ABSTRACT. In this talk, the early stages of development of two emerging radiotherapy (RT) techniques will be discussed. First, kilovoltage x-ray beam arc therapy (KVAT) will be introduced and its development from the conception using analytical calculations on phantoms through Monte Carlo (MC) simulations of patient plans to building a prototype machine will be presented. Second, RT delivered with very high-energy electron (VHEE) beams, including MC simulations, their experimental validation and its potential biological benefits will be presented.

KVAT has been proposed as a treatment modality to improve global access to RT. It is meant to be delivered with a low-energy x-ray beam generated by a high output x-ray tube. The proposed design eliminates the use of expensive linac technology and would result in reduces shielding cost. MC simulations of the KVAT system consisting of a 200 kV linear source mounted on a gantry to enable arc treatments were first performed. After initial phantom simulations, lung and partial breast irradiations on patients were simulated with MC and optimized using McGill’s Radify software. The results of these planning studies and their comparisons to current treatments in terms of dose distributions and treatment times will be presented. A simple prototype proof-of-the-principle experiment set up a table-top system and informed by MC simulations will also be demonstrated.

VHEE RT is an intriguing modality with a potential to improve patient outcomes through ultrafast FLASH therapy. Treatment plans for a machine capable of delivery of magnetically steered 60-100 MeV electron pencil beams were calculated with MC and optimized in a research version of RayStation by RaySearch Laboratories. VHEE treatment plans for a number of cases including pediatric brain, H&N, pelvis and lung cases will be compared to state-of-the-art treatments in terms of dose distributions and speed of delivery. Experimental validation of MC dose distributions for a number of VHEE pencil beams will be also presented. The potential of FLASH irradiations with VHEE beams will be discussed and future directions of experimental VHEE FLASH research at TRIUMF will be outlined.

ABSTRACT. After decades of development and practice, the physics interaction models implemented in electron-photon Monte Carlo codes have become quite homogeneous, at least those referring to electromagnetic interactions. I will summarize the theoretical basis of models currently in use for photon, electron/positron, and proton transport simulation, and discuss their limitations and associated uncertainties. For energies lower than about 1 GeV, the tendency is to use either numerical databases of calculated differential cross sections (DCSs) or analytical models with parameters determined empirically. Photon interactions are frequently modeled by using total cross sections from the Evaluated Photon Data Library, which is frequently complemented with approximate angular distributions, electron shell Compton profiles, empirical electron binding energies, etc. Elastic collisions of electrons/positrons with atoms are simulated from DCS calculated by the partial-wave expansion method; in the case of protons and heavier charged particles, partial-wave calculations are unfeasible and the DCS is calculated (to a similar degree of accuracy) by the eikonal approximation, which was first used by Molière in the formulation of his popular multiple-scattering theory. The description of inelastic interactions of charged particles with atoms is the main dissimilarity between the various codes; a consistent simulation scheme requires modeling the response of individual electron shells to account for the correlation between inelastic collisions and x-ray and Auger-electron emission. Unfortunately, several codes still consider mixed schemes that combine the continuous slowing down approximation and the DCS for close collisions with stationary free electrons or, even less reliably, an energy-straggling distribution of Landau’s or Vavilov’s type. The process of bremsstrahlung emission is relevant only for electrons and positrons; it is usually described by means of the extensive tabulations by Seltzer and Berger. With the most elaborate databases available, simulations of photons, electrons, and positrons are expected to meet essentially all practical needs for energies higher than about 1 keV; proton simulations are expected to be reliable for energies higher than about 1 MeV.

ABSTRACT. Advances in patient-specific information acquired from anatomical/functional imaging and molecular biotechnology are contributing to a new era in radiation oncology. Radiotherapy is a major clinical treatment of cancer that applies a physical tool based on high energy particles to modify the biological environment of targeted cells. Hence, multiple heterogeneous factors may jointly contribute to form a patient’s response to radiotherapy.  Categorization of these factors will enhance our ability to characterize the way ionizing radiation interacts with living tissues and improve its therapeutic efficacy. In addition to traditional clinical and dosimetric factors, imaging and molecular biomarkers are currently being integrated into radiobiological modeling and are enriching our understanding of treatment response in the new emerging fields of radiomics and radiogenomics, respectively with machine learning algorithms in their driver seat. Here, we will provide an overview of these new areas and draw examples from our work and others to highlight their current status and potentials for reshaping treatment management of cancer.  

ABSTRACT. Purpose: Intensity Modulated Radiation Therapy (IMRT) allows for significant dose reductions to organs at risk in prostate cancer patients. However, the accurate delivery of IMRT plans can be compromised by patient positioning errors. The purpose of this study was to determine if the modeling of grade≥2 acute rectal toxicity could be used to monitor the quality of IMRT protocols. Materials and Methods: 79 patients treated with Image and Fiducial Markers Guided IMRT (FMIGRT) and 302 patients treated with trans-abdominal ultrasound guided IMRT (USGRT) were selected for this study. Treatment plans were available for the FMIGRT group, and hand recorded dosimetric indices were available for both groups. We modeled toxicity in the FMIGRT group using the Lyman Kutcher Burman (LKB) and Univariate Logistic Regression (ULR) models, and we modeled toxicity in USGRT group using the ULR model. We performed Receiver Operating Characteristics (ROC) analysis on all of the models and compared the area under the ROC curve (AUC) for the FMIGRT and the USGRT groups. Results: The observed Incidence of grade ≥ 2 rectal toxicity was 20% in FMIGRT patients and 54% in USGRT patients. LKB model parameters in the FMIGRT group were TD_50=56.8Gy , slope m=0.093, and exponent n=0.131. The most predictive indices in the ULR model for the FMIGRT group were D_(25%) and V_50Gy. AUC for both models in the FMIGRT group was similar (AUC=0.67). The FMIGRT URL model predicted less than a 37% incidence of grade ≥ 2 acute rectal toxicity in the USGRT group. A fit of the ULR model to USGRT data did not yield a predictive model (AUC = 0.5). Conclusions: Modeling of acute rectal toxicity provided a quantitative measure of the correlation between planning dosimetry and this clinical endpoint. Our study suggests that an unusually weak correlation may indicate a persistent patient positioning error.

ABSTRACT. Purpose/Objective(s): We investigate the use of objective biomarkers extracted from CT imaging to study the evolution of RILD up to 24 months after RT.

Materials/Methods: Twenty subjects treated in a non-randomized phase I/II trial of isotoxic chemoradiation in stage II/III non-small cell lung cancer were analysed. CT imaging was available pre-RT and at fixed serial time-points following RT: 3, 6, 12 and 24 months. The average CT image resolution was 0.77×0.77×2.7 (±0.09×0.09×1.8) mm3. All scans were acquired at breath-hold with similar inhalation level across time-points for each subject. To investigate how RILD appears and develops, four CT-based biomarkers were calculated over serial time-points: volume of consolidation (RV), pleural change (ΔP [%]), normal lung volume shrinkage (∆NV [%]), and anterior junction line rotation (Δβ [°]). These biomarkers were recently developed and are representative of parenchymal, pleural and lung volume change, and anatomical distortions after RT, respectively. The biomarkers were calculated using semi-automated image analysis pipelines.

Results: The values measured for the biomarkers varied according to time since RT (t={3, 6, 12, 24}months for all measures), providing complementary information on the evolution of RILD. RV peaked at 6 months and reduced at following time-points (RV={2.4±0.9, 2.9±1.4, 2.5±1.1, 2.5±1.1}). ∆P was variable across the patient group (∆P={10±14, 13±17, 10±13, 12±18}%). ∆NV became increasingly more severe over time, with the largest variation occurring from 3 to 6 months (∆NV={12±11, 24±14, 25±15, 27±16}%). Δβ continually increased over time (∆β={1.4±2.0, 3.0±3.3, 5.0±3.5, 6.3±4.2}°. These findings are indicative of an evolution of RILD from early acute inflammation phase (3-6 months), characterised by reversible parenchymal change, into chronic inflammation (6-24 months), characterised by irreversible scarring, progressive lung volume loss and anatomical distortion.

Conclusion: Our findings show that CT-based imaging biomarkers provide objective quantitative information about the evolution of RILD.

ABSTRACT. Purpose/Objective: Accurately predicting distant failure for patients with high risk is critical to achieve better treatment outcomes with intensified treatment modalities. To build a PET based reliable and automatic predictive model, we develop an automated multi-objective (AutoMO) learning model for predicting distant failure in cervix cancer after radiotherapy. Materials/Methods: The dataset used in this study includes 70 patients (stage IB1 to IVA) with definitive intent between 2009 and 2012. All the tumors were manually contoured by radiation oncologists. The extracted features include intensity, texture and geometry. The training stage in AutoMO is to generate the Pareto-optimal model set. In this stage, feature selection and model parameter training are performed simultaneously. Our iterative multi-objective immune algorithm is adopted to maximize the two objective functions including sensitivity and specificity. The test stage consists of weight calculation and predicted probability fusion. In the first step, the weights are calculated based on the trade-off between sensitivity and specificity, as well as the area under the curve (AUC). On the other hand, the weights for models with extremely imbalanced sensitivity and specificity are set as zero. In the second step, the final predicted probability is obtained by fusing all the individual model predicted probability through the evidential reasoning approach which is good at obtaining more reliable predicted results. Then the label with the maximal output probability is considered as the final label. Results: In this study, 90% data is used for training and the remaining is for testing. AUC, accuracy, sensitivity and specificity for AutoMO are 0.8333, 0.8571, 1.0000, and 0.8333, respectively. Furthermore, AutoMO outperforms traditional multi-objective model which achieves 0.6667, 0.7143, 1.0000, and 0.6667 for AUC, accuracy, sensitivity and specificity, respectively. Conclusion: We proposed new AutoMO for predicting distant failure in cervix cancer. The experimental results demonstrated that AutoMO outperformed the traditional multi-objective model.

ABSTRACT. We have used the University of Virginia data set from 92 lung cancer patients undergoing Stereotactic Body radiation treatment (SBRT). The regression of lymphocyte count was done using both the radiation plan input (RT volumes and doses) and clinical variables (initial and final patient absolute lymphocyte count, age and treatment details). Neural network topologies were fine-tuned for minimum mean-square error of predicted final lymphocyte count using the Keras application programming interface (API) with TensorFlow backend. We discuss how one can discriminate between radiation plans by both maximizing the tumor volume dose and minimizing the lymphocyte loss.

ABSTRACT. Quantitative information extracted from patient’s scans (radiomics) can provide additional information to support treatment decision in radiation oncology, besides traditionally accepted semantic clinical features. However, due to the large number of extracted radiomic features, dimensionality reduction and a robust methodology for signature developments is mandatory. Machine learning (ML) is a powerful tool to achieve the above-mentioned goals. In our preliminary study, we compare different ML-based algorithms and classifiers for selecting and combining radiomic features to predict positive lymph nodes in oral cavity squamous cell carcinoma. 134 patients with both biopsies proven oral cavity squamous cell carcinoma (OSCC) and pre-operative diagnostic MRI who eventually underwent curative surgical extirpation including neck dissection with or without adjuvant radiotherapy were included in the study. 130 radiomic features were extracted from manually delineated primary GTV for each patient using the open source software PyRadiomics. All the possible combinations of six different unsupervised ML algorithms and three different classifiers were considered to develop the radiomic signature. Training and validation were repeated 100 times using stratified sub-sampled portions of the data, 80 and 20% respectively, without replacement. The AUC (Area Under the curve) was used to evaluate signature’s performances. Overall, the decision tree classifier had larger performances in the validation, compared to elastic net and logistic regression. Dimensionality reduction applied before classification led to larger performances with respect to directly injecting computed features in the classifiers. The best combination included a decision tree classifier with PCA (Principal component analysis) and ICA (Independent Component Analysis) as reduction techniques: AUC 0.83 and AUC 0.81 on training and validation respectively. Our results show the positive impact of ML unsupervised methods for dimensionality reduction and radiomic signature developments. Comparison with traditional accepted clinical predictors and optimization of feature extraction is needed to further strengthen the study.

ABSTRACT. Purpose Recent studies show the importance of cardiac dose in predicting overall survival following radiotherapy of lung cancer patients. The aim of this study was to employ 4DCT acquired with contrast to optimally identify vascular calcifications. Moreover, association between overall survival and vascular calcifications, as surrogate for vascular health, was studied for patients treated with radiotherapy. Material and Methods Data from 334 lung cancer patients treated with Stereotactic Body Radiation Therapy was analyzed. Calcifications within the thoracic cavity were automatically segmented, by employing image processing algorithms, on the average and all 10 phases of the planning 4DCT scans. Correlation between cardiac comorbidities and calcification volumes was inspected for the subset of patients with recorded Adult Co-Morbidity Evaluation (n=303). Log-rank analysis and Cox-regression model were employed for investigating associations between the obtained calcifications by volume and dose and other known prognostic factors. Results There were no significant differences in the volume of calcifications obtained from each respiratory phase. However, the volume of calcifications from average scan were significantly lower than each phase (p<0.001). Highest level of correlations between myocardial infarction/overall cardiac comorbidities and volume of the calcifications were found for phase 1, 5, and 10 (p<0.005). Volume of the calcifications was subsequently averaged over phase 1, 5, and 10 (extreme inhale and exhale scans). These volumes were used as vascular calcification measure in the thoracic cavity. The adjusted Cox-model suggested that tumor volume and calcification volume are independent factors predicting overall survival.

Discussion&Conclusions This study shows the advantage of using 4DCT scan over average scan for identifying thoracic calcification. Calcification volumes from inhale-exhale phases had the highest correlation with myocardial infarction/ overall cardiac comorbidities and were independently associated with overall survival. Therefore, thoracic calcification, obtained by 4DCT, is a useful indicator for poor prognosis in lung cancer patients treated with SBRT.

ABSTRACT. There is evidence that radiotherapy (RT) affects the morphology and the function of small vessels of healthy and neoplastic tissues exposed to radiation. We propose a computational model able to capture the interplay between microcirculatory vessels and interstitial tissue perfusion in a healthy organ exposed to RT. The model embraces enough of the fundamental physics regulating the tumor microenvironment such that it is suitable to describe also the effects of radiation on the fluid balance at the microvascular level. Here, we take into account two of the main acute effects already discussed in the literature: a) the vasodilation (as a consequence of the inflammatory response) and b) the increased membrane permeability (recently, found to be dose-dependent). Therefore, we have modified the model parameters associated to these features (arterial pressure and endothelial permeability coefficient) to describe a biforcation of metarterioles (microvesseles that precede and feed the capillary bed) during the first weeks of treatment. Indeed, we have also imposed a positive interstitial pressure (+3mmHg, while physiological values are between -5 and 0 mmHg) as indirect effect of the condition a) and b). It mimes the slow impairment of the structures after a certain number of dose fractions and the consequent moderate fluid extravasaton (+5mmHg is the standard interstitial pressure in the condtiton of edema). Reasonably, a similar scenario can occur earlier or later in time depending on the dose per fraction and, possibly, on the radiation source. The results of this study represent a first step towards the challenging objective of understanding, and describing in a mechanistic way the effect of radiation on the vascular microenvironment. It is worth highlighting that this project will include a tuning process of the simulation parameters based on in vivo (sublingual microscopy and spectrophotometric imaging) and in vitro (microfluidi chip) measurments.

ABSTRACT. Combinations of radiation (RT) and hyperthermia (HT, heating a tumour to temperatures above the physiological level) show promise for treating radio-resistant tumours such as those comprising hypoxic subregions. Such combination treatments are planed using biological equivalent dose (BEQD) calculations to estimate their combined effects. These are based on clonogenic survival, and thus account for intrinsic tumour cell sensitivity, but not for microenvironmental factors or dynamic growth. 3D tumour spheroids provide a more physiologically relevant environment in an in vitro setting. A systems oncology framework which simulates the response of tumour spheroids to RT and HT is presented here as an extension to BEQD calculation for multimodality treatment planning. Within the framework of a cellular automaton model, 3D geometries, oxygen diffusion, RT, HT and RTHT treatment delivery were simulated for a large number (104-107) of individual cells forming a spheroid. The large number of cells modelled enabled a calibration and validation of simulations to experimental growth data for HCT116 (human colorectal carcinoma) spheroids for RT (0-10 Gy), HT (0-240 CEM43), and combinations thereof. An analytical spheroid oxygenation model was used for quantitative comparison with the iterative microenvironment simulation. Despite comparable clonogenic survival levels, experimental spheroid growth curves differed significantly depending on the treatment modality chosen. This dynamic response was described well by the framework (coefficients of determination>0.85), once differences in cell death mechanism following HT or RT were considered. Heat-induced cell death was implemented as a fast, proliferation independent process, allowing spheroid reoxygenation and rapid repopulation, whereas radiation-induced cell death mimicked mitotic catastrophe which was restricted to the proliferating cell layers. The strengths of this framework are its experimental validation and novel ability to simulate RT, HT and RTHT treatments in tumour spheroids. It provides a more accurate description of growth response following multimodality treatments where BEQD calculation alone was insufficient.

ABSTRACT. Monte Carlo (MC) simulations are widely used for medical application and nuclear reaction models are fundamental for the simulations of the particles interaction with patients in Ion therapy. Therefore, it is of utmost importance to have reliable models in MC simulations for such interactions. Geant4 is one of the most used toolkits for MC simulation. However, its models showed severe limitations in reproducing the yields measured in the interaction of ion beams below 100 MeV/n with thin targets. For this reason, we interfaced a model dedicated to simulate such reactions, BLOB (“Boltzmann-Langevin One Body”), with Geant4. BLOB is a self-consistent one-body approaches to solve the Boltzmann-Langevin equation. It includes the treatment of the mean-field propagation, on the basis of an effective interaction and fluctuations are included in a two-body interaction term. Furthermore, we implemented a correction to the excitation energy calculated for the light fragments emerging from the simulations and a simple coalescence model. BLOB has been developed to simulate the heavy ion interactions. With the implemented corrections described before, it shows very good results in reproducing the experimental yields of light fragments, up to alpha particles, obtained in the interaction of 12C with a thin carbon target at 62 MeV/n. Such a promising result stresses the importance to integrate it in the Geant4 toolkit.

ABSTRACT. New treatment techniques in radiotherapy make it easier to protect healthy tissues from high doses, reducing short-term deterministic effects and increasing the life expectancy of patients. This, however, might be at the cost of a greater exposure to low doses far from the radiation field, linked to a higher incidence of late effects such as secondary cancer. It could become decisive to appropriately calculate this dose for each treatment, as it tends to increase with modern techniques. Today, medical physicists have close to no reliable and fast tool to predict this peripheral dose, as it is usually overlooked by treatment planning systems. The best way to theoretically calculate this dose is Monte Carlo simulation, which is bound by the noticeable obstacle of computing time. The out-of-field dose being almost exclusively deposited by scattered particles, areas far for the radiation beam tend to be reached by very few particles. Achieving a statistically significant result can then require a considerable amount of time. The solution introduced here is the implementation of a variance reduction technique, based on the pseudo-deterministic transport method, in the possibly soon-to-be-released Monte Carlo code Ines. Whenever a Compton scattering occurs, this technique deterministically scatters and transports a newly created artificial particle to an arbitrarily positioned sphere surrounding the area of interest. Energy estimators remain unbiased by properly adjusting the particle’s weight and not allowing regular particles to enter the sphere during the step following the Compton scattering. Preliminary results suggest that this will allow a much better sampling of the targeted area, as every Compton interaction will end with the creation of a photon on the sphere surface. This, in turn, should make out-of-field dose calculation possible in a reasonable time.

ABSTRACT. Purpose: Monte Carlo (MC) simulations supporting magnetic fields (bfields) are increasing in popularity due to the increasing commercial use of machines combining a linear accelerator and an MRI. This study focusses on practical considerations for employing MC simulations supporting magnetic fields in treatment planning software. Methods: Two aspects of MC simulations in bfields are studied: electrons spiraling in air and end of track algorithms. Spiraling electrons cause a massive performance penalty which is handled by the development of a method to identify those spiraling electrons and the usage of a simplified transport algorithm. The simplified transport involves approximating the helical trajectory of the electrons. Simulations without and with air around the phantom are performed to study the impact of using or not using the simplified transport. The CSDA end of track algorithm for electrons is adapted to simulations with bfield by modifying the “straight line” approximation of a non-bfield implementation to one that considers the helical trajectory of electrons in a bfield. A geometry designed to show differences between incorrect and correct implementations is used, in conjunction with high and low electron transport energy cut-off values. Results: The identification of spiraling electrons and further simplification of the particle transport results in a drastic reduction of the penalty between a simulation with and without air added around the patient contour from 3.85x to 1.3x. No dose differences in non-air voxels are observed between the full transport in air and the simplified transport. The modified CSDA end of track algorithm resulted in no performance difference. The dose results show that the simulations with a low cut-off and the simulation with a high cut-off agree only if the simulation with high cut-off uses the modified end of track algorithm. Conclusions: This study looked at two practical aspects of MC simulations within bfields.

ABSTRACT. Proton-boron fusion is a neutron-less reaction with a reaction Q-value of 8.7 MeV. The reaction goes through the resonant creation of a 12C in its excited state (with a resonance when protons have an energy of about 675 keV) that mostly splits into three alphas with energy about 4 MeV. This reaction was recently proposed by Do-Kun Y. et al. for medical purposes: the creation of high-LET alpha particles might enhance protontherapy, whose limit is the low effectiveness when treating radio-resistant tumours compared to carbon treatments. This technique is called Proton Boron Capture Therapy (PBCT). Radiobiological experiments recently conducted by Cirrone et al. for the first time experimentally demonstrated an increase in proton lethality when boron atoms are introduced inside the tumour mass. In this work, we will discuss the possibility to use the Geant4 Monte Carlo simulation code in order to estimate the alpha yield and understand if its increment as due to the Boron presence can justify the observed radiobiological effects. Some Geant4 hadronic physics models (Bertini cascade, binary cascade, Liège intranuclear cascade and ParticleHP) have been investigated using a modified version of the HADR03 extended example and compared to available experimental data taken from EXFOR. Comparisons against TALYS-1.9 analytical calculations have been performed, as well. We found that all available hadronic Geant4 models were unable to satisfactorily describe this reaction. Even the parameterised ParticleHP model, recently included in Geant4 to better deal with low energy range, failed at energies below 1 MeV, missing the 675 keV resonance. An improved version of the ParticleHP model is proposed here, where the proton-boron cross section is changed to better agree with the experimental data, and allowing more precise simulations of experiments where proton-boron fusion is important, like in PBCT.

ABSTRACT. Recently, there has been growing interest in heavy ion therapy to use 4He ion beams to treat patients as an alternative to proton and 12C ion therapy, with 4He presenting a balance between proton and 12C therapy. Compared with proton therapy 4He provides significantly less lateral scattering as well as greater effectiveness for radio-resistant tumours. While compared to 12C therapy 4He has a significantly reduced fragment tail, making it attractive for paediatric cases which have traditionally avoided using 12C therapy due to the prominence of the fragment tail. However, despite the reduced fragmentation of 4He compared to 12C beams, the fragment tail is still present and due to the mix of primary and secondary particles it is important to consider when planning the treatment of a patient, especially for paediatric cases. Monte Carlo codes are extensively used in radiotherapy to study the radiation field, particularly in hadron therapy; as such it is essential that the accuracy and the response of the fragmentation models are understood and are reliable as possible in the energy range of interest for hadron therapy. This work compares the performance of three alternative fragmentation models available in Geant4 against different experimental measurements. The three models compared were the Binary Intranuclear Cascade (BIC), the Quantum Molecular Dynamics (QMD) and the Liege Intranuclear Cascade (INCL). Comparisons were made against four different experimental data sets, available in the literature, comprising of fragment yields as well as angular and energy distributions. The experimental measurements used therapeutic energies of 4He ion beams from 90-220 MeV/u incident upon water, PMMA and graphite targets. The different models studied were seen to vary quite differently between one another and, depending on the fragment type, agreement with experiment varied between ~15-60%.

ABSTRACT. Coupled simulation of water radiolysis following the energy deposition of charged particles goes a step further in extending the modeling of the biological effects of radiation from first principles. Sophisticated Monte Carlo simulation tools facilitate such modeling by allowing the combination, within the same simulation, of complex geometry models, detailed physics models and an extended scheme of chemical species and reactions. In this work, we present the extended capabilities of the Geant4-DNA based TOPAS-nBio tool for water radiolysis modeling. G-values calculated with the step-by-step method and the independent reaction times (IRT) method, recently implemented in TOPAS-nBio, were compared as a function of the track-averaged LET100eV (0.1-250 keV/μm) for short-tracks of e–, protons and alpha particles. IRT was up to 145 times faster and allowed inclusion of an extended scheme of reactions that improved the agreement of LET-dependent G-values with experimental data from the literature (figure 1). As a demonstration, the simplicity of using TOPAS-nBio to set up a plasmid supercoiled DNA geometry irradiated by 60Co gamma rays for the quantification of single and double strand breaks of DNA will also be presented.

ABSTRACT. Geant4 is a Monte Carlo code widely used in medical physics, from dosimetry to nanodosimetry, imaging to radiation protection. Validation and performance monitoring focused on physics quantities relevant to this domain are of crucial need. In addition, it is important to identify the most appropriate physics models for a specific medical physics application. To respond to these needs, we developed G4-Med, a fully automated benchmarking and regression testing suite of Geant4 for medical physics. The testing suite currently includes 15 tests, from basic physical quantities (stopping powers, cross sections, electron scattering, dose point kernels, etc.) to tests of realistic set-ups typical of medical physics applications (hadron therapy, brachytherapy, external electron beam therapy). Both electromagnetic and hadronic physics processes/models are tested. The tests have been integrated in the Geant-val platform and executed for Geant4 development tags and public releases. The physical observables are compared to reference data for validation and to previous Geant4 versions for regression testing. Users can access and download the results of the benchmarking suite from the web. At the conference we will show the results obtained for Geant4 10.5 and how to use the testing suite.

ABSTRACT. The relativistic impulse approximation (RIA) for the Compton interaction incorporates both binding effects and Doppler broadening and yields an expression for the DDCS differential in outgoing photon angle and energy. The key ingredient in the calculation of the RIA cross sections is the Compton profile (CP) of each electronic orbital, which is computed from the corresponding linear momentum distribution. Since only calculations involving atomic CPs have been reported, In this work we investigated for three materials of dosimetric interest (air, water, and carbon) the impact of using molecular CPs on the Compton energy-transfer cross section derived within the RIA. The CPs were integrated from momentum densities obtained through self-consistent Hartree-Fock calculations, with wave functions expanded in a cc-pVTZ Gaussian basis set. The new energy-transfer cross sections are relevant for trC/ values in the tens of keV range, where for these materials Compton becomes the dominant interaction. We find that there is very little difference between the RIA with atomic and molecular CPs, thus the RIA does not seem to be particularly sensitive to the specific shape of the CPs. We also studied the difference between the RIA and the Waller-Hartree (WH) approach to modelling binding effects and found that, surprisingly, both RIA curves (molecular and atomic) are significantly closer to the Klein-Nishina mass energy-transfer coefficients than the WH model. The latter can differ by a large amount from the other models in the tens of keV range (e.g. 6-10% at 20 keV).

ABSTRACT. Geant4-DNA [1,2] is a package of Geant4, a Monte Carlo toolkit that simulates the passage of particles through matter. It contains the models of interaction of ionizing radiation at DNA scale. Currently, the models for ionization and excitation of liquid water molecules by protons have an upper energy limit of 100 MeV. However, clinical proton beams used in radiotherapy can reach up to 250MeV. Thus, the goal of this work is to extend the range of applicability of Geant4-DNA for track-structure calculations of protons through liquid water up to roughly 300 MeV, so that the energy range of interest in proton therapy can be covered. The calculation of the doubly-differential cross section (DDCS) in energy loss and momentum transfer is framed in quantum relativistic mechanics, based on [3]. The interaction is treated at first order perturbation, concretely under Relativistic Plane Wave Born Approximation (RPWBA), with separate projectile and target contributions to the DDCS. The latter is represented by the generalized oscillator strength (GOS), which characterizes the response of the media. We modelled the GOS of liquid water for each shell, splitting each shell GOS in two regimes of momentum transfer. In the low momentum regime, the GOS is calculated using models based on the dielectric functions of liquid water [4]. From DDCS, we have computed the mass stopping power with our model and compared it with PSTAR tabulated values. We found a remarkable agreement for a wide range of proton incident energy, included the values of interest. [1] S. Incerti et al., Med. Phys. 37 (2010) 4692-4708. [2] M. A. Bernal et Al., Phys. Med. 31 (2015) 861-874. [3] F. Salvat, Nucl. Instrum. Meth. B 316 (2013) 144-159. [4] D. Emfietzoglou, Radiation Research 164, (2005) 202–211.

ABSTRACT. Proton therapy has become a most popular in radiation oncology due the superior dose distribution of proton beam. However, the advantage of proton therapy cannot be fully utilized without proper measurement of in-patient proton dose. Currently, no clinically applicable method is available. Detecting secondary gammas has been proposed as a potential method to measure in-patient proton dose since treatment protons stop within the patient as they deliver the dose. One possibility is the development of an imaging device to measure the scattered gamma-rays produced during a proton therapy treatment. During the design of this imaging device, Monte Carlo simulations have been performed to understand the production of these secondary (prompt) gamma-rays, particularly from tissue. Discrepancies have been reported in the Geant4 prompt gamma production specifically in the most prominent elements (12C and 16O) of tissue. The goal of this study is to compare the measured and simulated prompt gamma cross section of 4.438 MeV of 12C over the proton energy range of 80 – 125 MeV using Geant4 AFRODITE model. The measurement was carried at iThemba LABS using the AFRODITE clover detector system. A proton beam over the range of 80 -125 MeV were used to hit a natural carbon target of thickness 8.400.07 mg/cm3. In the simulation study, the geometry of the AFRODITE detection system was carefully modelled to mimic the actual geometry by importing CAD models into the Geant4 code. The physics of the AFRODITE model was tested. Once the model was validated, the experimental runs were simulated and the same procedures were followed to calculate cross-section. As with the experimental 4.438 MeV cross-section data, Geant4 simulated cross section results appear to be higher than the expected values, but due to the scarcity of data, it is hard to determine if these data points are indeed too high.

ABSTRACT. H. Piersma [a]

Introduction

Current complex modulated treatment plans and treatment planning algorithms require advanced verification methods. VMAT treatment plans are mostly created with low energy photon beams where photoelectric and Compton interactions dominate[1]. WebGL 2.0 is a web standard[2] for rendering 3D content that conforms closely to the OpenGL ES 3.0 API and is available on every system with a modern GPU and web browser.

Materials & Methods

A platform independent web application was developed that uses the WebGL 2.0 API to render 3D content and simulate low energy Monte Carlo photon interactions using parallel programming techniques. New WebGL 2.0 features were used to generate a good quality parallel random number generator, multiple draw buffers made it possible to transfer photon state data after each iteration. Klein-Nishina directional sampling was optimized and branching was minimized to improve parallel performance. Phantom and patient anatomy is represented as a triangular mesh in order to make use of standard 3D rendering algorithms.

Results

The first results where the interaction of 250.000 photons in a small mesh cube geometry is tested show that fast (few seconds on integrated GPU of an Intel Celeron N4100 CPU) dose calculations are possible.

Discussion & Conclusions

It is feasible to implement a photon dose calculation engine using the WebGL 2.0 web standard. The next steps will be to validate and optimize the dose calculation engine. After this it will be investigated if it can be used to verify advanced treatment planning algorithms and use it for patient specific QA.

References

[1] A simple algorithm for the transport of gamma rays in a medium, AJP 71, 38 (2003) [2] WebGL 2.0 Specification, khronos.org

[a] Medisch Spectrum Twente, Koningsplein 1, 7512 KZ Enschede, The Netherlands

ABSTRACT. Purpose:Particle imaging offers potential improvement of range uncertainties in particle radiotherapy by directly sampling the stopping power. However, no end-to-end characterization of the image quality (signal-to-noise ratio (SNR) and spatial resolution) as a function of the dose/energy is available. This work investigates these characteristics in proton (pCT) and helium CT (HeCT) and describes their relationship with object size, with the aim of minimising range uncertainties and establishing a fair comparison with X-ray CT (XCT). Methods:The imaging noise originates from Coulomb scattering with the nucleus and energy loss straggling with electrons. Monte Carlo simulations were used to transport proton/Helium (n=107) through a water cylinder (r=15 cm), and noise profiles were recorded. The tomographic noise over a set of projections was extrapolated through adapted filtered back projection. Spatial resolution was expressed as the modulation transfer functions (MTF) extracted from simulations of a Catphan (The Phantom Laboratory, NY) CTP528 module. Results:Tomographic pCT/HeCT noise increases with phantom thickness and decreases with decreasing energy. Scattering noise is dominant around the cylinder edge whereas straggling noise is maximal in the center. HeCT noise is four times lower than pCT noise. For an equivalent dose, the SNR is ten times more between particle and X-ray imaging at 200 MeV/u and decreases with increasing energy. The pCT MTF is limited by Coulomb scattering with the nucleus and lower than XCT at low energy. The MTF 10% level is smaller by 50% at 230 MeV but increases with increasing energy. The HeCT MTF10% is 1.4 times higher than the pCT MTF10%. Conclusion:Particle imaging shows a significantly lower spatial resolution and higher SNR than XCT. Increasing the beam energy reduces the SNR, thus introducing range uncertainty, but increases the spatial resolution and reduces unaccounted range-mixing due to partial volume effects. The cumulated impact on range uncertainty will be subsequently studied.

ABSTRACT. A distributed computing system (RTGrid) has been configured and deployed to provide a statistically robust comparison of Monte Carlo (MC) and analytical dose calculations. 52 clinical oesophageal radiotherapy plans were retrospectively re-calculated using the Pencil Beam Enhanced (PBE) and Collapsed Cone Enhanced (CCE) algorithm within the Oncentra v4.3 Treatment Planning System. Simulations were performed using the BEAMnrc and DOSXYZnrc codes. The Computing Environment for Radiotherapy Research (CERR) has been used to calculate Dose Volume Histogram (DVH) parameters such as the V95% for the Planning Target Volume (PTV) for the PBE, CCE and MC calculated dose distributions. An initial sample of 12 oesophageal radiotherapy treatment plans were simulated using the RTGrid system. The differences in the DVH parameters between the dose calculation methods, and the variance in the 12 cases, were used to calculate the sample size needed, which yielded a required sample size of 37. Therefore, a further 40 oesophageal cases were simulated, following the same method. The median difference in the PTV V95% between CCE and MC in the group of 40 cases was found to be 3%. The differences between the two sets of PTV V95% values did not follow a Gaussian, so the Wilcoxon matched pairs test was used to compare the distributions. This showed that the null hypothesis (i.e. that the distributions are the same) was rejected with a p-value less than 0.001, so there is very strong evidence for a difference in the two sets of values of PTV V95%. Similar statistical analyses were performed for other DVH parameters, as well as Conformance Indices used to describe the agreement between the 95% dose and the PTV, and estimates of the Tumour Control Probability (TCP). From the results, the use of MC simulations are recommended when non-soft tissue voxels make up > 60% of the PTV.

ABSTRACT. High Z value of gold allows greater local photo-absorption followed by photoelectrons and Auger electron emission, in which, contributes to local dose enhancement. The local dose enhancement is maximized when AuNp interacts with low energy x-ray source like palladium-103. To test the possibility of constructing dissolvable brachytherapy seed, palladium-103 core and a gold shell structured nanoparticle (18±2 nm for Pd-103 core and 50±5 nm total NP) has been created. Then we investigated the relative nanoscale dose deposition around a single palladium core gold nanoparticle (Pd-GNP) by varying the gold shell thickness and palladium core size, through the use of a general purpose Monte Carlo code, GEANT4. As a result, cubic shaped core palladium nanoparticles with monocrystalline structure of size 18±2 nm were produced and Pd-GNP showed mono-dispersed size of 50±5 nm, smooth and equal thickness of gold layer and sphere shaped morphology. Nanoparticle stability and cytotoxicity assays both in vitro and in vivo demonstrated, Pd/Au nanoparticles are highly stable and no cytotoxic in biological system. Monte carlo simulation indicate that greater and greater levels of dose deposition occur when the gold shell thickness decreases, possibly suggesting that gold in this configuration does not enhance dose. In addition, overall enhancement for this configuration might actually occur for smaller palladium sizes as well, which is indicated by the fact that the relative dose is maximized at smaller palladium and a special no gold case. Larger scale studies are necessary to determine if the results presented in this work lead to the conclusion that Pd-GNPs do not enhance dose if used clinically.

ABSTRACT. Introduction Intraoperative electron radiotherapy (IOERT) refers to the administration of a high single fraction dose to the patient immediately after surgery. Due to the direct contact of employed dedicated accelerators and patient during this treatment technique as well as high prescribed dose levels, in-field radiation contamination during the irradiation may be of concern. The aim of this study is to evaluate the photon and neutron contamination dose during the IOERT. Materials & Methods LIAC12 dedicated IOERT accelerator was considered in this study. Accelerator was modeled by MCNPX 2.6 Monte Carlo code and in-field radiation contamination including photons and neutrons was quantitatively evaluated through scoring the contaminant dose along both central axis and off-axis distances from the primary incident electron beam. All of dose calculations were performed using 10 cm reference applicator inside an ICRP soft tissue equivalent phantom and for electron energies of 6, 8, 10 and 12 MeV. Results The photon contamnination was observed at all electron energies and its intensity increases with electron energy increment. With increasing the lateral distance from central axis, the contaminat photons dose continiously decreses while, a buildup effect was observed along the central axis. In contast to the photons, neutron contamination was only existed at 12 MeV electron energy. The neutron dose gradually decrements with increasing the distance along beam central-axis and off-axis. Neutron intensity was siginificantly lower than the photon one so that the neutron dose could be neglicted in comparison with the photon dose. Discussion & Conclusions In-field radiation contamination of dedicated IOERT accelerators is kept at a very low level. This fact is manily attributed to the employment of low atomic number materials such as PMMA and PEEK for different components of accelerator head as well as absence of tanguestan/lead made movable jaws in design of these dedicated IOERT accelerators.

ABSTRACT. Introduction In radiotherapy, Gafchromic films are widely used for dosimetry purposes. EBT2 and RTQA2 film are Gafchromic films, the difference lies in their composition of different layers. It is essential to know the energy response of different layers of film to use RTQA2 film for dosimetry purpose over of EBT2 film. Materials & Methods Different layers of EBT2 and RTQA2 films were simulated using EGSnrc Monte Carlo Code. A parallel beam of radius 5 cm is used as a source of photon beam of energies Co-60, 4-24 MV. Both the films were positioned on the surface and at 5 cm depth in water phantom of radius 15 cm and thickness 25 cm placed at a distance of 100 cm. Results The dose difference in the active layer of RTQA2 film and EBT2 film when placed on the surface was found to 0.4%, 3.62%, 7.85, 8.12%, 1.61%, 0.5% for Co-60, 4, 6, 10, 15, 24 MV respectively. At 5 cm depth, the dose difference was found within 2% for all energies. At the surface, the absorbed-dose in all the layers of both film decreases as the energy of photon increases and at 5 cm depth, the absorbed-dose in all layers increases linearly as energy of photon increases. Discussion & Conclusions The energy response of different layers of both the films depends on the energy of the incident photon and the depth of measurement. As the energy responce from active layer of RTQA2 film is within 2% with EBT2 film at 5cm depth, therefore, RTQA2 film can be used as an alternative to EBT2 film for dosimetry purpose at lower depths. References [1] M.J. Butson, K.P.N. Yua, T. Cheunga, P. Metcalfe, Radiochromic film for medical radiation dosimetry. Materials Sci Eng R 2003; Acknowledgements NIL †Corresponding Author: gmargarita.mora@gmail.com

ABSTRACT. Introduction

We first describe the modelling of the MLC-based M6 Cyberknife accelerator in EGSnrc. The model is integrated in the Moderato platform and patient plans are re-calculated. Finally, a new machine learning (ML) method is proposed to predict the electron beam parameters, allowing significant time gain in the linac modelling.

Materials and methods

Electron beam energy and spot size for the M6 Cyberknife were optimized from measured dose profiles in water using BEAMnrc/DOSXYZnrc. The accelerator model was then integrated in Moderato, a Monte Carlo platform offering independent verification of dose distributions calculated by the Treatment Planning System (TPS). Dose distributions from algorithms included in the TPS (Finite-Size Pencil Beam FSPB and Accuray Monte Carlo AMC) and from Moderato were compared for patient plans. Finally, a new ML regression algorithm was trained to predict the parameters of the electron beam by extracting features from simulated dose profiles with varying spot size (1 to 4 mm) and energy (4 to 8 MeV).

Results

From dose profiles comparisons, a monoenergetic electron beam of 6.75 MeV with a gaussian spatial distribution of 2.4 mm FWHM was selected. Patient plans showed good agreement (< 2 %) between the three algorithms, although significant differences (> 5 %) appeared for cases including so-called ”peripheral” fields. These are being investigated through measurements. Finally, a 10-fold cross-validation of the ML algorithm showed that energy and spot size could be predicted with a mean absolute error (MAE) of 0.4 MeV and 0.2 mm respectively.

Conclusions

The Moderato platform now includes the MLC-based Cyberknife in its supported accelerators, allowing for routine verification of patient plans. In addition, a ML algorithm was tested to validate the concept of predicting electron beam parameters from profile data. Further work is ongoing to reduce the uncertainty on energy, and apply this principle to other devices.

ABSTRACT. Introduction In recent years, small transmission type x-ray tubes have attracted significant attention for electronic brachytherapy in radiotherapy. The purpose of this study is to determine the optimum target thickness on uniform angular dose distributions for a transmission type X-ray tube with a truncated conical-shaped anode.

Methods X-ray characteristics of a small transmission type X-ray tube were simulated with the MCNP6 Monte Carlo (MC) code. Angular dose distributions of the different thicknesses of tungsten target were investigated for a range of electron energies from 30 keV and 70 keV.

Results The MC simulation results showed the effective intensity of Bremsstrahlung X-ray were generated for the tungsten targets with the thicknesses of 0.6-23 μm. Uniform angular dose distributions at polar angles of 0°≤ θ ≤150° in 10 increments were shown at the optimum target thicknesses for electron energies of 30-70 keV, respectively.

Conclusions We present the optimum target thicknesses for producing the effective Bremsstrahlung X-ray and uniform angular dose distributions of the truncated conical-shaped transmission type X-ray target at different incident electron beam energies.

Acknowledgements This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2015R1C1A1A02036358).

ABSTRACT. The Advanced Examples released as part of Geant4 demonstrate the use of this Monte Carlo simulation toolkit in a variety of realistic application scenarios, ranging from high energy physics to space science and medical physics applications. In this work, we will illustrate the recent developments of the Advanced Examples in the field of medical physics, including the refinement of existing applications (e.g. brachytherapy, hadrontherapy, IORT_therapy, medical_linac, microdosimetry) and the inclusion, for the first time, of an example (called doiPET) dedicated to the characterisation of a depth-of-interaction PET system. We will describe the recent developments as released in Geant4 10.5.

ABSTRACT. In view of the development of nuclear technology in various areas of life, study on use of different materials in nuclear shields have grown recently. During the recent years and throughout the world, application of nanotechnology and nano structured materials in the development of nuclear radiation shields is taken into consieration. This technology right now is used for multipurpose radiation shields. In the other hand, several studies have extensively shown that shields with nano materials show more efficiency, less toxicity and hurt and also more radiation flux-weakening power. Moreover, nano-structure materials can improve the mechanical performance of the shield. Therefore, in this study, which is a combination of nuclear shielding and nano-technology application, design and development of a nuclear radiation shield using nano-diamond has been considered. In this research, purification and crushing of the nano-diamond hunks in view of its thermodynamic features have been studied by mechanical tensions. Method of the construction of polyethylene cylindrical shields coated with diamond nano-particles has also been presented. To examine the shield accurately, SEM, TEM, XRD and EDAX analyses have been conducted on the produced samples. The results show that nano-particles have appropriate size and purity which cause specific surface increment in the nano-diamond structure. Several tests have been conducted to examine neutron moderation and gamma absorption of the shield. The results have been compared and analyzed through simulation by help of Mont Carlo code (Geant4). Simulated results data have been well consistent with experimental tests. The shields containing nano-particles have a very well moderating properties compared to the typical shields (polyethylene, graphite, water …).

ABSTRACT. We compared the simulation results by EGS5 and PENELOPE 2014 for a very simplified geometry and radiation source. EGS5 and PENELOPE give the similar energy deposition when the photon energy is below Co-60 energy and the scored region is larger than 0.1 cm, however there seems an apparent difference at higher energy than 2 MeV, which is important in the radiation therapy with a clinical Linac.

ABSTRACT. A new system of glasses such as 10WO3-xMoO3-(90-x)TeO2, x = 10-40 mol% have been investigated to check their potentiality for radiation shielding properties in medical science. For this objective, the FLUKA Monte Carlo code is used to estimate some radiation shielding quantities like mass attenuation coefficient, mean free path, half value layer and effective atomic number for WO3-MoO3-TeO2 glass systems. The energy dependence of these shielding parameters has also been reported for the commonly used gamma sources with energies 59.5 (241Am), 80.8, 276.4, 302.8, 356, 383.8 (133Ba), 661.6 (137Cs) and 1173.2 (60Co) keV using the FLUKA simulation program. A simple rectangular geometry with the axis along the z-direction was described in the input file. A beam of 1x105 gamma photons was directed towards the materials in the z-direction and attenuated in rectangular parallelepiped glass samples. The results of photon transmission were obtained from output files for each of the material thicknesses using the USRBDX score card. The results obtained with FLUKA have been compared with experimental data available in literature. The current theoretical values of shielding parameters centered on FLUKA Monte Carlo code are in good agreement, relative difference less than 1%, with available experimental data and NIST XCOM database. Among the investigated samples, 10WO3-10MoO3-80TeO2 glass is found to have excellent shielding properties when compared to lead based glass and ordinary concretes. It is concluded from the study that FLUKA can be an alternative way to estimate various shielding modalities, for glasses with reinforcement, especially when it is hard to set up an experiment. The exploration of heavy metal oxide (HMO) glasses for radiation shielding properties may open many new technological applications in field of radiation dosimetry and protection.

ABSTRACT. In recent years there has been an increasing interest in the combination of high atomic number materials and radiotherapy, due to the enhancement in dose deposition resulting from the boosted production of secondary electrons. This effect is particularly evident when tissues are irradiated with X-ray energies just above the absorption edge of the metal. In this study, Monte Carlo simulations were employed to evaluate the contribution of gold-containing compounds to the depth dose profiles, when utilising monochromatic X-rays in the kiloelectronvolt energy range. Materials and methods A 2x2 cm2 uniform monochromatic X-ray beam of energies ranging from 30 to 120 keV was perpendicularly impinging on the phantom. The physics processes were described with the G4EmStandardPhysics_option4 libraries included in GEANT4 v.10.4. The PMMA phantom exploited had an approximate size of 50x50x90 mm3. Two 14 mm slabs of various concentrations (up to 30 mg/ml) of gold-containing materials were inserted at a depth of 20 and 54 mm. The phantom was voxelised using the GEANT4 mesh to allow for the assessment of dose deposition. The dose enhancement ratio was calculated as the simulated dose with and without gold-containing materials. Results obtained in-silico through GEANT4 at a gold concentration of 1.8 mg/ml were compared with experimental data collected at the European Synchrotron Radiation Facility with the aid of a PTW semiflex ionisation chamber. Results and conclusions The simulated dose enhancement factors showed a neat photon energy- and concentration- dependency. The calculations clearly highlighted a dose enhancement in the contrast medium areas, and the experimental measurements carried out in PMMA just before or after those regions overlapped the depth dose deposition simulated with GEANT4.

ABSTRACT. TBA

ABSTRACT. An outcomes analysis tool, PROPEL (Platform for Radiotherapy Plan Evaluation and Learning), has been developed to collect DICOM data and outcomes data from multi-center studies in a single framework. PROPEL has been developed to support the NHS England Commissioning through Evaluation program for SBRT. The aim of the platform is to allow many sources of data relevant for radiotherapy analysis to be queried from a single platform. Both DICOM and outcomes data are entered and submitted via a web application. Summary dosimetric data such as DVHs are calculated and passed into an analysis database. An application has been created to query the analysis database using RESTful web services. This allows data analysis to be performed using and software capable of making an http request. We show the use of this system to extract dose volume metrics and display the results. We find that PROPEL allows data to be gathered from a range of systems and multiple institutions. A very simple analysis is demonstrated to show that as expected the ratio of OAR dose to PTV dose falls the further apart the structure are in space. Currently this framework is being used to perform an outcomes analysis of the SABR Commissioning through Evaluation data using the DICOM dosimetric and planning data.

ABSTRACT. The gamma index is a useful tool for quantifying the deviation of a reference point dose value from a three-dimensional dose distribution with a single number. In practical applications one may be required to quantify the difference between two or three-dimensional dose distributions. This can for example be a difference between two calculated dose distributions or a 2D film measurement against a calculated dose. A widely used method is to calculate the gamma index for all points in one of the distributions and then to find percentage of gamma error values smaller than a threshold value. This method that is suitable for special evaluation applications can be called Gamma Index Pass Rate (GIPR). This is not a good metric if a priori knowledge of the degree of similarity of the two dose distributions is not available. This is because if the average dissimilarity is much smaller or larger than required, the pass rate is always 100%, or always close to 0%, respectively, neither of which is very useful information. By inverting the GIPR we get the gamma index threshold value corresponding to a given pass rate or in other words Minimum Achievable Gamma Index threshold Coefficient (MAGIC). Finally, we obtain Varian Dose Distribution Distance Metric (VDDDM) by combining several MAGIC metrics that correspond to different dose regions (e.g. high dose rate region, surface region etc.). This single value metric allows to quantify differences in 3D dose distributions in effective and intuitive way. It allows to have different evaluation criteria applied to multiple regions of the dose distributions and percentiles of the gamma index distribution but combined to a single value that can be used to assess the quality of the agreement between different observations.

ABSTRACT. Medical software and particularly treatment planning systems are increasingly offering scripting capabilities. Scripts for such systems can improve clinical workflows and provide otherwise missing functionality. Automated code testing is an important aspect of best practice development and maintenance strategies. This is particularly important where updates to the parent application have the possibility to cause scripting errors or give erroneous results. A script testing framework has been developed for RayStation (RaySearch, Stockholm, and Sweden) and is released under an open source license. The framework makes it possible to create and quickly run a set of automated tests which can validate the continued correctness of scripts in clinical use. The framework has been used in clinical practice to develop a set of over 250 separate script tests that run in under 30 seconds. It has been used to check for changes in scripting syntax over four separate upgrades of the parent application.

ABSTRACT. This project aims to create an open source, automated, tool to present the dosimetric consequences of deviations between the reported log file data and the original planned parameters. Data within the high-resolution log files are converted to the DICOM RT Plan file format. These RT Plan files are sent to dose calculation software for reconstruction of delivered dose, which is then compared to the original dose distribution. The source code for this software is available within the PyMedPhys python library (https://pymedphys.com).

ABSTRACT. The open source treatment planning toolkit “matRad”, published in 2015, has built up a substantial community (>70 forks on Github) and has been used in research projects at more than 25 institutions worldwide. Here we report on novel developments within matRad which are of general interest to the research community. (1) Interfaces to the open source photon and proton Monte Carlo (MC) dose calculation engines “ompMC” and “MCsquare” have been integrated into matRad’s planning workflow. The lightweight photon (and electron) MC dose calculation engine ompMC was modified to compile directly as a shared library MEX interface. The high-performance proton dose calculation engine MCsquare is called as standalone executable from Matlab. Both interfaces conform to the existing standards of matRad’s native analytical dose calculation algorithms enabling modular switching between analytical and MC dose calculation with minimal effort for dose influence matrix computation. For photons and protons we show a lung cancer treatment plan that was optimized based on a MC beamlet calculation within 168min and 120min, respectively. (2) The modularity of matRad has been improved to better cater the needs of specific research projects. It is now possible to use custom resolutions (independent from the CT grid) for dose calculation, enabling users to define a custom trade-off between performance and accuracy. Additionally, a new object-oriented interface for optimizers, optimization problems and objectives/constraints was implemented. (3) A continuous integration framework has been implemented for matRad’s Github repository based on “TravisCI”. This enables automatic integrity tests for novel internal as well as external developments (via pull requests) with low-dimensional dose calculation and treatment planning problems. The reported new developments enable researchers to work with comparable algorithms and realize their own research projects intuitively, while keeping track of distributed development and code integrity.

ABSTRACT. Particle therapy with raster-scanning ion-beams has become an increasingly available modality to treat cancer. Compared to conventional (photon) therapy, it offers improved tumour targeting, while optimally sparing healthy tissue. In particle therapy, multipurpose Monte Carlo (MC) simulations are considered a gold standard in dose calculation but are too time-consuming to be clinically viable. Although fast MC codes were recently introduced, they don’t support the four ion-species available at the Heidelberg Ion-Beam Therapy Center (HIT): protons, helium, carbon and oxygen ions. To that end, Fast dose Recalculation on GPU (FRoG) has been developed in-house at HIT and Centro Nazionale di Adroterapia Oncologica (CNAO). FRoG affords fast dose calculation for all four ion-species by employing NVIDIAs CUDA architecture wrapped by the Python programming language for graphics processing unit (GPU) accelerated dose calculation. FRoGs base-data is FLUKA MC generated and uses a triple Gaussian beam model in water. Inaccuracies in dose calculation caused by tissue heterogeneities are resolved by an advanced pencil beam splitting algorithm with each pencil beam subdivided in up to 700 sub-pencil beams. As a validation study, five head and neck patients that have been treated with protons at HIT are compared to experimentally validated FLUKA MC simulations. Agreement was analysed by means of dose volume histogram metrics D5, D50, and D95 that correspond to the dose that 5%, 50% or 95% of the target receives. The highest D95 deviation is seen for a patient, where high Z material (metal implants) is located within the beam path. Overall, excellent agreement with FLUKA simulations is found with average D05 and D50 values below 1%, whereas D95<1.9%. With calculation times of up to 300 times faster than FLUKA simulations and comparable accuracy, FRoG is a powerful tool to support the clinical activity and large-scale patient cohort analyses at HIT.

ABSTRACT. Software libraries for deriving radiomics features and in-turn outcomes models have grown rapidly since the term “radiomics” was first introduced by Lambin et al in 2012. Similarly, automatic image segmentation, which is crucial for reproducible extraction of radiomics features and treatment planning, has seen a great deal of recent progress with deep learning. However, there is lack of centralized repository which provides implementations of radiomics and segmentation models. Such a library of model implementations will be useful to (i) test existing models on datasets collected at different institutions, (ii) automate segmentation, (iii) create ensembles for improving performance and (iv) incorporate validated models in clinical workflow. In this work, we present a platform based on Computational Environment for Radiological Research (CERR) to create a library of radiomics and segmentation models. CERR is a natural choice to centralize model implementations due to its comprehensiveness, popularity and ease of use (Matlab-based). CERR’s radiomics feature extraction is validated and a user has fine control over calculation settings. This allows users to select appropriate feature calculation used in model derivation. Models for automatic segmentation are distributed via Singularity containers, with seamless i/o to and from CERR. Singularity containers allow for model implementations to be independent of operating system. Radiomics and segmentation models with their associated parameters are defined via json file. This makes it convenient for users to plug in models without having to write any software code. Additionally, the wrapper code in CERR can be called programmatically for batch evaluation of these models. To demonstrate the utility of our centralized repository we used three radiomics models from the published literature and two deep learning-based image segmentation models created using Tensorflow and Pytorch with standard architectures. Model Implementation Library is distributed as an open source, GNU-copyrighted software at https://www.github.com/cerr/CERR.

ABSTRACT. Accurate values of the beam quality conversion factor, kQ, are required for reference dosimetry measurements for external radiation therapy beams following dosimetry protocols like the AAPM's TG-51 and the IAEA's TRS 398. These factors can be determined using MC simulations by taking the ratio of calculations of absorbed dose to a small volume of water in a water phantom to calculations of the absorbed dose to the gas in an ion chamber model in a linac beam of quality Q, normalized to the same ratio in a cobalt-60 beam. The addendum to TG-51 recommends the use of MC calculated kQ factors, and these can be dependent on the accuracy of specifications provided by manufacturers. In this light, kQ factor calculations would benefit from a publicly available repository of accurate chamber models. This is particularly relevant for small detectors now used routinely for small-field dosimetry (compared to traditional reference-class chambers). This sensitivity is illustrated in this work by comparing the differences in kQ between a detailed and simplified model of the Exradin A26 micro ionization chamber, both provided by the manufacturer. Simulations were performed using the EGSnrc egs_chamber application, for a range of photon beams modeling the spectra of a Varian Truebeam. In the simplified model, the only changes were the electrode and teflon regions being extended toward the collector by 0.635 mm and 0.889 mm, respectively. These small changes result in systematically reduced kQ factors, on the order of 0.2%. Manufacturer specifications also indicate the "true" sensitive volume, based on the shape of electric field lines. This effect is investigated, but found to have negligible impact. The small differences observed in this study illustrate that millimeter-scale geometry deviations near the cavity may be detectable, but likely within the systematic uncertainties in calculated kQ.

ABSTRACT. The new RefleXion biology-guided radiotherapy (BgRT) system combines PET/CT imaging with 6MV radiotherapy for better localization and tracking the tumor during the treatment delivery. The maximum achievable treatment field size of the BgRT system is 2 cm or 3 cm in the IEC 61217 Y dimension at the SAD of 85 cm. The closest clinically used field size to the conventional reference field in this system is 10 x 2 cm2 at the isocenter. This field size does not meet the lateral charged particle equilibrium (LCPE) condition of the machine-specific reference (msr) field defined in the IAEA TRS-483 Code of Practice (CoP). Consequently, the TRS-483 cannot be directly used for the calibration of this system. In this study, two methodologies are proposed for calibration of the BgRT system. In the first method, the generic correction factors are calculated using EGSnrc Monte Carlo code for five chambers (Exradin-A1SL, A26, A14, A14SL and PTW-31010). In the second method, the TRS-483 protocol is extended to 2 cm field size. The beam quality specifier and equivalent square field are determined and used to determine the beam quality correction factor analytically (using the TRS-398). The beam quality correction is also corrected for the differences between the WFF and FFF beams (the volume averaging and water to air stopping power ratio corrections). The equivalent square fields size for the BgRT system is found to be 3.6 cm and 4.8 cm for the 10 x 2 cm2 and 10 x 3 cm2 fields respectively. Good agreement (to within 0.3%) is observed between the corrections calculated using the two approaches. If the MC correction is not available for the user’s chamber, the user can use the second approach for estimating the beam quality conversion factors provided that the chamber electrode is not made of high atomic number material.

ABSTRACT. The radial response function has been shown as a method, to unify radiation measurements results gathered with ionization chambers of different size and shape in small radiotherapy fields. The response function is a mathematical concept that relates the actual and observed signal with each other by convolution operation, and therefore the physical background of the response function is not self-evident. The aim of this study is to use Monte Carlo techniques to provide insight into the physical interactions and geometrical dependencies to understand the physical phenomena that govern the radial response function.

We use EGSnrc software (NRC, Ottawa, Canada) to simulate ‘measured’ dose distributions with the PTW34001 Roos parallel plate ionization chamber (PTW, Freiburg, Germany) and corresponding ‘true’ dose distributions in water. Additionally, observations with a modified Roos chamber, which materials except cavity air have been changed to water, are simulated. Details of the radiation beam and collimation with cone collimators are implemented with BEAMnrc application, while the detailed detector models are simulated with egs_chamber application. The radial response functions of studied detector models are solved using a non-parametric super-resolution deconvolution method.

When the Roos chamber response function based on simulated data is compared with corresponding response function based on experimental data, the functions are found to have strong similarities. The impact of chamber body material is demonstrated to be clear in shape and moderate in size.

ABSTRACT. The lack of standard for reference dosimetry in MR guided radiotherapy (MRgRT) has motivated the investigation of dosimeters response in magnetic field (B-field). Measurements and Monte Carlo (MC) simulations have demonstrated that in the presence of B-field the charged particles will be influenced by the Lorenz force modifying the dose response of ion-chambers. Among other effects of detectors in B-field that will increase the uncertainty on dose measurement are the air-gaps around the dosimeters. The aim of this work was to use MC simulations to investigate the effect of the air-gaps on the response of the alanine, which can act as an alternative reference dosimeter to ion-chamber. A bespoke waterproof holder, shaped like a Farmer-type ion-chamber, was built to accommodate alanine pellets (2.3mm height and 5mm diameter) for reference dose measurements in B-field. Air-gaps that exist inside the holder might modify the alanine signal due to the electron return effect. MC simulations, using EGSnrc, were performed to investigate this effect. The experimental setup, which involves the irradiation of alanine in a water phantom between the poles of an electromagnet (alanine orientated perpendicularly to both radiation beam and B-field, which was ranging from 0T-1.5T) using Co-60, 6MV and 8MV, was modelled. A detailed model of the air-gaps inside the holder was also included. Additional simulations comprise the validation of the experimental setup and the alanine model, including the holder. The MC model agrees with the experiments within 0.5%. Simulations demonstrated that the air-gaps inside the holder can introduce an effect of up to 0.6%. Because of the possible asymmetry in the positions of the pellets inside the holder, this effect must be included as a component in the measurement uncertainty, which increases with B-field strength and reaches 0.6% (k=1) at 1.5T.

ABSTRACT. The mean glandular dose (MGD) is the reference quantity for dosimetry in 2D mammography. We computed normalized glandular dose (DgN) coefficients (mGy absorbed dose/mGy incident air kerma) using GEANT4 toolkit version 10.00. Our aim is the determination of DgN coefficients for MGD estimation in mammography, taking into account a different skin thickness of the breast (1.45 mm) with respect to the currently adopted models (skin-equivalent tissue of thickness 4-5 mm). The breast was modelled as a cylinder with a semi-circular cross section with a radius of 10 cm, made of a homogeneous mixture of adipose and glandular tissue surrounded by a 1.45-mm thick skin layer. The compressed breast thickness ranged between 3 cm and 8 cm and the modelled breast with a glandular fraction by mass of 0%, 14.3%, 25%, 50%, 75% and 100%. The monoenergetic DgN coefficients were calculated for photon energies between 4.25 keV and 49.25 keV (step 0.5 keV) and then fitted with polynomial curves. Polyenergetic DgNp coefficients were then computed for spectra obtained for various anode/filter combinations as adopted in routine clinical practice: Mo/Mo 30 μm (25-40 kV), Mo/Rh 25 μm (25-40 kV), Rh/Rh 25 μm (25-40 kV), W/Ag 50 μm (26-34 kV), W/Al 500 μm (26-38 kV), W/Al 700 μm (28-40 kV) and W/Rh 50 μm (24-35 kV). DgNp coefficients were 6% higher than those provided in the reference literature, on average. The differences ranged between 18% and 30%; 50% of the computed coefficients differed by less than 10%. DgNp coefficients were provided as tables for varying compressed breast thickness and glandular fraction by mass. We developed also a computer code for generating user specific coefficients DgNp for user defined X-ray spectra up to 49 kV, calculated by spectral weighting from the dataset of monoenergetic DgN coefficients.

ABSTRACT. Introduction: Minibeam radiation therapy (MBRT) is a novel therapeutic strategy to reduce normal tissue toxicity, whose exploration was hindered due to its confinement to large synchrotrons. However, our recent implementation of MBRT in a widely spread small animal irradiator offers the possibility to perform comprehensive radiobiological studies. The aim of this work was to perform a complete dosimetric evaluation, to develop a set of dosimetric tools to reliably guide the biological experiments in the irradiator.

Methods: A Monte Carlo (Geant4 V10.01)-based dose calculation engine was developed. It was then benckmarched against a series of dosimetric measurements performed with gafchromic films in both homogeneous and heterogeneous phantoms. Those included depth dose curves, lateral profiles and peak-to-valley dose ratios. Two voxelized rat phantoms (ROBY, computer tomography) were used to evaluate the treatment plan of F98 tumor-bearing rats irradiations. The response of a group of seven animals receiving an unilateral irradiation of 58 Gy peak dose as measured in a water phantom was compared to a group of non-irradiated controls

Results: The good agreement between calculations and the experimental data allowed the validation of the dose-calculation engine. The latter was first used to compare the dose distributions in a rat's head computer tomography images and in a digital model of rat's head (ROBY), obtaining a general good agreement. Finally, we have used the developed tools to plan a pilot in vivo experiment.

Conclusions: The developed dosimetric tools were used to reliably guide the first MBRT treatments of intracranial glioma-bearing rats outside synchrotrons. The significant tumor control obtained with respect to the non-irradiated controls, despite the heterogenous dose coverage of the target, might indicate the participation of non-targeted effects. Dose escalation and the use of interlaced geometries to increase the overall tumor dose are the future directions.

ABSTRACT. With the advent of MR-guided radiotherapy (MRgRT), the response of detectors need to be characterised in the presence of magnetic fields (B-field). Monte Carlo (MC), like in conventional radiotherapy, will be an essential method for reference dosimetry in MRgRT by characterising physical properties of beams, radiation detectors and contributing valuable data for clinical dosimetry. The aim of this study was to benchmark MC calculations of quality correction factors for the presence of B-field, kQB, against measurement for suitable ion-chambers.

Measurements involve the irradiation of Farmer-type ion chambers placed in a water phantom between the poles of an electromagnet in 6 and 8MV x-rays on an Elekta Synergy, The chambers were irradiated both in perpendicular and parallel orientation to the x-ray beam and always perpendicular to B-field (from -1.5T to +1.5T). Alanine/EPR was used as a reference detector and corrected for the effect of the B-field.

MC simulations of chamber response were carried out using EGSnrc and benchmarked using an adapted Fano-test. Further simulations were performed to validate the experimental setup by comparing depth dose profiles between the magnetic poles with EBT3 films. Finite element modelling (COMSOL) was also used to determine the true collecting volume of the ion chamber.

A self-consistency of 0.1% or better was achieved for ion chamber response calculations in magnetic fields with an EM ESTEPE of 1% for all B-field strengths. Calculated depth dose profiles agreed with measured film data to within 1% or better along the whole profile. Initial benchmarking results indicate that EGSnrc can reproduce measured ion-chamber response and kQB to within 1% or better at all B-field strengths. Interestingly, in the perpendicular orientation with negative B-fields (electrons deviated towards the stem), closest agreement was obtained when the actual (geometrical) chamber volume was used in the simulations.

ABSTRACT. Purpose: To model and experimentally confirm the dosimetric impact of magnetic fields on clinical electron beams for different beam energies, magnetic field strengths and orientations in a water equivalent phantom. Results will be useful for assessing the feasibility of future MR-guided electron beam therapy. Methods: A permanent magnet apparatus was used to generate a strong magnetic field encompassing a solid water phantom. Gafchromic EBT3 film was placed in the phantom for dose measurements of 6, 12 and 20 MeV electron beams of a Varian Clinac 2100 C linear accelerator in transverse (main magnetic field perpendicular to primary beam), inline (main magnetic field parallel to primary beam) and a zero magnetic field setup. A multiple source Monte Carlo beam model was commissioned and applied to generate phase space files for the measured electron beams. Geant4 version 10.2 was configured to simulate the particle transport from the phase space files into the phantom geometry including the magnetic field. Results: Film data confirmed Monte Carlo predictions of substantial deflection of the electron beam in the transverse setup for all three initial energies (6, 12 and 20 MeV) and for two non-zero field strengths (Bmax = 0.2 T and 0.4 T) compared to the reference measurements without a magnetic field. For an inline magnetic field with Bmax = 0.4 T as well as 0.7 T, a dose enhancement of the profiles was observed compared to the B = 0 T measurement. Conclusions: An experimental and corresponding in-silico framework to measure and simulate clinical electron beams in magnetic fields of different strengths and relative orientations was established and successfully tested. Substantial dosimetric impacts of the magnetic field in transverse and inline orientation were observed and agreed with the theoretical prediction.

ABSTRACT. Medium energy X-ray beams can be calibrated by determining the absorbed dose to water at a depth of 2 cm in a water phantom. In order to correct an ionization chamber's response between the in-phantom and calibration conditions, an overall chamber correction factor is required. This correction factor accounts for the changes in chamber response due to the displacement of water by the chamber cavity, the presence of the stem and the change in incident photon energy and angular distribution in the phantom to that in air. The effects of a waterproof sheath (if required) are accounted for in another factor. In the context of the TRS-398 protocol update, This work aims to calculate chamber correction factors through Monte Carlo simulations and compare them to experimental values obtained at PTB with their recently-developed water calorimetry-based absorbed dose to water primary standard. The Monte Carlo software EGSnrc was used for all simulations in this work. The dose to water at a depth of 2 cm in a water phantom, air kerma, mean water to air mass-energy absorption coefficient ratios and chamber simulations were carried out for radiation qualities between 50 and 300 kV. Chamber simulations were carried out for the PTW TM30013, NE2571, A12, IBA FC65-G, IBA FC65-P and Exradin A12 chambers. All simulated chamber correction factors are within 2 % of unity. The simulated factors agree with experimental values to within 0.1-1.5 % for all chambers. Despite this, there are trends in the comparison that merit further investigation.

ABSTRACT. With the advent of radiotherapy guided by MRI, Monte Carlo (MC) simulations will play an essential role in determining dosimetry correction factors. Since high-resolution measurements require small cavity ionization chambers, MC models require careful validation with experiments. The aim of this study is to characterize small chamber response in the presence of B-fields using an experimental environment and compare results to detector responses simulated with MC. Small chambers are irradiated in water by a conventional 6 MV Elekta linac beam for various B-field strengths at two polarities: 0 T, +_0.35 T, +-0.5 T, +-1 T, +-1.5 T. The B-field is perpendicular to the irradiation beam and the chamber orientation. Two orientations are used, and the detector models are PTW31010, PTW31021, PTW31016 and PTW30122 with respective volumes of 0.125, 0.070, 0.030, 0.016 cm3. The experimental setup in the perpendicular orientation is simulated using EGSnrc using default and enhanced EMF macros with the following simulation parameters: ESTEPE=0.25, EMESTEPE=0.01, AE=0.512 MeV and AP=0.001 MeV. The sensitive volume is reduced to account for the inefficiency adjacent the guard electrode (i.e., dead volume) based on COMSOL simulations of electric potentials. The B-field affects the chamber response by up to 4.1% and 4.5% in the parallel and perpendicular orientations, respectively. The parallel-orientation experimental response shows an unexpected asymmetry of up to 0.81%. For the largest chamber, the MC simulation agrees only within 1.8%. For the smallest chamber, the discrepancy is 1.5% for the positive polarity and to 3.9% for the negative polarity. The reduction of the sensitive volume improves the agreement for the positive polarity and smallest chamber. Results suggest that the significant disagreement between experiments and simulations highlights the sensitivity of small chamber measurements in B-fields. The reliability of design specifications could play an essential role in dose response characterization, including the dead volume considerations.

ABSTRACT. Purpose: The development of volumetric scintillation detectors has enabled millimeter-resolution, water-equivalent, and real-time 3D dosimetry. The proven advantages of volumetric scintillation dosimeters could be further expanded by developing deformable scintillation detectors. The aim of this work is to identify and quantify the factors impacting the cumulative light measurement of a volumetric deformable scintillation detector.

Methods: Monte Carlo simulations using the Geant4 toolkit (v4.10.04) were performed to compare the optical photon transport from two different geometries. The first one consisted of a 27 cm3 cubic plastic scintillator immersed in a water-filled acrylic tank and the second of a 2 cm radius, 27 cm3, cylindrical plastic scintillator. Optical photons were created isotropically within the scintillator’s volume. Different refractive indexes were attributed to the surrounding medium to quantify the impact of refractive index mismatches. Finally, a shell-shaped scorer recorded the spatial distribution of the exiting photons at radial distances of 50 cm, 60 cm and 70 cm from the center of the phantom.

Results: 99.13% of the initial photons produced exited the phantom for the cubic scintillator whereas 94.13% of the initial photons exited the phantom for the cylindrical scintillator case. Missing photons were mostly trapped in the scintillators because of total internal refraction. The signal difference between both scintillators geometries ranged from 5.8±0.1% to 7.4±0.1% when facing the flat surface (0°) or the curved face (90°) of the cylinder, respectively. However, differences up to 30% arise radially to the acrylic tank’s corners. Changing the refractive index of the surrounding medium from 1.33 (water) to 1.58 (scintillators) decreased those differences under 1.4±0.1%. Light differences decreased to 4.2±0.1% (90°) and 6.9±0.1% (0°) at 70 cm.

Conclusion: The refractive indexes mismatches between the scintillator and its surrounding medium was demonstrated to be the greatest challenge to the development of a volumetric deformable scintillation detector.

ABSTRACT. With this work we present an innovative system for calculating occupational doses, as it is now being developed within the PODIUM (Personal Online DosImetry Using computational Methods) project. Individual monitoring of workers is essential to follow up regulatory dose limits and to apply the ALARA principle. However, current personal dosimeters are subject to large uncertainties, especially in non-homogeneous fields, like those found interventional cardiology. Workers in these fields also need to wear several dosimeters (extremity, eye lens, above/below apron), which causes practical problems. As the capabilities of computational methods are increasing, it will become feasible to calculate doses in place of physical dosimeters. In our concept system, operational and protection quantities are calculated by fast Monte Carlo methods. Our dose calculation accounts for the real radiation field (including intensity, energy and angular distributions) and for the relative position of different body parts of the worker. The movements of exposed workers are captured using depth cameras. This information is translated to a flexible anthropomorphic phantom, and then in Monte-Carlo simulations. For the moment this is done off-line, after the procedure is finished, and the parameters of the procedure are collected. For validating our system, we performed tests in interventional radiology rooms. In total, we followed 15 procedures in Cath-labs at UZ-VUB and CHU- Liège. An accurate analysis of the staff position was performed, and as a first step, we compared simulated Hp(10) and measured Hp(10) with electronic personal dosimeter (EPD) during an angiography procedure for some of these procedures. The results showed good agreement between the calculated doses and the ones measured by the EPD dosimeter. This method has big advantages in interventional radiology workplaces where the fields are non-homogeneous and doses to staff can be relatively high. This method can also help in ALARA applications and for education or training.

ABSTRACT. TBA

ABSTRACT. Introduction Brachytherapy is an integral component of the treatment of cervical cancer. Outcome in terms of disease control and toxicity is a function of dose to the tumor and organs at risk (OARs). By utilizing intensity modulated brachytherapy (IMBT), where rotating metallic shields are introduced into brachytherapy catheters, dose delivery can be optimized to better conform to the tumor while reducing OARs doses. In this work, we investigate novel rotating shields, compatible with MRI-compatible tandems used for cervix brachytherapy. Three unique shield models were evaluated using the traditionally used Ir-192 source. Additionally, Se-75 and Yb-169 were investigated as alternative sources.

Materials & Methods Three shields for use in IMBT were modeled and simulated in RapidBrachyMCTPS, a MC based treatment planning system, built on the Geant4 toolkit. Shields were designed to fit inside MRI-compatible tandems. Shield material was set to either platinum or tungsten. Simulations were performed in a (40 cm)3 water phantom. The active core of the source was replaced with Ir-192, Se-75 and Yb-169. Absorbed dose was normalized at 1 cm from the tandem at 0° on the transverse plane. Transmission factors (TFs) were calculated and defined as the dose ratio at 1 cm on opposite sides of the tandem.

Results TFs are favorable for IMBT and ranged between 4.1%-24.1% for Ir-192, 0.7%-8.6% for Se-75 and 0.1-2.7% for Yb-169. On average, platinum shields attenuated 34% more than tungsten. Shields had a TF of at least 50% over an average arc of 245°, 208° and 283° for Ir-192, Se-75 and Yb-169, respectively.

Conclusions This study has quantitatively assessed the influence of various IMBT tandem shields on dose attenuation. By developing and making cervix IMBT clinically available, the therapeutic potential of brachytherapy will be significantly improved.

ABSTRACT. Purpose: To update the Carleton Laboratory for Radiotherapy Physics (CLRP) TG-43 dosimetry database for 51 low- and high-energy brachytherapy sources utilizing the EGSnrc Monte Carlo open-source application egs_brachy. A comprehensive set of TG-43 parameters including dose-rate constants (DRCs), radial-dose (g(r)) and anisotropy functions (F(r,ɵ)), along-away dose-rate tables, and primary- and scatter-separated dose-data are evaluated and compared to the literature.

Method: The detailed geometry modelling of each brachytherapy source is systematically re-checked and updated. Air-kerma strength per history is calculated in voxels representing NIST’s WAFAC detector’s solid angle. Full scatter water phantoms in cylindrical coordinates are used for dose calculations. To calculate source-volume-corrections with more precision for voxels containing part of the source, random-point-number-densities from 10^4-10^11 cm-3 are investigated. TG-43 parameters for all sources are extracted.

Results: In general, results improve overall statistical uncertainties, volume-corrections, and possible source geometry errors compared to previous CLRP calculations. Some inaccuracy in geometry models of sources (e.g., STM1251, Thinseed-9011, PharmaSeed-ST1 & ST2, selectSeed-130.002, AgX100) lead to changes in DRCs of up to 2.6%, g(r) deviations up to 2.6% close to the source (0.05

Conclusions: Some small changes in geometry models have been implemented. The requirement for higher random-point-number-density values to avoid large uncertainties from volume-corrections has been identified; since the required changes are in very high-dose regions close to source they don’t affect clinical treatment. This comprehensive update provides the medical physics community with more accurate TG-43 dose evaluation parameters, as well as fully-benchmarked source models that will be distributed with egs_brachy.

ABSTRACT. Purpose:  To evaluate the dosimetric parameters for a new 169Yb source model designed for intensity modulated brachytherapy (IMBT) with the recently developed AIM-Brachy system. Methods: Simulations were performed using RapidBrachyMC, a validated Monte Carlo (MC) code for brachytherapy applications based on Geant4. TG-43U1 parameters were generated for the source model: air kerma strength, dose rate constant, radial dose function, and 2D anisotropy function. In addition, dose distributions were simulated for the source in combination with a platinum shield with an emission window of 180° and a maximum thickness of 0.8 mm. A dosimetric error analysis was included to estimate the uncertainties in the dosimetric parameters. Results:  The total photon yield was 3.803 photons per disintegration. The air kerma strength per unit activity was 1.22±0.03 U mCi-1. The dose rate per unit activity at 1 cm off-axis was 1.47±0.03 cGy h-1 mCi-1. The dose rate constant was 1.20±0.03 cGy h-1 U-1. Tabulated values for the radial dose function and 2D anisotropy function were provided. The radial dose function reaches a maximum of 1.17 at r = 5 cm. A fourth order polynomial function was obtained for the radial dose function by curve fitting. The 2D anisotropy function ranged between 0.45 and 1.0 over all polar angles at r = 1 cm. The 2D anisotropy function decreased at low polar angles, and increased with increasing distance from the source. The platinum shield reduced the dose on the shielded side at 1 cm off-axis by 82% compared to the unshielded side. Conclusion: Dosimetric parameters were calculated for the bare and shielded 169Yb source model. Validation of dose distributions will be performed with measurements in-air and in-water. 

ABSTRACT. The INTRABEAM electronic brachytherapy source represents a convenient alternative to conventional radionuclides since it delivers a low energy (50 kVp) x-ray beam, reducing the regulatory and shielding requirements without compromising the dose delivery rates. Despite its benefits and extended use as a brachytherapy source, it has not been characterized according to the AAPM TG-43 specifications, restricting its modeling in commercial treatment planning systems (TPS) for efficient dose calculations. In the present study, a brachytherapy dosimetry characterization of the INTRABEAM source based on the AAPM TG-43 protocol is presented. The TG-43 parameters were calculated using egs_brachy, a user code of EGSnrc, which allows rapid dose calculations via a tracklength estimator. The azimuthal symmetry of the geometry was exploited by scoring the dose in a set of annular bins of water in 0.5 mm steps in the radial and axial directions from the source tip. The rectangular grid of data was then transformed to polar coordinates varying in steps of 1 mm from 0.4 to 5 cm and steps of 1° from 0° to 180° to determine the TG-43 parameters. Depth dose results in water along the source axis were compared with previous dosimetric studies of the INTRABEAM resulting in differences lower than 3%. The radial dose function diminished close to the source (< 1 cm) with a steep gradient higher than that of conventional brachytherapy radionuclides (192Ir, 103Pd and 125I), but it is partially flattened at larger distances with a similar fall-off as the Xoft electronic brachytherapy source. The simulated polar anisotropy values were mainly uniform along θ = 0° and gradually decreased close to the border of the source, affected by the beam attenuation in the elements of the source walls. The TG-43 parameters provided in this paper serve as preliminary data required for 3D TPS.

ABSTRACT. Purpose To develop a framework for performing dynamic model-based dosimetry in target organs exhibiting geometry changes during brachytherapy. The framework is applied to an investigation of the dosimetric effects of edema during permanent implant prostate brachytherapy (PIPB).

Methods From a single post-implant CT image, the framework generates intermediate geometries corresponding to different edema magnitudes. Prostate swelling is modelled by expansion of the prostate contours with consistent shifting of seed locations and calcifications. Dose distributions are calculated using the MC code egs_brachy and are subsequently mapped to a common geometry using an energy mapping technique. The cumulative dose for a specific temporal model of edema resolution is calculated by weighting each mapped dose distribution by the proportion of total decays that occur within the corresponding edema magnitude step. Ten edema resolution models from the literature are implemented. Dynamic MC calculations are compared to static 30 day post-implant MC and TG43 calculations for 20 PIPB patients.

Results The minimum number of intermediate geometries required for convergence of calculated metrics is found to correspond to intervals of prostate volume approximately equal to 6% of the maximum prostate volume. Static MC dosimetry calculates considerably reduced dose coverage compared to TG43 and dose coverage predicted by dynamic MC is further reduced. The discrepancy between a static TG43 dosimetry and a dynamic MC dosimetry is found to depend on both the choice of edema model and the initial treatment geometry.

Conclusion Beginning with a single post-implant CT, dynamic model-based dose accumulation can be performed with an arbitrary number of intermediate geometries modelling changes in patient geometry and seed positions during the treatment course. The framework was successfully applied to 20 PIPB patient treatments considering 10 unique edema models each, demonstrating considerable discrepancies with static calculations. Edema must be considered for accurate dosimetry of PIPB treatment.

ABSTRACT. bGPUMCD is a GPU accelerated Monte Carlo dose calculation algorithm for brachytherapy. We present here the first steps in the validation of this algorithm for clinical use by evaluating its performances for the WG-MBDCA-Ir192 generic source defined by the AAPM working group on model-based dose calculation algorithms in brachytherapy (WGDCAB). We start by comparing against the TG43 reference data. We then compare its performances against the reference data of the first three cases defined by the WGDCAB. The results show that the gL(r) and F(r, theta) data obtained with bGPUMCD are in accordance with the TG43 data published for the generic source with an error no greater than 2%. Furthermore, The data for the test cases show that the relative error compared the the MCNP6 reference data is under 5% for all test cases up until 10cm away from the source. Each simulation result was obtained with a generation of 1e9 primary gamma in a voxelized geometry of 201x201x201 voxels of 1mm3. Each simulation took under 20 minutes to complete on an Nvidia Titan X (2016).

ABSTRACT. Intraoperative radiotherapy (IORT) using The INTRABEAM System (Carl Zeiss Meditec AG, Jena, Germany), a miniature 50 kVp x-ray source, has shown to be an effective modality in the treatment of breast cancer. However, recent investigations have suggested that the absorbed dose to water delivered by the INTRABEAM can deviate significantly from the reported dose. In this work, the effect of spherical applicator size on the delivered dose was investigated by performing ionization chamber measurements in a water phantom. The dose at the applicator surface and 1 cm depth was calculated using two methods: a) the manufacturer-recommended formula (Zeiss), and b) the CQ method, which uses a chamber air-kerma to dose to water conversion factor calculated with monte carlo (EGSnrc). These results were normalised to a prescription dose of 20 Gy to the applicator surface following the TARGIT breast protocol. The dose at 1 cm depth from the applicator surface for the Xoft Axxent, a competing miniature x-ray source, was also calculated for comparison.

The dose delivered to the INTRABEAM applicator surface was found to range from 25.2 Gy to 31.7 Gy according to the CQ method for the largest (5 cm) and smallest (1.5 cm) diameter applicator, respectively. The results of the Zeiss method were 7% to 10% lower. This suggests that TARGIT breast patients are receiving 26% to 59% greater dose than the 20 Gy prescription depending on the size of applicator used. This dose inaccuracy could have implications for any studies wishing to compare outcomes of IORT treatments performed with INTRABEAM and other radiation modalities.

ABSTRACT. Bone is recognized as a common site for metastasis. Prostate, breast and lung cancers have high prevalence and are most probable to spread to the bone. One of the typical sites of secondary bone cancers is the spine. Vertebroplasty is a process by which vertebral bone tumours can be removed surgically, and the cavity is filled by a bone cement to maintain structural integrity. Adjuvant external radiation therapy is applied following vertebroplasty in order to destroy residual tumour cells and plays a role as pain relief. In this study, we are proposing a novel approach to bone brachytherapy as an alternative to external radiation therapy. Although conventional brachytherapy is based on sealed radiation sources, in this research the radioactive material is mixed with the bone cement and injected to fill the cavity. For vertebral bone cancer, the spinal cord is the main organ at risk that must be considered for this procedure. Three properties of potential radioactive sources, namely the half-life, radioactive decay and intensity of energy, are important to consider for this novel approach. 125I, 131I, 192Ir, 103Pd, 169Yb and bioglass bone cement have been considered as radioactive sources and bone cement, respectively. This investigational work is performed in the Monte Carlo code TOPAS MC. The goal of this research is to introduce a novel bone brachytherapy technique with the goal of delivering maximum dose to the residual cancerous cells and minimal dose to surrounding organs at risk such as the spinal cord. Along with the dosimetry consideration, the mechanical properties of modified bone cement will be considered.

ABSTRACT. This study aimed to design a module type sealed source applicator for patient specific brachytherapy. According to tumor shape, the sealed source was arranged to cover to skin. To deliver homogeneous dose to target, the shape of skin attachment was designed. Dose distributions were evaluated by a Monte Carlo method to optimize the structure and materials of a module type sealed source applicator. Beta and low-energy photon and emitter were selected as isotope. The internal structure of applicator was designed in consideration of shielding, sealing and encapsulation. For cylindrical and hexagonal models, the dose distribution was calculated at flat surface using various source. The minimum percent dose of between neighboring applicator was 12.5% and 22.2% for cylindrical and hexagonal models, respectively. The dose uniformity at 0.125 cm depth was 0.217 and 0.363 for cylindrical and hexagonal models, respectively. In terms of dose uniformity, hexagonal geometry turned out to be superior to cylindrical geometry. The applicator with hexagonal shape have flatter dose distribution than that with cylindrical for single. When ten applicators was arranged to cover the surface, dose distributions of hexagonal type has more uniform than those of cylindrical type. To deliver more uniform dose, the thickness of encapsulation is needed to be minimized.

ABSTRACT. This paper presents the comparison of absorbed doses in brachytherapy plans and Monte Carlo simulation for brachytherapy treatment of a female patient with cervix carcinoma. At the Department of Brachytherapy at the Clinical Center Kragujevac, the microSelectron after loading device is used for intracavitary brachytherapy in the HDR regime. This device uses a miniature radioactive source 192Ir in the form of a cylinder, active dimensions of 0.6 mm × 3.5 mm, and a high initial activity of about 370 GBq. Before therapy, computer planning is performed, which represents a computer reconstruction of the position of the source guide in the patient based on two radiographic images, and isodose planning in relation to the desired dosimetry points. Essential planning data are the daily dose and number of fractions. In this case, the daily dose is 700 cGy and is delivered in three fractions once a week. This means that the duration of this brachytherapy treatment will be a total of three weeks. Monte Carlo simulations by using MCNP6 code version 2.0 were applied for brachytherapy treatment to estimate the dose distribution in uterus and several critical organs at risk (bladder and colon). The MCNP tally F6 (MeV/g) was chosen due to easy convert energy deposition to absorbed dose. The computational ORNL and voxel phantoms were used to prepare input files which simulate brachytherapy. By comparing measured and calculated values, it can be seen that Monte Carlo techniques are a powerful tool for application in brachytherapy planning References [1] MCNP6.2 Monte Carlo N–Particle Transport Code System; Version 6.2. Report LA-UR-17-29981. Los Alamos, 2018. [2] Eckerman K F, Cristy M, Ryman J C. Oak Ridge National Laboratory. TN 37831, USA; 1996. [3] ICRP 110: Adult Reference Computational Phantoms. Realistic reference phantoms: An ICRP/ICRU joint effort. Ann. ICRP 39 3-5; Elsevier. 2009.

ABSTRACT. Introduction An increasing number of cardiovascular patients with cardiac implantable electronic devices (CIED) are encountered in Radiotherapy. Irradiating CIEDs may cause device failures or malfunction. And according to the AAPM TG-34 the CIED’s cumulated dose should be estimated before or measured during treatment and should not exceed 2 Gy. Mouton et al. reported an important failure at cumulative dose of 0.15 Gy. It is difficult to precisely estimate the dose to CIEDs since it is usually located out of the treatment field where the accuracy of the treatment planning system (TPSs) is known to decrease. Measuring this dose using dosimeters needs special equipment and phantoms. Monte Carlo (MC) simulation can provide an elegant alternative to estimate the cumulative dose at CIEDs. Materials & Methods Five patients with CIEDs were treated on Varian Trilogy linear accelerator using IMRT or VMAT techniques. The treatment plans were calculated using Varian Eclipse TPS. The plans were also simulated with MC using PRIMO software. Calculated and simulated doses were normalized to the mean dose at the planning target volume (PTV). Both dose distributions for each plan were compared within the treatment fields using the Gamma index methodology (2mm, 2%). Then, cumulated doses at CIEDs were compared. Results Simulated doses matched well with calculated ones within the treatment fields. In all cases more than 95% of Gamma points passed the criteria of (2mm, 2%). As the distance between the treatment fields and the CIEDs increases the cumulated dose at CIEDs decreases and the discrepency between MC and TPS calculations increases. However, this is not critical since the doses are bellow the CIED tolerance dose. Eclipse tends to under estimate the dose at CIEDs unless PTV is partially or completely located in the lungs. In this particular case, the calculated dose was higher than the simulated one.

ABSTRACT. Introduction: Monte Carlo beam modelling is a tedious and time-consuming work but defines the accuracy of dose calculation. We present an automated method to determine accurate beam model parameters of a proton treatment beam. Modern proton therapy centres employ the scanned ion beam delivery (SIBD) technique, where narrow proton beams so-called pencil beams (PB) are scanned over the target area.

Material & Methods: Experimental measurements were performed for 5 proton energies in the clinical energy range (62-253MeV). Integrated depth dose profiles in water and spot sizes in air at various iso-centre distances (ISD) in 10cm intervals were measured. Using a unique feature, that the treatment nozzle can be removed, two data sets were acquired, with and without Nozzle in the beam path. Proton ranges and spot sizes at the same positions were derived from GATE/Geant4 based Monte Carlo simulations using a first guess of the initial energy, energy distribution, spot size, divergence and emittance. These beam parameters were then automatically adapted by minimizing the sum of squared differences between measured and simulated values using the gradient-based Nelder-Mead simplex algorithm. Beam models with and without clinical nozzle were created.

Results: Simulated beam ranges in water were within 0.05mm of measurements, which is below the measurement uncertainty. Beam behaviour was accurately modelled for converging (w/o Nozzle) as well as for diverging beams (with Nozzle). Spot sizes using the automatically derived beam model without nozzle agreed well with a maximum deviation of 0.7mm for the lowest energy. For the beam model including the clinical treatment nozzle, maximum deviations of less than 0.3mm were observed.

Conclusions: Beam optics parameters were automatically derived for GATE/Geant4 for two proton beam lines, enabling the creation of clinically acceptable beam models in very short time. Furthermore, the flexibility of this approach allows its use for generic beam lines.

ABSTRACT. Introduction Complex irradiation techniques such as VMAT require appropriate QA to ensure that the intended treatment for individual patients will be correctly delivered. This should include TPS dose calculation verification, for which Monte Carlo methods are often considered as gold standard. In the different Monte Carlo software available for radiotherapy applications, PRIMO appears to be the only one self-containing (geometry included and validated), free, publically available, and able to handle VMAT plans for TrueBeam linacs. By combining dedicated features available in beta-version of PRIMO and in-house scripts, process of dose verification was automated with PRIMO. The aim of this study is to check if PRIMO is suitable for systematic dose verification of VMAT plans. Methods As TrueBeam beam production part is not implemented in PRIMO, 6 MV PSF provided by Varian was used. We used dedicated features in a PRIMO beta-version combined with in-house scripts to automate VMAT plan dose estimation with PRIMO. We compared dose from Eclipse 13.7 (AcurosXB, dose to medium), with PRIMO (DPM) for 50 VMAT plans with Gamma Index 3%/3mm and 2%/2mm, including noise reduction. Failing plans with DPM were recomputed with PENELOPE. Results PTV median dose is on average 0.95% higher with DPM than Acuros. Combining DPM and PENELOPE, all 50 plans pass 3%/3mm analysis. 45 plans pass the 2%/2mm analysis. Human time needed is a few seconds per plan. Calculation time vary from 19 min to 2 h 17 min, with an observed correlation to PTV volume. Conclusions PRIMO v0.3.1 can perform systematic VMAT plan Monte Carlo dose verification, for TrueBeam and Acuros, using Varian PSF. This requires low setup efforts. Combining dedicated features in PRIMO beta-version with in-house scripts automates the dose verification, with the perspective to include it systematically in pre-treatment QA.

ABSTRACT. Development of the TOPAS-based independent assessment system including an automated DICOM-RT interface is performed to accurately verify the pencil beam scanning (PBS) treatment plan. IBA universal beam nozzle installed at the Korea national cancer center (KNCC) was modeled by TOPAS based on manufacturer’s information. In the MC commissioning, beam sizes and integrated depth dose (IDD) were matched to measured data. DICOM-RT interface was developed to automatically set the beam condition and patient model by extracting the patient-dependent parameters stored in a treatment plan. In KNCC, beam sizes and IDDs were measured for 28 and 141 proton energies, respectively. As the results of MC commissioning, beam sizes and IDDs were well matched for all energies within ± 0.5 mm and ± 1 mm, respectively. Maximum deviation of the beam size and range is 0.43 mm at 227.1 MeV at isocenter and 0.69 mm at 225.1 MeV, respectively. Also the beam positions changed by the bending magnets on the isocenter were consistent with the planning at three proton energies (100.51, 170.12, and 210.87 MeV). With the validated proton beam parameters, the in-house system is able to accurately calculate the patient dose distribution through the DICOM-RT interface and assess that with a dose volume histogram and three-dimensional gamma index analysis. The developed TOPAS-based independent system shows the potential to precisely evaluate the PBS treatment plan.

ABSTRACT. The global electron cutoff energy (ECUT) equal to 700 keV is often used for the Monte Carlo treatment planning dose calculations. Previous studies [1], performed in heterogeneous phantoms, reported that DVHs for targets including large low-density regions may be significantly affected by the use of the high value of ECUT. In the present work, we investigate the effect of ECUT on the calculation accuracy of Monte Carlo Treatment Planning for lung patients. Clinical lung cases including large low-density cavities in the target were selected for this study. Dose calculations were performed on the CT based phantoms using MCSIM code, different number of histories and ECUT equal to 700 keV. MCSIM was modified to score independently the dose deposited on air voxels of the target. And consequently determine the contribution of the air dose uncertainty to the total uncertainty of the target dose. Calculations with different number of histories (RUN1 and RUN2) generated average dose value on the target with uncertainty lower than 2%. The value of Dmax occurred in the air voxel, and the uncertainty on Dmax varies from 2% (RUN1) to 11% (RUN2). Comparing the target DVH(RUN1) and target DVH(RUN2) we observed maximum differences of about 10%. While the differences between DMH(RUN1) and DMH(RUN2) are less than 2%. DVHs for lung targets including large (about 45 %) low-density regions may be significantly affected by the ECUT value used for the Monte Carlo dose calculations. In contrast, DMH was not affected by high ECUT in the treatment plan evaluation, and thus we can use high ECUT and DMH to improve the efficiency and accuracy of Monte Carlo based Treatment Planning for lung patients. 1- G Mora, A Eldib, J Li, C M Ma. AAPM 2017. Influence of Electron Transport Parameters on MC Treatment Planning

ABSTRACT. Purpose : The critical organ of patient was exposed for a long time during interventional procedure compared with another diagnostic radiography. To reduce dose of patient, it is designed the variable collimator system differently from previous study for patient safety through reduction of unnecessary patient dose in interventional procedure.

Material & Design : The variable collimator system of interventional procedure was designed by commercial engineering software and it is used universal simulation code (MCNPX, LANL, USA) to evaluate shielding rate of variable collimator system. The shielding material was SKD 11 alloy tool steel that have 7.89 g/cm3 Using the tally card 6, we evaluated shielding rate of variable collimator system depending on the energy.

Results : We designed two modules to develop variable collimator system of interventional procedure. First module is the multileaf collimator possible to change shape of exposure field and the module was attached to the X-ray tube’s head of C- arm at the exit slit of the X-ray beam. Second was user interface module. User can designate resign of interest (ROI) as treatment area was exposed for a long time during interventional procedure and set shielding resign of critical organs. And the shielding rate was that 98 % (80kVp), 96 % (100kVp), 94 % (120 kVp) using the Monte Carlo simulation, according to energy.

Discussion & Conclusion : Previous studies on dose reduction have been carried out using shielding materials (Lead (Pb), Bismuth (Bi)) and used selective lead collimator has limitation that was invariable shape, provided that the shielding itself does not affect the diagnosis. Our designed system was possible to change shape, freely and reduce the unnecessary dose of patient. The study would be look forward to increase the quality of life for patient undergoing interventional procedure.

ABSTRACT. A very fast coupled photon/electron Monte Carlo dose algorithm has been released in the latest version of the treatment planning system RayStation (8B). It is a new code base written directly for GPU in CUDA. The Monte Carlo code transports photons, electrons and positrons in a voxelized patient geometry. Photons are transported with Woodcock tracking and Compton scattering, photoelectric absorption and pair creation events are simulated. Electrons are transported with a class II condensed history method, where Møller scattering and bremsstrahlung are treated as discrete events. The modelling of multiple scattering uses the theory of Goudsmit-Saunderson. Cross sections and electron stopping power are computed for any material composition. Positrons are treated as electrons except that they can undergo annihilation, either at rest or in flight. The fluence from the linac head is computed analytically in the same framework as for the Collapsed Cone dose engine. The photons and contamination electrons are subsequently sampled form the computed fluence distribution for further transport in the patient. The algorithm has been successfully validated for photons against measurements and against EGSnrc for energies between 500 keV and 20 MeV in various materials and geometries. Final dose can be computed to 1% uncertainty within 5s for a 7 beam IMRT prostate case with 3mm voxels. A dual arc lung case with 2mm voxels take approximately 16s for 1% uncertainty.

ABSTRACT. The EGS5-MPI is a parallelization package for the EGS5. A file IO error was often caused in a EGS5-MPI simulation, when each MPI process called the HATCH subroutine with the USEGSD option. Because the HATCH subroutine included in the EGS5 generated a large-size file to calculate electron multiple scattering with Goudsmit-Saunderson Ditribution (GSD) model. In the present study, we revised the EGS5-MPI to avoid the file IO error. We also demonstrated the MC simulation of a high-energy electron beam from a clinical linac with the USEGSD option.

ABSTRACT. Background and Objectives: In Bangladesh, a huge number of breast cancer patients are being treated using a multimodality approach of tangential beam 3D-CRT and IMRT and VMAT. It has been seen in any case that the dose of OARs is decreasing by using Hybrid IMRT than other multimodality approaches. The purpose of this study is to investigate heart doses, lung parameters (V5, V10, V20, and V30) and contralateral breast respectively in Hybrid-IMRT plans according to RTOG protocol. Materials and Methods: More than eight patients with left breast carcinoma who underwent whole breast irradiation was planned using Hybrid-IMRT. Target and critical structures were delineated by one radiation oncologist according to protocol. Hybrid IMRT plans were created for prescription of 40.5Gy in 15 fractions. 3DCRT tangential open fields delivered 80 percent of the prescribed dose and two IMRT fields delivered 20 percent of the prescribed dose. IMRT fields were optimized for open field base doses, MLC was used in the two open fields to block OARs as much as possible without compromising PTV coverage. Results: Hybrid IMRT average mean heart dose was (1.64 ± 0.55) Gy, V5(%) was 3.55 ± 2.65, V20 (%) was 2.5 ± 0.55 & V30 (%) was 1.05 ± 0.65 respectively. Hybrid IMRT average mean left lung dose was (7.5Gy ± 2.24) Gy. V5 (%) was 28% ± 4.5%, V10 was 20.24% ± 5.6%, V20 (%) was 14.77% ± 6.9% and V30(%) was 11.98% ± 2.8% respectively. However, hybrid IMRT offered excellent results comparable to that of full IMRT. Conclusion: In the context of limiting mean and low dose to lung and heart, Hybrid-IMRT is found to achieve good OAR sparing with acceptable PTV coverage for free breathing, left breast irradiation. The most important advantage of Hybrid-IMRT is treatment planning time better than IMRT.

ABSTRACT. Introduction: This study aims to determine the severity of the effects on calculations of dose distributions caused by varying statistical uncertainties in a Monte Carlo based treatment planning system (TPS).

Materials & Methods: For this study, multiple archived patient plans were selected for recalculation. Treatment sites included prostate and head and neck. These plans were each recalculated seven times with varying uncertainty levels using Elekta’s Monaco Version 5.11.00 Monte Carlo Gold Standard XVMC dose calculation algorithm. The statistical uncertainty values ranged from 0.5% to 3% at intervals of 0.5%, with a final calculation at the maximum allowable statistical uncertainty of 5.0%. The grid spacing was set at 3mm for all calculations. Indices defined by the RTOG describing conformity, coverage, and homogeneity were recorded for each recalculation. Additionaly, basic PTV DVH statistics were recorded.

Results: Of the observed DVH statistics, the most affected was the volume of the target receiving the prescription dose (TVPI). There was a negative correlation between the statistical uncertainty and this volume. As the uncertainty increased, the TVPI decreased. In some cases, the difference was as large as 19.3 cubic centimeters. On average, the maximum dose to the target increased with an increase in uncertainty while the coverage and homogeneity decreased. As the uncertainty increases, the isodose lines become more irregular, and the target coverage and dose homogeneity both suffer.

Discussion & Conclusions: Increasing the statistical uncertainty per calculation correlates with a decrease in TVPI and produces more irregular isodose lines. While time is spared using higher uncertainty levels, the plan quality is decreased. The default calculation properties for Monaco are 0.7% uncertainty per calculation, with a 3 mm grid spacing. Additional plans will be recalculated to further improve statistical accuracy.

ABSTRACT. Introduction As the ionizing radiation can be used in the treatment of cancers, it also caused to have serious injuries to the patient bodies. Radiation-induced secondary cancers in patients under radiation therapy are one of the damages associated with these beams. The excess relative risk (ERR) and excess absolute risk (EAR) radiobiology models used to evaluate the secondary cancers rate organs under radiation therapy for breast cancer.

Materials & Methods 60 patients with left breast cancer were chosen and treatment planning was performed using TPS. Dose-volume histogram curves and patients prescription doses, the average received dose are achieved to their critical organs including heart, lung, liver, thyroid and contralateral breast, respectively. Finally using the age of patient under radiation and the ERR and EAR models that have been proposed by the BEIR VII Committee, the incidence of secondary cancers in the mentioned organs was obtained. Results The results indicate that the ERR model predicts the relative risk of breast radiation dangers in critical organs such as lung, thyroid, heart and liver, 21.23, 5.015, 1.58 and 1.36, respectively. The risk ratio of patients in the lung was 4.23, 13.43, and 15.61 relative to the thyroid, heart, and liver, respectively. The EAR model has predicted the excessive risk of lung, heart, contralateral breast and liver organs as 105, 45.75, 15.85 and 4.35, respectively that the excessive risk of lung relative to the heart, contralateral breast and liver organs is achieved as 4.15, 5.84 and 11.40 respectively based on the excess risk value. Discussion & Conclusions The rate of secondary cancer in the lung is higher than other organs of patients with breast cancer and then it is most likely to increase in heart, thyroid and contralateral breast, respectively. The rate of secondary cancers in the liver is negligible compared to other organs.

ABSTRACT. Purpose: This study is to compare the dosimetrical and radiobiological parameters among various volumetric modulated arc therapy (VMAT) techniques using restricted and continuous arc beams for the left sided breast cancer. Materials and Methods: 10 patients with left-sided breast cancer without regional nodes were retrospectively selected and prescribed the dose of 42.6 Gy in 16 fractions to the planning target volume. For each patient, three plans were generated by using the EclipseTM system (Varian Medical System, Palo Alto, CA) with one partial arc VMAT (1pVMAT), two partial arcs VMAT (2pVMAT), and two tangential arcs VMAT (2tVMAT). All plans were calculated by using anisotropic analytic algorithm and photon optimizer with 6 MV photon beam of VitalBEAMTM. The same dose objectives for each plan were used to accomplish fair comparison during optimization. Results: For PTV, dosimetrical parameters such as the HI, CI, and CN were superior in 2pVMAT than those in both techniques. V95% which indicates PTV coverage was 91.86%, 96.60%, and 96.65% for 1pVMAT, 2pVMAT, and 2tVMAT, respectively. In most of OARs, 2pVMAT was reduced significantly the delivered doses compared with the other techniques, excluded in doses to contralateral lung. For the analysis of radiobiological parameters, the significant difference of NTCP was in ipsilateral lung while no difference was observed in the other OARs. Conclusion: Our study founded that 2pVMAT was better plan quality and normal tissues sparing than 1pVMAT and 2tVMAT, although not for all parameters. Therefore, the 2pVMAT could be considered the priority choice on treatment planning for left breast cancer.

ABSTRACT. Purpose: The aim of this study is to develope and evaluate the volumetric independence dose calculation (vIDC) system for brachytherapy by comparison with practical cases.

Method: Five patients for squamous cell carcinoma of the cervix were applied image-guided adaptive brachytherapy (IGABT) using 192Ir HDR-source and adapted tandem and ring technique. Oncentra Brachy treatment planning system (TPS) is used to establish treatment planning for cervical cancer. Dose of volume which is enclosed by 90% of the prescribed dose (V90) of high-risk critical target volume (HR-CTV) is established to 550 cGy. Dose constraints of organ-at-risks (OARs) are implemented to minimum dose which delivered to 2 cc volume (D2cc) for each OARs. Main formalism of vIDC was based on the AAPM TG-43UI. To evaluate the accuracy of the vIDC, comparisons of point-dose were performed. In addition, parameters of dose-volume assessment based on DVH such as D10cc, D5cc, D2cc, D1cc, D0.1cc, V90, and V100 were calculated to evaluate the treatment planning. Each evaluation of the vIDC was performed with varying of the dose grid size such as 2.5 x 2.5 (G2.5), 1.0 x 1.0 (G1.0) cm2, and resolution of treatment planning images (default) to verify the grid size effect.

Results: Point-doses of original coordination were presented the most similarity to those of TPS calculation. The percentage difference (%diff) between these values were within -1.79. The averaged V90 of HR-CTV is indicated as 514.80 cGy which is similar to prescripted dose of treatment planning. In contrast, the V90 of calculation by vIDC are 448.23, 486.87, 497.20 cGy for G2.5, G1.0, and default grid-szie, respectively. In OAR, overall D2cc are acceptable to dose constraints of planning.

Conclusion: It will be possible to verify the accuracy of the treatment planning in IGABT and expect the quality of treatment to be increasing by using vIDC.

ABSTRACT. Purpose: The aim of this study is to determine dosimetric parameters of the vacuum-sealed miniature X-ray tube (mXT) to calculate the distribution of radiation doses in brachytherapy. Method: Spatial dose distribution from the mXT source was measured with EBT3 films located at 1 cm below the mXT source in a manufactured acrylonitrile-butadiene-styrene (ABS) phantom. . The dose distribution in virtual ABS phantom was also calculated with the MCNP6.1 code at the same condition of the measurement. The MCNP simulation was validated by comparing the measured dose distribution to the calculated dose distribution in the ABS phantom. Dose distribution in water at the same source geometric conditions in the ABS phantom was calculated with the MCNP simulation, then dose-rate constant, radial dose function, and anisotropic function were derived from the calculated dose distribution. Results: The relative doses from MC simulation were well agreed with measured doses, and the root mean square of the difference between normalized measured doses and calculated doses was 0.0287. The measured dose rate at 1 cm below the mXT source in ABS phantom was 265.24 cGy/min and corresponding dose rate in water phantom was 322.84 cGy/min. The dose-rate constant was calculated to 1354.71 cGy•/h•μA by dividing published air kerma strength of the mXT (Nanoscale Research Letters 2012, 7:258). Radial dose function in the water phantom was decreased from 1.00 at 1 cm from the source to 0.26 at 4 cm. The anisotropic function was slightly increased with radial distance and angle from transverse axis, but had large variations according to the radial distances and the angles. Conclusion: Derived dosimetric parameters of the new mXT source could be applied to the dose calculation in brachytherapy treatment planning system.

ABSTRACT. Accurate calculation of the radiation transport in an irradiated object is a necessity for determination of dosimetric quantities such as dose, fluence, stopping power ratios, etc. For dose calculations in treatment planning systems (TPS) beam modelling is critical for dose accuracy, and is an obligatory step in the commissioning procedures for any dose algorithm, hence not unique for Monte Carlo (MC) simulations. However, since MC can explicitly provide all of the above exemplified dosimetric quantities, it’s beam modelling process might require extra awareness. MC is also a convenient tool to investigate various limitations of approximations in clinical TPS, which can make beam modelling critical.

Beam modelling for MC require explicit generation of full particle phase spaces, while beam modelling for semi-analytical dose calculations in a TPS rely on macroscopic quantities such as fluence and energy spectrum,  where (spatial) correlations between the two is often neglected. Certain MC packages provide detailed interfaces for describing the geometry of the treatment head/beam nozzle and are thus able to generate a beam phase space at a beam reference plane downstream of all beam shaping/collimating devices. The beam properties upstream of the beam shaping devices are important for the outcome, but not accessible for MC simulation per se, thus requiring iterative approaches for validation of assumed properties. The common modelling approaches for photon and particle beams will be reviewed and a comparison to clinical TPS modelling approaches made to highlight critical areas. Collimating devices with partial leakage, scattering and generation of secondary particles can be challenging to commonly used approximations in clinical TPS and thus be of interest for more detailed MC analysis. The penumbra regions also yield gradients in dosimetric quantities of importance for detector response calculations.

ABSTRACT. The evaluation of the absorbed dose deposited on a patient following a pre-ordained clinical plan has undergone a renaissance during the last decade. The combined role played by increasingly powerful computational architectures together with new refined algorithms allows the clinical user to reach an accuracy only dreamed of during the last decades.  Among the different radiation modalities, it is brachytherapy the one that has evolved more drastically in the last few years.

Monte Carlo (MC) calculations, an advance dose calculation algorithm itself, have been used since 1994 for the evaluation of absorbed dose in clinical brachytherapy by means of the TG-43 formalism. Such formalism makes use of state-of-the-art MC simulations to obtain the absorbed dose deposited by the brachytherapy sources. However, in order to be used in clinical practice, such dose kernels are pre-obtained assuming that the patient is made of water and immersed in an infinite volume of water. Therefore, effects like interseed attenuation, tissue heterogeneities, patient geometry or applicator materials are not explicitly considered when creating a clinical plan. This implies that the role played by MC simulations until few years ago was restricted to the development and validation of new source models and applicator designs.

Fortunately, in the last years, new Model Based Dose Calculation Algorithms (MBDCAs) have been developed. In 2012, AAPM, ESTRO, ABS, and ABG released a report, TG-186, to provide guidance for early adopters of MBDCAs for brachytherapy, and to ensure clinical practice uniformity. TG186 goes one-step further, defining MC as the gold standard to which any new dose calculation algorithms should be compared and commissioned. Therefore, any MBDCA algorithm has to be verified against state-of-the-art MC calculations to ensure: i) that it reproduces correctly the “true” absorbed dose as described by MC, ii) to evaluate possible discrepancies due to the unavoidable numerical approximations required, and iii) to guide the clinical user when moving away from the current TG-43 clinical planning.

In this lecture, we will describe the new MBDCAs commercially available in clinical practice, describing their main properties and focusing on the role that MC simulations play in their commissioning.

ABSTRACT. The use of Monte Carlo -based dose algorithms has become mainstream in the clinic today.  Several major commercially available treatment planning systems incorporate Monte Carlo algorithms for photon and electron transport.  New delivery technologies, such as the MRI-linac systems include the use of MC-based dose algorithms for treatment planning.  Clinical examples motivating the need for MC-based dose algorithms will be presented for photon and electron beams. This session will include discussion of issues, such as dose reporting methods (dose-to-water or medium), statistical uncertainties and dose denoising.  Methods for modeling of the treatment head geometry will be reviewed based on the AAPM Task Group Report No. 157, aimed to provide guidance to clinical physicists on the tasks of acceptance testing and commissioning of MC-based treatment planning systems for photon and electron beam dose calculation.  The specific recommendations on methods and practical procedures to clinically accept and commission source models for treatment planning systems employing MC-based dose algorithms will be discussed, along with clinical examples.

ABSTRACT. Breast radiotherapy is commonly performed with a linear accelerator and with photon spectra of 610 MV. After previous studies of UC Davis, we developed a Geant4 Monte Carlo simulation code for investigating the use of kilovoltage X-ray sources rotating around the breast, for external beam radiotherapy of breast cancer (kV-EBRT). Here, the patient is in prone position on a translating and rotating supporting bed, with her breast pending from a hole in the bed. The photon beam, produced by an under-table X-ray tube, is collimated horizontally and vertically in order to irradiate the target volume while rotating in a circular orbit around a vertical axis passing through the target. The code is based on Geant4 toolkit v. 10.00. The pendant breast is modelled as a cylindrical phantom made either of polyethylene (for validation purpose) or of homogeneous breast tissue. We scored the dose distribution within the simulated breast for X-ray tube spectra at 320 kV, 300 kV and 150 kV (the last two adopted in the experimental validation) with a spatial resolution of 1 mm3. We evaluated the skin-to-lesion dose ratio, as a figure of merit for the skin sparing effect produced in rotational kV-EBRT. We studied the influence of the distance between the cylindrical lesion (target) and the skin (phantom surface), of the irradiation protocol (full scan or partial scan) and of the beam intensity modulation. The skin-to-tumor dose ratio ranged between 13% and 9% for the adopted spectra and a 14-cm diameter breast, and for a tumor located at the central axis of the phantom: this ratio increased for a lesion close to the surface. For a tumor at 5.25 cm from the breast axis, the dose ratio decreased from 34% to a value of 20% of the tumor dose, by adopting a beam intensity modulation.

ABSTRACT. Purpose: Pulsed low dose rate (PLDR) radiotherapy has potential to reduce normal tissue toxicities while still providing significant tumor control for re-irradiation and adjuvant treatments. This work investigates treatment planning and dosimetry verification criteria for PLDR radiotherapy using advanced delivery techniques. Methods: Different treatment sites were investigated including breast, pancreas, prostate, head and neck, and lung. PLDR plans were generated using the Varian Eclipse system with 6, 10 and 15MV photon beams. Each plan consisted of either 10 gantry angles for IMRT delivery or two arcs for VMAT delivery to achieve a daily dose of 2Gy. The dosimetry requirement was to deliver 20cGy per sub-fraction (pulse) with a 3min interval to achieve an effective dose rate of 6.7cGy/min. Monte Carlo simulations were performed to calculate the actual dose delivered to the planning target volume (PTV) considering beam attenuation/scattering and intensity modulation. The maximum, minimum and mean doses to the PTV were analyzed together with the dose volume histograms and isodose distributions. Results: Both IMRT and VMAT could achieve superior dose distributions for complex treatment geometry to spare critical structures effectively. For IMRT patients, the mean PTV dose for each sub-fraction ranged 15-35cGy per gantry angle and the minimum doses ranged from 8.2 cGy to 17.5 cGy. The VMAT dose varied between 8.56 and 31.2cGy/arc for breast, 12.9 and 27.5cGy/arc for pancreas, 12.6 and 28.3cGy/arc for prostate, 12.1 and 30.4cGy/arc for H&N, and 16.2 and 27.6cGy/arc for lung, respectively. Good agreement was achieved between the planned doses and Monte Carlo simulations. Matrix measurements showed 95-98% passing rates (3%/3mm). Conclusions: Advanced radiotherapy delivery techniques can provide superior target coverage and normal tissue sparing for PLDR radiotherapy either for re-irradiation or adjuvant treatment. IMRT is more suited for large and irregular targets while VMAT is more effective for centrally located tumors.

ABSTRACT. Accuray’s Ray-Tracing algorithm is able to quickly calculate doses for CyberKnife plans through significant approximations of the beam’s physics. Although demonstrated to be inaccurate in heterogeneous regions, all CyberKnife plans were calculated via Ray-Tracing prior to the introduction of Accuray’s Monte Carlo algorithm. However, for non-small cell lung cancer (NSCLC) patients treated with stereotactic body radiation therapy, local control and survival rates have been shown to be strongly correlated to the dose delivered to the tumor. To assess the extent of the error between planned and delivered doses, 115 NSCLC patient plans treated with CyberKnife were recalculated on an independent Monte Carlo model. The model was built on EGSnrc and was validated such that output factors agreed with commissioning measurements within 1%. In this retrospective study, dosimetric differences between the doses calculated by the Ray-Tracing algorithm, Accuray’s Monte Carlo algorithm and the EGSnrc Monte Carlo algorithm were analysed. No significant differences were found between the two Monte Carlo models. The Ray Tracing algorithm was found to overestimate the mean dose in the planning target volume (PTV) by more than 25%. Near-maximum doses, both in the PTV and organs at risk, were however not found to be statistically significantly different when recalculated. Some Ray-Tracing plans were noted to have larger in-PTV doses when recalculated. This study thus confirms the risk of significantly undercovering the PTV when planning lung plans with Ray-Tracing doses, while also highlighting the large variability involved when estimating delivered Ray-Tracing doses.

ABSTRACT. GammaPod is a breast-specific Co60 multi-source stereotactic radiotherapy device which uses 25 sources distributed over 25 degrees of latitudinal angles to create 25 non-overlapping conical arcs and achieve highly focused dose distribution. The 25 beams span from 18 to 42 degree off the horizontal plane, with 1 degree latitude intervals. The rotational focusing of the current GammaPod design yielded rapid dose falloff from the edge of target, with excellent sparing of normal breast and surrounding organs. We describe Treatment Planning System based upon Monte Carlo kernels to compute and optimize planning dose according to the objectives and constrains. We present our development pipeline starting with Co60 hardware sources and breast cups CAD description through Monte Carlo calculations based upon GEANT4 and EGSnrc. We describe in details our cloud-based home grown Monte Carlo cloud system solution, which, coupled with power of Google Compute Engine and Kubernetes allows us to compute Monte Carlo dosimetric kernels in fast and predictable way. We demonstrate good agreement with single shot films and ionization chambers measurements as well as comparison for optimized clinical multishot plans. Applicability of Monte Carlo kernels based treatment planning to head and neck tumors is discussed as well.

ABSTRACT. Introduction: Independent verification of treatment planning dose calculations is an essential part of radiotherapy quality assurance program. As of August 2015 our department implemented Monte Carlo (MC) based verification of VMAT and IMRT as well as complex conformal plans.

Method: Vancouver Island Monte Carlo system (VIMC) [1, 2] is used for routine verification of clinical VMAT/IMRT and conformal plans that were generated with Eclipse AAA. “Quick MC” option within the VIMC system includes “Fast Jaw Tracking” (FAJT) [3] combined with VCU particle DMLC models for particle transport through the linac head. VMC++ code is used for dose calculation in the phantom produced from patient CT dataset. MC simulation times are of the order of 5-10 minutes for majority of cases. MC dose distribution is imported to the planning system (Eclipse), compared to AAA dose and kept as a part of the patient record.

Results: Over 3600 plans have been verified using our MC system with current use rate being about 35-40 plans per week. Overall agreement of mean PTV dose between AAA and MC across all treatment sites was -0.5% (SD=1.4%), ranging from -29.2% to +5.6%. Site-specific mean PTV dose agreement for abdomen, breast & chest-wall, brain, esophagus, head & neck, lung and pelvis was within +-1% (SD=1%). As expected, the worst agreement between AAA and MC calculations was observed for lung-SABR plans with mean PTV dose difference of -3.2%, SD=4.0%, ranging from -29.2% to +2.9%.

Conclusion: MC verification of the treatment plans has shown to be efficient and reliable process. It highlighted deficiencies in commercial planning algorithms in the regions of heterogeneity and allowed to make appropriate corrections.

References: 1. S.Zavgorodni, K.Bush, C.Locke and W.Beckham, Radiother. Oncol. 84, Supplement 1, S49,(2007). 2. K.Bush, R.Townson and S.Zavgorodni, Phys.Med.Biol., N359-370 (2008). 3. R.Townson, H.Egglestone and S.Zavgorodni J.Appl.Clin. Med. Phys. (2018).

ABSTRACT. We present a non-hybrid Monte Carlo (MC) based inverse treatment planning for trajectory-based volumetric modulated arc therapy (TVMAT). With the use of continuous and simultaneous gantry and couch rotation, a higher dosimetric plan quality can be achieved. However, commercial treatment planning systems do not provide such a capability. It has been shown that a full MC based optimization greatly reduces the optimization convergence errors. Previously published approaches to MC based optimization have not been clinically implemented, and none have been proposed for VMAT or TVMAT so far. In this work, we have developed a method that reflects the dynamic multi-leaf collimator (MLC) and gantry-couch trajectory of the actual beam delivery at all stages of the optimization. Dose optimization is performed in a single Monte Carlo simulation, thereby greatly reducing calculation time. We select the initial trajectory (i.e. the range of the gantry, collimator and couch angles) and the initial set of leaf positions, corresponding to a dynamic beam conformal to the target. The MC simulation starts from a phase space scored at the top of the MLC module and uses isource=20 of DOSXYZnrc. We modified DOSXYZnrc in order to generate a 4D dose file that scores individual, time-stamped, energy deposition events in the voxels of the planning target volume (PTV) and organs at risk (OAR). Consequently, a relation is established between the space and time (i.e. MU index) coordinates of source particles in the phase space and their contribution to energy deposition. A direct-aperture optimization, with a dose-volume histogram based quadratic objective function, is performed using an in-house code, taking rigorously into account the continuous movement of the MLC, gantry and couch between adjacent control points. Clinically acceptable PTV coverage and OAR sparing have been achieved with this trajectory-based MC optimization.

ABSTRACT. This research proposes a novel concept of the stomach cancer diagnosis focusing on the reactions of the thermal neutron capture γ-ray emission using two Monte Carlo simulation programs, MCNPX and GATE. Chlorine is one of trace elements being possible to diagnose stomach cancer due to the concentration difference between normal and cancerous tissues. Thermalized neutrons having about 0.025 eV or under interact with the chlorine, then emit significant prompt characteristic γ-rays having specific energy levels since the thermal neutron capture reaction cross section by chlorine is highly large compared to the other trace elements consisting of stomach tissue. Optimized geometry including a thermal neutron source, samples of normal and cancerous stomach tissues, and a circular array to form the detection system was designed and the results on energy spectra of γ-rays were obtained using numerical simulations of the interactions between thermal neutrons and chlorines. Three specific peaks at the energy levels of 6.12, 7.80 and 8.58 MeV were recognized then the net count areas for both normal and cancerous tissues were compared. Since the concentration of the chlorine in the normal tissue is as twice as that in the cancerous tissue of the stomach, the significant difference of net count areas provides a concise but accurate approach for the diagnosis of the stomach cancer. Even though there are many methods to diagnose cancer nowadays, blind spots still exist. This study could contribute to making up for the weakness of the diagnosis system for the detection of cancer cells.

ABSTRACT. Objective: To validate 4D Monte Carlo (MC) simulations of dose delivery to a deformable phantom using realistic respiratory patterns.

Methods: A tissue-equivalent deformable lung phantom was used to simulate realistic breathing motion. A piston, attached to a programmable motor, provides variable motion along the sup-inf direction. Irradiations were performed on an Elekta Infinity linac for 3 different breathing traces. Dose within the tumor, was measured using film and the RADPOS 4D dosimetry system. Outside the tumor, two RADPOS detectors measured dose on the top and bottom surfaces of the plug. RADPOS position tracker recorded the phantom motion with a temporal resolution of a 100 ms. Square static and VMAT plans were created on the end-of-inhale (0 cm) CT scans of the phantom in Monaco TPS V.5.11.01 to deliver 100 cGy to the center of the tumor. A validated BEAMnrc model of our 6 MV linac along with the 4DdefDOSXYZnrc user code were used for 4DMC simulations of the dose delivery to the breathing phantom. Delivery log files were used to generate simulation input files. Deformation vectors were obtained by registering CT scans of the end-of-exhale (3 cm) to end-of-inhale states using Velocity AI 3.2.0. Deformation vectors and phantom motion trace measured with RADPOS, were used to model the phantom motion.

Results: Dose values from simulations and measurements at the center of the tumor and bottom surface of the plug were found to be within 3%. Agreements on the top surface of the plug (high dose gradient region) were found to be better slightly over 5.0%. Simulated dose profiles agreed within 2%/2 mm with film for over 94% of points.

Conclusions: Our 4DMC code accurately calculates dose delivered to a realistic breathing anatomy. This tool can be used for adaptive purposes to calculate the cumulative dose delivered to patients during treatments.

ABSTRACT. To ensure accuracy during radiotherapy treatment deliveries, the radiation beam produced by a linac must match the beam model in the treatment planning system (TPS), usually assuming an ideal beam set-up. All medical linacs therefore undergo routine quality assurance (QA) testing in order to ensure they are able to achieve the desired accuracy for delivering clinical treatments as planned within the TPS. Current QA guidelines provide medical physicists with specific beam quality tests together with allowed tolerances to be used within a QA program.

However, these tests can be affected by drifts in electron beam energy as well as angle of incidence on target and lateral offset position. The tolerances for all QA tests should be set such that they are sensitive to beam drift instability and effects across the entire range of beam arrangements.

Full Monte Carlo simulations of the medical linac treatment head can help determine the sensitivity of different QA tests to observed drifts in linac operating parameters. We have developed an entire high performance computing experimental test-bed that simulates the linac operating at 6MV, 10MV and 18MV to within a 2% - 2mm accuracy. We are using this model to test how drifts in linac operating parameters can best be measured within a QA program.

ABSTRACT. Microbeam and minibeam radiotherapy have become areas of great interest in radiotherapy as they offer the possibility of a very substantial increase in the therapeutic ratio. At the same time geometric accuracy becomes of less importance and issues that affect current IGRT techniques such as setup and delineation uncertainties are greatly reduced. We study the possibility of producing a collimator that could be attached to a high dose rate linear accelerator to produce linac based mini-beam radiotherapy. This study compared different collimator designs to determine whether a realisable design could provide the field sizes required for a minibeam treatment. Collimator designs were simulated in a Geant 4 based simulation to determine the peak-to-valley ratio (PVR) that resulted. The collimators were designed to match the divergence of the beam. It was found that with a 1 MeV electron beam PVRs of better than 10 were achievable in the centre of the field. This represents a value that is probably useful clinically however the design would be challenging to manufacture, particularly with a very high-z material. Reducing the density by using brass for the collimator reduced the PVR to significantly less than 10 for a 5 cm thick collimator. We found that PVRs that are likely to be clinically useful could be produced with a collimated photon beam however the design needs to be further refined to be manufactured.