I believe that it is easier to detect (and have a more reliable value of) the pitch if a larger part of the ROI covers the lateral of the patient i.e. the side of the chest wall. We use an ROI that more or less covers the target area.
To your concern regarding the vertebra and sternum. In kV-images we have also seen some disagreement in the longitudinal direction, both for patients treated in FB and DIBH. When using the vertebra as a primary matching structure (in kV-images) we quite frequently performed couch shifts in longitudinal direction and the corresponding MV-(verification)images showed that the LNG-shift (compared to the DRR) was unnecessary and even inaccurate.
Today we use the sternum and the distance between the chest wall and vertebras as primary structures when matching kV-images and the vertebras alone have lower priority. The MV’s correspond better and we don´t have the same longitudinal displacement systemetics.
We haven’t looked into the couch vrt vs. the surface vertical.
Kind regards Annika