Premium
SU‐E‐J‐75: Automated Two Loop Optimization of a Treatment Planning System
Author(s) -
Wang H,
Xing L
Publication year - 2015
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4924162
Subject(s) - radiation treatment planning , computer science , planner , task (project management) , process (computing) , eclipse , plan (archaeology) , medical physics , artificial intelligence , radiation therapy , programming language , medicine , engineering , radiology , systems engineering , history , physics , archaeology , astronomy
Purpose: To establish a strategy of a two loop optimization with recorded interactions between a planner and a commercial treatment planning system (TPS) and to apply it to facilitate VMAT/IMRT planning with incorporation of prior knowledge‐guidance. Methods: We first record some commonly used planner‐ TPS interactions as subroutines using Microsoft Visual Studio Coded UI. A recorded action is called back in C# application programming when the corresponding task needs to be accomplished. We implement a prior knowledge‐guided VMAT/IMRT plan selection algorithm in a two loop optimization framework with an Eclipse TPS (Varian Medical Systems, Palo Alto, CA). In this implementation, the DVHs of a prior treatment plan are used to guide the search for a clinically optimal VMAT/IMRT treatment plan by iteratively evaluating and modifying the optimization parameters of a series of Eclipse plans. The approach is applied to the treatment planning of three clinical cases, a prostate case and two head and neck cases. Results: An automated two loop optimization is developed. The method is applied to guide VMAT/IMRT planning process and our results show that it is capable of finding clinically sensible treatment plans with little planner interactions. The process mimics a planner's planning process and provides a solution that would otherwise require a huge amount of trial‐and‐error interactions of a planner. The results obtained using the approach are found to be either clinically acceptable or close to be acceptable. Conclusion: The proposed technique provides a valuable way to harness a commercial TPS by application programming via the use of recorded human‐computer interactions. A prior knowledge guided plan selection is developed in the platform, which greatly facilitates the search of clinically acceptable plans. The development brings us a big step closer toward the goal of automated treatment planning in a clinical environment.