Discretization of optimal beamlet intensities in IMRT: A binary integer programming approach

Abstract The intensity modulated radiation therapy (IMRT) treatment planning problem is usually divided into three smaller problems that are solved sequentially: the geometry problem, intensity problem, and realization problem. That division has the consequence of causing a plan quality deterioration arising from the transition between the intensity problem and the realization problem. Typically, on the beamlet-based approach, after the optimal beamlet intensities are determined, they are discretized over a range of values using a distance criterion (rounding). However, that decision criterion is not appropriate and we present empirical evidence that this can lead to a significant deterioration of the treatment plan quality regardless of the model used to tackle the intensity problem. We propose a combinatorial optimization approach and a probabilistic binary tabu search algorithm to enable an improved transition from optimized to delivery fluence maps in IMRT by minimizing the deterioration of the treatment plan quality and improving organ sparing at the same time. Four head and neck clinical examples were used to test the ability of the proposed formulation and resolution method to obtain improved plans compared to the usual rounding procedure. The results obtained present a clear improvement of the treatment plan quality both in terms of target coverage and also in terms of parotid sparing.