Thesis supervisor: Prof. Patera, Prof. Schiavi (Dipartimento di Scienze di base ed applicate per l'Ingegneria)
Thesis title: Optimization of hadron therapy proton beam using Monte Carlo code on GPU
Nowadays one of the major issues related to the Treatment Planing Softwares is the hig computational time required in order to meet the quest of high accuracy. A software toolkit (FRED) has been developed for this aim. The short computational time is obtained through the code implementation on Graphics Processing Units (GPU). Thanks to their strong processes parallelization capability, GPUs significatly reduce the CPU time consuption, up to factor ~100 respect to a standard CPU running software. The software accuracy is ensured through the use of Monte Carlo methods to simulate the interaction of proton beams with tissues. The physical processes reproduced are: Multiple Coulomb Scattering (double and triple Gaussian models), energy straggling, elastic nuclear interactions (proton-proton and proton-oxygen) and inelastic nuclear interactions of protons with the main nuclei composing the biological tissues (N, O, C, P, Ca). Improvements in the reliability of the FRED simulation results are also coming from dropping the water equivalent assumption in order to start from the patient's anatomical information from Computed Tomography scans, reconstructing and simulating the atomic composition of each tissue crossed. Thanks to its features Fred can be used as an efficient tool for dose recalculation, on the day of the treatment. In fact it can provide in less than one minute the dose map obtained combining the treatment plan, earlier computed by the commercial TPS, and the current patient anatomic arrangement.