Hi Charlotte
We use open and closed 5 point masks, open masks with faceROI. We are going towards 5 point open masks. We have not used vacuums or other additional supports under the neck/shoulders, which could be beneficial, but sometimes expensive.
We had 3 point open masks in use for something like 30-40 H&N patients. We used face ROI and positioned the shoulders with postural video. Our results showed finally slightly larger deviation to the clavicles in the CBCT images with 3 point open masks than with 5 point masks. This was as well more time consuming and in general not so popular way to setup the shoulders. We also used deformation tool with treatment capture to understand the displacements at the setup from whole upper body. We have also good results and feedback from brain setup and treatment with 3 point open masks.
What to do to make 3 point open masks popular again here? I think we need good results from the users, publication of good achieved posture of the patient (see the text later) and easy way to setup the shoulders, which obviously are the main topic with the short masks.
When talking about the results, I do not consider translational and rotational results as a relevant results in H&N mask comparisons. If we think about H&N CBCT images, we want to see the whole cervical spine, mandibule, base of the skull and shoulders with small residual errors after typically matching the images to the middle of C- spine. This is as well what was our goal at the AlignRT setup with 3 point open masks: face ROI to setup the head and to thereby verify the mandible/base of the skull location, video to setup the shoulders. May be some additional ROI to cranial part of chest could be useful if we continue this in the future to approximately verify the lower part of C-spine and chest. Some use this together with faceROI and some separately (first setup with the faceROI and then switch to caudalROI or combinationROI. With combination of FaceROI+ cranial chestROI it could be useful to use live deformation tool, this is something that we have not tested, but is interesting.
However, I know that pretty much all are interested to position C- spine, head and shoulders as well as possible with the AlignRT. Why do not we evaluate and publish the succeeding of this based on the CBCTs? We need much more CBCT data on how this accuracy has really happened. Results of translational/isocenter and rotation accuracy is not telling the whole truth. I think in the study 10.1016/j.ijrobp.2008.11.035 t Setup Uncertainties of Anatomical Sub-Regions in Head-and-Neck Cancer Patients After Offline CBCT Guidance - ScienceDirect this H&N posture in the CBCT images is evaluated well.
It is easy to do that kind of H&N CBCT study in the Offline Review. First we match mandible for 30 times (if 30 fr.), we get the residual error in 6D to the mandible (systematic and random for the patient). This is repeated for 25 patients in 3 point open mask and 5 point masks groups, then we get the systematic and random errors to the mandible in the groups for group comparison. We can select which bony anatomy landmarks around the H&N PTV for comparison, but mandible, base of the skull, c1, c7, sternum, clavicles are good starting points for the evaluation. If in your H&N matching guidelines you perform your automatic match to the c3-4, then the residual errors to the named landmarks are compared to the online matched c3-4 match. This can as well be done in Offline Review if we think that online match location varies too much, to automatch c3-4 and after that compare and calculate the residual errors of individual bony structures to that landmark in Excel or some other program. My main point is that we need much more detailed data of SGRT setup accuracy with different kind of masks than only the results of online shifts made in VRT, LNG, LAT, ROT, ROLL, PITCH. H&N bony structures are not definitely acting that simply, on the contrary several important bony structures can be shifted in the background of such results with showing remarkable under/overdose at the edges of H&N PTV without isocenter results showing it.
Marko