(Sunday, 7/29/2018) 3:00 PM - 6:00 PM
Room: Exhibit Hall
Purpose: Optimization of treatment plans using automated techniques to achieve a consistently optimal plan with efficient and accurate delivery for a given site still remains an open question. Despite recent advances in automated planning techniques, significant time is still spent in iterative plan optimization in order to achieve set Treatment Planning Objectives (TPO). Evaluation for further improvements in plan quality (PQ) is often speculative. For the purpose of this study, optimal plan is defined as the treatment plan that maximizes organ at risk (OAR) sparing without increasing plan complexity or sacrificing Target Coverage (TC). The objective of this study is to extract features from a small database of prostate plans to predict optimal PQ.
Methods: Seven prostate plans optimized using PACE protocol were used in this study with PTV prescribed to 62 Gy in 20 fractions. TC, Equivalent Uniform Dose (EUD) of OARs and percentage of volume of OAR overlapping with PTV were extracted for all plans. Deliverability of plans was verified using ArcCheck with a local Gamma of 3%,3mm > 95% .
Results: This study showed there exists a linear correlation between percentages of OAR overlapped with PTV as compared to optimal OAR EUD. All plans show similar complexity as analyzed by delivery QA and in-house delivery emulator. Study also indicated a certain threshold of OAR overlap, above which TC has to be compromised to achieve the TPO.
Conclusion: Intelligent prediction for optimal plan based on bladder and rectum target overlap with PTV while minimizing plan complexity can be achieved. Baseline prediction derived from this study has been used as a starting objective value for plan optimization which has greatly reduced the time spent for iterative optimization. This can also be used to assess if further improvements without compromising plan complexity can be achieved.
Keywords
Prostate Therapy, Optimization, Quality Control
Taxonomy
TH- External beam- photons: VMAT dose optimization algorithms
Contact Email
