A Bayesian Approach for Measurements of Stray Neutrons at Proton Therapy Facilities: Quantifying Neutron Dose Uncertainty.

Bonner sphere measurements are typically analyzed using unfolding codes. It is well known that it is difficult to get reliable estimates of uncertainties for standard unfolding procedures. An alternative approach is to analyze the data using Bayesian parameter estimation. This method provides reliable estimates of the uncertainties of neutron spectra leading to rigorous estimates of uncertainties of the dose. We extend previous Bayesian approaches and apply the method to stray neutrons in proton therapy environments by introducing a new parameterized model which describes the main features of the expected neutron spectra. The parameterization is based on information that is available from measurements and detailed Monte Carlo simulations. The validity of this approach has been validated with results of an experiment using Bonner spheres carried out at the experimental hall of the OncoRay proton therapy facility in Dresden.

Characterization of the microbunch time structure of proton pencil beams at a clinical treatment facility.

Proton therapy is an advantageous treatment modality compared to conventional radiotherapy. In contrast to photons, charged particles have a finite range and can thus spare organs at risk. Additionally, the increased ionization density in the so-called Bragg peak close to the particle range can be utilized for maximum dose deposition in the tumour volume. Unfortunately, the accuracy of the therapy can be affected by range uncertainties, which have to be covered by additional safety margins around the treatment volume. A real-time range and dose verification is therefore highly desired and would be key to exploit the major advantages of proton therapy. Prompt gamma rays, produced in nuclear reactions between projectile and target nuclei, can be used to measure the proton's range. The prompt gamma-ray timing (PGT) method aims at obtaining this information by determining the gamma-ray emission time along the proton path using a conventional time-of-flight detector setup. First tests at a clinical accelerator have shown the feasibility to observe range shifts of about 5 mm at clinically relevant doses. However, PGT spectra are smeared out by the bunch time spread. Additionally, accelerator related proton bunch drifts against the radio frequency have been detected, preventing a potential range verification. At OncoRay, first experiments using a proton bunch monitor (PBM) at a clinical pencil beam have been conducted. Elastic proton scattering at a hydrogen-containing foil could be utilized to create a coincident proton-proton signal in two identical PBMs. The selection of coincident events helped to suppress uncorrelated background. The PBM setup was used as time reference for a PGT detector to correct for potential bunch drifts. Furthermore, the corrected PGT data were used to image an inhomogeneous phantom. In a further systematic measurement campaign, the bunch time spread and the proton transmission rate were measured for several beam energies between 69 and 225 MeV as well as for variable momentum limiting slit openings. We conclude that the usage of a PBM increases the robustness of the PGT method in clinical conditions and that the obtained data will help to create reliable range verification procedures in clinical routine.

Experimental investigation of irregular motion impact on 4D PET-based particle therapy monitoring.

Particle therapy positron emission tomography (PT-PET) is an in vivo and non-invasive imaging technique to monitor treatment delivery in particle therapy. The inevitable patient respiratory motion during irradiation causes artefacts and inaccurate activity distribution in PET images. Four-dimensional (4D) maximum likelihood expectation maximisation (4D MLEM) allows for a compensation of these effects, but has up to now been restricted to regular motion for PT-PET investigations. However, intra-fractional motion during treatment might differ from that during acquisition of the 4D-planning CT (e.g. amplitude variation, baseline drift) and therefore might induce inaccurate 4D PET reconstruction results. This study investigates the impact of different irregular analytical one-dimensional (1D) motion patterns on PT-PET imaging by means of experiments with a radioactive source and irradiated moving phantoms. Three sorting methods, namely phase sorting, equal amplitude sorting and event-based amplitude sorting, were applied to manage the PET list-mode data. The influence of these sorting methods on the motion compensating algorithm has been analysed. The event-based amplitude sorting showed a superior performance and it is applicable for irregular motions with ⩽ 4 mm amplitude elongation and drift. For motion with 10 mm baseline drift, the normalised root mean square error was as high as 10.5% and a 10 mm range deviation was observed.

Detection of mixed-range proton pencil beams with a prompt gamma slit camera.

With increasing availability of proton and particle therapy centers for tumor treatment, the need for in vivo range verification methods comes more into the focus. Imaging of prompt gamma rays emitted during the treatment is one of the possibilities currently under investigation. A knife-edge shaped slit camera was recently proposed for this task and measurements proved the feasibility of range deviation detection in homogeneous and inhomogeneous targets. In the present paper, we concentrate on laterally inhomogeneous materials, which lead to range mixing situations when crossed by one pencil beam: different sections of the beam have different ranges. We chose exemplative cases from clinical irradiation and assembled idealized tissue equivalent targets. One-dimensional emission profiles were obtained by measuring the prompt gamma emission with the slit camera. It could be shown that the resulting range deviations can be detected by evaluation of the measured data with a previously developed range deviation detection algorithm. The retrieved value, however, strongly depends on the target composition, and is not necessarily in direct relation to the ranges of both parts of the beam. By combining the range deviation detection with an analysis of the slope of the distal edge of the measured prompt gamma profile, the origin of the detected range deviation, i.e. the mixed range of the beam, is also identified. It could be demonstrated that range mixed prompt gamma profiles exhibit less steep distal slopes than profiles from beams traversing laterally homogeneous material. For future application of the slit camera to patient irradiation with double scattered proton beams, situations similar to the range mixing cases are present and results could possibly apply.

Measurement of prompt gamma profiles in inhomogeneous targets with a knife-edge slit camera during proton irradiation.

Proton and ion beam therapies become increasingly relevant in radiation therapy. To fully exploit the potential of this irradiation technique and to achieve maximum target volume conformality, the verification of particle ranges is highly desirable. Many research activities focus on the measurement of the spatial distributions of prompt gamma rays emitted during irradiation. A passively collimating knife-edge slit camera is a promising option to perform such measurements. In former publications, the feasibility of accurate detection of proton range shifts in homogeneous targets could be shown with such a camera. We present slit camera measurements of prompt gamma depth profiles in inhomogeneous targets. From real treatment plans and their underlying CTs, representative beam paths are selected and assembled as one-dimensional inhomogeneous targets built from tissue equivalent materials. These phantoms have been irradiated with monoenergetic proton pencil beams. The accuracy of range deviation estimation as well as the detectability of range shifts is investigated in different scenarios. In most cases, range deviations can be detected within less than 2 mm. In close vicinity to low-density regions, range detection is challenging. In particular, a minimum beam penetration depth of 7 mm beyond a cavity is required for reliable detection of a cavity filling with the present setup. Dedicated data post-processing methods may be capable of overcoming this limitation.

Experimental verification of a 4D MLEM reconstruction algorithm used for in-beam PET measurements in particle therapy.

In-beam positron emission tomography (PET) has been proven to be a reliable technique in ion beam radiotherapy for the in situ and non-invasive evaluation of the correct dose deposition in static tumour entities. In the presence of intra-fractional target motion an appropriate time-resolved (four-dimensional, 4D) reconstruction algorithm has to be used to avoid reconstructed activity distributions suffering from motion-related blurring artefacts and to allow for a dedicated dose monitoring. Four-dimensional reconstruction algorithms from diagnostic PET imaging that can properly handle the typically low counting statistics of in-beam PET data have been adapted and optimized for the characteristics of the double-head PET scanner BASTEI installed at GSI Helmholtzzentrum Darmstadt, Germany (GSI). Systematic investigations with moving radioactive sources demonstrate the more effective reduction of motion artefacts by applying a 4D maximum likelihood expectation maximization (MLEM) algorithm instead of the retrospective co-registration of phasewise reconstructed quasi-static activity distributions. Further 4D MLEM results are presented from in-beam PET measurements of irradiated moving phantoms which verify the accessibility of relevant parameters for the dose monitoring of intra-fractionally moving targets. From in-beam PET listmode data sets acquired together with a motion surrogate signal, valuable images can be generated by the 4D MLEM reconstruction for different motion patterns and motion-compensated beam delivery techniques.

4D particle therapy PET simulation for moving targets irradiated with scanned ion beams.

Particle therapy positron emission tomography (PT-PET) allows for an in vivo and in situ verification of applied dose distributions in ion beam therapy. Since the dose distribution cannot be extracted directly from the ß(+)-activity distribution gained from the PET scan the validation is done by means of a comparison between the reconstructed ß(+)-activity distributions from a PT-PET measurement and from a PT-PET simulation. Thus, the simulation software for generating PET data predicted from the treatment planning is an essential part of the dose verification routine. For the dose monitoring of intra-fractionally moving target volumes the PET data simulation needs to be upgraded by using time resolved (4D) algorithms to account correctly for the motion dependent displacement of the positron emitters. Moreover, it has to consider the time dependent relative movement between target volume and scanned beam to simulate the accurate positron emitter distribution generated during irradiation. Such a simulation program is presented which properly proceeds with motion compensated dose delivery by scanned ion beams to intra-fractionally moving targets. By means of a preclinical phantom study it is demonstrated that even the sophisticated motion-mitigated beam delivery technique of range compensated target tracking can be handled correctly by this simulation code. The new program is widely based on the 3D PT-PET simulation program which had been developed at the Helmholtz-Zentrum Dresden-Rossendorf, Germany (HZDR) for application within a pilot project to simulate in-beam PET data for about 440 patients with static tumor entities irradiated at the former treatment facility of the GSI Helmholtzzentrum für Schwerionenforschung, Darmstadt, Germany (GSI). A simulation example for a phantom geometry irradiated with a tracked (12)C-ion beam is presented for demonstrating the proper functionality of the program.

On the feasibility of automatic detection of range deviations from in-beam PET data.

In-beam PET is a clinically proven method for monitoring ion beam cancer treatment. The objective is predominantly the verification of the range of the primary particles. Due to different processes leading to dose and activity, evaluation is done by comparing measured data to simulated. Up to now, the comparison is performed by well-trained observers (clinicians, physicists). This process is very time consuming and low in reproducibility. However, an automatic method is desirable. A one-dimensional algorithm for range comparison has been enhanced and extended to three dimensions. System-inherent uncertainties are handled by means of a statistical approach. To test the method, a set of data was prepared. Distributions of ß(+)-activity calculated from treatment plans were compared to measurements performed in the framework of the German Heavy Ion Tumor Therapy Project at GSI Helmholtz Centre for Heavy Ion Research, Darmstadt, Germany. Artificial range deviations in the simulations served as test objects for the algorithm. Range modifications of different depth (4, 6 and 10 mm water equivalent path length) can be detected. Even though the sensitivity and specificity of a visual evaluation are higher, the method is feasible as the basis for the selection of patients from the data pool for retrospective evaluation of treatment and treatment plans and correlation with follow-up data. Furthermore, it can be used for the development of an assistance tool for a clinical application.