Purpose: To study the impact on dose coverage and the dose to the healthy tissue applying optimized margins in single isocenter multiple brain metastases radiosurgery (SIMM-SRS) in linac machine based on setup rotations/translations induced errors calculated by a genetic algorithm (GA).
Method: The following quality indices of SIMM-SRS were analyzed for 32 plans (256 lesions): Paddick conformity index (PCI), gradient index (GI), maximum (Dmax) and mean (Dmean) doses, local and global V12 for the healthy brain. A GA based on Python packages were used to determine the maximum shift produced by induced errors of 0.2º/0.2 mm, and 0.5º/0.5 mm in 6 degrees of freedom.
Results: In terms of Dmax, and Dmean, the quality of the optimized-margin plans remains unchanged (p>0.072) concerning the original plan. However, considering the 0.5º/0.5 mm plans, PCI and GI decreased for ≥ 10 metastases, and local, and global V12 increased considerably in all cases. To consider 0.2º/0.2 mm plans, PCI and GI get worse but local, and global V12 improved in all cases.
Conclusion: GA facilities to find the individualized margins automatically among the number of possible permutations of the setup order. The user-dependent margins are avoided. This computational approach takes into account more SRS sources of uncertainty, enabling the protection of the healthy brain by “smartly” reducing the margins, and maintaining clinically acceptable target volumes’ coverage in most cases.