Validation of a pencil beam model-based treatment planning system for fast neutron therapy.

Treatment planning systems (TPSs) are used to compute dosedelivered to the patient. In the case of fast neutron therapy, TPSs are mostly not of general purpose but are dedicated to one facility. This is due to the few fast neutron facilities worldwide and due to the high variation in the neutron energy distributions. Efforts have been undertaken to develop a new TPS that could be applied to all the existing fast neutron facilities. The University Hospital of Essen operates a d (14 MeV )+ Be fast neutron beam and the TPS used is based on an empirical model. In a previous study, the empirical model has been evolved to a pencil beam model of 35 monoenergetic neutron beams. Monte Carlo techniques have been utilized to compute distributions of the energy deposition due to primary and scatteredneutrons in a simple geometry water phantom. The experimental validation of the method is now presented. Depth dose curves in water of monoenergetic neutrons have been derived from the distributions of energy deposition. The resultant depth dose curves have been utilized in order to determine the depth dose curves of the fast neutron beam of the Essen facility for the 14 radiationfield sizes available in this facility. This determination requires the initial neutron spectrum. As this spectrum could not be measured at the Essen facility, the initial neutron spectrum of the Physikalisch Technische Bundesanstalt, Braunschweig, Germany, which operates the same cyclotron, was used. The calculated depth dose curves were compared to experimental depth dose curves that have been obtained in water at the University Hospital of Essen. The comparison between calculated and experimental depth dose curves showed significant deviations in the case of large radiation fields and of depth less than 5 cm. In the case of radiation field areas less than 150 cm 2 and depth more than 5 cm (usual clinical situation), the measured and calculated values are in a good agreement. In the case of clinical situation, the dependence on the radiationfield size is relatively well taken into account by the model presented here.