Sonja Wegener, Robert Schindhelm, Otto A. Sauer

• Purpose: Any Linac will show geometric imprecisions, including non-ideal alignment of the gantry, collimator and couch axes, and gantry sag or wobble. Their angular dependence can be quantified and resulting changes of the dose distribution predicted (Wack, JACMP 20(5), 2020). We analyzed whether it is feasible to correct geometric shifts during treatment planning. The successful implementation of such a correction procedure was verified by measurements of different stereotactic treatment plans. Methods: Isocentric shifts were quantified forPurpose: Any Linac will show geometric imprecisions, including non-ideal alignment of the gantry, collimator and couch axes, and gantry sag or wobble. Their angular dependence can be quantified and resulting changes of the dose distribution predicted (Wack, JACMP 20(5), 2020). We analyzed whether it is feasible to correct geometric shifts during treatment planning. The successful implementation of such a correction procedure was verified by measurements of different stereotactic treatment plans. Methods: Isocentric shifts were quantified for two Elekta Synergy Agility Linacs using the QualiForMed ISO-CBCT+ module, yielding the shift between kV and MV isocenters, the gantry flex and wobble as well as the positions of couch and collimator rotation axes. Next, the position of each field's isocenter in the Pinnacle treatment planning system was adjusted accordingly using a script. Fifteen stereotactic treatment plans of cerebral metastases (0.34 to 26.53 cm3) comprising 9–11 beams were investigated; 54 gantry and couch combinations in total. Unmodified plans and corrected plans were measured using the Sun Nuclear SRS-MapCHECK with the Stereophan phantom and evaluated using gamma analysis. Results: Geometric imprecisions, such as shifts of up to 0.8 mm between kV and MV isocenter, a couch rotation axis 0.9 mm off the kV isocente,r and gantry flex with an amplitude of 1.1 mm, were found. For eight, mostly small PTVs D98 values declined more than 5% by simulating these shifts. The average gamma (2%/2 mm, absolute, global, 20% threshold) was reduced from 0.53 to 0.31 (0.32 to 0.30) for Linac 1 (Linac 2) when including the isocentric corrections. Thus, Linac 1 reached the accuracy level of Linac 2 after correction. Conclusion: Correcting for Linac geometric deviations during the planning process is feasible and was dosimetrically validated. The dosimetric impact of the geometric imperfections can vary between Linacs and should be assessed and corrected where necessary.…