Publikationsrepositorium - Helmholtz-Zentrum Dresden-Rossendorf

Ion beam therapy (IBT) is a promising treatment option in radiotherapy. The characteristic physical and biological properties of light ion beams allow for the delivery of highly tumour conformal dose distributions. Related to the sparing of surrounding healthy tissue and nearby organs at risk, it is feasible to escalate the dose in the tumour volume to reach higher tumour control and survival rates. Remarkable clinical outcome was achieved with IBT for radio-resistant, deep-seated, static and well fixated tumour entities. Presumably, more patients could benefit from the advantages of IBT if it would be available for more frequent tumour sites. Those located in the thorax and upper abdominal region are commonly subjected to intra-fractional, respiration related motion. Different motion compensated dose delivery techniques have been developed for active field shaping with scanned pencil beams and are at least available under experimental conditions at the GSI Helmholtzzentrum für Schwerionenforschung (GSI) in Darmstadt, Germany.
High standards for quality assurance are required in IBT to ensure a safe and precise dose application. Both underdosage in the tumour and overdosage in the normal tissue might endanger the treatment success. Since minor unexpected anatomical changes e.g. related to patient mispositioning, tumour shrinkage or tissue swelling could already lead to remarkable deviations between planned and delivered dose distribution, a valuable dose monitoring system is desired for IBT. So far, positron emission tomography (PET) is the only in vivo, in situ and non-invasive qualitative dose monitoring method applied under clinical conditions. It makes use of the tissue autoactivation by nuclear fragmentation reactions occurring along the beam path. Among others, +-emitting nuclides are generated and decay according to their half-life under the emission of a positron. The subsequent positron-electron annihilation creates two 511 keV photons which are emitted in opposite direction and can be detected as coincidence event by a dedicated PET scanner. The induced three-dimensional (3D) +- activity distribution in the patient can be reconstructed from the measured coincidences. Conclusions about the delivered dose distribution can be drawn indirectly from a comparison between two +-activity distributions: the measured one and an expected one generated by a Monte-Carlo simulation. This workflow has been proven to be valuable for the dose monitoring in IBT when it was applied for about 440 patients, mainly suffering from deep-seated head and neck tumours that have been treated with 12C ions at GSI.
In the presence of intra-fractional target motion, the conventional 3D PET data processing will result in an inaccurate representation of the +-activity distribution in the patient. Fourdimensional, time-resolved (4D) reconstruction algorithms adapted to the special geometry of in-beam PET scanners allow to compensate for the motion related blurring artefacts. Within this thesis, a 4D maximum likelihood expectation maximization (MLEM) reconstruction algorithm has been implemented for the double-head scanner Bastei installed at GSI. The proper functionality of the algorithm and its superior performance in terms of suppressing motion related blurring artefacts compared to an already applied co-registration approach has been demonstrated by a comparative simulation study and by dedicated measurements with moving radioactive sources and irradiated targets. Dedicated phantoms mainly made up of polymethyl methacrylate (PMMA) and a motion table for regular one-dimensional (1D) motion patterns have been designed and manufactured for the experiments. Furthermore, the general applicability of the 4D MLEM algorithm for more complex motion patterns has been demonstrated by the successful reduction of motion artefacts from a measurement with rotating (two-dimensional moving) radioactive sources. For 1D cos2 and cos4 motion, it has been clearly illustrated by systematic point source measurements that the motion influence can be better compensated with the same number of motion phases if amplitudesorted instead of time-sorted phases are utilized. In any case, with an appropriate parameter selection to obtain a mean residual motion per phase of about half of the size of a PET crystal size, acceptable results have been achieved. Additionally, it has been validated that the 4D MLEM algorithm allows to reliably access the relevant parameters (particle range and lateral field position and gradients) for a dose verification in intra-fractionally moving targets even from the intrinsically low counting statistics of IBT-PET data.
To evaluate the measured +-activity distribution, it should be compared to a simulated one that is expected from the moving target irradiation. Thus, a 4D version of the simulation software is required. It has to emulate the generation of +-emitters under consideration of the intra-fractional motion, their decay at motion state dependent coordinates and to create listmode data streams from the simulated coincidences. Such a revised and extended version that has been compiled for the special geometry of the Bastei PET scanner is presented within this thesis. The therapy control system provides information about the exact progress of the motion compensated dose delivery. This information and the intra-fractional target motion needs to be taken into account for simulating realistic +-activity distributions. A dedicated preclinical phantom simulation study has been performed to demonstrate the correct functionality of the 4D simulation program and the necessity of the additional, motionrelated input parameters.
Different to the data evaluation for static targets, additional effort is required to avoid a potential misleading interpretation of the 4D measured and simulated +-activity distributions in the presence of deficient motion mitigation or data processing. It is presented that in the presence of treatment errors the results from the simulation might be in accordance to the measurement although the planned and delivered dose distribution are different. In contrast to that, deviations may occur between both distributions which are not related to anatomical changes but to deficient 4D data processing. Recommendations are given in this thesis to optimize the 4D IBT-PET workflow and to prevent the observer from a mis-interpretation of the dose monitoring data. In summary, the thesis contributes on a large scale to a potential future application of the IBT-PET monitoring for intra-fractionally moving target volumes by providing the required reconstruction and simulation algorithms. Systematic examinations with more realistic, multi-directional and irregular motion patterns are required for further improvements. For a final rating of the expectable benefit from a 4D IBT-PET dose monitoring, future investigations should include real treatment plans, breathing curves and 4D patient CT images.