Dosimetric comparison of RapidPlan and manually optimized plans in volumetric modulated arc therapy for prostate cancer

• Published in: Physica medica Vol. 44; pp. 199 - 204
• Main Authors: Kubo, Kazuki, Monzen, Hajime, Ishii, Kentaro, Tamura, Mikoto, Kawamorita, Ryu, Sumida, Iori, Mizuno, Hirokazu, Nishimura, Yasumasa
• Format: Journal Article
• Language: English
• Published: Italy Elsevier Ltd 01-12-2017
• Subjects: Knowledge-based planning; Prostate cancer; RapidPlan; VMAT; Knowledge-based planning; VMAT; Prostate cancer; RapidPlan
• ISSN: 1120-1797; 1724-191X
• DOI: 10.1016/j.ejmp.2017.06.026

•We used 51 previously administered VMAT plans to train a RapidPlan model.•Thirty RapidPlan based plans compared with 30 clinical manual optimization plans.•RapidPlan based plans were created by a single optimization without intervention.•The dose volume parameter values for the PTV were significantly similar.•Rectal and bladder doses in RapidPlan based plans were comparable to clinical plans.
This study evaluated whether RapidPlan based plans (RP plans) created by a single optimization, are usable in volumetric modulated arc therapy (VMAT) for patients with prostate cancer.
We used 51 previously administered VMAT plans to train a RP model. Thirty RP plans were created by a single optimization without planner intervention during optimization. Differences between RP plans and clinical manual optimization (CMO) plans created by an experienced planner for the same patients were analyzed (Wilcoxon tests) in terms of homogeneity index (HI), conformation number (CN), D95%, and D2% to planning target volume (PTV), mean dose, V50Gy, V70Gy, V75Gy, and V78Gy to rectum and bladder, monitor unit (MU), and multi-leaf collimator (MLC) sequence complexity.
RP and CMO values for PTV D95%, PTV D2%, HI, and CN were significantly similar (p<0.05 for all). RP mean dose, V50Gy, and V70Gy to rectum were superior or comparable to CMO values; RP V75Gy and V78Gy were higher than in CMO plans (p<0.05). RP bladder dose-volume parameter values (except V78Gy) were lower than in CMO plans (p<0.05). MU values were RP: 730±55MU and CMO: 580±37MU (p<0.05); and MLC sequence complexity scores were RP: 0.25±0.02 and CMO: 0.35±0.03 (p<0.05).
RP plans created by a single optimization were clinically acceptable in VMAT for patient with prostate cancer. Our simple model could reduce optimization time, independently of planner’s skill and knowledge.