Small beams, fast predictions: a comparison of machine learning dose prediction models for proton minibeam therapy.

BACKGROUND: Dose calculations for novel radiotherapy cancer treatments such as proton minibeam radiation therapy is often done using full Monte Carlo (MC) simulations. As MC simulations can be very time consuming for this kind of application, deep learning models have been considered to accelerate dose estimation in cancer patients. PURPOSE: This work systematically evaluates the dose prediction accuracy, speed and generalization performance of three selected state-of-the-art deep learning models for dose prediction applied to the proton minibeam therapy. The strengths and weaknesses of those models are thoroughly investigated, helping other researchers to decide on a viable algorithm for their own application. METHODS: The following recently published models are compared: first, a 3D U-Net model trained as a regression network, second, a 3D U-Net trained as a generator of a generative adversarial network (GAN) and third, a dose transformer model which interprets the dose prediction as a sequence translation task. These models are trained to emulate the result of MC simulations. The dose depositions of a proton minibeam with a diameter of 800µm and an energy of 20-100 MeV inside a simple head phantom calculated by full Geant4 MC simulations are used as a case study for this comparison. The spatial resolution is 0.5 mm. Special attention is put on the evaluation of the generalization performance of the investigated models. RESULTS: Dose predictions with all models are produced in the order of a second on a GPU, the 3D U-Net models being fastest with an average of 130 ms. An investigated 3D U-Net regression model is found to show the strongest performance with overall 61.0 % ± $\%\pm$ 0.5% of all voxels exhibiting a deviation in energy deposition prediction of less than 3% compared to full MC simulations with no spatial deviation allowed. The 3D U-Net models are observed to show better generalization performance for target geometry variations, while the transformer-based model shows better generalization with regard to the proton energy. CONCLUSIONS: This paper reveals that (1) all studied deep learning models are significantly faster than non-machine learning approaches predicting the dose in the order of seconds compared to hours for MC, (2) all models provide reasonable accuracy, and (3) the regression-trained 3D U-Net provides the most accurate predictions.

The radiosensitizing effect of platinum nanoparticles in proton irradiations is not caused by an enhanced proton energy deposition at the macroscopic scale.

Objective.Due to the radiosensitizing effect of biocompatible noble metal nanoparticles (NPs), their administration is considered to potentially increase tumor control in radiotherapy. The underlying physical, chemical and biological mechanisms of the NPs' radiosensitivity especially when interacting with proton radiation is not conclusive. In the following work, the energy deposition of protons in matter containing platinum nanoparticles (PtNPs) is experimentally investigated.Approach.Surfactant-free monomodal PtNPs with a mean diameter of (40 ± 10) nm and a concentration of 300 µg ml-1, demonstrably leading to a substantial production of reactive oxygen species (ROS), were homogeneously dispersed into cubic gelatin samples serving as tissue-like phantoms. Gelatin samples without PtNPs were used as control. The samples' dimensions and contrast of the PtNPs were verified in a clinical computed tomography scanner. Fields from a clinical proton machine were used for depth dose and stopping power measurements downstream of both samples types. These experiments were performed with a variety of detectors at a pencil beam scanning beam line as well as a passive beam line with proton energies from about 56-200 MeV.Main results.The samples' water equivalent ratios in terms of proton stopping as well as the mean proton energy deposition downstream of the samples with ROS-producing PtNPs compared to the samples without PtNPs showed no differences within the experimental uncertainties of about 2%.Significance.This study serves as experimental proof that the radiosensitizing effect of biocompatible PtNPs is not due to a macroscopically increased proton energy deposition, but is more likely caused by a catalytic effect of the PtNPs. Thus, these experiments provide a contribution to the highly discussed radiobiological question of the proton therapy efficiency with noble metal NPs and facilitate initial evidence that the dose calculation in treatment planning is straightforward and not affected by the presence of sensitizing PtNPs.

No more glowing in the dark: how deep learning improves exposure date estimation in thermoluminescence dosimetry.

The time- or temperature-resolved detector signal from a thermoluminescence dosimeter can reveal additional information about circumstances of an exposure to ionising irradiation. We present studies using deep neural networks to estimate the date of a single irradiation with 12 mSv within a monitoring interval of 42 days from glow curves of novel TL-DOS personal dosimeters developed by the Materialprüfungsamt NRW in cooperation with TU Dortmund University. Using a deep convolutional network, the irradiation date can be predicted from raw time-resolved glow curve data with an uncertainty of roughly 1-2 days on a 68% confidence level without the need for a prior transformation into temperature space and a subsequent glow curve deconvolution (GCD). This corresponds to a significant improvement in prediction accuracy compared to a prior publication, which yielded a prediction uncertainty of 2-4 days using features obtained from a GCD as input to a neural network.

Impact of air gap, range shifter, and delivery technique on skin dose in proton therapy.

OBJECTIVE: Side effects of radiation therapy may include skin damage. The surface dose is of great interest and contains the buildup effect. In particular, the proton therapy community requires further experimental data to quantify doses in the surface region. This specification includes the skin dose, which is defined according to ICRU Report No. 39 at 70 µm water equivalent depth. The aim of this study is to gather more knowledge of the skin dose by varying key parameters defined by the patient treatment plan. This consists of clinical aspects such as the influence of the air gap, the application of a range shifter (RS), or the proton delivery technique. MATERIAL/METHODS: Skin doses were determined with a PTW 23391 extrapolation chamber with three thin Kapton® entrance windows operated as a conventional ionization chamber. The impact on the skin dose for quasi-monoenergetic pencil beam scanning (PBS) proton beams was evaluated for clinical air gaps between 3.5 and 51.1 cm. The differences in skin dose were assessed by irradiating equivalent fields with an RS of 51 mm water equivalent thickness (RS51) and without. Furthermore, the delivery techniques PBS, uniform scanning (US), and double scattering (DS) were compared by defining a spread-out Bragg peak (SOBP). TOPAS (V.3.1.2) was used to model an IBA nozzle with PBS and to score dose to water at the surface of a water phantom. RESULTS: For the monoenergetic fields without the application of the RS the skin dose was constant down to an air gap of 6.2 cm. A lower air gap of 3.5 cm showed a variation in skin dose by up to 2.4% compared to the results obtained with larger air gaps. With the inserted RS51 an increase in the skin dose was found for air gaps smaller than 11.3 cm. Experimentally, a dose difference of 1.4% was recorded for an air gap of 6.2 cm by inserting an RS and none. With the Monte Carlo calculations the largest dose increase was observed at the air gap of 3.5 cm with 1.7% and 4.0% relative to the skin dose results without the RS and to the largest evaluated air gap of 51.1 cm, respectively. The SOBP comparison of the beam modalities at the measuring plane at the isocenter revealed higher skin doses without RS (including RS) by up to +1.9% (+1.5%) for DS and +1.3% (+1.1%) for US compared to PBS. For all three techniques an approx. 2% rise in skin dose was observed for the largest evaluated air gap of 37.7 cm to an air gap of 6.2 cm when using an RS51. CONCLUSION: The study investigated aspects of skin dose of a water equivalent phantom by varying key parameters of a proton treatment plan. Parameters like the RS, the air gap, and the delivery modality have an impact on the order of 4.0% for the skin dose at the depth of 70 µm. The increases in skin dose are the effects of the contribution of the increased electron fluence at small air gaps and the emitted hadronic particles produced by the RS.

Determination of surface dose in pencil beam scanning proton therapy.

PURPOSE/OBJECTIVE: Quantification of surface dose within the first few hundred water equivalent µm is challenging. Nevertheless, it is of large interest for the proton therapy community to study dose effects in the skin. The experimental determination is affected by the detector properties, such as the detector volume and material. The International Commission on Radiation Units and Measurements in its report 39 recommends assessing the skin dose at a depth of 0.07 mm. The aim of this study is the estimation of the absorbed dose at and around a depth of 70 µm. We used various dosimetric approaches in conjunction with proton pencil beam scanning delivery to determine the skin dose in a clinical setting. MATERIAL/METHODS: Five different detectors were tested for determining the surface dose in water: EBT3 and HD-V2 GAFCHROMIC™ radiochromic film, LiF:Mg,Ti thermoluminescent dosimeter, IBA PPC05 plane-parallel ionization chamber, and PTW 23391 extrapolation chamber. The irradiation setup consisted of quasi-monoenergetic scanned proton pencil beams with kinetic energies of 100, 150, and 226.7 MeV, respectively. Radiochromic films were placed within a vertical stack and in wedge geometry and were analyzed with FilmQA Pro™ adopting triple channel dosimetry. The extrapolation chamber PTW 23391, which served as a reference in the current work, was used in a conventional ionization chamber setup with a fixed electrode gap of 2 mm. Three Kapton® entrance windows with thicknesses of 25, 50, and 75 µm were employed. Thermoluminescent dosimeters were provided as powder and were pressed onto a sheet of aluminum. Furthermore, the Monte Carlo code TOol for PArticle Simulation (TOPAS) in version 3.1.p2 was used to model an IBA pencil beam scanning nozzle and score dose to water in a water phantom. RESULTS: The resulting depth dose curves were normalized to their 100% dose at the reference depth of 3 cm. We obtained the skin doses with the extrapolation chamber and with TOPAS. For the experimental approach this resulted in 79.7 ± 0.3%, 86.0 ± 0.6%, and 87.1 ± 0.1% for the proton energies 100, 150, and 226.7 MeV, respectively. The results for TOPAS were 80.1 ± 0.2% (100 MeV), 87.1 ± 0.5% (150 MeV), and 86.9 ± 0.4% (226.7 MeV), respectively. Based on the experimental results of the skin dose, we provided a clinically relevant surface extrapolation factor for the common measurement methods. This allows the result of the first measurement depth of a detector to be scaled to the dose at the skin depth. Most practical would be the use of the surface extrapolation factor for the PPC05 chamber, due to its direct reading, the wide availability in clinics and the low uncertainties. The calculated factors were 0.986 ± 0.004 for 100 MeV, 0.961 ± 0.008 for 150 MeV, and 0.963 ± 0.003 for 226.7 MeV. CONCLUSIONS: In this study, dissimilar experimental approaches were evaluated with respect to measurements at depths close to the surface. The experimental depth dose curves are in good agreement with the simulation with TOPAS Monte Carlo. To the author's knowledge this was the first experimental determination of the skin dose according to the International Commission on Radiation Units and Measurements 39 definition in proton pencil beam scanning.

A NEW TL-DOS NEUTRON DOSEMETER FOR MEASUREMENTS OF THE PERSONAL DOSE EQUIVALENT HP(10).

A new thermoluminescence albedo dosemeter with improved properties is developed as part of the TL-DOS project. The dosemeter measures the neutron and photon whole-body dose of radiation workers. The TL-DOS neutron dosemeter is presented and its results of well-known field measurements as well as field calibrations are shown. Its advantages, such as its potential to measure the high-linear energy transfer peaks, its improved detector sensitivity and long detector lifetime, are explained. The new dosemeter is compared to a thermoluminescence albedo dosemeter already used in routine dosimetry in Germany.