(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: Decreasing the run-time of optimizing Intensity Modulated Proton Therapy (IMPT) treatment plans, particularly the run-time of the in-house Proton Therapy Center IMPT Optimizer (OPT) at MD Anderson Cancer Center, in order to make the IMPT optimization process more conducive to real-time clinical use.
Methods: redesigned OPT (OPT+) improves speed at the expense of readability. Order of complexity reductions and factor optimizations were utilized. Data structures were designed to reduce minimizer time. Sparse data structures were introduced. Multithreading was fine-tuned, reducing false sharing. OPT, OPT+, and the optimizer in the Varian Eclipse TPS (Eclipse) were run on patients of various levels of computational complexity.
Results: a prostate patient took 6.6 minutes with OPT, 44 seconds with OPT+, and 1.4 minutes with Eclipse. Processing a more complex pelvis patient took 47.4 minutes with OPT, 6.7 minutes with OPT+, and 69.5 minutes with Eclipse. The OPT+ is observed to be 7 to 9 times faster than the OPT. The less computationally extensive the plan the faster the OPT+ is compared to the OPT. The core optimization algorithms are speed up by a factor of 4.6 to 9.2. The heavily used algorithm to recalculate dose is speed up by a factor of 5.1 to 29. The OPT+ is observed to be 1.9 (less complex plans) to 10.4 (more complex plans) times faster than Eclipse. There were no significant differences in the optimized results produced by OPT, OPT+, and Eclipse.
Conclusion: is able to optimize simple plans in less than a minute and complex plans in less than 10 minutes. OPT+ provides fast real-time optimization, overcoming the time bottleneck provided by systems like Eclipse. With this development we estimate that Robust LET and RBE optimization for even the most complicated cases could be done in the near future within five minutes.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
