Stability and Reproducibility of Extended Region of Interest for Surface Guided Proton Therapy

The objective of this presentation is to understand the challenges of implementing SGRT in proton therapy, to explore the impact of ROI design on SGRT in proton therapy, and to evaluate the effectiveness of extended ROIs in maintaining stable and reproducible service tracking.

A little bit about Inova. We’re a health system right outside of DC in Northern Virginia. We have several hospital sites spread out across the region. Our Department of Radiation Oncology is part of the Schar Cancer Institute, and we’re located in five different locations. The Fairfax location is our main medical center, and it has some of the special machines like Proton and CyberKnife. But across the system, we have about 11 to 12 LINACs, and each LINAC is equipped with AlignRT system, including C-arm and ring-based.

So yeah, talk about the Inova Schar Cancer Institute and the proton therapy in Inova. So yeah, Inova Cancer Center proton therapy is located in the main campus, the Schar Cancer Center. Okay. So there we have a IBA Proteus Plus system, which contains two rotation gantry treatment rooms. So the two rooms are identical so that patients can be transferred freely from one room to the other. So the system has a pencil beam scanning with small spot size, and then we have a six-degree robotic arm couch. Both rooms have dual kV and a CBCT OBI system. And then the treatment field is 30 by 40, it’s a large field. For the breath-hold, we use SDX system. We have two AlignRT systems in both of our rooms for setup and the monitoring of the patients.

Okay. So the Inova Proton Therapy Center started in March 2020. Over the past several years, we have treated now about 1,500 patients. So among the patient size, 30% of them are head and neck patients, and then we have about one-fourth of prostate and pelvis patients, which is divided half by half, prostate and pelvis. Then we have about 15% of chest wall and breast patients, about 10% of brain patients, and we do have a small amount, 2% CSI patients. Then we also have the thoracic patients and the lung patients, about 6%. And then abdomen patients and liver patients. Yeah. Basically all sites.

We’ve been active SGRT users for several years now, both on the photon and proton sites. SGRT has been fully integrated into our clinical workflow and plays a key role in our day-to-day patient setup. Our proton workflow is pretty similar to the photon one. We’ll first do an initial setup using SGRT once the patient gets on the treatment couch. For every proton patient, they always do a cone beam CT afterwards. And based on the cone beam CT, they’ll do fine adjustments before verification. And once that is done, we’ll capture a new surface reference on AlignRT, and we’ll continuously monitor the patient during treatment. SGRT has allowed us to improve our efficiency in patient setup. Our RTTs are very reliant on it. That’s the first thing that they go to once the patients are in the room on the couch. It also allows us to eliminate skin marks and tattoos. Several of our sites have gone tattoo-less and marker-less now. So eventually, we’re going to try to do that across the entire system. SGRT also streamlines our clinical workflow overall.

Proton therapy has a unique challenge that’s very different from photon. It has gantry-induced occlusion of the SGRT camera’s field of view. So for example, for left-sided breast patient, based on the treatment plan, the gantry and the snout might block the SGRT cameras completely. So if you look at this photo examples, so this left-sided patient, if you look at the AlignRT, it’s not getting enough information due to the gantry blockage. So all we can get data from is just inferior to the breast tissues, and that’s usually what the ROI we draw for each patient. That’s based on where the treatment site is. And proton, again, based on how the plan is done, it could block a significant amount of the signals from the camera. So when that happens, it reduces the surface tracking continuity, also the accuracy in treatment monitoring. And because of that, that might cause interruption to treatment delivery, and we might have to go back to cone beam CT and increases our reliance on image guidance. And might also prolong our treatment duration overall.

So in that case, we decided to investigate using an extended region of interest for some of our proton patients. So in this study, we expanded the ROIs to include surfaces adjacent to the treatment site. So for a left-sided breast patient, we might extend it over to the right side as well. And if you look at those photo examples, without the gantry in the way, the cameras are able to see the entire ROI. And then during treatment, even though the gantry and the snout are blocking some of the cameras, the cameras can still see the other side of the ROI. So that will give us enough information to continue the treatment and ensure that patient’s in the same position, is not moving.

So in this study, we divided our data into four patient groups. We looked at thorax, pelvis, prostate, and breast patients. So we used each of their log files. After each fraction, we extracted those data from those log files and analyzed them for each patient.

Okay. So the log file from VMRT actually can give us a full recording throughout the treatment. Here we show the deviation of the magnitude throughout recording. This example. From here, basically we can clearly see the patient that went through the whole process during the treatment, starting with the initial setup with AlignRT. Then we proceeded with the CBCT to adjust the patient position. After that, we do a rebaseline, and then here we can monitor the treatment throughout the course. And if we take a closer look at the recordings during the beam on, you can see. This is an example of 200 seconds during the treatments. You can see that the oscillations of the deviations, and then basically here you see two trends. One is that you see a little bit of drift throughout the course, but it’s still within two millimeters, basically. And another is that the curves oscillated. This reflected the breathing of the patient.

So we can do analysis on these recordings and then for different sites and for different patients. Here we show the results. For each site, starting with the breast and chest wall patients. And here we show three patients’ data just to see the variations among different patients, and also that taken in different days, from day one to day five in consecutive days. Each data point here is the mean value of the deviation with the standard deviation within that fraction. And also across five different fractions, we can get an average of the mean deviation and also standard error of the mean for that mean deviation. We can see from among different patients that the mean deviation is about one or two millimeters for the five days. So this is for the breast patients. So with this, we can see the intra-beam reproducibility and also inter-fraction reproducibility.

So this shows the result for the thorax patients, also three patients. We can get the mean and standard deviation for each fraction, and also that across the five consecutive days. And we can get the inter-fraction reproducibility with the mean value and the standard error of mean. And here we can actually compare this with the breast patients. We can see a little bit increase of the inter-fraction, the mean deviation. Seems that the thorax patients have a little bit larger deviation compared to the breast patients. Also, here we show the pelvic patients across the five days, and then each day of the mean and standard deviation for three patients. And we get similarly, the mean value and the standard error of the means among the five days. And also the same thing for the prostate patients.

From our study, we found that SGRT provides precise patient positioning and real-time treatment monitoring without compromising workflow efficiency. And with proton therapy, gantry-induced occlusion of the SGRT camera is a major challenge. However, extended ROIs might help with establishing stability and reproducibility intra and inter-fractionally. In the future, we will explore the application of SGRT for beam gating in proton therapy, for example, for DIBH breast patients or any thorax patient, and evaluate its performance compared to other breath-hold methods like ABC. We currently use SDX for all the breath-hold patients.

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