A geometry based optimisation algorithm for conformal external beam orientation

We consider topology based criteria for the selection of optimal beam orientations for conformal external radiotherapy. The advantage of such an approach is that computation time is considerably reduced since no dose calculation is required. The algorithm preclassifies all possible irradiation directions by assigning to each a fitness measure, the geometry fitness factor (GFF). This factor is a measure of the degree of fitting of the beam 3D shape to the planning target volume (PTV) by simultaneously avoiding any intersection with the critical structures. Individual shielding and table-gantry collision as well as userdefined margins are taken into account. After classifying all possible beam directions, adaptive simulated annealing (ASA) has been used to search for an optimal configuration of a specific number of beams by minimizing a cost function that takes into account individual beam feasibility and beam spacing. The appropriateness of choosing ASA as optimisation procedure has been checked by comparing its performance with that of the standard genetic algorithm (SGA). A phantom case where the best directions are a priori known was used to test the beam selection criteria. The robustness of the initial hypothesis is tested by comparing the dosimetric results of standard and optimised plans for a set of three clinical cases. E Schreibmann et al: A geometrical-based beam orientation optimization algorithm for conformal external beam radiotherapy 2