Identification Of Gamma Knife Planning Strategies By Calculation Of Pareto Dominant Plans Using Genetic AlgorithmsKeywords: dosimetry, gamma knife, dose planning, radiosurgery, techniqueInteractive ManuscriptAsk Questions of this Manuscript: What is the background behind your study? Traditional computerized Gamma Knife optimization is usually implemented to maximize a chosen conformality index, often by exhaustively testing various possibilities. What is the purpose of your study?We show an approach differing from this in three ways. First, we seek novel planning strategies easily completed by a human user, rather than complete plans. Second, we optimize using the concept of Pareto dominance. Specifically, plan A is Pareto dominant to plan B if, when using plan A, more of the prescription isodose volume (PIV) is contained in the tumor and more of the tumor is contained in the PIV. A plan is a Pareto dominant plan if it is not dominated by any other plan. Two different Pareto dominant plans represent different tradeoffs between tumor dose and dose to normal tissue, because neither plan can dominate the other (by definition). We calculate a list of Pareto dominant plans, viewing them as competing strategies to be assessed by the user. Third, we use efficient genetic algorithms rather than brute force to find the Pareto dominant plans. Describe your patient group. Describe what you did.Custom software calculated dose distributions for a ‘virtual Gamma Knife’ using standard algorithms, along with the fraction f1 of the PIV contained in the tumor and the fraction f2 of the tumor contained in the PIV. The Matlab Genetic Algorithm Toolbox was used to automatically find Pareto dominant plans based on f1 and f2. Five irregular tumor shapes were tested, using trials of two 8 mm shots, three 8 mm shots, one 8 and one 14 mm shot, and two 8 and one 14 mm shots. Describe your main findings.Custom software calculated dose distributions for a ‘virtual Gamma Knife’ using standard algorithms, along with the fraction f1 of the PIV contained in the tumor and the fraction f2 of the tumor contained in the PIV. The Matlab Genetic Algorithm Toolbox was used to automatically find Pareto dominant plans based on f1 and f2. Five irregular tumor shapes were tested, using trials of two 8 mm shots, three 8 mm shots, one 8 and one 14 mm shot, and two 8 and one 14 mm shots. Describe the main limitation of this study.This is a retrospective study. Describe your main conclusion.Custom software calculated dose distributions for a ‘virtual Gamma Knife’ using standard algorithms, along with the fraction f1 of the PIV contained in the tumor and the fraction f2 of the tumor contained in the PIV. The Matlab Genetic Algorithm Toolbox was used to automatically find Pareto dominant plans based on f1 and f2. Five irregular tumor shapes were tested, using trials of two 8 mm shots, three 8 mm shots, one 8 and one 14 mm shot, and two 8 and one 14 mm shots. Describe the importance of your findings and how they can be used by others.This question was not answered by the author Traditional computerized Gamma Knife optimization is usually implemented to maximize a chosen conformality index, often by exhaustively testing various possibilities. We show an approach differing from this in three ways. First, we seek novel planning strategies easily completed by a human user, rather than complete plans. Second, we optimize using the concept of Pareto dominance. Specifically, plan A is Pareto dominant to plan B if, when using plan A, more of the prescription isodose volume (PIV) is contained in the tumor and more of the tumor is contained in the PIV. A plan is a Pareto dominant plan if it is not dominated by any other plan. Two different Pareto dominant plans represent different tradeoffs between tumor dose and dose to normal tissue, because neither plan can dominate the other (by definition). We calculate a list of Pareto dominant plans, viewing them as competing strategies to be assessed by the user. Third, we use efficient genetic algorithms rather than brute force to find the Pareto dominant plans. Custom software calculated dose distributions for a ‘virtual Gamma Knife’ using standard algorithms, along with the fraction f1 of the PIV contained in the tumor and the fraction f2 of the tumor contained in the PIV. The Matlab Genetic Algorithm Toolbox was used to automatically find Pareto dominant plans based on f1 and f2. Five irregular tumor shapes were tested, using trials of two 8 mm shots, three 8 mm shots, one 8 and one 14 mm shot, and two 8 and one 14 mm shots. Custom software calculated dose distributions for a ‘virtual Gamma Knife’ using standard algorithms, along with the fraction f1 of the PIV contained in the tumor and the fraction f2 of the tumor contained in the PIV. The Matlab Genetic Algorithm Toolbox was used to automatically find Pareto dominant plans based on f1 and f2. Five irregular tumor shapes were tested, using trials of two 8 mm shots, three 8 mm shots, one 8 and one 14 mm shot, and two 8 and one 14 mm shots. This is a retrospective study. Custom software calculated dose distributions for a ‘virtual Gamma Knife’ using standard algorithms, along with the fraction f1 of the PIV contained in the tumor and the fraction f2 of the tumor contained in the PIV. The Matlab Genetic Algorithm Toolbox was used to automatically find Pareto dominant plans based on f1 and f2. Five irregular tumor shapes were tested, using trials of two 8 mm shots, three 8 mm shots, one 8 and one 14 mm shot, and two 8 and one 14 mm shots. Project Roles:
|
