On the feasibility of using radioactive ion beams in hadrontherapy: dosimetric and imaging studies

In the context of hadrontherapy, whilst ions are capable of effectively destroying radio resistant, deep seated tumors, their treatment localization must be well assessed to ensure the sparing of surrounding healthy tissue and treatment effectiveness. In clinical practice, information on the in vivo beam range can be of paramount importance, for it provides confirmation that the tumor was targeted effectively. Range verification tech- niques, such as positron–emission–tomography (PET) imaging, hold great potential in clinical practice, in order to check the accuracy of the Bragg Peak in the delivered treat- ment by means of the comparison of the acquired signal with a corresponding prediction. In the aforementioned case of PET, the activity distribution arises from inelastic nu- clear interaction products between the ion and the tissue nuclei. It is therefore often deemed more advantageous to have the PET scanner integrated into the beam line, with scanning occuring during or immediately after irradiation, the so called in beam PET. This allows to mitigate acquisition errors related to patient positioning and biological washout, while benefiting from shorter half–life β$^+$ emitters contributing to the total PET signal. The overall success of the carbon ion therapy projects in Japan and Europe, along with the recent advances in accelerator technology and medical imaging, contributed to a renewed interest in innovative solutions for hadrontherapy applications. In particular, this work aimed at investigating the advantages of using β$^+$ emitting, radioactive ion beams of $^{11}$C and $^{15}$O in hadrontherapy, in comparison to the performance of their stable counterparts. To this end, the FLUKA Monte Carlo particle transport and interaction code was used to simulate radioactive ion beams in clinical environments, including the use of PET scans with equivalent dose delivery for different online– and offline–PET acquisition scenarios. The dosimetric performance evaluation with FLUKA benefited from its recent developments in charged hadron transport and fragment production models at relatively low energies of therapeutic interest. During the course of this work, the code has been used to simulate both mono–energetic and Spread Out Bragg Peaks (SOBP) in water and in an anthropomorphic head voxel phantom. $^{12}$C and $^{16}$O ion beam dose delivery was modeled in the simulations using an approximation of the synchroton beam delivery and beam line applied by the Heidelberg Ion Therapy Center (HIT). The research treatment planning system data were then used in extrapolations for radioactive ion beams. The imaging potential, particularly for range verification, was assessed with the newly developed FLUKA PET tools. This required a more detailed modeling of the Siemens Biograph mCT PET/CT (PET scanner model used at HIT) geometry and signal response. Furthermore, calculations of the annihilations events at rest ensuing from the β$^+$ emitters were performed. The time dependence of the corresponding PET signal was also included in the calculations, so that the effects of both the beam time structure and scan time could be reproduced in the final result. In the simulations performed using synchrotron–like irradiation schemes and the ap- proximate HIT beam line elements, it was verified that radioactive ion beams imaging results clearly outperform stable ion beam irradiations for every PET acquisition scenario, with a comparable dose delivery. In particular, it was observed approximately an order of magnitude higher amount of annihilation events at rest occurring when employing radioactive ion beams, for online PET acquisitions (130 seconds, including spill time) using $^{15}$O as well as for offline PET acquisition (5 to 30 minutes acquisition time after beam) using $^{11}$C compared to the stable counterparts. Furthermore, not only a considerable gain in coincidence events was observed, but also the quality of the reconstruction images was improved. Namely, radioactive ion beam results allowed for a better identification of the the distal edge of the SOBP (within 1 mm), with superior definition of proximal rise to distal fall–off regions with respect to their stable counterparts by up to a factor of 2 in $^{15}$O, in room and almost 3 for $^{11}$C, in offline PET acquisitions. In a second stage, experimental data acquired at the Heavy Ion Medical Accelerator at Chiba (HIMAC) were used to benchmark the simulation results. These data were obtained in collaboration with colleagues from the Japanese National Institute of Radiological Sciences (NIRS) Physics Imaging Team. The synchroton primary $^{12}$C and $^{16}$O ion beam were respectively converted into radioactive ion beams of $^{11}$ C and $^{15}$O using the projectile fragmentation separation method. The FLUKA code was employed for calculating energy deposition of Bragg Peak curves in water and polymethyl methacrylate (PMMA). The comparison of simulations and experimental results in water showed a good agreement, with range deviations below 1 mm. The amount and shape of β$^+$ activity were also calculated, and later compared with the ones obtained with an openPET scanner prototype, which collected data in between spills and some minutes afterwards. Despite the production method employed, which broadened considerably the Bragg Peak curves of radioactive ion beams by a factor 4–5 in water in comparison to the stable ion beams, the images reconstructed in PMMA using these species featured approximately a factor 2 better definition of the region between the proximal rise and distal fall–off compared to stable ion beams. These findings were also confirmed by the FLUKA simulations and are in line with previous observations. Moreover, the reconstructed signals indicate a considerable gain in magnitude, of at least one order of magnitude compared to stable ion beams, which is as well corroborated by FLUKA simulations. Concluding, the results in this work indicate that β$^+$ emitting radioactive ion beams can enhance the imaging signal output available for beam range verification and treatment monitoring with respect to stable ion beams in hadrontherapy.