Katia Parodi

Objective. Prompt gamma (PG) detection enables beam range verification and margin reduction in proton therapy, but its effectiveness is limited by counting statistics. Spot boosting has been proposed to enhance the signal-to-noise ratio by increasing the number of protons in selected spots of a pencil beam scanning treatment plan. This work integrates PG monitoring into an multi-criteria optimization (MCO) framework to investigate trade-offs between proton counts in boosted spots and dosimetric metrics. Approach. Intensity modulated proton therapy plans for a prostate-tumor and a brain-tumor patient were optimized in OpenTPS. Dose, LET weighted dose (LETxdose) and PG distributions were simulated for each pencil beam using the fast Monte Carlo (MC) engine MCSquare. For each field, ten candidate spots were selected based on three metrics quantifying the quality of PG monitoring and an even spatial distribution across the clinical target volume. Initial pencil beam weights and corresponding dose and LETxdose distributions were imported into an MCO framework for re-optimization, simultaneously considering both dosimetric goals and boosting objectives. Main results. Boosting up to 5 ⋅ 10 8 protons/spot maintained the plan quality without violating constraints in the prostate case. In the brain case, spot boosting required moderate deviations from the initial target dose constraints. For both cases, only minor boosting gains were obtained by relaxing dose metrics for organs at risk. Significance. This study is the first to visualize spot boosting trade-offs via Pareto fronts to support an informed decision, integrating verification quality aspects into the treatment planning stage. The approach enables selection of plans that preserve dose and LETxdose objectives while enhancing PG signal in targeted spots for reliable in-vivo range verification. The framework is broadly applicable and can provide a strategy for embedding verification-driven objectives into treatment optimization beyond PG monitoring.

Objective. Range uncertainties in proton therapy largely arise from the semi-empirical conversion of x-ray computed tomography (CT) Hounsfield Units to stopping power relative to water (RSP), requiring generous safety margins or robust optimization. Pencil beam-wise integrated depth-dose profiles (IDDs), obtained via proton transmission imaging using a particle integrating detector, can be combined with the treatment planning CT to obtain patient-specific conversion curves. Existing calibration approaches rely on projection images or water-equivalent path length (WEPL) histograms derived from IDDs via signal decomposition or deconvolution. A direct optimization approach based on the acquired IDD signal is proposed, avoiding the need for decomposition or deconvolution and their associated WEPL accuracy and resolution limitations. Approach. The RSP conversion curve is iteratively optimized using Monte Carlo-simulated IDDs acquired from two orthogonal projection angles in a line-scan proton imaging setup. The differentiable implementation of the forward projection operator is embedded in a gradient-based optimization scheme. Main results. The proposed IDD-based optimization approach significantly reduces the calibration error compared to the initial guess. The median mean absolute percentage error on a set of axial CT slices decreases from 1.88% to 0.58%, with a median computation time of 5.47 h per slice. The optimized calibration improves dose accuracy for an exemplary treatment pencil beam. Increasing the spot spacing by a factor of ten decreases the computation time to 0.60 h with minimal loss of accuracy. Using a single projection angle yields comparable results to using both projections (0.64% vs 0.57%). With current detector depth resolutions of ∼ 2 mm, calibration accuracy of ∼ 0.7% is achievable, with potential improvements using finer-resolution detectors. Significance. Particle integrating detectors offer potential for clinical translation due to lower cost and complexity. Direct optimization on IDDs using a differentiable forward operator enables accurate, fully automated calibration, improving upon earlier manual IDD-based or automated WEPL-based approaches.

Objective. The design and base-validation of a small-animal microwave hyperthermia applicator for the adaptation of the clinically used BSD-500 hyperthermia device to function with mice. Approach. We iteratively designed and tested a 3D-printable applicator, using mouse cadavers as well as the life sciences simulation software Sim4life to optimise the applicator. The applicator uses an agarose phantom to couple the antenna to the tissue, and we determined the dielectric and thermal properties thereof. To monitor the temperatures in the cadavers during the experiments, electromagnetically stable 4-wire sensors from the clinical device were read out independently. Main Results. The agarose coupling phantom was characterised dielectrically and thermally and closely approximated muscle tissue ( ϵ r = 51.91 , σ = 0.99 S m − 1 at 915 MHz) and c p = ( 2592 ± 40 ) J kg − 1 K − 1 . The independent sensor readout achieved 3 Hz data collection with a sensor standard deviation from the reference thermometer smaller than 0.01 ∘ C . In each of the validation experiments visual positioning was by itself sufficient to ensure the tumour-surrogate reached therapeutic temperatures (40 ∘ C – 43 ∘ C ) using only a control sensor in the agarose gel in contact with the mouse body. Extrapolating from our data, the highest temperature should occur in the gel at the antenna tip, reaching approximately 43 ∘ C . The therapeutic zone extends to a depth of 9.3 mm–17.6 mm (95% CI). The electromagnetic simulation predicted the heating-rate distribution well and was robust to changes in material properties and problem geometry. Perfusion-aware simulations based on the Pennes bioheat equation predicted at most an ≈ 21% reduction in tumour heating under physiological conditions, indicating that therapeutic temperatures should remain achievable in vivo . Significance. This work substantially narrows the gap towards an accessible, easy-to-use small-animal hyperthermia applicator and provides the base-validation needed before progressing to in vivo experiments.

Objective. A low-energy x-rays camera–based system has been introduced to image secondary electron bremsstrahlung x-rays, enabling external visualization of proton beam profiles during irradiation. To date, this approach has been demonstrated only in phantom studies and simulations, and has not yet been successfully applied in patients. To overcome this limitation, it is essential to establish the feasibility of SEB x-ray imaging in a clinical condition. Approach. We first optimized the energy window and threshold settings for range estimation by acquiring SEB x-ray images of a water phantom comparable in size to the human lower abdomen, using a cerium doped yttrium aluminum perovskite (YAP(Ce)) x-ray camera with a pinhole collimator. We then performed the first in vivo SEB x-ray measurement by imaging proton beams during treatment of a patient of prostate cancer with two opposed fields, each delivering an RBE-weighted dose of 1.5 Gy. Main results. In the patient study, SEB x-ray distributions were observed in the measured images using the same energy window as that established in the water phantom study. From these images, the beam ranges were estimated to be within approximately 1 cm of the values calculated by the treatment planning system using the optimized profile threshold. Dynamic imaging at 1 s time intervals was achieved, enabling temporal evaluation of count rate curves. Significance. We present, for the first time, measurements of SEB x-ray images acquired in a patient during proton beam irradiation. These preliminary results indicate that in vivo SEB x-ray imaging could be a promising approach for clinical real-time beam monitoring in proton therapy. Ethics approval. Ethics approval was obtained from the institutional review board of Kobe Proton Center: approval number: 06-10.

Treatment adaptation is particularly critical in particle therapy, where even small range deviations can compromise target coverage or lead to unintended dose delivered to surrounding healthy tissues. In-beam positron emission tomography (PET) has emerged as a promising approach for range verification during irradiation with protons or stable carbon ion beams. However, its clinical use is limited by low signal-to-noise ratio and by the spatial mismatch between activity and dose distributions, reducing verification accuracy and limiting timely intervention. We used radioactive ion beams for real-time range adaptation in 10 weeks old C3H/ HeNRj female mice bearing LM8-osteosarcoma tumors. Three ¹¹C beam range settings were planned: short (S), right (R), and long (L). The range of a collimated monoenergetic probing beam was monitored in real-time with the SIRMIO in-beam PET scanner by tracking the activity peak along the beam path, while range adaptation was achieved with a remotely controlled range shifter. Each plan (S, R and L) was also delivered, and tumor growth, toxicity assays, and histological analyses were performed to evaluate each treatment outcomes. Dynamic repositioning of the ¹¹C beam produced spatially resolved PET signals that correlated with distinct biological outcomes. Toxicity was observed only in the L group, while adequate tumor coverage was achieved in both R and L groups. In contrast, the S group showed continued tumor growth. We provide the first demonstration that in-beam imaging of radioactive ion beams can enable real-time range-guided radiotherapy in a living organism. These findings establish radioactive ion beams as a promising platform for precision range-guided particle therapy.

Objective. As carbon ion therapy is highly sensitive to range uncertainties, it can benefit greatly from range verification. Positron emission tomography (PET) provides a valuable approach, in which predicted positron emitter distributions (PED) can be used to calculate the expected activity distribution, which is then compared with the irradiation-induced PET signal. In a previous work, we developed an analytical approach to predict 3D positron emitter distributions from dose distributions, providing a faster alternative to Monte Carlo (MC) simulations at a comparable accuracy, and with potential for integration into analytical treatment planning systems (TPS). The purpose of this work is to validate this analytical approach using real clinical cases where offline PET/Computed tomography (CT) monitoring was employed. Approach. Four carbon ion therapy patients treated at the Heidelberg Ion Beam Therapy Center (HIT) were selected, and their treatment plans and CT images were used for MC simulations and analytical prediction of β +activity distributions. The analytically predicted activity distributions, derived from the simulated dose with our analytical approach, were then compared with both simulated results and measured offline PET data. M ain results. The analytical and MC activity distributions demonstrated a good match in range with mean deviations less than 0.5 mm, and in amplitude with mean normalized rootmean-square error (NRMSE) less than 2%. Range shifts between the measured PET signals and the analytical activity patterns were evaluated and found to be consistent with published results. Signif icance. The obtained results demonstrate the capability of our analytical approach to predict PET images for range verification in carbon ion therapy under real clinical scenarios, offering faster predictions than MC simulations while maintaining comparable accuracy. The solution proposed also offers the possibility of a straightforward integration into TPS by leveraging the commonly used pencil beam algorithms (PBAs) present in analytical carbon ion dose engines.

Objective. In proton therapy, dual-energy computed tomography (DECT) has shown to improve proton range estimation and treatment planning, but it is not yet widely implemented clinically as technology differences may restrict application (e.g. field-of-view or temporal separation). Recently, photon-counting CT (PCCT) was introduced clinically. PCCT offers multi-energy CT imaging in a single scan and with low imaging dose options. Therefore, this study aimed to evaluate proton range differences between PCCT and DECT across varying anthropomorphic phantom sizes, tissue types, and imaging doses. Approach. Virtual monoenergetic images (70 and 180 keV) were generated from DECT (80/140 kVp) and PCCT (120 kVp) scans at three imaging doses (CTDI _vol,32cm : 5, 10, and 20 mGy) using a commercial head phantom and a 3D-printed abdomen phantom with exchangeable fat rings for varying dimensions. Proton therapy plans were created in RayStation (RaySearch laboratories) with varying proton beam configurations (e.g. crossing different tissues) to assess dependency of phantom size, tissue type, and proton energy. The proton range, R ₈₀ , was extracted for each plan, and range differences were evaluated both between imaging doses and between modalities (DECT vs PCCT) at 20 mGy (inter-modality). Main results. Under reduced imaging dose, DECT showed substantial proton range differences for the large phantom sizes. It resulted in median range differences up to 20.1 mm due to visible imaging artifacts. Similar trends were seen for long proton path lengths. In contrast, for PCCT, range deviations stayed below 0.5 mm across all phantom sizes and through all tissue types. Inter-modality range variations increased when protons traversed denser tissues or longer path lengths. Significance. This work showed consistent R ₈₀ estimations for PCCT across phantom sizes and traversing different tissue types, even under reduced imaging dose in PCCT. This offers possibilities to lower the imaging dose and thus cumulative patient dose, without compromising treatment accuracy.