Multiobjective evolutionary optimization of number of beams, their orientations and weights for IMRT

We propose a hybrid multiobjective (MO) evolutionary optimization algorithm (MOEA) for intensity modulated radiotherapy inverse planning and apply it to optimize the number of incident beams, their orientations and intensity profiles. The algorithm produces a set of efficient solutions, which represent different clinical tradeoffs and contains information such as variety of dose distributions and dosevolume histograms (DVH). No importance factors are required and solutions can be obtained in regions not accessible by conventional weighted sum approaches. The application of the algorithm using a test case, a prostate and a head and neck tumor case is shown. The results are compared with MO inverse planning using a gradient-based optimization algorithm. E. Schreibmann et al: Multiobjective evolutionary optimization 2/25