Advanced radiotherapy techniques use complex beam delivery systems like focused multipole beams, in order to obtain high tumor control probability with reduced normal tissue complication rate. The Treatment Planning System (TPS) is the complex software that controls the accelerator parameters used during the treatment: intensity, energies, transverse positions, etc, needed to obtain the optimal dose deliver to the patient. Usually, the commercial TPS provided by the accelerator vendors are analytical code and are outperformed by Monte Carlo based TPS especially for those cases with very heterogeneous tissue. The advent of general programming Graphics Processing Units (GPU) has prompted the development of MC codes that can dramatically reduce the plan recalculation time with respect to full MC codes in Central Processing Unit (CPU) hardware. The impressive speed gain compared to CPU-based calculations is due to both algorithmic simplification and hardware acceleration. FRED code has been developed in this framework. It is a MC-based code that runs on GPU developed to recalculate and optimize ion beam treatment planning within minutes, opening the way to many clinical applications where the calculation time is important. As far as proton beams are concerned FRED is already used as a quality assurance tool and as research.
The Intra Operative Radiation Therapy (IORT) may represent one of the first clinical modalities of a Flash clinical treatment. Within IORT, whenever needed and possible, temporarily beam modifiers (such as the protection disc for breast carcinoma treatment) are used to protect the underlying healthy tissues during the irradiation. This project is focused on the development of the electromagnetic FRED code in order to extend the use of this MC-on-GPU based as a tool for dose calculation and treatment optimization for IORT and Flash Therapy treatments.
As already said today the main IORT limitation is the absence of a TPS: the effects of beam misalignment, gaps, bolus, changes in the penumbra, and tissue inhomogeneities in realistic patient geometries are not properly investigated. Therefore dose planning is not optimized to the patient-specific tumor and there is not dose report: the information concerning the amount of dose released into the tumor and into the surrounding healthy organs is not known "a posteriori".
Until now it has not been possible to elaborate a specific treatment plan for each patient during the surgical operation, due to the dose calculations time provided by the standard MC tools. Therefore the tumor patient is irradiated using a uniform electron beam and in order to protect the surrounding healthy organs a metallic disk is inserted under the tumor bed. Since the TPS must be calculate during the surgery, where the patient is highly exposed, it is essential to minimize the simulation time. For this reason, for the exceptional speed of the proton tracking algorithms implemented in FRED and for the excellent results achieved, the purpose of my project is to develop for the first time a TPS for IORT based on the FRED code. To this aim became essential a patient intra-operative image in order to provide informations about the material shape and density that the electron beam will cross. Developing a "local imaging" method to reconstruct the anatomy of the modified area during the operation, for example by ultrasound or optical sensors, is possible to "correct" the original CT and evaluate the changes made by the surgery. Taking into account the new imaging is than possible, using FRED code, to have a patient specific TPS that takes only few minutes of computing time. FRED code allows to leave the uniform irradiation method in favor of a pencil beam one that takes into account the specific tumor site morphology, thus allowing to treat the residual tumor cells with a better efficiency and accuracy. This project will also has an application to the FLASH therapy as the IORT technique may represents one of the first clinical modalities of a FLASH clinical treatment.