Program Information
Cloud-Based Radiation Treatment Planning: Performance Evaluation of Dose Calculation and Plan Optimization
Y Na1*, D Kapp1 , Y Kim1 , T Suh2 , L Xing1 , (1) Stanford University School of Medicine, Stanford, CA, (2) Catholic Univ Medical College, Seoul, Seoul
Presentations
SU-D-BRD-1 Sunday 2:05PM - 3:00PM Room: Ballroom DPurpose: To report the first experience on the development of a cloud-based treatment planning system and investigate the performance improvement of dose calculation and treatment plan optimization of the cloud computing platform.
Methods: A cloud computing-based radiation treatment planning system (cc-TPS) was developed for clinical treatment planning. Three de-identified clinical head and neck, lung, and prostate cases were used to evaluate the cloud computing platform. The de-identified clinical data were encrypted with 256-bit Advanced Encryption Standard (AES) algorithm. VMAT and IMRT plans were generated for the three de-identified clinical cases to determine the quality of the treatment plans and computational efficiency. All plans generated from the cc-TPS were compared to those obtained with the PC-based TPS (pc-TPS). The performance evaluation of the cc-TPS was quantified as the speedup factors for Monte Carlo (MC) dose calculations and large-scale plan optimizations, as well as the performance ratios (PRs) of the amount of performance improvement compared to the pc-TPS.
Results: Speedup factors were improved up to 14.0-fold dependent on the clinical cases and plan types. The computation times for VMAT and IMRT plans with the cc-TPS were reduced by 91.1% and 89.4%, respectively, on average of the clinical cases compared to those with pc-TPS. The PRs were mostly better for VMAT plans (1.0 ≤ PRs ≤ 10.6 for the head and neck case, 1.2 ≤ PRs ≤ 13.3 for lung case, and 1.0 ≤ PRs ≤ 10.3 for prostate cancer cases) than for IMRT plans. The isodose curves of plans on both cc-TPS and pc-TPS were identical for each of the clinical cases.
Conclusion: A cloud-based treatment planning has been setup and our results demonstrate the computation efficiency of treatment planning with the cc-TPS can be dramatically improved while maintaining the same plan quality to that obtained with the pc-TPS.
Funding Support, Disclosures, and Conflict of Interest: This work was supported in part by the National Cancer Institute (1R01 CA133474) and by Leading Foreign Research Institute Recruitment Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (MSIP) (Grant No.2009-00420).
Contact Email: