Michael Tallhamer, MSC, DABR
A number of things can be happening here and understanding which is important. We have been using Vision RT’s AlignRT (same thing as OSMS) for SRS for over 5 or 6 years now and have characterized the system performance under many SRS conditions. There is a white paper and at least 1 presentation from Ben Waghorn Ph.D. DABR (a physicist who works for Vision RT) that discusses the aspects of the ACO camera calibration that Dustin referred to above. The whitepaper covers the expected improvements it can bring to a system and the results of that have been seen at my institution as well as documented by the data out of UAB. With you being an OSMS customer and not a AlignRT customer I do not know what the logistics are in getting the ACO camera calibration for your site and I would recommend reaching out to Varian to see what that might take.
However, you don’t have ACO and I have found a non ACO system can maintain a compound uncertainty with all variables under the worst conditions of around 0.7-0.8mm across all couch positions. That was with throwing every unfavorable characteristic (i.e. poor head position, obstruction, extreme couch angles, etc.) at the system and letting the errors compound at the absolute edge of the FOV for the cameras. I have listed a few items you can consider below and referenced a couple white papers and talks as well.
You should be characterizing your system for SRS by looking at the rotational stability and obstruction characteristics of your cameras and isolating various confounding variables. I have written a whitepaper (pre ACO) outlining a number of tests you can use and the expected results of those tests under a full range of isocenter placement positions in the cranial vault. It also covers the impact of cranial positioning and ROI effects that I can send to you if you are interested.
The long and short of it is obstruction characteristics of the cameras are, or should be, on the order of 0.1mm – 0.2mm (can vary with isocenter location). The rotational walkout of the deltas followed our couch walkout within 0.1mm when the radiographic WL images are corrected for gantry sag effects (very important – I can send you the slide from my AAPM 2018 talk that shows this). Head position and isocenter position can have a large impact on the deltas in a non ACO system as the couch rotates over the full range of possibilities for SRS.
Head tilt should be kept to neutral or chin down for SRS as chin up orientations result in shadowing or obstruction of the ROI by the patient’s own anatomy at couch angles approaching 90 and 270. With treatment arcs often passing over the head and obstructing the opposite camera this obstruction effect can cause serious issues with the camera’s ability to see the ROI and result in poor performing deltas.
If the patient has a poor head position the mask opening and ROI size becomes much more important and in the case of Qfix I have seen on a number of my consulting cases where the opening in the mask is far too small for the orientation of the patient’s head during delivery. This seems very common for Qfix users but is by no means limited to them. I would encourage some good mask making training to make sure the mask opening is of the proper size going into treatment planning.
The calibration of the camera is also important. The conditions of the calibration should be controlled and there is in fact a Customer Information Bulletin from back in 2017 (ID # CIB2017SN0005) that discusses this and makes recommendations related to the conditions under which the cameras should take place and gives an outline of the procedure.
I cannot say I have ever seen things like what Dustin mentioned in his points 1 and 2 without some of these other issues being at the root cause nor have we ever had to limit our couch rotation angles for SRS deliveries. I wouldn’t venture to say we have eliminated ALL delta discrepancy issues entirely but we can predict with a very high degree of success the very small fraction of patients that are going to potentially have one during delivery because 1) we are operating at the extreme limit of the system or 2) the patient’s position is very unfavorable well before we see it during treatment. We can do this because the rest of the system is fairly easy to characterize and control even without ACO.
If you need a copy of anything I have mentioned my email is firstname.lastname@example.org. I would be happy to share what I have with you or others.