Deep TL: progress of a machine learning aided personal dose monitoring system.

Personal dosemeters using thermoluminescence detectors can provide information about the irradiation event beyond the pure dose estimation, which is valuable for improving radiation protection measures. In the presented study, the glow curves of the novel TL-DOS dosemeters developed by the Materialprüfungsamt NRW in cooperation with the TU Dortmund University are analysed using deep learning approaches to predict the irradiation date of a single-dose irradiation of 10 mGy within a monitoring interval of 41 d. In contrast of previous work, the glow curves are measured using the current routine read-out process by pre-heating the detectors before the read-out. The irradiation dates are predicted with an accuracy of 2-5 d by the deep learning algorithm. Furthermore, the importance of the input features is evaluated using Shapley values to increase the interpretability of the neural network.

Accurate and Fast Deep Learning Dose Prediction for a Preclinical Microbeam Radiation Therapy Study Using Low-Statistics Monte Carlo Simulations.

Microbeam radiation therapy (MRT) utilizes coplanar synchrotron radiation beamlets and is a proposed treatment approach for several tumor diagnoses that currently have poor clinical treatment outcomes, such as gliosarcomas. Monte Carlo (MC) simulations are one of the most used methods at the Imaging and Medical Beamline, Australian Synchrotron to calculate the dose in MRT preclinical studies. The steep dose gradients associated with the 50µm-wide coplanar beamlets present a significant challenge for precise MC simulation of the dose deposition of an MRT irradiation treatment field in a short time frame. The long computation times inhibit the ability to perform dose optimization in treatment planning or apply online image-adaptive radiotherapy techniques to MRT. Much research has been conducted on fast dose estimation methods for clinically available treatments. However, such methods, including GPU Monte Carlo implementations and machine learning (ML) models, are unavailable for novel and emerging cancer radiotherapy options such as MRT. In this work, the successful application of a fast and accurate ML dose prediction model for a preclinical MRT rodent study is presented for the first time. The ML model predicts the peak doses in the path of the microbeams and the valley doses between them, delivered to the tumor target in rat patients. A CT imaging dataset is used to generate digital phantoms for each patient. Augmented variations of the digital phantoms are used to simulate with Geant4 the energy depositions of an MRT beam inside the phantoms with 15% (high-noise) and 2% (low-noise) statistical uncertainty. The high-noise MC simulation data are used to train the ML model to predict the energy depositions in the digital phantoms. The low-noise MC simulations data are used to test the predictive power of the ML model. The predictions of the ML model show an agreement within 3% with low-noise MC simulations for at least 77.6% of all predicted voxels (at least 95.9% of voxels containing tumor) in the case of the valley dose prediction and for at least 93.9% of all predicted voxels (100.0% of voxels containing tumor) in the case of the peak dose prediction. The successful use of high-noise MC simulations for the training, which are much faster to produce, accelerates the production of the training data of the ML model and encourages transfer of the ML model to different treatment modalities for other future applications in novel radiation cancer therapies.

Evaluation of a New Inverse, Globally Convex Treatment Planning System Algorithm for Gamma Knife Radiation Surgery Within a Prospective Trial: Advantages and Disadvantages in Practical Application.

Purpose: A new inverse planning software called IntuitivePlan (IP) based on a global convex optimization algorithm was adopted for the Gamma Knife radiation surgery. We investigated IP's suitability for daily clinical use and its applicability for different cerebral entities. Methods and Materials: For 230 target volumes, IP was tested in a prospective trial. The computed treatment plans were compared with conventional expert preplans, which included forward planning by the expert and local internal optimization. Based on the same dose constraints, we used the default settings for the inverse calculation of the treatment plans. Plan quality metrics such as the Paddick conformity index were compared for both planning techniques with additional subdivisions into the 3 selectable IP planning strategies and different entity groups. Results: IP calculated treatment plans of quality similar to that of preplans created by expert planners. Some plan quality metrics, especially those related to conformity and dose gradient, attained statistically significantly higher scores combined with high coverage for the inversely generated plans except for the selectivity optimizing strategy. Normal brain volume receiving 10 Gy or 12 Gy or higher (V 1 0 Gy or V 1 2 Gy ) did not show significant differences for the coverage optimizing strategies. The IP software demonstrated significantly shorter planning times versus manual planning as well as greater numbers of isocenters, often associated with longer treatment times. In terms of total time, these differences almost balanced out again. Conclusions: Our results suggest that IP is advantageous for complex tumors. We observed general clinical significance for conformity and superiority for the selectivity optimizing strategy. In addition, the high-quality calculation from IP enables novices in the profession to achieve pre-treatment plans of a quality similar to that of expert planners. IP allows for optimizing the sparing of surrounding tissue and conformity for benign tumors within a short time. Thus, IP forms a solid basis for further planning on the treatment day.

Fast and accurate dose predictions for novel radiotherapy treatments in heterogeneous phantoms using conditional 3D-UNet generative adversarial networks.

PURPOSE: Novel radiotherapy techniques like synchrotron X-ray microbeam radiation therapy (MRT) require fast dose distribution predictions that are accurate at the sub-mm level, especially close to tissue/bone/air interfaces. Monte Carlo (MC) physics simulations are recognized to be one of the most accurate tools to predict the dose delivered in a target tissue but can be very time consuming and therefore prohibitive for treatment planning. Faster dose prediction algorithms are usually developed for clinically deployed treatments only. In this work, we explore a new approach for fast and accurate dose estimations suitable for novel treatments using digital phantoms used in preclinical development and modern machine learning techniques. We develop a generative adversarial network (GAN) model, which is able to emulate the equivalent Geant4 MC simulation with adequate accuracy and use it to predict the radiation dose delivered by a broad synchrotron beam to various phantoms. METHODS: The energy depositions used for the training of the GAN are obtained using full Geant4 MC simulations of a synchrotron radiation broad beam passing through the phantoms. The energy deposition is scored and predicted in voxel matrices of size 140 × 18 × 18 with a voxel edge length of 1 mm. The GAN model consists of two competing 3D convolutional neural networks, which are conditioned on the photon beam and phantom properties. The generator network has a U-Net structure and is designed to predict the energy depositions of the photon beam inside three phantoms of variable geometry with increasing complexity. The critic network is a relatively simple convolutional network, which is trained to distinguish energy depositions predicted by the generator from the ones obtained with the full MC simulation. RESULTS: The energy deposition predictions inside all phantom geometries under investigation show deviations of less than 3% of the maximum deposited energy from the simulation for roughly 99% of the voxels in the field of the beam. Inside the most realistic phantom, a simple pediatric head, the model predictions deviate by less than 1% of the maximal energy deposition from the simulations in more than 96% of the in-field voxels. For all three phantoms, the model generalizes the energy deposition predictions well to phantom geometries, which have not been used for training the model but are interpolations of the training data in multiple dimensions. The computing time for a single prediction is reduced from several hundred hours using Geant4 simulation to less than a second using the GAN model. CONCLUSIONS: The proposed GAN model predicts dose distributions inside unknown phantoms with only small deviations from the full MC simulation with computations times of less than a second. It demonstrates good interpolation ability to unseen but similar phantom geometries and is flexible enough to be trained on data with different radiation scenarios without the need for optimization of the model parameter. This proof-of-concept encourages to apply and further develop the model for the use in MRT treatment planning, which requires fast and accurate predictions with sub-mm resolutions.

Enhancement of Proton Therapy Efficiency by Noble Metal Nanoparticles Is Driven by the Number and Chemical Activity of Surface Atoms.

Proton-based radiotherapy is a modern technique for the treatment of solid tumors with significantly reduced side effects to adjacent tissues. Biocompatible nanoparticles (NPs) with high atomic numbers are known to serve as sensitizers and to enhance treatment efficacy, which is commonly believed to be attributed to the generation of reactive oxygen species (ROS). However, little systematic knowledge is available on how either physical effects due to secondary electron generation or the particle surface chemistry affect ROS production. Thereto, ligand-free colloidal platinum (Pt) and gold (Au) NPs with well-controlled particle size distributions and defined total surface area are proton-irradiated. A fluorescence-based assay is developed to monitor the formation of ROS using terephthalic acid as a cross-effect-free dye. The findings indicate that proton irradiation (PI)-induced ROS formation sensitized by noble metal NPs is driven by the total available particle surface area rather than particle size or mass. Furthermore, a distinctive material effect with Pt being more active than Au is observed which clearly indicates that the chemical reactivity of the NP surface is a main contributor to ROS generation upon PI. These results pave the way towards an in-depth understanding of the NP-induced sensitizing effects upon PI and hence a well-controlled enhanced therapy.

Clinical Implementation of Proton Therapy Using Pencil-Beam Scanning Delivery Combined With Static Apertures.

Proton therapy makes use of the favorable depth-dose distribution with its characteristic Bragg peak to spare normal tissue distal of the target volume. A steep dose gradient would be desired in lateral dimensions, too. The widespread spot scanning delivery technique is based, however, on pencil-beams with in-air spot full-widths-at-half-maximum of typically 1 cm or more. This hampers the sparing of organs-at-risk if small-scale structures adjacent to the target volume are concerned. The trimming of spot scanning fields with collimating apertures constitutes a simple measure to increase the transversal dose gradient. The current study describes the clinical implementation of brass apertures in conjunction with the pencil-beam scanning delivery mode at a horizontal, clinical treatment head based on commercial hardware and software components. Furthermore, clinical cases, which comprised craniopharyngiomas, re-irradiations and ocular tumors, were evaluated. The dosimetric benefits of 31 treatment plans using apertures were compared to the corresponding plans without aperture. Furthermore, an overview of the radiation protection aspects is given. Regarding the results, robust optimization considering range and setup uncertainties was combined with apertures. The treatment plan optimizations followed a single-field uniform dose or a restricted multi-field optimization approach. Robustness evaluation was expanded to account for possible deviations of the center of the pencil-beam delivery and the mechanical center of the aperture holder. Supplementary apertures improved the conformity index on average by 15.3%. The volume of the dose gradient surrounding the PTV (evaluated between 80 and 20% dose levels) was decreased on average by 17.6%. The mean dose of the hippocampi could be reduced on average by 2.9 GyRBE. In particular cases the apertures facilitated a sparing of an organ-at-risk, e.g. the eye lens or the brainstem. For six craniopharyngioma cases the inclusion of apertures led to a reduction of the mean dose of 1.5 GyRBE (13%) for the brain and 3.1 GyRBE (16%) for the hippocampi.

Evaluation of the activation of brass apertures in proton therapy using gamma-ray spectrometry and Monte Carlo simulations.

Collimating apertures are used in proton therapy to laterally conform treatment fields to the target volume. While this is a standard technique in passive spreading treatment heads, patient-specific apertures can supplement pencil-beam scanning (PBS) techniques to sharpen the lateral dose fall-off. A radiation protection issue is that proton-induced nuclear reactions can lead to the formation of radionuclides in the apertures. In the experiments of the current study, cylindrical, thick brass targets were irradiated with quasi-monoenergetic proton fields of 100.0 MeV and of 226.7 MeV in PBS mode. The radioactivation of these two brass samples was characterised with a low-level gamma-ray spectrometer. The activation products were scored in a Monte Carlo simulation, too, and compared with the experimental activities. For the high-energy field, 63Zn, 60Cu, and 61Cu were the most important short-lived isotopes regarding the measured specific activity. After irradiation with the 100.0 MeV field, 62Cu, 63Zn, and 60Cu had the highest activity. Regarding long-lived isotopes, which determine the storage time of the used apertures, the isotopes 57Co, 65Zn, 54Mn, 56Co had the largest contribution to the activity. The relative difference of activities between simulation and experiment was typically between 10%-20% for short-lived nuclides and were up to a factor of five larger for long-lived nuclides. Summarising experiments and simulations for both incident proton energies, 62Cu was the most important detected residual nucleus regardless if specific activity or equivalent dose is considered.

Small field output correction factors of the microSilicon detector and a deeper understanding of their origin by quantifying perturbation factors.

PURPOSE: The aim of this study is the experimental and Monte Carlo-based determination of small field correction factors for the unshielded silicon detector microSilicon for a standard linear accelerator as well as the Cyberknife System. In addition, a detailed Monte Carlo analysis has been performed by modifying the detector models stepwise to study the influences of the detector's components. METHODS: Small field output correction factors have been determined for the new unshielded silicon diode detector, microSilicon (type 60023, PTW Freiburg, Germany) as well as for the predecessors Diode E (type 60017, PTW Freiburg, Germany) and Diode SRS (type 60018, PTW Freiburg, Germany) for a Varian TrueBeam linear accelerator at 6 MV and a Cyberknife system. For the experimental determination, an Exradin W1 scintillation detector (Standard Imaging, Middleton, USA) has been used as reference. The Monte Carlo simulations have been performed with EGSnrc and phase space files from IAEA as well as detector models according to manufacturer blueprints. To investigate the influence of the detector's components, the detector models have been modified stepwise. RESULTS: The correction factors for the smallest field size investigated at the TrueBeam linear accelerator (equivalent dosimetric square field side length Sclin  = 6.3 mm) are 0.983 and 0.939 for the microSilicon and Diode E, respectively. At the Cyberknife system, the correction factors of the microSilicon are 0.967 at the smallest 5-mm collimator compared to 0.928 for the Diode SRS. Monte Carlo simulations show comparable results from the measurements and literature. CONCLUSION: The microSilicon (type 60023) detector requires less correction than its predecessors, Diode E (type 60017) and Diode SRS (type 60018). The detector housing has been demonstrated to cause the largest perturbation, mainly due to the enhanced density of the epoxy encapsulation surrounding the silicon chip. This density has been rendered more water equivalent in case of the microSilicon detector to minimize the associated perturbation. The sensitive volume itself has been shown not to cause observable field size-dependent perturbation except for the volume-averaging effect, where the slightly larger diameter of the sensitive volume of the microSilicon (1.5 mm) is still small at the smallest field size investigated with corrections <2%. The new microSilicon fulfils the 5% correction limit recommended by the TRS 483 for output factor measurements at all conditions investigated in this work.

Contrast-enhanced spectral mammography with a compact synchrotron source.

For early breast cancer detection, mammography is nowadays the commonly used standard imaging approach, offering a valuable clinical tool for visualization of suspicious findings like microcalcifications and tumors within the breast. However, due to the superposition of anatomical structures, the sensitivity of mammography screening is limited. Within the last couple of years, the implementation of contrast-enhanced spectral mammography (CESM) based on K-edge subtraction (KES) imaging helped to improve the identification and classification of uncertain findings. In this study, we introduce another approach for CESM based on a two-material decomposition, with which we expect fundamental improvements compared to the clinical procedure. We demonstrate the potential of our proposed method using the quasi-monochromatic radiation of a compact synchrotron source-the Munich Compact Light Source (MuCLS)-and a modified mammographic accreditation phantom. For direct comparison with the clinical CESM approach, we also performed a standard dual-energy KES at the MuCLS, which outperformed the clinical CESM images in terms of contrast-to-noise ratio (CNR) and spatial resolution. However, the dual-energy-based two-material decomposition approach achieved even higher CNR values. Our experimental results with quasi-monochromatic radiation show a significant improvement of the image quality at lower mean glandular dose (MGD) than the clinical CESM. At the same time, our study indicates the great potential for the material-decomposition instead of clinically used KES to improve the quantitative outcome of CESM.

Improvements to the muon veto of the Dortmund Low Background Facility.

The Dortmund Low Background Facility is a germanium gamma-ray spectrometry laboratory situated above ground. A massive artificial shielding, corresponding to 10m of water equivalent in combination with an active muon veto results in a background level comparable to laboratories situated underground. Due to the recent completion of the muon veto, the background is lowered by 20% compared to previously reported values (Gastrich et al., 2016). Additionally, Monte Carlo simulations of the cosmic muon induced components of the background spectrum are described.

The Dortmund Low Background Facility - Low-background gamma ray spectrometry with an artificial overburden.

The Dortmund Low Background Facility is an instrument for low-level gamma ray spectrometry with an artificial overburden of ten meters of water equivalent, an inner shielding, featuring a neutron absorber, and an active muon veto. An integral background count rate between 40keV and 2700keV of (2.528±0.004)counts/(kgmin) enables low-background gamma ray spectrometry with sensitivities in the range of some 10mBq/kg within a week of measurement time.