Room: AAPM ePoster Library
Purpose: To improve the inverse optimization speed of intensity modulated proton therapy (IMPT) treatment planning.
Methods: A new spot weight optimization (SWO) algorithm based on fast inverse dose optimization (FIDO) was developed in MATLAB using the open-source toolkit matRad. Like the original FIDO algorithm, this SWO-FIDO algorithm introduces modifications to the objective function used in SWO, as well as modifications to the SWO workflow. These modifications help reduce the computational cost of calculating the objective function value and gradient vector during optimization, as well as reduce the number of iterations performed during optimization.
A retrospective planning study was conducted on 15 datasets to evaluate the performance and plan quality obtained with this FIDO-inspired SWO algorithm. The selected 15 cases represent a variety of treatment planning sites, of varying size and complexity. Two plans were generated on each dataset, one using the FIDO algorithm and the other using a conventional SWO method. Both SWO algorithms performed multi-field optimization and used the same optimization objectives. Different objective penalty weights were used on occasion so that comparable target volume coverage was achieved in both plans. The performance of each SWO algorithm was compared based on their optimization time, the number of iterations, and the quality of plans they produced.
Results: On average (standard deviation), the FIDO algorithm enhanced the optimization speed by 18.6 (20.7) fold, taking 19.4s (23.9s) over 68 (48) iterations to converge to a solution while the conventional algorithm took 85.0s (65.7s) over 216 (149) iterations to converge its solution. Plans of similar or better quality were obtained with the FIDO algorithm.
Conclusion: Promising plan quality and speed enhancement was obtained with the proposed FIDO SWO algorithm. This algorithm could significantly improve the efficiency of IMPT treatment planning and could allow for more sophisticated optimization and planning techniques to become feasible.