We use tilted breast board and upside-down t- shaped ROI for all of our resection (n0) breast patients. In our small clinical study of ten patients we found that the upside- down t- ROI (does not include diaphragm, includes small portion of both breasts) correlates best with the entire bony chest wall position (achieved in the online match). Based on larger study, the realized setup margin needed for the soft tissue was around 5 mm with that ROI (n=25). The ROI which covered the whole breast caused larger random errors in the LNG isocenter and ROLL in comparison to bony online match (compromise match between sternum and the ribs in the middle of the PTV in LNG,LAT,VRT), correlating with the daily position variation of the soft tissue. Nothing else statistically significant was found between the ROIs, so I can not much argue with users which draws plenty of soft tissue into ROIs. Some may create a full publication of this, take a catch:). Do you want to roll the entire patient in the setup, based on the roll which comes from the roll of the soft tissue, or how do you know is the whole patient rolled or just the soft tissue with the whole breast- ROI? In your case I could do an additional ROIs, one for the soft tissue and one for the rigid structures to see how much they differ in the setup or setup the patient with upside-down t-ROI and acquire treatment capture to see the position of the soft tissue. Can that colour view- feature be useful? In our current practice we finally verify the location of the soft tissue with the tangential images and it should be inside 1 cm and inside treatment field. In VMAT- treatments we acquire CBCTs and want the soft tissue to locate exactly at the planned position, mostly in other cases the margins covers the daily variation of the soft tissue after bony match. If not, finally re-planning may be needed. Your solution depends on the ideology to treat the breasts.