3072 - AToS: On-line Adaptive Proton SBRT by Adapt-To-Setup

Purpose/Objective(s): Proton SBRT dose distributions can be very sensitive to setup and anatomical variations. Specifically, back tissue correspondence in posterior beam spine SBRT and changes in lung density or pleural effusion in lung SBRT are scenarios in which plan adaptation would be beneficial. To this end, we propose a fast adaptive algorithm called AToS (Adapt To Setup), which is intended to compensate for anatomical changes upstream of the target.

Materials/Methods: The energies of individual proton spots were adapted to account for changes in the proton’s range due to anatomical differences. Energy of each spot in the adapted plan was tuned to stop in the same position as the original plan. A dosimetric comparison between the original plan, a verification plan on an QA CT (CT-on-Rail used for image guidance), and an adapted plan on the same QA CT was completed to assess the performance of AToS.

Results: AToS adaptation time is ~10 seconds per field in Matlab. The resulting adapted DICOM plan files were imported to Eclipse for dose calculation. Results on two spine SBRT (40/35Gy SIB in 5fx) and three lung SBRT (50Gy in 5fx) cases were summarized in table 1. Columns of on-line verification reported the dose on QA CT that patient would receive during treatment. Verification results showed that tissue thickness change at the order of 3~5mm can pull anterior target dose back to cord or spill posterior target dose into cord hence increase cord dose by 3.4~7.6Gy in spine SBRT. In addition, setup variation, lung density and pleural effusion change can pull the coverage back significantly by 20~40% in lung SBRT. Results in the column of on-line adaptive manifested that AToS adaptive can account for anatomy variation and restore the original coverage and sparing.

Conclusion: Dosimetric studies demonstrated the effectiveness of a novel and efficient on-line adaptation strategy for proton SBRT, allowing for reduced patient setup times, higher quality treatments, and a reduction of offline adaptive resources.