Verdi Vanreusel

Ultra-high dose rate radiotherapy (UHDR-RT) is an emerging cancer treatment technique that triggers the FLASH effect which minimizes healthy tissue damage while controlling tumors. However, clinical implementation of UHDR-RT is hindered by the lack of dedicated machines and standardized dosimetric procedures. Accurate UHDR-RT dosimetry requires advanced detectors that address challenges such as saturation and dose rate dependence. Radioluminescence has emerged as a promising technique for UHDR dosimetry, offering high spatial and temporal resolution for real time multidimensional measurements. This study evaluates 2D RL films based on natural Brazilian calcium fluoride (CaF₂:N) for UHDR electron beam dosimetry. Films with different grain sizes and CaF₂ concentrations were irradiated using an ElectronFlash linac, assessing pulse length dependence, field size, and dose linearity. The CaF₂:N films were explored for dosimetry application in conventional mode for comparison. Results demonstrated linear RL dose-response in both conventional and UHDR modes, with higher RL intensity for films containing 54% CaF₂ and larger grains. Field size analysis matched Gafchromic films, though signal attenuation occurred due to phantom surface artifacts. Despite energy dependence, CaF₂:N films exhibited spatial homogeneity and flexibility, and desirable dosimetric features, making them suitable for 2D dose mapping. With optimized acquisition systems and reduced luminescent lifetime, CaF₂:N films could advance UHDR-RT dosimetry by combining real time, high-resolution, and cost-effective dose monitoring.

Objective. The dielectric wall accelerator (DWA) is proposed as a low-cost, compact system for proton therapy. It offers additional degrees of freedom over conventional accelerators for controlling beam parameters. This study seeks to determine clinically desirable proton beam characteristics by evaluating their implications for treatment plan quality; particularly, the influence of variations in lateral spot size and beam energy spread were investigated. Approach. A linear optics model of the DWA, developed in TRANSOPTR , was used to define realistic beam parameters at the accelerator exit, from which an analytical method selected those yielding the desired spot sizes and energy spreads at isocenter. These parameters were subsequently passed through a generic nozzle model in TOPAS. Lateral spot size was varied from 1 to 10 mm. Beam energy spread was varied from 0.00005 % and 15 % . Robustly optimized treatment plans were generated in RayStation for a homogeneous water phantom and a sample pediatric central nervous system cancer case. Results were validated against conventional proton beam data. Main results. In the water phantom study, lower energy spreads decreased the normalized surface dose and increased the depth of maximum dose. In both phantom and patient cases, reducing the lateral spot size yielded greater organs-at-risk (OAR) sparing while maintaining comparable target coverage, which is in agreement with the literature. A similar trend of improved OAR sparing was observed with reduced beam energy spread for all cases; albeit, the effect was less pronounced. For realistic optimization of the DWA system, the beam parameters identified as most favorable were a lateral spot size of approximately 1 mm and an energy spread on the order of 0.05 % to 0.5% for 20 to 226 MeV. Significance. These results should prove useful to both hardware and numerical efforts, and form the basis for future studies of the clinical characteristics of prospective DWA machines.

Current dosimetry protocols typically recommend multiple measurements to determine recombination correction factors (ks), increasing the time required for dose measurements in the quality assurance workflow. We propose a novel dual-gap ionization chamber (DGIC) design for reference dosimetry featuring two air gaps of different thicknesses within a single device. This design enables the determination of ks directly from the same measurements required to determine absorbed dose-to-water. Thus, eliminating the need for separate measurements to correct for recombination losses. The approach relies on analyzing the charge ratio between the two gaps, which can also be correlated with the average dose rate under Ultra High Dose Rates (UHDR) conditions. A DGIC prototype with electrode distances of 1 and 0.6 mm was developed and tested using different beam qualities: (1) a 240 MeV/n clinical carbon ion beam at conventional dose rates, (2) a 226 MeV continuous proton beam with a current between 5 and 800 nA, where the maximum approximately corresponds to 200 Gy/s in the treatment room and (3) a 9 MeV electron beam with a DPP from 0.03 to 4.2 Gy. ks-factors were derived for the top cavity using the DGIC method and compared against the following: For proton and carbon ions, comparisons were made with the Jaffe plot method. For the electron beam, it was compared with a dose rate independent device, a flashDiamond detector. A DGIC prototype was able to successfully correct for recombination losses under different beam modalities: for initial recombination in a clinical carbon ion beam, volume recombination in UHDR proton beam with average dose rates of 200 Gy/s and in UHDPP electron beams tested up to a DPP of 4.2 Gy. A DGIC design and its inherent method provides a practical and accurate way of determining dose and dose rate in emerging radiotherapy treatment modalities.

Over 50% of all cancer patients receive radiotherapy during their treatment. The efficacy hereof is limited by normal tissue toxicity. Recent discovery of the FLASH effect showed that normal tissue toxicity can be drastically reduced, with conservation of the anti-tumoral effect, by delivering the dose at ultra-high dose rates (UHDR). This poses dosimetric challenges as most conventional dosimeters saturate. The aim of this PhD was to provide the tools for a dosimetric framework for UHDR electron radiotherapy. Working dosimeters for real-time and time-resolved point dosimetry, and for passive and time-resolved 2D dosimetry were investigated, focusing on average dose rate, dose per pulse (pulse durations < 4 µs), pulse repetition frequency (PRF), instantaneous dose rate and energy (in)dependence. Via collaborations, we showed the UHDR suitability of the flashDiamond, a SiC thin membrane dosimeter and an ultra-thin ionization chamber up to 20 Gy, 2 Gy and 10 Gy per pulse, respectively. A theoretical description of a gas-filled ionization chamber, reliable up to a dose per pulse of 40 Gy was presented. Also, a proof-of-principle study, investigating a novel method to determine the ion recombination correction factor with AI, was performed. Point scintillator dosimeters were manufactured and characterized for time-resolved UHDR point dosimetry. In-house made point scintillators had sufficient temporal resolution to resolve the pulse structure of the beam (PRF ≤ 245 Hz), but showed saturation effects at a dose per pulse > 1 Gy. The commercial Hyperscint-RP200 system showed no saturation and superior stem signal removal, but could only measure individual pulses for PRF < 12 Hz. An optically stimulated luminescence sheet was investigated as solution for passive 2D dosimetry for UHDR. The system showed suitable for dose per pulse up to at least 4.3 Gy and superior ease of use compared to radiochromic film. However, it was susceptible to physical damage, resulting in systematic errors and need for recalibration. Finally, a scintillating sheet-camera setup was investigated for time-resolved 2D UHDR dosimetry. It showed suitable for a dose per pulse up to at least 1.8 Gy, and had sufficient temporal resolution to measure the pulsed beam structure. The use of this setup was tested in a preclinical setting and found adequate. However, some of the sheets showed a dose dependence that needed correction. This work provides the basis to develop a dosimetric framework for UHDR electron beams. It presents potential tools for real-time and time-resolved point dosimetry, and for passive and time-resolved 2D dosimetry.

Background: Plane‐parallel ionization chambers are the recommended secondary standard systems for clinical reference dosimetry of electrons. Dosimetry in high dose rate and dose‐per‐pulse (DPP) is challenging as ionization chambers are subject to ion recombination, especially when dose rate and/or DPP is increased beyond the range of conventional radiotherapy. The lack of universally accepted models for correction of ion recombination in UDHR is still an issue as it is, especially in FLASH‐RT research, which is crucial in order to be able to accurately measure the dose for a wide range of dose rates and DPPs.Purpose: The objective of this study was to show the feasibility of developing an Artificial Intelligence model to predict the ion‐recombination factor—k_sat for a plane‐parallel Advanced Markus ionization chamber for conventional and ultra‐high dose rate electron beams based on machine parameters. In addition, the predicted k_sat of the AI model was compared with the current applied analytical models for this correction factor.Methods: A total number of 425 measurements was collected with a balanced variety in machine parameter settings. The specific k_sat values were determined by dividing the output of the reference dosimeter (optically stimulated luminescence [OSL]) by the output of the AM chamber. Subsequently, a XGBoost regression model was trained, which used the different machine parameters as input features and the corresponding k_sat value as output. The prediction accuracy of this regression model was characterized by R²‐coefficient of determination, mean absolute error and root mean squared error. In addition, the model was compared with the Two‐Voltage (TVA) method and empirical Petersson model for 19 different dose‐per‐pulse values ranging from conventional to UDHR regimes. The Akiake Information criterion (AIC) was calculated for the three different models.Results: The XGBoost regression model reached a R²‐score of 0.94 on the independent test set with a MAE of 0.067 and RMSE of 0.106. For the additional 19 random data points, the k_sat values predicted by the XGBoost model showed to be in agreement, within the uncertainties, with the ones determined by the Petersson model and better than the TVA method for doses per pulse >3.5 Gy with a maximum deviation from the ground truth of 14.2%, 16.7%, and −36.0%, respectively, for DPP >4 Gy.Conclusion: The proposed method of using AI for k_sat determination displays efficiency. For the investigated DPPs, the k_sat values obtained with the XGBoost model were in concurrence with the ones obtained with the current available analytical models within the boundaries of uncertainty, certainly for the DPP characterizing UDHR. But the overall performance of the AI model, taking the number of free parameters into account, lacked efficiency. Future research should optimize the determination of the experimental k_sat, and investigate the determination the k_sat for DPPs higher than the ones investigated in this study, while also evaluating the prediction of the proposed XGBoost model for UDHR machines of different centers.

FLASH-RT has proven beneficial in preclinical studies. However, the lack of accurate real time 2D dosimetry is a limiting factor. In this work, an innovative solution for 2D real time dosimetry in UHDR electron beams is presented. The in-house developed ImageDosis system consists of a scientific camera, with high temporal resolution, and a coating, containing 12% of Y$_3$Al$_5$O$_{12}$:Ce$^{3+}$ as scintillating material. Reference dosimetry was performed by means of radiochromic film, and a (C$_{38}$H$_{34}$P$_2$)MnBr$_4$ point scintillator was used to validate the pulse discrimination properties of the ImageDosis system. Irradiations were performed in two centers (Antwerp and Orsay), with an ElectronFlash accelerator, with varying number of pulses, pulse length, pulse repetition frequency (PRF) and energy. Also, the temporal resolution and 2D properties were investigated. For doses > 3.5 Gy, the ImageDosis showed a linear dose response up to at least 13 Gy. No dose rate dependence was found for an average dose rate up to 140 Gy/s, a dose per pulse up to 2 Gy and a PRF up to 300 Hz. The ImageDosis system showed capable of measuring the dose of the individual pulses up to a PRF of 250 Hz, but did not detect 3% of the pulses, because these pulses were delivered during the dead time of the camera. The maximal difference in FWHM of the field size between the ImageDosis system and the reference was 3.6%. For a nominal field size of 100 mm and 120 mm, a decreased output was observed on the superior part of the field. The ImageDosis system showed linear dose response and no dose rate nor energy dependence. It showed capable of discriminating and measuring the dose of individual pulses and promising 2D characteristics that need further optimization.