| As an advanced technique of three-dimensional conformal radiotherapy (3D-CRT), intensity modulated radiotherapy (IMRT) has been applied into clinical oncology world widely. The intensity of the radiation beam is modulated across the treatment field. Comparatively even dosage can be achieved by using modulated irradiation on the tumor target. At the same time, less dose falls into the near ambient normal and sensitive tissue. IMRT has significant potential for improving the therapeutic ratio and reducing toxicity.IMRT treatment plans are often generated using inverse planning, which is different from traditional forward planning. First, doctor determines the prescribe dose. Then optimization techniques are used to help determine the distribution of intensities across the target volume. The optimization of the treatment planning is the crucial part of the whole inverse planning system, which directly influence the precision and efficiency of the therapy.This paper is focused on the research of the optimization for the inverse planning of IMRT. First optimization methods are introduced, which include gradient methods and stochastic methods, such as simulated annealing, genetic algorithm, etc. Secondly, the concept of multi-objective optimization has been analyzed in detail, especially the NSGA-II algorithm. Thirdly, the dose calculation models of both regular fields and irregular fields have been presented, where the dose calculation based on the pencil beam model is mainly introduced. Finally two optimization methods have been studied for IMRT planning. One is based on gradient method and the other is based on multi-objective genetic algorithm. They have been applied to some simulated data and have been compared with each other. The first method is based on L-BFGS algorithm, where objective functions are transformed to a single objective one. The method converges quickly, but often falls into local optimal, and only one optimal solution can be obtained at one time. In comparison, NSGA-II method can provide Pareto optimal set for clinical choice. It's more flexible and meets the clinical requirements better. |
