Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures

Final Master Thesis Double Master’s Degree in Industrial and Nuclear Engineering Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures THESIS Author: Álvaro Merino Canete Director: María Amor Duch Data: March 2018 Escola Tècnica Superior d’Enginyeria Industrial de Barcelona You must be the change you wish to see in the World Mahatma Gandhi 2 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 3 ACKNOWLEDGMENTS ACKNOWLEDGMENTS Este proyecto significa para mí la culminación de mis años universitarios, por este motivo, antes de exponerlo me gustaría dedicar unas palabras a todas aquellas personas que me han ayudado, con las que he compartido bueno y malos momentos y de las que he aprendido durante todos mis años universitarios. En primer lugar, quiero dar las gracias a mis padres por dármelo absolutamente todo y gracias a los cuales he cumplido un sueño, nunca les estaré lo suficientemente agradecido; a mi hermano Eduardo por su apoyo incondicional, por levantarme los ánimos cuando flaqueaban las fuerzas y por mostrarme su filosofía; y a mis abuelos por su confianza ciega, incluso cuando nos las tenía todas conmigo. Tampoco me olvido de quien, a pesar de las dificultades, siempre ha estado a mi lado y me ha apoyado de forma incansable incluso en los peores días. Capaz de comprender cuando necesitaba espacio y cuando apoyo, así como proporcionármelos, siempre le estaré agradecido. También quiero agradecer a todos mis amigos de Madrid: Guille, Mayte, Nuria, Chechu, Sandra, Alicia, Pablo, Steven, María y a todos los demás que, aunque por la distancia no tengamos tanto contacto, siempre me han animado. De mis primeros años en la ETSII tengo mucho que agradecer a los “Industrialitos” por hacerme el primer año mucho más llevadero de lo que a priori yo esperaba y de los cursos sucesivos es necesario mencionar a aquellos con los que he compartido silla día a día y un trabajo tras otro como Refoyo y Tere. De mi etapa en el Máster de Ingeniería Industrial en la ETSEIB hay mucha gente con la que he trabajado y de la que he aprendido multitud de cosas y son tantas personas que aquí solo pueden estar recogidos algunos. Gracias al grupo de “Padelmania” por acogerme como uno más desde el primer día; gracias al grupo de “Blancanieves y los siete enanitos” especialmente a Esther y Miquel por trabajar como un verdadero equipo; gracias a los chicos de Girona por hacerme sentir como en casa y gracias a Cristian, Ana y David por hacerme más llevaderos estos años. A mis compañeros del Máster de Ingeniería Nuclear les deseo lo mejor y les mando mis más sinceros agradecimientos a Alfonso, Eduard, Bartek, Miriam y Nathan, así como a todos los integrantes de “Nuclear Party”, “Spanish Mafia” y a mis predecesores del doble máster que me marcaron el camino, mención especial para Soriano. Dentro del apartado nuclear, quiero hacer un aparte para mi querida Laura Gómez. Una persona con la que no sólo he compartido proyectos en la Universidad sino prácticamente toda la vida fuera durante el último año de máster, una persona que, en poco tiempo, se ha convertido en una de las personas más importantes de mi vida y con la que espero volverme a cruzar a nivel profesional en el futuro. Para terminar con los agradecimientos a mis compañeros de Universidad, hay dos personas que se merecen una mención individual por lo que han significado para mí dentro de las clases, pero sobre todo por lo que han sido, son y serán fuera de ellas. Por un lado, David Arias, quien tuvo que aguantarme día tras día en Madrid y con quien estudiar, incluso las peores asignaturas, era algo que esperaba con ganas. Por otro lado, Arnau Prats, con quien he compartido mis años en Barcelona desde que salía el sol hasta que se ponía. De la mano de ambos he terminado el Grado y el Doble Máster, tal 4 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures y como se esperaba de nosotros, ambos se merecen el mayor de mis agradecimientos y les deseo lo mejor porque sé que se lo merecen. A mis profesores, tanto a los que soy afín como a los que no, les quiero agradecer todo lo aprendido, en concreto a José María Cabanellas, mi profesor de dibujo y director del TFG; a Carlos Molpeceres, mi profesor de electromagnetismo; y a Eva Gallardo, mi profesora de recursos humanos, por hacerme sacar lo mejor de mí en cada momento y por enseñarme cosas que transcienden a la ingeniería. Por último, quiero agradecer a Lluis Batet por facilitarme las cosas en el Doble Máster; a Mercè Ginjaume y María Amor por darme la oportunidad de trabajar en este proyecto y confiar en mí; a Sara Principi y María Rodríguez por aguantarme durante las prácticas y ayudarme siempre que lo necesitaba, incluso cuando estaban ocupadas; al INTE en general por su calurosa acogida y al CSN por su financiación para desarrollar este proyecto. 5 ABSTRACT ABSTRACT This project was developed as part of the European MEDIRAD project in the initial phase of Work package 2, Subtask 2.2.2 Real-time patient dose monitoring in fluoroscopicallyguided interventional and PODIUM project. During this project, a program called MCGPU, what is a Monte Carlo simulation code that can simulate real interventional radiological procedures conditions, was used. The main objectives fulfilled in this project are the MC-GPU code improvement, the validation and verification of the MC-GPU results and the elaboration of a MC-GPU manual for beginner users. The MC-GPU materials generator was updated due to a 3 %variation of the photoelectric MFPs between the materials of the versions of PENELOPE 2006 and 2014. The update was completed successfully obtaining more accurate material data files than the originals. The computational validation phase consists of comparing the MC-GPU dose results with the obtained ones with a standard Monte Carlo simulation code, PENELOPE/penEasy. In most cases, the main MC-GPU results vary around 1 % with respect to those obtained with PENELOPE/penEasy but the simulation times are between 300 and 500 times shorter. At the end of the computational validation, a simple comparison was made with experimental measures. This first comparison between the experimental measurements made in the Hospital and the MC-GPU simulations showed good results, giving rise to the beginning of the validation phase with real measurements. In this phase, more complex distributions in terms of materials and rotations must be used. In the medium-long term, the validation of the new MC-GPU features as the presence of the operator, the influence of shielding materials and the us of conversion factors will be the lines of future work. UNESCO codes: 210115, 220212, 320111, 320112, 320401, 330723 Keywords: MEDIRAD; Monte Carlo Simulation; MC-GPU; PENELOPE; PENELOPE/penEasy; CUDA; voxelized geometry; interventional radiology; patient dose monitoring; dose distribution; X-rays; validation. 6 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 7 CONTENTS 1. CONTENTS ACKNOWLEDGMENTS __________________________________________________ 3 ABSTRACT ____________________________________________________________ 5 1. CONTENTS ________________________________________________________ 7 2. GLOSSARY _______________________________________________________ 11 2.1 SYMBOLS AND UNITS ________________________________________________ 13 2.2 ABBREVIATIONS AND ACRONYMS ______________________________________ 13 3. PREFACE ________________________________________________________ 15 3.1 ORIGIN OF PROJECT __________________________________________________ 17 3.2 MOTIVATION _______________________________________________________ 17 3.3 STATE-OF-THE-ART __________________________________________________ 17 4. INTRODUCTION ___________________________________________________ 19 4.1 PROJECT OBJECTIVES _________________________________________________ 19 4.2 SCOPE OF THE PROJECT _______________________________________________ 19 4.3 CONTEXTUALIZATION ________________________________________________ 20 4.4 TECHNOLOGY CHALLENGES ____________________________________________ 22 4.5 BUSINESS AND COMMERCIAL OPPORTUNITIES ____________________________ 23 5. METHODOLOGY __________________________________________________ 25 5.1 MC-GPU ___________________________________________________________ 25 5.2 PENELOPE/penEasy __________________________________________________ 27 5.3 THE CLUSTER _______________________________________________________ 27 6. RESULTS AND DISCUSSION __________________________________________ 29 6.1 THE USER MANUAL __________________________________________________ 29 6.2 MC-GPU UPDATE ____________________________________________________ 30 6.2.1 Materials generator code ___________________________________________________ 30 6.2.2 MC-GPU library modification ________________________________________________ 32 6.3 MATERIALS VALIDATION ______________________________________________ 33 6.3.1 Simulation 1: Air voxel ______________________________________________________ 36 6.3.2 Simulation 2: PMMA block __________________________________________________ 38 6.3.3 Conclusions ______________________________________________________________ 40 6.4 LIBRARIES __________________________________________________________ 41 6.4.1 Materials library __________________________________________________________ 41 6.4.2 Spectra library ____________________________________________________________ 41 6.5 MC-GPU VALIDATION ________________________________________________ 42 6.5.1 Simulation 3: One air voxel in PMMA block _____________________________________ 42 6.5.2 Simulation 4: Three materials phantom ________________________________________ 43 6.5.3 Simulation 5: Zubal chest ___________________________________________________ 45 6.5.4 Simulation 6: Duke tests ____________________________________________________ 46 6.5.5 Simulation 7: Duke’s chest tests ______________________________________________ 48 14 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 15 PREFACE 3. PREFACE The project explained in this thesis is encompassed in a European project named MEDIRAD123. The Horizon 2020 MEDIRAD project on implications of medical low dose radiation exposure aims to enhance the scientific bases and clinical practice of radiation protection in the medical field and thereby addresses the need to understand and evaluate the health effects of low dose ionising radiation exposure from diagnostic and therapeutic imaging and from off-target effects in radiotherapy. MEDIRAD pursues three major operational objectives: - First, it will improve organ dose estimation and registration to inform clinical practice, optimise doses, set recommendations and provide adequate dosimetry for clinical-epidemiological studies of effects of medical radiation. - Second, it aims to evaluate and understand the effects of medical exposures, focusing on the two major endpoints of public health relevance: cardiovascular effects of low to moderate doses of radiation from radiotherapy in breast cancer treatment incl. understanding of mechanisms; and long-term effects on cancer risk of low doses from CT in children. - Third, it will develop science-based consensus policy recommendations for the effective protection of patients, workers and the general public. INTE-UPC work is included in one of the six work packages, the WP2. The main objective of this work package is to develop novel methodologies to reduce patient and staff radiation dose and potential radiation-related risks of cancer and non-cancer outcomes from chest imaging while maintaining or improving diagnostic information from existing and emerging techniques. Work will focus on state-of-the-art CT, fluoroscopically-guided interventional procedures and hybrid systems. Detailed dosimetry data will be produced, which will be valuable for optimising RP of patients from high-dose diagnostic and interventional procedures, as well as for input to epidemiological RP research studies and development of models of radiation induced risk. To be more precise, the subtask in which INTE-UPC is working is the Subtask 2.2.2 Real-time patient dose monitoring in fluoroscopically-guided interventional procedures. Following a brief explanation of this subtasks: A state-of-the-art method has been described for real time patient dose monitoring for interventional cardiology procedures based on the freely available accelerated Monte Carlo (MC) code, MC-GPU. The system uses two independent computer codes, for the simulation of the X-ray transport including detailed voxelized phantoms and for the virtual X-ray control console where the operator manually introduces the relevant acquisition parameters. The programme has not yet been benchmarked against standard simulation codes such as EGSnrc, MCNP or PENELOPE, nor tested in the clinical environment. The objective of this Subtask will be to validate and improve MC-GPU to determine skin dose distribution for specific realistic clinical set-up. First the X-ray transport will be validated by simulating a realistic clinical set-up with PENELOPE and MCNP MC codes. Corresponding measurements will be carried out using a Rando anthropomorphic phantom and passive dosimeters in an interventional cardiology suite in Belgium. Validation will be performed first for single projections and then for a complete treatment. In a second step the virtual X-ray source console module will be reviewed and improved, so that acquisition conditions can be read automatically from the console. The prototype performance will further be assessed in patients undergoing interventional procedures. 1 Official webpage: http://www.medirad-project.eu/#about-medirad 2 Brief explanation on: http://www.eibir.org/projects/h2020-projects/medirad/ 3 Project members on: https://cordis.europa.eu/project/rcn/211042_en.html 16 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Clinical tests will be performed at least in two hospitals (Barcelona and Belgium). Systems able to accurately monitor the dose to the patient in real time are the state-ofthe-art solutions to the issue of minimizing the likelihood of radiogenic skin injuries and estimating organ doses from fluoroscopically-guided procedures. The real time dose monitoring system developed in this Subtask will fulfil the requests of recent EU Council Directive 2013/59/EURATOM Basic Safety Standards4, Articles 60 and Article 63. The extract of this Council Directive that concerns in this project and thesis is the following: THE COUNCIL OF THE EUROPEAN UNION, Having regard to the Treaty establishing the European Atomic Energy Community, and in particular Articles 31 and 32 thereof, Having regard to the proposal from the European Commission, drawn up after having obtained the opinion of a group of persons appointed by the Scientific and Technical Committee from among scientific experts in the Member States, and after having consulted the European Economic and Social Committee, Having regard to the opinion of the European Parliament, Having regard to the opinion of the European Economic and Social Committee, Whereas: (28) In the medical area, important technological and scientific developments have led to a notable increase in the exposure of patients. In this respect, this Directive should emphasise the need for justification of medical exposure, including the exposure of asymptomatic individuals and should strengthen the requirements concerning information to be provided to patients, the recording and reporting of doses from medical procedures, the use of diagnostic reference levels and the availability of dose-indicating devices. Also, the mentioned article number 60 refers to Equipment and the number 63 refers to accidental and unintended exposures. To summarize, the project explained in this thesis is part of the Horizon 2020 MERIRAD project in the framework of WP 2, Subtask 2.2.2 Real-time patient dose monitoring in fluoroscopically-guided interventional. Additionally, another European project named PODIUM5 has begun concurrently to MEDIRAD project. The acronym PODIUM stands for Personal Online Dosimetry Using Computational Methods, which main objective is to improve personal dosimetry by using an innovative tool based on an on-line application. The validation and verification of results and user manual developed during this project will be also used in this new project. 4 https://ec.europa.eu/energy/sites/ener/files/documents/CELEX-32013L0059-EN-TXT.pdf 5 Some additional information on: http://www.irrs.eu/documents/IRRS2017Programme.pdf 17 PREFACE 3.1 ORIGIN OF PROJECT Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures project takes place at the beginning of the work requested to INTE-UPC. It is its first task. Before using the MC-GPU code for real-time monitoring, is necessary validate the code in static situations. The validation and verification of the results is done in ascending order of difficulty, starting with the most basic modules of the programme and ending with the most realistic simulations. 3.2 MOTIVATION The main purpose of this project is to explain in detail the validation and verification process of the MC-GPU code. The motivation and choice of this topic responds to two different circumstances: one institutional and other personal. First, the ultimate goal of the subtask 2.2.2 of MEDIRAD and PODIUM projects is the obtaining of a software capable of monitor the dose received by patients and medical personnel in fluoroscopically-guided interventional procedures. The motivation of this WP is fulfilling the European regulation of dose monitoring, in order to protect workers and general public. Second, the personal motivation to get involved in this project is the remarkable social character of it. The last beneficiaries of the project will be the patient who must pass through this kind of procedures. The development of a final commercial product is also one of the main attractive of the project. 3.3 STATE-OF-THE-ART As previously mentioned, this project is the first work developed in the INTE-UPC as part of MEDIRAD project and it starts few months after the beginning of MEDIRAD project. At this point, the programme is completely written and some test had been already made. The documentation about the main features of the code and how to compile the codes were provided by the coder. A folder full of materials files usually used in this kind of simulations and some spectra were also provided. Last modifications of the codes are dated in 2014 as the provided documentation. Few papers in which the code and test made are published during the last years before the project. About the MEDIRAD project, the budget, objectives, scope and WPs are clearly defined. 18 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 19 INTRODUCTION 4. INTRODUCTION Previous section of this thesis provides a global picture in which this project and this thesis are framed. In this part, the main challenges and objectives of the project are highlighted as well as an overall view of the project. The project has been developed at the Institut de Tècniques Energètiques of the Universitat Politècnica de Catalunya with partial funding support of the Consejo de Seguridad Nuclear, the Spanish Nuclear Regulatory Body. The project includes Monte Carlo simulations of different scenarios carried out by two different programmes, MC-GPU and PENELOPE/penEasy, the comparative between them and with experimental data measured in Vall d’Hebron Hospital of Barcelona. This project may be considered as a concept validation of the MC-GPU programme. For this reason, this project can be included between TRL 3, experimental proof of concept, and TRL 4, technological validity in a laboratory. 4.1 PROJECT OBJECTIVES The aforementioned code was developed few years ago, and some test were developed with it. However, a most exhaustive use of it must be done, mainly validation the results of the code with already tests codes as PENELOPE/penEasy and with experimental measurements. Within the before explained framework, this thesis has the following main aims: 1. Analyse, understand, improve and update to the extent possible the MC-GPU code and all the complementary codes and files, in order to obtain more realistic and accurate simulations. 2. Validation and verification of the MC-GPU and MC-GPU Beta, an extended version not yet publicly released, results. A comparative of simulations run by them and with PNELOPE/penEasy must be included. Moreover, a comparative of the results obtained by simulations with experimental data measured in a laboratory. This is the main objective of the project. 3. Elaboration of a MC-GPU user manual during the first months of this project to allow other users to use the programme and future personnel who will be involved in the MEDIRAD and PODIUM projects. This handbook must include information complementary to the code, explanation of the code and how to use it. 4.2 SCOPE OF THE PROJECT The European project in which this project is included will last 4 years, but the duration of this project is shorter. This project and the members in charge of it are the responsible of the validation and verification of the results, that is, the pre-project of what will be a commercial product. The scope of this project and thesis finishes in the experimental data validation phase. At this stage measured results obtained in an operating room will be compared with computational results obtained in the MC-GPU simulations. 20 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 4.3 CONTEXTUALIZATION As it is explained in previous sections, the project is based on the validation of a code which simulates real fluoroscopically-guided interventional procedures. An interventional radiological (IR) procedure is any procedure using radiological imaging equipment in order to guide a therapeutic/invasive procedure on a patient. Examples of such procedures include angiography, angioplasty, embolization, biopsy and drainage, dilations and stent placements. Fluoroscopy is predominantly used, which is a way of working in real time. Currently almost exclusively fluoroscopy is used. Interventional procedures are complex and generally involve the use of long fluoroscopy times. Consequently, there is a potential for high radiation doses to patients and staff as compared to other X-ray examinations. Because fluoroscopy involves the use of X-rays, an ionizing radiation, fluoroscopic procedures pose a potential for increasing the patient's risk of radiation-induced cancer. Radiation doses to the patient depend on the size of the patient as well as duration of the procedure. Exposure times vary depending on the procedure being performed. Because of the long duration of some procedures, additionally to the cancer risk and other stochastic radiation effects, deterministic radiation effects have also been observed ranging from mild erythema, equivalent of a sun burn, to more serious burns. While deterministic radiation effects are a possibility, radiation burns are not typical of standard fluoroscopic procedures. Most procedures sufficiently long in duration to produce radiation burns are part of necessary life-saving operations. Moreover, some procedures are very complex and they can lead to relatively high doses to the medical staff, henceforth the operators, who stand close to the primary radiation field if they do not protect themselves with the appropriate shield. Moreover, as the procedures performed in IC/IR require the interventional cardiologists and radiologists stand close to the patient, the patient represents the main source of scattered radiation. From a radiation protection point of view, under-couch (x-ray tube) operation is preferable as it reduces the amount of scattered radiation on operators. Following figure shows a real interventional radiological procedure using a radiological imaging equipment. Figure 1: Photo of a IC/IR procedure using Phillips Allura 21 INTRODUCTION Next figure shows a real radiological imaging equipment to provide a better idea of the equipment and environment which are simulated during this project. Figure 2: Real radiological imaging equipment (Philips Allura) The photo is taken from the side where the medical staff usually works. The picture shows the couch where the patient lies down, the screen in which the image appears and the machine C-arm, which has the X-ray source at the bottom and the image intensifier (detector) at the top. This image also includes the coordinate axis employed in the MC-GPU simulations. All information related to MC-GPU simulation may be consulted in the MC-GPU User Manual at the end of this thesis. 22 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 4.4 TECHNOLOGY CHALLENGES In this section, two different topics are explained. On the one hand, the technologies necessaries in which this project is based on and its state of art. On the other hand, possible activities with which the project may transfer technology bidirectionally. The code in which the software is based on is written in CUDA language CUDA 5.0. MC-GPU was developed using the CUDA programming model from NVIDIA to achieve maximum performance on NVIDIA GPUs. The code can also be compiled with a standard C compiler to be executed in a regular CPU. However, the code that creates the material files necessary in the simulations is written in Fortran, the same programming language in which PENELOPE and penEasy are written. At the beginning of the project the MC-GPU software and the material generator code were completed but their last modifications were made in 2014. The PENELOPE/penEasy [4] and PENELOPE [5] version employed in this project are the latest version published by their authors. Complementary to these four codes, some other programmes are necessary to analyse the results and do the comparatives: - MATLAB6 : some scripts to read the data obtained in MC-GPU and PENELOPE/penEasy simulations need to be run on it. - IMAGEJ7 : programme used to visualize the raw binary files, which is the format in what the code reports the results. It is mainly used to obtain images but also to obtain some numerical results. - GNUPLOT8 : this programme is used to visualize the tallies obtained in penEasy. - XCOMP5r9 : this software is employed to create the X-ray spectrum files. Most of these programmes are public and free for student and research personnel. All are tested programmes and belong to different users’ communities. Although this project is based on code simulations and all stuff related to the computational framework, experimental data is also needed, thus and for these activities dosimeters and real x-ray tubes are need. In this case, the relation between the discipline of medicine and computing is clear. Regarding to the possible activities related to this project, it is important to point out that this programme is focused on interventional procedures in which X-rays are employed. Nevertheless, this programme and technology have two possible paths: - The objective which this programme was created for, the real on-time monitoring of patient and medicine personnel during the interventional procedures taken place in an operating room using tracking cameras. In this case, the relation is between the software (code) and hardware (cameras) disciplines. - Parallel to the main path, the technology could be transferred to other activities (mainly medical procedures) in which radiation is used. The programme (and materials) would need some modifications but it could be employed. 6 Source: https://es.mathworks.com/products/matlab.html 7 Source (free): https://imagej.nih.gov/ij/ 8 Source (free): http://www.gnuplot.info/download.html 9 This is an old DOS programme and very difficult to find, it is in possession of the INTE personnel. 23 INTRODUCTION 4.5 BUSINESS AND COMMERCIAL OPPORTUNITIES The ultimate aim of the European projects is the developing of a commercial product which monitor the doses received by patients and medical personnel in interventional procedures. The idea for the final product is a software connected to tracking cameras to track the personal and to the x-ray tube to know the energy and time of the pulses generated. At the moment this thesis is written, there is not a commercial product which provides the required information (doses and maximum dose location) in a reasonable time using Monte Carlo methods, but there are ones that use other methods. These estimations are required by physicians for two different reasons: - Patient doses: in patients who undergo lot of procedures or lot of projections of different energies is important to know the received dose to avoid deterministic effects, mainly on the skin where the maximum dose is usually located. - Personnel doses: personnel who usually work with radiation equipment must wear passive dosimeters to monitor the received dose. The problem is that the passive dosimeters do not give a rapid measure and there exist some limitations on their use. Moreover, the directive10 about the dose received is changing and the collection of this information will not be optional but compulsory. The interest shown by the medical personnel who has been consulted by the project member joined to the lack of commercial products with the mentioned characteristics creates a commercial opportunity with possible profitable results in medium to long term. 10 EU Council Directive 2013/59/EURATOM Basic Safety Standards 30 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 6.2 MC-GPU UPDATE 6.2.1 Materials generator code Once the initial stage of learning was considered as finished, the code update process started. MC-GPU codes were modified to the newest version possible. As mentioned before, the latest version of MC-GPU code was released in 2012 and it has not suffered any modification since then. However the programme in which is based on, PENELOPE, has been updated as well as the materials that it uses. The latest version of PENELOPE and which is used for PENELOPE/penEasy simulation in this project is PENELOPE 201412. There have been few changes between the PENELOPE versions of 2006 and 2014, one of them what concerns to this project is that, as the NEA announces in its webpage, programmes for the calculation of mass energy-absorption coefficients for photons, and linear energy transfers for charged particles have been included in the distribution package. Therefore, the cross sections and the mean free paths of both versions are not the same. From material data files the MFP's have been obtained and the relative differences between both versions, 2006 and 2014, are plotted below. Analysing the plots, it may be concluded: - Compton MFP difference is high at low energies but rapidly decreases at high energies. - Rayleigh MFP difference is also high at low energies and although decrease at high energies, it does slower. - Photoelectric MFP difference is roughly constant in all the energy range between 2 and 3 %. - Total MFP difference may be considered medium values at low energies but at high energies decreases. Above 60 keV, the difference is almost negligible. Figure 4: MFP's PENELOPE 2006 and 2014 comparative for Dry Air 12 Available on https://www.oecd-nea.org/tools/abstract/detail/nea-1525 0% 1% 2% 3% 4% 5% 0,00E+00 2,00E+04 4,00E+04 6,00E+04 8,00E+04 1,00E+05 1,20E+05 MFP's relative difference Energy (eV) Rayleigh Compton Photoelectric TOTAL 31 RESULTS AND DISCUSSION Figure 5: MFP's PENELOPE 2006 and 2014 comparative for PMMA As it has been concluded before, the difference in the total MFP's is small at low energies and almost negligible at high energies. These variations between versions could have disregarded if the simulations used a broad energy range of spectrum. Figure 4 and Figure 5 just show the MFP’s relative difference for dry air and PMMA. However, the IR energy range of spectra is 60 to 120 keV, range in which the most probable interaction is the photoelectric effect and as it has shown before the variation of this interaction is roughly constant of 2 - 3 % depending on the material of interest. Figure 6: Mass attenuation coefficient for Dry Air13 13 Source: MIT open course ware: https://ocw.mit.edu/courses/nuclear-engineering/22-101applied-nuclear-physics-fall-2006/ 0% 1% 2% 3% 4% 5% 6% 0,00E+00 2,00E+04 4,00E+04 6,00E+04 8,00E+04 1,00E+05 1,20E+05 MFP's relative difference Energy (eV) Rayleigh Compton Photoelectric TOTAL 32 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures To sum up, as the MFP’s of materials used in the MC-GPU simulations are based on PENELOPE 2006 data and the MFP’s of materials used in the penEasy simulations are based on latest version of PENELOPE, and also the interaction most important in this kind of simulations has the larger variations thus the provided material files for MC-GPU simulations cannot be used to validate the code comparing with penEasy. For these reasons, the code which creates the materials in MC-GPU format (.mcgpu) from PENELOPE material data files (.mat) was updated. All the steps and of this process and all the modifications made in the code are in the APPENDIX B: UPDATED CODE AND NEW FILES. 6.2.2 MC-GPU library modification Due to the modifications explained in the previous section and the difference in the size and number of data included in the new PENELOPE material file, the MC-GPU library in where the variables are declared, named MC-GPU_v1.3.h, had little modifications. - The maximum number of shells for materials has been enlarged from 30 to 52. - The maximum number of energy bins has been enlarged from 25005 to 26005. 33 RESULTS AND DISCUSSION 6.3 MATERIALS VALIDATION After updating the materials generator code and modifying the MC-GPU variables library, it was necessary to check if the material data files generated by the new executable were the same as in PENELOPE 2014. Also, to check that all the tables and variables were correctly written. The new MC-GPU material file (.mcgpu) looks like this: Figure 7: MC-GPU material file: MFP's and Rayleigh max cumul prob table Figure 8: MC-GPU material file: Rayleigh interactions table Figure 9: MC-GPU material file: Shell information table 34 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures These figures correspond to the material Dry Air, in the energy range 5 – 120 keV. The columns are written correctly and in their places. Also, the Rayleigh and Compton tables are updated to the new materials. Following a comparative between the MFP’s generated with the new executable and in PENELOPE 2014 is shown: Figure 10: Rayleigh MFP comparative Figure 11: Compton MFP comparative 0 50 100 150 200 250 300 5,00E+03 2,50E+04 4,50E+04 6,50E+04 8,50E+04 1,05E+05 MFP (cm) Energy (eV) New Rayleigh Rayleigh PENELOPE 2014 0 1 2 3 4 5 6 7 8 9 10 5,00E+03 2,50E+04 4,50E+04 6,50E+04 8,50E+04 1,05E+05 MFP (cm) Energy (eV) New Compton Compton PENELOPE 2014 35 RESULTS AND DISCUSSION Figure 12: Photoelectric MFP comparative Figure 13: Total MFP comparative Figure 14: MC-GPU material file Rayleigh cumulative probability 0 200 400 600 800 1.000 5,00E+03 2,50E+04 4,50E+04 6,50E+04 8,50E+04 1,05E+05 MFP (cm) Energy (eV) New Photoelectric Photoelectric PENELOPE 2014 0 2 4 6 8 10 5,00E+03 2,50E+04 4,50E+04 6,50E+04 8,50E+04 1,05E+05 MFP (cm) Energy (eV) New Total Total PENELOPE 2014 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1 0,00E+00 2,00E+04 4,00E+04 6,00E+04 8,00E+04 1,00E+05 1,20E+05 1,40E+05 Rayleigh mac cumul prob () Energy (eV) 36 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures In this thesis, only the comparative between PENELOPE 2014 and the new material generator for Dry Air is shown, but the same process was made for different materials. As the graphs show, the results obtained with the new executable are the same obtained in PENELOPE 2014. In the light of these results, new materials may be considered as updated but the complete validation of the materials is made by running simulations with them. 6.3.1 Simulation 1: Air voxel The simulation run in the first place was the simplest possible, one voxel of air. The features what MC-GPU demands for run a simulation are shown in the following table: SIMULATION 1 Phantom Nº Voxels 1 x 1 x 1 Voxel size (cm3) 10 x 10 x 10 Dimensions (cm3) 10 x 10 x 10 Material Dry air ( ρ = 0.001205 g/cm3) Source Focal position (cm) 5 -15 5 Spectrum Monoenergetic spectra (60 – 120 keV) Histories 1.0E9 Detector size (cm2) 20 x 20 Source-to-detector distance (cm) 30 Table 3: Simulation 1 This simulation consists on irradiate a voxel of air perpendicularly to the XZ plane. The simulation is run for only one projection but repeated for the photon range of energies used in the interventional radiology (60 – 120 keV). The number of simulated histories was high in order to reduce as much as possible the uncertainties of the dose and energy deposited in the phantom but having in mind the simulation time. Figure 15: Simulation 1 scheme 37 RESULTS AND DISCUSSION From these simulation characteristics, a package of simulations was run with PENELOPE/penEasy simulating exactly the same elements. The results obtained from the simulations using the same inputs with new and old material data files, and the comparative with the PENELOPE/penEasy 2014 are shown below. Table 4: Simulation 1 new comparative Figure 16: Simulation 1 dose relative errors Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference New Difference Old Relative error New Relative error 60 11.61226 11.47738 11.436 0.176 0.042 1.543% 0.364% 70 11.54258 11.44943 11.418 0.124 0.031 1.088% 0.273% 80 12.22802 12.15716 12.115 0.113 0.042 0.935% 0.350% 90 13.34371 13.28865 13.256 0.087 0.032 0.659% 0.244% 100 14.75085 14.71027 14.650 0.100 0.060 0.686% 0.409% 110 16.36093 16.32939 16.252 0.109 0.077 0.668% 0.474% 120 18.08912 18.06191 17.984 0.105 0.078 0.582% 0.431% 200 33.94609 33.830 0.116 0.344% 300 54.81751 54.566 0.252 0.462% Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference New Difference Old Relative error New Relative error 60 13.99 13.83 13.780 0.210 0.050 1.523% 0.362% 70 13.91 13.80 13.759 0.151 0.041 1.097% 0.297% 80 14.73 14.65 14.598 0.132 0.052 0.903% 0.355% 90 16.08 16.01 15.974 0.106 0.036 0.664% 0.226% 100 17.77 17.73 17.654 0.116 0.076 0.659% 0.432% 110 19.71 19.68 19.584 0.126 0.096 0.643% 0.490% 120 21.80 21.76 21.671 0.129 0.089 0.594% 0.410% 200 40.91 40.765 0.145 0.356% 300 66.06 65.752 0.309 0.469% ENERGY DEPOSITION (eV/hist) DOSE (eV/g·hist) 0,0% 0,2% 0,4% 0,6% 0,8% 1,0% 1,2% 1,4% 1,6% 1,8% 60 70 80 90 100 110 120 Relative error Energy (keV) Old Relative error New Relative error 38 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Figure 17: Simulation 1 energy deposition relative errors In light of these results, a little more complex simulation was proposed. 6.3.2 Simulation 2: PMMA block For the second simulation, a PMMA block was chosen. The parallelepiped phantoms used in experimental measurements in dosimetry laboratories are usually made of this material. Moreover, the dimensions of the block are similar to a thick person’s chest. SIMULATION 2 Phantom Nº Voxels 6 x 3 x 6 Voxel size (cm3) 5 x 10 x 5 Dimensions (cm3) 30 x 30 x 30 Material PMMA ( ρ = 1.19 g/cm3) Source Focal position (cm) 15 -45 15 Spectrum Monoenergetic spectra (60 – 120 keV) Histories 1.0E9 Detector size (cm2) 54 x 54 Source-to-detector distance (cm) 81 Table 5:Simulation 2 Some features are different between first and second simulation but the most important are the change in the material and that in this second simulation the voxel dimensions are not equal in the three axes. 0,0% 0,2% 0,4% 0,6% 0,8% 1,0% 1,2% 1,4% 1,6% 60 70 80 90 100 110 120 Relative error Energy (keV) Old Relative error New Relative error 39 RESULTS AND DISCUSSION Using the same inputs but with the new and old material data files, the results obtained from the simulations and the comparative with the penEasy/PENELOPE 2014 are shown below. Table 6: Simulation 2 new comparative Figure 18: Simulation 2 dose relative errors Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference Old Relative error New Difference New Relative error 60 0.808 0.801 0.802 0.0053 0.666% 0.0009 0.116% 70 0.888 0.882 0.883 0.0049 0.555% 0.0011 0.124% 80 0.983 0.978 0.979 0.0045 0.456% 0.0011 0.114% 90 1.090 1.085 1.086 0.0041 0.378% 0.0011 0.100% 100 1.206 1.201 1.202 0.0038 0.318% 0.0010 0.086% 110 1.328 1.324 1.325 0.0036 0.273% 0.0010 0.074% 120 1.456 1.452 1.453 0.0033 0.231% 0.0009 0.065% 200 2.591 2.592 0.0008 0.030% 300 4.114 4.115 0.0006 0.015% Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference Old Relative error New Difference New Relative error 60 25947.550 25745.970 25776.000 171.550 0.666% 30.030 0.117% 70 28540.420 28347.670 28382.900 157.520 0.555% 35.230 0.124% 80 31596.950 31417.570 31453.500 143.450 0.456% 35.930 0.114% 90 35020.840 34854.070 34888.900 131.940 0.378% 34.830 0.100% 100 38732.970 38577.050 38610.300 122.670 0.318% 33.250 0.086% 110 42673.680 42526.050 42557.500 116.180 0.273% 31.450 0.074% 120 46793.210 46655.240 46685.500 107.710 0.231% 30.260 0.065% 200 83241.110 83266.300 25.190 0.030% 300 132191.460 132211.000 19.540 0.015% DOSE (eV/g·hist) ENERGY DEPOSITION (eV/hist) 0,0% 0,1% 0,2% 0,3% 0,4% 0,5% 0,6% 0,7% 60 70 80 90 100 110 120 Relative error Energy (keV) MC-GPU NEW MC-GPU 46 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 6.5.4 Simulation 6: Duke tests 21 These simulations were carried out in order to avoid the discrepancies due to a large number of materials. In this case, the entire Duke voxelized by 0.5 x 0.5 x 0.5 cm3 voxels was used. Although the phantom is a representation of a whole human body, the radiation field is focus on the chest. Same simulation was repeated with a 100 keV monoenergetic and with a realistic RQR 6 spectrum, both simulated a PA projection. The first simulation used a 100 keV monoenergetic in which the results are also compared with the obtained with the MC-GPU original. In the second one, the spectrum was the RQR 6 and only the comparison between the updated MC-GPU and PENELOPE/penEasy is shown 22 . SIMULATION 6 Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Focal position (cm) 30.5 -30.5 130 Spectra 100 keV monoenergetic and RQR 6 Histories 1006720000 Detector size (cm2) 143 x 143 Source-to-detector distance (cm) 71.5 Table 13: Simulation 6 Table 14: Simulation 6 with RQR6 results 21 Source: https://www.itis.ethz.ch/virtual-population/virtual-population/overview/ 22 Results of both simulations at: SIMULATION 6 Dose (eV/g·hist) MC-GPU updated PENELOPE/penEasy Updated Relative Error Air 0.13787 0.1379 0.029% Muscle 0.22679 0.2269 0.053% Soft tissue 0.13792 0.1371 0.572% Bone 0.63196 0.6299 0.332% Cartilage 0.06394 0.0634 0.923% Adipose 0.16574 0.1661 0.228% Blood 0.10704 0.1067 0.309% Skin 0.26648 0.2677 0.439% Lung 0.29497 0.2930 0.667% Maximum dose point 7.232274 7.3318 1.357% Sigma 0.001633 0.067 Sigma/mean 0.02% 0.91% Coordinates (58,5,271) (58,5,271) Real simulation time (s) 112.96 33641.11 47 RESULTS AND DISCUSSION Regarding to the simulation that uses a monoenergetic spectrum, it can be appreciated that blood and skin have relative errors above 1 % when the old materials are used but when the new materials are employed, all relative errors are below 0.17 %. Maximum dose point is also obtained accurately even with almost 60 % sigma in PENELOPE/penEasy results. When the energy deposition standard deviations obtained in PENELOPE/penEasy for each material are taken into account, the higher maximum relative error is 0.18 % and the higher minimum relative error is 0.14 % (air is neglected). These results may be considered as accurate enough. As the updated MC-GPU results are more precise, the simulation using RQR 6 spectrum is just carried out with this code. The comparative between the MC-GPU and PENELOPE/penEasy results are, one more time, below 1 %. In this case, the maximum dose sigma for PENELOPE/penEasy simulation is approximately 1 % and the relative error in the maximum dose between codes are 1.4 % being the same coordinates. It is easy to check that when the obtained PENELOPE/penEasy sigmas are taken into account the relative errors improve. In the lights of all these results, it may be concluded that the updated MC-GPU provides results as good as the PENELOPE/penEasy ones but in much less time. The simulation which uses 100 keV monoenergetic is repeated with PENELOPE/penEasy simulating 1.8·1010 histories and dose and maximum dose precision do not improve significantly, however the real simulation time increases exponentially 23 . Figure 20: Simulation 6 dose distribution using RQR 6 23 Results at: SIMULATION 6 LONG 48 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 6.5.5 Simulation 7: Duke’s chest tests This section explains the last package of simulations carried out by MC-GPU programme that were performed in order to analyse the influence of voxels’ size and different spectra on the dose and simulation time. For this analysis, the phantom employed was only the Duke’s chest, formed by 0.1 x 0.1 x 0.1 cm3 voxels. The performance of the programme in a realistic range of energy spectrum wanted to be test so that the spectra used were 100 keV monoenergetic and a bunch of RQR spectra from 80 kVp (RQR 6) to 120 kVp (RQR 9). In all simulations X-ray beams are parallel to Y axis, PA projection. SIMULATION 7 Phantom Nº Voxels 580 x 260 x 300 Voxel size (cm3) 0.1 x 0.1 x 0.1 Dimensions (cm3) 58 x 26 x 30 Materials Human body materials Source Focal position (cm) 29 -32.65 15 Spectra 100 keV monoenergetic and RQRs 6 - 9 Histories 1006720000 Detector size (cm2) 122.8 x 61.4 Source-to-detector distance (cm) 68.65 Table 15: Simulation 7 The simulation which uses a monoenergetic spectrum is also repeated with PENELOPE/penEasy using 1.2·1010 histories. In this case, dose and maximum dose precision do not improve significantly, however the real simulation time increases. These results are at SIMULATION 7 LONG. All the results tables obtained in this package of simulations are included in the APPENDIX C: SIMULATIONS’ RESULTS because the tables which include the values and the comparison between programmes are too large 24 . The following tables and plots summarize the most important results. Table 16: Simulation 7 dose relative errors 24 Results at: SIMULATION 7 100 keV Monoenergetic RQR 6 RQR 7 RQR 8 RQR 9 Air 0.17% 0.18% 0.14% 0.08% 0.20% Muscle 0.09% 0.15% 0.16% 0.16% 0.17% Soft tissue 0.09% 0.88% 0.72% 0.54% 0.36% Bone 0.12% 0.26% 0.19% 0.14% 0.06% Cartilage 0.11% 1.12% 0.75% 0.69% 0.50% Adipose 0.02% 0.30% 0.28% 0.24% 0.18% Blood 0.10% 0.19% 0.10% 0.19% 0.02% Skin 0.03% 0.45% 0.42% 0.36% 0.28% Lung 0.17% 0.65% 0.51% 0.42% 0.21% Dose Relative Errors 49 RESULTS AND DISCUSSION Table 17: Simulation 7 real simulation times Figure 21: Simulation 7 dose relative errors Figure 22: Simulation 7 real simulation time Spectra MC-GPU PENEASY 100 keV Monoenergetic 128.54 61326.63 RQR 6 108.12 49790.51 RQR 7 110.86 51640.70 RQR 8 113.65 51355.24 RQR 9 116.13 54053.39 Real Simulation Time (s) 0,00% 0,10% 0,20% 0,30% 0,40% 0,50% 0,60% 0,70% 0,80% 0,90% 1,00% 1,10% 1,20% Air Muscle Soft tissue Bone Cartilage Adipose Blood Skin Lung Dose Relative Error Materials 100 keV Monoenergetic RQR 6 RQR 7 RQR 8 RQR 9 110 100 1000 10000 100000 100 keV Monoenergetic RQR 6 RQR 7 RQR 8 RQR 9 Real Simulation Time (s) PENEASY MC-GPU 50 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Analysing the graphs, some conclusion can be extracted: - The widely distributed materials (skin) are materials with large relative errors. - All materials have low dose relative errors for a 100 keV monoenergetic spectrum. - For most of materials, the higher the mean energy of the spectrum, the lower dose relative error. - The used spectrum has not influence in the simulation time. The ratio between the MC-GPU and PENELOPE/penEasy simulation times is within 400 and 500. In spite of everything mentioned before, the maximum relative error is approximately 1.1 %, what highlights the precision of the MC-GPU code in terms of dose calculations. The repetition of the 100 keV monoenergetic spectrum simulation using a larger number of histories does not provide a significate results due to the large standard deviation in the maximum dose point in PENELOPE/penEasy. The dose relative errors of the materials are as precise as the simulations explained above. Once few simulations are carried out and the obtained results are satisfactory, the MCGPU Beta validation begins. At this point, the new material files were considered correct and the algorithm that compute the dose in MC-GPU validated. Figure 23: Simulation 7 using RQR 9 dose distribution 51 RESULTS AND DISCUSSION 6.6 MC-GPU BETA VALIDATION MC-GPU Beta is an extended version of the basic programme and includes a lot of new features as operator inclusion, shielding and one spectrum for each projection. However, this project is only focused on the validation of patient dose calculation and the source rotation is included. In this section, the Duke phantom, new materials and realistic spectra were used. The complexity of this version are the angles and the rotations. Once the results obtained in MC-GPU were considered as enough accurate, the results obtained with MC-GPU Beta were compared to MC-GPU results and experimental measurements. 6.6.1 Simulation 8: Two projections The first simulation run with MC-GPU Beta had two projections, PA and AP, with the Duke’s phantom and focusing the beam to the chest. The simulation was repeated with MC-GPU to compare the results. 25 SIMULATION 8 Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Initial focal position (cm) 30.5 -30.5 130 C-arm radius (cm) 46 Spectra RQR 8 (PA) and RQR 9 (AP) MC-GPU Beta Simulation time (s) 150 MC-GPU histories per simulation 1006720000 Detector size (cm2) 143 x 143 Source-to-detector distance (cm) 71.5 Table 18: Simulation 8 Table 19: Simulation 8 dose results The energy deposition relative errors for each material are all below 0.1 % (excluding air). Moreover, even with a higher standard deviation in the maximum point dose, the relative errors are less than 0.7 %. The coordinates of the maximum are the same in both codes. 25 Results at: SIMULATION 8 Dose (eV/g·hist) MC-GPU 1 MC-GPU 2 MC-GPU Total MC-GPU Beta Relative Error Air 0.12815 0.18678 0.31493 0.32 0.174% Muscle 0.23431 0.19600 0.43031 0.43 0.005% Soft tissue 0.16054 0.20567 0.36621 0.37 0.003% Bone 0.68175 0.46335 1.1451 1.15 0.000% Cartilage 0.07868 0.18722 0.2659 0.27 0.101% Adipose 0.17586 0.16412 0.33998 0.34 0.000% Blood 0.13217 0.34773 0.4799 0.48 0.023% Skin 0.25623 0.20848 0.46471 0.46 0.026% Lung 0.33691 0.31644 0.65335 0.65 0.009% 52 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 6.6.2 Simulation 9: Main projections After the outstanding results, a simulation with eight projections was performed. This simulation is more complex because the rotation angles gain importance. The position of each projection is shown in the figure. For this simulation the Duke’s phantom is also used focusing the beam into the chest 26 . Figure 24: Projections and patient orientation (stylized phantoms) 27 SIMULATION 9 Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Initial focal position (cm) 30.5 -29.5 124 C-arm radius (cm) 45 Spectra RQR 6 and RQR 7 MC-GPU Beta Simulation time (s) 150 MC-GPU histories per simulation 1006720000 Detector size (cm2) 55 x 55 Source-to-detector distance (cm) 81 Table 20: Simulation 9 The rotations of the projections are referred to the patient reference system and the PA initial position. The election of these projections is because they are the most common on interventional procedures. Projections PA, AP and the rotated around Z axis used the RQR 6 spectrum and the projection caudal and cranial used the RQR 7 spectrum. This was to check the automatic change of spectrum in different projections, one of the MC-GPU Beta new features. 26 Results at: SIMULATION 9 27 EURADOS Report 2012-02 ORAMED: Optimization of Radiation Protection of Medical Staff 53 RESULTS AND DISCUSSION The obtained MC-GPU results for different projects were added and averaged to be compared with the MC-GPU Beta which does not sum the values automatically but it averages them. All the obtained results are included in SIMULATION 9 in the appendices section. The comparison generates exceptional results, the dose and energy deposition obtained in each projection are almost the same, the individual differences between them are below 0.1 %. Total energy deposition relative error is a little higher (0.2 %) than the individual ones but it is an outstanding result. It is important to point out that the Air relative error is 100 % due to the MC-GPU Beta eliminates automatically the energy deposited in materials with such low density. The conclusion is that the rotation of the source around the reference system axis are the expected and the results are incredible even when the spectra are not as simple as a monoenergetic one. In this case, the simulation time comparison cannot be performed because in MC-GPU Beta the simulations are configured by time and not by the number of histories. 54 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 6.7 EXPERIMENTAL MEASUREMENTS This is the last part of the validation and it includes the preparation of the experimental measurements, the simulations and the comparative between them. 6.7.1 Experiment preparation Before the experimental measurement a preparation process was carried out. In this process the different experiments were established and the thermoluminescent dosimeters (TLDs) were selected and treated at the INTE's TLD Laboratory. With the available slabs of the equivalent materials (bone, lung and plastic water) and different thickness, some phantoms were designed. This project only focuses on plastic water phantom whose dimensions are 30 x 15 x 30 cm3. Regarding the TLDs, they were erased by a thermal treatment and calibrated. 6.7.2 Experimental measurement The photo shows the real set up of the plastic water sheets at the Vall d’Hebron Hospital. The image intensifier, the plastic supports and the couch also appear in the photo. As it may be appreciated, the couch is without the mattress. Below the phantom there was an ionization chamber to measure the air kerma and the irradiation time for each projection. All data referred to the distance source-to-detector and the spectrum was provided by the computer (DICOM). After the experiments, some irradiation where performed allocated the ionization chamber below and over the couch to compute the attenuation due to the couch. Table 21: Experimental measurement geometry data Floor-phantom 84.5 cm Floor-source 29.5 cm Intensifier 25x25 cm2 FDD 100 cm Phantom 30*15*30 cm3 Phantom-source 55 cm DATA Figure 25: Experimental measurement assembly 55 RESULTS AND DISCUSSION Following image show a schematic draw of the whole assembly for a better visualization of the experiment. Figure 26: Experimental measurement scheme With this configuration, different irradiations were performed: •PA and TLD’s at 1 cm depth within the plastic water phantom, central axis •PA and TLD’s at 7 cm depth within the plastic water phantom, central axis •PA and TLD’s at 13 cm depth within the plastic water phantom, central axis 6.7.3 Simulation 10 Before carry out the simulations, some additional tasks were performed: - Material creation: knowing the chemical composition and the density of the plastic water the slabs are made of, the material was generated in PENELOPE 2014 (.mat) and converted into MC-GPU material data file format (.mcgpu). - Couch thickness estimation: knowing the air kerma for a spectrum below and above the couch without and knowing the inherent and added filtration of the xray tube employed, the XCOMP5r was used to estimate the thickness of equivalent aluminium, it was 0.8 mm Al. - Voxelized geometry creation: this was created taking into account the 0.8 mm Al, 2 cm air and the real dimensions of the phantom. The result was a 9 x 453 x 9 voxels phantom whose unitary dimensions are 3.33 x 0.04 x 3.33 cm3. Figure 27: Simulation 10 scheme 62 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 63 PROJECT EXECUTION 10. PROJECT EXECUTION This part of the thesis shows the scheduling of the project, its budget and the work packages in which the project is divided. It is important to note that this section explains how the real project was carried out, which is different from what was originally planned. 10.1 WORK SCHEDULE Following the work packages are explained. Those are separated in six stages chronologically organized. Figure 28: Work Packages Stage 1 •Bibliography study •PENELOPE/penEasy learning •MC-GPU learning Stage 2 •User Manual elaboration Stage 3 •Version comparison •Material generator code updating •Materials validation •LIbraries elaboration Stage 4 •MC-GPU second simulations •Results analysis Stage 5 •MC-GPU Beta simulations •Results analysis Stage 6 •Thesis report writing •Thesis presentation preparation 64 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures STAGE 1 • Bibliography study: the project begins with a learning stage through the reading of books and articles on physics and computation, related papers and the documentation provided by the MC-GPU code author. • PENELOPE/penEasy learning: since this programme is used during the project it is necessary to know how to use it. Reading of the bibliography provided by the authors and some explained examples were carried out. Also, learning how to create geometries and materials. • MC-GPU learning: understand the necessary inputs to carry out the simulations and the output files as well as learn how to obtained are the tasks included in this step. STAGE 2 • User Manual elaboration: the preparation of the user manual takes place during the whole stage, as the initial tests were carried out. STAGE 3 (Delayed. It was not executed as planned) • Version comparison: the comparison of the PENELOPE’s versions. • Material generator code updating: once the option to update the material generator code is chosen, the study of the code takes place. Since the code is written in Fortran, learning basic concepts of this programming language is necessary. • Materials validation: material files are generated using the updated code. These materials are used to repeat the simulations carried out in the previous stage and the results are compared. • Libraries elaboration: a materials library is generated with the updated material generator code and a spectra library with the XCOMP5r. STAGE 4 • MC-GPU second simulations: this stage was planned but delayed by the tasks not initially planned. Simulations with more complex geometries and spectra are carried out. Same simulations are done with PENELOPE/penEasy. • Results analysis: results of both programmes are compared to validate de results. STAGE 5 • MC-GPU Beta simulations: this stage was also delayed. The most complex simulations emulating a real procedure are carried out in this stage. • Results analysis: results are compared with some experimental measurements. STAGE 6 • Thesis report writing: documentation of all the project in the Final Master Thesis. • Thesis presentation preparation: preparation of the audio-visual support used in the presentation of the project. 65 PROJECT EXECUTION 10.1.1 GANTT DIAGRAM Figure 29: Gantt Diagram 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 STAGE 1 Bibliography study 1 4 1 4 100% PENELOPE/penEasy learning 2 3 2 3 100% MC-GPU learning 3 2 3 2 100% STAGE 2 User Manual elaboration 5 11 511 100% STAGE 3 Version comparison 5 9 5 9 100% Material generator code updating 0 0 7 7 100% Materials validation 0 0 14 1100% Libraries elaboration 0 0 15 1100% Christmas Holidays 16 2 STAGE 4 MC-GPU second simulations 0 0 18 4100% Results analysis 0 0 22 1100% STAGE 5 MC-GPU Beta simulations 0 0 18 4100% Results analysis 0 0 22 1100% STAGE 6 Thesis report writting 23 423 4100% Thesis presentation preparation 27 127 1100% Holy Week Holidays 28 1 JANUARY FEBRUARY MARCH REAL DURATION NOT EXECUTED AS PLANNED Legend: ACTIVITY PLANNED BEGINNING PLANNED DURATION REAL BEGINNING REAL DURATION COMPLETED PERCENTAGE REAL BEGINNING EXECUTED AS PLANNED SEPTEMBER OCTOBER NOVEMBER DICEMBER Holidays 66 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 10.1.2 PERT DIAGRAM Figure 30: Pert Diagram Table 23: Work Packages Duration Precedent Duration Precedent A Bibliography study 4 - I MC-GPU second simulations 4 H B PENELOPE/penEasy learning 3 - J Results analysis 1 I C MC-GPU learning 2 - K MC-GPU Beta simulations 4 H D User Manual elaboration 11 A,B,C L Results analysis 1 K E MC-GPU first simulations 9 A,B,C M Thesis report writing 4 J,L F Material generator code updating 7 - N Thesis presentation preparation 1 M G Materials validation 1 E,F H Libraries elaboration 1 G Task Task STAGE 1 STAGE 2 STAGE 3 STAGE 4 STAGE 5 STAGE 6 67 PROJECT EXECUTION 10.1.3 CRITICAL RISKS IN THE IMPLEMENTATION As mentioned at the beginning of this section, the scheduling showed in this thesis (Gantt and PERT diagrams) are the real and the executed ones. For this reason, the critical path and the critical tasks that delay the projects are known. First of all, the stage of learning, stage one, is one of the most important due to it is in the critical path of the project whose duration depends on it. This step is important because huge volume of information has to be analysed. The version comparison took more time than the expected This incident changed the original schedule delaying the following tasks. Once that problem was solved, the simulation package was carried out, and this is one of the most important points. The time of simulations preparation, the simulation time with PENELOPE/penEasy and the number of simulations is the third key parameter which was critical in the project duration. To avoid problems due to these three factors, the simulations are carried out in the INTEUPC cluster Argos2, which allows running simulations in parallel. Moreover, as the simulations duration are too large, the simulations are launched to run during the nights. On the other hand, as MC-GPU and PENELOPE/penEasy simulations are run in different ports of the cluster, they are prepared and run in parallel. Last steps about results analysis and documentation do not show complexity. 68 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 10.2 TIME AND COST ANALYSIS This section includes the used equipment during the project and tables of costs and dedication time. This is a seven months’ project, from September 2017 to March 2018. The diary dedication is of 4 hours. The total number of hours is approximately 560 hours. The labour cost of this project is summered in the following table: CONCEPT COST (€) Junior Engineer 6720 Senior / Advisors 4200 TOTAL 10920 Table 24: Human Resources costs The budget of this project is very simple due to the most of the project is based on computer simulations. CONCEPT UNITARY COST (€/unit) UNITS (unit) TOTAL COST (€) PC 800 1 800 Pack of 500 sheets 3 1 3 TOTAL 803 Table 25: Equipment costs Complementary costs of consumption of the project are approximately: CONCEPT UNITARY COST UNITS TOTAL COST (€) Electricity29 0.13 €/kWh 1 kW · 560h 72.80 Computer electricity 0.13 €/kWh 0.5 kW · 560 h 36.40 Water30 0.00166 €/L 50 L/d · 140 d 11.62 TOTAL 120.82 Table 26: Complementary costs The estimated cost of the whole project is 11843.82€ 29 Electricity mean price obtained from ENDESA webpage. 30 Data provided by OCU. The Price is the Spanish mean Price. 69 APPENDICES 11. APPENDICES In this section, three different appendices are included. Each appendix corresponds to a stage of the project and they are organized chronologically. Appendix A: MC-GPU User Manual v1.0 Appendix B: This appendix includes the updated material generator code as well as part of a material data file and entire spectrum file. The original versions of codes and materials may be found in the download package provided by the MC-GPU coder. Appendix C: Results of all simulations, included or not in this thesis, carried out during the project are collected in different Excel files, most of the results are attached in this appendix. 11.1 APPENDIX A: USER MANUAL The User Manual elaborated during the project is a completely independent document with its own cover, index and bibliography, for this reason, it is a document attached to and not included in this thesis. 70 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 11.2 APPENDIX B: UPDATED CODE AND NEW FILES 11.2.1 Code update process The first matter to take into account before modify Fortran codes based on PENELOPE code is to know that in the latest version of PENELOPE, 2014 version, the RITA initialization functions are in a separate subroutine package called rita.f. Attending to the code MC-GPU_create_material_data.f itself, it may be noticed that the variables used on it are the same which were used in PENELOPE 2006 and they have been named different in PENELOPE 2014, even some of them also have changed its meaning. For this reason, the first task is identifying in the new PENELOPE code the variables necessary in the materials generator code. Following the modifications and additions in the new MC-GPU materials generator code are showing. Also, some comparative screenshots between the old and new codes are presented. • At the beginning, the Fortran files penelope.f and rita.f files (both 2014 version) are included. • The penelope.f (2014 version) had been previously modified: - PENELOPE_mod and TRACK-mod are disabled - MAXMAT and NEGP parameters are added manually and also few COMMON blocks. • Variables initialization Figure 31: Variables initialization - The main modifications are made in the COMMON blocks which depends directly of the penelope.f file. - Moreover, few new variables are added in the auxiliary array section. 71 APPENDICES • Inputs demand: the part in which the user must introduce the variables and the name of the input and output files has not been modified with the only exception of names format. Figure 32: Inputs demand • PENELOPE initialization with material information: the parameters which initialize PENELOPE are the same. In both cases the Fortran code calls the PENELOPE subroutine PEINIT to get the material information. Figure 33: PENELOPE initialization with material information • Next part of the code has not suffered any modification because is just the part in which the material (the only one) and the type of particle (photon) are chosen and also the header of the output file is written. 78 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 11.2.3 Air_5_120keV.mcgpu Figure 37: Air 5-120 keV material file […] […] 79 APPENDICES 11.2.4 RQR6.spc Figure 38: RQR 6 spectrum file 80 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures This spectrum is shown in the following plot to visualize it: Figure 39: RQR 6 spectrum distribution 0 500 1000 1500 2000 2500 3000 010000 20000 30000 40000 50000 60000 70000 80000 Intensity [] Energy [eV] 81 APPENDICES 82 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 11.3 APPENDIX C: SIMULATIONS’ RESULTS SIMULATION 1 SIMULATION 1 Phantom Nº Voxels 1 x 1 x 1 Voxel size (cm3) 10 x 10 x 10 Dimensions (cm3) 10 x 10 x 10 Material Dry air ( ρ = 0.001205 g/cm3) Source Focal position (cm) 5 -15 5 Spectrum Monoenergetic spectra (60 – 120 keV) Histories 1.0E9 Detector size (cm2) 20 x 20 Source-to-detector distance (cm) 30 83 APPENDICES Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference New Difference Old Relative error New Relative error Improvement factor 60 11.61226 11.47738 11.436 0.176 0.042 1.543% 0.364% 4.244 70 11.54258 11.44943 11.418 0.124 0.031 1.088% 0.273% 3.992 80 12.22802 12.15716 12.115 0.113 0.042 0.935% 0.350% 2.669 90 13.34371 13.28865 13.256 0.087 0.032 0.659% 0.244% 2.702 100 14.75085 14.71027 14.650 0.100 0.060 0.686% 0.409% 1.678 110 16.36093 16.32939 16.252 0.109 0.077 0.668% 0.474% 1.409 120 18.08912 18.06191 17.984 0.105 0.078 0.582% 0.431% 1.351 200 33.94609 33.830 0.116 0.344% 300 54.81751 54.566 0.252 0.462% Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference New Difference Old Relative error New Relative error Improvement factor 60 13.99 13.83 13.780 0.210 0.050 1.523% 0.362% 4.206 70 13.91 13.80 13.759 0.151 0.041 1.097% 0.297% 3.689 80 14.73 14.65 14.598 0.132 0.052 0.903% 0.355% 2.544 90 16.08 16.01 15.974 0.106 0.036 0.664% 0.226% 2.939 100 17.77 17.73 17.654 0.116 0.076 0.659% 0.432% 1.524 110 19.71 19.68 19.584 0.126 0.096 0.643% 0.490% 1.313 120 21.80 21.76 21.671 0.129 0.089 0.594% 0.410% 1.450 200 40.91 40.765 0.145 0.356% 300 66.06 65.752 0.309 0.469% ENERGY DEPOSITION (eV/hist) DOSE (eV/g·hist) 84 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures SIMULATION 2 SIMULATION 2 Phantom Nº Voxels 6 x 3 x 6 Voxel size (cm3) 5 x 10 x 5 Dimensions (cm3) 30 x 30 x 30 Material PMMA ( ρ = 1.19 g/cm3) Source Focal position (cm) 15 -45 15 Spectrum Monoenergetic spectra (60 – 120 keV) Histories 1.0E9 Detector size (cm2) 54 x 54 Source-to-detector distance (cm) 81 85 APPENDICES Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference Old Relative error New Difference New Relative error 60 0.808 0.801 0.802 0.0053 0.666% 0.0009 0.116% 70 0.888 0.882 0.883 0.0049 0.555% 0.0011 0.124% 80 0.983 0.978 0.979 0.0045 0.456% 0.0011 0.114% 90 1.090 1.085 1.086 0.0041 0.378% 0.0011 0.100% 100 1.206 1.201 1.202 0.0038 0.318% 0.0010 0.086% 110 1.328 1.324 1.325 0.0036 0.273% 0.0010 0.074% 120 1.456 1.452 1.453 0.0033 0.231% 0.0009 0.065% 200 2.591 2.592 0.0008 0.030% 300 4.114 4.115 0.0006 0.015% Energy Spectrum (keV) MC-GPU NEW MC-GPU penEasy Old Difference Old Relative error New Difference New Relative error 60 25947.550 25745.970 25776.000 171.550 0.666% 30.030 0.117% 70 28540.420 28347.670 28382.900 157.520 0.555% 35.230 0.124% 80 31596.950 31417.570 31453.500 143.450 0.456% 35.930 0.114% 90 35020.840 34854.070 34888.900 131.940 0.378% 34.830 0.100% 100 38732.970 38577.050 38610.300 122.670 0.318% 33.250 0.086% 110 42673.680 42526.050 42557.500 116.180 0.273% 31.450 0.074% 120 46793.210 46655.240 46685.500 107.710 0.231% 30.260 0.065% 200 83241.110 83266.300 25.190 0.030% 300 132191.460 132211.000 19.540 0.015% DOSE (eV/g·hist) ENERGY DEPOSITION (eV/hist) 86 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures SIMULATION 3 SIMULATION 3 Phantom Nº Voxels 3 x 3 x 3 Voxel size (cm3) 10 x 10 x 10 Dimensions (cm3) 30 x 30 x 30 Materials PMMA and Dry Air (1voxel) Source Focal position (cm) 15 -45 15 Spectrum N100 Histories 1006720000 Detector size (cm2) 54 x 54 Source-to-detector distance (cm) 81 87 APPENDICES MC-GPU original MC-GPU updated PENELOPE/penEasy Original Relative Error Updated Relative Error Total Energy Absorbed (eV/hist) 32322.85 32144.36 32180.97 0.441% 0.114% Maximum voxel dose (ev/g·hist) 2.851025 2.831048 2.8371 0.493% 0.212% Sigma 0.000845 0.000838 0.00060 Coordinates (1,0,1) (1,0,1) (2,1,2)=(1,0,1) Dose ROI (ev/g·hist) 1.057630 1.05179 1.05299 0.441% 0.114% Sigma << 1% << 1% E dp (ev/hist) 32321.29 32142.82 32179.40 0.441% 0.114% Sigma - - 1.90 Mass (g) Dose ROI (ev/g·hist) 1.292280 1.28424 1.301278008 0.691% 1.309% Sigma 0.011590 0.01153 E dp (ev/hist) 1.56 1.55 1.5680 0.513% 1.150% Sigma - - 0.014 Mass (g) Real simulation time (s) 49.30 59.72 20357.19 30560.000010 1.205 PMMA Dry Air 94 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Energy dep (eV/hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 53.83 53.67 0.297% 53.48 0.650% 0.351% Muscle 6935.92 6948.03 0.175% 6924.55 0.164% 0.339% Soft tissue 270.94 272.9 0.723% 271.00 0.021% 0.702% Bone 7474.56 7339.13 1.812% 7321.85 2.086% 0.236% Cartilage 44.34 44.3 0.090% 44.41 0.163% 0.254% Adipose 2.62 2.62 0.000% 2.62 0.120% 0.120% Blood 497.01 490.56 1.298% 461.29 7.744% 6.346% Skin 6066.87 5979.67 1.437% 6292.37 3.584% 4.970% Lung 961.86 967.77 0.614% 931.39 3.271% 3.906% Glands 5.1 5.07 0.588% 4.98 2.428% 1.826% Brain 47.17 48.15 2.078% 48.03 1.781% 0.260% Red Marrow 285.84 289.34 1.224% 288.18 0.813% 0.402% Liver 1018.19 1018.33 0.014% 1017.24 0.093% 0.107% Stomach 2140.63 2134.55 0.284% 2124.25 0.771% 0.485% Water 323.32 322.93 0.121% 321.57 0.545% 0.424% TOTAL 26128.19 25917.03 0.808% 26107.2111 0.080% 0.728% Maximum dose point (eV/g·hist) 6.125987 5.989736 3.7912 Sigma 0.002147 0.001928 0.018 Sigma/mean 0.035% 0.032% 0.475% Coordinates (72,2,116) (72,15,106) (65,24,131) Real simulation time (s) 87.25 128.32 7.63E+05 95 APPENDICES PENELOPE/penEasy Sigma MAX PenEasy MIN PenEasy Superior Relative Error Relative error Inferior Relative Error Minimum Error Maximum Error Air 53.48 0.0 5.35E+01 5.35E+01 0.32% 0.35% 0.39% 0.32% 0.39% Muscle 6924.55 0.2 6.92E+03 6.92E+03 0.34% 0.34% 0.34% 0.34% 0.34% Soft tissue 271.00 0.0 2.71E+02 2.71E+02 0.69% 0.70% 0.72% 0.69% 0.72% Bone 7321.85 0.3 7.32E+03 7.32E+03 0.23% 0.24% 0.24% 0.23% 0.24% Cartilage 44.41 0.0 4.44E+01 4.44E+01 0.29% 0.25% 0.21% 0.21% 0.29% Adipose 2.62 0.0 2.63E+00 2.62E+00 0.26% 0.12% 0.02% 0.02% 0.26% Blood 461.29 0.1 4.61E+02 4.61E+02 6.33% 6.35% 6.36% 6.33% 6.36% Skin 6292.37 0.2 6.29E+03 6.29E+03 4.97% 4.97% 4.97% 4.97% 4.97% Lung 931.39 0.1 9.31E+02 9.31E+02 3.90% 3.91% 3.92% 3.90% 3.92% Glands 4.98 0.0 4.98E+00 4.97E+00 1.71% 1.83% 1.94% 1.71% 1.94% Brain 48.03 0.0 4.80E+01 4.80E+01 0.22% 0.26% 0.30% 0.22% 0.30% Red Marrow 288.18 0.0 2.88E+02 2.88E+02 0.39% 0.40% 0.42% 0.39% 0.42% Liver 1017.24 0.1 1.02E+03 1.02E+03 0.10% 0.11% 0.12% 0.10% 0.12% Stomach 2124.25 0.1 2.12E+03 2.12E+03 0.48% 0.48% 0.49% 0.48% 0.49% Water 321.57 0.1 3.22E+02 3.22E+02 0.41% 0.42% 0.44% 0.41% 0.44% Updated MC-GPU validation 96 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures SIMULATION 6 SIMULATION 6 (Duke) Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Focal position (cm) 30.5 -30.5 130 Spectra 100 keV monoenergetic and RQR 6 Histories 1006720000 Detector size (cm2) 143 x 143 Source-to-detector distance (cm) 71.5 97 APPENDICES Spectrum: 100 keV monoenergetic Energy dep (eV/hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 44.41 44.29 0.270% 44.27 0.323% 0.052% Muscle 11959.95 11906.22 0.449% 11914.90 0.378% 0.073% Soft tissue 1461.58 1454.86 0.460% 1457.04 0.312% 0.150% Bone 6513.28 6521.7 0.129% 6527.99 0.225% 0.096% Cartilage 293.2 291.94 0.430% 292.07 0.386% 0.046% Adipose 3712.12 3682.66 0.794% 3684.46 0.751% 0.049% Blood 377.72 371.22 1.721% 371.39 1.704% 0.046% Skin 1610.44 1592.23 1.131% 1592.02 1.157% 0.013% Lung 1587.82 1581.7 0.385% 1584.38 0.217% 0.169% TOTAL 27560.52 27446.82 0.413% 27468.5211 0.335% 0.079% Dose (eV/g·hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 0.13024 0.12989 0.269% 0.1298 0.330% 0.061% Muscle 0.33006 0.32858 0.448% 0.3288 0.377% 0.073% Soft tissue 0.32366 0.32217 0.460% 0.3226 0.313% 0.149% Bone 0.82896 0.83003 0.129% 0.8308 0.225% 0.096% Cartilage 0.18801 0.18720 0.431% 0.1873 0.386% 0.047% Adipose 0.30844 0.30599 0.794% 0.3061 0.751% 0.050% Blood 0.31869 0.31321 1.720% 0.3134 1.703% 0.046% Skin 0.2969 0.29354 1.132% 0.2935 1.158% 0.013% Lung 0.62977 0.62735 0.384% 0.6284 0.217% 0.168% Maximum dose point 4.188542 4.000672 4.2011 Sigma 0.001671 0.001422 2.5 Sigma/mean 0.040% 0.036% 59.508% Coordinates (40,4,273) (45,2,256) (70,1,256) Real simulation time (s) 76.25 127.95 42225.31 98 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures PENELOPE/penEasy Sigma Sigma/Mean MAX PenEasy MIN PenEasy Superior Relative Error Relative error Inferior Relative Error Minimum Error Maximum Error 44.27 0.07 0.16% 44.34 44.20 0.11% 0.05% 0.21% 0.05% 0.21% 11914.90 1.40 0.01% 11916.30 11913.50 0.08% 0.07% 0.06% 0.06% 0.08% 1457.04 0.50 0.03% 1457.54 1456.54 0.18% 0.15% 0.12% 0.12% 0.18% 6527.99 1.30 0.02% 6529.29 6526.69 0.12% 0.10% 0.08% 0.08% 0.12% 292.07 0.20 0.07% 292.27 291.87 0.11% 0.05% 0.02% 0.02% 0.11% 3684.46 0.64 0.02% 3685.10 3683.82 0.07% 0.05% 0.03% 0.03% 0.07% 371.39 0.24 0.06% 371.63 371.15 0.11% 0.05% 0.02% 0.02% 0.11% 1592.02 0.44 0.03% 1592.46 1591.58 0.01% 0.01% 0.04% 0.01% 0.04% 1584.38 0.52 0.03% 1584.90 1583.86 0.20% 0.17% 0.14% 0.14% 0.20% Updated MC-GPU validation 99 APPENDICES Spectrum: RQR 6 Energy dep (eV/hist) MC-GPU updated PENELOPE/penEasy Updated Relative Error Air 47.01 47.03 0.040% Muscle 8217.67 8222.13 0.054% Soft tissue 622.83 619.28 0.573% Bone 4965.44 4949.00 0.332% Cartilage 99.72 98.80 0.929% Adipose 1994.75 1999.26 0.226% Blood 126.87 126.47 0.313% Skin 1445.43 1451.83 0.441% Lung 743.69 738.77 0.666% TOTAL 18263.42 18252.5758 0.059% Dose (eV/g·hist) MC-GPU updated PENELOPE/penEasy Updated Relative Error Air 0.13787 0.1379 0.029% Muscle 0.22679 0.2269 0.053% Soft tissue 0.13792 0.1371 0.572% Bone 0.63196 0.6299 0.332% Cartilage 0.06394 0.0634 0.923% Adipose 0.16574 0.1661 0.228% Blood 0.10704 0.1067 0.309% Skin 0.26648 0.2677 0.439% Lung 0.29497 0.2930 0.667% Maximum dose point 7.232274 7.3318 1.357% Sigma 0.001633 0.067 Sigma/mean 0.02% 0.91% Coordinates (58,5,271) (58,5,271) Real simulation time (s) 112.96 33641.11 100 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures PENELOPE/penEasy Sigma Sigma/Mean MAX PenEasy MIN PenEasy Superior Relative Error Relative error Inferior Relative Error Minimum Error Maximum Error 47.03 0.08 0.16% 47.10 46.95 0.20% 0.04% 0.12% 0.04% 0.20% 8222.13 0.97 0.01% 8223.10 8221.16 0.07% 0.05% 0.04% 0.04% 0.07% 619.28 0.30 0.05% 619.58 618.98 0.52% 0.57% 0.62% 0.52% 0.62% 4949.00 0.87 0.02% 4949.87 4948.13 0.31% 0.33% 0.35% 0.31% 0.35% 98.80 0.12 0.12% 98.92 98.68 0.81% 0.93% 1.05% 0.81% 1.05% 1999.26 0.47 0.02% 1999.73 1998.79 0.25% 0.23% 0.20% 0.20% 0.25% 126.47 0.13 0.10% 126.60 126.34 0.21% 0.31% 0.42% 0.21% 0.42% 1451.83 0.41 0.03% 1452.24 1451.42 0.47% 0.44% 0.41% 0.41% 0.47% 738.77 0.33 0.04% 739.10 738.44 0.62% 0.67% 0.71% 0.62% 0.71% Updated MC-GPU validation 101 APPENDICES SIMULATION 6 LONG SIMULATION 6 (Duke) Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Focal position (cm) 30.5 -30.5 130 Spectra 100 keV monoenergetic Histories MC-GPU 1006720000 Histories penEasy 1.8 · 1010 Detector size (cm2) 143 x 143 Source-to-detector distance (cm) 71.5 102 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Energy dep (eV/hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 44.41 44.29 0.270% 44.26 0.328% 0.057% Muscle 11959.95 11906.22 0.449% 11915.00 0.377% 0.074% Soft tissue 1461.58 1454.86 0.460% 1456.98 0.316% 0.146% Bone 6513.28 6521.7 0.129% 6528.23 0.229% 0.100% Cartilage 293.2 291.94 0.430% 292.19 0.345% 0.086% Adipose 3712.12 3682.66 0.794% 3685.18 0.731% 0.068% Blood 377.72 371.22 1.721% 371.37 1.711% 0.040% Skin 1610.44 1592.23 1.131% 1592.09 1.153% 0.009% Lung 1587.82 1581.7 0.385% 1584.08 0.236% 0.150% TOTAL 27560.52 27446.82 0.413% 27469.3837 0.332% 0.082% Dose (eV/g·hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 0.13024 0.12989 0.269% 0.1298 0.336% 0.066% Muscle 0.33006 0.32858 0.448% 0.3288 0.376% 0.074% Soft tissue 0.32366 0.32217 0.460% 0.3226 0.317% 0.145% Bone 0.82896 0.83003 0.129% 0.8309 0.229% 0.100% Cartilage 0.18801 0.18720 0.431% 0.1874 0.345% 0.087% Adipose 0.30844 0.30599 0.794% 0.3062 0.731% 0.069% Blood 0.31869 0.31321 1.720% 0.3133 1.709% 0.039% Skin 0.2969 0.29354 1.132% 0.2935 1.154% 0.009% Lung 0.62977 0.62735 0.384% 0.6283 0.236% 0.150% Maximum dose point 4.188542 4.000672 3.8884 7.719% 2.887% Sigma 0.001671 0.001422 0.014 Sigma/mean 0.040% 0.036% 0.360% Coordinates (40,4,273) (45,2,256) (58,6,267) Real simulation time (s) 76.25 127.95 7.62E+05 103 APPENDICES PENELOPE/penEasy Sigma Sigma/Mean MAX PenEasy MIN PenEasy Superior Relative Error Relative error Inferior Relative Error Minimum Error Maximum Error 44.26 0.02 0.04% 44.28 44.25 0.02% 0.06% 0.10% 0.02% 0.10% 11915.00 0.32 0.00% 11915.32 11914.68 0.08% 0.07% 0.07% 0.07% 0.08% 1456.98 0.12 0.01% 1457.10 1456.86 0.15% 0.15% 0.14% 0.14% 0.15% 6528.23 0.30 0.00% 6528.53 6527.93 0.10% 0.10% 0.10% 0.10% 0.10% 292.19 0.05 0.02% 292.24 292.14 0.10% 0.09% 0.07% 0.07% 0.10% 3685.18 0.15 0.00% 3685.33 3685.03 0.07% 0.07% 0.06% 0.06% 0.07% 371.37 0.06 0.02% 371.42 371.31 0.05% 0.04% 0.02% 0.02% 0.05% 1592.09 0.10 0.01% 1592.19 1591.99 0.00% 0.01% 0.02% 0.00% 0.02% 1584.08 0.12 0.01% 1584.20 1583.96 0.16% 0.15% 0.14% 0.14% 0.16% Updated MC-GPU validation 110 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures SIMULATION 7 LONG SIMULATION 7 (Duke Chest) Phantom Nº Voxels 580 x 260 x 300 Voxel size (cm3) 0.1 x 0.1 x 0.1 Dimensions (cm3) 58 x 26 x 30 Materials Human body materials Source Focal position (cm) 29 -32.65 15 Spectra 100 keV monoenergetic Histories MC-GPU 1006720000 Histories penEasy 1.3 · 1010 Detector size (cm2) 122.8 x 61.4 Source-to-detector distance (cm) 68.65 111 APPENDICES Energy dep (eV/hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 20.54 20.51 0.146% 20.50 0.200% 0.053% Muscle 11138.25 11084.77 0.480% 11094.40 0.395% 0.087% Soft tissue 1216.29 1210.71 0.459% 1211.75 0.375% 0.086% Bone 5613.02 5616 0.053% 5622.48 0.168% 0.115% Cartilage 211.89 210.94 0.448% 211.15 0.351% 0.099% Adipose 2777.28 2757.97 0.695% 2758.77 0.671% 0.029% Blood 509.96 501.21 1.716% 501.63 1.660% 0.084% Skin 1419.75 1404.08 1.104% 1404.41 1.092% 0.023% Lung 2873.84 2861.56 0.427% 2866.16 0.268% 0.160% TOTAL 25780.81 25667.74 0.439% 25691.2511 0.349% 0.092% Dose (eV/g·hist) MC-GPU original MC-GPU updated Relative difference PENELOPE/penEasy Original Relative Error Updated Relative Error Air 0.71345 0.71235 0.154% 0.7119 0.221% 0.067% Muscle 1.12123 1.11584 0.481% 1.1168 0.396% 0.087% Soft tissue 0.75935 0.75586 0.460% 0.7565 0.375% 0.086% Bone 2.89662 2.89816 0.053% 2.9015 0.168% 0.115% Cartilage 0.64531 0.64241 0.449% 0.6431 0.350% 0.101% Adipose 0.93262 0.92614 0.695% 0.9264 0.671% 0.029% Blood 0.74960 0.73674 1.716% 0.7374 1.659% 0.085% Skin 1.10749 1.09527 1.103% 1.0955 1.092% 0.024% Lung 1.13985 1.13498 0.427% 1.1368 0.268% 0.161% Maximum dose point 173.3247 179.033718 3.294% 24.1286 618.337% 641.998% Sigma 0.010588 0.010435 24 Sigma/mean 0.006% 0.006% 99.467% Coordinates (575,126, 220) (171,36,24) (179,2,48) Real simulation time (s) 89.18 128.02 7.62E+05 112 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures PENELOPE/penEasy Sigma MAX PenEasy MIN PenEasy Superior Relative Error Relative error Inferior Relative Error Minimum Error Maximum Error 2.05E+01 0.0 2.05E+01 2.05E+01 0.02% 0.05% 0.12% 0.02% 0.12% 1.11E+04 0.4 1.11E+04 1.11E+04 0.09% 0.09% 0.08% 0.08% 0.09% 1.21E+03 0.1 1.21E+03 1.21E+03 0.10% 0.09% 0.08% 0.08% 0.10% 5.62E+03 0.3 5.62E+03 5.62E+03 0.12% 0.12% 0.11% 0.11% 0.12% 2.11E+02 0.0 2.11E+02 2.11E+02 0.12% 0.10% 0.08% 0.08% 0.12% 2.76E+03 0.2 2.76E+03 2.76E+03 0.03% 0.03% 0.02% 0.02% 0.03% 5.02E+02 0.1 5.02E+02 5.02E+02 0.10% 0.08% 0.07% 0.07% 0.10% 1.40E+03 0.1 1.40E+03 1.40E+03 0.03% 0.02% 0.02% 0.02% 0.03% 2.87E+03 0.2 2.87E+03 2.87E+03 0.17% 0.16% 0.15% 0.15% 0.17% Updated MC-GPU validation 113 APPENDICES SIMULATION 8 The main characteristics of the simulations and the results are the following: SIMULATION 8 Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Initial focal position (cm) 30.5 -30.5 130 C-arm radius (cm) 46 Spectra RQR 8 (PA) and RQR 9 (AP) MC-GPU Beta Simulation time (s) 150 MC-GPU histories per simulation 1006720000 Detector size (cm2) 143 x 143 Source-to-detector distance (cm) 71.5 Voxels x number 122.00 x dim vox 0.50 x dim 61 y number 62.00 y dim vox 0.50 y dim 31 z number 372.00 z dim vox 0.50 z dim 186 Volume 0.13 Total volume 351726 114 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Energy dep (eV/hist) MC-GPU 1 MC-GPU 2 MC-GPU Total MC-GPU Beta Relative Error Air 43.7 63.7 107.4 0.00 - Muscle 8490.1 7102.18 15592.28 15591.80 0.003% Soft tissue 724.98 928.76 1653.74 1653.79 0.003% Bone 5356.65 3640.61 8997.26 8997.30 0.000% Cartilage 122.71 291.97 414.68 415.08 0.096% Adipose 2116.53 1975.19 4091.72 4091.70 0.000% Blood 156.65 412.14 568.79 568.91 0.021% Skin 1389.83 1130.86 2520.69 2521.40 0.028% Lung 849.45 797.84 1647.29 1647.12 0.010% TOTAL 19250.61 16343.24 35593.85 35487.10 0.301% Dose (eV/g·hist) MC-GPU 1 MC-GPU 2 MC-GPU Total MC-GPU Beta Relative Error Air 0.12815 0.18678 0.31493 0.32 0.174% Muscle 0.23431 0.19600 0.43031 0.43 0.005% Soft tissue 0.16054 0.20567 0.36621 0.37 0.003% Bone 0.68175 0.46335 1.1451 1.15 0.000% Cartilage 0.07868 0.18722 0.2659 0.27 0.101% Adipose 0.17586 0.16412 0.33998 0.34 0.000% Blood 0.13217 0.34773 0.4799 0.48 0.023% Skin 0.25623 0.20848 0.46471 0.46 0.026% Lung 0.33691 0.31644 0.65335 0.65 0.009% Maximum Point MC-GPU 1 MC-GPU 2 Relative Error Relative Error Edep (eV/hist) 1.688116 1.185994 1.6841 1.1933 -0.238% 0.614% Dose (eV/g·hist) 6.786396 4.767817 6.770278 4.80E+00 -0.238% 0.614% Sigma 0.00164 0.001418 0.06 0.05 Sigma/mean 0.02% 0.03% 0.86% 1.06% Coordinates (58,5,271) (56,44,265) (58,5,271) (56,44,265) Material 4 4 4 4 Voxel mass (g) 0.25 0.25 0.2488 0.2488 MC-GPU Beta 115 APPENDICES SIMULATION 9 The main characteristics of the simulations and the results are the following: SIMULATION 9 Phantom Nº Voxels 122 x 62 x 372 Voxel size (cm3) 0.5 x 0.5 x 0.5 Dimensions (cm3) 61 x 31 x 186 Materials Human body materials Source Initial focal position (cm) 30.5 -29.5 124 C-arm radius (cm) 45 Spectra RQR 6 and RQR 7 MC-GPU Beta Simulation time (s) 150 MC-GPU histories per simulation 1006720000 Detector size (cm2) 55 x 55 Source-to-detector distance (cm) 81 116 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures Source 30.5 -29.5 124 30.5 60.5 124 Direction 0 1 0 0 -1 0 Detector 55 055 55 055 Source to detector dis Spectrum Nº histories MCGPU 81 81 PA RQR 6 1006720000 AP RQR 6 1006720000 Rotation angle 0 0 -90 0 0 90 Direction 0 1 0 0 1 0 Detector 55 055 55 055 Source to detector dis Spectrum Nº histories MCGPU Rotation angle 0 0 -45 0 0 45 Direction 0 1 0 0 1 0 Detector 55 055 55 055 Source to detector dis Spectrum Nº histories MCGPU Rotation angle -45 0 0 45 0 0 Direction 0 1 0 0 1 0 Detector 55 055 55 055 Source to detector dis Spectrum Nº histories MCGPU CAUD 45° CRAN 45° RAO 45° LAO 45° RQR 6 RQR 6 81 81 PA Reference System 1006720000 1006720000 RQR 6 RQR 6 1006720000 1006720000 81 81 RAO 90° LAO 90° 81 81 RQR 7 RQR 7 1006720000 1006720000 117 APPENDICES MC-GPU Results Energy dep (eV/hist) PA AP RAO 90 LAO 90 RAO 45 LAO 45 CAUD 45 CRAN 45 MC-GPU Sum MC-GPU Average Air 27.46 61.62 87.93 98.34 44.36 44.00 29.13 34.35 427.19 53.40 Muscle 13488.84 11794.61 11502.33 10201.78 14402.08 13239.53 12040.85 13703.69 100373.71 12546.71 Soft tissue 2435.16 3258.59 920.41 2251.29 1885.85 2660.19 1284.67 1721.39 16417.55 2052.19 Bone 7693.59 4261.25 4623.38 4196.68 6286.11 6661.93 9468.68 9234.10 52425.72 6553.22 Cartilage 177.07 788.51 252.99 214.15 149.94 152.82 148.46 211.84 2095.78 261.97 Adipose 3420.25 4026.97 2503.23 2445.42 3530.87 3791.75 2942.58 4483.96 27145.03 3393.13 Blood 260.80 950.15 414.90 277.90 335.43 279.71 357.24 160.92 3037.05 379.63 Skin 1637.74 1910.23 1859.83 1719.45 2303.53 2136.18 2425.35 2101.56 16093.87 2011.73 Lung 1375.32 1445.02 525.56 626.64 1017.85 1068.25 3047.78 642.93 9749.35 1218.67 TOTAL 30516.22 28496.94 22690.57 22031.66 29956.03 30034.37 31744.74 32294.75 227765.25 28470.66 Dose (eV/g·hist) PA AP RAO 90 LAO 90 RAO 45 LAO 45 CAUD 45 CRAN 45 MC-GPU Sum MC-GPU Average Air 0.08051 0.18069 0.25785 0.28838 0.13007 0.12902 0.08543 0.10073 1.253 0.157 Muscle 0.37226 0.32550 0.31743 0.28154 0.39746 0.36538 0.33230 0.37819 2.770 0.346 Soft tissue 0.53925 0.72159 0.20382 0.49853 0.41761 0.58908 0.28448 0.38119 3.636 0.454 Bone 0.97918 0.54234 0.58843 0.53412 0.80005 0.84788 1.20510 1.17524 6.672 0.834 Cartilage 0.11355 0.50562 0.16223 0.13732 0.09615 0.09799 0.09520 0.13584 1.344 0.168 Adipose 0.28419 0.33460 0.20799 0.20319 0.29338 0.31506 0.24450 0.37257 2.255 0.282 Blood 0.22005 0.80167 0.35007 0.23448 0.28301 0.23600 0.30141 0.13578 2.562 0.320 Skin 0.30193 0.35217 0.34287 0.31699 0.42467 0.39382 0.44713 0.38744 2.967 0.371 Lung 0.54549 0.57313 0.20845 0.24854 0.40371 0.42370 1.20883 0.25500 3.867 0.483 118 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures MC-GPU Beta Results Energy dep (eV/hist) PA AP RAO 90 LAO 90 RAO 45 LAO 45 CAUD 45 CRAN 45 MC-GPU Beta Sum MC-GPU Beta Average Air 27.48 61.59 87.98 98.34 44.43 44.03 29.13 34.35 427.33 0.00 Muscle 13488.41 11794.14 11501.44 10201.92 14401.78 13239.91 12041.74 13702.26 100371.60 12546.45 Soft tissue 2435.46 3258.97 920.29 2251.64 1885.47 2659.74 1284.39 1721.68 16417.64 2052.21 Bone 7693.52 4261.01 4623.40 4196.59 6287.40 6662.19 9469.82 9235.64 52429.57 6553.70 Cartilage 177.02 788.15 253.00 214.01 149.96 152.83 148.58 211.75 2095.30 261.91 Adipose 3420.88 4026.75 2503.30 2445.69 3531.05 3791.36 2942.70 4484.09 27145.81 3393.23 Blood 260.69 950.33 415.02 277.93 335.56 279.83 357.14 160.98 3037.50 379.69 Skin 1637.82 1910.04 1860.02 1719.34 2303.59 2135.61 2425.69 2101.76 16093.87 2011.73 Lung 1375.28 1444.88 525.68 626.38 1017.92 1068.11 3047.57 643.01 9748.84 1218.61 TOTAL 30489.08 28434.26 22602.14 21933.50 29912.74 29989.58 31717.63 32261.18 227767.45 28417.52 Dose (eV/g·hist) PA AP RAO 90 LAO 90 RAO 45 LAO 45 CAUD 45 CRAN 45 MC-GPU Sum MC-GPU Average Air 0.08059 0.18061 0.25800 0.28837 0.13029 0.12911 0.08542 0.10073 1.253 0.000 Muscle 0.37225 0.32549 0.31741 0.28155 0.39745 0.36539 0.33232 0.37815 2.770 0.346 Soft tissue 0.53931 0.72167 0.20379 0.49861 0.41752 0.58898 0.28442 0.38125 3.636 0.454 Bone 0.97917 0.54231 0.58843 0.53411 0.80021 0.84791 1.20524 1.17544 6.673 0.834 Cartilage 0.11351 0.50539 0.16223 0.13723 0.09616 0.09800 0.09527 0.13578 1.344 0.168 Adipose 0.28424 0.33458 0.20800 0.20321 0.29340 0.31502 0.24451 0.37258 2.256 0.282 Blood 0.21995 0.80183 0.35016 0.23450 0.28313 0.23610 0.30133 0.13583 2.563 0.320 Skin 0.30194 0.35213 0.34291 0.31697 0.42468 0.39372 0.44720 0.38748 2.967 0.371 Lung 0.54547 0.57308 0.20850 0.24844 0.40374 0.42364 1.20875 0.25503 3.867 0.483 119 APPENDICES Comparison Energy dep (eV/hist) PA AP RAO 90 LAO 90 RAO 45 LAO 45 CAUD 45 CRAN 45 Relative Error Sum Relative Error Average Air 0.08% -0.05% 0.06% 0.00% 0.16% 0.07% 0.00% 0.00% 0.03% -100.00% Muscle 0.00% 0.00% -0.01% 0.00% 0.00% 0.00% 0.01% -0.01% 0.00% 0.00% Soft tissue 0.01% 0.01% -0.01% 0.02% -0.02% -0.02% -0.02% 0.02% 0.00% 0.00% Bone 0.00% -0.01% 0.00% 0.00% 0.02% 0.00% 0.01% 0.02% 0.01% 0.01% Cartilage -0.03% -0.05% 0.00% -0.07% 0.01% 0.00% 0.08% -0.04% -0.02% -0.02% Adipose 0.02% -0.01% 0.00% 0.01% 0.01% -0.01% 0.00% 0.00% 0.00% 0.00% Blood -0.04% 0.02% 0.03% 0.01% 0.04% 0.04% -0.03% 0.04% 0.01% 0.01% Skin 0.00% -0.01% 0.01% -0.01% 0.00% -0.03% 0.01% 0.01% 0.00% 0.00% Lung 0.00% -0.01% 0.02% -0.04% 0.01% -0.01% -0.01% 0.01% -0.01% -0.01% TOTAL -0.09% -0.22% -0.39% -0.45% -0.14% -0.15% -0.09% -0.10% 0.00% -0.19% Dose (eV/g·hist) PA AP RAO 90 LAO 90 RAO 45 LAO 45 CAUD 45 CRAN 45 Relative Error Sum Relative Error Average Air 0.10% -0.05% 0.06% 0.00% 0.17% 0.07% -0.01% 0.00% 0.03% -100.00% Muscle 0.00% 0.00% -0.01% 0.00% 0.00% 0.00% 0.01% -0.01% 0.00% 0.00% Soft tissue 0.01% 0.01% -0.01% 0.02% -0.02% -0.02% -0.02% 0.02% 0.00% 0.00% Bone 0.00% -0.01% 0.00% 0.00% 0.02% 0.00% 0.01% 0.02% 0.01% 0.01% Cartilage -0.03% -0.05% 0.00% -0.07% 0.01% 0.01% 0.08% -0.04% -0.02% -0.02% Adipose 0.02% -0.01% 0.00% 0.01% 0.01% -0.01% 0.00% 0.00% 0.00% 0.00% Blood -0.04% 0.02% 0.03% 0.01% 0.04% 0.04% -0.03% 0.03% 0.01% 0.01% Skin 0.00% -0.01% 0.01% -0.01% 0.00% -0.03% 0.01% 0.01% 0.00% 0.00% Lung 0.00% -0.01% 0.02% -0.04% 0.01% -0.01% -0.01% 0.01% -0.01% -0.01% 126 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 127 BIBLIOGRAPHY 14. BIBLIOGRAPHY The last section of this thesis lists all the books, articles and publications consulted during the project. The bibliography is divided in two categories: the first one includes the most important references used in the learning stage; the second one shows the complementary reading to go into detail of the project and the matter dealt with. It is important to note that most of the references used are indicated when have been used during the thesis. 14.1 BIBLIOGRAPHICAL REFERENCES [1] Andreu Badal. MC-GPU v1.3 documentation generated by Doxygen. [2] Andreu Badal, Fahad Zafar, Han Dong, and Aldo Badano. A real-time radiation dose monitoring system for patients and staff during interventional fluoroscopy using a GPU-accelerated Monte Carlo simulator and an automatic 3D localization system based on a depth camera. SPIE 8668, Medical Imaging 2013: Physics of Medical Imaging, 866828 (19 March 2013); DOI: 10.1117/12.2008031 [3] Andreu Badal and Aldo Badano. Accelerating Monte Carlo simulations of photon transport in a voxelized geometry using a massively parallel graphics processing unit. Medical Physics 36, 4878-4880 (2009); DOI: 10.1118/1.3231824 [4] Josep Sempau, Andreu Badal and Lorenzo Brualla. A PENELOPE-based system for the automated Monte Carlo simulation of clinacs and voxelized geometriesapplication to far-from-axis fields. Med. Phys 38. Year 2011. PMID: 22047353 DOI: 10.1118/1.3643029 [5] Francesc Salvat. PENELOPE-2014. A Code System for Monte Carlo Simulation of Electron and Photon Transport (Nuclear Energy Agency OECD, 2015). Available at https://www.oecd-nea.org/science/docs/2015/nsc-doc2015-3.pdf [6] Xavier Ortega Aramburu and Jaume Jorba Bisbal. Radiaciones Ionizantes: Utilización y riesgos vol. I and II. ISBN: 84-8301-170-0 [7] Sara Principi. Doctoral Dissertation. Development of Methodologies for Estimating the Dose to the Eye Lens in Interventional Radiology. Operational Implications of the Eye Lens New Dose Limit. [8] Glenn F.Knoll. Radiation Detection and Measurement. John Wiley & Sons Inc; 4th Edition. ISBN-10: 0470131489. ISBN-13: 978-0470131480 [9] Frank Herbert Attix. Introduction to radiological physics and radiation dosimetry. Online ISBN: 9783527617135. DOI: 10.1002/9783527617135 [10] ICRP, 2007. The 2007 Recommendations of the International Commission on Radiological Protection. ICRP Publication 103. Ann. ICRP 37 (2-4). [11] ICRP, 2009. Adult Reference Computational Phantoms. ICRP Publication 110. Ann. ICRP 39 (2). 128 Optimization of patient dose monitoring in fluoroscopically-guided interventional procedures 14.2 FURTHER READING Vanhavere F., Carinou E., Gualdrini G., Clairand I., Sans Merce M., Ginjaume M., Nikodemova D., Jankowski J., Bordy J-M., Rimpler A., Wach S., Martin P., Struelens L., Krim S., Koukorava C., Ferrari P., Mariotti F., Fantuzzi E., Donadille L., Itié C., Ruiz N., Carnicer A., Fulop M., Domienik J., Brodecki M., Daures J., Barth I., Bilski P. EURADOS Report 2012-02. ORAMED: Optimization of Radiation Protection of Medical Staff. Implications of Medical Low Dose Radiation Exposure. MEDIRAD. NFRP-2016-2017 EU Council Directive 2013/59/EURATOM of 5 December 2013 laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation, and repealing Directives 89/618/Euratom, 90/641/Euratom, 96/29/Euratom, 97/43/Euratom and 2003/122/Euratom. Published in the Official Journal of the European Union. MC-GPU HANDBOOK Version 1.0 2017 -2018 MC-GPU USER MANUAL ÁLVARO MERINO CANETE UPC | INTE 1 OUTLINE MC-GPU DESCRIPTION 4 PURPOSE 4 PREVIOUS KNOWLEDGE 5 Interventional Radiology (IR) 5 X-Rays / Equipment 5 X-Ray Circuit 6 X-ray tube (Source) 6 Image intensifier (Detector) 7 Risks 7 Patient Risks 7 Operator Risks 8 Monte Carlo simulation 8 Voxelized geometry 8 Physics 10 X-Ray 10 Bremsstrahlung 10 Characteristic X-rays 11 Units and Definitions 12 Ionizing radiation 12 Energy 12 Energy fluence 12 KERMA (Kinetic Energy Released per unit Mass) 12 Air KERMA 12 Absorbed Dose D 13 Absorbed-Dose Rate 𝐷 13 Equivalent dose, HT 13 Effective dose, E 13 Mass attenuation coefficient 14 Photon Interactions 14 Photoelectric absorption 14 Compton scattering 15 Rayleigh scattering 15 Pair production 15 Biological effects of ionizing radiation effect 15 EXTENDED KNOWLEDGE 16 SIMULATION TIPS 16 2 PRE-SIMULATION INFORMATION 17 GENERAL SCHEME 18 MAIN INPUT FILE SECTIONS 19 SIMULATION CONFIGURATION SECTION 19 SOURCE SECTION 19 IMAGE DETECTOR SECTION 20 CT SCAN TRAJECTORY SECTION 20 DOSE DEPOSITION SECTION 21 VOXELIZED GEOMETRY SECTION 21 MATERIAL FILES LIST SECTION 21 OTHER INPUT FILES 22 SPECTRUM FILE 22 VOXEL GEOMETRY FILE 23 MATERIAL FILES 24 PROGRAM EXECUTION 24 Initialization 24 Launching steps 24 OUTPUT FILES AND POST PROCESSED 25 MAIN OUTPUT FILE 25 VOXEL ROI DOSE TALLY REPORT 25 MATERIALS TOTAL DOSE TALLY REPORT 25 TEXT OUTPUT FILES 26 IMAGE DETECTOR TEXT FILES 26 DOSE TEXT FILE 26 IMAGE OUTPUT FILES 27 IMAGE DETECTOR RAW FILE 27 DOSE RAW FILE 28 OTHER METHODS 28 IMAGEJ 29 3 MC-GPU BETA VERSION 30 MC-GPU BETA DESCRIPTION 30 PRE-SIMULATION INFORMATION 31 GENERAL SCHEME 32 INPUT FILES 33 MAIN INPUT FILE 33 SECTION OPERATOR DOSE 33 SECONDARY INPUT FILE 34 OPERATOR DATA SECTION 34 SOURCE DATA SECTION 34 SHIELD SECTION 35 PROGRAM EXECUTION 35 STEP ONE 36 Initialization 36 Launching steps 36 STEP TWO 36 Initialization 36 Launching steps 36 OUTPUT FILES 37 MAIN OUTPUT FILE 37 TEXT OUTPUT FILES 37 IMAGE DETECTOR TEXT FILES 37 DOSE TEXT FILES 38 IMAGE OUTPUT FILES 38 IMAGE DETECTOR RAW FILES 38 DOSE RAW FILES 38 4 MC-GPU DESCRIPTION MC-GPU is a Monte Carlo simulation code that generates synthetic radiographic images and computed tomography (CT) scans of realistic models of the human anatomy using the computational power of commodity Graphics Processing Unit (GPU) cards. The code implements a massively multi-threaded Monte Carlo simulation algorithm for the transport of x rays in a voxelized geometry. The x ray interaction models and material properties have been adapted from PENELOPE 2006. MC-GPU was developed using the CUDA programming model from NVIDIA to achieve maximum performance on NVIDIA GPUs. The code can also be compiled with a standard C compiler to be executed in a regular CPU. In a typical medical imaging simulation, the use of GPU computing with MC-GPU has been shown to provide a speed up of between 20 and 40 times, compared to the execution on a single CPU core. The MC-GPU code has been described in different scientific publications. The main reference of this work, which the users should cite, is the following: Andreu Badal and Aldo Badano, "Accelerating Monte Carlo simulations of photon transport in a voxelized geometry using a massively parallel Graphics Processing Unit", Medical Physics 36, pp. 4878-4880 (2009) The main developer of MC-GPU is Andreu Badal, working at the U.S. Food and Drug Administration (Centre for Devices and Radiological Health, Office of Science and - Engineering Laboratories, Division of Imaging and Applied Mathematics). The source code of MC-GPU is free and open software in the public domain, as explained in the Disclaimer section below. The source code of MC-GPU and its auxiliary files are distributed from the website: http://code.google.com/. PURPOSE The purpose of this manual is to help the new users to use the MC-GPU code to simulate a real interventional radiology (IR), also known as vascular and interventional radiology (VIR), in order to obtained in the most precise way the dose received by the patient and by the physician during the intervention. This manual is structured to learn how to use the MC-GPU program in which only the patient is simulated and then learn the extended version (MC-GPU_Beta) in which the physician, hereinafter operator, is also simulated. It is divided in sections in which the previous and the posterior steps of a simulation, as well as how to run a simulation, are explained. 5 PREVIOUS KNOWLEDGE Before carrying out the simulations it is important to have some basic knowledge about the procedure to be simulated, physics, Monte Carlo simulations and practical aspects of the program configuration itself. Interventional Radiology (IR) An interventional radiological procedure is any procedure using radiological imaging equipment in order to guide a therapeutic/invasive procedure on a patient. Examples of such procedures include angiography, angioplasty, embolization, biopsy and drainage, dilations and stent placements. Interventional radiology has gained importance in the fields of cardiology and neurology, vascular and non-vascular medicine and often represents an alternative to more hazardous surgery. Fluoroscopy is predominantly used, which is a way of working in real time. Currently almost exclusively fluoroscopy is used. Interventional procedures are complex and generally involve the use of long fluoroscopy times. Consequently, there is a potential for high radiation doses to patients and staff as compared to other X-ray examinations. X-Rays / Equipment Fluoroscopy or radioscopy is an imaging technique used in medicine to obtain real-time images of the internal structures of patients by using a fluoroscope. The basic fluoroscopic equipment includes an x-ray source designed to give off pulse or continuous radiation and an image intensifier between which the patient is placed. Both elements are mounted on a C-arm. In its simplest form, a fluoroscope consists of an X-ray source and a fluorescent screen between which the patient is placed. However, modern fluoroscopes couple the screen to an X-ray image intensifier and a CCD video camera, allowing images to be recorded and played back on a monitor. The use of X-rays, a type of ionizing radiation, requires that the potential risks of a procedure be carefully weighed against the expected benefits to the patient. Although physicians always try to use low doses of radiation during fluoroscopies, the duration of a typical procedure often results in a relatively high absorbed dose to the patient. Recent advances include the digitization of captured images and flat panel detector systems that further reduce the radiation dose for patients. In operation, when the x-ray source emits x-rays toward the patient, some radiation typically passes through the patient and impinges on the detector. As the radiation passes through the patient, anatomical structures of different densities inside the patient cause intensity variations of the transmitted radiation that goes into the detector which transmits the image to a monitor. Nowadays, the use of intensifier tubes and modern digital flat panel receptors make it possible to optimize the balance of patient exposure with high image quality so as not to expose the patient to unnecessary radiation levels. "C-arm" mobile fluoroscopy machines are often colloquially referred to as image intensifiers, however strictly speaking the image intensifier is only one part of the machine (namely the detector). 12 Units and Definitions Ionizing radiation Ionization produced by particles is the process by which one or more electrons are liberated in collisions of the particles with atoms or molecules. This can be distinguished from excitation, which is a transfer of electrons to higher energy levels in atoms or molecules and generally requires less energy. When charged particles have slowed down sufficiently, ionization becomes less likely or impossible, and the particles increasingly dissipate their remaining energy in other processes such as excitation or elastic scattering. Thus, near the end of their range, charged particles that were ionizing can be considered to be non-ionizing. The term ionizing radiation refers to charged particles (electrons or protons) and uncharged particles (photons or neutrons) that can produce ionizations in a medium or can initiate nuclear or elementary-particle transformations that then result in ionization or the production of ionizing radiation. Energy The traditional unit for measurement of radiation energy is the electron volt or eV, defined as the kinetic energy gained by an electron by its acceleration through a potential difference of 1 V. The multiples of kiloelectron volt (keV) and megaelectron volt (MeV) are more common in the measurement of energies for ionizing radiation. The electron volt is a convenient unit when dealing with particulate radiation because the energy gained from an electric field can easily be obtained by multiplying the potential difference by the number of electronic charges carried by the particle 1 𝑒𝑉 =1.602·10−19 𝐽 Energy fluence It is quotient of dR by da, where dR is the radiant energy (energy of particles excluding the rest energy) on a sphere of cross-sectional area da. Its unit is J/m2 or eV/cm2. The use of a sphere of cross-sectional area da expresses in the simplest manner the fact that one considers an area da perpendicular to the direction of each particle. KERMA (Kinetic Energy Released per unit Mass) The kerma, K, for ionizing uncharged particles, is the mean energy transferred (sum of the initial kinetic energies of all the charged particles liberated) to charged particles by uncharged ionizing radiation per unit mass. Its unit is gray Gy (J/kg) Air KERMA It is of importance in the practical calibration of instruments for photon measurement, where it is used for the traceable calibration of gamma instrument metrology facilities using a "free air" ion chamber to measure air kerma. The IAEA safety report 16 states "The quantity air kerma should be used for calibrating the reference photon radiation fields and reference instruments. Radiation protection monitoring instruments should be calibrated in terms of dose equivalent quantities. Area dosimeters or dose parameters should be calibrated in terms of the ambient dose equivalent, H*(10), or the directional dose equivalent, H′(0.07),without any phantom present, i.e. free in air." 13 Conversion coefficients from air kerma in Gy to equivalent dose in Sv are published in the International Commission on Radiological Protection (ICRP) report 74. Absorbed Dose D The absorbed dose, D, is the mean energy imparted by ionizing radiation to matter per unit mass. Its unit is gray Gy (J/kg) Absorbed-Dose Rate 𝐷󰇗 It is defined as the increment of absorbed dose in a time interval. Its unit is Gy/s (J/kg·s) Mean absorbed dose in a tissue or organ, DT It is the absorbed dose averaged over the tissue or organ T. 𝐷𝑇=𝜀𝑇 𝑚𝑇 Where 𝜀𝑇 is the mean total energy imparted in a tissue or organ T and mT is the mass of that tissue or organ. Equivalent dose, HT For a biological material, the probability of radiation effects depends not only on the absorbed dose, but also on the type and energy of the radiation causing the dose. The dose in a tissue or organ T is given by: 𝐻𝑇=∑𝑤𝑅·𝐷𝑇,𝑅 𝑅 Where DT,R is the mean absorbed dose from radiation R in a tissue or organ T, and wR is the radiation weighting factor. Since wR is dimensionless, the unit for the equivalent dose is the same as for absorbed dose, J/kg, and its special name is sievert (Sv). The radiation weighting factors are set by International Commission on Radiological Protection (ICRP) Effective dose, E As different tissues have different sensitivity to ionizing radiation the effective dose is defined as the tissue-weighted sum of the equivalent doses in all specified tissues and organs of the body, given by the expression 𝐸 =∑𝑤𝑇 𝑇∑𝑤𝑅·𝐷𝑇,𝑅 𝑅=∑𝑤𝑇·𝐻𝑇 𝑇 where HT or wR·DT,R is the equivalent dose in a tissue or organ, T, and wT is the tissue weighting factor. The unit for the effective dose is the same as for absorbed dose, J/kg, and its special name is sievert (Sv). The tissue weighting factors are set by International Commission on Radiological Protection (ICRP) 14 Mass attenuation coefficient The mass attenuation coefficient of the volume of a material characterizes how easily it can be penetrated by a beam of light or other energy or matter. In addition to visible light, mass attenuation coefficients can be defined for other electromagnetic radiation such as X-rays. The SI unit of mass attenuation coefficient is the square metre per kilogram (m2/kg). Other common units include cm2/g (the most common unit for X-ray mass attenuation coefficients). The values of mass attenuation coefficients are dependent upon the absorption and scattering of the incident radiation caused by several different mechanisms. Photon Interactions Although a large number of possible interaction mechanisms are known for gamma rays in matter, only four major types play an important role in radiation measurements (figure). All these processes lead to the partial or complete transfer of the photon energy to electron energy. They result in sudden and abrupt changes in the photon history, in that the photon either disappears entirely or is scattered through a significant angle. This behaviour is in marked contrast to the charged particles discussed earlier in this chapter, which slow down gradually through continuous, simultaneous interactions with many absorber atoms. Figure 9: Photon interactions 4 Photoelectric absorption In the photoelectric absorption process, a photon undergoes an interaction with an absorber atom in which the photon completely disappears. In its place, an energetic photoelectron is ejected by the atom from one of its bound shells. The interaction is with the atom as a whole and cannot take place with free electrons. For photon of sufficient energy, the most probable origin of the photoelectron is the most tightly bound or K shell of the atom. In addition to the photoelectron, the interaction also creates an ionized absorber atom with a vacancy in one of its bound shells. This vacancy is quickly filled through capture of a free electron from the medium and/or rearrangement of electrons from other shells of the atom. Therefore, one or more characteristic X-ray photons may also be generated. Although in most cases these X-rays are reabsorbed close to the original site through photoelectric absorption involving less tightly bound shells, their migration and possible escape from radiation detectors can influence their response. In some fraction of the cases, the emission of an Auger electron may substitute for the characteristic X-ray in carrying away the atomic excitation energy. The threshold for ejection of an electron of a given shell is just the ionization energy of the shell. 4 PENELOPE: A code system for Monte Carlo simulation of electron and photon transport 15 Compton scattering The interaction process of Compton scattering takes place between the incident photon and an electron in the absorbing material. It is most often the predominant interaction mechanism for gamma-ray energies typical of radioisotope sources. In Compton scattering, the incoming photon is deflected through an angle with respect to its original direction. The photon transfers a portion of its energy to the electron (assumed to be initially at rest), which is then known as a recoil electron. Because all angles of scattering are possible, the energy transferred to the electron can vary from zero to a large fraction of the photon energy. This process has a threshold for each shell of bound electrons corresponding to the ionization energy of the shell. Rayleigh scattering In addition to Compton scattering, another type of scattering can occur in which the photon interacts coherently with all the electrons of an absorber atom. This coherent scattering or Rayleigh scattering process neither excites nor ionizes the atom, and the photon retains its original energy after the scattering event. Because virtually no energy is transferred, this process is often neglected in basic discussions of interactions. This term is usually associate to the scattering of light by particles of size much smaller than the wavelength. Its cross section increases rapidly with photon energy. This elastic scattering from atoms has the only cross-section that has no energy threshold. Pair production If the photon energy exceeds twice the rest-mass energy of an electron (1.02 MeV), the process of pair production is energetically possible. As a practical matter, the probability of this interaction remains very low until the photon energy approaches several MeV and therefore pair production is predominantly confined to high-energy gamma rays. In the interaction (which must take place in the coulomb field of a nucleus), the photon disappears and is replaced by an electron-positron pair. All the excess energy carried in by the photon above the 1.02 MeV required to create the pair goes into kinetic energy shared by the positron and the electron. Because the positron will subsequently annihilate after slowing down in the absorbing medium, two annihilation photons are normally produced as secondary products of the interaction. Biological effects of ionizing radiation effect The biological effects due to an ionizing radiation are divided in two groups: - Deterministic Effects: most of organs and tissues are not affected by the loss of certain number of cells but if the amount of killed cells becomes large, a loss of functional capacity of the tissue will be observed. These reactions that may occur early or late after irradiation are the deterministic effects. The appearance of these effects depends on a threshold dose, below which the effects are null. This limit depends on dose delivery mode and may differ in different persons. - Stochastic Effects: the killing of one or a small number of cells will, in most cases, have no consequences in tissue, but modification in single cells, such as genetic changes or transformations leading ultimately to malignancy, may have serious consequences, this is the case of stochastic effects. The severity of these effects is not dose dependent but the probability increases with the dose. 16 Extended knowledge All information presented in the previous sections mean a general overview of the most important concepts need to understand the program but it is only a brief explanation. For To go deeper and learn more about the subject, the recommended lectures where this information were obtained are the following: - The Physics Of Radiation Therapy – Faiz M. Khan - PENELOPE-2014: A Code System for Monte Carlo Simulation of Electron and Photon Transport – Francesc Salvat - ICRU Report nº 85 Fundamental Quantities and Units for Ionizing Radiation - ICRP Publication 103: The 2007 Recommendations of the International Commission on Radiological Protection - Introduction to radiological physics and radiation dosimetry – Frank Herbert Attix - Radiation Detection and Measurement – Glenn F.Knoll - Practical Gamma-ray Spectrometry – Gordon R.Gilmore - Radiaciones ionizantes, Utilización y riesgos I y II – Xavier Ortega y Jaume Jorba (Spanish) - Documentation given by MC-GPU and penEasy distributors Simulation tips A few little advices for creating the plain text inputs: - The location of all the values in the main input file must be respected. The program needs all the default lines in its adequate position. The only section may increase or decrease its extension is [SECTION MATERIAL FILE LIST] which depends on the voxelized geometry. - All lines must have the number of values what is expected. - The space between values in each line must be respected. - In order to avoid some simulation problems is recommended to respect the space for the minus even for positive values and to type most of the values with one decimal even for integer numbers (except in the [SECTION SIMULATION CONFIG]) - For all the plain text input files the symbol # means a comment and the executables do not read the lines which have it at the beginning. 17 PRE-SIMULATION INFORMATION MC-GPU has been developed and tested only in the Linux operating system. A Makefile script is provided to compile the MC-GPU code in Linux. The CUDA libraries and the GNU GCC compiler must be previously installed. All information need to code compilation are explained in MC-GPU_v1.3_README.pdf which is included in the compressed directory which can be downloaded from https://code.google.com/archive/p/mcgpu/downloads. Once the directory has been decompressed, the user has the necessary libraries to compile the code, some important documentation, the material files library, the gnuplot’s scripts and some examples. All information provided in this handbook henceforth is in case of using the Argos Cluster located into the INTE/SEN facilities (Barcelona) Argos2 server Port 222 where all the libraries and resources need to compile de code are previously installed. The user just need to compile de code following the instructions collected in the mentioned MC-GPU_v1.3_README.pdf. It is important to highlight that the user needs to have permission to access to this Cluster and its servers and ports. In order to use the MC-GPU system efficiently some auxiliary tools that are not included in the distribution package will be needed. They are the following: 1. A powerful file manager will also make some tasks considerably easier. To change files from the cluster to the personal computer, WinSCP is recommended. It can be downloaded here: https://winscp.net/eng/download.php 2. An SSH and telnet client which allow you to run simulations in the cluster. In this case the recommendation is Putty It can be downloaded here: http://www.putty.org/ 3. A plain text editor. A text editor serves to edit plain text (ASCII) files, which do not contain ‘implicit’ formatting instructions and it is also useful an output text files. The use of Notepad++ is recommended. It can be downloaded here: https://notepad-plus-plus.org 4. A graphic program. The recommendation is Gnuplot, a powerful 2D and 3D plotting program, also from the GNU project, that is free and open source. It can be downloaded here: http://www.gnuplot.info 5. A graphics program to visualize raw files and volumes. Those files may be read with Octave, Image J, Paraview, Slicer or other programs but the recommendation is ImageJ. It can be downloaded here: https://imagej.nih.gov/ij/ 6. Spectrum file generator. Spekcal and XCOMP5R (free) are specific programs in this area. This last one is recommended due to it is free and easy to use. However, this program runs only on computers with MS-DOS operating systems, for this reason, a MS-DOS simulator like DOSBOX is need. 18 GENERAL SCHEME The scheme shown below gives a general view of the necessary files to run a simulation as well as the process to follow before and after a simulation. Figure 10: MC-GPU general scheme It is recommended to have all the inputs (spectrum, geometry, materials and main input file) and the executable in the same directory. In addition, the creation of a directory for each simulation is recommended. 19 MAIN INPUT FILE SECTIONS SIMULATION CONFIGURATION SECTION LINE NAME EXAMPLE DESCRIPTION 1 Total number of histories 1E8 Number of histories5 to simulate. If it is lower than 1E5, program will take this value as simulation time in seconds. This value set the number of X-rays. 2 Random seed 1234567890 The seed determines the sequence of random numbers utilised during the simulation. Use the same number to compare different simulations 3 GPU number to use when MPI is not used 0 The use of a 0 value avoids MPI runs 4 GPU threads per CUDA block 128 This value must be multiple of 32. Recommended value: 128 5 Simulated histories per GPU thread 100 Recommended value: 100 Note: time simulation depends directly on the total number of histories required. SOURCE SECTION The emitted cone beam is computationally collimated to produce a rectangular field on the detector plane, realistic approach, within the azimuthal and polar angles specified by the user. LINE NAME EXAMPLE DESCRIPTION 1 X-Ray energy spectrum 90keV.spc This spectrum file is described in Other input files section 2 Source position 15.0 -45.0 15.0 Cartesian coordinates in centimetres of source position origin 3 Source direction cosines 0.0 1.0 0.0 This normal vector determines the direction of the beam 4 Polar and azimuthal apertures for the fan beam -45.0 -45.0 Angles in degrees that determine the aperture of the pyramid. Input negative to automatically cover the whole detector (recommended) 5 In Monte Carlo simulation of radiation transport, the history (track) of a particle is viewed as a random sequence of free flights that end with an interaction event where the particle changes its direction of movement, loses energy and, occasionally, produces secondary particles. 20 IMAGE DETECTOR SECTION The detector plane is automatically located at the specified distance right in front of the source focal spot, with the collimated cone beam pointing towards the geometric centre of the detector. LINE NAME EXAMPLE DESCRIPTION 1 Output image file name mc-gpu_image.dat Name of the .dat and .raw output files 2 Number of pixels in the detector image 120.0 120.0 Number of pixels in X and Z direction of the detector image 3 Detector image size 120.0 120.0 Width (x-axis) and height (z-axis) in centimetres of the detector image 4 Source-to-detector distance 100.0 Perpendicular distance from point source to detector centre Note: in order to obtain a good detector image is recommended use a value for number of pixels multiple of the detector image features. CT SCAN TRAJECTORY SECTION MC-GPU simulates a single projection image or a full CT scan. The CT is simulated generating many projection images around the static voxelized geometry. The code is limited to perform a simple CT trajectory rotating around the Z axis. LINE NAME EXAMPLE DESCRIPTION 1 Number of projections 2 Beam must be perpendicular to Z axis and detector. Set to 1 for a single projection 2 Angle between projections 45.0 Angle in degrees between the different projections required. For full CT, 360/n_projections 3 Angles of interest 0.0 3600.0 Projections outside the input interval will be skipped The recommend values are shown 4 Source-to-rotation axis distance 60.0 The rotation radius around the Z axis 5 Vertical translation between projections 0.0 This is use to obtain a helical scan Note: the results are given separately and jointly 21 DOSE DEPOSITION SECTION 6 The tally 7 cards are used to specify what type of information the user wants to gain from the Monte Carlo calculation. This information is requested by the user by using a combination of the following cards. For each answer the fractional standard deviation, relative error, is provided. LINE NAME EXAMPLE DESCRIPTION 1 Tally material dose YES X-ray locally deposited at interaction Options: YES / NO 2 Tally 3D voxel dose YES Dose measured separately for each voxel. Options: YES / NO 3 Output voxel dose file name mc-gpu_dose.dat Name of the .dat and .raw output files 4 VOXEL DOSE ROI: X-index 1 3 Range in X axis of ROI Xmin Xmax 5 VOXEL DOSE ROI: Y-index 1 1 Range in Y axis of ROI Ymin Ymax 6 VOXEL DOSE ROI: Z-index 1 2 Range in X axis of ROI Zmin Zmax VOXELIZED GEOMETRY SECTION In this section the voxelized geometry to simulate is included, there are two different ways of introduce it: - The name of the file if it is in the same directory of the input - The name of the path to the file MATERIAL FILES LIST SECTION In this section the material files are included, there are two different ways of introduce them: - The name of the files if they are in the same directory of the input - The name of the path to the files Each material must be listed in the adequate position to correspond with the ID number of the material in the voxelized geometry. All must have the same number of bins. 6 ROI: Region Of Interest 7 MCNPX User’s Manual