little bit. And then you can use 3D dosimetry, 3D dosimetry, okay? Even some people using the real 3D like a jail fountain, and then use like an MRI or some way to scan, or you have a different way, okay? But this is not quite mature yet, okay? These are commercially available. These two fountain, okay? And we decide to try to use this ARC check, okay? I'm going to use these to show you. But I'm kind of a little bit. This is the 2D fountain, okay? This is for our proton now, okay? This kind of side check real quick. For our proton, we use in this matrix, okay? So we use this either without or with extension, and then in order to make a different depth, you can make a decent fountain. This is really the way we're doing the QA. This is for the proton. It's very, I'm trying to say it's very primitive, but as long as it works, that's the way it is, okay? Even the proton machine is very fancy, or QA has a little bit of problem, okay? Still trying to improve. Huh? Huh? Pardon me? Like a... Compos, compost, IBS program? Yeah, IBS, yeah, yeah. But anyway, in order to make the measurement real quick, we have to repeat the different depths and just add a different build up. Anyway, this is the proton. But then you can use a map check for the QA. We're not using this. I just show you it is possible. In order to make a different depth, you can use their fountain called a map fan, okay? And the list is the detector depth. So many detect this like a matrix. It's like a matrix that you can do, okay? And also you can use the epic. I think he talked about the epic last talk, is that right? So maybe I should pass this. So epic is really good because it's attached to the machine, but it's really bad because it is so hard to calibrate, okay? You need to be very careful. I just give you really quick information about ARC check. When we first switch to the VMAG, the physician asked for more measurement to verify and to a point where to find the different QA fountain to show, and this is what we're going to show them. And this is the specification. Okay, this is the ARC check, we call it ARC check fountain. And you have a different diameter. Okay, the advantage to use the ARC check is basically you can do 3D dose measurement. The bin is always normal to the detector services and it's allow for ion chamber measurement. At this point, we only do the ion chamber measurement when the dial measurement failed. And the next day, we do dial on the chamber at the same time, okay? And then you can do the real-time measurement just real quick if it's not like the film, you have to go process it, do another, you know, film writer, do another film analysis program, okay? And it is usually easy to set up. You can make two measurements at one time and then you can see the composite or each control point that you can separate them. So it gives you more information. Not necessarily good, but for some people, we have to show them this is the way it is, okay? And then you need a commissioning this system. Basically this is the way of commissioning. You need a scan this fountain. Same thing for the ARC fountain, I just show you. You need a scan this fountain, CT scan is for fountain. And then this, which means this fountain is a patient. You're going to treat this patient, okay? Use the plan. So this will be, then after you scan, you verify your scanner, verify the fountain according to whatever the manufacturer specification, okay? You do these measurements. See all this measurement to compare what is the manufacturer saying, what is we have from our CT. And then check whether this is uniform, check this Hansper unit uniformity. See what I'm saying? Because when you calculate in the system, 10 centimeter is the 10 centimeter. You'll have a different attenuation. If your image, your image you come out, instead of 10 centimeter give you 11 or nine centimeter, of course you are not going to measure correctly. See what I'm saying? So you have to make sure this is according to the spec. And then you check for each individual system, you check the house, the dosing linearity, meaning assume your machine is, MU is linear, okay? So make sure give different dose, the response is linear because you may have some patient give 400 a day, some patient give maybe 200 a day, okay? And for this particular detector, maybe see the entrance dose, maybe see the exit dose, the dose may be different. Also this should be dose rate independent, okay? This is different dose rate, like a different MU per minute, see what's the dose rate. We do see this when the dose rate goes very low, you have like a 0.4% away from this expect number, okay? This is where you expect. But for different dose rate, you have a different response. That's the best things for the dial, for the film, for the film now. For the dial, this is the problem, this is the detector issue, okay? And chamber usually independent of dose rate, okay? And then you check all this patient, we compare this arc phantom, check to see this, because we are switching to this new phantom. So when they compare with our old phantom and the new phantom CC, whether they are giving us the same result aspect. So arc check and the previous phantom, we use all the previous measurement, previous plan and the compare, okay? With 25 case and the five-vehicle case, and we don't see really big difference. See, this is where we see, okay? For the gamma analysis, this is where the gamma analysis, or for the in the middle, the ion chamber measurement. But as you can see, the ion chamber measurement jump up and down. So we don't really rely on ion chamber measurement that much, because all these dials supposedly to measure absolute dose for you already, okay? So as long as this peripheral dose is good enough, we just take it, okay? But if it's not, then we do. Then we do this data, we call the error test analysis to have a different dose rate, different location. If we intentionally set up error, see whether this system can pick up this error for us to see how sensitive, okay? And as you can see, it's not quite sensitive, but still good for us, okay? We evaluate, we intentionally change this parameter to see, in principle, we supposed to see this fail, this, but see whether the system tells us this is fail because the setup is wrong, okay? You need to do this kind of test real quick. And then you can see, you can do the analysis, see this called the real time, that when we do that, this look like it's a plus state plan. This is the plus state arc we have from here to here, okay? Usually we have two arcs, we have two arcs. What it's trying to show you, say when this is moving, I don't know if you can see it. When this is moving, see this moving here, this moving, you can see the real time. What do you have, okay? If you see individual cursor here, you have some outliers, see this? The red one, meaning you fail, how many you fail? The green one, how many dials you pass, see it? You can see each individual angle, and they even see the real time, what happened? For this particular arc, it seems here you have a lot of failures, see what I'm saying? Can you see it? This is like a, this going arc like a clockwise, this is like a counterclockwise, from here to all the way to here. So for each individual location, you may see more problem, but the problem is when you add up together, they average out, okay? That's the point. See, then you put all together, and then this gives you, and as you can do like this, the line to see the profile, then they all agree with the other very well, okay? If you just show one, maybe, get your angle at zero, we have a lot of failure. Did you see that? But you see the overall, it's fine. So basically we decide, decide, okay? Take a long time to decide. We are using 3% and the three millimeter gamma analysis for the most cases, okay? So when we report our PA, Physics Assistant, need to show this 3% and 3 millimeter. Where it's more than 90% meaning pass. No question asked, okay? What happened if fail? What happened if fail? Sometimes we see the large fail size because diode is like sensitive to the scatter radiation. We only use 10 by 10 to do the calibration. There's a lot of this issue. Anyway, when it's failed, we use 3%, I mean 3 millimeter and the 5%. If 3 millimeter and 5% pass, usually we pass, okay? Unless physician, particularly the head and neck doctor, say I want to verification, okay? Then we have to do next day. For most other services, 3%, I mean 3 millimeter, 5%, they will just go ahead past, okay? And another thing is sometimes, because this is compared to measurement with the calculations, this measurement, just like the film, you need the conversion, you need the calibration curve so that you can convert the number of your readings to the real dose, relative dose, okay? Even though the manufacturer recommend, probably only need to do this every two or three years. But we do know, I think I have another thing. Okay, maybe this is the last one, okay? Basically, right now we have a two R checks commissioning, we have a two Fenton, we have a two different building when we walk over there, it's like 10, 15 minutes. So there's one PA doing measurement over here, another measurement doing over there, okay? Then, but you need to do this comprehensive analysis, just like I show you, two commissioners Fenton. And then you need to develop the QA program to monitor how it's performed. That's what I'm trying to say. Because the dial is drifting, okay? Even though the manufacturer say you only need to do calibration curve maybe once every year or once every two years, I don't know the number, but pretty soon we realize. At the beginning, everything calibrate, we get a patient plan, pass, pass, pass, pass. And then suddenly after maybe three months or four months, gradually this number getting worse and worse, and suddenly most of patients getting worse and worse. And what happened is because the dial drifting. So we realize we need to do this calibration more often than they recommend the period, okay? So all we can monitor, now we ask them monitor, see what's the trend. So we decide, okay, now it's time for you to redo the calibration to do this calibration curve, okay? And this happened to us. That's the one time always fail. So we have to go back to repeat again, fail, repeat again because we realize actually it's not a measurement fail. It's the calibration curve, calibration files. No longer valid because the dial response, the sensitivity changed it, okay? And then the issue is also most of the time physician really interest in the high dose volume, but our check, Fenton, only give you prefer, prefer dose and which means usually it's like a low dose region. And sometimes we are saying, oh, for this particular patient I want an additional measurement. They don't satisfy with this prefer dose which is making more measurement, okay? Another problem is the dial response, the sensitivity is really related to this feel of small to big. We realize for the small feel it seems we are fine. But every time we have a large feel like for the pelvis they have this large, you know, feel or the sarcoma case with a large feel, we always fail because there's a lot of scatter, a lot of exit dose cause this problem, okay? And then we don't know how to solve this. It seems like we need two different calibration, two different calibration curve or calibration files, one for the small feel, one for the large feel, okay? Anyway, so this is what, and when we do need the something in between, we put the insert, you can get a different insert there. And the two, we are making this different insert or we have a machine shop so that we can put the un-chamber in different location wherever different, we made ourself, okay? So what I'm, what I just show is we, we do need a QA tool for patient before the treatment so that we know whatever we see on the paper, the treatment plan is what we're going to deliver to patient. However, all the check we have right now is doing, basically it's a homogeneous phantom. That's another thing for the heart check because this dial is not homogeneous, okay? So it's really, that's a different issue. And eventually, gradually we realize we can really not check precise within 2% or 2 millimeter. That's what we are asking for. The, you know, at the beginning, we talk about the religion therapy. See, we want the goals to get within 2%, 2 millimeter, even though eventually the biological dose, we want within 5%, but from a physics point of view, we want within 2%, it's all right? And the 2 millimeter. But whereas we can never get that, okay? So what we are really checking for the MRT QA is probably the patient's safety. The patient's safety is more important. Meaning, as long as it's a reasonable number, we're probably going to take it, okay? If we really were worried, then this pre-check treatment really cannot verify because patient is not homogeneous, okay? If a certain patient, like we treat a total scalp, we treat a total scalp very close to the surface and a lot of bone, and then the physician will prescribe, say, I need the TOD measurement, like in vivo dose. So often times when you have the new technique using the MRT or VMAID, the physician, you need to do this, we call it in vivo dose to verify your plan, yes? Okay, I'm done. This is the last slide, actually. I'm done. So this is the last slide. Sorry. Thank you, sir. Thank you.