Kanokkarn Kuekkong. Application of the RapidPlan knowledge-based treatment planning system for radiation therapy of prostate cancer patients. Master's Degree(Medical Physics). Mahidol University. Mahidol University Library and Knowledge Center. : Mahidol University, 2019.
Application of the RapidPlan knowledge-based treatment planning system for radiation therapy of prostate cancer patients
Abstract:
RapidPlan knowledge-based treatment planning was developed and adopted in volumetric modulated arc radiotherapy (VMAT) planning to improve plan quality and planning efficiency. RapidPlan used plan database to build a model for predicting organ-at-risk (OAR) dose-volume-histograms (DVHs) of the new treatment plan. Therefore, the purpose of this study was to generate and evaluate the performance of the RapidPlan knowledge-based treatment planning of VMAT for definitive radiotherapy of prostate cancer. Three RapidPlan models based on a number of 20, 40, and 60 previously VMAT plans were trained and validated with 10 new prostate cancer patients. The coefficient of determination; R2, the average chi-square; χ2 and the model goodness with mean square error (MSE) provided from the RapidPlan system were used to optimal model determination. Dosimetric parameters including the D2%, D95%, D98%, homogeneity index (HI), and conformation number (CN) for planning target volume (PTV), V65Gy, V70Gy, V75Gy for bladder, V50Gy, V60Gy, V65Gy, V70Gy, V75Gy for rectum, V50Gy, Dmax for both of femurs, and Dmean for penile bulb among 3 models were collected and compared by using pair-t-test or Wilcoxon sign-rank test (p<0.05). After that, the optimal model results were selected and compared with the manually optimized (MO) plans from beginner planner and expert planner. Moreover, the planning times of both optimization methods were also collected. The results showed that in term of models comparison, PTV coverage and OARs dose parameters were not statistically different. The robustness Model60 was selected and compared with MO. Comparison of the model and MO plans showed the model had similar results of D95%, D98% for PTV but a significant higher result of D2%, and CN from RapidPlan (82.2 Gy for D2% and 0.9 for CN) when compared with MO (81.3-81.5 Gy for D2% and 0.8 for CN). For HI value, model plans were inferior to MO-beginner planners (p<0.05), while similar average HI was presented from the MO-expert planners. For OARs, all dose-volume parameters of RapidPlan were significantly lower than MO (p<0.05) only in rectum V75Gy, and both of femurs Dmax, were similar to MO by expert planner. The average planning time was reduced from 33-105 minutes by manual plans to 10 minutes when the model was used. In conclusion, RapidPlan knowledge-based treatment planning in this investigation presented an acceptable VMAT plan quality for definitive radiotherapy for prostate cancer in only single optimization.