Experiments and FLUKA simulations of $^{12}C$ and $^{16}O$ beams for therapy monitoring by means of in-beam Positron Emission Tomography

Since 1997 at the experimental C-12 ion therapy facility at Gesellschaft fuer Schwerionenforschung (GSI), Darmstadt, Germany, more than 350 patients have been treated. The therapy is monitored with a dedicated positron emission tomograph, fully integrated into the treatment site. The measured beta+-activity arises from inelastic nuclear interactions between the beam particles an the nuclei of the patients tissue. Because the monitoring is done during the irradiation the method is called in-beam PET. The underlying principle of this monitoring is a comparison between the measured activity and a simulated one. The simulations are presently done by the PETSIM code which is dedicated to C-12 beams. In future ion therapy centers like the Heidelberger Ionenstrahl Therapiezentrum (HIT), Heidelberg, Germany, besides C-12 also proton, $^3$He and O-16 beams will be used for treatment and the therapy will be monitored by means of in-beam PET. Because PETSIM is not extendable to other ions in an easy way, a code capable to predict the beta+-activity created by all ions of interrest, also possible future ones, is needed. A candidate for such a role is the multi purpose particle transport and interaction Monte Carlo code FLUKA. The objective of this thesis is to investigate the ability of FLUKA to predict the beta+-activity induced by C-12 and O-16 beams, necessary for in-beam PET. Experiments with O-16 and C-12 beams on homogeneous targets of water, polymethyl methacrylate (PMMA) and graphite were performed and the created beta+-activity was measured by means of in-beam PET. In case of the O-16 beams this was done for the first time. The build up and decay of the beta+-active nuclei, their spatial distribution and total amount of produced beta+-activity was investigated. The experimental data was used for benchmarking FLUKA and its implemented nuclear reaction modells. Of special interrest was the performance of the recently added event generator BME which handels nucleus-nucleus interactions at low energies. FLUKA was interfaced with the part of PETSIM which modells the detector response and stores the detected events in list mode data format. This enables to process the simulated data exactly the same way than in the experimental case and avoids uncertainties due to the detection and backprojection. To reduce the computing time, full advantage of the biasing options in FLUKA was taken. Because of the double head geometry of the in-beam PET only a small fraction of the created annihilation photons can be detected. Because most of the annihilation photons produced are not detected, a lot of computing time is wasted. To futher speed up the simulations decay direction biasing for annihilation photons was introduced. This biasing preferentially emits the annihilation photons into a wanted direction, i.e. the detector heads. The use of this biasing additionally with other standard biasing methods brought an dramatic improvement in terms of computing time. In the final simulations good agreement between measured and simulated beta+-activity could be obtained within reasonable computing times. However, also limits of the new BME event generator in its present implementetion were pointed out.