In this paper we introduce a p-median problem based clustering heuristic for selecting efficient beam angles for intensity-modulated radiation therapy. The essence of the method described here is the clustering of beam angles according to probability that an angle will be observed in the final solution and similarities among different angles and the selection of a representative angle from each of the p resulting cluster cells. We conduct experiments using several combinations of modeling parameters to find the conditions where the heuristic best performs. We found a combination of such parameters that outperformed all other parameters on three of the four tested instances.
Lim, Gino J.; Holder, Allen; and Reese, Josh, "A clustering approach for optimizing beam angles in IMRT planning" (2009). Mathematical Sciences Technical Reports (MSTR). 14.