Final Report Summary - FULLBEAM (Fully digital in-beam PET for hadron therapy)
The possibility of controlling the penetration range of hadronic nuclei in radiotherapy makes their use advantageous with respect to photons: the better conformity of the irradiated dose to the tumour, in principle, implies better tumour control and less toxicity to the surrounding healthy tissues. The most popular method for verifying that hadrons are delivered with the correct range is PET monitoring. It works by imaging positron emitters generated in the patient after (off-beam) or during treatment (in-beam) and using the images to indirectly measure delivered dose. Successful estimation of range from the recorded activity requires high-quality PET images, but this is not always straightforward to achieve. In-beam PET systems are subject to more geometrical constraints than in conventional PET imaging (it must allow beam delivery and bed movement), therefore, they can’t be as sensitive as in the best clinical arrangement. In-beam PET must filter out data acquired during particles extraction, because the instantaneous beam-induced background noise (e.g. prompt gamma) is much stronger than the useful signal. Finally, the positron-emitting species involved in hadron therapy (e.g. 15-O and 11-C) have generally much shorter half-lives than in oncological PET based on 18F-FDG, thus, their emissions must be acquired very quickly. The FULLBEAM project aims at providing a proof of concept fully-digital PET system able to discriminate positron annihilation photons from the prompt gamma background present during in-beam acquisitions. This is achieved by using a new acquisition architecture and new coincidence processing algorithms designed to better reject the noise radiation.
The activity of the first year has been carried out following two parallel research directions, as originally proposed:
1) specific in-beam experiments have been carried out in order to demonstrate the feasibility of successful PET acquisitions during the irradiation. These experiments were required to produce the necessary knowledge to understand under what conditions beam-on acquisitions are possible, and to have a quantitative estimation of the technological limits that need to be overcome. The already available prototype system (QPEM) was updated and used at the CATANA cyclotron-based proton therapy center, INFN-LNS, Catania (Italy) and at the CNAO synchrotron-based proton/carbon therapy center, CNAO foundation, Pavia (Italy). The QPEM system consisted of two planar LYSO detector heads of 10 cm x 10 cm, each one coupled to photomultiplier tubes (PMT). Updates involved the reduction of electronics dead time at the front-end, a reduction of the coincidence window, an increase of the acquisition throughput of the mainboard and the optimization of calibration procedures to correct the different system response at different input count rates. Results have been published as open access article on Physics in Medicine and Biology (“First full-beam PET acquisitions in proton therapy with a modular dual-head dedicated system”, G. Sportelli et al., PMB, Vol. 59, n. 1, pp. 43-60). The article has been listed as one of the ten most downloaded articles on PMB during the first month of publication and has been covered by a news article on medicalphysicsweb.org.
2) An improved acquisition system has been designed according to the results of such experimental activity. The new system is compatible with the SiPM-based PET detectors being developed for the Italian INSIDE project (MIUR PRIN 2010-2011) and is able to sustain input rates two orders of magnitude higher than the QPEM system. With such a higher bandwidth it is possible to digitize all individual photons and then match them in pairs using complex coincidence filtering based on energy and time constraints, including phase relationships with the particle acceleration cycles. The construction of the acquisition electronics was delegated to the collaborating company AGE Scientific and took the whole second year to be completed.
In the second year, the fellow has visited the Department of Radiation Oncology of the Massachusetts General Hospital (MGH), Boston (U.S.A.) for four months. In this period, he followed specialised training on particle beam physics and studied by simulation the characteristics of the radiation produced by beams of protons on a PMMA phantom. Special attention was paid to time and energy characteristics of the prompt-radiation that overlaps the beta-decays while the beam is on. This allowed comparing and validating the results previously published with a simulated replica of the experimental setup.
Results
A proof of concept PET system able to acquire in “full-beam” conditions has been realised and tested with proton beams at cyclotron and synchrotron based facilities. With cyclotron beams irradiating at 62 MeV (as in typical eye melanoma treatments), it was showed experimentally that full-beam acquisitions improve the image quality and therefore the capability of detecting the particles range. This can be achieved with a coincidence window of 3 ns, even if no random coincidences correction is applied. With higher energies or with synchrotron beams, however, experiments and simulations confirmed the following two disruptive effects that incur during the irradiation.
1) The old front-end electronics paralyses because of the too high single photon input rate. Early evidences suggested that it is not only the acquisition electronics that is paralysed, but that the PMT is not capable of charging its dynodes sufficiently quickly to sustain such rates. This partially impairs the advantage of using faster fully-digital acquisition electronics; however, it reinforces the validity of the SiPM-based approach. SiPMs are faster and more compact than PMTs: by assembling small independent SiPM modules, it is possible to better parallelise their readout and therefore to increase the maximum input rate by more than one order of magnitude with respect to PMTs.
2) Decay coincidence rates are too small with respect to the total input radiation to be successfully reconstructed. By means of simulation, the background radiation has been discriminated as coming from three source types: prompt gamma, prompt neutrons interactions and annihilations following pair productions. Each type of background has a specific time-energy distribution that can be used to improve the coincidence filtering. Prompt-gamma and pair-productions happen in time ranges of the order of 1 ns with respect to the beginning of the particle extraction. Since the extraction period ranges from 10 ns (in cyclotrons) to a few hundreds of nanoseconds (in synchrotrons), it is possible in principle to synchronise the acquisition system with the extraction cycle and filter out all the radiation coming in the first few nanoseconds of each cycle. Moreover, the deposited energy distribution of prompt-gamma and neutrons is flat in a range that exceeds several MeV, hence, they can be better filtered by narrowing the energy window around the 511 keV peak (see Figure 1).
Impact
Acquiring data during the irradiation in hadron therapy has become a major priority in PET monitoring for hadron therapy. In the last two years a general consensus is growing on the necessity to give a verification feedback in situ and shortly after the treatment. The FULLBEAM project followed a pioneering approach in this aspect by answering on one hand to the long standing demand for a faster and more sophisticated PET acquisition system and, on the other hand, providing evidence that proofs the feasibility of full-beam acquisitions even with cyclotron-based particle beams.
The acquisition system developed within the FULLBEAM project is one of the new enabling technologies adopted in the national "INSIDE" (Innovative Solutions for Dosimetry in Hadrontherapy) project, funded by the Italian Ministry of Education University and Research. INSIDE is coordinated by the University of Pisa, it involves 3 other universities and the National Institute of Nuclear Physics. Its main objective is to provide a commercial-grade verification tool for hadron therapy, made of the latest available technologies, including a full-beam SiPM-based PET scanner.
Another project that will benefit from FULLBEAM technological developments is the EU-funded "TRIMAGE" project (FP7 HEALTH.2013.2.2.1-2 602621). TRIMAGE aims at developing a low cost trimodality PET/MR/EEG system for brain imaging. Its is coordinated by the University of Pisa and involves 10 other entities, including public research institutions, small and medium enterprises. In TRIMAGE the PET system will be very sensitive, i.e. it will have to handle high input count rates, hence the idea of using the same acquisition system as for FULLBEAM.
The activity of the first year has been carried out following two parallel research directions, as originally proposed:
1) specific in-beam experiments have been carried out in order to demonstrate the feasibility of successful PET acquisitions during the irradiation. These experiments were required to produce the necessary knowledge to understand under what conditions beam-on acquisitions are possible, and to have a quantitative estimation of the technological limits that need to be overcome. The already available prototype system (QPEM) was updated and used at the CATANA cyclotron-based proton therapy center, INFN-LNS, Catania (Italy) and at the CNAO synchrotron-based proton/carbon therapy center, CNAO foundation, Pavia (Italy). The QPEM system consisted of two planar LYSO detector heads of 10 cm x 10 cm, each one coupled to photomultiplier tubes (PMT). Updates involved the reduction of electronics dead time at the front-end, a reduction of the coincidence window, an increase of the acquisition throughput of the mainboard and the optimization of calibration procedures to correct the different system response at different input count rates. Results have been published as open access article on Physics in Medicine and Biology (“First full-beam PET acquisitions in proton therapy with a modular dual-head dedicated system”, G. Sportelli et al., PMB, Vol. 59, n. 1, pp. 43-60). The article has been listed as one of the ten most downloaded articles on PMB during the first month of publication and has been covered by a news article on medicalphysicsweb.org.
2) An improved acquisition system has been designed according to the results of such experimental activity. The new system is compatible with the SiPM-based PET detectors being developed for the Italian INSIDE project (MIUR PRIN 2010-2011) and is able to sustain input rates two orders of magnitude higher than the QPEM system. With such a higher bandwidth it is possible to digitize all individual photons and then match them in pairs using complex coincidence filtering based on energy and time constraints, including phase relationships with the particle acceleration cycles. The construction of the acquisition electronics was delegated to the collaborating company AGE Scientific and took the whole second year to be completed.
In the second year, the fellow has visited the Department of Radiation Oncology of the Massachusetts General Hospital (MGH), Boston (U.S.A.) for four months. In this period, he followed specialised training on particle beam physics and studied by simulation the characteristics of the radiation produced by beams of protons on a PMMA phantom. Special attention was paid to time and energy characteristics of the prompt-radiation that overlaps the beta-decays while the beam is on. This allowed comparing and validating the results previously published with a simulated replica of the experimental setup.
Results
A proof of concept PET system able to acquire in “full-beam” conditions has been realised and tested with proton beams at cyclotron and synchrotron based facilities. With cyclotron beams irradiating at 62 MeV (as in typical eye melanoma treatments), it was showed experimentally that full-beam acquisitions improve the image quality and therefore the capability of detecting the particles range. This can be achieved with a coincidence window of 3 ns, even if no random coincidences correction is applied. With higher energies or with synchrotron beams, however, experiments and simulations confirmed the following two disruptive effects that incur during the irradiation.
1) The old front-end electronics paralyses because of the too high single photon input rate. Early evidences suggested that it is not only the acquisition electronics that is paralysed, but that the PMT is not capable of charging its dynodes sufficiently quickly to sustain such rates. This partially impairs the advantage of using faster fully-digital acquisition electronics; however, it reinforces the validity of the SiPM-based approach. SiPMs are faster and more compact than PMTs: by assembling small independent SiPM modules, it is possible to better parallelise their readout and therefore to increase the maximum input rate by more than one order of magnitude with respect to PMTs.
2) Decay coincidence rates are too small with respect to the total input radiation to be successfully reconstructed. By means of simulation, the background radiation has been discriminated as coming from three source types: prompt gamma, prompt neutrons interactions and annihilations following pair productions. Each type of background has a specific time-energy distribution that can be used to improve the coincidence filtering. Prompt-gamma and pair-productions happen in time ranges of the order of 1 ns with respect to the beginning of the particle extraction. Since the extraction period ranges from 10 ns (in cyclotrons) to a few hundreds of nanoseconds (in synchrotrons), it is possible in principle to synchronise the acquisition system with the extraction cycle and filter out all the radiation coming in the first few nanoseconds of each cycle. Moreover, the deposited energy distribution of prompt-gamma and neutrons is flat in a range that exceeds several MeV, hence, they can be better filtered by narrowing the energy window around the 511 keV peak (see Figure 1).
Impact
Acquiring data during the irradiation in hadron therapy has become a major priority in PET monitoring for hadron therapy. In the last two years a general consensus is growing on the necessity to give a verification feedback in situ and shortly after the treatment. The FULLBEAM project followed a pioneering approach in this aspect by answering on one hand to the long standing demand for a faster and more sophisticated PET acquisition system and, on the other hand, providing evidence that proofs the feasibility of full-beam acquisitions even with cyclotron-based particle beams.
The acquisition system developed within the FULLBEAM project is one of the new enabling technologies adopted in the national "INSIDE" (Innovative Solutions for Dosimetry in Hadrontherapy) project, funded by the Italian Ministry of Education University and Research. INSIDE is coordinated by the University of Pisa, it involves 3 other universities and the National Institute of Nuclear Physics. Its main objective is to provide a commercial-grade verification tool for hadron therapy, made of the latest available technologies, including a full-beam SiPM-based PET scanner.
Another project that will benefit from FULLBEAM technological developments is the EU-funded "TRIMAGE" project (FP7 HEALTH.2013.2.2.1-2 602621). TRIMAGE aims at developing a low cost trimodality PET/MR/EEG system for brain imaging. Its is coordinated by the University of Pisa and involves 10 other entities, including public research institutions, small and medium enterprises. In TRIMAGE the PET system will be very sensitive, i.e. it will have to handle high input count rates, hence the idea of using the same acquisition system as for FULLBEAM.
