I try to comment your concern 2. We take two CTs one in free breeze (FB) and one in breath hold (BH). The FB CT is taken with a low dose protocol, as we only need to extract the body structure for FB initial setup using AlignRT. Our ROI is on the breast. Once this is done, we switch to the BH body contour. And yes it is true, depending on patient the difference between FB and BH can be small. To confirm BH position we don’t use CBCT but an AP or PA kV-image and an MV-image in the direction of one treatment field. Before the imaging we let the patient hold the breath and check with AlignRT, mainly vrt. With the AP/PA image we correct only lat and rtn (based on lung/rip interface). With the MV-image from beam direction we correct vrt and lng. Immediately before applying the shift, we let the patient again hold ist breath and check with AlignRT. If vrt is ok, we apply the shift (still under BH) and take immediatly after the shift an new reference surface (still under BH). Like this we can be pretty sure that the new reference surface was captured with the same BH Level as during CT. Before applying the treat field, we check in BH (and new reference surface) with an MV-“before” image if the thorax wall matches the DRR structure. Then we Beam On and take an MV-movie. So you can check offline the level of BH. Most of the time this works quite well.
I guess similar/adapted procedures can be applied to other regions.