Optimized beam planning for linear accelerator-based stereotactic radiosurgery.

Abstract

PURPOSE: Current treatment planning for linear accelerator-based stereotactic radiosurgery and radiotherapy is a lengthy and iterative procedure. The planner has to manually select the beam arcs and carefully consider many different selections to ensure target volume coverage while sparing dose to critical organs. In this article we report an optimization procedure that can automatically select the beam arcs based on geometric and dosimetric analysis of the treatment parameters. METHODS AND MATERIALS: The optimization problem is introduced by using a Beam's Eye View (BEV) map where a pattern of lines represents a beam arc combination for a treatment plan. The collection of all possible treatment plans is described by using the concept of phase space where each point corresponds to a particular configuration of the system under consideration, and in this case, a particular beam arc combination. A geometric reduction of the phase space is performed by excluding static beam ports that irradiate too much critical organs and too little target volume. The phase space is further reduced by excluding beam arc combinations that do not comply with treatment convenience considerations and established planning experiences. These reductions significantly reduces the number of beam arc combinations to be considered and thus dramatically simplifies the computational complexity. The method of simulated annealing is then used to the reduced phase space to select the set of beam arcs that provides the best surface dose distribution for the target volume. The optimization procedure is applied to a radiosurgery case to compare the optimized beam arcs with the previously manually planned beam arcs. The procedure is also applied to 10 randomly selected cases for a comparison in terms of tissue-volume ratio calculations. RESULTS: The system is a highly automated beam arc planning tool for stereotactic radiosurgery and stereotactic radiotherapy. Its interactive nature allows the planner to rapidly consider many treatment plans to search for the best option. For the case presented, it is shown that the optimized beams substantially reduce the dose to the postrema. The tissue-volume ratio calculations demonstrate that the optimization often produces clinically superior treatment plans than the manual beam planning method. CONCLUSIONS: Our method of phase space reduction proves to be very useful in approaching the complex problem of treatment planning optimization. Not only does it substantially reduce the number of beam arcs that need to be considered, but it also simplifies the evaluation of the beam arc options. Both of these greatly reduce the computational complexity of the optimization and make the procedure fast and efficient. Moreover, the reduction of phase space adds another layer of interaction between the user and the beam selection procedure, so that the optimization process is well controlled and thus very effective.