Experimental verification of an algorithm for inverse radiation therapy planning.

In inverse radiotherapy planning, the traditional dose planning sequence is reversed. This makes it possible to calculate the optimal incident beam profiles required to produce the desired dose distribution in the target volume by solving an integral equation with an iterative algorithm. The major advantage, compared with conventional treatment planning, is that the trial and error part is avoided, and replaced by a deterministic calculation of the optimal treatment plan. In the present paper this algorithm is briefly described and compared with experimental results and an analytical inversion formula which is valid for a cylindrical geometry. The experiments were performed with non-homogeneous beams shaped with compensators designed using the algorithm. The agreement between the experimental results and the predictions of the algorithm are quite good, generally within about 5%. The differences are caused by discretization noise due to the finite resolution of the calculation matrix, imperfections in the experimental situation, and by the assumption of spatial invariant dose distribution kernels.