Treatment-plan optimization for soft-tissue sarcoma brachytherapy using a genetic algorithm

To describe a treatment-plan optimization system for temporary implant of soft-tissue sarcomas using a genetic algorithm, and evaluate its potential advantages over manual planning.

A planning system that optimizes the distribution of radioactive seeds needed for adequate coverage of the target in the treatment of soft-tissue sarcomas has been designed and implemented. The treatment-planning procedures include simulation, film digitization, target-volume definition, optimized planning, and plan evaluation. The input to the optimization program consists of seed coordinates reconstructed from isocentric films, prescription points, and a list of available seed activities. The optimization is performed using a genetic algorithm.

Case studies are presented, which compare plans generated by computer optimization or by trial and error (manually). As expected, computer-optimized plans are often (but not always) superior to manual plans. This is particularly evident for situations where (unavoidably) catheters are far apart or irregularly spaced, in which case the advantages of optimized planning in terms of tumor coverage can be quite dramatic. When the target volume is well contained, the optimized plan minimizes the dose to normal tissue. Computer-based optimization has the additional advantage of being much faster than manual planning; this is valuable because it often reduces the total time the patient will spend in the hospital before implantation.

Optimized planning with a genetic algorithm and seeds of different activities significantly improves planning efficiency and generally results in improved plan quality. The utility of this optimization system is not limited to sarcoma implants.