VisionRT has advised us to perform isocenter couch shift in DIBH such, that patient takes a BH to the reference surface used in the imaging, apply the couch shifts based on online match, acquire a new reference surface for the treatment and let patient breath again. In that way breath hold level should remain the same after the couch shifts. Still it is not known what happens in the chest wall/BH while the couch is moving and does the patient take the BH exactly to the same delta value as she did in the imaging. Therefore I would always verify the succeeding of the action with tangential image before treatment and to be honest on the other hand therefore I would not much care about 1-2 mm couch shifts that you mentioned in practice. In the first fractions, when we want to achieve optimal reference surfaces for the setup and eliminate the systematic errors, I would finetune the setup surfaces with that kind of accuracy as you described. Then I would continue with daily tangential images and weekly with n+, mastectomy DIBHs verify that the vertebra and sternum remain inside 3mm in the orthogonal images in vertical direction, if so, continue four next fractions with the tangential images again. If there is a need to do the couch corrections based on orthogonal images in the weekly check, I can describe some examples in the different online match situations, concentrating on vertical directions only. In LAT image, in vertical direction…:
1. both vertebra and sternum have >3mm displacement in same direction, nearly as much, in comparison to DRR = only isocenter error-> couch shift described in the beginning of the text
2. error mostly in the sternum is > 3 mm-> ask patient to take a BH delta outside of the zero value as much as the error in the sternum in the image was and acquire a new reference surface. Not necessarily need for the couch shifts, if the vertebra accuracy was good.
3. there is >3 mm displacement mostly in the vertebra, couch shift to vertebra. Not necessarily need for the new reference surface if the chest wall accuracy in the images was good.
In the point 2 and 3, you may need to take account the effect of changed LNG chest wall motion, since we changed the BHL and new orthogonal images may be needed. With the optimized workflow there is not often need for such rather repugnant online corrections 1-3, which is at least my goal with AlignRT DIBH, since using the IGRT for mostly verification only with certain action levels seems easier for the patient and the rtt.